1 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
4 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
6 (*cmp_ite0): Add enabled_for_depr_it attribute.
9 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
11 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
14 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
16 * tree-vrp.c (vrp_int_const_binop): Return true on success and
17 return the value by pointer.
18 (extract_range_from_multiplicative_op_1): Update accordingly.
19 Return as soon as an operation fails.
21 2017-11-05 Tom de Vries <tom@codesourcery.com>
24 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
25 (DEF_SANITIZER_BUILTIN): ... here.
26 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
27 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
29 2017-11-05 Tom de Vries <tom@codesourcery.com>
32 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
34 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
35 ASM_OUTPUT_BEFORE_CASE_LABEL call.
36 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
38 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
39 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
40 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
42 2017-11-05 Tom de Vries <tom@codesourcery.com>
45 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
48 2017-11-04 Michael Clark <michaeljclark@mac.com>
50 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
51 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
53 (*addsi3_extended): Likewise.
54 (*addsi3_extended2): Likewise.
55 (<optab>si3): Likewise.
56 (<optab>di3): Likewise.
57 (<optab><mode>3): Likewise.
58 (<*optabe>si3_internal): Likewise.
59 (zero_extendqi<SUPERQI:mode>2): Likewise.
60 (*add<mode>hi3): Likewise.
61 (*xor<mode>hi3): Likewise.
62 (<optab>di3): Likewise.
63 (*<optab>si3_extend): Likewise.
64 (*sge<u>_<X:mode><GPR:mode>): Likewise.
65 (*slt<u>_<X:mode><GPR:mode>): Likewise.
66 (*sle<u>_<X:mode><GPR:mode>): Likewise.
68 2017-11-04 Andrew Waterman <andrew@sifive.com>
70 * config/riscv/riscv.c (riscv_option_override): Conditionally set
71 TARGET_STRICT_ALIGN based upon -mtune argument.
73 2017-11-04 Andrew Waterman <andrew@sifive.com>
75 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
77 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
79 config/i386/i386.c (choose_basereg): Use optional scratch
80 register and add assertion.
81 (x86_emit_outlined_ms2sysv_save): Use scratch register when
82 needed, and don't allocate stack.
83 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
84 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
85 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
87 2017-11-03 Jeff Law <law@redhat.com>
89 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
90 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
91 to probe at the start of a noreturn function.
93 2017-11-03 Jakub Jelinek <jakub@redhat.com>
95 PR tree-optimization/78821
96 * gimple-ssa-store-merging.c: Update the file comment.
97 (MAX_STORE_ALIAS_CHECKS): Define.
98 (struct store_operand_info): New type.
99 (store_operand_info::store_operand_info): New constructor.
100 (struct store_immediate_info): Add rhs_code and ops data members.
101 (store_immediate_info::store_immediate_info): Add rhscode, op0r
102 and op1r arguments to the ctor, initialize corresponding data members.
103 (struct merged_store_group): Add load_align_base and load_align
105 (merged_store_group::merged_store_group): Initialize them.
106 (merged_store_group::do_merge): Update them.
107 (merged_store_group::apply_stores): Pick the constant for
108 encode_tree_to_bitpos from one of the two operands, or skip
109 encode_tree_to_bitpos if neither operand is a constant.
110 (class pass_store_merging): Add process_store method decl. Remove
111 bool argument from terminate_all_aliasing_chains method decl.
112 (pass_store_merging::terminate_all_aliasing_chains): Remove
113 var_offset_p argument and corresponding handling.
114 (stmts_may_clobber_ref_p): New function.
115 (compatible_load_p): New function.
116 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
117 if there is overlap and rhs_code is not INTEGER_CST. For
118 non-overlapping stores terminate group if rhs is not mergeable.
119 (get_alias_type_for_stmts): Change first argument from
120 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
121 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
122 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
124 (get_location_for_stmts): Change first argument from
125 auto_vec<gimple *> & to vec<gimple *> &.
126 (struct split_store): Remove orig_stmts data member, add orig_stores.
127 (split_store::split_store): Create orig_stores rather than orig_stmts.
128 (find_constituent_stmts): Renamed to ...
129 (find_constituent_stores): ... this. Change second argument from
130 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
131 to info structures rather than the statements.
132 (split_group): Rename ALLOW_UNALIGNED argument to
133 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
134 it. Adjust find_constituent_stores caller.
135 (imm_store_chain_info::output_merged_store): Handle rhs_code other
136 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
137 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
138 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
139 (mem_valid_for_store_merging): New function.
140 (handled_load): New function.
141 (pass_store_merging::process_store): New method.
142 (pass_store_merging::execute): Use process_store method. Adjust
143 terminate_all_aliasing_chains caller.
145 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
147 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
148 Return true for more constants, symbols and label references.
149 (aarch64_valid_floating_const): Remove unused function.
151 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
154 * config/aarch64/aarch64.c (aarch64_layout_frame):
155 Undo forcing of LR at bottom of frame.
157 2017-11-03 Jeff Law <law@redhat.com>
160 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
161 for int_registers_saved.
163 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
164 extracted from tree-ssa-dom.c.
165 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
166 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
167 (record_equivalences_from_incoming_edge): Add additional argument
168 to single_pred_edge_ignoring_loop_edges call.
169 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
170 (uncprop_dom_walker::before_dom_children): Add additional argument
171 to single_pred_edge_ignoring_loop_edges call.
172 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
173 single_pred_edge_ignoring_loop_edges rather than open coding.
174 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
176 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
178 * match.pd (-(-A)): Rewrite.
180 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
182 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
183 (rs6000_emit_int_cmove): New declaration.
184 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
185 (rs6000_emit_sISEL): Delete.
186 (rs6000_emit_int_cmove): Make non-static.
187 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
188 instead of rs6000_emit_sISEL.
190 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
192 * asan.c (create_cond_insert_point): Maintain profile.
193 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
195 * basic-block.h (struct basic_block_def): Remove frequency.
196 (EDGE_FREQUENCY): Use to_frequency
197 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
199 (find_traces): Update to use to_frequency.
200 (find_traces_1_round): Likewise; use only IPA counts.
201 (bb_to_key): Likewise.
202 (connect_traces): Use IPA counts only.
203 (copy_bb_p): Update to use to_frequency.
204 (fix_up_crossing_landing_pad): Likewise.
205 (sanitize_hot_paths): Likewise.
206 * bt-load.c (basic_block_freq): Likewise.
207 * cfg.c (init_flow): Set count_max to uninitialized.
208 (check_bb_profile): Remove frequencies; check counts.
209 (dump_bb_info): Do not dump frequencies.
210 (update_bb_profile_for_threading): Update counts only.
211 (scale_bbs_frequencies_int): Likewise.
212 (MAX_SAFE_MULTIPLIER): Remove.
213 (scale_bbs_frequencies_gcov_type): Update counts only.
214 (scale_bbs_frequencies_profile_count): Update counts only.
215 (scale_bbs_frequencies): Update counts only.
216 * cfg.h (struct control_flow_graph): Add count-max.
217 (update_bb_profile_for_threading): Update prototype.
218 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
219 (find_many_sub_basic_blocks): Likewise.
220 * cfgcleanup.c (try_forward_edges): Likewise.
221 (try_crossjump_to_edge): Likewise.
222 * cfgexpand.c (expand_gimple_cond): Likewise.
223 (expand_gimple_tailcall): Likewise.
224 (construct_init_block): Likewise.
225 (construct_exit_block): Likewise.
226 * cfghooks.c (verify_flow_info): Check consistency of counts.
227 (dump_bb_for_graph): Do not dump frequencies.
228 (split_block_1): Do not update frequencies.
229 (split_edge): Do not update frequencies.
230 (make_forwarder_block): Do not update frequencies.
231 (duplicate_block): Do not update frequencies.
232 (account_profile_record): Do not update frequencies.
233 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
234 for global heuristics.
235 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
236 (expected_loop_iterations_unbounded): Use counts only.
237 * cfgloopmanip.c (scale_loop_profile): Simplify.
238 (create_empty_loop_on_edge): Simplify
240 (duplicate_loop_to_header_edge): Simplify
241 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
242 (update_br_prob_note): Take care of removing note when profile
244 (relink_block_chain): Do not dump frequency.
245 (rtl_account_profile_record): Use to_frequency.
246 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
247 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
248 (cgraph_update_edges_for_call_stmt_node): Likewise.
249 (cgraph_edge::verify_count_and_frequency): Update.
250 (cgraph_node::verify_node): Temporarily disable frequency verification.
251 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
253 (cgraph_edge::rebuild_edges): Convert to ipa counts.
254 * cgraphunit.c (init_lowered_empty_function): Do not initialize
256 (cgraph_node::expand_thunk): Update profile.
257 * except.c (dw2_build_landing_pads): Do not update frequency.
258 * final.c (compute_alignments): Use to_frequency.
259 (dump_basic_block_info): Do not dump frequency.
260 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
261 (dump_gimple_bb_header): Do not dump frequency.
262 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
264 * gimple-streamer-in.c (input_bb): Do not stream frequency.
265 * gimple-streamer-out.c (output_bb): Do not stream frequency.
266 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
267 (init_before_recovery): Do not update frequency.
268 (sched_create_recovery_edges): Do not update frequency.
269 * hsa-gen.c (convert_switch_statements): Do not update frequency.
270 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
271 (ipa_cp_c_finalize): Set max_count to uninitialized.
272 * ipa-fnsummary.c (get_minimal_bb): Use counts.
273 (param_change_prob): Use counts.
274 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
276 * ipa-split.c (consider_split): Use to_frequency.
277 (split_function): Use to_frequency.
278 * ira-build.c (loop_compare_func): Likewise.
279 (mark_loops_for_removal): Likewise.
280 (mark_all_loops_for_removal): Likewise.
281 * loop-doloop.c (doloop_modify): Do not update frequency.
282 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
284 * lto-streamer-in.c (input_function): Update count_max.
285 * omp-expand.c (expand_omp_taskreg): Update count_max.
286 * omp-simd-clone.c (simd_clone_adjust): Update profile.
287 * predict.c (maybe_hot_frequency_p): Use to_frequency.
288 (maybe_hot_count_p): Use ipa counts only.
289 (maybe_hot_bb_p): Simplify.
290 (maybe_hot_edge_p): Simplify.
291 (probably_never_executed): Do not take frequency argument.
292 (probably_never_executed_bb_p): Do not pass frequency.
293 (probably_never_executed_edge_p): Likewise.
294 (combine_predictions_for_bb): Check that profile is nonzero.
295 (propagate_freq): Do not set frequency.
296 (drop_profile): Simplify.
297 (counts_to_freqs): Simplify.
298 (expensive_function_p): Use to_frequency.
299 (propagate_unlikely_bbs_forward): Simplify.
300 (determine_unlikely_bbs): Simplify.
301 (estimate_bb_frequencies): Add hack to silence graphite issues.
302 (compute_function_frequency): Use ipa counts.
303 (pass_profile::execute): Update.
304 (rebuild_frequencies): Use counts only.
305 (force_edge_cold): Use counts only.
306 * profile-count.c (profile_count::dump): Dump new count types.
307 (profile_count::differs_from_p): Check compatiblity.
308 (profile_count::to_frequency): New function.
309 (profile_count::to_cgraph_frequency): New function.
310 * profile-count.h (struct function): Declare.
311 (enum profile_quality): Add profile_guessed_local and
312 profile_guessed_global0.
313 (class profile_proability): Decrease number of bits to 29;
314 update from_reg_br_prob_note and to_reg_br_prob_note.
315 (class profile_count: Update comment; decrease number of bits
316 to 61. Check compatibility.
317 (profile_count::compatible_p): New private member function.
318 (profile_count::ipa_p): New member function.
319 (profile_count::operator<): Handle global zero correctly.
320 (profile_count::operator>): Handle global zero correctly.
321 (profile_count::operator<=): Handle global zero correctly.
322 (profile_count::operator>=): Handle global zero correctly.
323 (profile_count::nonzero_p): New member function.
324 (profile_count::force_nonzero): New member function.
325 (profile_count::max): New member function.
326 (profile_count::apply_scale): Handle IPA scalling.
327 (profile_count::guessed_local): New member function.
328 (profile_count::global0): New member function.
329 (profile_count::ipa): New member function.
330 (profile_count::to_frequency): Declare.
331 (profile_count::to_cgraph_frequency): Declare.
332 * profile.c (OVERLAP_BASE): Delete.
333 (compute_frequency_overlap): Delete.
334 (compute_branch_probabilities): Do not use compute_frequency_overlap.
335 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
336 * sched-ebb.c (rank): Use counts only.
337 * shrink-wrap.c (handle_simple_exit): Use counts only.
338 (try_shrink_wrapping): Use counts only.
339 (place_prologue_for_one_component): Use counts only.
340 * tracer.c (find_best_predecessor): Use to_frequency.
341 (find_trace): Use to_frequency.
342 (tail_duplicate): Use to_frequency.
343 * trans-mem.c (expand_transaction): Do not update frequency.
344 * tree-call-cdce.c: Do not update frequency.
345 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
346 (gimple_merge_blocks): Likewise.
347 (gimple_split_edge): Likewise.
348 (gimple_duplicate_sese_region): Likewise.
349 (gimple_duplicate_sese_tail): Likewise.
350 (move_sese_region_to_fn): Likewise.
351 (gimple_account_profile_record): Likewise.
352 (insert_cond_bb): Likewise.
353 * tree-complex.c (expand_complex_div_wide): Likewise.
354 * tree-eh.c (lower_resx): Update profile.
355 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
357 (initialize_cfun): Do not initialize frequencies
358 (freqs_to_counts): Delete.
359 (copy_cfg_body): Ignore count parameter.
361 (expand_call_inline): Update count_max.
362 (optimize_inline_calls): Update count_max.
363 (tree_function_versioning): Update count_max.
364 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
365 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
367 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
368 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
369 (try_peel_loop): Likewise.
370 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
372 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
373 (tree_transform_and_unroll_loop): Do not use frequencies
374 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
375 Use reliable prediction only.
376 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
377 * tree-ssa-sink.c (select_best_block): Use to_frequency.
378 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
380 * tree-ssa-threadupdate.c (create_block_for_threading): Do
382 (any_remaining_duplicated_blocks): Likewise.
383 (update_profile): Likewise.
384 (estimated_freqs_path): Delete.
385 (freqs_to_counts_path): Delete.
386 (clear_counts_path): Delete.
387 (ssa_fix_duplicate_block_edges): Likewise.
388 (duplicate_thread_path): Likewise.
389 * tree-switch-conversion.c (gen_inbound_check): Use counts.
390 * tree-tailcall.c (decrease_profile): Do not update frequency.
391 (eliminate_tail_call): Likewise.
392 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
393 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
394 (optimize_mask_stores): Likewise.
395 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
396 * ubsan.c (ubsan_expand_null_ifn): Update profile.
397 (ubsan_expand_ptr_ifn): Update profile.
398 * value-prof.c (gimple_ic): Simplify.
399 * value-prof.h (gimple_ic): Update prototype.
400 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
401 * ipa-inline.c (compute_uninlined_call_time): Be sure that
403 (want_inline_self_recursive_call_p): Likewise.
404 (resolve_noninline_speculation): Only cummulate defined counts.
405 (inline_small_functions): Use nonzero_p.
406 (ipa_inline): Do not access freed node.
408 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
410 * config/aarch64/aarch64.c (aarch64_override_options_internal):
411 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
413 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
415 * config/riscv/riscv.c (riscv_legitimize_move): Handle
416 non-legitimate address.
418 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
420 * config/rs6000/rs6000.md (*lt0_disi): Delete.
421 (*lt0_<mode>di, *lt0_<mode>si): New.
423 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
425 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
428 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
429 Jim Wilson <jim.wilson@linaro.org>
431 * config/aarch64/aarch64-cores.def (saphira): New CPU.
432 * config/aarch64/aarch64-tune.md: Regenerated.
433 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
434 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
436 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
438 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
439 (arc_expand_prologue): Restore blink for millicode.
440 * config/arc/linux.h (LINK_EH_SPEC): Defined.
442 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
445 * config/i386/i386.c (ix86_vector_duplicate_value): Use
446 gen_vec_duplicate after forcing the scalar into a register.
448 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
450 * combine (try_combine): Print the insns input to try_combine to the
453 2017-11-02 Steve Ellcey <sellcey@cavium.com>
456 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
457 Remove second argument from aarch64_process_target_attr call.
458 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
460 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
462 (aarch64_handle_attr_arch): Remove second argument.
463 (aarch64_handle_attr_cpu): Ditto.
464 (aarch64_handle_attr_tune): Ditto.
465 (aarch64_handle_attr_isa_flags): Ditto.
466 (aarch64_process_one_target_attr): Ditto.
467 (aarch64_process_target_attr): Ditto.
468 (aarch64_option_valid_attribute_p): Remove second argument.
469 on aarch64_process_target_attr call.
471 2017-11-02 David Malcolm <dmalcolm@redhat.com>
473 * diagnostic.c: Include "selftest-diagnostic.h".
474 (selftest::assert_location_text): New function.
475 (selftest::test_diagnostic_get_location_text): New function.
476 (selftest::diagnostic_c_tests): Call it.
478 2017-11-02 David Malcolm <dmalcolm@redhat.com>
480 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
481 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
482 (class selftest::test_diagnostic_context): Move to...
483 * selftest-diagnostic.c: New file.
484 * selftest-diagnostic.h: New file.
486 2017-11-02 James Bowman <james.bowman@ftdichip.com>
488 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
489 offset range for FT32B.
490 * config/ft32/ft32.h: option "mcompress" enables relaxation.
491 * config/ft32/ft32.md: Add TARGET_NOPM.
492 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
493 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
495 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
497 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
499 2017-11-02 Jeff Law <law@redhat.com>
501 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
502 virtual keyword on FINAL OVERRIDE members.
504 * tree-ssa-propagate.h (ssa_propagation_engine): Group
505 virtuals together. Add virtual destructor.
506 (substitute_and_fold_engine): Similarly.
508 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
510 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
512 2017-11-02 Richard Biener <rguenther@suse.de>
514 PR tree-optimization/82795
515 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
517 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
519 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
521 (gcc_GAS_CHECK_FEATURE): Remove.
522 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
523 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
524 * configure: Regenerate.
526 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
528 * config/arc/arc.c (hwloop_optimize): Account for empty
531 2017-11-02 Richard Biener <rguenther@suse.de>
534 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
535 Truncate ARRAY_REF index and element size.
537 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
539 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
541 2017-11-01 Jeff Law <law@redhat.com>
543 * tree-ssa-ccp.c (ccp_folder): New class derived from
544 substitute_and_fold_engine.
545 (ccp_folder::get_value): New member function.
546 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
547 (ccp_fold_stmt): Remove prototype.
548 (ccp_finalize): Call substitute_and_fold from the ccp_class.
549 * tree-ssa-copy.c (copy_folder): New class derived from
550 substitute_and_fold_engine.
551 (copy_folder::get_value): Renamed from get_value.
552 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
553 * tree-vrp.c (vrp_folder): New class derived from
554 substitute_and_fold_engine.
555 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
556 (vrp_folder::get_value): New member function.
557 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
558 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
559 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
560 provide a class interface to folder/substitute routines.
561 (ssa_prop_fold_stmt_fn): Remove typedef.
562 (ssa_prop_get_value_fn): Likewise.
563 (subsitute_and_fold): Remove prototype.
564 (replace_uses_in): Likewise.
565 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
566 Renamed from replace_uses_in. Call the virtual member function
567 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
568 (substitute_and_fold_dom_walker): Remove initialization of
569 data member entries for calbacks. Add substitute_and_fold_engine
570 member and initialize it.
571 (substitute_and_fold_dom_walker::before_dom_children0: Use the
572 member functions for get_value, replace_phi_args_in c
573 replace_uses_in, and fold_stmt calls.
574 (substitute_and_fold_engine::substitute_and_fold): Renamed from
575 substitute_and_fold. Remove assert. Update ctor call.
577 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
578 (ssa_prop_visit_phi_fn): Likewise.
579 (class ssa_propagation_engine): New class to provide an interface
581 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
583 (ssa_prop_visit_phi): Likewise.
584 (ssa_propagation_engine::simulate_stmt): Moved into class.
585 Call visit_phi/visit_stmt from the class rather than via
586 file scoped static variables.
587 (ssa_propagation_engine::simulate_block): Moved into class.
588 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
589 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
590 set file scoped statics for the visit_stmt/visit_phi callbacks.
591 * tree-complex.c (complex_propagate): New class derived from
592 ssa_propagation_engine.
593 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
594 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
595 (tree_lower_complex): Call ssa_propagate via the complex_propagate
597 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
598 ssa_propagation_engine.
599 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
600 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
601 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
602 * tree-ssa-copy.c (copy_prop): New class derived from
603 ssa_propagation_engine.
604 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
605 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
606 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
607 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
608 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
609 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
610 (execute_vrp): Call ssa_propagate from the vrp_prop class.
612 2017-11-01 Jakub Jelinek <jakub@redhat.com>
614 PR rtl-optimization/82778
615 PR rtl-optimization/82597
616 * compare-elim.c (struct comparison): Add in_a_setter field.
617 (find_comparison_dom_walker::before_dom_children): Remove killed
618 bitmap and df_simulate_find_defs call, instead walk the defs.
619 Compute last_setter and initialize in_a_setter. Merge definitions
620 with first initialization for a few variables.
621 (try_validate_parallel): Use insn_invalid_p instead of
622 recog_memoized. Return insn rather than just the pattern.
623 (try_merge_compare): Fix up comment. Don't uselessly test if
624 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
626 (execute_compare_elim_after_reload): Remove df_chain_add_problem
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/aarch64.c (aarch64_rtx_costs): Use
634 aarch64_hard_regno_nregs to get the number of registers
637 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
638 Alan Hayward <alan.hayward@arm.com>
639 David Sherwood <david.sherwood@arm.com>
641 * config/aarch64/constraints.md (Upl): Rename to...
643 * config/aarch64/aarch64.md
644 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
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_add_constant_internal)
652 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
655 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
656 Alan Hayward <alan.hayward@arm.com>
657 David Sherwood <david.sherwood@arm.com>
659 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
660 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
661 (aarch64_evpc_dup): Generate rtl direcly, rather than using
663 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
665 * config/aarch64/iterators.md: Add a comment above the permute
666 unspecs to say that they are generated directly by
667 aarch64_expand_vec_perm_const.
668 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
670 2017-11-01 Nathan Sidwell <nathan@acm.org>
672 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
674 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
676 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
677 and medany code models, and describe what they do.
679 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
681 Revert accidental duplicate:
683 * combine.c (can_change_dest_mode): Reject changes in
684 REGMODE_NATURAL_SIZE.
686 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
688 PR rtl-optimization/64682
689 PR rtl-optimization/69567
690 PR rtl-optimization/69737
691 PR rtl-optimization/82683
692 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
693 register mentioned in the note, drop the note, unless it came from I3,
694 in which case it should go to I3 again.
696 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
698 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
699 and return false if not.
700 (clear_bytes_written_by, live_bytes_read): Update accordingly.
702 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
704 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
705 range is known to be empty.
707 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
708 Alan Hayward <alan.hayward@arm.com>
709 David Sherwood <david.sherwood@arm.com>
711 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
712 and CONST_VECTOR_NUNITS instead of computing the number of units from
713 the byte sizes of the vector and element.
714 (simplify_binary_operation_1): Likewise.
715 (simplify_const_binary_operation): Likewise.
716 (simplify_ternary_operation): Likewise.
718 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
719 Alan Hayward <alan.hayward@arm.com>
720 David Sherwood <david.sherwood@arm.com>
722 * var-tracking.c (INT_MEM_OFFSET): Replace with...
723 (int_mem_offset): ...this new function.
724 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
725 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
726 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
729 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
730 Alan Hayward <alan.hayward@arm.com>
731 David Sherwood <david.sherwood@arm.com>
733 * lower-subreg.c (interesting_mode_p): New function.
734 (compute_costs, find_decomposable_subregs, decompose_register)
735 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
736 (resolve_clobber, dump_choices): Use it.
738 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
739 Alan Hayward <alan.hayward@arm.com>
740 David Sherwood <david.sherwood@arm.com>
742 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
744 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
745 Alan Hayward <alan.hayward@arm.com>
746 David Sherwood <david.sherwood@arm.com>
748 * alias.c (find_base_value, find_base_term): Only process integer
749 truncations. Check the precision rather than the size.
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 * machmode.h (is_narrower_int_mode): New function
756 * optabs.c (expand_float, expand_fix): Use it.
757 * dwarf2out.c (rotate_loc_descriptor): Likewise.
759 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
760 Alan Hayward <alan.hayward@arm.com>
761 David Sherwood <david.sherwood@arm.com>
763 * rtl.h (narrower_subreg_mode): New function.
764 * ira-color.c (update_costs_from_allocno): Use it.
766 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
767 Alan Hayward <alan.hayward@arm.com>
768 David Sherwood <david.sherwood@arm.com>
770 * optabs-query.h (convert_optab_p): New function, split out from...
771 (convert_optab_handler): ...here.
772 (widening_optab_handler): Delete.
773 (find_widening_optab_handler): Remove permit_non_widening parameter.
774 (find_widening_optab_handler_and_mode): Likewise. Provide an
775 override that operates on mode class wrappers.
776 * optabs-query.c (widening_optab_handler): Delete.
777 (find_widening_optab_handler_and_mode): Remove permit_non_widening
778 parameter. Assert that the two modes are the same class and that
779 the "from" mode is narrower than the "to" mode. Use
780 convert_optab_handler instead of widening_optab_handler.
781 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
782 instead of widening_optab_handler.
783 * expr.c (expand_expr_real_2): Update calls to
784 find_widening_optab_handler.
785 * optabs.c (expand_widen_pattern_expr): Likewise.
786 (expand_binop_directly): Take the insn_code as a parameter.
787 (expand_binop): Only call find_widening_optab_handler for
788 conversion optabs; use optab_handler otherwise. Update calls
789 to find_widening_optab_handler and expand_binop_directly.
790 Use convert_optab_handler instead of widening_optab_handler.
791 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
792 find_widening_optab_handler and use scalar_mode rather than
794 (convert_plusminus_to_widen): Likewise.
796 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
797 Alan Hayward <alan.hayward@arm.com>
798 David Sherwood <david.sherwood@arm.com>
800 * machmode.h (fixed_size_mode): New class.
801 * rtl.h (get_pool_mode): Return fixed_size_mode.
802 * gengtype.c (main): Add fixed_size_mode.
803 * target.def (get_raw_result_mode): Return a fixed_size_mode.
804 (get_raw_arg_mode): Likewise.
805 * doc/tm.texi: Regenerate.
806 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
807 * targhooks.c (default_get_reg_raw_mode): Likewise.
808 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
809 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
810 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
811 (msp430_get_raw_result_mode): Likewise.
812 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
813 * dbxout.c (dbxout_parms): Require fixed-size modes.
814 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
815 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
816 * omp-low.c (lower_oacc_reductions): Likewise.
817 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
818 (simplify_subreg): Update accordingly.
819 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
820 (force_const_mem): Update accordingly. Return NULL_RTX for modes
821 that aren't fixed-size.
822 (get_pool_mode): Return a fixed_size_mode.
823 (output_constant_pool_2): Take a fixed_size_mode.
825 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
826 Alan Hayward <alan.hayward@arm.com>
827 David Sherwood <david.sherwood@arm.com>
829 * doc/rtl.texi (vec_series): Document.
830 (const): Say that the operand can be a vec_series.
831 * rtl.def (VEC_SERIES): New rtx code.
832 * rtl.h (const_vec_series_p_1): Declare.
833 (const_vec_series_p): New function.
834 * emit-rtl.h (gen_const_vec_series): Declare.
835 (gen_vec_series): Likewise.
836 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
837 (gen_vec_series): Likewise.
838 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
839 * simplify-rtx.c (simplify_unary_operation): Handle negations
841 (simplify_binary_operation_series): New function.
842 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
843 (test_vector_ops_series): New function.
844 (test_vector_ops): Call it.
845 * config/powerpcspe/altivec.md (altivec_lvsl): Use
846 gen_const_vec_series.
847 (altivec_lvsr): Likewise.
848 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
850 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
851 Alan Hayward <alan.hayward@arm.com>
852 David Sherwood <david.sherwood@arm.com>
854 * doc/rtl.texi (const): Update description of address constants.
855 Say that vector constants are allowed too.
856 * common.md (E, F): Use CONSTANT_P instead of checking for
858 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
859 checking for CONST_VECTOR.
860 * expmed.c (make_tree): Use build_vector_from_val for a CONST
862 * expr.c (expand_expr_real_2): Check for vector modes instead
863 of checking for CONST_VECTOR.
864 * rtl.h (const_vec_p): New function.
865 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
866 (unwrap_const_vec_duplicate): Handle them here too.
868 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
869 David Malcolm <dmalcolm@redhat.com>
870 Alan Hayward <alan.hayward@arm.com>
871 David Sherwood <david.sherwood@arm.com>
873 * rtl.h (vec_duplicate_p): New function.
874 * selftest-rtl.c (assert_rtx_eq_at): New function.
875 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
876 (assert_rtx_eq_at): Declare.
877 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
878 * selftest-run-tests.c (selftest::run_tests): Call it.
879 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
880 (simplify_unary_operation_1): Recursively handle vector duplicates.
881 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
883 (simplify_subreg): Handle subregs of vector duplicates.
884 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
885 (selftest::simplify_rtx_c_tests): New functions.
887 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
888 Alan Hayward <alan.hayward@arm.com>
889 David Sherwood <david.sherwood@arm.com>
891 * emit-rtl.h (gen_const_vec_duplicate): Declare.
892 (gen_vec_duplicate): Likewise.
893 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
895 (gen_const_vector): ...here.
896 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
897 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
898 whose elements are all equal.
899 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
900 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
901 (simplify_relational_operation): Likewise.
902 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
904 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
905 (aarch64_expand_vector_init): Likewise.
906 * config/arm/arm.c (neon_vdup_constant): Likewise.
907 (neon_expand_vector_init): Likewise.
908 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
909 (arm_block_set_unaligned_vect): Likewise.
910 (arm_block_set_aligned_vect): Likewise.
911 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
912 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
913 (expand_vec_perm_even_odd_pack): Likewise.
914 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
915 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
916 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
917 gen_const_vec_duplicate.
918 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
919 * config/mips/mips.c (mips_gen_const_int_vector): Use
920 gen_const_vec_duplicate.
921 (mips_expand_vector_init): Use CONST0_RTX.
922 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
923 (define_split): Use gen_const_vec_duplicate.
924 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
925 (define_split): Use gen_const_vec_duplicate.
926 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
927 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
928 * config/spu/spu.c (spu_const): Likewise.
930 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
931 Alan Hayward <alan.hayward@arm.com>
932 David Sherwood <david.sherwood@arm.com>
934 * combine.c (can_change_dest_mode): Reject changes in
935 REGMODE_NATURAL_SIZE.
937 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
939 * configure.ac (--enable-libssp): New.
940 (gcc_cv_libc_provides_ssp): Check for explicit setting before
941 trying to determine target-specific default. Adjust indentation.
942 * configure: Regenerated.
943 * doc/install.texi (Configuration): Expand --disable-libssp
946 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
948 config/i386/i386.c (ix86_expand_epilogue): Correct stack
951 2017-10-31 Martin Jambor <mjambor@suse.cz>
954 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
956 2017-10-31 David Malcolm <dmalcolm@redhat.com>
958 * auto-profile.c (autofdo_source_profile::read): Use
959 UNKNOWN_LOCATION rather than 0.
960 * diagnostic-core.h (warning_at_rich_loc): Rename to...
961 (warning_at): ...this overload.
962 (warning_at_rich_loc_n): Rename to...
963 (warning_n): ...this overload.
964 (error_at_rich_loc): Rename to...
965 (error_at): ...this overload.
966 (pedwarn_at_rich_loc): Rename to...
967 (pedwarn): ...this overload.
968 (permerror_at_rich_loc): Rename to...
969 (permerror): ...this overload.
970 (inform_at_rich_loc): Rename to...
971 (inform): ...this overload.
972 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
973 (diagnostic_n_impl_richloc): Rename to...
974 (diagnostic_n_impl): ...this rich_location *-based decl.
975 (inform_at_rich_loc): Rename to...
976 (inform): ...this, and add an assertion.
977 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
978 (warning_at_rich_loc): Rename to...
979 (warning_at): ...this, and add an assertion.
980 (warning_at_rich_loc_n): Rename to...
981 (warning_n): ...this, and add an assertion.
982 (warning_n): Update location_t-based implementation for removal of
983 location_t-based diagnostic_n_impl.
984 (pedwarn_at_rich_loc): Rename to...
985 (pedwarn): ...this, and add an assertion.
986 (permerror_at_rich_loc): Rename to...
987 (permerror): ...this, and add an assertion.
988 (error_n): Update for removal of location_t-based diagnostic_n_impl.
989 (error_at_rich_loc): Rename to...
990 (error_at): ...this, and add an assertion.
991 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
992 (driver::do_spec_on_infiles): Likewise.
993 * substring-locations.c (format_warning_va): Update for renaming
994 of inform_at_rich_loc.
996 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
998 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
999 _Float<N>X built-in functions so that the variant without the
1000 "__builtin_" prefix is only enabled for the GNU C and Objective C
1001 languages when they are in non-strict ANSI/ISO mode.
1002 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1003 * target.def (floatn_builtin_p): Add a target hook to control
1004 whether _Float<N> and _Float<N>X built-in functions without the
1005 "__builtin_" prefix are enabled, and return true for C and
1006 Objective C in the default hook. Include langhooks.h in
1008 * targhooks.h (default_floatn_builtin_p): Likewise.
1009 * targhooks.c (default_floatn_builtin_p): Likewise.
1010 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1011 floatn_builtin_p target hook.
1012 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1014 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1015 Eric Botcazou <ebotcazou@adacore.com>
1017 PR rtl-optimization/81803
1018 * lra-constraints.c (curr_insn_transform): Also reload the whole
1019 register for a strict subreg no wider than a word if this is for
1020 a WORD_REGISTER_OPERATIONS target.
1022 2017-10-31 Jason Merrill <jason@redhat.com>
1024 * gdbinit.in: Skip over inlines from timevar.h.
1026 2017-10-31 Martin Liska <mliska@suse.cz>
1028 * doc/gcov.texi: Document new option.
1029 * gcov.c (print_usage): Likewise print it.
1030 (process_args): Support the argument.
1031 (format_count): New function.
1032 (format_gcov): Use the function.
1034 2017-10-31 Martin Liska <mliska@suse.cz>
1036 * gcov.c (struct name_map): do not use typedef.
1037 Define operator== and operator<.
1038 (name_search): Remove.
1039 (name_sort): Remove.
1040 (main): Do not allocate names.
1041 (process_file): Add vertical space.
1042 (generate_results): Use std::find.
1043 (release_structures): Do not release memory.
1044 (find_source): Use std::find.
1046 2017-10-31 Martin Liska <mliska@suse.cz>
1048 * gcov.c (struct line_info): Remove it's typedef.
1049 (line_info::line_info): Add proper ctor.
1050 (line_info::has_block): Do not use a typedef.
1051 (struct source_info): Do not use typedef.
1052 (circuit): Likewise.
1053 (get_cycles_count): Likewise.
1054 (output_intermediate_file): Iterate via vector iterator.
1055 (add_line_counts): Use std::vector methods.
1056 (accumulate_line_counts): Likewise.
1057 (output_lines): Likewise.
1059 2017-10-31 Martin Liska <mliska@suse.cz>
1061 * gcov.c (struct source_info): Remove typedef.
1062 (source_info::source_info): Add proper ctor.
1063 (accumulate_line_counts): Use struct, not it's typedef.
1064 (output_gcov_file): Likewise.
1065 (output_lines): Likewise.
1066 (main): Do not allocate an array.
1067 (output_intermediate_file): Use size of vector container.
1068 (process_file): Resize the vector.
1069 (generate_results): Do not preallocate, use newly added vector
1071 (release_structures): Do not release sources.
1072 (find_source): Use vector methods.
1073 (add_line_counts): Do not use typedef.
1075 2017-10-31 Martin Liska <mliska@suse.cz>
1077 * doc/gcov.texi: Document that.
1078 * gcov.c (add_line_counts): Mark lines with a non-executed
1080 (output_line_beginning): Handle such lines.
1081 (output_lines): Pass new argument.
1082 (output_intermediate_file): Print it in intermediate format.
1084 2017-10-31 Martin Liska <mliska@suse.cz>
1086 * color-macros.h: New file.
1087 * diagnostic-color.c: Factor out color related to macros to
1089 * doc/gcov.texi: Document -k option.
1090 * gcov.c (INCLUDE_STRING): Include string.h.
1091 (print_usage): Add -k option.
1092 (process_args): Parse it.
1093 (pad_count_string): New function.
1094 (output_line_beginning): Likewise.
1095 (DEFAULT_LINE_START): New macro.
1096 (output_lines): Support color output.
1098 2017-10-31 Martin Liska <mliska@suse.cz>
1100 PR gcov-profile/82633
1101 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1102 their interaction with GCOV infrastructure.
1103 * configure.ac: Add -fkeep-{inline,static}-functions to
1105 * configure: Regenerate.
1107 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1110 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1112 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1115 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1116 into a register if it does not fit into an immediate offset field.
1118 2017-10-31 Olivier Hainque <hainque@adacore.com>
1120 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1122 2017-10-31 Julia Koval <julia.koval@intel.com>
1124 * config.gcc: Add gfniintrin.h.
1125 * config/i386/gfniintrin.h: New.
1126 * config/i386/i386-builtin-types.def
1127 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1128 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1129 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1130 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1131 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1132 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1133 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1134 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1135 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1136 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1137 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1138 * config/i386/immintrin.h: Include gfniintrin.h.
1139 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1141 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1143 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1145 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1147 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1148 (arm_ashldi3_1bit): Remove pattern.
1149 (ashrdi3): Remove shift by 1 expansion.
1150 (arm_ashrdi3_1bit): Remove pattern.
1151 (lshrdi3): Remove shift by 1 expansion.
1152 (arm_lshrdi3_1bit): Remove pattern.
1153 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1154 cost of ashldi3 by 1.
1155 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1156 (<shift>di3_neon): Likewise.
1158 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1160 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1161 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1162 and (*aarch64_simd_mov<VQ:mode>).
1163 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1164 pattern alternative.
1165 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1166 attributes to match pattern alternative.
1168 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1170 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1171 * config/rs6000/emmintrin.h: New file.
1172 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1174 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1176 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1177 * (movdi_vfp_cortexa8): Remove pattern.
1179 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1181 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1183 (Specific, alpha*-dec-osf5.1): Remove.
1184 (Specific, mips-sgi-irix5): Remove.
1185 (Specific, mips-sgi-irix6): Remove.
1187 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1190 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1191 arguments to clear_bit_region_be.
1193 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1195 * gimplify.c: Include memmodel.h.
1197 2017-10-30 Martin Jambor <mjambor@suse.cz>
1199 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1200 condition whether loop should be executed at all.
1202 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1204 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1205 gimple folding of vec_madd() intrinsics.
1206 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1207 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1208 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1210 2017-10-30 Richard Biener <rguenther@suse.de>
1212 PR tree-optimization/82762
1214 2017-10-23 Richard Biener <rguenther@suse.de>
1216 PR tree-optimization/82129
1218 2017-08-01 Richard Biener <rguenther@suse.de>
1220 PR tree-optimization/81181
1221 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1222 (compute_antic): ... end of iteration here.
1224 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1226 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1227 -std=iso9899:2017 and -std=gnu17.
1228 * doc/standards.texi (C Language): Document C17 support.
1229 * doc/cpp.texi (Overview): Mention -std=c17.
1230 (Standard Predefined Macros): Document C11 and C17 values of
1231 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1232 * doc/extend.texi (Inline): Do not list individual options for
1233 standards newer than C99.
1234 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1236 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1239 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1241 * asan.c (asan_finish_file): Align asan globals array by shadow
1244 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1247 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1248 (struct store_immediate_info): Add bitregion_start and bitregion_end
1250 (store_immediate_info::store_immediate_info): Add brs and bre
1251 arguments and initialize bitregion_{start,end} from those.
1252 (struct merged_store_group): Add bitregion_start, bitregion_end,
1253 align_base and mask fields. Drop unnecessary struct keyword from
1254 struct store_immediate_info. Add do_merge method.
1255 (clear_bit_region_be): Use memset instead of loop storing zeros.
1256 (merged_store_group::do_merge): New method.
1257 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
1258 stores as long as the surrounding bitregions have no gaps.
1259 (merged_store_group::merge_overlapping): Use do_merge.
1260 (merged_store_group::apply_stores): Test that bitregion_{start,end}
1261 is byte aligned, rather than requiring that start and width are
1262 byte aligned. Drop unnecessary struct keyword from
1263 struct store_immediate_info. Allocate and populate also mask array.
1264 Make start of the arrays relative to bitregion_start rather than
1265 start and size them according to bitregion_{end,start} difference.
1266 (struct imm_store_chain_info): Drop unnecessary struct keyword from
1267 struct store_immediate_info.
1268 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
1269 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
1270 struct keyword from struct store_immediate_info.
1271 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
1272 between stores as long as the surrounding bitregions have no gaps.
1274 (struct split_store): Add orig non-static data member.
1275 (split_store::split_store): Initialize orig to false.
1276 (find_constituent_stmts): Return store_immediate_info *, non-NULL
1277 if there is exactly a single original stmt. Change stmts argument
1278 to pointer from reference, if NULL, don't push anything to it. Add
1279 first argument, use it to optimize skipping over orig stmts that
1280 are known to be before bitpos already. Simplify.
1281 (split_group): Return unsigned int count how many stores are or
1282 would be needed rather than a bool. Add allow_unaligned argument.
1283 Change split_stores argument from reference to pointer, if NULL,
1284 only do a dry run computing how many stores would be produced.
1285 Rewritten algorithm to use both alignment and misalign if
1286 !allow_unaligned and handle bitfield stores with gaps.
1287 (imm_store_chain_info::output_merged_store): Set start_byte_pos
1288 from bitregion_start instead of start. Compute allow_unaligned
1289 here, if true, do 2 split_group dry runs to compute which one
1290 produces fewer stores and prefer aligned if equal. Punt if
1291 new count is bigger or equal than original before emitting any
1292 statements, rather than during that. Remove no longer needed
1293 new_ssa_names tracking. Replace num_stmts with
1294 split_stores.length (). Use 32-bit stack allocated entries
1295 in split_stores auto_vec. Try to reuse original store lhs/rhs1
1296 if possible. Handle bitfields with gaps.
1297 (pass_store_merging::execute): Ignore bitsize == 0 stores.
1298 Compute bitregion_{start,end} for the stores and construct
1299 store_immediate_info with that. Formatting fixes.
1301 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
1304 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
1305 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
1307 2017-10-29 Jim Wilson <wilson@tuliptree.org>
1309 * gimplify.c: Include tm_p.h.
1311 * common.opt (gcoff): Re-add as ignored option.
1312 (gcoff1, gcoff2, gcoff3): Likewise.
1314 * Makefile.in (OBJS): Delete sdbout.o.
1315 (GTFILES): Delete $(srcdir)/sdbout.c.
1316 * debug.h: Delete sdb_debug_hooks.
1317 * final.c: Delete sdbout.h include.
1318 (final_scan_insn): Delete SDB_DEBUG check.
1319 (rest_of_clean_state): Likewise.
1320 * output.h: Delete sdb_begin_function_line.
1323 * toplev.c: Delete sdbout.h include.
1324 (process_options): Delete SDB_DEBUG check.
1325 * tree-core.h (tree_type_common): Delete pointer field of
1327 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
1328 TYPE_SYMTAB_POINTER.
1329 * tree.h (TYPE_SYMTAB_POINTER): Delete.
1330 (TYPE_SYMTAB_IS_POINTER): Delete.
1331 (TYPE_SYMTAB_IS_DIE): Renumber.
1332 * xcoffout.c: Refer to former sdbout.c file.
1333 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
1335 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
1336 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
1337 (Debugging Options): Delete -gcoff.
1338 (-gstabs): Delete SDB reference.
1340 (-gcoff@var{level}): Delete.
1341 * doc/passes.texi (Debugging information output): Delete SDB and
1342 sdbout.c references.
1343 * doc/tm.texi: Regenerate.
1344 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
1345 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
1347 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
1348 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
1349 (SDB_DEBUGGING_INFO): Delete.
1350 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
1351 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
1352 * target.def (output_source_filename): Delete COFF reference.
1354 * common.opt (gcoff): Delete.
1355 (gxcoff+): Update Negative chain.
1356 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
1358 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
1359 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
1360 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
1362 (expand_function_start): Change sdb reference to past tense.
1363 (expand_function_end): Change sdb reference to past tense.
1364 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
1365 * opts.c (debug_type_names): Delete coff entry.
1366 (common_handle_option): Delete OPT_gcoff case.
1367 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
1369 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
1370 * config/cris/cris.h: Delete SDB reference in comment.
1371 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
1372 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
1373 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
1374 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
1376 (ix86_expand_prologue): Likewise.
1377 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
1378 * config/ia64/ia64.h: Likewise.
1379 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
1380 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
1382 * config/mmix/mmix.h: Likewise.
1383 * config/nds32/nds32.c: Likewise.
1384 * config/stormy/storym16.h: Likewise.
1385 * config/visium/visium.h: Likewise.
1386 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
1388 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
1390 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
1391 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
1392 FRAME_POINTER_REGNUM point at high end of local var area.
1394 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
1396 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
1397 Move comment around. Do not reset best_edge for a copiable
1398 destination if the copy would cause a partition change.
1399 (better_edge_p): Remove redundant check.
1401 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1403 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
1405 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1407 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
1408 for math functions that have _Float<N> and _Float<N>X variants.
1409 (mathfn_built_in_2): Add support for math functions that have
1410 _Float<N> and _Float<N>X variants.
1411 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
1412 (expand_builtin_mathfn_ternary): Add support for fma with
1413 _Float<N> and _Float<N>X variants.
1414 (expand_builtin): Likewise.
1415 (fold_builtin_3): Likewise.
1416 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
1417 create math function _Float<N> and _Float<N>X variants as external
1419 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
1420 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
1421 the __builtin_ prefix and if not strict ansi, without the prefix.
1422 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
1423 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
1424 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
1425 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
1426 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
1427 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
1428 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
1429 function signatures for fma _Float<N> and _Float<N>X variants.
1430 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
1431 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
1432 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
1433 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
1434 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
1435 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
1436 * gencfn-macros.c (print_case_cfn): Add support for math functions
1437 that have _Float<N> and _Float<N>X variants.
1438 (print_define_operator_list): Likewise.
1439 (fltfn_suffixes): Likewise.
1441 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
1442 for math functions that have _Float<N> and _Float<N>X variants.
1443 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
1444 and _Float<N>X variants.
1445 (COPYSIGN): Likewise.
1448 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
1449 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
1451 (integer_valued_read_call_p): Likewise.
1452 * fold-const-call.c (fold_const_call_ss): Likewise.
1453 (fold_const_call_sss): Add support for copysign, fmin, and fmax
1454 _Float<N> and _Float<N>X variants.
1455 (fold_const_call_ssss): Add support for fma _Float<N> and
1456 _Float<N>X variants.
1457 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
1458 support for copysign and fma _Float<N> and _Float<N>X variants.
1459 (backprop::process_builtin_call_use): Likewise.
1460 * tree-call-cdce.c (can_test_argument_range); Add support for
1461 sqrt _Float<N> and _Float<N>X variants.
1462 (edom_only_function): Likewise.
1463 (get_no_error_domain): Likewise.
1464 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
1465 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
1466 copysign _Float<N> and _Float<N>X variants.
1467 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
1468 handled by machine independent code.
1469 (FMAF128): Likewise.
1470 * doc/cpp.texi (Common Predefined Macros): Document defining
1471 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
1472 fma _Float<N> and _Float<N>X variants.
1474 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1477 * config/i386/i386-modes.def (CCFPU): Remove definition.
1478 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
1479 (ix86_cc_modes_compatible): Ditto.
1480 (ix86_expand_carry_flag_compare): Ditto.
1481 (ix86_expand_int_movcc): Ditto.
1482 (ix86_expand_int_addcc): Ditto.
1483 (ix86_reverse_condition): Ditto.
1484 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
1485 Return true/false for unordered/ordered fp comparisons.
1486 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
1487 (ix86_prepare_fp_compare_args): Update for rename.
1488 (ix86_expand_fp_compare): Update for rename. Generate unordered
1489 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
1490 (ix86_expand_sse_compare_and_jump): Ditto.
1491 * config/i386/predicates.md (fcmov_comparison_operator):
1492 Remove CCFPU mode handling.
1493 (ix86_comparison_operator): Ditto.
1494 (ix86_carry_flag_operator): Ditto.
1495 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
1496 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
1497 (*cmpu<mode>_cc_i387): Ditto.
1498 (FPCMP): Remove mode iterator.
1499 (unord): Remove mode attribute.
1500 (unord_subst): New define_subst transformation
1501 (unord): New define_subst attribute.
1503 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
1504 (*cmpi<unord>xf_i387): Ditto.
1505 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
1506 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
1507 using unord_subst transformation.
1508 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
1509 (round_saeonly): Also handle CCFP mode.
1510 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
1511 Remove UNSPEC_SAHF unspec handling.
1513 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1515 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
1517 2017-10-27 Jeff Law <law@redhat.com>
1519 * gimple-ssa-sprintf.c: Include domwalk.h.
1520 (class sprintf_dom_walker): New class, derived from dom_walker.
1521 (sprintf_dom_walker::before_dom_children): New function.
1522 (struct call_info): Moved into sprintf_dom_walker class
1523 (compute_formath_length, handle_gimple_call): Likewise.
1524 (sprintf_length::execute): Call the dominator walker rather
1525 than walking the statements.
1527 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
1528 gimple statement locations.
1529 (check_array_bounds): Corresponding changes. Get the statement's
1530 location directly from wi->stmt.
1532 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
1535 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
1537 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1539 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
1540 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
1542 2017-10-27 Jakub Jelinek <jakub@redhat.com>
1545 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
1546 * config/i386/i386.c (maybe_get_pool_constant): Removed.
1547 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
1548 maybe_get_pool_constant.
1549 * config/i386/predicates.md (zero_extended_scalar_load_operand):
1552 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1554 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
1555 2.26 caveat. Update gas and gld versions.
1556 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
1559 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1561 * cgraph.h (set_malloc_flag): Declare.
1562 * cgraph.c (set_malloc_flag_1): New function.
1563 (set_malloc_flag): Likewise.
1564 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
1565 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
1567 (read_ipa_call_summary): Add support for reading is_return_callee.
1568 (write_ipa_call_summary): Stream is_return_callee.
1569 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
1570 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
1571 ipa-prop.h, ipa-fnsummary.h.
1572 (pure_const_names): Change to static.
1573 (malloc_state_e): Define.
1574 (malloc_state_names): Define.
1575 (funct_state_d): Add field malloc_state.
1576 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
1577 (check_retval_uses): New function.
1578 (malloc_candidate_p): Likewise.
1579 (analyze_function): Add support for malloc attribute.
1580 (pure_const_write_summary): Stream malloc_state.
1581 (pure_const_read_summary): Add support for reading malloc_state.
1582 (dump_malloc_lattice): New function.
1583 (propagate_malloc): New function.
1584 (warn_function_malloc): New function.
1585 (ipa_pure_const::execute): Call propagate_malloc and
1586 ipa_free_fn_summary.
1587 (pass_local_pure_const::execute): Add support for malloc attribute.
1588 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
1589 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
1591 2017-10-27 Martin Liska <mliska@suse.cz>
1593 PR gcov-profile/82457
1594 * doc/invoke.texi: Document that one needs a non-strict ISO mode
1595 for fork-like functions to be properly instrumented.
1597 2017-10-27 Richard Biener <rguenther@suse.de>
1600 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
1601 info when we redirected EH.
1603 2017-10-26 Michael Collison <michael.collison@arm.com>
1605 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
1607 (<optab>_trunchf<GPI:mode>2: New pattern.
1608 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
1609 * config/aarch64/iterators.md (wv): New mode attribute.
1610 (vf, VF): New mode attributes.
1611 (vgp, VGP): New mode attributes.
1612 (s): Update attribute with SImode and DImode prefixes.
1614 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
1616 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
1617 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
1618 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
1619 (nios2_option_overide): Initialize it. Don't allow R0-relative
1620 addressing with PIC.
1621 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
1622 (nios2_symbolic_constant_p): Likewise.
1623 (nios2_legitimate_address_p): Likewise.
1624 (nios2_r0rel_section_name_p): New.
1625 (nios2_symbol_ref_in_r0rel_data_p): New.
1626 (nios2_emit_move_sequence): Handle r0rel_constant_p.
1627 (r0rel_constant_p): New.
1628 (nios2_print_operand_address): Handle r0rel_constant_p.
1629 (nios2_cdx_narrow_form_p): Likewise.
1630 * config/nios2/nios2.opt (mr0rel-sec=): New option.
1631 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
1632 (Nios II Options): Document -mr0rel-sec.
1634 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
1636 * config/nios2/nios2.c: Include xregex.h.
1637 (nios2_gprel_sec_regex): New.
1638 (nios2_option_overide): Initialize it. Don't allow GP-relative
1639 addressing with PIC.
1640 (nios2_small_section_name_p): Check for regex match.
1641 * config/nios2/nios2.opt (mgprel-sec=): New option.
1642 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
1643 (Nios II Options): Document -mgprel-sec.
1645 2017-10-26 Jim Wilson <wilson@tuliptree.org>
1647 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
1649 2017-10-26 Tom de Vries <tom@codesourcery.com>
1651 PR tree-optimization/82707
1652 * gimple.c (gimple_copy): Fix unsharing of
1653 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
1655 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
1657 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
1658 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
1659 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
1660 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
1661 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
1662 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
1663 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
1664 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
1665 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
1666 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
1667 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
1668 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
1669 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
1670 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
1671 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
1672 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
1673 _mm512_mask_cmpunord_ps_mask): New intrinsics.
1675 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1677 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
1679 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
1680 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
1681 warning to rs6000.c. Remove the Undocumented flag, since it has
1683 (-mabi=ibmlongdouble): Likewise.
1684 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
1685 already set, set the default format for long double.
1686 (rs6000_debug_reg_global): Print whether long double is IBM or
1688 (rs6000_option_override_internal): Rework setting long double
1689 format. Only warn if the user is changing the long double default
1690 and they did not use -Wno-psabi.
1691 * doc/invoke.texi (PowerPC options): Update the documentation for
1692 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
1694 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1695 Alan Hayward <alan.hayward@arm.com>
1696 David Sherwood <david.sherwood@arm.com>
1698 * rtl.h (wider_subreg_mode): New function.
1699 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
1700 rather than an unsigned int *.
1701 * ira-color.c (regno_max_ref_width): Replace with...
1702 (regno_max_ref_mode): ...this new variable.
1703 (coalesced_pseudo_reg_slot_compare): Update accordingly.
1704 Use wider_subreg_mode.
1705 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
1706 rather than an unsigned int *.
1707 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
1708 (process_alt_operands): Likewise.
1709 (invariant_p): Likewise.
1710 * lra-spills.c (assign_mem_slot): Likewise.
1711 (add_pseudo_to_slot): Likewise.
1712 * lra.c (collect_non_operand_hard_regs): Likewise.
1713 (add_regs_to_insn_regno_info): Likewise.
1714 * reload1.c (regno_max_ref_width): Replace with...
1715 (regno_max_ref_mode): ...this new variable.
1716 (reload): Update accordingly. Update call to
1717 ira_sort_regnos_for_alter_reg.
1718 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
1719 (init_eliminable_invariants): Update to use regno_max_ref_mode.
1720 (scan_paradoxical_subregs): Likewise.
1722 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1724 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
1725 (aarch64_frame): Add emit_frame_chain boolean.
1726 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1727 Move eh_return case to aarch64_layout_frame.
1728 (aarch64_layout_frame): Initialize emit_frame_chain.
1729 (aarch64_expand_prologue): Use emit_frame_chain.
1731 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1733 * config/aarch64/aarch64.c (aarch64_layout_frame):
1734 Ensure LR is always stored at the bottom of the callee-saves.
1735 Remove rarely used frame layout which saves callee-saves at top of
1736 frame, so the store of LR can be used as a valid probe in all cases.
1738 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1740 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
1741 Improve unaligned TImode/TFmode base/offset split.
1743 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1744 Alan Hayward <alan.hayward@arm.com>
1745 David Sherwood <david.sherwood@arm.com>
1747 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
1748 * combine.c (find_single_use_1): Likewise.
1749 (expand_field_assignment): Likewise.
1750 (move_deaths): Likewise.
1751 * lra-constraints.c (simplify_operand_subreg): Likewise.
1752 (curr_insn_transform): Likewise.
1753 * lra.c (collect_non_operand_hard_regs): Likewise.
1754 (add_regs_to_insn_regno_info): Likewise.
1755 * rtlanal.c (reg_referenced_p): Likewise.
1756 (covers_regno_no_parallel_p): Likewise.
1758 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1760 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
1761 Don't print any bits outside the precision of the value.
1762 * wide-int.cc (test_printing): Add some new tests.
1764 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1766 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
1767 supports -xbrace_comment option.
1768 * configure: Regenerate.
1769 * config.in: Regenerate.
1770 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
1771 (ASM_CPU_SPEC): Use it.
1773 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1775 * target.def (static_rtx_alignment): New hook.
1776 * targhooks.h (default_static_rtx_alignment): Declare.
1777 * targhooks.c (default_static_rtx_alignment): New function.
1778 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
1779 * doc/tm.texi: Regenerate.
1780 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
1781 instead of targetm.constant_alignment. Remove call to
1783 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1784 (cris_preferred_mininum_alignment): New function, split out from...
1785 (cris_constant_alignment): ...here.
1786 (cris_static_rtx_alignment): New function.
1787 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
1789 (ix86_constant_alignment): ...here.
1790 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1791 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1792 (mmix_static_rtx_alignment): New function.
1793 * config/spu/spu.c (spu_static_rtx_alignment): New function.
1794 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1796 2017-10-26 Tamar Christina <tamar.christina@arm.com>
1799 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
1800 Add flag_trapping_math and flag_fp_int_builtin_inexact.
1802 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
1804 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
1805 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
1806 mark as a sign-extending load.
1807 (local_pic_load_u): Define.
1809 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
1812 * fold-const.c (operand_equal_for_comparison_p): Also return true
1813 if ARG0 is a simple variant of ARG1 with narrower precision.
1814 (fold_ternary_loc): Always pass unstripped operands to the predicate.
1816 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
1818 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
1820 * i386.h (processor_costs): Add gather_static, gather_per_elt,
1821 scatter_static, scatter_per_elt.
1822 * x86-tune-costs.h: Add new cost entries.
1824 2017-10-25 Richard Biener <rguenther@suse.de>
1826 * tree-ssa-sccvn.h (vn_eliminate): Declare.
1827 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
1828 class pass_fre): Move to ...
1829 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
1830 class pass_fre): ... here and adjust for statistics.
1832 2017-10-25 Jakub Jelinek <jakub@redhat.com>
1835 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
1836 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
1837 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
1838 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
1841 2017-10-25 Richard Biener <rguenther@suse.de>
1843 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
1844 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
1845 eliminate_push_avail, eliminate_insert): Move inside...
1846 (class eliminate_dom_walker): ... this class in preparation
1848 (fini_eliminate): Remove by merging with ...
1849 (eliminate): ... this function. Adjust for class changes.
1850 (pass_pre::execute): Remove fini_eliminate call.
1851 (pass_fre::execute): Likewise.
1853 2017-10-24 Jakub Jelinek <jakub@redhat.com>
1856 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
1857 (VPERMI2, VPERMI2I): New mode iterators.
1858 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
1859 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
1861 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
1862 mode iterator. Remove 3 old define_insn patterns.
1863 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
1864 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
1865 VPERMI2 mode iterator, remove the other two expanders.
1866 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
1867 to use VPERMI2 mode iterator, add another alternative for vpermi2*
1868 instructions, remove the other two patterns.
1869 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
1870 mode iterator, remove the other two patterns.
1871 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
1872 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
1873 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
1874 and adjust argument order accordingly.
1875 (ix86_expand_vec_perm): Adjust caller.
1876 (expand_vec_perm_1): Likewise.
1877 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
1878 (expand_vec_perm_vpermt2_vpshub2): ... this.
1879 (ix86_expand_vec_perm_const_1): Adjust caller.
1880 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
1883 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
1884 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
1885 (vec_shl_<mode>): Remove unused expander.
1886 (avx512bw_<shift_insn><mode>3): New define_insn.
1887 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
1888 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
1890 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
1893 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
1896 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
1898 PR rtl-optimization/82396
1899 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
1900 (autopref_multipass_init): Simplify initialization.
1901 (autopref_rank_data): Simplify sort order.
1902 * gcc/sched-int.h (autopref_multipass_data_): Remove
1903 multi_mem_insn_p, min_offset and max_offset.
1905 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
1908 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1909 Check special value of flag_omit_frame_pointer.
1910 (aarch64_can_eliminate): Likewise.
1911 (aarch64_override_options_after_change_1): Simplify handling of
1912 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
1914 2017-10-24 Richard Biener <rguenther@suse.de>
1916 PR tree-optimization/82697
1917 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
1918 zero for conditional load and unconditional store.
1920 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
1922 * doc/install.texi: Document bootstrap-cet.
1924 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
1927 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
1928 ENDBR instruction at function entrance if function is only
1931 2017-10-24 Jakub Jelinek <jakub@redhat.com>
1934 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
1935 patterns to better describe from which operation the CF is computed.
1936 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
1937 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
1938 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
1939 is 0, use _0 suffixed expanders instead of emitting a comparison
1942 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
1944 * config/i386/i386.md(*movsf_internal, *movdf_internal):
1945 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
1947 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
1950 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
1951 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
1952 * loop-iv.c (iv_get_reaching_def): Likewise.
1953 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
1954 variable is promoted and the partition contains undefined values.
1956 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1958 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
1960 (nios2_address_cost): Define.
1961 (nios2_legitimize_address): Recognize (exp + constant) directly.
1962 (TARGET_ADDRESS_COST): Define.
1964 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1966 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
1967 (nios2_symbolic_memory_operand_p): Declare.
1968 (nios2_split_large_constant): Declare.
1969 (nios2_split_symbolic_memory_operand): Declare.
1970 * config/nios2/nios2.c: Adjust includes.
1971 (nios2_symbolic_constant_allowed): New.
1972 (nios2_symbolic_constant_p): New.
1973 (nios2_plus_symbolic_constant_p): New.
1974 (nios2_valid_addr_expr_p): Recognize addresses involving
1976 (nios2_legitimate_address_p): Likewise, also LO_SUM.
1977 (nios2_symbolic_memory_operand_p): New.
1978 (nios2_large_constant_p): New.
1979 (nios2_split_large_constant): New.
1980 (nios2_split_plus_large_constant): New.
1981 (nios2_split_symbolic_memory_operand): New.
1982 (nios2_legitimize_address): Code refactoring. Handle addresses
1983 involving symbolic constants.
1984 (nios2_emit_move_sequence): Likewise.
1985 (nios2_print_operand): Improve error output.
1986 (nios2_print_operand_address): Handle LO_SUM.
1987 (nios2_cdx_narrow_form_p): Likewise.
1988 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
1989 operands involving symbolic constants.
1990 (movhi_internal, movsi_internal): Likewise.
1991 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
1992 (extendhisi2, extendqi<mode>2): Likewise.
1994 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1996 * tree-pass.h (PROP_rtl_split_insns): Define.
1997 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
1999 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2001 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2003 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2006 * target.def (const_not_ok_for_debug_p): Default to
2007 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2008 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2009 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2010 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2011 which targetm.const_not_ok_for_debug_p returned true.
2012 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2014 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2016 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2017 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2018 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2020 (ix86_const_not_ok_for_debug_p): New function.
2021 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2022 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2024 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2027 * system.h: Conditionally include "unique-ptr.h" if
2028 INCLUDE_UNIQUE_PTR is defined.
2029 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2030 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2032 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2034 * config/rl78/rl78.md: New define_expand "subdi3".
2036 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2039 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2040 DF_REF_INSN if DF_REF_INSN_INFO is false.
2042 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2044 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2045 xmm_move instead of sse_move.
2046 (sse_store_index): New function.
2047 (ix86_register_move_cost): Be more sensible about mismatch stall;
2048 model AVX moves correctly; make difference between sse->integer and
2050 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2051 moves; make difference between SSE and AVX.
2052 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2053 and zmm_move. Increase size of sse load and store tables;
2054 add unaligned load and store tables; add ssemmx_to_integer.
2055 * x86-tune-costs.h: Update all entries according to real
2056 move latencies from Agner Fog's manual and chip documentation.
2058 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2061 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2062 * config/i386/constraints.md (Wf): New constraint.
2063 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2064 (cmp<dwi>_doubleword): Removed.
2065 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2066 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2067 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2068 expand with cmp<dwi>_doubleword. For LTU and GEU use
2069 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2071 * common.opt (gcolumn-info): Enable by default.
2072 * doc/invoke.texi (gcolumn-info): Document new default.
2074 2017-10-23 Richard Biener <rguenther@suse.de>
2076 PR tree-optimization/82672
2077 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2078 Fold the stmt if we propagated into it.
2080 2017-10-23 Richard Biener <rguenther@suse.de>
2082 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2083 (bitmap_remove_expr_from_set): ... this. All callers call this
2084 for non-constant values.
2085 (bitmap_set_subtract): Rename to...
2086 (bitmap_set_subtract_expressions): ... this. Adjust and
2088 (bitmap_set_contains_value): Remove superfluous check.
2089 (bitmap_set_replace_value): Inline into single caller ...
2090 (bitmap_value_replace_in_set): ... here and simplify.
2091 (dependent_clean): Merge into ...
2092 (clean): ... this using an overload. Adjust.
2093 (prune_clobbered_mems): Adjust.
2094 (compute_antic_aux): Likewise.
2095 (compute_partial_antic_aux): Likewise.
2097 2017-10-23 Richard Biener <rguenther@suse.de>
2099 PR tree-optimization/82129
2101 2017-08-01 Richard Biener <rguenther@suse.de>
2103 PR tree-optimization/81181
2104 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2105 (compute_antic): ... end of iteration here.
2107 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2109 * target.def (starting_frame_offset): New hook.
2110 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2111 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2112 * doc/tm.texi.in: Regenerate.
2113 * hooks.h (hook_hwi_void_0): Declare.
2114 * hooks.c (hook_hwi_void_0): New function.
2115 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2116 STARTING_FRAME_OFFSET.
2117 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2118 * reload1.c (reload): Likewise.
2119 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2120 instead of STARTING_FRAME_OFFSET.
2121 * function.c (try_fit_stack_local): Likewise.
2122 (assign_stack_local_1): Likewise
2123 (instantiate_virtual_regs): Likewise.
2124 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2125 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2126 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2127 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2128 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2129 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2130 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2131 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2132 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2133 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2134 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2135 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2136 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2137 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2138 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2139 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2140 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2141 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2142 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2143 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2144 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2145 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2146 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2147 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2148 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2149 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2150 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2151 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2152 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2153 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2154 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2155 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2156 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2157 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2158 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2159 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2160 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2161 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2162 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2163 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2164 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2165 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2166 return a HOST_WIDE_INT.
2167 (avr_builtin_setjmp_frame_value): Use it instead of
2168 STARTING_FRAME_OFFSET.
2169 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2170 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2171 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2173 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2174 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2175 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2176 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2177 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2178 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2179 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2180 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2181 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2182 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2183 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2184 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2186 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2187 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2188 * config/mips/mips.c (mips_compute_frame_info): Refer to
2189 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2190 (mips_starting_frame_offset): New function.
2191 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2192 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2193 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2194 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2195 and return a HOST_WIDE_INT.
2196 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2197 (mmix_initial_elimination_offset): Refer to
2198 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2199 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2200 * config/pa/pa.c (pa_starting_frame_offset): New function.
2201 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2202 (pa_expand_prologue): Likewise.
2203 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2204 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2205 !FRAME_GROWS_DOWNWARD handling to...
2206 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2207 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2208 !FRAME_GROWS_DOWNWARD handling to...
2209 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2210 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2211 !FRAME_GROWS_DOWNWARD handling to...
2212 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2213 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2215 (rs6000_starting_frame_offset): New function.
2216 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2217 !FRAME_GROWS_DOWNWARD handling to...
2218 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2219 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2220 !FRAME_GROWS_DOWNWARD handling to...
2221 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2222 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2223 !FRAME_GROWS_DOWNWARD handling to...
2224 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2225 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2226 (rs6000_starting_frame_offset): New function.
2227 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2228 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2229 * config/vax/vax.c (vax_starting_frame_offset): New function.
2230 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2231 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2232 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2233 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2234 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2235 * system.h (STARTING_FRAME_OFFSET): Poison.
2237 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2239 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2240 SCALAR_TYPE_MODE instead of TYPE_MODE.
2242 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2243 Alan Hayward <alan.hayward@arm.com>
2244 David Sherwood <david.sherwood@arm.com>
2246 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2248 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2249 Alan Hayward <alan.hayward@arm.com>
2250 David Sherwood <david.sherwood@arm.com>
2252 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
2254 2017-10-23 Richard Biener <rguenther@suse.de>
2256 PR tree-optimization/82129
2257 * tree-ssa-pre.c (bitmap_set_and): Remove.
2258 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
2259 canonicalizing expressions in the set to those with lowest
2260 ID rather than taking that from the first edge.
2262 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2264 * combine.c (rtx_equal_for_field_assignment_p): Use
2265 byte_lowpart_offset.
2267 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2268 Alan Hayward <alan.hayward@arm.com>
2269 David Sherwood <david.sherwood@arm.com>
2271 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
2272 to a SUBREG_PROMOTED_VAR.
2274 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2275 Alan Hayward <alan.hayward@arm.com>
2276 David Sherwood <david.sherwood@arm.com>
2278 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
2279 (expand_debug_source_expr): Likewise.
2280 * combine.c (combine_simplify_rtx): Likewise.
2281 * cse.c (fold_rtx): Likewise.
2282 * optabs.c (expand_float): Likewise.
2283 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2284 (simplify_binary_operation_1): Likewise.
2286 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2287 Alan Hayward <alan.hayward@arm.com>
2288 David Sherwood <david.sherwood@arm.com>
2290 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
2291 (record_promoted_value): Likewise.
2292 * expr.c (expand_expr_real_2): Likewise.
2293 * ree.c (update_reg_equal_equiv_notes): Likewise.
2294 (combine_set_extension): Likewise.
2295 * rtlanal.c (low_bitmask_len): Likewise.
2296 * simplify-rtx.c (neg_const_int): Likewise.
2297 (simplify_binary_operation_1): Likewise.
2299 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2300 Alan Hayward <alan.hayward@arm.com>
2301 David Sherwood <david.sherwood@arm.com>
2303 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
2304 * regcprop.c (maybe_mode_change): Likewise.
2305 * reload1.c (alter_reg): Likewise.
2307 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2309 * inchash.h (inchash::hash::add_wide_int): New function.
2310 * lto-streamer-out.c (hash_tree): Use it.
2312 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2314 * inchash.h (inchash::hash::add_wide_int): Rename to...
2315 (inchash::hash::add_hwi): ...this.
2316 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
2317 (polymorphic_call_target_hasher::hash): Likewise.
2318 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
2319 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
2320 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
2321 * lto-streamer-out.c (hash_tree): Likewise.
2322 * optc-save-gen.awk: Likewise.
2323 * tree.c (add_expr): Likewise.
2325 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2328 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
2329 for ordered inequality comparisons even with TARGET_IEEE_FP.
2331 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2334 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
2335 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2336 Expand with cmp<dwi>_doubleword.
2338 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2340 * extend.texi: Add x86 specific to 'nocf_check' attribute.
2341 List CET intrinsics.
2342 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
2343 specific to -fcf-protection option.
2345 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2347 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
2348 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
2349 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
2350 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
2351 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
2352 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
2353 (extra_objs): Add cet.o for Linux/x86 targets.
2354 (tmake_file): Add i386/t-cet for Linux/x86 targets.
2355 * config/i386/cet.c: New file.
2356 * config/i386/cetintrin.h: Likewise.
2357 * config/i386/t-cet: Likewise.
2358 * config/i386/cpuid.h (bit_SHSTK): New.
2359 (bit_IBT): Likewise.
2360 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
2361 pass IBT and SHSTK bits.
2362 * config/i386/i386-builtin-types.def
2363 (VOID_FTYPE_UNSIGNED_PVOID): New.
2364 (VOID_FTYPE_UINT64_PVOID): Likewise.
2365 * config/i386/i386-builtin.def: Add CET intrinsics.
2366 * config/i386/i386-c.c (ix86_target_macros_internal): Add
2367 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
2368 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
2369 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
2371 * config/i386/i386.c (rest_of_insert_endbranch): New.
2372 (pass_data_insert_endbranch): Likewise.
2373 (pass_insert_endbranch): Likewise.
2374 (make_pass_insert_endbranch): Likewise.
2375 (ix86_notrack_prefixed_insn_p): Likewise.
2376 (ix86_target_string): Add -mibt, -mshstk flags.
2377 (ix86_option_override_internal): Add flag_cf_protection
2379 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
2380 (ix86_print_operand): Add 'notrack' prefix output.
2381 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
2382 (ix86_expand_builtin): Expand CET intrinsics.
2383 (x86_output_mi_thunk): Add 'endbranch' instruction.
2384 * config/i386/i386.h (TARGET_IBT): New.
2385 (TARGET_IBT_P): Likewise.
2386 (TARGET_SHSTK): Likewise.
2387 (TARGET_SHSTK_P): Likewise.
2388 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
2389 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
2390 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
2391 (builtin_setjmp_setup): New pattern.
2392 (builtin_longjmp): Likewise.
2393 (rdssp<mode>): Likewise.
2394 (incssp<mode>): Likewise.
2395 (saveprevssp): Likewise.
2396 (rstorssp): Likewise.
2397 (wrss<mode>): Likewise.
2398 (wruss<mode>): Likewise.
2399 (setssbsy): Likewise.
2400 (clrssbsy): Likewise.
2401 (nop_endbr): Likewise.
2402 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
2404 * config/i386/immintrin.h: Include <cetintrin.h>.
2405 * config/i386/linux-common.h
2406 (file_end_indicate_exec_stack_and_cet): New prototype.
2407 (TARGET_ASM_FILE_END): New.
2409 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2411 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
2412 latencies instead of having separate table; make difference between
2413 integer and float costs.
2414 * i386.h (processor_costs): Remove scalar_stmt_cost,
2415 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
2416 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
2418 * x86-tune-costs.h: Remove entries which has been removed in
2419 procesor_costs from all tables; make cond_taken_branch_cost
2420 and cond_not_taken_branch_cost COST_N_INSNS based.
2422 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2424 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
2426 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2428 * config/i386/i386.md (isa): Remove fma_avx512f.
2429 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
2430 <avx512>_fmadd_<mode>_mask3<round_name>,
2431 <avx512>_fmsub_<mode>_mask<round_name>,
2432 <avx512>_fmsub_<mode>_mask3<round_name>,
2433 <avx512>_fnmadd_<mode>_mask<round_name>,
2434 <avx512>_fnmadd_<mode>_mask3<round_name>,
2435 <avx512>_fnmsub_<mode>_mask<round_name>,
2436 <avx512>_fnmsub_<mode>_mask3<round_name>,
2437 <avx512>_fmaddsub_<mode>_mask<round_name>,
2438 <avx512>_fmaddsub_<mode>_mask3<round_name>,
2439 <avx512>_fmsubadd_<mode>_mask<round_name>,
2440 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
2441 (*vec_widen_umult_even_v16si<mask_name>,
2442 *vec_widen_smult_even_v16si<mask_name>): Likewise.
2443 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
2445 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2447 * extend.texi: Add 'nocf_check' documentation.
2448 * gimple.texi: Add second parameter to
2449 gimple_build_call_from_tree.
2450 * invoke.texi: Add -fcf-protection documentation.
2451 * rtl.texi: Add REG_CALL_NOTRACK documenation.
2453 2017-10-20 Richard Biener <rguenther@suse.de>
2455 PR tree-optimization/82473
2456 * tree-vect-loop.c (vectorizable_reduction): Properly get at
2457 the largest input type.
2459 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2461 * c-attribs.c (handle_nocf_check_attribute): New function.
2462 (c_common_attribute_table): Add 'nocf_check' handling.
2463 * gimple-parser.c: Add second argument NULL to
2464 gimple_build_call_from_tree.
2465 * attrib.c (comp_type_attributes): Check nocf_check attribute.
2466 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
2468 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
2469 * common.opt: Add fcf-protection flag.
2470 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
2471 * flag-types.h: Add enum cf_protection_level.
2472 * gimple.c (gimple_build_call_from_tree): Add second parameter.
2473 Add 'nocf_check' attribute propagation to gimple call.
2474 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
2475 (gimple_build_call_from_tree): Update prototype.
2476 (gimple_call_nocf_check_p): New function.
2477 (gimple_call_set_nocf_check): Likewise.
2478 * gimplify.c: Add second argument to gimple_build_call_from_tree.
2479 * ipa-icf.c: Add nocf_check attribute in statement hash.
2480 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
2481 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
2482 * toplev.c (process_options): Add flag_cf_protection handling.
2484 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2486 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
2488 2017-10-20 Richard Biener <rguenther@suse.de>
2490 PR tree-optimization/82603
2491 * tree-if-conv.c (predicate_mem_writes): Make sure to only
2492 remove false predicated stores.
2494 2017-10-20 Richard Biener <rguenther@suse.de>
2496 * graphite-isl-ast-to-gimple.c
2497 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
2498 Remove return value and simplify, dump copied stmt after lhs
2500 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
2501 Reduce dump verbosity.
2502 (gsi_insert_earliest): Likewise.
2503 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
2504 * graphite.c (print_global_statistics): Adjust dumping.
2505 (print_graphite_scop_statistics): Likewise.
2506 (print_graphite_statistics): Do not dump loops here.
2507 (graphite_transform_loops): But here.
2509 2017-10-20 Nicolas Roche <roche@adacore.com>
2511 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
2512 * configure: Regenerate.
2514 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2517 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
2518 functions when optimizing replace GIMPLE_RETURN stmts with
2519 calls to __builtin_unreachable ().
2522 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
2523 for -fsanitize=thread link of executables.
2524 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
2525 link of executables.
2528 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
2530 (<shift_insn><mode>3<mask_name>): Change the last of the 3
2531 define_insns for logical vector shifts to use VI248_AVX512BW
2532 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
2533 condition, useless isa and prefix attributes. Change the first
2534 2 of these define_insns to ...
2535 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
2536 define_insn for avx512vl.
2537 (<shift_insn><mode>3): ... and this, new define_insn without
2538 masking for non-avx512vl.
2541 * config/i386/sse.md (*andnot<mode>3,
2542 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
2543 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
2544 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
2545 not applied use empty suffix even for TARGET_AVX512VL.
2546 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
2547 is applied, supply evex,evex or evex,evex,evex instead of just
2550 2017-10-20 Julia Koval <julia.koval@intel.com>
2552 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
2553 (OPTION_MASK_ISA_GFNI_UNSET): New.
2554 (ix86_handle_option): Handle OPT_mgfni.
2555 * config/i386/cpuid.h (bit_GFNI): New.
2556 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
2557 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
2558 * config/i386/i386.c (ix86_target_string): Add -mgfni.
2559 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
2560 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
2561 * config/i386/i386.opt: Add mgfni.
2563 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
2565 * config/msp430/msp430.c (msp430_option_override): Disable
2566 -fdelete-null-pointer-checks.
2567 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
2569 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2571 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
2572 of x87 and SSE instructions.
2574 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2576 * asan.c (create_cond_insert_point): Do not update edge count.
2577 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
2578 (afdo_propagate_circuit): Likewise.
2579 (afdo_calculate_branch_prob): Likewise.
2580 (afdo_annotate_cfg): Likewise.
2581 * basic-block.h (struct edge_def): Remove count.
2582 (edge_def::count): New accessor.
2583 * bb-reorder.c (rotate_loop): Update.
2584 (find_traces_1_round): Update.
2585 (connect_traces): Update.
2586 (sanitize_hot_paths): Update.
2587 * cfg.c (unchecked_make_edge): Update.
2588 (make_single_succ_edge): Update.
2589 (check_bb_profile): Update.
2590 (dump_edge_info): Update.
2591 (update_bb_profile_for_threading): Update.
2592 (scale_bbs_frequencies_int): Update.
2593 (scale_bbs_frequencies_gcov_type): Update.
2594 (scale_bbs_frequencies_profile_count): Update.
2595 (scale_bbs_frequencies): Update.
2596 * cfganal.c (connect_infinite_loops_to_exit): Update.
2597 * cfgbuild.c (compute_outgoing_frequencies): Update.
2598 (find_many_sub_basic_blocks): Update.
2599 * cfgcleanup.c (try_forward_edges): Update.
2600 (try_crossjump_to_edge): Update
2601 * cfgexpand.c (expand_gimple_cond): Update
2602 (expand_gimple_tailcall): Update
2603 (construct_exit_block): Update
2604 * cfghooks.c (verify_flow_info): Update
2605 (redirect_edge_succ_nodup): Update
2606 (split_edge): Update
2607 (make_forwarder_block): Update
2608 (duplicate_block): Update
2609 (account_profile_record): Update
2610 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
2611 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
2612 * cfgloopmanip.c (scale_loop_profile): Update.
2614 (lv_adjust_loop_entry_edge): Update.
2615 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
2616 (force_nonfallthru_and_redirect): Update.
2617 (purge_dead_edges): Update.
2618 (rtl_flow_call_edges_add): Update.
2619 * cgraphunit.c (init_lowered_empty_function): Update.
2620 (cgraph_node::expand_thunk): Update.
2621 * gimple-pretty-print.c (dump_probability): Update.
2622 (dump_edge_probability): Update.
2623 * gimple-ssa-isolate-paths.c (isolate_path): Update.
2624 * haifa-sched.c (sched_create_recovery_edges): Update.
2625 * hsa-gen.c (convert_switch_statements): Update.
2626 * ifcvt.c (dead_or_predicable): Update.
2627 * ipa-inline-transform.c (inline_transform): Update.
2628 * ipa-split.c (split_function): Update.
2629 * ipa-utils.c (ipa_merge_profiles): Update.
2630 * loop-doloop.c (add_test): Update.
2631 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
2632 * lto-streamer-in.c (input_cfg): Update.
2633 (input_function): Update.
2634 * lto-streamer-out.c (output_cfg): Update.
2635 * modulo-sched.c (sms_schedule): Update.
2636 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
2637 * predict.c (maybe_hot_edge_p): Update.
2638 (unlikely_executed_edge_p): Update.
2639 (probably_never_executed_edge_p): Update.
2640 (dump_prediction): Update.
2641 (drop_profile): Update.
2642 (propagate_unlikely_bbs_forward): Update.
2643 (determine_unlikely_bbs): Update.
2644 (force_edge_cold): Update.
2645 * profile.c (compute_branch_probabilities): Update.
2646 * reg-stack.c (better_edge): Update.
2647 * shrink-wrap.c (handle_simple_exit): Update.
2648 * tracer.c (better_p): Update.
2649 * trans-mem.c (expand_transaction): Update.
2650 (split_bb_make_tm_edge): Update.
2651 * tree-call-cdce.c: Update.
2652 * tree-cfg.c (gimple_find_sub_bbs): Update.
2653 (gimple_split_edge): Update.
2654 (gimple_duplicate_sese_region): Update.
2655 (gimple_duplicate_sese_tail): Update.
2656 (gimple_flow_call_edges_add): Update.
2657 (insert_cond_bb): Update.
2658 (execute_fixup_cfg): Update.
2659 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
2660 * tree-complex.c (expand_complex_div_wide): Update.
2661 * tree-eh.c (lower_resx): Update.
2662 (unsplit_eh): Update.
2663 (cleanup_empty_eh_move_lp): Update.
2664 * tree-inline.c (copy_edges_for_bb): Update.
2665 (freqs_to_counts): Update.
2666 (copy_cfg_body): Update.
2667 * tree-ssa-dce.c (remove_dead_stmt): Update.
2668 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
2669 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
2670 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
2671 (unloop_loops): Update.
2672 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
2673 * tree-ssa-loop-split.c (connect_loops): Update.
2674 (split_loop): Update.
2675 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
2676 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
2677 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
2678 * tree-ssa-reassoc.c (branch_fixup): Update.
2679 * tree-ssa-tail-merge.c (replace_block_by): Update.
2680 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
2681 (compute_path_counts): Update.
2682 (update_profile): Update.
2683 (recompute_probabilities): Update.
2684 (update_joiner_offpath_counts): Update.
2685 (estimated_freqs_path): Update.
2686 (freqs_to_counts_path): Update.
2687 (clear_counts_path): Update.
2688 (ssa_fix_duplicate_block_edges): Update.
2689 (duplicate_thread_path): Update.
2690 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
2691 (case_bit_test_cmp): Update.
2692 (collect_switch_conv_info): Update.
2693 (gen_inbound_check): Update.
2694 (do_jump_if_equal): Update.
2695 (emit_cmp_and_jump_insns): Update.
2696 * tree-tailcall.c (decrease_profile): Update.
2697 (eliminate_tail_call): Update.
2698 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
2699 (vect_do_peeling): Update.
2700 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
2701 * ubsan.c (ubsan_expand_null_ifn): Update.
2702 (ubsan_expand_ptr_ifn): Update.
2703 * value-prof.c (gimple_divmod_fixed_value): Update.
2704 (gimple_mod_pow2): Update.
2705 (gimple_mod_subtract): Update.
2706 (gimple_ic): Update.
2707 (gimple_stringop_fixed_value): Update.
2709 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
2712 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
2714 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
2716 PR rtl-optimization/82395
2717 * ira-color.c (allocno_priority_compare_func): Fix comparison step
2718 based on non_spilled_static_chain_regno_p.
2720 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
2722 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
2723 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
2724 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
2726 2017-10-19 Martin Sebor <msebor@redhat.com>
2728 PR tree-optimization/82596
2729 * tree.c (array_at_struct_end_p): Handle STRING_CST.
2731 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
2733 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
2734 (get_mem_refs_of_builtin_call): Likewise.
2735 * builtins.c (expand_builtin_apply): Adjust call to
2736 allocate_dynamic_stack_space.
2737 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
2738 the third argument to allocate_dynamic_stack_space, otherwise -1.
2739 (expand_builtin): Deal with all alloca variants.
2740 (is_inexpensive_builtin): Likewise.
2741 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
2742 * calls.c (special_function_p): Deal with all alloca variants.
2743 (initialize_argument_information): Adjust call to
2744 allocate_dynamic_stack_space.
2745 (expand_call): Likewise.
2746 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
2747 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
2748 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
2749 use it for the stack usage computation.
2750 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
2751 * function.c (gimplify_parameters): Call build_alloca_call_expr.
2752 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
2753 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
2754 (in_loop_p): Remove first argument and useless check.
2755 (pass_walloca::execute): Remove useless test and adjust call to above.
2756 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
2757 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
2758 (gimplify_call_expr): Deal with all alloca variants.
2759 * hsa-gen.c (gen_hsa_alloca): Likewise.
2760 (gen_hsa_insns_for_call): Likewise.
2761 * ipa-pure-const.c (special_builtin_state): Likewise.
2762 * tree-chkp.c (chkp_build_returned_bound): Likewise.
2763 * tree-object-size.c (alloc_object_size): Likewise.
2764 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2765 (call_may_clobber_ref_p_1): Likewise.
2766 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
2767 (ccp_fold_stmt): Likewise.
2768 (optimize_stack_restore): Likewise.
2769 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
2770 (mark_all_reaching_defs_necessary_1): Likewise.
2771 (propagate_necessity): Likewise.
2772 (eliminate_unnecessary_stmts): Likewise.
2773 * tree.c (build_common_builtin_nodes): Build
2774 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
2775 (build_alloca_call_expr): New function.
2776 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
2777 (CASE_BUILT_IN_ALLOCA): Likewise.
2778 (build_alloca_call_expr): Declare.
2779 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
2781 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
2784 * dwarf2out.c (new_die_raw): New static inline function.
2785 (new_die): Use it to create the DIE.
2786 (add_AT_external_die_ref): Likewise.
2787 (clone_die): Likewise.
2788 (clone_as_declaration): Likewise.
2789 (dwarf2out_vms_debug_main_pointer): Likewise.
2790 (base_type_die): Likewise. Remove early return for corner cases.
2791 Do not call add_pubtype on the DIE here.
2792 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
2793 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
2794 typedefs for base types with DW_AT_endianity. Make sure a DIE with
2795 native order exists for base types, attach the DIE manually and call
2796 add_pubtype on it. Do not equate a reverse order DIE to the type.
2798 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
2800 * config/arm/arm.c (align_ok_ldrd_strd): New function.
2801 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
2803 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
2805 2017-10-19 Jakub Jelinek <jakub@redhat.com>
2807 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
2808 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
2809 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
2810 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
2811 * opts.c (sanitizer_opts): Add builtin.
2812 * ubsan.c (instrument_builtin): New function.
2813 (pass_ubsan::execute): Call it.
2814 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
2815 * doc/invoke.texi: Document -fsanitize=builtin.
2817 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
2818 builtins, store max (log2 (align), 0) into uchar field instead of
2819 align into uptr field.
2820 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
2821 store uchar 0 field instead of uptr 0 field.
2822 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
2823 instead of passing one address of struct with 2 locations pass
2824 two addresses of structs with 1 location each.
2825 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
2826 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
2827 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
2828 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
2829 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
2830 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
2831 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
2832 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
2834 2017-10-19 Martin Liska <mliska@suse.cz>
2837 * file-find.c (remove_prefix): Remove.
2838 * file-find.h (remove_prefix): Likewise.
2839 * gcc-ar.c: Remove smartness of lookup.
2841 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
2843 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
2844 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
2845 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
2847 2017-10-19 Jakub Jelinek <jakub@redhat.com>
2850 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
2851 (setcc + and to xor + setcc): New peephole2.
2853 2017-10-19 Tom de Vries <tom@codesourcery.com>
2855 * doc/sourcebuild.texi (Test Directives, Variants of
2856 dg-require-support): Add dg-require-stack-size.
2858 2017-10-19 Martin Liska <mliska@suse.cz>
2861 * gimplify.c (gimplify_decl_expr): Do not instrument variables
2862 that have a large alignment.
2863 (gimplify_target_expr): Likewise.
2865 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
2867 PR rtl-optimization/82602
2868 * ira.c (rtx_moveable_p): Return false for volatile asm.
2870 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
2873 * config/i386/i386-modes.def (CCGZ): New CC mode.
2874 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
2875 * config/i386/predicates.md (ix86_comparison_operator):
2877 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2878 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
2879 with double-word subtraction.
2880 (put_condition_code): Handle CCGZmode.
2882 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
2884 * wide-int.cc (debug (const wide_int &)): New.
2885 (debug (const wide_int *)): New.
2886 (debug (const widest_int &)): New.
2887 (debug (const widest_int *)): New.
2889 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
2892 * lra-constraints.c (curr_insn_transform): Use non-input operand
2893 instead of output one for matched reload.
2895 2017-10-18 Bin Cheng <bin.cheng@arm.com>
2897 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
2898 (tree-ssa-loop-ivopts.h): New header file.
2899 (struct builtin_info): New fields.
2900 (classify_builtin_1): Compute and record base and offset parts for
2901 memset builtin partition by calling strip_offset.
2902 (offset_cmp, fuse_memset_builtins): New functions.
2903 (finalize_partitions): Fuse adjacent memset partitions by calling
2905 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
2906 Expose the interface.
2907 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
2909 2017-10-18 Bin Cheng <bin.cheng@arm.com>
2911 PR tree-optimization/82574
2912 * tree-loop-distribution.c (find_single_drs): New parameter. Check
2913 that data reference must be executed exactly once per iteration
2914 against the outermost loop in nest.
2915 (classify_partition): Update call to above function.
2917 2017-10-18 Richard Biener <rguenther@suse.de>
2919 PR tree-optimization/82591
2920 * graphite.c (graphite_transform_loops): Move code gen message
2922 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2923 Here. Handle scop_to_isl_ast failing.
2924 (scop_to_isl_ast): Limit the number of ISL operations.
2926 2017-10-18 Richard Biener <rguenther@suse.de>
2928 * graphite-isl-ast-to-gimple.c
2929 (translate_isl_ast_to_gimple::set_rename): Simplify.
2930 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
2931 (graphite_copy_stmts_from_block): ... here.
2932 (copy_bb_and_scalar_dependences): Simplify.
2933 (add_parameters_to_ivs_params): Canonicalize.
2934 (generate_entry_out_of_ssa_copies): Simplify.
2935 * graphite-sese-to-poly.c (extract_affine_name): Simplify
2936 by passing in ISL dimension.
2937 (parameter_index_in_region_1): Rename to ...
2938 (parameter_index_in_region): ... this.
2939 (extract_affine): Adjust assert, pass down parameter index.
2940 (add_param_constraints): Use range-info when available.
2941 (build_scop_context): Adjust.
2942 * sese.c (new_sese_info): Adjust.
2943 (free_sese_info): Likewise.
2944 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
2945 Remove unused typedefs.
2946 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
2948 2017-10-18 Martin Liska <mliska@suse.cz>
2950 * combine.c (simplify_compare_const): Add gcc_fallthrough.
2952 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2954 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
2955 (s390_sched_init): Do not reset s390_sched_state if we entered the
2956 current basic block via a fallthru edge and all others are unlikely.
2958 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2960 * config/s390/s390.c (NUM_SIDES): New variable.
2961 (LONGRUNNING_THRESHOLD): New variable.
2962 (LATENCY_FACTOR): New variable.
2963 (s390_sched_score): Decrease score for long-running instructions on
2965 (s390_sched_variable_issue): Perform bookkeeping for long-running
2968 2017-10-18 Richard Biener <rguenther@suse.de>
2970 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
2971 Simplify with removal of the parameter rename map.
2972 (set_rename): Likewise.
2973 (should_copy_to_new_region): Likewise.
2974 (graphite_copy_stmts_from_block): Likewise.
2975 (copy_bb_and_scalar_dependences): Remove initialization of
2976 unused copied_bb_map.
2978 (copy_internal_parameters): Likewise.
2979 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
2980 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
2981 Use INTEGRAL_TYPE_P.
2982 (parameter_index_in_region_1): Rename to ...
2983 (assign_parameter_index_in_region): ... this. Assert we have
2984 a parameter we handle.
2985 (scan_tree_for_params): Adjust.
2986 * sese.h (parameter_rename_map_t): Remove.
2987 (struct sese_info_t): Remove unused parameter_rename_map and
2988 copied_bb_map members.
2989 * sese.c (new_sese_info): Adjust.
2990 (free_sese_info): Likewise.
2992 2017-10-18 Martin Liska <mliska@suse.cz>
2995 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
2996 on an abnormal edge.
2998 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3000 * doc/invoke.texi (ffunction-sections and fdata-sections):
3003 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3005 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3006 the use statement can throw internally.
3008 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3010 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3011 any RTX present on the RHS of a SET.
3012 * compare-elim.c (try_eliminate_compare): Restore comment.
3014 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3016 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3017 may be also called on tcc_type nodes.
3018 * langhooks.c (lhd_tree_size): Likewise.
3020 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3022 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3023 format_warning_at_substring.
3024 (maybe_warn): Convert source_range * param to a location_t. Pass
3025 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3026 (format_directive): Remove code to extract source_ranges and
3027 source_range * in favor of just a location_t.
3028 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3030 * substring-locations.c (format_warning_va): Convert
3031 source_range * param to a location_t.
3032 (format_warning_at_substring): Likewise.
3033 * substring-locations.h (format_warning_va): Likewise.
3034 (format_warning_at_substring): Likewise.
3036 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3038 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3040 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3041 and scatter/gather ops.
3043 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3044 vec_gather_load and vec_scatter_store.
3045 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3046 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3047 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3048 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3049 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3050 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3052 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3054 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3055 Detect FTST instruction and handle its register pops. Only pop
3056 second operand if can_pop_second_op is true.
3057 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3058 set can_pop_second_op to false in the compare_for_stack_reg call.
3060 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3061 output_fp_compare for stack register operands.
3062 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3063 instructions here. Do not emit stack register pops here. Assert
3064 that FCOMPP pops next to top stack register. Rewrite function.
3066 2017-10-17 Nathan Sidwell <nathan@acm.org>
3069 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3070 use DECL_ASSEMBLER_NAME_RAW.
3073 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3076 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3077 Wilco Dijkstra <wilco.dijkstra@arm.com>
3079 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3080 converstion to Pmode from the buf_addr.
3082 2017-10-17 Richard Biener <rguenther@suse.de>
3084 * graphite-dependences.c (scop_get_reads_and_writes): Change
3085 output parameters to references.
3087 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3089 PR 71026/tree-optimization
3090 * fold-const.c (distribute_real_division): Removed.
3091 (fold_binary_loc): Remove calls to distribute_real_divison.
3093 2017-10-17 Richard Biener <rguenther@suse.de>
3095 * graphite-scop-detection.c
3096 (scop_detection::stmt_has_simple_data_refs_p): Always use
3097 the full nest as region.
3098 (try_generate_gimple_bb): Likewise.
3099 * sese.c (scalar_evolution_in_region): Simplify now that
3100 SCEV can handle instantiation in regions.
3101 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3102 in the non-loop part of a function if requested.
3104 2017-10-17 Richard Biener <rguenther@suse.de>
3106 PR tree-optimization/82563
3107 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3109 (graphite_regenerate_ast_isl): Call it.
3110 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3112 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3114 PR tree-optimization/82549
3115 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3116 Formatting fixes. Instead of calling make_bit_field_ref with negative
3119 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3121 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3122 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3123 _mm_maskz_reduce_ss): New.
3124 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3125 __builtin_ia32_reducess_mask): Ditto..
3126 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3127 * config/i386/sse.md (reduces<mode>): Renamed to ...
3128 (reduces<mode><mask_scalar_name>): ... this.
3129 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3131 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3132 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3134 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3136 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3137 * selftest-run-tests.c (selftest::run_tests): Call
3138 selftest::unique_ptr_tests_cc_tests.
3139 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3140 * unique-ptr-tests.cc: New file.
3142 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3145 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3147 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3148 (make_hard_regno_born, make_hard_regno_dead): Update
3149 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3151 2017-10-16 Jeff Law <law@redhat.com>
3153 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3155 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3157 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3159 2017-10-16 Olivier Hainque <hainque@adacore.com>
3161 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3162 with_cpu if we were configured for an e500v2 target cpu name.
3164 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3166 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3167 * doc/invoke.texi: Document +nodsp as a valid extension for
3170 2017-10-16 Martin Liska <mliska@suse.cz>
3172 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3173 (test_set_range): Likewise.
3174 (test_range_functions): Rename to ...
3175 (test_bit_in_range): ... this.
3176 (sbitmap_c_tests): Add new test.
3178 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3180 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3182 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3183 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3185 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3187 * config/aarch64/aarch64-builtins.c
3188 (aarch64_types_quadopu_lane_qualifiers): New.
3189 (TYPES_QUADOPU_LANE): New.
3190 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3191 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3192 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3193 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3194 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3195 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3196 (Vdottype, DOTPROD): New.
3197 (sur): Add SDOT and UDOT.
3199 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3201 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3202 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3203 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3205 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3206 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3207 Enable TARGET_DOTPROD.
3208 (cortex-a75.cortex-a55): Likewise.
3209 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3211 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3213 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3214 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3216 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3218 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3219 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3220 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3221 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3222 * config/arm/types.md (neon_dot, neon_dot_q): New.
3223 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3225 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3227 * config/arm/arm.h (TARGET_DOTPROD): New.
3228 * config/arm/arm.c (arm_arch_dotprod): New.
3229 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3230 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3231 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3232 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3233 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3234 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3235 * doc/invoke.texi (armv8.2-a): Document dotprod
3237 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3239 * i386.c (ix86_vec_cost): New function.
3240 (ix86_rtx_costs): Handle vector operations better.
3241 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3242 * x86-tune-costs.h: Add new costs to all tables.
3244 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3246 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3248 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3249 divsd, sqrtss and sqrtsd
3250 * x86-tune-costs.h: Add new entries to all costs.
3251 (znver1_cost): Fix to match real instruction latencies.
3253 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3254 Michael Collison <michael.collison@arm.com>
3256 * compare-elim.c: Include emit-rtl.h.
3257 (can_merge_compare_into_arith): New function.
3258 (try_validate_parallel): Likewise.
3259 (try_merge_compare): Likewise.
3260 (try_eliminate_compare): Call the above when no previous clobber
3262 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
3265 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3269 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
3270 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
3274 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
3275 to be any operand_equal_p operands. For & (B - 1) require
3276 B to be power of 2. Recognize
3277 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
3279 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3282 * optabs.c (expand_memory_blockage): Fix call of
3283 targetm.have_memory_blockage.
3285 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3288 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
3289 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
3290 objects to extra_objs instead of overwriting it.
3292 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3294 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
3295 Use any_fp_register_operand as operand[3] predicate. Simplify
3296 equality test for operands[2] and operands[4] memory location.
3297 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
3298 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
3299 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
3300 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
3301 any_fp_register_operand as operand[1] predicate. Simplify
3302 equality test for operands[0] and operands[3] memory location.
3303 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
3304 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
3305 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
3307 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3309 * target-insns.def: Add memory_blockage.
3310 * optabs.c (expand_memory_blockage): New function.
3311 (expand_asm_memory_barrier): Rename ...
3312 (expand_asm_memory_blockage): ... to this.
3313 (expand_mem_thread_fence): Call expand_memory_blockage
3314 instead of expand_asm_memory_barrier.
3315 (expand_mem_singnal_fence): Ditto.
3316 (expand_atomic_load): Ditto.
3317 (expand_atomic_store): Ditto.
3318 * doc/md.texi (Standard Pattern Names For Generation):
3319 Document memory_blockage instruction pattern.
3321 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
3323 * config/rl78/rl78.c (rl78_emit_libcall): New function.
3324 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
3325 * config/rl78/rl78.md: New define_expand "adddi3".
3327 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3329 * cfghooks.c (verify_flow_info): Disable check that all probabilities
3332 2017-10-13 Jeff Law <law@redhat.com>
3334 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
3336 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3339 * internal-fn.c (expand_mul_overflow): If both operands have
3340 the same highpart of -1 or 0 and the topmost bit of lowpart
3341 is different, overflow is if res <= 0 rather than res < 0.
3343 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
3345 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
3346 TARGET_P9_VECTOR code for unaligned_load case.
3348 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3350 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
3352 2017-10-13 Nathan Sidwell <nathan@acm.org>
3354 * tree-core.h (tree_contains_struct): Make bool.
3355 * tree.c (tree_contains_struct): Likewise.
3356 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
3357 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
3358 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
3359 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
3360 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
3362 2017-10-13 Richard Biener <rguenther@suse.de>
3364 * graphite-isl-ast-to-gimple.c
3365 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
3366 parameters and dominance check.
3367 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
3368 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
3369 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
3370 Do not update SSA form here or do intermediate IL verification.
3371 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
3372 (graphite_initialize): Remove check on the number of loops in
3373 the function and inline into graphite_transform_loops.
3374 (graphite_finalize): Inline into graphite_transform_loops.
3375 (graphite_transform_loops): Perform SSA update and IL verification
3377 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
3379 2017-10-13 Richard Biener <rguenther@suse.de>
3381 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
3382 graphite_expression_type_precision): Avoid global constructor
3384 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
3385 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
3386 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
3387 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
3388 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
3389 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
3391 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
3394 * config/i386/i386.h (ix86_red_zone_size): New.
3395 * config/i386/i386.md (push peephole2s): Replace
3396 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
3398 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3399 Alan Hayward <alan.hayward@arm.com>
3400 David Sherwood <david.sherwood@arm.com>
3402 * combine.c (can_change_dest_mode): Reject changes in
3403 REGMODE_NATURAL_SIZE.
3405 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3406 Alan Hayward <alan.hayward@arm.com>
3407 David Sherwood <david.sherwood@arm.com>
3409 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
3410 (expand_debug_source_expr): Likewise.
3411 * combine.c (combine_simplify_rtx): Likewise.
3412 * cse.c (fold_rtx): Likewise.
3413 * fwprop.c (canonicalize_address): Likewise.
3414 * targhooks.c (default_shift_truncation_mask): Likewise.
3416 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3417 Alan Hayward <alan.hayward@arm.com>
3418 David Sherwood <david.sherwood@arm.com>
3420 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
3421 (widened_mode): Likewise.
3422 (expand_unop): Likewise.
3423 * ree.c (transform_ifelse): Likewise.
3424 (merge_def_and_ext): Likewise.
3425 (combine_reaching_defs): Likewise.
3426 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3428 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3429 Alan Hayward <alan.hayward@arm.com>
3430 David Sherwood <david.sherwood@arm.com>
3432 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
3433 * combine.c (gen_lowpart_for_combine): Likewise.
3434 * dwarf2out.c (rtl_for_decl_location): Likewise.
3435 * final.c (alter_subreg): Likewise.
3436 * rtlhooks.c (gen_lowpart_general): Likewise.
3437 (gen_lowpart_if_possible): Likewise.
3439 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3440 Alan Hayward <alan.hayward@arm.com>
3441 David Sherwood <david.sherwood@arm.com>
3443 * calls.c (expand_call): Use subreg_lowpart_offset.
3444 * cse.c (cse_insn): Likewise.
3445 * regcprop.c (copy_value): Likewise.
3446 (copyprop_hardreg_forward_1): Likewise.
3448 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3451 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
3452 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
3453 =Q constraints to +Q and into insn condition add check
3454 that operands[0] and operands[1] are equal.
3455 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
3456 =Q constraints to +Q and into insn condition add check
3457 that operands[0] is equal to either operands[1] or operands[2].
3460 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
3461 instead of handling MINUS_EXPR twice (once for each argument),
3462 canonicalize operand order and handle just once, use rtype where
3463 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
3466 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
3467 any values of __C while still being pattern recognizable as a simple
3470 2017-10-13 Richard Biener <rguenther@suse.de>
3472 PR tree-optimization/82451
3474 2017-10-02 Richard Biener <rguenther@suse.de>
3476 PR tree-optimization/82355
3477 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3478 a mapping for the enclosing loop but avoid generating one for
3480 (copy_bb_and_scalar_dependences): Remove premature codegen
3481 error on PHIs in blocks duplicated into multiple places.
3482 * graphite-scop-detection.c
3483 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3484 in the region use it as loop and nest to analyze the DR in.
3485 (try_generate_gimple_bb): Likewise.
3486 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3487 (add_loop_constraints): For blocks in a loop not in the region
3488 create a dimension with a single iteration.
3489 * sese.h (gbb_loop_at_index): Remove assert.
3491 * cfgloop.c (loop_preheader_edge): For the loop tree root
3492 return the single successor of the entry block.
3493 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3494 Reset the SCEV hashtable and niters.
3495 * graphite-scop-detection.c
3496 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
3497 assert that we only have POLYNOMIAL_CHREC that vary in loops
3498 contained in the region.
3499 (scop_detection::graphite_can_represent_expr): Adjust.
3500 (scop_detection::stmt_has_simple_data_refs_p): For loops
3501 not in the region set loop to NULL. The nest is now the
3502 entry edge to the region.
3503 (try_generate_gimple_bb): Likewise.
3504 * sese.c (scalar_evolution_in_region): Adjust for
3505 instantiate_scev change.
3506 * tree-data-ref.h (graphite_find_data_references_in_stmt):
3507 Make nest parameter the edge into the region.
3508 (create_data_ref): Likewise.
3509 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
3510 entry edge into a region and adjust instantiate_scev calls.
3511 (create_data_ref): Likewise.
3512 (graphite_find_data_references_in_stmt): Likewise.
3513 (find_data_references_in_stmt): Pass the loop preheader edge
3514 from the nest argument.
3515 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
3516 parameter the edge into the region.
3517 (instantiate_parameters): Use the loop preheader edge as entry.
3518 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
3520 (get_instantiated_value_entry): Make instantiate_below parameter
3521 the edge into the region.
3522 (instantiate_scev_name): Likewise. Adjust dominance checks,
3523 when we cannot use loop-based instantiation instantiate by
3524 walking use-def chains.
3525 (instantiate_scev_poly): Adjust.
3526 (instantiate_scev_binary): Likewise.
3527 (instantiate_scev_convert): Likewise.
3528 (instantiate_scev_not): Likewise.
3529 (instantiate_array_ref): Remove.
3530 (instantiate_scev_3): Likewise.
3531 (instantiate_scev_2): Likewise.
3532 (instantiate_scev_1): Likewise.
3533 (instantiate_scev_r): Do not blindly handle N-operand trees.
3534 Do not instantiate array-refs. Handle all constants and invariants.
3535 (instantiate_scev): Make instantiate_below parameter
3536 the edge into the region.
3537 (resolve_mixers): Use the loop preheader edge for the region
3538 parameter to instantiate_scev_r.
3539 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
3541 2017-10-13 Richard Biener <rguenther@suse.de>
3543 PR tree-optimization/82525
3544 * graphite-isl-ast-to-gimple.c
3545 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
3547 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
3548 Fail code generation when we cannot represent the isl integer.
3549 (binary_op_to_tree): Elide modulo operations that are no-ops
3550 in the type we code generate. Remove now superfluous code
3553 2017-10-13 Richard Biener <rguenther@suse.de>
3555 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
3556 (scop_detection::harmful_loop_in_region): Remove premature
3557 IV type restriction.
3558 (scop_detection::graphite_can_represent_scev): We can handle
3559 pointer IVs just fine.
3561 2017-10-13 Alan Modra <amodra@gmail.com>
3563 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
3564 "Clobbers and Scratch Registers". Add paragraph on
3565 alternative to clobbers for scratch registers and OpenBLAS
3568 2017-10-13 Alan Modra <amodra@gmail.com>
3570 * doc/extend.texi (Clobbers): Correct vax example. Delete old
3571 example of a memory input for a string of known length. Move
3572 commentary out of table. Add a number of new examples
3573 covering array memory inputs.
3575 2017-10-12 Martin Liska <mliska@suse.cz>
3577 PR tree-optimization/82493
3578 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
3579 (test_range_functions): New function.
3580 (sbitmap_c_tests): Likewise.
3581 * selftest-run-tests.c (selftest::run_tests): Run new tests.
3582 * selftest.h (sbitmap_c_tests): New function.
3584 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3586 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
3588 * config/rs6000/amo.h: Fix spacing issue.
3590 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3593 * config/i386/i386.md (*ashl<mode>3_mask_1,
3594 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
3595 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
3598 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3600 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
3601 (profile_probability): Set max_probability
3602 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
3604 * profile-count.c (profile_probability::differs_from_p): Do not
3605 rely on max_probaiblity == 10000
3607 2017-10-12 Jeff Law <law@redhat.com>
3609 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
3612 2017-10-12 Martin Sebor <msebor@redhat.com>
3616 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
3617 (handle_alias_pairs): Call it.
3618 * common.opt (-Wattribute-alias): New option.
3619 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
3620 * doc/invoke.texi (-Wattribute-alias): Document.
3622 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
3625 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
3627 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3629 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3630 (make_hard_regno_born, make_hard_regno_dead): Update
3631 bb_killed_pseudos and bb_gen_pseudos.
3633 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3635 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
3637 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
3639 * config/alpha/alpha.c (alpha_split_conditional_move):
3640 Use std::swap instead of manually swapping.
3641 (alpha_stdarg_optimize_hook): Ditto.
3642 (alpha_canonicalize_comparison): Ditto.
3644 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3646 * tree-loop-distribution.c (struct builtin_info): New struct.
3647 (struct partition): Refactor fields into struct builtin_info.
3648 (partition_free): Free struct builtin_info.
3649 (build_size_arg_loc, build_addr_arg_loc): Delete.
3650 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
3651 information from struct builtin_info.
3652 (find_single_drs): New function refactored from classify_partition.
3653 Also moved builtin validity checks to this function.
3654 (compute_access_range, alloc_builtin): New functions.
3655 (classify_builtin_st, classify_builtin_ldst): New functions.
3656 (classify_partition): Refactor code into functions find_single_drs,
3657 classify_builtin_st and classify_builtin_ldst.
3658 (distribute_loop): Don't do runtime alias check when distributing
3660 (find_seed_stmts_for_distribution): New function.
3661 (pass_loop_distribution::execute): Refactor code finding seed
3662 stmts into above function. Support distribution for the innermost
3663 two-level loop nest. Adjust dump information.
3665 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3667 * tree-loop-distribution.c: Adjust the general comment.
3668 (NUM_PARTITION_THRESHOLD): New macro.
3669 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
3670 (classify_partition): Skip builtin pattern of loop nest's inner loop.
3671 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
3672 in call to build_partition_graph.
3673 (finalize_partitions): New parameter. Make loop distribution more
3674 conservative by fusing more partitions.
3675 (distribute_loop): Don't do runtime alias check in case of loop nest
3677 (find_seed_stmts_for_distribution): New function.
3678 (prepare_perfect_loop_nest): New function.
3679 (pass_loop_distribution::execute): Refactor code finding seed stmts
3680 and loop nest into above functions. Support loop nest distribution.
3681 Adjust dump information accordingly.
3683 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3685 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
3686 and set PTYPE_SEQUENTIAL for merged partition.
3688 2017-10-12 Richard Biener <rguenther@suse.de>
3690 PR tree-optimization/69728
3692 2017-09-19 Richard Biener <rguenther@suse.de>
3694 PR tree-optimization/69728
3695 * graphite-sese-to-poly.c (schedule_error): New global.
3696 (add_loop_schedule): Handle empty domain by failing the
3698 (build_original_schedule): Handle schedule_error.
3700 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
3701 domain by returning an unchanged schedule.
3703 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3705 * genrecog.c (validate_pattern): For VEC_SELECT verify that
3706 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
3708 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
3710 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
3713 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3716 * expr.c (store_field): Don't optimize away bitsize == 0 store
3717 from CALL_EXPR with addressable return type.
3719 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
3721 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
3722 * config/rs6000/rs6000.md (sel): Delete mode attribute.
3723 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
3724 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
3725 TARGET_ISEL instead of TARGET_ISEL<sel>.
3727 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
3729 * config/rs6000/rs6000.c
3730 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
3732 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
3734 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
3736 (reg_or_cint_operand): Fix comment.
3737 (reg_or_zero_operand): New predicate.
3738 * config/rs6000/rs6000-protos.h (output_isel): Delete.
3739 * config/rs6000/rs6000.c (output_isel): Delete.
3740 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
3741 instead of reg_or_cint_operand. Output instruction directly (not via
3743 (isel_unsigned_<mode>): Ditto.
3744 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
3745 gpc_reg_operand. Add an instruction alternative for this. Output
3746 instruction directly.
3747 (*isel_reversed_unsigned_<mode>): Ditto.
3749 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3751 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
3752 (TARGET_CANONICALIZE_COMPARISON): Define.
3754 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
3757 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
3758 Check whether the dest is REG before adding REG_EQUIV note.
3760 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
3763 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3765 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3766 (make_hard_regno_born, make_hard_regno_dead): Update
3767 bb_killed_pseudos and bb_gen_pseudos.
3769 2017-10-11 Nathan Sidwell <nathan@acm.org>
3771 * incpath.h (enum incpath_kind): Name enum, prefix values.
3772 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
3773 * incpath.c (heads, tails): Use INC_MAX.
3774 (add_env_var_paths, add_standard_paths): Use incpath_kind.
3775 (merge_include_chains, split_quote_chain,
3776 register_include_chains): Update incpath_kind names.
3777 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
3778 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
3780 (add_framework_path, darwin_register_objc_includes): Likewise.
3781 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
3783 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3785 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
3786 Do not use float_operator operator predicate.
3787 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
3788 * config/i386/predicates.md (float_operator): Remove predicate.
3790 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3792 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
3793 (*jccxf_i387): Ditto.
3794 (*jcc<mode>_i387): Ditto.
3795 (*jccu<mode>_i387): Ditto.
3796 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
3797 (*jcc_*_i387 splitters): Remove.
3798 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
3799 * config/i386/i386.c (ix86_split_fp_branch): Remove.
3800 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
3803 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3805 * profile-count.h (slow_safe_scale_64bit): New function.
3806 (safe_scale_64bit): New inline.
3807 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
3808 * profile-count.c: Include wide-int.h
3809 (slow_safe_scale_64bit): New.
3811 2017-10-11 Nathan Sidwell <nathan@acm.org>
3813 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
3814 HAS_DECL_ASSEMBLER_NAME_P.
3815 * gimple-expr.c (gimple_decl_printable_name: Check
3816 HAS_DECL_ASSEMBLER_NAME_P too.
3817 * ipa-utils.h (type_in_anonymous_namespace_p): Check
3818 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
3819 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
3820 * passes.c (rest_of_decl_compilation): Check
3821 HAS_DECL_ASSEMBLER_NAME_P too.
3822 * recog.c (verify_changes): Likewise.
3823 * tree-pretty-print.c (dump_decl_name): Likewise.
3824 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
3826 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
3827 (SET_DECL_ASSEMBLER_NAME): Use it.
3828 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
3829 (COPY_DECL_ASSEMBLER_NAME): Likewise.
3830 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
3832 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3834 * config.gcc (i386, x86_64): Add extra objects.
3835 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
3836 (ix86_min_insn_size): Declare.
3837 (ix86_issue_rate): Declare.
3838 (ix86_adjust_cost): Declare.
3839 (ia32_multipass_dfa_lookahead): Declare.
3840 (ix86_macro_fusion_p): Declare.
3841 (ix86_macro_fusion_pair_p): Declare.
3842 (ix86_bd_has_dispatch): Declare.
3843 (ix86_bd_do_dispatch): Declare.
3844 (ix86_core2i7_init_hooks): Declare.
3845 (ix86_atom_sched_reorder): Declare.
3846 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
3847 (COSTS_N_BYTES): Move to x86-tune-costs.h.
3848 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
3849 (rip_relative_addr_p): Rename to ...
3850 (ix86_rip_relative_addr_p): ... this one; export.
3851 (memory_address_length): Update.
3852 (ix86_issue_rate): Move to x86-tune-sched.c.
3853 (ix86_flags_dependent): Move to x86-tune-sched.c.
3854 (ix86_agi_dependent): Move to x86-tune-sched.c.
3855 (exact_dependency_1): Move to x86-tune-sched.c.
3856 (exact_store_load_dependency): Move to x86-tune-sched.c.
3857 (ix86_adjust_cost): Move to x86-tune-sched.c.
3858 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
3859 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
3860 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
3861 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
3862 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
3863 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
3864 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
3865 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
3866 (min_insn_size): Rename to ...
3867 (ix86_min_insn_size): ... this one; export.
3868 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
3869 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
3870 (core2i7_first_cycle_multipass_backtrack): Move to
3871 x86-tune-sched-core.c.
3872 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
3873 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
3874 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
3875 (ix86_avoid_jump_mispredicts): Update.
3876 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
3877 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
3878 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
3879 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
3880 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
3881 (MAX_INSN): Move to ix86-tune-sched-bd.c.
3882 (MAX_IMM): Move to ix86-tune-sched-bd.c.
3883 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
3884 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
3885 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
3886 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
3887 (MAX_STORE): Move to ix86-tune-sched-bd.c.
3888 (BIG): Move to ix86-tune-sched-bd.c.
3889 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
3890 (enum insn_path): Move to ix86-tune-sched-bd.c.
3891 (get_mem_group): Move to ix86-tune-sched-bd.c.
3892 (is_cmp): Move to ix86-tune-sched-bd.c.
3893 (dispatch_violation): Move to ix86-tune-sched-bd.c.
3894 (is_branch): Move to ix86-tune-sched-bd.c.
3895 (is_prefetch): Move to ix86-tune-sched-bd.c.
3896 (init_window): Move to ix86-tune-sched-bd.c.
3897 (allocate_window): Move to ix86-tune-sched-bd.c.
3898 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
3899 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
3900 (process_end_window): Move to ix86-tune-sched-bd.c.
3901 (allocate_next_window): Move to ix86-tune-sched-bd.c.
3902 (find_constant): Move to ix86-tune-sched-bd.c.
3903 (get_num_immediates): Move to ix86-tune-sched-bd.c.
3904 (has_immediate): Move to ix86-tune-sched-bd.c.
3905 (get_insn_path): Move to ix86-tune-sched-bd.c.
3906 (get_insn_group): Move to ix86-tune-sched-bd.c.
3907 (count_num_restricted): Move to ix86-tune-sched-bd.c.
3908 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
3909 (add_insn_window): Move to ix86-tune-sched-bd.c.
3910 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
3911 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
3912 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
3913 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
3914 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
3915 (do_dispatch): Move to ix86-tune-sched-bd.c.
3916 (has_dispatch): Move to ix86-tune-sched-bd.c.
3917 * config/i386/t-i386: Add new object files.
3918 * config/i386/x86-tune-costs.h: New file.
3919 * config/i386/x86-tune-sched-atom.c: New file.
3920 * config/i386/x86-tune-sched-bd.c: New file.
3921 * config/i386/x86-tune-sched-core.c: New file.
3922 * config/i386/x86-tune-sched.c: New file.
3924 2017-10-11 Liu Hao <lh_mouse@126.com>
3926 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
3927 the generic version below instead.
3928 (should_colorize): Recognize Windows consoles as terminals
3930 * pretty-print.c [__MINGW32__] (write_all): New function.
3931 [__MINGW32__] (find_esc_head): Likewise.
3932 [__MINGW32__] (find_esc_terminator): Likewise.
3933 [__MINGW32__] (eat_esc_sequence): Likewise.
3934 [__MINGW32__] (mingw_ansi_fputs): New function that handles
3936 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
3939 2017-10-11 Richard Biener <rguenther@suse.de>
3941 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
3942 Properly call analyze_scalar_evolution with the loop of the stmt.
3944 2017-10-11 Richard Biener <rguenther@suse.de>
3946 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
3947 * tree-core.h (tree_base): Add chrec_var union member.
3948 * tree.h (CHREC_VAR): Remove.
3949 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
3950 * tree-chrec.h (build_polynomial_chrec): Adjust.
3951 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
3952 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
3954 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
3956 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
3957 * match.pd: ... here.
3958 ((T) X == (T) Y): Relax condition.
3960 2017-10-11 Bin Cheng <bin.cheng@arm.com>
3962 PR tree-optimization/82472
3963 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
3965 (break_alias_scc_partitions): Update postorder number.
3967 2017-10-11 Martin Liska <mliska@suse.cz>
3970 * opts.c (parse_no_sanitize_attribute): Do not use error_value
3972 * opts.h (parse_no_sanitize_attribute): Remove last argument.
3974 2017-10-11 Martin Liska <mliska@suse.cz>
3976 * print-rtl.c (print_insn): Move declaration of idbuf
3977 to same scope as name.
3979 2017-10-11 Martin Liska <mliska@suse.cz>
3984 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
3987 2017-10-11 Martin Liska <mliska@suse.cz>
3990 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
3993 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
3995 * config.gcc (default_gnu_indirect_function): Default to yes for
3996 arm*-*-linux* with glibc.
3998 2017-10-11 Richard Biener <rguenther@suse.de>
4000 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4001 default-defs and types we do not want to analyze.
4002 (interpret_loop_phi): Replace unreachable code with an assert.
4003 (compute_scalar_evolution_in_loop): Remove and inline ...
4004 (analyze_scalar_evolution_1): ... here, replacing condition with
4005 what makes the intent clearer. Remove handling of cases
4006 get_scalar_evolution now handles.
4008 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4010 PR rtl-optimization/81434
4011 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4012 comment for main loop. In sched_group_found if, also add checks for
4013 pass and min_cost_group.
4015 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4017 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4018 (rs6000_insn_cost): New function.
4019 * config/rs6000/rs6000.md (cost): New attribute.
4021 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4022 H.J. Lu <hongjiu.lu@intel.com>
4026 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4027 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4028 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4029 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4030 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4031 OPTION_MASK_ISA_MMX and some other bit set require both
4032 mmx and the ISAs without the mmx bit.
4033 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4034 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4035 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4036 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4037 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4038 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4039 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4040 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4041 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4042 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4043 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4044 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4045 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4046 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4047 Add OPTION_MASK_ISA_MMX.
4049 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4051 * config.gcc (armv7*-*-freebsd*): New target.
4052 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4054 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4056 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4057 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4060 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4062 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4063 a template parameter.
4064 (WIDE_INT_REF_FOR): Update accordingly.
4065 * tree.h (wi::int_traits <const_tree>): Delete.
4066 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4067 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4068 (wi::tree_to_wide_ref): New typedef.
4069 (wi::to_wide): New function.
4070 * calls.c (get_size_range): Use wi::to_wide when operating on
4072 * cgraph.c (cgraph_node::create_thunk): Likewise.
4073 * config/i386/i386.c (ix86_data_alignment): Likewise.
4074 (ix86_local_alignment): Likewise.
4075 * dbxout.c (stabstr_O): Likewise.
4076 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4077 * expr.c (const_vector_from_tree): Likewise.
4078 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4079 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4080 (fold_negate_expr_1, int_const_binop_1, const_binop)
4081 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4082 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4083 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4084 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4085 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4086 (fold_not_const, round_up_loc): Likewise.
4087 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4088 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4089 (alloca_call_type): Likewise.
4090 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4091 * godump.c (go_output_typedef): Likewise.
4092 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4093 * internal-fn.c (get_min_precision): Likewise.
4094 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4095 * ipa-polymorphic-call.c
4096 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4097 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4098 (ipa_modify_call_arguments): Likewise.
4099 * match.pd: Likewise.
4100 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4101 * print-tree.c (print_node_brief, print_node): Likewise.
4102 * stmt.c (expand_case): Likewise.
4103 * stor-layout.c (layout_type): Likewise.
4104 * tree-affine.c (tree_to_aff_combination): Likewise.
4105 * tree-cfg.c (group_case_labels_stmt): Likewise.
4106 * tree-data-ref.c (dr_analyze_indices): Likewise.
4107 (prune_runtime_alias_test_list): Likewise.
4108 * tree-dump.c (dequeue_and_dump): Likewise.
4109 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4110 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4111 * tree-pretty-print.c (dump_generic_node): Likewise.
4112 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4113 (simple_iv_with_niters): Likewise.
4114 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4115 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4116 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4117 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4118 (refine_value_range_using_guard, number_of_iterations_ne_max)
4119 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4120 (get_cst_init_from_scev, record_nonwrapping_iv)
4121 (scev_var_range_cant_overflow): Likewise.
4122 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4123 * tree-ssa-pre.c (compute_avail): Likewise.
4124 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4125 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4126 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4127 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4128 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4129 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4130 (dump_case_nodes, try_switch_expansion): Likewise.
4131 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4132 (vect_do_peeling): Likewise.
4133 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4134 * tree-vect-stmts.c (vectorizable_load): Likewise.
4135 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4136 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4137 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4138 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4139 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4140 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4141 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4142 (evrp_dom_walker::before_dom_children): Likewise.
4143 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4144 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4145 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4146 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4147 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4148 (get_range_pos_neg): Likewise.
4149 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4150 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4151 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4152 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4153 * config/avr/avr.c (avr_fold_builtin): Likewise.
4154 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4155 * config/msp430/msp430.c (msp430_attr): Likewise.
4156 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4157 * config/powerpcspe/powerpcspe-c.c
4158 (altivec_resolve_overloaded_builtin): Likewise.
4159 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4160 (rs6000_expand_ternop_builtin): Likewise.
4161 * config/rs6000/rs6000-c.c
4162 (altivec_resolve_overloaded_builtin): Likewise.
4163 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4164 (rs6000_expand_ternop_builtin): Likewise.
4165 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4167 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4169 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4170 when copying loop nest with only one inner loop.
4172 2017-10-10 Richard Biener <rguenther@suse.de>
4174 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4175 blocks if SCEV is active.
4176 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4178 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4179 (scev_initialize): Assert we are not yet initialized.
4181 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4183 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4184 inner loop's exit stmt by making it always exit the loop, otherwise
4185 we would generate an infinite empty loop.
4187 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4189 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4190 renaming variables in new preheader if it's deleted.
4192 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4194 * tree-loop-distribution.c (struct partition): Remove unused field
4195 loops of the structure.
4196 (partition_alloc, partition_free): Ditto.
4197 (build_rdg_partition_for_vertex): Ditto.
4199 2017-10-09 Jeff Law <law@redhat.com>
4201 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4202 return type to match prototype and documentation.
4204 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4206 * config/rs6000/rs6000.c (processor_costs): Move to ...
4207 * config/rs6000/rs6000.h: ... here.
4208 (rs6000_cost): Declare.
4210 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4212 * except.c (setjmp_fn): New global variable.
4213 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4214 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4215 if DONT_USE_BUILTIN_SETJMP is defined.
4217 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4219 * target.def (insn_cost): New hook.
4220 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4221 * doc/tm.texi: Regenerate.
4222 * rtlanal.c (insn_cost): Use the new hook.
4224 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4226 * combine.c (combine_validate_cost): Compute the new insn_cost,
4227 not just pattern_cost.
4228 (try_combine): Adjust comment.
4230 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4232 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4234 * combine.c (uid_insn_cost): Adjust comment.
4235 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4237 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4238 * dse.c (find_shift_sequence): Ditto.
4239 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4240 (bb_valid_for_noce_process_p): Use pattern_cost.
4241 * rtl.h (insn_rtx_cost): Delete.
4242 (pattern_cost): New prototype.
4243 (insn_cost): New prototype.
4244 * rtlanal.c (insn_rtx_cost): Rename to...
4245 (pattern_cost): ... this.
4248 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4250 * config/i386/i386.md (*jcc_2): Remove insn pattern.
4251 (*jcc<mode>_0_r_i387): Ditto.
4252 (*jccxf_r_i387): Ditto.
4253 (*jcc<mode>_r_i387): Ditto.
4254 (*jccu<mode>_r_i387): Ditto.
4255 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
4256 (*jcc): Rename from *jcc_1.
4258 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4260 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
4261 deferred rescans after the lvx/stvx recombination pre-pass.
4263 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
4265 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
4266 memory operation instruction support.
4267 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
4268 (rs6000-ibm-aix[789]*): Likewise.
4269 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
4270 Document new functions.
4272 2017-10-09 Richard Biener <rguenther@suse.de>
4274 PR tree-optimization/82397
4275 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
4276 equality only for semantically equal trees.
4278 2017-10-09 Richard Biener <rguenther@suse.de>
4280 PR tree-optimization/82449
4281 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
4282 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
4283 allow constant addresses.
4284 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
4287 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4289 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
4292 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4295 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
4298 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4301 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
4303 2017-10-09 Jakub Jelinek <jakub@redhat.com>
4306 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
4307 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
4309 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
4311 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
4312 (WI_BINARY_PREDICATE_RESULT): Likewise.
4313 (wi::binary_traits::operator_result): New type.
4314 (wi::binary_traits::predicate_result): Likewise.
4315 (generic_wide_int::operator~, unary generic_wide_int::operator-)
4316 (generic_wide_int::operator==, generic_wide_int::operator!=)
4317 (generic_wide_int::operator&, generic_wide_int::and_not)
4318 (generic_wide_int::operator|, generic_wide_int::or_not)
4319 (generic_wide_int::operator^, generic_wide_int::operator+
4320 (binary generic_wide_int::operator-, generic_wide_int::operator*):
4322 (operator~, unary operator-, operator==, operator!=, operator&)
4323 (operator|, operator^, operator+, binary operator-, operator*): New
4325 * expr.c (get_inner_reference): Use wi::bit_and_not.
4326 * fold-const.c (fold_binary_loc): Likewise.
4327 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
4328 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
4329 (bit_value_binop): Likewise.
4330 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
4331 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
4332 (extract_range_from_binary_expr_1): Likewise.
4333 (masked_increment): Likewise.
4334 (simplify_bit_ops_using_ranges): Likewise.
4336 2017-10-09 Martin Jambor <mjambor@suse.cz>
4339 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
4340 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
4341 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
4343 (hsa_fixup_mov_insn_type): New function.
4344 (hsa_op_with_type::get_in_type): Use it.
4345 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
4346 immediates in an assert.
4347 (hsa_op_with_type::extend_int_to_32bit): New method.
4348 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
4349 types. Convert to dest type if necessary.
4350 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
4351 (reg_for_gimple_ssa): Pass false as min32int to
4352 hsa_type_for_scalar_tree_type.
4353 (gen_hsa_addr): Fixup type when creating addresable temporary.
4354 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
4355 (gen_hsa_unary_operation): Extend operands and convert to dest type if
4356 necessary. Call hsa_fixup_mov_insn_type.
4357 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
4358 extend operands and convert to dest type if necessary.
4359 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
4360 to dest type if necessary.
4361 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
4362 if conversion nt necessary and size matches.
4363 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
4364 to dest type if necessary.
4365 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
4366 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
4368 (gen_hsa_clrsb): Likewise.
4369 (gen_hsa_ffs): Likewise.
4370 (gen_hsa_divmod): Extend operands and convert to dest type if
4372 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
4374 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4376 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
4377 Remove empty default arguments. Use a brace block as output
4379 (conditional return): Ditto.
4381 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
4382 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
4384 (group_ending_nop): Ditto.
4385 (doloop_end): Ditto.
4386 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
4387 (splitters for those): Ditto.
4389 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4391 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
4392 a conditional jump (and the compare for it) so that pc_rtx is the
4394 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
4395 for the deleted and renamed ctr<mode>_internal[234] patterns.
4396 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
4397 Delete second conditional return pattern.
4398 (ctr<mode>_internal2): Delete this second bdnz pattern.
4399 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
4400 (ctr<mode>_internal4): Delete this second bdz pattern.
4402 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4404 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
4405 (always_initialized_rtx_for_ssa_name_p): New predicate.
4406 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
4407 (finish_out_of_ssa): Free new field of SA.
4408 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
4409 * tree-ssa-coalesce.c: Include tree-ssa.h.
4410 (get_parm_default_def_partitions): Remove extern keyword.
4411 (get_undefined_value_partitions): New function.
4412 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
4413 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
4415 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
4417 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4419 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
4421 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4423 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
4424 for targets that preffer 128bit.
4426 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4428 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
4430 2017-10-08 Olivier Hainque <hainque@adacore.com>
4432 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
4433 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
4435 (thumb_set_return_address): Likewise.
4437 2017-10-08 Olivier Hainque <hainque@adacore.com>
4439 * common/config/arm/arm-common.c (arm_except_unwind_info):
4440 Handle DWARF2_UNWIND_INFO.
4442 2017-10-07 Michael Collison <michael.collison@arm.com>
4444 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
4447 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
4449 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
4450 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
4451 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
4452 defined, force the creation of a new block for a dispatch label.
4454 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
4456 * invoke.texi (Wsuggest-attribute=cold): Document.
4457 * common.opt (Wsuggest-attribute=cold): New
4458 * ipa-pure-const.c (warn_function_cold): New function.
4459 * predict.c (compute_function_frequency): Use it.
4460 * predict.h (warn_function_cold): Declare.
4462 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
4464 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
4467 2017-10-06 Martin Liska <mliska@suse.cz>
4469 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
4470 keyword for member functions.
4471 (struct sanopt_tree_couple): New struct.
4472 (struct sanopt_tree_couple_hash): New function.
4473 (struct sanopt_ctx): Add new hash_map.
4474 (has_dominating_ubsan_ptr_check): New function.
4475 (record_ubsan_ptr_check_stmt): Likewise.
4476 (maybe_optimize_ubsan_ptr_ifn): Likewise.
4477 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
4478 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
4480 2017-10-06 Sudakshina Das <sudi.das@arm.com>
4483 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
4484 aarch64_simd_valid_immediate on CONST_VECTORs.
4485 (aarch64_reg_or_bic_imm): Likewise.
4487 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
4489 PR rtl-optimization/82396
4490 * haifa-sched.c (ready_sort_real): Disable qsort checking.
4492 2017-10-06 Sebastian Pop <sebpop@gmail.com>
4494 * graphite-dependences.c (scop_get_reads): Move code to...
4495 (scop_get_must_writes): Move code to...
4496 (scop_get_may_writes): Move code to...
4497 (scop_get_reads_and_writes): ... here.
4498 (scop_get_dependences): Call scop_get_reads_and_writes.
4500 2017-10-06 Jakub Jelinek <jakub@redhat.com>
4502 PR tree-optimization/82434
4503 * fold-const.h (can_native_encode_type_p,
4504 can_native_encode_string_p): Remove.
4505 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
4506 don't encode anything, just return what would be otherwise returned.
4507 (native_encode_fixed, native_encode_complex, native_encode_vector):
4509 (native_encode_string): Likewise. Inline by hand
4510 can_native_encode_string_p.
4511 (can_native_encode_type_p): Remove.
4512 (can_native_encode_string_p): Remove.
4513 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
4514 STRING_CSTs using can_native_encode_string_p, test all
4515 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
4516 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
4517 argument from native_encode_expr.
4518 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
4519 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
4522 2017-10-06 Richard Biener <rguenther@suse.de>
4524 PR tree-optimization/82397
4525 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
4526 operand_equal_p but rely on data_ref_compare_tree for detecting
4528 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
4529 to match up with dr_group_sort_cmp.
4531 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4534 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
4536 * config/s390/s390-builtin-types.def: Regenerate.
4538 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4541 * config/s390/s390-builtin-types.def: Regenerate.
4542 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
4543 Change flag from B_VXE to B_VX.
4544 (s390_vec_min_dbl): Remove B_VXE flag.
4546 2017-10-06 Richard Biener <rguenther@suse.de>
4548 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
4549 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
4550 translate_isl_ast_to_gimple::is_valid_rename,
4551 translate_isl_ast_to_gimple::get_rename,
4552 translate_isl_ast_to_gimple::get_def_bb_for_const,
4553 translate_isl_ast_to_gimple::get_new_name,
4554 translate_isl_ast_to_gimple::collect_all_ssa_names,
4555 translate_isl_ast_to_gimple::copy_loop_phi_args,
4556 translate_isl_ast_to_gimple::collect_all_ssa_names,
4557 translate_isl_ast_to_gimple::copy_loop_phi_args,
4558 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
4559 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
4560 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
4561 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
4562 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
4563 translate_isl_ast_to_gimple::copy_cond_phi_args,
4564 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
4565 translate_isl_ast_to_gimple::edge_for_new_close_phis,
4566 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
4567 translate_isl_ast_to_gimple::rename_uses,
4568 translate_isl_ast_to_gimple::rename_all_uses): Remove.
4569 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
4570 (set_rename_for_each_def): Likewise.
4571 (graphite_copy_stmts_from_block): Handle debug stmt resetting
4572 here. Handle rewriting SCEV analyzable uses here.
4573 (copy_bb_and_scalar_dependences): Generate code for PHI
4575 (graphite_regenerate_ast_isl): Adjust.
4576 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
4577 (add_write, add_read): New functions.
4578 (build_cross_bb_scalars_def): Use it and simplify.
4579 (build_cross_bb_scalars_use): Likewise.
4580 (graphite_find_cross_bb_scalar_vars): Inline into...
4581 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
4582 simulating out-of-SSA. Compute liveout and add dependencies.
4583 (build_scops): Force an empty entry block.
4584 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
4586 (sese_build_liveouts): Declare.
4587 (sese_trivially_empty_bb_p): Likewise.
4588 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
4589 compute liveout and debug_liveout.
4590 (sese_bad_liveouts_use): Remove.
4591 (sese_reset_debug_liveouts_bb): Likewise.
4592 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
4593 (sese_build_liveouts): Build liveout and debug_liveout and store
4595 (new_sese_info): Adjust.
4596 (free_sese_info): Likewise.
4597 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
4598 do not build liveout here.
4599 (move_sese_in_condition): Adjust region entry.
4600 (scev_analyzable_p): Match up with chrec_apply requirements.
4601 (sese_trivially_empty_bb_p): New.
4602 * tree-into-ssa.c (get_reaching_def): Properly support generating
4603 default-defs for incremental rewrite of anonymous names.
4605 2017-10-06 Richard Biener <rguenther@suse.de>
4607 * graphite-sese-to-poly.c (extract_affine): For casts increasing
4608 precision do not perform modulo reduction.
4610 2017-10-06 Richard Biener <rguenther@suse.de>
4612 PR tree-optimization/82436
4613 * tree-vect-slp.c (vect_supported_load_permutation_p): More
4614 conservatively choose the vectorization factor when checking
4615 whether we can perform the required load permutation.
4616 (vect_transform_slp_perm_load): Assert when we may not fail.
4618 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
4620 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
4621 message for incompatible -msdata=* and -mcall-* options.
4623 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4625 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
4626 rate for post-reload scheduling.
4628 2017-10-05 Tamar Christina <tamar.christina@arm.com>
4630 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
4632 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4634 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
4635 to improve monte carlo in scimark.
4637 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4639 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
4640 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
4641 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
4642 pentium4_cost, nocona_cost): Set reassociation width to 1.
4643 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
4644 width to 2 for fp operations and 1 otherwise.
4645 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
4647 (atom_cost): Set reassociation width to 2.
4648 (slm_cost, generic_cost): Set fp reassociation width
4649 to 2 and 1 otherwise.
4650 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
4651 (core_cost): Set fp reassociation width to 4 and vector to 2.
4652 (ix86_reassociation_width): Rewrite using cost table; special case
4653 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
4654 and TARGET_AVX128_OPTIMAL.
4655 * config/i386/i386.h (processor_costs): Add
4656 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
4657 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
4658 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
4659 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
4660 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
4661 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
4663 2017-10-05 Nathan Sidwell <nathan@acm.org>
4665 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
4667 2017-10-05 Tamar Christina <tamar.christina@arm.com>
4669 * config/arm/arm.c (arm_test_fpu_data): New.
4670 (arm_run_selftests): Call arm_test_fpu_data.
4672 2017-10-04 Nathan Sidwell <nathan@acm.org>
4674 * toplev.c (toplev::main): Remove excess parens on pretty_printer
4676 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
4678 2017-10-04 Sudakshina Das <sudi.das@arm.com>
4680 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
4681 check type for aarch64_simd_valid_immediate.
4682 (aarch64_output_simd_mov_immediate): Update prototype.
4683 (aarch64_simd_valid_immediate): Update prototype.
4684 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
4685 support for ORR-immediate.
4686 (and<mode>3): modified pattern to add support for BIC-immediate.
4687 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
4688 now checks for valid immediate for BIC and ORR based on new enum
4690 (aarch64_output_simd_mov_immediate): Function now used to output
4691 BIC/ORR imm as well based on new enum argument.
4692 * config/aarch64/constraints.md (Do): New vector immediate constraint.
4694 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
4695 (aarch64_reg_or_bic_imm): Likewise.
4697 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4699 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
4700 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
4702 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
4706 PR rtl-optimization/82396
4707 * haifa-sched.c (autopref_multipass_init): Simplify
4709 (autopref_rank_data): Simplify sort order.
4710 * sched-int.h (autopref_multipass_data_): Remove
4711 multi_mem_insn_p, min_offset and max_offset.
4713 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4715 * doc/sourcebuild.texi: Document vect_peeling_profitable.
4717 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4719 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
4722 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4724 * doc/sourcebuild.texi: Document vect_long_mult.
4726 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
4728 PR tree-optimization/82413
4729 * fold-const.c (build_range_check): Use widest_int when comparing
4730 the maximum ETYPE value with HIGH.
4732 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
4734 PR rtl-optimization/82396
4735 * haifa-sched.c (autopref_multipass_init): Simplify
4737 (autopref_rank_data): Simplify sort order.
4738 * sched-int.h (autopref_multipass_data_): Remove
4739 multi_mem_insn_p, min_offset and max_offset.
4741 2017-10-04 Jakub Jelinek <jakub@redhat.com>
4743 PR tree-optimization/82381
4744 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
4745 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
4748 PR tree-optimization/82374
4749 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
4750 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
4751 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
4752 current_function_decl to the new decl.
4754 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
4756 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
4757 helper macro for IEEE float128 hardware built-in functions.
4758 (SQRTF128_ODD): Add built-in functions with the round-to-odd
4760 (TRUNCF128_ODD): Likewise.
4761 (ADDF128_ODD): Likewise.
4762 (SUBF128_ODD): Likewise.
4763 (MULF128_ODD): Likewise.
4764 (DIVF128_ODD): Likewise.
4765 (FMAF128_ODD): Likewise.
4766 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
4767 UNSPEC_TRUNC_ROUND_TO_ODD.
4768 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
4769 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
4770 floating point round to odd instructions.
4771 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
4772 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
4773 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
4774 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
4775 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
4776 (trunc<mode>sf2_hw): Change the truncate with round to odd
4777 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
4778 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
4779 to odd hardware instructions.
4780 (sub<mode>3_odd): Likewise.
4781 (mul<mode>3_odd): Likewise.
4782 (div<mode>3_odd): Likewise.
4783 (sqrt<mode>2_odd): Likewise.
4784 (fma<mode>4_odd): Likewise.
4785 (fms<mode>4_odd): Likewise.
4786 (nfma<mode>4_odd): Likewise.
4787 (nfms<mode>4_odd): Likewise.
4788 (trunc<mode>df2_odd): Change the truncate with round to odd
4789 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
4791 * doc/extend.texi (PowerPC built-in functions): Update documentation
4792 for existing IEEE float128-bit built-in functions. Add built-in
4793 functions that generate the IEEE 128-bit floating point round to
4796 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
4798 PR rtl-optimization/77729
4799 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
4800 to (X&(C1&~C2))|C2 transformations.
4802 2017-10-03 Martin Jambor <mjambor@suse.cz>
4804 PR tree-optimization/82363
4805 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
4806 mismatch, mark lacc written regardless of racc.
4808 2017-10-03 Jakub Jelinek <jakub@redhat.com>
4810 PR tree-optimization/82381
4811 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
4812 stmt_to_insert nor wheather SSA_NAMEs are default defs.
4813 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
4814 fallthrough into reassoc_stmt_dominates_stmt_p.
4817 * combine.c (combine_instructions): Don't combine in unreachable
4820 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4823 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
4824 function to not use the have_cpu variable. Do not set cpu_index,
4825 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
4827 (rs6000_valid_attribute_p): Remove duplicate initializations of
4828 old_optimize and func_optimize.
4829 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
4830 (rs6000_activate_target_options): Make global.
4831 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
4834 2017-10-02 Jakub Jelinek <jakub@redhat.com>
4836 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
4837 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
4838 Set *poffset to 0 and *psize and *pmax_size to -1 if
4839 *poffset + *psize overflows in HOST_WIDE_INT.
4841 PR tree-optimization/82387
4842 PR tree-optimization/82388
4843 PR tree-optimization/82389
4844 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
4845 instead of live_bytes non-NULL.
4847 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
4850 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
4853 2017-10-02 Richard Biener <rguenther@suse.de>
4855 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
4856 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
4857 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
4859 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
4861 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
4862 requested precision matches the type's.
4863 * calls.c (alloc_max_size): Calculate the new candidate size as
4864 a widest_int and use wi::to_widest when comparing it with the
4865 current candidate size.
4866 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
4867 zero rather than integer_zero_node.
4868 * match.pd: Check for a no-op conversion before using wi::add
4869 rather than after. Use tree_to_uhwi when summing small shift
4870 counts into an unsigned int.
4872 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
4873 Alan Hayward <alan.hayward@arm.com>
4874 David Sherwood <david.sherwood@arm.com>
4877 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
4878 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
4879 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
4880 POINTER_AND_FP_REGS.
4882 2017-10-02 Richard Biener <rguenther@suse.de>
4884 PR tree-optimization/82355
4885 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4886 a mapping for the enclosing loop but avoid generating one for
4888 (copy_bb_and_scalar_dependences): Remove premature codegen
4889 error on PHIs in blocks duplicated into multiple places.
4890 * graphite-scop-detection.c
4891 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4892 in the region use it as loop and nest to analyze the DR in.
4893 (try_generate_gimple_bb): Likewise.
4894 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4895 (add_loop_constraints): For blocks in a loop not in the region
4896 create a dimension with a single iteration.
4897 * sese.h (gbb_loop_at_index): Remove assert.
4899 2017-10-01 Kevin Buettner <kevinb@redhat.com>
4901 * omp-expand.c (adjust_context_scope): New function.
4902 (expand_parallel_call): Call adjust_context_scope.
4904 2017-10-01 Jeff Law <law@redhat.com>
4906 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
4907 dom_opt_dom_walker class with direct access to private members.
4908 Add comments. Call test_for_singularity.
4909 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
4910 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
4911 m_dummy_cond anymore.
4912 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
4914 (pass_dominator:execute): Build the dummy_cond here and pass it
4915 to the dom_opt_dom_walker ctor.
4916 (test_for_singularity): New function.
4918 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
4919 Maya Rashish <coypu@sdf.org>
4921 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
4922 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
4923 (alpha*-*-netbsd*) Use nbsd_tm_file.
4924 (arm*-*-netbsdelf*) Likewise.
4925 (i[34567]86-*-netbsdelf*) Likewise.
4926 (x86_64-*-netbsd*) Likewise.
4927 (mips*-*-netbsd*) Likewise.
4928 (powerpc-*-netbsd*) Likewise.
4929 (sh*-*-netbsd*) Likewise.
4930 (sparc-*-netbsdelf*) Likewise.
4931 (sparc64-*-netbsd*) Likewise.
4932 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
4934 (vax-*-netbsdelf*) Likewise.
4935 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
4936 (UINT_FAST8_TYPE) Likewise.
4937 (INT_FAST16_TYPE) Check CHAR_FAST16.
4938 (UINT_FAST16_TYPE) Likewise.
4940 2017-09-30 Jakub Jelinek <jakub@redhat.com>
4943 * config/i386/i386.md
4944 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
4945 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
4946 *divmodsi4_zext_2): New define_insn_and_split.
4947 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
4948 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
4949 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
4950 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
4951 New define_insn_and_split.
4952 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
4953 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
4954 operands[1] having DImode when mode is SImode.
4956 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
4957 always SImode for DIV and MOD in REG_EQUAL notes.
4959 2017-09-29 Yury Gribov <tetra2005@gmail.com>
4962 * match.pd: Fix handling of NaNs in pattern.
4964 2017-09-29 Jeff Law <law@redhat.com>
4966 * sbitmap.c (bitmap_bit_in_range_p): New function.
4967 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
4968 * tree-ssa-dse.c (live_bytes_read): New function.
4969 (dse_classify_store): Ignore reads of dead bytes.
4971 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
4972 typos and whitespace errors.
4973 * config/i386/predicates.md (address_no_seg_operand): Likewise.
4974 * config/s390/s390.c (s390_emit_prologue): Likewise.
4976 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
4979 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
4980 with a symbol for LRA.
4982 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
4984 PR rtl-optimization/82338
4985 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
4987 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
4989 * genmodes.c (calc_wider_mode): Suppress qsort macro.
4990 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
4991 (qsort_chk): Declare.
4992 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
4993 (qsort_chk): New function.
4995 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4997 PR tree-optimization/82337
4998 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
4999 phi definition if the PHI result appears in an abnormal PHI.
5000 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5001 basis appears in an abnormal PHI.
5003 2017-09-29 Richard Biener <rguenther@suse.de>
5005 * graphite-isl-ast-to-gimple.c
5006 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5007 (binary_op_to_tree): Use it.
5008 (get_rename_from_scev): Likewise.
5009 (copy_loop_phi_nodes): Likewise.
5010 (copy_bb_and_scalar_dependences): Likewise.
5011 (translate_pending_phi_nodes): Likewise.
5013 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5016 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5017 for movabsq $(i32 << shift), r64.
5019 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5021 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5022 index when encoding %esp as %rsp to avoid 0x67 prefix.
5024 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5026 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5027 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5029 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5031 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5032 Extensions with more than 16 double VFP registers.
5033 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5034 to_clear_mask and all code related to it. Replace the remaining
5035 entry by a sbitmap and adapt code accordingly.
5037 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5039 * brig-builtins.def: Change pure attributes to const.
5041 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5043 * config.gcc (default_gnu_indirect_function): Default to yes for
5044 sparc*-*-linux* with glibc.
5046 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5048 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5049 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5050 when creating .init_array and .fini_array sections with priority
5053 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5056 * config/aarch64/aarch64.c
5057 (aarch64_builtin_support_vector_misalignment): Always return false
5058 when misalignment is unknown.
5060 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5062 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5063 this function to return false if the definition used by the swap
5064 instruction is artificial, or if the memory address from which the
5065 constant value is loaded is not represented by a base address held
5066 in a register or if the base address register is a frame or stack
5067 pointer. Additionally, return false if the base address of the
5068 loaded constant is a SYMBOL_REF but is not considered to be a
5070 (replace_swapped_load_constant): New function.
5071 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5072 loaded constant vector with a load of a swapped constant vector.
5074 2017-09-27 Carl Love <cel@us.ibm.com>
5076 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5077 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5078 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5081 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5083 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5084 first, always call autopref_rank_data otherwise.
5086 2017-09-27 Richard Biener <rguenther@suse.de>
5088 * graphite-scop-detection.c (find_scop_parameters): Move
5089 loop bound handling ...
5090 (gather_bbs::before_dom_children): ... here, avoiding the need
5091 to build scop_info->loop_nest.
5092 (record_loop_in_sese): Remove.
5093 * sese.h (sese_info_t::loop_nest): Remove.
5094 * sese.c (new_sese_info): Do not allocate loop_nest.
5095 (free_sese_info): Do not free loop_nest.
5097 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5100 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5101 lhs from calls if the lhs has addressable type.
5103 2017-09-27 Richard Biener <rguenther@suse.de>
5105 * graphite.h (scop::max_alias_set): New member.
5106 * graphite-scop-detection.c: Remove references to non-existing
5107 --param in comments.
5108 (build_alias_sets): Record the maximum alias set used for drs.
5109 (build_scops): Support zero as unlimited for
5110 --param graphite-max-arrays-per-scop.
5111 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5113 (build_poly_sr_1): ... here. Compute alias set based on the
5114 maximum alias set used for drs rather than
5115 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5117 2017-09-27 Richard Biener <rguenther@suse.de>
5119 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5120 --param loop-block-tile-size=0 to disable tiling.
5122 2017-09-27 Richard Biener <rguenther@suse.de>
5124 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5125 (graphite-max-nb-scop-params): Document special value zero.
5126 * domwalk.h (dom_walker::STOP): New symbolical constant.
5127 (dom_walker::dom_walker): Add optional parameter for bb to
5129 (dom_walker::~dom_walker): Declare.
5130 (dom_walker::before_dom_children): Document STOP return value.
5131 (dom_walker::m_user_bb_to_rpo): New member.
5132 (dom_walker::m_bb_to_rpo): Likewise.
5133 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5134 mapping here if not provided by the user.
5135 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5136 provided by the user.
5137 (dom_walker::STOP): Define.
5138 (dom_walker::walk): Do not compute bb to RPO mapping here.
5139 Support STOP return value from before_dom_children to stop
5141 * graphite-optimize-isl.c (optimize_isl): If the schedule
5142 is the same still generate code if -fgraphite-identity
5143 or -floop-parallelize-all are given.
5144 * graphite-scop-detection.c: Include cfganal.h.
5145 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5147 (gather_bbs::before_dom_children): Return STOP for BBs
5149 (build_scops): Compute bb to RPO mapping and pass it to
5150 the domwalk. Treat --param graphite-max-nb-scop-params=0
5151 as not limiting the number of params.
5152 * graphite.c (graphite_initialize): Remove limit on the
5153 number of basic-blocks in a function.
5154 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5155 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5156 default value of 10.
5158 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5160 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5161 Adjust code to eliminate needing to do the shift right 32-bits
5162 operation after XSCVDPSPN.
5164 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5166 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5167 ((X / Y) != 0 -> X >= Y): Likewise.
5169 2017-09-26 Carl Love <cel@us.ibm.com>
5171 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5172 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5173 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5174 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5175 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5176 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5177 definitions and overloading.
5178 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5179 statement for P9V_BUILTIN_XST_LEN_R.
5180 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5181 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5182 define_expand and define_insn for the instructions and builtins.
5183 * doc/extend.texi: Update the built-in documentation file for the new
5185 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5186 define_insn for the instructions
5188 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5191 * gcc/config/netbsd-protos.h: New file.
5192 * gcc/config/netbsd.c: New file.
5193 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5194 * gcc/config/t-netbsd: New file.
5195 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5196 (tmake_file) Add t-netbsd.
5197 (extra_objs) Add netbsd.o.
5199 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5203 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5205 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5207 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5208 sign extension from a vector register to a GPR by doing a 32-bit
5209 direct move and then an EXTSW.
5210 (extendsi<mode>2 splitter): Likewise.
5211 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5212 right or vector extract after doing XSCVDPSPN. Use
5213 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5215 (movdi_from_sf_zero_ext): Likewise.
5216 (reload_gpr_from_vsxsf): Likewise.
5217 (p8_mfvsrd_4_disf): Delete, no longer used.
5218 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5219 then needing to move the SFmode to a GPR to use the XSCVDPSP
5220 instruction instead of FRSP and XSCVDPSPN.
5221 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5222 it is adjacent to the other XSCVSPDP insns.
5223 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5224 SFmode to be in traditional Altivec registers.
5225 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5226 (vsx_xscvspdpn): Likewise.
5227 (vsx_xscvspdpn_scalar): Likewise.
5229 2017-09-26 Martin Jambor <mjambor@suse.cz>
5231 * tree-sra.c (compare_access_positions): Put integral types first,
5232 stabilize sorting of integral types, remove conditions putting
5233 non-full-precision integers last.
5234 (sort_and_splice_var_accesses): Disable scalarization if a
5235 non-integert would be represented by a non-full-precision integer.
5237 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5239 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5240 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5241 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5242 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5243 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5244 conditionals inside the function instead of around it. Call
5245 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5246 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5248 2017-09-26 Richard Biener <rguenther@suse.de>
5250 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5252 (scop_detection::build_scop_breadth): ... this. Removed.
5253 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
5254 (scop_detection::harmful_stmt_in_bb): Likewise.
5255 (scop_detection::graphite_can_represent_stmt): Likewise.
5256 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
5257 (scop_detection::can_represent_loop): Remove recursion, fold in ...
5258 (scop_detection::can_represent_loop_1): ... this. Removed.
5259 (scop_detection::harmful_loop_in_region): Simplify after inlining
5260 the above and remove more quadraticness.
5261 (build_scops): Adjust.
5262 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
5265 2017-09-26 Jakub Jelinek <jakub@redhat.com>
5268 * config/i386/i386.c (ix86_print_operand_address_as): Only test
5269 REGNO (base) == SP_REG if base is a REG.
5272 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
5273 if it is different SSA_NAME.
5274 (optimize_range_tests_cmp_bitwise): New function.
5275 (optimize_range_tests): Call it.
5277 2017-09-26 Richard Biener <rguenther@suse.de>
5279 PR tree-optimization/82321
5280 * graphite.c (canonicalize_loop_closed_ssa): Properly check
5281 for the def being inside the loop.
5283 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5285 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
5287 * config/s390/s390-builtins.def: Fix constraint on op4.
5289 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5291 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
5292 independent expanders.
5293 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
5294 ("vec_ordered", "vec_unordered"): New expanders.
5296 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5298 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
5301 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5303 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
5304 vec_unpacks_lo_v16qi.
5305 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
5307 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5309 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
5310 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
5311 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
5313 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5315 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
5317 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
5319 ("*vec_slb<mode>"): New insn pattern.
5320 ("vec_shr_<mode>"): New expander.
5321 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
5322 and force the shift count operand to V16QImode.
5323 ("vec_srb<mode>"): Set shift count mode to V16QI.
5325 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5327 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
5328 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
5329 ("vec_widen_smult_hi_<mode>"): New expander definitions.
5331 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
5334 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
5336 2017-09-26 Richard Biener <rguenther@suse.de>
5338 PR tree-optimization/82320
5339 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
5342 2017-09-25 Jeff Law <law@redhat.com>
5344 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
5345 prototype for new argument.
5346 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
5347 mostly extracted from rs6000_emit_allocate_stack.
5348 (rs6000_emit_probe_stack_range_stack_clash): New function.
5349 (rs6000_emit_allocate_stack): Call
5350 rs6000_emit_probe_stack_range_stack_clash as needed.
5351 (rs6000_emit_probe_stack_range): Add additional argument
5352 to call to gen_probe_stack_range{si,di}.
5353 (output_probe_stack_range): New.
5354 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
5355 (output_probe_stack_range_stack_clash): New.
5356 (rs6000_emit_prologue): Emit notes into dump file as requested.
5357 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
5358 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
5359 Add additional operand and pass it to output_probe_stack_range.
5361 2017-09-25 Bin Cheng <bin.cheng@arm.com>
5363 PR tree-optimization/82163
5364 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
5365 (checking_verify_loop_closed_ssa): New parameter.
5366 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
5367 (check_loop_closed_ssa_stmt): Delete.
5368 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
5369 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
5370 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
5373 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
5375 * brig-builtins.def: Treat HSAIL barrier builtins as
5376 setjmp/longjump style functions.
5378 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5380 * target.def (constant_alignment): New hook.
5381 * defaults.h (CONSTANT_ALIGNMENT): Delete.
5382 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
5383 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
5384 * doc/tm.texi: Regenerate.
5385 * targhooks.h (default_constant_alignment): Declare.
5386 (constant_alignment_word_strings): Likewise.
5387 * targhooks.c (default_constant_alignment): New function.
5388 (constant_alignment_word_strings): Likewise.
5389 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
5390 instead of CONSTANT_ALIGNMENT.
5391 * varasm.c (align_variable, get_variable_align, build_constant_desc)
5392 (force_const_mem): Likewise.
5393 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
5394 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
5395 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
5396 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5397 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
5399 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
5400 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5401 constant_alignment_word_strings.
5402 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
5403 (CONSTANT_ALIGNMENT): Likewise.
5404 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5405 (arm_constant_alignment): New function.
5406 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
5407 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5408 constant_alignment_word_strings.
5409 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
5410 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5411 constant_alignment_word_strings.
5412 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
5413 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5414 (cris_constant_alignment): New function.
5415 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
5416 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5417 (epiphany_constant_alignment): New function.
5418 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
5419 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5420 constant_alignment_word_strings.
5421 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
5422 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5423 constant_alignment_word_strings.
5424 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
5425 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5426 constant_alignment_word_strings.
5427 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
5428 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
5429 * config/i386/i386.c (ix86_constant_alignment): Make static.
5430 Use the same interface as the target hook.
5431 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5432 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
5433 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5434 constant_alignment_word_strings.
5435 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
5436 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
5437 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5438 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
5439 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5440 constant_alignment_word_strings.
5441 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
5442 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5443 constant_alignment_word_strings.
5444 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
5445 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5446 constant_alignment_word_strings.
5447 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
5448 * config/microblaze/microblaze.c (microblaze_constant_alignment):
5450 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5451 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
5452 * config/mips/mips.c (mips_constant_alignment): New function.
5453 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5454 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
5455 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
5456 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5457 (mmix_constant_alignment): Make static. Use the same interface
5459 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
5460 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5461 constant_alignment_word_strings.
5462 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
5463 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5464 constant_alignment_word_strings.
5465 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
5466 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5467 constant_alignment_word_strings.
5468 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
5469 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5470 (rs6000_constant_alignment): New function.
5471 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
5472 * config/riscv/riscv.c (riscv_constant_alignment): New function.
5473 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5474 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
5475 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5476 (rs6000_constant_alignment): New function.
5477 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
5478 * config/s390/s390.c (s390_constant_alignment): New function.
5479 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5480 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
5481 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5482 constant_alignment_word_strings.
5483 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
5484 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5485 (sparc_constant_alignment): New function.
5486 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
5487 * config/spu/spu.c (spu_constant_alignment): New function.
5488 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5489 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
5490 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5491 constant_alignment_word_strings.
5492 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
5493 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5494 constant_alignment_word_strings.
5495 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
5496 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5497 constant_alignment_word_strings.
5498 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
5499 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5500 (visium_constant_alignment): New function.
5501 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
5502 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5503 (xtensa_constant_alignment): New function.
5504 * system.h (CONSTANT_ALIGNMENT): Poison.
5506 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
5508 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5509 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
5510 (rs6000_builtin_valid_without_lhs): New helper function.
5511 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5512 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
5514 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5516 * target.h (vec_perm_indices): Use unsigned short rather than
5518 (auto_vec_perm_indices): Likewise.
5519 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
5520 Use unsigned int rather than unsigned char.
5521 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
5523 2017-09-25 Richard Biener <rguenther@suse.de>
5525 * cfgloop.h (sort_sibling_loops): Declare.
5526 * cfgloop.c (sort_sibling_loops_cmp): New helper.
5527 (sort_sibling_loops): New function sorting the sibling loop list
5529 * graphite.c (graphite_transform_loops): Sort sibling loops.
5531 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
5533 * target.def (vec_perm_const_ok): Change sel parameter to
5535 * optabs-query.c (can_vec_perm_p): Update accordingly.
5536 * doc/tm.texi: Regenerate.
5537 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
5538 auto_vec_perm_indices and remove separate nelt field.
5539 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
5540 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
5541 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
5542 (aarch64_expand_vec_perm_const): Update accordingly.
5543 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
5544 to vec_perm_indices.
5545 * config/arm/arm.c (expand_vec_perm_d): Change perm to
5546 auto_vec_perm_indices and remove separate nelt field.
5547 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
5548 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
5549 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
5551 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
5552 to vec_perm_indices.
5553 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
5554 sel to vec_perm_indices.
5555 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
5556 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
5557 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
5559 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
5562 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
5565 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
5566 on the FUNCTION_DECL function context if it has a DIE that is a
5569 2017-09-25 Richard Biener <rguenther@suse.de>
5571 PR tree-optimization/82285
5572 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
5575 2017-09-25 Tom de Vries <tom@codesourcery.com>
5579 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
5582 2017-09-25 Richard Biener <rguenther@suse.de>
5584 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
5585 ISL errors other than isl_error_quota happen. Dump if the
5586 schedule is the same.
5587 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
5588 errors instead of aborting inside ISL.
5590 2017-09-25 Iain Sandoe <iain@codesourcery.com>
5593 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
5594 of libgcc_eh for m64.
5595 * config/i386/darwin64.h: Likewise.
5597 2017-09-25 Richard Biener <rguenther@suse.de>
5600 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
5601 attribute for incomplete types nor twice for complete ones.
5603 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
5606 * config/i386/i386.c (ix86_print_operand_address_as): Encode
5607 %esp as %rsp to avoid 0x67 prefix if there is no index or base
5610 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
5613 * config/i386/i386.opt (mprefer-avx256): Use
5614 ix86_target_flags variable.
5615 * config/i386/i386.c (ix86_target_string): Move
5616 -mprefer-avx256 to flag2_opts.
5618 2017-09-22 Jakub Jelinek <jakub@redhat.com>
5621 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
5622 and x != -1 | y != -1 into (x & y) != -1.
5624 2017-09-22 Steve Ellcey <sellcey@cavium.com>
5626 * config.gcc: Add new case statement to set
5627 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
5628 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
5629 s390x-*-linux* case statements. Added aarch64 to the list of
5630 supported architectures.
5632 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5634 PR tree-optimization/82289
5635 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
5636 STMT_VINFO_RELEVANT_P.
5638 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5639 Alan Hayward <alan.hayward@arm.com>
5640 David Sherwood <david.sherwood@arm.com>
5642 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
5643 for VR_RANGE only; don't allow VR_ANTI_RANGE.
5644 (extract_range_from_binary_expr_1): Don't call
5645 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
5647 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5648 Alan Hayward <alan.hayward@arm.com>
5649 David Sherwood <david.sherwood@arm.com>
5651 * target.def (preferred_vector_alignment): New hook.
5652 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
5654 * doc/tm.texi: Regenerate.
5655 * targhooks.h (default_preferred_vector_alignment): Declare.
5656 * targhooks.c (default_preferred_vector_alignment): New function.
5657 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
5659 (DR_TARGET_ALIGNMENT): New macro.
5660 (aligned_access_p): Update commentary.
5661 (vect_known_alignment_in_bytes): New function.
5662 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
5664 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
5665 Calculate the misalignment based on the target alignment rather than
5667 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
5668 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
5669 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
5670 the target alignment, rather than masking the element misalignment
5671 with the number of elements in a vector. Also use the target
5672 alignment when calculating the maximum number of peels.
5673 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
5674 instead of TYPE_ALIGN_UNIT.
5675 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
5676 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
5677 (vect_create_addr_base_for_vector_ref): Update call accordingly.
5678 (vect_create_data_ref_ptr): Likewise.
5679 (vect_setup_realignment): Realign by ANDing with
5680 -DR_TARGET_MISALIGNMENT.
5681 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
5682 the number of peels based on DR_TARGET_ALIGNMENT.
5683 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
5684 with the guaranteed alignment boundary when deciding whether
5686 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
5687 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
5688 (ensure_base_align): Remove stmt_info parameter. Get the
5689 target base alignment from DR_TARGET_ALIGNMENT.
5690 (vectorizable_store): Update call accordingly. Interpret
5691 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
5693 (vectorizable_load): Likewise.
5695 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5696 Alan Hayward <alan.hayward@arm.com>
5697 David Sherwood <david.sherwood@arm.com>
5699 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
5700 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
5701 (vect_enhance_data_refs_alignment): Likewise.
5703 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
5705 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
5706 error. Only quit immediately if parsing is complete.
5707 (BEGIN): Initialize fatal_err and parse_done.
5708 (begin fpu, end fpu): Check number of arguments.
5709 (begin arch, end arch): Likewise.
5710 (begin cpu, end cpu): Likewise.
5711 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5712 (optalias): Likewise.
5714 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
5716 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
5717 * config/arm/arm-isa.h: Delete. Move definitions to ...
5718 * arm-cpus.in: ... here. Use new feature and fgroup values.
5719 * config/arm/arm.c (arm_option_override): Use lower case for feature
5721 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
5722 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
5723 * config/arm/parsecpu.awk (END): Add new command 'isa'.
5725 (print_isa_bits_for): New function.
5726 (gen_isa): New function.
5727 (gen_comm_data): Use print_isa_bits_for.
5728 (define feature): New keyword.
5729 (define fgroup): New keyword.
5730 * config/arm/t-arm (TM_H): Remove.
5731 (GTM_H): Add arm-isa.h.
5732 (arm-isa.h): Add rule to generate file.
5733 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
5734 case for feature bit names.
5736 2017-09-22 Richard Biener <rguenther@suse.de>
5738 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
5740 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
5741 print of no dependency loops ...
5742 * graphite.c (graphite_transform_loops): ... here.
5743 (canonicalize_loop_closed_ssa_form): Work from inner to outer
5745 (same_close_phi_node, remove_duplicate_close_phi,
5746 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
5747 (canonicalize_loop_closed_ssa): ... here and simplify.
5748 * graphite-optimize-isl.c: Include tree-vectorizer.h.
5749 (optimize_isl): Use dump_printf_loc to tell when we stopped
5750 optimizing because of an ISL timeout.
5752 2017-09-22 Richard Biener <rguenther@suse.de>
5754 PR tree-optimization/82291
5755 * tree-if-conv.c (predicate_mem_writes): Make sure to
5756 remove writes in blocks predicated with false.
5758 2017-09-22 Richard Biener <rguenther@suse.de>
5760 * sese.c: Include cfganal.h.
5761 (if_region_set_false_region): Remove.
5762 (create_if_region_on_edge): Likewise.
5763 (move_sese_in_condition): Re-implement without destroying
5766 2017-09-22 Richard Biener <rguenther@suse.de>
5768 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
5769 Verify both BBs contain loop PHI nodes before dispatching to
5771 (graphite_regenerate_ast_isl): Do not recompute dominators,
5772 do not verify three times. Restructure for clarity.
5773 * graphite-scop-detection.c (same_close_phi_node,
5774 remove_duplicate_close_phi, make_close_phi_nodes_unique,
5775 defined_in_loop_p, canonicalize_loop_closed_ssa,
5776 canonicalize_loop_closed_ssa_form): Simplify, remove excess
5777 checking and SSA rewrite, move to ...
5778 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
5779 (graphite_initialize): Do not pass in ctx, do not reset the
5780 SCEV cache, compute only dominators.
5781 (graphite_transform_loops): Allocate ISL ctx after
5782 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
5783 Maintain post-dominators only around build_scops.
5784 * sese.c (if_region_set_false_region): Make static. Free
5785 and recompute dominators.
5786 (move_sese_in_condition): Assert we don't get called with
5787 post-dominators computed.
5788 * sese.h (if_region_set_false_region): Remove.
5790 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
5792 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
5793 mode attribute for TARGET_AVX512VL.
5795 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
5797 * config/i386/i386.opt (mprefer-avx256): New option.
5798 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
5800 (ix86_preferred_simd_mode): Return 256-bit AVX modes
5801 for TARGET_PREFER_AVX256.
5802 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
5804 2017-09-21 Jeff Law <law@redhat.com>
5806 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
5807 Fix dump output if the only stack space is for pushed registers.
5809 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5811 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
5814 2017-09-21 Martin Sebor <msebor@redhat.com>
5817 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
5818 code (in C++) or code that triggers warnings.
5820 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
5822 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
5824 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
5826 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
5827 * sched-rgn.c: Ditto.
5828 * sel-sched-ir.c: Ditto.
5830 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
5832 * toplev.h (set_random_seed): Adjust return type.
5833 * toplev.c (init_local_tick): Move eager initialization of random_seed
5834 to get_random_seed. Adjust comment.
5835 (init_random_seed): Inline to get_random_seed, delete.
5836 (get_random_seed): Initialize random_seed lazily.
5837 (set_random_seed): Do not return previous value.
5838 (print_switch_value): Do not call get_random_seed.
5840 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
5842 * cgraph.c (delete_function_version): New, broken out from...
5843 (cgraph_node::delete_function_version): ...here. Rename to
5844 cgraph_node::delete_function_version_by_decl. Update all uses.
5845 (cgraph_node::remove): Call delete_function_version.
5847 2017-09-21 Jakub Jelinek <jakub@redhat.com>
5850 * tree-inline.c (expand_call_inline): Emit clobber stmts for
5851 VAR_DECLs to which addressable non-volatile parameters are mapped
5852 and for id->retvar after the return value assignment. Clear
5853 id->retval and id->retbnd after inlining.
5855 2017-09-21 Richard Biener <rguenther@suse.de>
5857 PR tree-optimization/82276
5858 PR tree-optimization/82244
5859 * tree-vrp.c (build_assert_expr_for): Set
5860 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
5862 (remove_range_assertions): Revert earlier change.
5864 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
5867 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
5869 2017-09-21 Richard Biener <rguenther@suse.de>
5871 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5872 Restore valid IL after code generation errors.
5873 * graphite.c (graphite_transform_loops): Diagnose code
5874 generation issues as MSG_MISSED_OPTIMIZATION and continue
5875 with processing SCOPs.
5877 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5878 Alan Hayward <alan.hayward@arm.com>
5879 David Sherwood <david.sherwood@arm.com>
5881 * calls.c (compute_argument_addresses): Use simplify_gen_binary
5882 rather than choosing between plus_constant and gen_rtx_<CODE>.
5883 * expr.c (emit_push_insn): Likewise.
5884 (expand_expr_real_2): Likewise.
5886 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5887 Alan Hayward <alan.hayward@arm.com>
5888 David Sherwood <david.sherwood@arm.com>
5890 * loop-unroll.c (split_iv): Call copy_rtx on the step.
5892 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5893 Alan Hayward <alan.hayward@arm.com>
5894 David Sherwood <david.sherwood@arm.com>
5896 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
5897 calling tree_to_uhwi.
5899 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5900 Alan Hayward <alan.hayward@arm.com>
5901 David Sherwood <david.sherwood@arm.com>
5903 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
5904 INTEGER_CST rather than a negative test for ADDR_EXPR.
5906 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5907 Alan Hayward <alan.hayward@arm.com>
5908 David Sherwood <david.sherwood@arm.com>
5910 * tree-vrp.c (extract_range_from_binary_expr_1): Check
5911 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
5913 2017-09-21 Richard Biener <rguenther@suse.de>
5915 PR tree-optimization/71351
5916 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
5917 graphite_create_new_loop_guard): Remove, fold remaining parts
5919 (translate_isl_ast_node_for): ... here and simplify.
5921 2017-09-21 Jakub Jelinek <jakub@redhat.com>
5924 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
5925 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
5926 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
5927 alternative always use QI mode, for -Os imov (=R,R) alternative
5928 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
5931 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5932 Jeff Law <law@redhat.com>
5934 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
5935 (allocate_stack_space): New function, partially extracted from
5937 (s390_emit_prologue): Track offset to most recent stack probe.
5938 Code to allocate space moved into allocate_stack_space.
5939 Dump actions when no stack is allocated.
5940 (s390_prologue_plus_offset): New function.
5941 (s390_emit_stack_probe): Likewise.
5943 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
5945 * common.opt (Wa, Wl, Wp, g, gz=): Add
5947 (gno-column-info): Remove.
5948 (gcolumn-info): Drop RejectNegative.
5950 (gno-record-gcc-switches): Remove.
5951 (grecord-gcc-switches): Drop RejectNegative.
5952 (gno-split-dwarf): Remove.
5953 (gsplit-dwarf): Drop RejectNegative.
5954 (gno-strict-dwarf): Remove.
5955 (gstrict-dwarf): Drop RejectNegative.
5956 * config/darwin.opt (gfull, gused): Add RejectNegative.
5957 * dwarf2out.c (gen_producer_string): Drop
5958 gno-record-gcc-switches handler.
5959 * optc-gen.awk: Add g to prefixes with negative forms.
5960 * opts-common.c (remapping_prefix_p): New.
5961 (find_opt): Check it.
5962 (generate_canonical_option): Test g prefix.
5963 (option_map): Add -gno- mapping.
5964 (add_misspelling_candidates): Check remapping_prefix_p.
5966 2017-09-20 Jeff Law <law@redhat.com>
5968 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
5969 thinko in stack clash protection support.
5971 * explow.c (compute_stack_clash_protection_loop_data): Use
5972 CONST_INT_P instead of explicit test. Verify object is a
5973 CONST_INT_P before looking at INTVAL.
5974 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
5975 instead of explicit test.
5977 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
5980 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
5981 address instead of to r1 and r11.
5983 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
5985 * config.gcc: Support "knm".
5986 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
5987 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5989 * config/i386/i386.c (m_KNM): Define.
5990 (processor_target_table): Add "knm".
5992 (ix86_option_override_internal): Add "knm".
5993 (ix86_issue_rate): Add PROCESSOR_KNM.
5994 (ix86_adjust_cost): Ditto.
5995 (ia32_multipass_dfa_lookahead): Ditto.
5996 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
5997 (fold_builtin_cpu): Add M_INTEL_KNM.
5998 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
5999 (processor_type): Add PROCESSOR_KNM.
6000 * config/i386/x86-tune.def: Add m_KNM.
6001 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6003 2017-09-20 Richard Biener <rguenther@suse.de>
6005 PR tree-optimization/80213
6006 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6007 are allowed in empty BBs as well.
6008 (canonicalize_loop_closed_ssa): Also look for other complex
6010 (scop_detection::get_sese): Include the loop-closed PHI block
6012 (scop_detection::merge_sese): Remove code adding extra blocks.
6013 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6014 (build_scops): Assert the final returned scop is invalid.
6016 2017-09-20 Richard Biener <rguenther@suse.de>
6018 PR tree-optimization/82264
6019 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6021 (vn_phi_lookup): Likewise.
6022 (vn_phi_insert): Likewise.
6024 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6026 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6027 that fits into uhwi or shwi, add DW_AT_const_value regardless
6028 of early_dwarf without going through RTL, using add_AT_unsigned
6031 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6032 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6033 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6034 (DEBUG_MACRO_SECTION): Likewise.
6035 (DEBUG_LTO_MACRO_SECTION): Likewise.
6036 (DEBUG_STR_DWO_SECTION): Likewise.
6037 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6038 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6039 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6040 (DEBUG_LTO_LINE_STR_SECTION): Define.
6041 (init_sections_and_labels): Initialize debug_line_str_section
6042 variable. Initialize debug_loc_section for -gdwarf-5 to
6043 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6045 2017-09-20 Richard Biener <rguenther@suse.de>
6047 * graphite-sese-to-poly.c (extract_affine): Properly handle
6048 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6050 2017-09-20 Richard Biener <rguenther@suse.de>
6052 PR tree-optimization/81373
6053 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6054 Force SESE live-out defs to be handled even if they are
6057 2017-09-19 Jeff Law <law@redhat.com>
6059 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6060 nothing for stack adjustments with REG_STACK_CHECK.
6061 * sched-deps.c (parse_add_or_inc): Reject insns with
6062 REG_STACK_CHECK from dependency breaking.
6063 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6064 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6065 * reg-notes.def (STACK_CHECK): New note.
6067 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6068 (ix86_expand_prologue): Dump stack clash info as needed.
6069 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6071 * function.c (dump_stack_clash_frame_info): New function.
6072 * function.h (dump_stack_clash_frame_info): Prototype.
6073 (enum stack_clash_probes): New enum.
6075 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6076 flag_stack_clash_protection.
6077 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6078 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6079 (arm_frame_pointer_required): Likewise.
6080 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6081 (ia64_expand_prologue): Likewise.
6082 * config/mips/mips.c (mips_expand_prologue): Likewise.
6083 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6084 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6085 (sparc_flat_expand_prologue): Likewise.
6086 * config/spu/spu.c (spu_expand_prologue): Likewise.
6088 * explow.c: Include "params.h".
6089 (anti_adjust_stack_and_probe_stack_clash): New function.
6090 (get_stack_check_protect): Likewise.
6091 (compute_stack_clash_protection_loop_data): Likewise.
6092 (emit_stack_clash_protection_loop_start): Likewise.
6093 (emit_stack_clash_protection_loop_end): Likewise.
6094 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6095 Use anti_adjust_stack_and_probe_stack_clash.
6096 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6097 (emit_stack_clash_protection_loop_start): Likewise.
6098 (emit_stack_clash_protection_loop_end): Likewise.
6099 * rtl.h (get_stack_check_protect): Prototype.
6100 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6101 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6102 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6104 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6106 * doc/tm.texi: Rebuilt.
6107 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6108 get_stack_check_protect.
6109 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6110 * config/arm/arm.c (arm_expand_prologue): Likewise.
6111 (arm_frame_pointer_required): Likewise.
6112 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6113 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6114 * config/mips/mips.c (mips_expand_prologue): Likewise.
6115 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6116 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6117 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6118 (sparc_flat_expand_prologue): Likewise.
6120 * common.opt (-fstack-clash-protection): New option.
6121 * flag-types.h (enum stack_check_type): Note difference between
6122 -fstack-check= and -fstack-clash-protection.
6123 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6124 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6125 * toplev.c (process_options): Issue warnings/errors for cases
6126 not handled with -fstack-clash-protection.
6127 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6128 (-fstack-check): Note additional problem with -fstack-check=generic.
6129 Note that -fstack-check is primarily for Ada and refer users
6130 to -fstack-clash-protection for stack-clash-protection.
6131 Document new params for stack clash protection.
6133 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6135 * config/i386/i386.c (ix86_split_long_move): Do not handle
6136 address used for LEA in a special way.
6138 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6140 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6142 2017-09-19 Martin Sebor <msebor@redhat.com>
6145 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6146 of incompatible types.
6148 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6150 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6151 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6152 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6153 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6155 2017-09-19 Richard Biener <rguenther@suse.de>
6157 PR tree-optimization/82244
6158 * tree-vrp.c (remove_range_assertions): Do not propagate
6159 a constant to abnormals but replace the assert with a copy.
6161 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6163 PR rtl-optimization/57878
6164 PR rtl-optimization/68988
6165 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6166 avoidance test involving non_reload_pseudos. Move frequency test
6167 below the general fragmentation avoidance test.
6169 2017-09-19 Richard Biener <rguenther@suse.de>
6171 PR tree-optimization/69728
6172 * graphite-sese-to-poly.c (schedule_error): New global.
6173 (add_loop_schedule): Handle empty domain by failing the
6175 (build_original_schedule): Handle schedule_error.
6177 2017-09-19 Richard Biener <rguenther@suse.de>
6179 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6180 Do not iterate to sibling loops but only to siblings of inner
6183 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6186 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6187 that the registers are different.
6189 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6191 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6192 to processor_model and "amdfam17h" to arch_names_table.
6193 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6195 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6198 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6200 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6201 Alan Hayward <alan.hayward@arm.com>
6202 David Sherwood <david.sherwood@arm.com>
6204 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6206 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6208 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6209 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6210 vect_schedule_slp_instance.
6211 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6212 Update call to vect_slp_analyze_node_operations. Simplify return
6214 (vect_slp_analyze_bb_1): Update call accordingly.
6215 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6216 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6217 (vect_schedule_slp): Update call accordingly.
6219 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6220 Alan Hayward <alan.hayward@arm.com>
6221 David Sherwood <david.sherwood@arm.com>
6223 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6224 with types that aren't in fact scalar.
6226 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6228 * tree-vect-slp.c (vect_record_max_nunits): New function,
6230 (vect_build_slp_tree_1): ...here.
6231 (vect_build_slp_tree_2): Call it for phis too.
6233 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6235 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6236 to vect_get_vec_def_for_operand when getting the mask operand.
6238 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6239 Alan Hayward <alan.hayward@arm.com>
6240 David Sherwood <david.sherwood@arm.com>
6242 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
6245 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6246 Alan Hayward <alan.hayward@arm.com>
6247 David Sherwood <david.sherwood@arm.com>
6249 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
6250 calculation for vector booleans.
6252 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6253 Alan Hayward <alan.hayward@arm.com>
6254 David Sherwood <david.sherwood@arm.com>
6256 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
6258 (vect_transform_stmt): ...here.
6259 (vect_analyze_stmt): Use it instead of calling
6260 vectorizable_live_operation directly.
6262 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
6264 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
6265 non-SIMT targets in acc vector loops.
6267 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
6269 * configure.ac: Add arc and check if assembler supports gdwarf2.
6270 * configure: Regenerate.
6272 2017-09-18 Richard Biener <rguenther@suse.de>
6274 PR tree-optimization/82220
6275 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
6276 epilogue niters from the min_profitable_iters compute.
6278 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6281 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
6282 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
6283 (ix86_init_pic_reg): Revert 2017-09-01 changes.
6285 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
6288 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
6289 switching to a new text section.
6291 2017-09-18 Richard Biener <rguenther@suse.de>
6293 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
6295 (build_alias_set): Reject aliases with no access function.
6297 2017-09-18 Richard Biener <rguenther@suse.de>
6299 PR tree-optimization/79622
6300 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
6302 (build_cross_bb_scalars_use): Likewise.
6304 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
6306 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
6308 2017-09-18 Alan Modra <amodra@gmail.com>
6311 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
6312 stack_pointer_rtx for count 0. Update comments. Break up
6313 large rtl expression.
6315 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
6317 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
6318 Increase to 20 bytes.
6319 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
6320 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
6321 or avx version of the stub.
6323 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
6326 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
6327 compute the minimum stack alignment. Also update preferred stack
6328 boundary for leaf functions.
6330 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
6332 PR tree-optimization/82228
6333 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
6336 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6338 * common/config/nds32/nds32-common.c
6339 (nds32_option_optimization_table): Refine formatting.
6340 (nds32_option_optimization_table): Use -fsched-pressure and
6341 -fomit-frame-pointer for specific optimization level.
6343 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6345 * config/nds32/nds32.c: Refine formatting and comments.
6346 * config/nds32/nds32.h: Likewise.
6347 * config/nds32/nds32.md: Likewise.
6348 * config/nds32/nds32-cost.c: Likewise.
6349 * config/nds32/nds32-isr.c: Likewise.
6350 * config/nds32/nds32-md-auxiliary.c: Likewise.
6351 * config/nds32/nds32-multiple.md: Likewise.
6352 * config/nds32/nds32-predicates.c: Likewise.
6354 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
6355 Jakub Jelinek <jakub@redhat.com>
6357 Add support for -std=c++2a.
6358 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
6360 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
6362 2017-09-15 Steve Ellcey <sellcey@cavium.com>
6365 * doc/extend.texi (Common Function Attributes): Add
6366 references to ARM, AArch64, and S/390 specific attributes.
6367 (Function Specific Option Pragmas): Add AArch64 and S/390
6368 to list of back ends that support the target pragma.
6370 2017-09-15 Nathan Sidwell <nathan@acm.org>
6372 * doc/standards.texi: Fix C++17 description. Update URLs for
6375 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
6377 * common.opt (Wcast-align=strict): New warning option.
6378 * doc/invoke.texi: Document -Wcast-align=strict.
6380 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
6382 * cgraph.h (cgraph_thunk_info): Add comments.
6383 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
6384 assert for VIRTUAL_* arguments stricter.
6386 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
6388 PR tree-optimization/71026
6389 * match.pd: Move RDIV patterns from fold-const.c
6390 * fold-const.c (distribute_real_division): Removed.
6391 (fold_binary_loc): Remove calls to distribute_real_divison.
6393 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6395 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
6396 c++1z and gnu++1z as deprecated. Change other references to
6397 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
6398 Change -Wc++1z-compat to -Wc++17-compat.
6399 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
6400 * dwarf2out.c (highest_c_language): Handle C++17.
6401 (gen_compile_unit_die): Likewise.
6403 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6405 PR rtl-optimization/82192
6406 * combine.c (make_extraction): Don't look through non-paradoxical
6407 SUBREGs or TRUNCATE if pos + len is or might be bigger than
6410 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6411 Alan Hayward <alan.hayward@arm.com>
6412 David Sherwood <david.sherwood@arm.com>
6414 * target.def (function_arg_offset): New hook.
6415 * targhooks.h (default_function_arg_offset): Declare.
6416 * targhooks.c (default_function_arg_offset): New function.
6417 * function.c (locate_and_pad_parm): Use
6418 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
6419 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
6420 (TARGET_FUNCTION_ARG_OFFSET): ...this.
6421 * doc/tm.texi: Regenerate.
6422 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
6423 * config/spu/spu.c (spu_function_arg_offset): New function.
6424 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
6425 * system.h (FUNCTION_ARG_OFFSET): Poison.
6427 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6428 Alan Hayard <alan.hayward@arm.com>
6429 David Sherwood <david.sherwood@arm.com>
6431 * target.def (truly_noop_truncation): New hook.
6432 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
6433 than TRULY_NOOP_TRUNCATION.
6434 * hooks.h (hook_bool_uint_uint_true): Declare.
6435 * hooks.c (hook_bool_uint_uint_true): New function.
6436 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
6437 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
6438 * doc/tm.texi: Regenerate.
6439 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
6440 rather than TRULY_NOOP_TRUNCATION in comments.
6441 (simplify_comparison): Likewise.
6442 (record_truncated_value): Likewise.
6443 * expmed.c (extract_bit_field_1): Likewise.
6444 (extract_split_bit_field): Likewise.
6445 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
6446 instead of TRULY_NOOP_TRUNCATION.
6447 * function.c (assign_parm_setup_block): Likewise.
6448 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
6449 * rtlhooks.c: Include target.h.
6450 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
6451 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
6452 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
6453 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
6454 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
6455 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
6456 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
6457 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
6458 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
6459 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
6460 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
6461 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
6462 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
6463 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
6464 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
6465 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
6466 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
6467 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
6468 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
6469 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
6470 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
6471 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
6472 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
6473 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
6474 * config/mips/mips.c (mips_truly_noop_truncation): New function.
6475 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6476 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6477 rather than TRULY_NOOP_TRUNCATION in comments.
6478 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
6479 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
6480 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
6481 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
6482 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
6483 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
6484 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
6485 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
6486 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
6487 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
6488 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
6489 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6490 rather than TRULY_NOOP_TRUNCATION in comments.
6491 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
6492 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
6493 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
6494 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
6495 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
6496 TRULY_NOOP_TRUNCATION condition.
6497 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
6498 (TRULY_NOOP_TRUNCATION): Delete.
6499 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
6500 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
6501 * config/spu/spu.c (spu_truly_noop_truncation): New function.
6502 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6503 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
6504 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
6505 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
6506 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6507 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6508 rather than TRULY_NOOP_TRUNCATION in comments.
6509 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
6510 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
6511 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
6512 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
6513 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
6514 * system.h (TRULY_NOOP_TRUNCATION): Poison.
6516 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
6519 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
6520 (*cmp_ior): Likewise.
6521 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
6522 (*ior_scc_scc_cmp): Likewise.
6523 (*and_scc_scc): Likewise.
6524 (*and_scc_scc_cmp): Likewise.
6526 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6527 Alan Hayard <alan.hayward@arm.com>
6528 David Sherwood <david.sherwood@arm.com>
6530 * target.def (can_change_mode_class): New hook.
6531 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
6532 (hard_regno_nregs): Likewise.
6533 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
6534 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
6535 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
6536 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
6537 (LOAD_EXTEND_OP): Update accordingly.
6538 * doc/tm.texi: Regenerate.
6539 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
6540 CANNOT_CHANGE_MODE_CLASS.
6541 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
6542 (REG_CAN_CHANGE_MODE_P): ...this new macro.
6543 * combine.c (simplify_set): Update accordingly.
6544 * emit-rtl.c (validate_subreg): Likewise.
6545 * recog.c (general_operand): Likewise.
6546 * regcprop.c (mode_change_ok): Likewise.
6547 * reload1.c (choose_reload_regs): Likewise.
6548 (inherit_piecemeal_p): Likewise.
6549 * rtlanal.c (simplify_subreg_regno): Likewise.
6550 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
6551 instead of CANNOT_CHANGE_MODE_CLASS.
6552 (reload_cse_simplify_operands): Likewise.
6553 * reload.c (push_reload): Use targetm.can_change_mode_class
6554 instead of CANNOT_CHANGE_MODE_CLASS.
6555 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
6556 REG_CANNOT_CHANGE_MODE_P.
6557 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6558 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
6559 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6560 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6561 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6562 (arm_can_change_mode_class): New function.
6563 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6564 than CANNOT_CHANGE_MODE_CLASS in comments.
6565 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6566 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
6567 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
6568 (ix86_can_change_mode_class): ...this new function, inverting the
6569 sense of the return value.
6570 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6571 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6572 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6573 (ia64_can_change_mode_class): New function.
6574 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6575 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
6576 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
6577 (m32c_can_change_mode_class): ...this new function, inverting the
6578 sense of the return value.
6579 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6580 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6581 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
6582 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
6583 (mips_can_change_mode_class): ...this new function, inverting the
6584 sense of the return value.
6585 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6586 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6587 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6588 (msp430_can_change_mode_class): New function.
6589 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6590 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
6591 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6592 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6593 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6594 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
6595 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6596 (pa_cannot_change_mode_class): Replace with...
6597 (pa_can_change_mode_class): ...this new function, inverting the
6598 sense of the return value.
6599 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6600 than CANNOT_CHANGE_MODE_CLASS in comments.
6601 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6602 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
6603 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6604 (pdp11_cannot_change_mode_class): Replace with...
6605 (pdp11_can_change_mode_class): ...this new function, inverting the
6606 sense of the return value.
6607 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6608 * config/powerpcspe/powerpcspe-protos.h
6609 (rs6000_cannot_change_mode_class_ptr): Delete.
6610 * config/powerpcspe/powerpcspe.c
6611 (rs6000_cannot_change_mode_class_ptr): Delete.
6612 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6613 (rs6000_option_override_internal): Assign to
6614 targetm.can_change_mode_class instead of
6615 rs6000_cannot_change_mode_class_ptr.
6616 (rs6000_cannot_change_mode_class): Replace with...
6617 (rs6000_can_change_mode_class): ...this new function, inverting the
6618 sense of the return value.
6619 (rs6000_debug_cannot_change_mode_class): Replace with...
6620 (rs6000_debug_can_change_mode_class): ...this new function.
6621 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6622 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
6623 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6624 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6625 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
6627 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
6628 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6629 (rs6000_option_override_internal): Assign to
6630 targetm.can_change_mode_class instead of
6631 rs6000_cannot_change_mode_class_ptr.
6632 (rs6000_cannot_change_mode_class): Replace with...
6633 (rs6000_can_change_mode_class): ...this new function, inverting the
6634 sense of the return value.
6635 (rs6000_debug_cannot_change_mode_class): Replace with...
6636 (rs6000_debug_can_change_mode_class): ...this new function.
6637 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6638 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
6639 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
6640 (s390_can_change_mode_class): ...this new function, inverting the
6641 sense of the return value.
6642 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6643 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6644 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
6645 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6646 (sh_cannot_change_mode_class): Replace with...
6647 (sh_can_change_mode_class): ...this new function, inverting the
6648 sense of the return value.
6649 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6650 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6651 (sparc_can_change_mode_class): New function.
6652 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6653 * config/spu/spu.c (spu_can_change_mode_class): New function.
6654 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6655 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6656 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6657 (visium_can_change_mode_class): New function.
6658 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
6660 2017-09-15 Richard Biener <rguenther@suse.de>
6662 PR tree-optimization/82217
6663 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
6664 but not undefined case.
6666 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6669 * postreload.c (reload_cse_simplify_operands): Skip
6670 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
6672 2017-09-15 Richard Biener <rguenther@suse.de>
6674 PR tree-optimization/68823
6675 * graphite-scop-detection.c (build_alias_set): If we have a
6676 possible dependence check whether we can handle them by just
6677 looking at the DRs DR_ACCESS_FNs.
6678 (build_scops): If build_alias_set fails, fail the SCOP.
6680 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6682 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
6683 to support float128 built-in functions that require the ISA 3.0
6685 (BU_FLOAT128_3_HW): Likewise.
6686 (SQRTF128): Add support for the IEEE 128-bit square root and fma
6688 (FMAF128): Likewise.
6690 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
6691 support for built-in functions that need the ISA 3.0 IEEE 128-bit
6692 floating point instructions.
6693 (rs6000_invalid_builtin): Likewise.
6694 (rs6000_builtin_mask_names): Likewise.
6695 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
6696 (RS6000_BTM_FLOAT128_HW): Likewise.
6697 (RS6000_BTM_COMMON): Likewise.
6698 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
6700 * doc/extend.texi (RS/6000 built-in functions): Document the
6701 IEEE 128-bit floating point square root and fused multiply-add
6704 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
6706 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
6707 reg (r2) isn't in the set of registers defined in the prologue.
6709 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6710 Alan Hayward <alan.hayward@arm.com>
6711 David Sherwood <david.sherwood@arm.com>
6713 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
6714 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
6715 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
6716 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
6717 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
6719 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
6720 max_vectorization_factor.
6721 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
6723 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6724 Alan Hayward <alan.hayward@arm.com>
6725 David Sherwood <david.sherwood@arm.com>
6727 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
6728 (vect_worthwhile_without_simd_p): Declare.
6729 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
6730 (vectorizable_reduction): Use it.
6731 * tree-vect-stmts.c (vectorizable_shift): Likewise.
6732 (vectorizable_operation): Likewise.
6734 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6735 Alan Hayward <alan.hayward@arm.com>
6736 David Sherwood <david.sherwood@arm.com>
6738 * tree-vectorizer.h (vect_get_num_copies): New function.
6739 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
6740 * tree-vect-loop.c (vectorizable_reduction): Likewise.
6741 (vectorizable_induction): Likewise.
6742 (vectorizable_live_operation): Likewise.
6743 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6744 (vectorizable_bswap): Likewise.
6745 (vectorizable_call): Likewise.
6746 (vectorizable_conversion): Likewise.
6747 (vectorizable_assignment): Likewise.
6748 (vectorizable_shift): Likewise.
6749 (vectorizable_operation): Likewise.
6750 (vectorizable_store): Likewise.
6751 (vectorizable_load): Likewise.
6752 (vectorizable_condition): Likewise.
6753 (vectorizable_comparison): Likewise.
6754 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
6756 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6757 Alan Hayward <alan.hayward@arm.com>
6758 David Sherwood <david.sherwood@arm.com>
6760 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
6761 of vect_init_vector.
6763 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6764 Alan Hayward <alan.hayward@arm.com>
6765 David Sherwood <david.sherwood@arm.com>
6767 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
6768 an inline wrapper that provides a location.
6769 (gimple_build_vector): Likewise.
6770 * gimple-fold.c (gimple_build_vector_from_val): New function.
6771 (gimple_build_vector): Likewise.
6772 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
6773 functions to build the initial value. Always return a gimple value.
6774 (get_initial_defs_for_reduction): Likewise. Only compute
6776 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
6777 vect_init_vector on the results from get_initial_def(s)_for_reduction.
6778 (vectorizable_induction): Use gimple_build_vector rather than
6781 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6782 Alan Hayward <alan.hayward@arm.com>
6783 David Sherwood <david.sherwood@arm.com>
6785 * target.h (vec_perm_indices): New typedef.
6786 (auto_vec_perm_indices): Likewise.
6787 * optabs-query.h: Include target.h
6788 (can_vec_perm_p): Take a vec_perm_indices *.
6789 * optabs-query.c (can_vec_perm_p): Likewise.
6790 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
6791 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
6792 * tree-vect-generic.c (lower_vec_perm): Likewise.
6793 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
6794 (vect_grouped_load_supported): Likewise.
6795 (vect_shift_permute_load_chain): Likewise.
6796 (vect_permute_store_chain): Use auto_vec_perm_indices.
6797 (vect_permute_load_chain): Likewise.
6798 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
6799 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
6800 Update uses of can_vec_perm_p.
6801 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
6802 mode with a number of elements. Take a vec_perm_indices *.
6803 (vect_create_epilog_for_reduction): Update accordingly.
6804 Use auto_vec_perm_indices.
6805 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
6806 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
6807 (vect_transform_slp_perm_load): Likewise.
6808 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
6809 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
6810 (vect_gen_perm_mask_checked): Likewise.
6811 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
6812 (vect_gen_perm_mask_checked): Likewise.
6813 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
6814 (vectorizable_store): Likewise.
6815 (vectorizable_load): Likewise.
6816 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
6817 (vectorizable_bswap): Likewise.
6819 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6820 Alan Hayward <alan.hayward@arm.com>
6821 David Sherwood <david.sherwood@arm.com>
6823 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
6824 * tree.c (build_vector): Likewise.
6825 (build_vector_from_ctor): Update accordingly.
6826 (build_vector_from_val): Likewise.
6827 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
6828 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
6829 * tree-vect-generic.c (add_rshift): Likewise.
6830 (expand_vector_divmod): Likewise.
6831 (optimize_vector_constructor): Likewise.
6832 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
6833 (vect_transform_slp_perm_load): Likewise.
6834 (vect_schedule_slp_instance): Likewise.
6835 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
6836 (vectorizable_call): Likewise.
6837 (vect_gen_perm_mask_any): Likewise. Add elements in order.
6838 * expmed.c (make_tree): Likewise.
6839 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
6840 a vector passed to build_vector.
6841 (fold_convert_const): Likewise.
6842 (exact_inverse): Likewise.
6843 (fold_ternary_loc): Likewise.
6844 (fold_relational_const): Likewise.
6845 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
6846 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
6847 (const_unop): Likewise. Store the reduction accumulator in a
6848 variable rather than an array.
6849 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
6850 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
6851 the new vector, rather than constructing it after the input arrays.
6852 (native_interpret_vector): Use auto_vec<tree> when building
6853 a vector passed to build_vector. Add elements in order.
6854 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
6855 auto_vec<tree> when building a vector passed to build_vector.
6856 (vect_create_epilog_for_reduction): Likewise.
6857 (vectorizable_induction): Likewise.
6858 (get_initial_def_for_reduction): Likewise. Fix indentation of
6860 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
6862 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
6863 passed to build_vector.
6865 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6866 Alan Hayward <alan.hayward@arm.com>
6867 David Sherwood <david.sherwood@arm.com>
6869 * tree-core.h (tree_base::u): Add an "nelts" field.
6870 (tree_vector): Use VECTOR_CST_NELTS as the length.
6871 * tree.c (tree_size): Likewise.
6872 (make_vector): Initialize VECTOR_CST_NELTS.
6873 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
6874 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
6875 TYPE_VECTOR_SUBPARTS.
6876 * expr.c (const_vector_mask_from_tree): Consistently use "units"
6877 as the number of units, setting it from VECTOR_CST_NELTS.
6878 (const_vector_from_tree): Likewise.
6879 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
6880 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
6881 (fold_negate_expr_1): Likewise.
6882 (fold_convert_const): Likewise.
6883 (const_binop): Likewise. Differentiate the number of output and
6885 (const_unop): Likewise.
6886 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
6887 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
6888 in cases that did the opposite.
6890 2017-09-14 Richard Biener <rguenther@suse.de>
6892 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
6895 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
6897 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
6899 2017-09-14 Jakub Jelinek <jakub@redhat.com>
6902 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
6903 if and where to split a bb, except for splitting before debug insn
6904 sequences followed by non-label real insn. Delete debug insns
6905 in between basic blocks.
6907 * combine.c (make_compound_operation_int): Formatting fixes.
6909 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
6910 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6911 * config/netbsd.h (LINK_EH_SPEC): Likewise.
6912 * config/sol2.h (LINK_EH_SPEC): Likewise.
6913 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6914 * config/s390/linux.h (LINK_SPEC): Likewise.
6915 * config/freebsd.h (LINK_EH_SPEC): Likewise.
6916 * config/openbsd.h (LINK_EH_SPEC): Likewise.
6917 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6918 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
6919 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
6920 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6921 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
6922 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
6924 2017-09-13 Jakub Jelinek <jakub@redhat.com>
6926 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
6928 (ENDFILE_LINUX_SPEC): Likewise.
6929 (LINK_EH_SPEC): Likewise.
6930 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
6931 (LINK_OS_LINUX_SPEC32): Likewise.
6932 (LINK_OS_LINUX_SPEC64): Likewise.
6933 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
6934 (LINK_OS_LINUX_SPEC): Likewise.
6936 2017-09-13 Martin Liska <mliska@suse.cz>
6939 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
6940 CASE_HIGH is NULL_TREE.
6942 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
6943 Alan Hayward <alan.hayward@arm.com>
6944 David Sherwood <david.sherwood@arm.com>
6946 * target.def (secondary_memory_needed): New hook.
6947 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
6948 instead of SECONDARY_MEMORY_NEEDED.
6949 (secondary_memory_needed_mode): Likewise.
6950 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
6951 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
6952 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
6953 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
6954 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
6955 * doc/tm.texi: Regenerate.
6956 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
6957 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
6958 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6959 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
6960 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
6961 * config/i386/i386.c (inline_secondary_memory_needed): Put the
6962 mode argument first and change the reg_class arguments to reg_class_t.
6963 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
6964 Make static. Update the call to inline_secondary_memory_needed.
6965 (ix86_register_move_cost): Update the call to
6966 inline_secondary_memory_needed.
6967 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6968 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
6970 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
6971 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6973 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
6974 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
6975 * config/mips/mips.c (mips_secondary_memory_needed): Make static
6976 and match hook interface. Add comment from mips.h.
6977 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6978 * config/mmix/mmix.md (truncdfsf2): Refer to
6979 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6981 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
6982 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
6983 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6984 (pa_secondary_memory_needed): New function.
6985 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
6986 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
6987 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6988 (pdp11_secondary_memory_needed): Make static and match hook interface.
6989 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
6990 * config/powerpcspe/powerpcspe-protos.h
6991 (rs6000_secondary_memory_needed_ptr): Delete.
6992 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
6994 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6995 (rs6000_option_override_internal): Assign to
6996 targetm.secondary_memory_needed rather than
6997 rs6000_secondary_memory_needed_ptr.
6998 (rs6000_secondary_memory_needed): Match hook interface.
6999 (rs6000_debug_secondary_memory_needed): Likewise.
7000 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7001 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7002 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7003 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7004 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7005 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7007 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7008 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7009 (rs6000_option_override_internal): Assign to
7010 targetm.secondary_memory_needed rather than
7011 rs6000_secondary_memory_needed_ptr.
7012 (rs6000_secondary_memory_needed): Match hook interface.
7013 (rs6000_debug_secondary_memory_needed): Likewise.
7014 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7015 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7016 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7017 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7018 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7019 (sparc_secondary_memory_needed): New function.
7020 * lra-constraints.c (check_and_process_move): Refer to
7021 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7023 (curr_insn_transform): Likewise.
7024 (process_alt_operands): Use targetm.secondary_memory_needed
7025 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7026 (check_secondary_memory_needed_p): Likewise.
7027 (choose_split_class): Likewise.
7028 * reload.c: Unconditionally include code that was previously
7029 conditional on SECONDARY_MEMORY_NEEDED.
7030 (push_secondary_reload): Use targetm.secondary_memory_needed
7031 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7032 (push_reload): Likewise.
7033 * reload1.c: Unconditionally include code that was previously
7034 conditional on SECONDARY_MEMORY_NEEDED.
7035 (choose_reload_regs): Use targetm.secondary_memory_needed
7036 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7037 (gen_reload): Likewise.
7038 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7040 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7041 Alan Hayward <alan.hayward@arm.com>
7042 David Sherwood <david.sherwood@arm.com>
7044 * target.def (secondary_memory_needed_mode): New hook:
7045 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7046 * targhooks.h (default_secondary_memory_needed_mode): New function.
7047 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7048 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7049 * doc/tm.texi: Regenerate.
7050 * lra-constraints.c (check_and_process_move): Use
7051 targetm.secondary_memory_needed_mode instead of
7052 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7053 (curr_insn_transform): Likewise.
7054 * reload.c (get_secondary_mem): Likewise.
7055 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7056 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7058 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7059 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7060 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7061 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7062 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7064 * config/powerpcspe/powerpcspe-protos.h
7065 (rs6000_secondary_memory_needed_mode): Delete.
7066 * config/powerpcspe/powerpcspe.c
7067 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7068 (rs6000_secondary_memory_needed_mode): Make static.
7069 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7070 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7072 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7074 (rs6000_secondary_memory_needed_mode): Make static.
7075 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7076 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7077 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7078 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7079 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7081 (sparc_secondary_memory_needed_mode): New function.
7082 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7084 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7086 * config/aarch64/constraints.md (Umq): New constraint.
7087 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7089 (mov<mode>): Update condition.
7091 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7093 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7094 when bitposition is the same.
7096 2017-09-13 Richard Biener <rguenther@suse.de>
7098 * dwarf2out.c (output_die_symbol): Remove.
7099 (output_die): Do not output a DIEs symbol.
7101 2017-09-13 Richard Biener <rguenther@suse.de>
7104 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7105 default-def to avoid breaking iterator update with the weird
7106 interaction with cgraph_update_edges_for_call_stmt_node.
7108 2017-09-13 Richard Biener <rguenther@suse.de>
7110 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7111 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7112 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7113 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7115 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7117 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7118 Disable pc relative literal load irrespective of
7119 TARGET_FIX_ERR_A53_84341 for default.
7121 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7123 * config/sparc/sparc.c (output_return): Output the source location of
7124 the insn in the delay slot, if any.
7125 (output_sibcall): Likewise.
7127 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7130 * common.opt (-static-pie): New alias.
7131 (shared): Negate static-pie.
7132 (-no-pie): Update help text.
7134 (static-pie): New option.
7135 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7136 -static-pie support.
7137 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7138 (LINK_EH_SPEC): Likewise.
7139 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7140 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7141 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7142 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7143 (init_gcc_specs): Likewise.
7144 (init_spec): Likewise.
7145 (display_help): Update help message for -pie.
7146 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7149 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7151 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7152 (movdi_aarch64): Likewise.
7153 (movti_aarch64): Likewise.
7155 2017-09-12 Simon Wright <simon@pushface.org>
7158 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7159 calculation of the minor version, always output as 0.
7161 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7164 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7165 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7166 on it early, rather than manual conversion late. For
7167 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7168 instead of performing manual conversion.
7170 2017-09-12 Carl Love <cel@us.ibm.com>
7172 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7173 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7175 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7176 VMULOSW): Add definitions.
7177 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7178 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7179 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7180 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7181 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7183 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7185 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7187 (movti_aarch64): Likewise.
7188 (movdf_aarch64): Likewise.
7189 (movtf_aarch64): Likewise.
7190 (load_pairdi): Likewise.
7191 (store_pairdi): Likewise.
7192 (load_pairdf): Likewise.
7193 (store_pairdf): Likewise.
7194 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7195 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7196 (ldr_got_small_<mode>): Likewise.
7197 (ldr_got_small_28k_<mode>): Likewise.
7198 (ldr_got_tiny): Likewise.
7199 * config/aarch64/iterators.md (ldst_sz): New.
7200 (ldpstp_sz): Likewise.
7201 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7203 (thunderx_load): Split load_8 to load_16.
7204 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7206 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7207 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7208 (xgene1_store_pair): Split store_8 to store_16.
7209 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7210 (falkor_st_0_st_sd): Split store_8 to store_16.
7212 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7214 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7215 and store1/2/3/4 to store_4/8/12/16.
7216 * config/aarch64/aarch64.md: Update for rename.
7217 * config/arm/arm.md: Likewise.: Likewise.
7218 * config/arm/arm.c: Likewise.
7219 * config/arm/thumb1.md: Likewise.
7220 * config/arm/thumb2.md: Likewise.
7221 * config/arm/vfp.md: Likewise.
7222 * config/arm/arm-generic.md: Likewise.
7223 * config/arm/arm1020e.md: Likewise.
7224 * config/arm/arm1026ejs.md: Likewise.
7225 * config/arm/arm1136jfs.md: Likewise.
7226 * config/arm/arm926ejs.md: Likewise.
7227 * config/arm/cortex-a15.md: Likewise.
7228 * config/arm/cortex-a17.md: Likewise.
7229 * config/arm/cortex-a5.md: Likewise.
7230 * config/arm/cortex-a53.md: Likewise.
7231 * config/arm/cortex-a57.md: Likewise.
7232 * config/arm/cortex-a7.md: Likewise.
7233 * config/arm/cortex-a8.md: Likewise.
7234 * config/arm/cortex-a9.md: Likewise.
7235 * config/arm/cortex-m4.md: Likewise.
7236 * config/arm/cortex-m7.md: Likewise.
7237 * config/arm/cortex-r4.md: Likewise.
7238 * config/arm/exynos-m1.md: Likewise.
7239 * config/arm/fa526.md: Likewise.
7240 * config/arm/fa606te.md: Likewise.
7241 * config/arm/fa626te.md: Likewise.
7242 * config/arm/fa726te.md: Likewise.
7243 * config/arm/fmp626.md: Likewise.
7244 * config/arm/iwmmxt.md: Likewise.
7245 * config/arm/ldmstm.md: Likewise.
7246 * config/arm/marvell-pj4.md: Likewise.
7247 * config/arm/xgene1.md: Likewise.
7248 * config/aarch64/thunderx.md: Likewise.
7249 * config/aarch64/thunderx2t99.md: Likewise.
7250 * config/aarch64/falkor.md: Likewise.
7252 2017-09-12 Martin Liska <mliska@suse.cz>
7254 * attribs.c (private_lookup_attribute): New function.
7255 * attribs.h (private_lookup_attribute): Declared here.
7256 (lookup_attribute): Called from this place.
7258 2017-09-12 Richard Biener <rguenther@suse.de>
7260 PR tree-optimization/82157
7261 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
7262 stmts with side-effects.
7264 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7265 Alan Hayward <alan.hayward@arm.com>
7266 David Sherwood <david.sherwood@arm.com>
7268 * target.def (hard_regno_nregs): New hook.
7269 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
7270 * targhooks.h (default_hard_regno_nregs): Declare.
7271 * targhooks.c (default_hard_regno_nregs): New function.
7272 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
7273 (TARGET_HARD_REGNO_NREGS): ...this hook.
7274 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
7275 (CLASS_MAX_NREGS): Likewise.
7276 * doc/tm.texi: Regenerate.
7277 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
7278 instead of HARD_REGNO_NREGS.
7279 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
7280 HARD_REGNO_NREGS in the comment.
7281 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
7282 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
7283 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
7284 Return an unsigned int.
7285 (TARGET_HARD_REGNO_NREGS): Redefine.
7286 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
7287 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
7288 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7289 (arc_hard_regno_nregs): New function.
7290 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
7291 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
7292 (arm_hard_regno_nregs): New function.
7293 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
7294 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
7295 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
7296 (TARGET_HARD_REGNO_NREGS): Redefine.
7297 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
7298 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
7299 (HARD_REGNO_NREGS): Delete.
7300 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
7301 (cr16_hard_regno_nregs): New function.
7302 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
7303 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
7304 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
7305 (cris_hard_regno_nregs): New function.
7306 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
7307 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
7308 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7309 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
7310 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
7311 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
7312 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
7313 (frv_hard_regno_nregs): Make static. Take and return an
7315 (frv_class_max_nregs): Remove outdated copy of documentation.
7316 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
7317 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
7318 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
7319 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
7320 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
7321 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
7322 (TARGET_HARD_REGNO_NREGS): Redefine.
7323 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
7324 (CLASS_MAX_NREGS): Update comment.
7325 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
7326 (ia64_hard_regno_nregs): New function.
7327 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
7328 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
7329 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
7330 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
7331 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
7333 (m32c_hard_regno_nregs): Likewise. Make static.
7334 (TARGET_HARD_REGNO_NREGS): Redefine.
7335 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
7336 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
7337 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
7338 (m68k_hard_regno_nregs): New function.
7339 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
7340 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
7341 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
7342 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
7343 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
7344 Take and return an unsigned int.
7345 (TARGET_HARD_REGNO_NREGS): Redefine.
7346 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
7347 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7348 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
7349 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
7350 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
7351 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
7352 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
7353 (msp430_hard_regno_nregs): Make static. Take and return an
7355 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
7356 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
7357 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
7358 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
7359 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
7360 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
7361 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
7362 (TARGET_HARD_REGNO_NREGS): Redefine.
7363 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
7364 (PA_HARD_REGNO_NREGS): ...this.
7365 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
7366 (PA_HARD_REGNO_NREGS): ...this.
7367 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7368 (pa_hard_regno_nregs): New function.
7369 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
7370 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
7371 (pdp11_hard_regno_nregs): New function.
7372 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
7373 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
7374 (rs6000_hard_regno_nregs_hook): New function.
7375 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
7376 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7377 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
7378 Take and return an unsigned int. Move earlier in file.
7379 (TARGET_HARD_REGNO_NREGS): Redefine.
7380 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
7381 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
7382 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
7383 (rl78_hard_regno_nregs): Make static. Take and return an
7385 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
7386 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
7387 (rs6000_hard_regno_nregs_hook): New function.
7388 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
7389 * config/rx/rx.c (rx_hard_regno_nregs): New function.
7390 (TARGET_HARD_REGNO_NREGS): Redefine.
7391 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
7392 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
7393 instead of HARD_REGNO_NREGS.
7394 (s390_hard_regno_nregs): New function.
7395 (s390_hard_regno_mode_ok): Add comment from s390.h.
7396 (TARGET_HARD_REGNO_NREGS): Redefine.
7397 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
7398 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
7399 (sh_hard_regno_nregs): New function.
7400 (sh_pass_in_reg_p): Use it.
7401 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
7402 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7403 (sparc_hard_regno_nregs): New function.
7404 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
7405 * config/spu/spu.c (spu_hard_regno_nregs): New function.
7406 (spu_function_arg_advance): Use it, supplying a valid register number.
7407 (TARGET_HARD_REGNO_NREGS): Redefine.
7408 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
7409 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
7410 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
7411 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
7412 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
7413 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
7414 (CLASS_MAX_NREGS): Remove copy of old documentation.
7415 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
7416 (visium_hard_regno_nregs): New function.
7417 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
7418 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
7419 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7420 xtensa_hard_regno_nregs): New function.
7421 * system.h (HARD_REGNO_NREGS): Poison.
7423 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7425 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
7426 hard_regno_nregs instead of HARD_REGNO_NREGS.
7427 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
7428 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
7429 (c6x_expand_epilogue): Likewise.
7430 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
7431 (frv_read_iacc_argument): Likewise.
7432 * config/sh/sh.c: Include regs.h.
7433 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7434 (regs_used): Likewise.
7435 (output_stack_adjust): Likewise.
7436 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
7437 * expmed.c: Include regs.h.
7438 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7439 * ree.c: Include regs.h.
7440 (combine_reaching_defs): Use hard_regno_nregs instead of
7442 (add_removable_extension): Likewise.
7444 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7446 * regs.h (hard_regno_nregs): Turn into a function.
7447 (end_hard_regno): Update accordingly.
7448 * caller-save.c (setup_save_areas): Likewise.
7449 (save_call_clobbered_regs): Likewise.
7450 (replace_reg_with_saved_mem): Likewise.
7451 (insert_restore): Likewise.
7452 (insert_save): Likewise.
7453 * combine.c (can_change_dest_mode): Likewise.
7454 (move_deaths): Likewise.
7455 (distribute_notes): Likewise.
7456 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
7457 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
7458 (rs6000_split_multireg_move): Likewise.
7459 (rs6000_register_move_cost): Likewise.
7460 (rs6000_memory_move_cost): Likewise.
7461 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
7462 (rs6000_split_multireg_move): Likewise.
7463 (rs6000_register_move_cost): Likewise.
7464 (rs6000_memory_move_cost): Likewise.
7465 * cselib.c (cselib_reset_table): Likewise.
7466 (cselib_lookup_1): Likewise.
7467 * emit-rtl.c (set_mode_and_regno): Likewise.
7468 * function.c (aggregate_value_p): Likewise.
7469 * ira-color.c (setup_profitable_hard_regs): Likewise.
7470 (check_hard_reg_p): Likewise.
7471 (calculate_saved_nregs): Likewise.
7472 (assign_hard_reg): Likewise.
7473 (improve_allocation): Likewise.
7474 (calculate_spill_cost): Likewise.
7475 * ira-emit.c (modify_move_list): Likewise.
7476 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
7477 (ira_hard_reg_in_set_p): Likewise.
7478 * ira.c (setup_reg_mode_hard_regset): Likewise.
7479 (clarify_prohibited_class_mode_regs): Likewise.
7480 (check_allocation): Likewise.
7481 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7482 (lra_setup_reg_renumber): Likewise.
7483 (setup_try_hard_regno_pseudos): Likewise.
7484 (spill_for): Likewise.
7485 (assign_hard_regno): Likewise.
7486 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
7487 * lra-constraints.c (in_class_p): Likewise.
7488 (lra_constraint_offset): Likewise.
7489 (simplify_operand_subreg): Likewise.
7490 (lra_constraints): Likewise.
7491 (split_reg): Likewise.
7492 (split_if_necessary): Likewise.
7493 (invariant_p): Likewise.
7494 (inherit_in_ebb): Likewise.
7495 * lra-lives.c (process_bb_lives): Likewise.
7496 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
7497 (get_hard_regs): Likewise.
7498 (do_remat): Likewise.
7499 * lra-spills.c (assign_spill_hard_regs): Likewise.
7500 * mode-switching.c (create_pre_exit): Likewise.
7501 * postreload.c (reload_combine_recognize_pattern): Likewise.
7502 * recog.c (peep2_find_free_register): Likewise.
7503 * regcprop.c (kill_value_regno): Likewise.
7504 (set_value_regno): Likewise.
7505 (copy_value): Likewise.
7506 (maybe_mode_change): Likewise.
7507 (find_oldest_value_reg): Likewise.
7508 (copyprop_hardreg_forward_1): Likewise.
7509 * regrename.c (check_new_reg_p): Likewise.
7510 (regrename_do_replace): Likewise.
7511 * reload.c (push_reload): Likewise.
7512 (combine_reloads): Likewise.
7513 (find_dummy_reload): Likewise.
7514 (operands_match_p): Likewise.
7515 (find_reloads): Likewise.
7516 (find_equiv_reg): Likewise.
7517 (reload_adjust_reg_for_mode): Likewise.
7518 * reload1.c (count_pseudo): Likewise.
7519 (count_spilled_pseudo): Likewise.
7520 (find_reg): Likewise.
7521 (clear_reload_reg_in_use): Likewise.
7522 (free_for_value_p): Likewise.
7523 (allocate_reload_reg): Likewise.
7524 (choose_reload_regs): Likewise.
7525 (reload_adjust_reg_for_temp): Likewise.
7526 (emit_reload_insns): Likewise.
7527 (delete_output_reload): Likewise.
7528 * rtlanal.c (subreg_get_info): Likewise.
7529 * sched-deps.c (sched_analyze_reg): Likewise.
7530 * sel-sched.c (init_regs_for_mode): Likewise.
7531 (mark_unavailable_hard_regs): Likewise.
7532 (choose_best_reg_1): Likewise.
7533 (verify_target_availability): Likewise.
7534 * valtrack.c (dead_debug_insert_temp): Likewise.
7535 * var-tracking.c (track_loc_p): Likewise.
7536 (emit_note_insn_var_location): Likewise.
7537 * varasm.c (make_decl_rtl): Likewise.
7538 * reginfo.c (choose_hard_reg_mode): Likewise.
7539 (init_reg_modes_target): Refer directly to
7540 this_target_regs->x_hard_regno_nregs.
7542 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7544 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
7545 instead of hard_regno_nregs.
7547 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7549 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
7550 end_hard_regno instead of hard_regno_nregs.
7551 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
7552 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
7553 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
7554 * ira-color.c (improve_allocation): Likewise.
7555 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7556 * lra-lives.c (mark_regno_live): Likewise.
7557 (mark_regno_dead): Likewise.
7558 * lra-remat.c (operand_to_remat): Likewise.
7559 * lra.c (collect_non_operand_hard_regs): Likewise.
7560 * postreload.c (reload_combine_note_store): Likewise.
7561 (move2add_valid_value_p): Likewise.
7562 * reload.c (regno_clobbered_p): Likewise.
7564 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7566 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
7568 * config/v850/v850.c (v850_reorg): Likewise.
7569 * reload.c (refers_to_regno_for_reload_p): Likewise.
7570 (find_equiv_reg): Likewise.
7571 * reload1.c (reload_reg_reaches_end_p): Likewise.
7573 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7575 * caller-save.c (add_used_regs): Use REG_NREGS instead of
7577 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
7578 * config/arm/arm.c (output_move_neon): Likewise.
7579 (arm_attr_length_move_neon): Likewise.
7580 (neon_split_vcombine): Likewise.
7581 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
7582 (c6x_mark_reg_written): Likewise.
7583 (c6x_dwarf_register_span): Likewise.
7584 * config/i386/i386.c (ix86_save_reg): Likewise.
7585 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
7586 (rws_access_reg): Likewise.
7587 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
7588 * mode-switching.c (create_pre_exit): Likewise.
7589 * ree.c (combine_reaching_defs): Likewise.
7590 (add_removable_extension): Likewise.
7591 * regcprop.c (find_oldest_value_reg): Likewise.
7592 (copyprop_hardreg_forward_1): Likewise.
7593 * reload.c (reload_inner_reg_of_subreg): Likewise.
7594 (push_reload): Likewise.
7595 (combine_reloads): Likewise.
7596 (find_dummy_reload): Likewise.
7597 (reload_adjust_reg_for_mode): Likewise.
7598 * reload1.c (find_reload_regs): Likewise.
7599 (forget_old_reloads_1): Likewise.
7600 (reload_reg_free_for_value_p): Likewise.
7601 (reload_adjust_reg_for_temp): Likewise.
7602 (emit_reload_insns): Likewise.
7603 (delete_output_reload): Likewise.
7604 * sel-sched.c (choose_best_reg_1): Likewise.
7605 (choose_best_pseudo_reg): Likewise.
7607 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7608 Alan Hayward <alan.hayward@arm.com>
7609 David Sherwood <david.sherwood@arm.com>
7611 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
7612 * target.def (slow_unaligned_access): New hook.
7613 * targhooks.h (default_slow_unaligned_access): Declare.
7614 * targhooks.c (default_slow_unaligned_access): New function.
7615 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
7616 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
7617 * doc/tm.texi: Regenerate.
7618 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
7619 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
7620 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
7622 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
7623 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
7625 (rs6000_slow_unaligned_access): New function.
7626 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
7627 (expand_block_compare): Likewise.
7628 (expand_strn_compare): Likewise.
7629 (rs6000_rtx_costs): Likewise.
7630 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
7631 (riscv_slow_unaligned_access): Likewise.
7632 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
7633 (riscv_slow_unaligned_access_p): ...this and make static.
7634 (riscv_option_override): Update accordingly.
7635 (riscv_slow_unaligned_access): New function.
7636 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
7637 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
7638 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
7639 (rs6000_slow_unaligned_access): New function.
7640 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
7641 (rs6000_rtx_costs): Likewise.
7642 * config/rs6000/rs6000-string.c (expand_block_compare)
7643 (expand_strn_compare): Use targetm.slow_unaligned_access instead
7644 of SLOW_UNALIGNED_ACCESS.
7645 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
7646 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
7647 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
7648 of SLOW_UNALIGNED_ACCESS.
7649 * expmed.c (simple_mem_bitfield_p): Likewise.
7650 * expr.c (alignment_for_piecewise_move): Likewise.
7651 (emit_group_load_1): Likewise.
7652 (emit_group_store): Likewise.
7653 (copy_blkmode_from_reg): Likewise.
7654 (emit_push_insn): Likewise.
7655 (expand_assignment): Likewise.
7656 (store_field): Likewise.
7657 (expand_expr_real_1): Likewise.
7658 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
7659 * lra-constraints.c (simplify_operand_subreg): Likewise.
7660 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7661 * gimple-ssa-store-merging.c: Likewise in block comment at start
7663 * tree-ssa-strlen.c: Include target.h.
7664 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
7665 of SLOW_UNALIGNED_ACCESS.
7666 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
7668 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7670 PR rtl-optimization/82185
7671 * expmed.c (emit_store_flag_int): Only test tem if it has been
7674 2017-09-12 Richard Biener <rguenther@suse.de>
7677 * match.pd ((FTYPE) N CMP CST): Fix typo.
7679 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
7681 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
7683 (mips_near_type_p): Add 'short_call' attribute as a synonym
7685 * doc/extend.texi (short_call): Document new function attribute.
7687 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7690 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
7691 assertion check that in the condition.
7692 (get_atomic_generic_size): Likewise. Before testing if parameter
7693 has pointer type, if it has array type, call for C++
7694 default_conversion to perform array-to-pointer conversion.
7696 2017-09-12 Richard Biener <rguenther@suse.de>
7698 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
7699 for operations we cannot scalarize.
7701 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
7703 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
7704 vectors heap vectors. Clean up comments.
7705 Make visited_bbs a reference.
7706 (profitable_jump_thread_path): Make GC
7707 vectors heap vectors. Clean up comments.
7709 (convert_and_register_jump_thread_path): Make GC vectors heap
7711 (check_subpath_and_update_thread_path): Same. Clean up comments.
7712 Make visited_bbs a reference.
7713 (handle_phi): Abstract common code to to
7714 register_jump_thread_path_if_profitable.
7715 Rename VAR_BB to DEF_BB.
7717 Make GC vectors heap vectors.
7718 Make visited_bbs a reference.
7719 (handle_assignment): Same.
7720 (register_jump_thread_path_if_profitable): New.
7721 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
7723 Make GC vectors heap vectors. Clean up comments.
7724 Make visited_bbs a reference.
7725 (find_jump_threads_backwards): Make visited_bbs live in the stack.
7726 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
7729 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
7732 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
7733 words of E_DImode object are reachable by xtensa_uimm8x4 access.
7735 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
7737 Revert r251800 and r251799.
7739 2017-09-11 Martin Jambor <mjambor@suse.cz>
7742 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
7743 arguments in advance.
7744 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
7745 use it to find predecessor edges.
7746 (naive_outof_ssa): Collect vector of predecessors.
7748 2017-09-08 Jason Merrill <jason@redhat.com>
7750 PR c++/70029 - ICE with ref-qualifier and -flto
7751 * langhooks.h (struct lang_hooks_for_types): Add
7752 copy_lang_qualifiers.
7753 * attribs.c (build_type_attribute_qual_variant): Use it.
7754 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
7756 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
7757 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
7759 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
7762 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
7763 (*mulsi3_sp64): New instruction.
7764 (mulsi3): New expander.
7766 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
7768 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
7770 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7772 * sancov.c: Include memmodel.h.
7774 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
7777 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
7780 2017-09-07 Carl Love <cel@us.ibm.com>
7782 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
7783 the sldi instruction.
7785 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
7787 * sancov.c: Include tm_p.h.
7789 2017-09-07 Jakub Jelinek <jakub@redhat.com>
7792 * output.h (switch_to_other_text_partition): New declaration.
7793 * varasm.c (switch_to_other_text_partition): New function.
7794 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
7795 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
7796 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
7797 to the other text partition before emitting LCL label and switch back
7798 after emitting the word after it.
7800 2017-09-07 Richard Biener <rguenther@suse.de>
7802 * passes.def (pass_split_crit_edges): Remove instance before PRE.
7803 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
7804 critical edges here, after loop init.
7805 (pass_data_pre): Remove PROP_no_crit_edges flags.
7806 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
7807 for valueization of call args to avoid leaking VN_TOP.
7808 (visit_use): Assert we do not visit default defs.
7809 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
7810 Use error_mark_node to more easily detect leaking VN_TOP.
7811 All default-defs are varying, not VN_TOP. Mark them visited.
7812 (run_scc_vn): Make code match comment.
7814 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7816 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
7817 OPTION_MASK_FLOAT128_KEYWORD.
7818 (POWERPC_MASKS): Likewise.
7819 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
7820 support for the -mfloat128-type option, and make -mfloat128
7821 default on PowerPC Linux systems. Define or undefine
7822 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
7823 Define __float128 to be __ieee128 if IEEE 128-bit support is
7824 enabled, or undefine it.
7825 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
7826 Delete defining __FLOAT128_TYPE__.
7827 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
7828 -mfloat128-type option and make -mfloat128 default on PowerPC
7830 (TARGET_FLOAT128_TYPE): Likewise.
7831 (-mfloat128-type): Likewise.
7832 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7833 Delete the -mfloat128-type option and make -mfloat128 default on
7834 PowerPC Linux systems. Always use __ieee128 to be the keyword for
7835 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
7836 128-bit floating point is enabled. Change tests from using
7837 -mfloat128-type to -mfloat128.
7838 (rs6000_mangle_type): Use the correct mangling for the __float128
7839 type even if normal long double is restricted to 64-bits.
7840 (floatn_mode): Enable the _Float128 type by default on VSX Linux
7842 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
7843 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
7844 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
7846 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
7847 documentation for -mfloat128.
7849 2017-09-06 Olivier Hainque <hainque@adacore.com>
7851 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
7853 2017-09-06 Wish Wu <wishwu007@gmail.com>
7854 Jakub Jelinek <jakub@redhat.com>
7856 * asan.c (initialize_sanitizer_builtins): Add
7857 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
7858 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
7859 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
7860 BT_FN_VOID_UINT64_PTR variables.
7861 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
7862 (BT_FN_VOID_UINT16_UINT16): Likewise.
7863 (BT_FN_VOID_UINT32_UINT32): Likewise.
7864 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
7865 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
7866 (BT_FN_VOID_UINT64_PTR): Likewise.
7867 * common.opt (flag_sanitize_coverage): New variable.
7868 (fsanitize-coverage=trace-pc): Remove.
7869 (fsanitize-coverage=): Add.
7870 * flag-types.h (enum sanitize_coverage_code): New enum.
7871 * fold-const.c (fold_range_test): Disable non-short-circuit
7872 optimization if flag_sanitize_coverage.
7873 (fold_truth_andor): Likewise.
7874 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
7875 * opts.c (COVERAGE_SANITIZER_OPT): Define.
7876 (coverage_sanitizer_opts): New array.
7877 (get_closest_sanitizer_option): Add OPTS argument, handle also
7878 OPT_fsanitize_coverage_.
7879 (parse_sanitizer_options): Adjusted to also handle
7880 OPT_fsanitize_coverage_.
7881 (common_handle_option): Add OPT_fsanitize_coverage_.
7882 * sancov.c (instrument_comparison, instrument_switch): New function.
7883 (sancov_pass): Add trace-cmp support.
7884 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
7885 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
7886 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
7887 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
7888 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
7889 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
7890 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
7891 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
7892 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
7893 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
7895 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
7897 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7898 error. Only quit immediately if parsing is complete.
7899 (BEGIN): Initialize fatal_err and parse_done.
7900 (begin fpu, end fpu): Check number of arguments.
7901 (begin arch, end arch): Likewise.
7902 (begin cpu, end cpu): Likewise.
7903 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7904 (optalias): Likewise.
7906 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
7908 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7909 * config/arm/arm-isa.h: Delete. Move definitions to ...
7910 * arm-cpus.in: ... here. Use new feature and fgroup values.
7911 * config/arm/arm.c (arm_option_override): Use lower case for feature
7913 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7914 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7915 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7917 (print_isa_bits_for): New function.
7918 (gen_isa): New function.
7919 (gen_comm_data): Use print_isa_bits_for.
7920 (define feature): New keyword.
7921 (define fgroup): New keyword.
7922 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
7923 (arm-isa.h): Add rule to generate file.
7924 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7925 case for feature bit names.
7927 2017-09-06 Richard Biener <rguenther@suse.de>
7929 * tree-ssa-pre.c (NECESSARY): Remove.
7930 (create_expression_by_pieces): Do not touch pass-local flags.
7931 (insert_into_preds_of_block): Likewise.
7932 (do_pre_regular_insertion): Likewise.
7933 (eliminate_insert): Likewise.
7934 (eliminate_dom_walker::before_dom_children): Likewise.
7935 (fini_eliminate): Do not look at inserted_exprs.
7936 (mark_operand_necessary): Remove.
7937 (remove_dead_inserted_code): Replace with simple work-list
7938 algorithm based on inserted_exprs and SSA uses.
7939 (pass_pre::execute): Re-order fini_eliminate and
7940 remove_dead_inserted_code.
7942 2017-09-06 Olivier Hainque <hainque@adacore.com>
7944 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
7945 for VxWorks 7. Adjust surrounding comments.
7947 2017-09-06 Richard Biener <rguenther@suse.de>
7949 * gimple-ssa-strength-reduction.c
7950 (find_candidates_dom_walker::before_dom_children): Also allow
7953 2017-09-06 Richard Biener <rguenther@suse.de>
7955 PR tree-optimization/82108
7956 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
7957 for gap in the non-permutation SLP case.
7959 2017-09-06 Martin Jambor <mjambor@suse.cz>
7961 PR tree-optimization/82078
7962 * tree-sra.c (sort_and_splice_var_accesses): Move call to
7963 add_access_to_work_queue...
7964 (build_accesses_from_assign): ...here.
7965 (propagate_all_subaccesses): Make sure racc is the group
7966 representative, if there is one.
7968 2017-09-06 Jakub Jelinek <jakub@redhat.com>
7971 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
7974 2017-09-06 Richard Biener <rguenther@suse.de>
7976 * gimple-ssa-strength-reduction.c
7977 (find_candidates_dom_walker::before_doom_children): Use a
7978 type and not a mode check.
7980 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
7983 * config/arm/predicates.md (arm_general_adddi_operand): Create new
7985 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
7987 2017-09-05 Jeff Law <law@redhat.com>
7989 PR tree-optimization/64910
7990 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
7991 cases where we have 3 or more operands.
7993 2017-09-05 Jakub Jelinek <jakub@redhat.com>
7996 * omp-low.c (lower_omp_for): Recompute tree invariant if
7997 gimple_omp_for_initial/final is ADDR_EXPR.
8000 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8001 into gimple val before gimplification fo the COND_EXPR.
8003 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8005 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8006 REGION_COPY argument.
8007 (thread_through_all_blocks): Remove unused argument to
8008 duplicate_thread_path.
8010 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8011 Alan Hayward <alan.hayward@arm.com>
8012 David Sherwood <david.sherwood@arm.com>
8014 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8015 Take a scalar_mode rather than a machine_mode.
8016 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8017 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8018 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8019 (aarch64_gen_adjusted_ldpstp): Likewise.
8020 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8022 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8023 Alan Hayward <alan.hayward@arm.com>
8024 David Sherwood <david.sherwood@arm.com>
8026 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8027 Take a scalar_int_mode instead of a machine_mode.
8028 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8029 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8030 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8031 (aarch64_simd_attr_length_rglist): Delete.
8032 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8033 a scalar_int_mode instead of a machine_mode.
8034 (aarch64_add_offset): Likewise.
8035 (aarch64_internal_mov_immediate): Likewise
8036 (aarch64_add_constant_internal): Likewise.
8037 (aarch64_add_constant): Likewise.
8038 (aarch64_movw_imm): Likewise.
8039 (aarch64_rtx_arith_op_extract_p): Likewise.
8040 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8041 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8042 Remove assert that the mode isn't a vector.
8043 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8044 (aarch64_expand_mov_immediate): Update calls after above changes.
8045 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8046 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8047 (aarch64_move_imm): Likewise.
8048 (aarch64_can_const_movi_rtx_p): Likewise.
8049 (aarch64_strip_extend): Likewise.
8050 (aarch64_extr_rtx_p): Likewise.
8051 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8052 a CONST_INT when the mode parameter is VOIDmode.
8053 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8055 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8057 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8058 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8059 (bitwise_type_for_mode): Update accordingly.
8061 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8063 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8064 * stor-layout.c (mode_for_size_tree): Likewise.
8065 (mode_for_array): Update accordingly.
8066 (layout_decl): Likewise.
8067 (compute_record_mode): Likewise. Only set the mode once.
8069 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8071 * target.def (get_mask_mode): Change return type to opt_mode.
8073 * doc/tm.texi: Regenerate.
8074 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8075 * targhooks.c (default_get_mask_mode): Likewise.
8076 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8077 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8078 targetm.get_mask_mode.
8079 * tree.c (build_truth_vector_type): Likewise.
8081 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8083 * machmode.h (mode_for_vector): Return an opt_mode.
8084 * stor-layout.c (mode_for_vector): Likewise.
8085 (mode_for_int_vector): Update accordingly.
8086 (layout_type): Likewise.
8087 * config/i386/i386.c (emit_memmov): Likewise.
8088 (ix86_expand_set_or_movmem): Likewise.
8089 (ix86_expand_vector_init): Likewise.
8090 (ix86_get_mask_mode): Likewise.
8091 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8093 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8094 * expmed.c (extract_bit_field_1): Likewise.
8095 * expr.c (expand_expr_real_2): Likewise.
8096 * optabs-query.c (can_vec_perm_p): Likewise.
8097 (can_vec_mask_load_store_p): Likewise.
8098 * optabs.c (expand_vec_perm): Likewise.
8099 * targhooks.c (default_get_mask_mode): Likewise.
8100 * tree-vect-stmts.c (vectorizable_store): Likewise.
8101 (vectorizable_load): Likewise.
8102 (get_vectype_for_scalar_type_and_size): Likewise.
8104 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8106 * machmode.h (mode_for_int_vector): New function.
8107 * stor-layout.c (mode_for_int_vector): Likewise.
8108 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8109 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8110 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8111 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8112 (s390_expand_vcond): Likewise.
8114 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8116 * machmode.h (opt_machine_mode): New type.
8117 (opt_mode<T>): Allow construction from anything that can be
8119 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8120 (mode_for_size): Return an opt_machine_mode.
8121 * stor-layout.c (mode_for_size): Likewise.
8122 (mode_for_size_tree): Update call accordingly.
8123 (bitwise_mode_for_mode): Likewise.
8124 (make_fract_type): Likewise.
8125 (make_accum_type): Likewise.
8126 * caller-save.c (replace_reg_with_saved_mem): Update call
8128 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8129 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8130 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8131 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8132 * expmed.c (extract_bit_field_1): Likewise.
8133 * reload.c (get_secondary_mem): Likewise.
8134 * varasm.c (assemble_integer): Likewise.
8135 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8138 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8140 * machmode.h (decimal_float_mode_for_size): New function.
8141 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8142 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8143 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8144 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8145 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8146 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8148 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8150 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8151 (get_builtin_sync_mode): Likewise.
8152 (expand_ifn_atomic_compare_exchange): Likewise.
8153 (expand_builtin_atomic_clear): Likewise.
8154 (expand_builtin_atomic_test_and_set): Likewise.
8155 (fold_builtin_atomic_always_lock_free): Likewise.
8156 * calls.c (compute_argument_addresses): Likewise.
8157 (emit_library_call_value_1): Likewise.
8158 (store_one_arg): Likewise.
8159 * combine.c (combine_instructions): Likewise.
8160 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8161 * config/arm/arm.c (arm_function_value): Likewise.
8162 (aapcs_allocate_return_reg): Likewise.
8163 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8164 * config/i386/i386.c (construct_container): Likewise.
8165 (ix86_gimplify_va_arg): Likewise.
8166 (ix86_expand_sse_cmp): Likewise.
8167 (emit_memmov): Likewise.
8168 (emit_memset): Likewise.
8169 (expand_small_movmem_or_setmem): Likewise.
8170 (ix86_expand_pextr): Likewise.
8171 (ix86_expand_pinsr): Likewise.
8172 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8173 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8175 * config/mips/mips.c (mips_function_value_1) Likewise.
8176 (mips_block_move_straight): Likewise.
8177 (mips_expand_ins_as_unaligned_store): Likewise.
8178 * config/powerpcspe/powerpcspe.c
8179 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8180 (rs6000_darwin64_record_arg_flush): Likewise.
8181 * config/rs6000/rs6000.c
8182 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8183 (rs6000_darwin64_record_arg_flush): Likewise.
8184 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8185 (sparc_function_value_1): Likewise.
8186 * config/spu/spu.c (adjust_operand): Likewise.
8187 (spu_emit_branch_or_set): Likewise.
8188 (arith_immediate_p): Likewise.
8189 * emit-rtl.c (gen_lowpart_common): Likewise.
8190 * expr.c (expand_expr_real_1): Likewise.
8191 * function.c (assign_parm_setup_block): Likewise.
8192 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8193 * reload1.c (alter_reg): Likewise.
8194 * stor-layout.c (mode_for_vector): Likewise.
8195 (layout_type): Likewise.
8197 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8199 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8200 (spu_convert_move): Likewise.
8201 * lower-subreg.c (resolve_simple_move): Likewise.
8203 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8206 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8207 define_insn to a define_expand.
8208 (altivec_vsum2sws_direct): New define_insn.
8209 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8211 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8213 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8216 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8221 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8222 realign stack if stack alignment needed is less than incoming
8225 2017-09-05 Marek Polacek <polacek@redhat.com>
8228 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8231 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8233 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8235 2017-09-05 Richard Biener <rguenther@suse.de>
8237 PR tree-optimization/82084
8238 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8240 2017-09-05 Richard Biener <rguenther@suse.de>
8242 PR tree-optimization/82102
8243 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8245 2017-09-05 Martin Liska <mliska@suse.cz>
8247 PR tree-optimization/82032
8248 * tree-cfg.c (generate_range_test): New function.
8249 * tree-cfg.h (generate_range_test): Declared here.
8250 * tree-cfgcleanup.c (convert_single_case_switch): New function.
8251 (cleanup_control_expr_graph): Use it.
8252 * tree-switch-conversion.c (try_switch_expansion): Remove
8254 (emit_case_nodes): Use generate_range_test.
8256 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8259 * config/i386/i386.md (*<btsc><mode>_mask): Add
8260 TARGET_USE_BT to insn constraint.
8261 (*btr<mode>_mask): Ditto.
8263 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
8265 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
8266 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
8268 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
8271 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
8272 TARGET_NEON and TARGET_IWMMXT.
8273 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
8274 TARGET_NEON and TARGET_IWMMXT.
8275 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
8277 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8279 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
8280 (ix86_rewrite_tls_address): Ditto.
8281 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
8282 (ix86_rewrite_tls_address_1): Ditto.
8283 (ix86_rewrite_tls_address): Ditto.
8284 * config/i386/predicates.md (tls_address_pattern): New predicate.
8285 * config/i386/i386.md (TLS address splitter): New splitter.
8287 2017-09-04 Richard Biener <rguenther@suse.de>
8289 PR tree-optimization/82084
8290 * fold-const.h (can_native_encode_string_p): Declare.
8291 * fold-const.c (can_native_encode_string_p): Factor out from ...
8292 (native_encode_string): ... here.
8293 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
8294 vectorizing stores from constants we later cannot handle.
8296 2017-09-04 Marek Polacek <polacek@redhat.com>
8299 * doc/invoke.texi: Update -Wtautological-compare documentation.
8301 2017-09-04 Jeff Law <law@redhat.com>
8303 PR tree-optimization/64910
8304 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
8305 swap the first and last operand if the last is a constant.
8307 2017-09-04 Marek Polacek <polacek@redhat.com>
8310 * convert.c (do_narrow): When sanitizing signed integer overflows,
8311 bail out for signed types.
8312 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
8314 2017-09-04 Richard Biener <rguenther@suse.de>
8316 PR tree-optimization/82060
8317 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8318 Move devirtualization after stmt folding and before EH/AB/noreturn
8319 cleanup to get the stmt refs canonicalized. Use a bool instead
8320 of gimple_modified_p since that doesn't work for NOPs. Schedule
8321 NOPs generated by folding for removal.
8323 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8324 Alan Hayward <alan.hayward@arm.com>
8325 David Sherwood <david.sherwood@arm.com>
8327 * coretypes.h (pad_direction): New enum.
8328 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
8329 (FUNCTION_ARG_PADDING): Likewise.
8330 * target.def (function_arg_padding): New hook.
8331 * targhooks.h (default_function_arg_padding): Declare.
8332 * targhooks.c (default_function_arg_padding): New function.
8333 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
8334 (TARGET_FUNCTION_ARG_PADDING): ...this.
8335 * doc/tm.texi: Regenerate.
8336 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
8337 instead of direction.
8338 (compute_argument_addresses): Likewise.
8339 (load_register_parameters): Likewise.
8340 (emit_library_call_value_1): Likewise.
8341 (store_one_arg): Use targetm.calls.function_arg_padding instead
8342 of FUNCTION_ARG_PADDING.
8343 (must_pass_in_stack_var_size_or_pad): Likewise.
8344 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
8345 (emit_group_store): Likewise.
8346 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
8347 instead of FUNCTION_ARG_PADDING.
8348 (emit_push_insn): Likewise, and propagate enum change throughout
8350 * function.h (direction): Delete.
8351 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
8353 * function.c (assign_parm_find_stack_rtl): Likewise.
8354 (assign_parm_setup_block_p): Likewise.
8355 (assign_parm_setup_block): Likewise.
8356 (gimplify_parameters): Likewise.
8357 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
8358 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
8360 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
8361 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8362 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
8363 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
8364 (aarch64_function_arg_padding): ...this new function.
8365 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
8366 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8367 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
8368 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8369 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
8370 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8371 (arm_pad_arg_upward): Replace with...
8372 (arm_function_arg_padding): ...this new function.
8373 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
8375 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
8376 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
8377 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8378 (ia64_hpux_function_arg_padding): Replace with...
8379 (ia64_function_arg_padding): ...this new function. Use pad_direction
8380 instead of direction. Check for TARGET_HPUX.
8381 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
8382 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8383 (iq2000_function_arg_padding): New function.
8384 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
8385 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
8386 (mips_function_arg_padding): ...this new function.
8387 (mips_pad_reg_upward): Update accordingly.
8388 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8389 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
8390 targetm.calls.function_arg_padding.
8391 (FUNCTION_ARG_PADDING): Delete.
8392 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8393 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
8394 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8395 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
8396 (nios2_block_reg_padding): Return pad_direction instead of direction.
8397 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
8398 instead of direction.
8399 (nios2_function_arg_padding): Likewise. Make static.
8400 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8401 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
8402 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
8403 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
8404 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8405 (pa_function_arg_padding): Make static. Return pad_direction instead
8407 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
8408 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8409 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
8410 instead of direction. Use targetm.calls.function_arg_padding.
8411 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
8412 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8413 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
8414 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
8415 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
8417 (function_arg_padding): Rename to...
8418 (rs6000_function_arg_padding): ...this. Make static. Return
8419 pad_direction instead of direction.
8420 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8421 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
8422 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8423 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
8424 instead of direction. Use targetm.calls.function_arg_padding.
8425 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
8426 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8427 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
8428 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
8429 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8430 (function_arg_padding): Rename to...
8431 (rs6000_function_arg_padding): ...this. Make static. Return
8432 pad_direction instead of direction.
8433 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8434 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
8435 * config/s390/s390.c (s390_function_arg_padding): New function.
8436 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8437 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
8438 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
8439 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8440 (function_arg_padding): Rename to...
8441 (sparc_function_arg_padding): ...this. Make static. Return
8442 pad_direction instead of direction.
8443 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
8444 * config/spu/spu.c (spu_function_arg_padding): New function.
8445 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8446 * system.h (FUNCTION_ARG_PADDING): Poison.
8448 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8449 Alan Hayward <alan.hayward@arm.com>
8450 David Sherwood <david.sherwood@arm.com>
8452 * target.def (modes_tieable_p): New hook.
8453 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
8454 (TARGET_MODES_TIEABLE_P): ...this.
8455 * doc/tm.texi.in: Regenerate.
8456 * hooks.h (hook_bool_mode_mode_true): Declare.
8457 * hooks.c (hook_bool_mode_mode_true): New function.
8458 * combine.c (subst): Use targetm.modes_tieable_p instead of
8460 * dse.c (find_shift_sequence): Likewise.
8461 * expmed.c (extract_low_bits): Likewise.
8462 * lower-subreg.c: Include target.h.
8463 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
8465 * rtlanal.c (rtx_cost): Likewise.
8466 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
8467 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
8468 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
8469 (TARGET_MODES_TIEABLE_P): Redefine.
8470 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
8471 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
8472 (TARGET_MODES_TIEABLE_P): Redefine.
8473 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
8474 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
8475 (arc_modes_tieable_p): New function.
8476 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
8477 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
8478 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
8479 (arm_modes_tieable_p): Make static.
8480 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
8481 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
8482 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
8483 (TARGET_MODES_TIEABLE_P): Redefine.
8484 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
8485 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
8486 (TARGET_MODES_TIEABLE_P): Redefine.
8487 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
8488 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
8489 (cr16_modes_tieable_p): New function.
8490 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
8491 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
8492 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
8493 (TRULY_NOOP_TRUNCATION): Update comment.
8494 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
8495 (TRULY_NOOP_TRUNCATION): Update comment.
8496 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
8497 (frv_modes_tieable_p): New function.
8498 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
8499 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
8500 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
8501 (TARGET_MODES_TIEABLE_P): Redefine.
8502 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
8503 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
8504 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
8505 (TARGET_MODES_TIEABLE_P): Redefine.
8506 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
8507 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
8508 (ia64_modes_tieable_p): New function.
8509 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
8510 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
8511 (iq2000_modes_tieable_p): New function.
8512 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
8513 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
8514 (lm32_modes_tieable_p): New function.
8515 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
8516 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
8517 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
8518 (TARGET_MODES_TIEABLE_P): Redefine.
8519 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
8520 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
8521 (m32r_modes_tieable_p): New function.
8522 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
8523 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
8524 (m68k_modes_tieable_p): New function.
8525 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
8526 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
8527 (mcore_modes_tieable_p): New function.
8528 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
8529 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
8531 (TARGET_MODES_TIEABLE_P): Redefine.
8532 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
8533 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
8534 * config/mips/mips.c (mips_modes_tieable_p): Make static.
8535 (TARGET_MODES_TIEABLE_P): Redefine.
8536 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
8537 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
8538 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
8539 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
8540 (mn10300_modes_tieable_p): ...this and make static.
8541 (TARGET_MODES_TIEABLE_P): Redefine.
8542 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
8543 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
8544 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
8545 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
8546 (msp430_modes_tieable_p): Make static.
8547 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
8548 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
8549 (TARGET_MODES_TIEABLE_P): Redefine.
8550 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
8551 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
8552 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
8553 (TARGET_MODES_TIEABLE_P): Redefine.
8554 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
8555 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
8556 * config/pa/pa.c (pa_modes_tieable_p): Make static.
8557 (TARGET_MODES_TIEABLE_P): Redefine.
8558 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
8559 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
8560 (pdp11_modes_tieable_p): New function.
8561 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
8562 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
8563 (rs6000_modes_tieable_p): New function.
8564 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8565 * config/powerpcspe/powerpcspe.md: Update comment.
8566 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
8567 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
8568 (TARGET_MODES_TIEABLE_P): Redefine.
8569 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
8570 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
8571 (rl78_modes_tieable_p): New function.
8572 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
8573 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
8574 (rs6000_modes_tieable_p): New function.
8575 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8576 * config/rs6000/rs6000.md: Update comment.
8577 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
8578 * config/rx/rx.c (rx_modes_tieable_p): New function.
8579 (TARGET_MODES_TIEABLE_P): Redefine.
8580 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
8581 * config/s390/s390.c (s390_modes_tieable_p): New function.
8582 (TARGET_MODES_TIEABLE_P): Redefine.
8583 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
8584 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
8585 (sh_modes_tieable_p): New function.
8586 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
8587 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
8588 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
8589 (sparc_modes_tieable_p): Make static.
8590 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
8591 * config/spu/spu.c (spu_modes_tieable_p): New function.
8592 (TARGET_MODES_TIEABLE_P): Redefine.
8593 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
8594 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
8595 (TARGET_MODES_TIEABLE_P): Redefine.
8596 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
8597 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
8598 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
8599 * config/v850/v850.c (v850_modes_tieable_p): New function.
8600 (TARGET_MODES_TIEABLE_P): Redefine.
8601 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
8602 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
8603 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
8604 (visium_modes_tieable_p): New function.
8605 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
8606 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
8607 (xtensa_modes_tieable_p): New function.
8608 * system.h (MODES_TIEABLE_P): Poison.
8610 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8611 Alan Hayward <alan.hayward@arm.com>
8612 David Sherwood <david.sherwood@arm.com>
8614 * target.def (hard_regno_mode_ok): New hook.
8615 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
8616 (TARGET_HARD_REGNO_MODE_OK): ...this.
8617 * doc/tm.texi.in: Regenerate.
8618 * hooks.h (hook_bool_uint_mode_true): Declare.
8619 * hooks.c (hook_bool_uint_mode_true): New function.
8620 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
8622 * genpreds.c (write_insn_preds_c): Add an include of target.h.
8623 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
8624 instead of HARD_REGNO_MODE_OK.
8625 * caller-save.c: Include target.h.
8626 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
8628 * combine.c (can_combine_p): Likewise.
8629 (combinable_i3pat): Likewise.
8630 (can_change_dest_mode): Likewise.
8631 * expr.c (init_expr_target): Likewise.
8632 (convert_move): Likewise.
8633 (convert_modes): Likewise.
8634 * ira.c (setup_prohibited_class_mode_regs): Likewise.
8635 (setup_prohibited_mode_move_regs): Likewise.
8636 * ira.h (target_ira): Likewise.
8637 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8638 * lra-constraints.c (process_alt_operands): Likewise.
8639 (split_reg): Likewise.
8640 * recog.c (peep2_find_free_register): Likewise.
8641 * ree.c (combine_reaching_defs): Likewise.
8642 * regcprop.c (maybe_mode_change): Likewise.
8643 * reginfo.c (init_reg_sets_1): Likewise.
8644 (choose_hard_reg_mode): Likewise.
8645 (simplifiable_subregs): Likewise.
8646 * regrename.c (check_new_reg_p): Likewise.
8647 * reload.c (find_valid_class): Likewise.
8648 (find_valid_class_1): Likewise.
8649 (reload_inner_reg_of_subreg): Likewise.
8650 (push_reload): Likewise.
8651 (combine_reloads): Likewise.
8652 (find_dummy_reload): Likewise.
8653 (find_reloads): Likewise.
8654 * reload1.c (find_reg): Likewise.
8655 (set_reload_reg): Likewise.
8656 (allocate_reload_reg): Likewise.
8657 (choose_reload_regs): Likewise.
8658 (reload_adjust_reg_for_temp): Likewise.
8659 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
8660 (simplify_subreg_regno): Likewise.
8661 * sel-sched.c (init_regs_for_mode): Likewise.
8662 * varasm.c (make_decl_rtl): Likewise.
8663 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
8664 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
8666 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
8667 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
8668 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8669 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
8670 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
8671 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8672 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
8673 (arc_mode_class): Delete.
8674 (HARD_REGNO_MODE_OK): Delete.
8675 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8676 (arc_hard_regno_mode_ok): Rename old array to...
8677 (arc_hard_regno_mode_ok_modes): ...this.
8678 (arc_conditional_register_usage): Update accordingly.
8679 (arc_mode_class): Make static.
8680 (arc_hard_regno_mode_ok): New function.
8681 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
8682 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
8683 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8684 (arm_hard_regno_mode_ok): Make static.
8685 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
8687 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
8688 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
8689 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
8691 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8692 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
8693 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
8694 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
8695 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
8696 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8697 * config/bfin/predicates.md (valid_reg_operand): Use
8698 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
8699 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
8700 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
8701 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8702 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
8703 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
8704 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8705 (cr16_hard_regno_mode_ok): Make static and return a bool.
8706 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
8707 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8708 (cris_hard_regno_mode_ok): New function.
8709 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
8710 (epiphany_mode_class): Delete.
8711 (HARD_REGNO_MODE_OK): Delete.
8712 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
8713 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8714 (hard_regno_mode_ok): Rename to...
8715 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
8716 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
8717 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
8719 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
8720 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
8721 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8722 (frv_hard_regno_mode_ok): Make static and return a bool.
8723 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
8725 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
8726 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
8727 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
8728 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
8730 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8731 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
8732 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
8733 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
8735 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8736 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
8737 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8738 (ia64_hard_regno_mode_ok): New function.
8739 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
8740 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8741 (iq2000_hard_regno_mode_ok): New function.
8742 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
8743 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8744 (lm32_hard_regno_mode_ok): New function.
8745 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
8746 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
8747 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
8748 instead of HARD_REGNO_MODE_OK.
8749 (m32c_hard_regno_ok): Rename to...
8750 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
8751 (m32c_cannot_change_mode_class): Update accordingly.
8752 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8753 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
8754 (m32r_mode_class): Delete.
8755 (HARD_REGNO_MODE_OK): Delete.
8756 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8757 (m32r_hard_regno_mode_ok): Rename to...
8758 (m32r_hard_regno_modes): ...this.
8759 (m32r_mode_class): Make static.
8760 (m32r_hard_regno_mode_ok): New function.
8761 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
8762 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
8763 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8764 (m68k_hard_regno_mode_ok): Make static.
8765 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
8766 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8767 (mcore_hard_regno_mode_ok): New function.
8768 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
8769 (HARD_REGNO_MODE_OK): Delete.
8770 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
8772 (microblaze_hard_regno_mode_ok_p): ...this and make static.
8773 (microblaze_hard_regno_mode_ok): New function.
8774 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8775 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
8776 (mips_hard_regno_mode_ok): Delete.
8777 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
8778 (mips_hard_regno_mode_ok_p): ...this and make static.
8779 (mips_hard_regno_mode_ok_p): Rename to...
8780 (mips_hard_regno_mode_ok_uncached): ...this.
8781 (mips_hard_regno_mode_ok): New function.
8782 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
8783 of HARD_REGNO_MODE_OK.
8784 (mips_option_override): Update after above name changes.
8785 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8786 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
8787 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
8788 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
8789 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
8790 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8791 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
8792 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
8793 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
8794 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8795 (msp430_hard_regno_mode_ok): Make static and return a bool.
8796 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
8797 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
8798 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
8800 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8801 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
8802 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
8803 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
8804 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
8805 (PA_HARD_REGNO_MODE_OK): ...this
8806 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
8807 (PA_HARD_REGNO_MODE_OK): ...this.
8808 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8809 (pa_hard_regno_mode_ok): New function.
8810 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
8811 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8812 (pdp11_hard_regno_mode_ok): New function.
8813 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
8814 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
8816 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
8818 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8819 (rs6000_hard_regno_mode_ok): Rename to...
8820 (rs6000_hard_regno_mode_ok_uncached): ...this.
8821 (rs6000_init_hard_regno_mode_ok): Update accordingly.
8822 (rs6000_hard_regno_mode_ok): New function.
8823 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
8824 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
8825 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
8826 (riscv_hard_regno_mode_ok): ...this and make static.
8827 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8828 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
8829 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
8830 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8831 (rl78_hard_regno_mode_ok): Make static and return bool.
8832 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
8833 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
8835 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
8836 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8837 (rs6000_hard_regno_mode_ok): Rename to...
8838 (rs6000_hard_regno_mode_ok_uncached): ...this.
8839 (rs6000_init_hard_regno_mode_ok): Update accordingly.
8840 (rs6000_hard_regno_mode_ok): New function.
8841 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
8842 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
8843 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8844 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
8845 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
8846 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
8847 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8848 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
8849 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
8850 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8851 (sh_hard_regno_mode_ok): Make static.
8852 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
8853 instead of HARD_REGNO_MODE_OK.
8854 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
8855 (sparc_mode_class): Delete.
8856 (HARD_REGNO_MODE_OK): Delete.
8857 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8858 (hard_regno_mode_classes): Make static.
8859 (sparc_mode_class): Likewise.
8860 (sparc_hard_regno_mode_ok): New function.
8861 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
8862 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
8863 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
8865 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8866 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
8867 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
8868 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
8869 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
8870 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8871 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
8872 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
8873 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8874 (visium_hard_regno_mode_ok): New function.
8875 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
8876 instead of HARD_REGNO_MODE_OK.
8877 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
8878 (HARD_REGNO_MODE_OK): Delete.
8879 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
8880 (xtensa_hard_regno_mode_ok_p): ...this and make static.
8881 (xtensa_option_override): Update accordingly.
8882 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8883 (xtensa_hard_regno_mode_ok): New function.
8884 * system.h (HARD_REGNO_MODE_OK): Poison.
8886 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8887 Alan Hayward <alan.hayward@arm.com>
8888 David Sherwood <david.sherwood@arm.com>
8890 * target.def (hard_regno_call_part_clobbered): New hook.
8891 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
8892 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
8893 * doc/tm.texi: Regenerate.
8894 * hooks.h (hook_bool_uint_mode_false): Declare.
8895 * hooks.c (hook_bool_uint_mode_false): New function.
8896 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8897 * cselib.c (cselib_process_insn): Use
8898 targetm.hard_regno_call_part_clobbered instead of
8899 HARD_REGNO_CALL_PART_CLOBBERED.
8900 * ira-conflicts.c (ira_build_conflicts): Likewise.
8901 * ira-costs.c (ira_tune_allocno_costs): Likewise.
8902 * lra-constraints.c (need_for_call_save_p): Likewise.
8903 * lra-lives.c: Include target.h.
8904 (check_pseudos_live_through_calls): Use
8905 targetm.hard_regno_call_part_clobbered instead of
8906 HARD_REGNO_CALL_PART_CLOBBERED.
8907 * regcprop.c: Include target.h.
8908 (copyprop_hardreg_forward_1): Use
8909 targetm.hard_regno_call_part_clobbered instead of
8910 HARD_REGNO_CALL_PART_CLOBBERED.
8911 * reginfo.c (choose_hard_reg_mode): Likewise.
8912 * regrename.c (check_new_reg_p): Likewise.
8913 * reload.c (find_equiv_reg): Likewise.
8914 * reload1.c (emit_reload_insns): Likewise.
8915 * sched-deps.c (deps_analyze_insn): Likewise.
8916 * sel-sched.c (init_regs_for_mode): Likewise.
8917 (mark_unavailable_hard_regs): Likewise.
8918 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
8919 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8920 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
8922 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8923 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8924 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
8926 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
8928 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8929 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8930 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
8932 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8933 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8934 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
8936 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8937 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
8939 * config/powerpcspe/powerpcspe.c
8940 (rs6000_hard_regno_call_part_clobbered): New function.
8941 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8942 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8943 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
8945 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8946 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8947 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
8949 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8950 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8951 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
8953 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8954 Alan Hayward <alan.hayward@arm.com>
8955 David Sherwood <david.sherwood@arm.com>
8957 * rtl.h (subreg_memory_offset): Declare.
8958 * emit-rtl.c (subreg_memory_offset): New function.
8959 * expmed.c (store_bit_field_1): Use it.
8960 * expr.c (undefined_operand_subword_p): Likewise.
8961 * simplify-rtx.c (simplify_subreg): Likewise.
8963 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
8965 PR rtl-optimization/57448
8968 * optabs.c (expand_atomic_load): Place compiler memory barriers if
8969 using atomic_load pattern.
8970 (expand_atomic_store): Likewise.
8972 2017-09-04 Jakub Jelinek <jakub@redhat.com>
8975 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
8976 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
8977 handling. Use replace_call_with_value with NULL instead of
8978 gsi_replace, unlink_stmt_vdef and release_defs.
8980 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
8983 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
8985 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8988 * rtl.h (emit_library_call_value_1): Declare.
8989 (emit_library_call): Replace declaration with a series of overloads.
8990 Remove the parameter count argument.
8991 (emit_library_call_value): Likewise.
8992 * calls.c (emit_library_call_value_1): Make global. Replace varargs
8993 with an "rtx_mode_t *".
8994 (emit_library_call_value): Delete.
8995 (emit_library_call): Likewise.
8996 * asan.c (asan_emit_stack_protection): Update calls accordingly.
8997 (asan_emit_allocas_unpoison): Likewise.
8998 * builtins.c (expand_builtin_powi): Likewise.
8999 (expand_asan_emit_allocas_unpoison): Likewise.
9000 * cfgexpand.c (expand_main_function): Likewise.
9001 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9002 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9003 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9004 * config/arm/arm.c (arm_trampoline_init): Likewise.
9005 (arm_call_tls_get_addr): Likewise.
9006 (arm_expand_divmod_libfunc): Likewise.
9007 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9008 (smulsi3_highpart): Likewise.
9009 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9010 (c6x_expand_compare): Likewise.
9011 (c6x_expand_movmem): Likewise.
9012 * config/frv/frv.c (frv_trampoline_init): Likewise.
9013 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9014 (ix86_expand_divmod_libfunc): Likewise.
9015 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9016 (ia64_expand_compare): Likewise.
9017 (ia64_profile_hook): Likewise.
9018 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9019 (nonlocal_goto): Likewise.
9020 (restore_stack_nonlocal): Likewise.
9021 * config/m32r/m32r.c (block_move_call): Likewise.
9022 (m32r_trampoline_init): Likewise.
9023 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9024 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9025 (m68k_call_m68k_read_tp): Likewise.
9026 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9027 (microblaze_expand_divide): Likewise.
9028 * config/mips/mips.h (mips_args): Likewise.
9029 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9030 (MIPS_ICACHE_SYNC): Likewise.
9031 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9032 (nios2_trampoline_init): Likewise.
9033 * config/pa/pa.c (hppa_tls_call): Likewise.
9034 (pa_trampoline_init): Likewise.
9035 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9036 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9037 (expand_strn_compare): Likewise.
9038 (rs6000_generate_compare): Likewise.
9039 (rs6000_expand_float128_convert): Likewise.
9040 (output_profile_hook): Likewise.
9041 (rs6000_trampoline_init): Likewise.
9042 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9043 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9044 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9045 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9046 (rs6000_generate_compare): Likewise.
9047 (rs6000_expand_float128_convert): Likewise.
9048 (output_profile_hook): Likewise.
9049 (rs6000_trampoline_init): Likewise.
9050 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9051 * config/sh/sh.c (sh_trampoline_init): Likewise.
9052 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9053 (sparc_emit_float_lib_cmp): Likewise.
9054 (sparc32_initialize_trampoline): Likewise.
9055 (sparc64_initialize_trampoline): Likewise.
9056 (sparc_profile_hook): Likewise.
9057 * config/spu/spu.c (ea_load_store): Likewise.
9058 * config/spu/spu.md (floatunssidf2): Likewise.
9059 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9060 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9061 * config/visium/visium.c (expand_block_move_4): Likewise.
9062 (expand_block_move_2): Likewise.
9063 (expand_block_move_1): Likewise.
9064 (expand_block_set_4): Likewise.
9065 (expand_block_set_2): Likewise.
9066 (expand_block_set_1): Likewise.
9067 (visium_trampoline_init): Likewise.
9068 (visium_profile_hook): Likewise.
9069 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9070 (xtensa_setup_frame_addresses): Likewise.
9071 (xtensa_trampoline_init): Likewise.
9072 * except.c (sjlj_emit_function_enter): Likewise.
9073 (sjlj_emit_function_exit): Likewise.
9074 * explow.c (allocate_dynamic_stack_space): Likewise.
9075 (probe_stack_range): Likewise.
9076 * expr.c (convert_mode_scalar): Likewise.
9077 * optabs.c (expand_binop): Likewise.
9078 (expand_twoval_binop_libfunc): Likewise.
9079 (expand_unop): Likewise.
9080 (prepare_cmp_insn): Likewise.
9081 (prepare_float_lib_cmp): Likewise.
9082 (expand_float): Likewise.
9083 (expand_fix): Likewise.
9084 (expand_fixed_convert): Likewise.
9085 (maybe_emit_sync_lock_test_and_set): Likewise.
9086 (expand_atomic_compare_and_swap): Likewise.
9087 (expand_mem_thread_fence): Likewise.
9088 (expand_atomic_fetch_op): Likewise.
9090 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9092 * doc/generic.texi (OpenACC): Adjust URL.
9093 * doc/invoke.texi (C Dialect Options): Ditto.
9095 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9097 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9098 predicate for operand 1. Add (m,<S>) constraint.
9099 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9100 Prevent memory operand 1 with register operand 2.
9102 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9104 PR rtl-optimization/82024
9105 * combine.c (try_combine): If the combination result is a PARALLEL,
9106 and we only need to retain the SET in there that would be placed
9107 at I2, check that we can place that at I3 instead, before doing so.
9109 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9112 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9114 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9115 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9118 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9119 Jeff Law <law@redhat.com>
9121 * varasm.c (bss_initializer_p): Do not put constants into .bss
9122 (categorize_decl_for_section): Handle bss_initializer_p returning
9123 false when DECL_INITIAL is NULL.
9125 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9128 * config/s390/s390.c (s390_can_inline_p): New function.
9130 2017-09-01 Jeff Law <law@redhat.com>
9132 PR tree-optimization/82052
9133 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9134 Always initialize the returned slot after a hash table miss
9135 when INSERT is true.
9137 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9139 * config/s390/s390.md (mem_signal_fence): Remove.
9140 * doc/md.texi (mem_signal_fence): Remove.
9141 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9143 * target-insns.def (mem_signal_fence): Remove.
9145 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9148 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9151 * asan.c (create_odr_indicator): Strip name encoding from assembler
9152 name before appending it after __odr_asan_.
9154 2017-09-01 Martin Liska <mliska@suse.cz>
9156 PR tree-optimization/82059
9157 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9158 frequency only when an edge is redirected.
9160 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9162 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9163 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9164 (arc_conditional_register_usage): Remove ARC600 lp_count
9166 (arc_file_start): Emit Tag_ARC_CPU_variation.
9167 (arc_can_use_doloop_p): New conditions to use ZOLs.
9168 (hwloop_fail): New function.
9169 (hwloop_optimize): Likewise.
9170 (hwloop_pattern_reg): Likewise.
9171 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9172 (arc_reorg_loops): New function, calls reorg_loops.
9173 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9174 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9176 (arc_loop_hazard): Remove function, functionality moved into
9178 (arc_hazard): Remove arc_loop_hazard call.
9179 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9180 into hwloop_optimize.
9181 (arc_label_align): Remove ZOL handling.
9182 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9183 * config/arc/arc.md (doloop_begin): Remove pattern.
9184 (doloop_begin_i): Likewise.
9185 (doloop_end_i): Likewise.
9186 (doloop_fallback): Likewise.
9187 (doloop_fallback_m): Likewise.
9188 (doloop_end): Reimplement expand.
9189 (arc_lp): New pattern for LP instruction.
9190 (loop_end): New pattern.
9191 (loop_fail): Likewise.
9192 (decrement_and_branch_until_zero): Likewise.
9193 * config/arc/arc.opt (mlpc-width): New option.
9194 * doc/invoke.texi (mlpc-width): Document option.
9196 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9198 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9199 (arc_ccfsm_advance): Fix checking for delay slots.
9200 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9202 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9204 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9206 (movhi_insn): Update store instruction constraint which are saving
9207 6-bit short immediates.
9208 (movsi_insn): Consider also short scaled load operations.
9209 (zero_extendhisi2_i): Use Usd constraint instead of T.
9210 (extendhisi2_i): Add q constraint.
9211 (arc_clzsi2): Add type and length attributes.
9212 (arc_ctzsi2): Likewise.
9213 * config/arc/constraints.md (Usc): Update constraint, the
9214 assembler can parse two relocations for a single instruction.
9216 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9218 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9219 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9221 2017-08-31 Olivier Hainque <hainque@adacore.com>
9223 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9224 match as powerpc-wrs-vxworks*.
9226 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9228 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9229 register constraint for by-element operand.
9230 (aarch64_mls_elt_merge<mode>): Likewise.
9232 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9234 * config/arc/arc.c (arc_can_follow_jump): Check for short
9237 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9239 * config.gcc: Use g.opt for arc.
9240 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9241 functionality moved to ...
9242 (legitimate_scaled_address_p): New function, ...here.
9243 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9244 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9245 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9247 (arc_override_options): Handle G option.
9248 (arc_output_pic_addr_const): Correct function definition.
9249 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9250 (arc_decl_anon_ns_mem_p): Delete.
9251 (arc_in_small_data_p): Overhaul this function to take into
9252 consideration the value given via G option.
9253 (arc_rewrite_small_data_1): Renamed and corrected old
9254 arc_rewrite_small_data function.
9255 (arc_rewrite_small_data): New function.
9256 (small_data_pattern): Don't use pic_offset_table_rtx.
9257 * config/arc/arc.h (CC1_SPEC): Recognize G option.
9258 * config/arc/simdext.md (movmisalignv2hi): Use
9259 prepare_move_operands function.
9261 (movmisalign*): Likewise.
9262 * doc/invoke.texi (ARC options): Document -G option.
9264 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9266 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
9268 * config/arc/arc.c (arc_print_operand): Output scalled address for
9269 sdata whenever is possible.
9270 (arc_in_small_data_p): Allow sdata for 64bit datum when double
9271 load/stores are available.
9272 (compact_sda_memory_operand): Check for the alignment required by
9273 code density instructions.
9274 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
9276 * config/arc/constraints.md (Usd): Update constraint.
9277 (Us0): New constraint.
9278 (Usc): Update constraint.
9280 2017-08-31 Richard Biener <rguenther@suse.de>
9283 * dwarf2out.c (dwarf2out_early_global_decl): Process each
9286 2017-08-31 Tamar Christina <tamar.christina@arm.com>
9288 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
9289 Resize type_signature.
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/aarch64.c (aarch64_base_register_rtx_p): Only allow
9296 subregs whose inner modes can be stored in GPRs.
9297 (aarch64_classify_index): Likewise.
9299 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9300 Alan Hayward <alan.hayward@arm.com>
9301 David Sherwood <david.sherwood@arm.com>
9303 * config/aarch64/iterators.md (V_cmp_result): Rename to...
9304 (V_INT_EQUIV): ...this.
9305 (v_cmp_result): Rename to...
9306 (v_int_equiv): ...this.
9307 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
9308 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
9309 (copysign<mode>3): Likewise.
9310 (aarch64_simd_bsl<mode>_internal): Likewise.
9311 (aarch64_simd_bsl<mode>): Likewise.
9312 (vec_cmp<mode><mode>): Likewise.
9313 (vcond<mode><mode>): Likewise.
9314 (vcond<v_cmp_mixed><mode>): Likewise.
9315 (vcondu<mode><v_cmp_mixed>): Likewise.
9316 (aarch64_cm<optab><mode>): Likewise.
9317 (aarch64_cmtst<mode>): Likewise.
9318 (aarch64_fac<optab><mode>): Likewise.
9319 (vec_perm_const<mode>): Likewise.
9320 (vcond_mask_<mode><v_cmp_result>): Rename to...
9321 (vcond_mask_<mode><v_int_equiv>): ...this.
9322 (vec_cmp<mode><v_cmp_result>): Rename to...
9323 (vec_cmp<mode><v_int_equiv>): ...this.
9325 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9326 Alan Hayward <alan.hayward@arm.com>
9327 David Sherwood <david.sherwood@arm.com>
9329 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
9331 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
9332 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
9333 (UNSPEC_LD4_DREG): New unspecs.
9334 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
9335 (aarch64_ld2<mode>_dreg_be): Replace with...
9336 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
9338 (aarch64_ld3<mode>_dreg_le)
9339 (aarch64_ld3<mode>_dreg_be): Replace with...
9340 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
9342 (aarch64_ld4<mode>_dreg_le)
9343 (aarch64_ld4<mode>_dreg_be): Replace with...
9344 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
9347 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9349 PR tree-optimization/81987
9350 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
9351 insert an initializer in a location not dominated by the stride
9354 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9356 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
9357 on the entire header of the finally block in the fallthru case.
9359 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9361 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
9363 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
9365 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
9366 rs6000_emit_move_from_cr and call renamed function.
9367 (rs6000_emit_prologue): Call renamed functions.
9368 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
9369 movesi_from_cr, remove volatile CRs.
9371 2017-08-30 Jon Beniston <jon@beniston.com>
9372 Richard Biener <rguenther@suse.de>
9374 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
9375 of VECTOR_MODE_P check.
9376 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
9377 element vector types.
9379 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9381 * df.h (df_read_modify_subreg_p): Remove in favor of...
9382 * rtl.h (read_modify_subreg_p): ...this new function. Take a
9383 const_rtx instead of an rtx.
9384 * cprop.c (local_cprop_find_used_regs): Update accordingly.
9385 * df-problems.c (df_word_lr_mark_ref): Likewise.
9386 * ira-lives.c (mark_pseudo_reg_live): Likewise.
9387 (mark_pseudo_reg_dead): Likewise.
9388 (mark_ref_dead): Likewise.
9389 * reginfo.c (init_subregs_of_mode): Likewise.
9390 * sched-deps.c (sched_analyze_1): Likewise.
9391 * df-scan.c (df_def_record_1): Likewise.
9392 (df_uses_record): Likewise.
9393 (df_read_modify_subreg_p): Remove in favor of...
9394 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
9395 const_rtx instead of an rtx.
9397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9398 Alan Hayward <alan.hayward@arm.com>
9399 David Sherwood <david.sherwood@arm.com>
9401 * rtl.h (partial_subreg_p): New function.
9402 * caller-save.c (save_call_clobbered_regs): Use it.
9403 * calls.c (expand_call): Likewise.
9404 * combine.c (combinable_i3pat): Likewise.
9405 (simplify_set): Likewise.
9406 (make_extraction): Likewise.
9407 (make_compound_operation_int): Likewise.
9408 (gen_lowpart_or_truncate): Likewise.
9409 (force_to_mode): Likewise.
9410 (make_field_assignment): Likewise.
9411 (reg_truncated_to_mode): Likewise.
9412 (record_truncated_value): Likewise.
9413 (move_deaths): Likewise.
9414 * cse.c (record_jump_cond): Likewise.
9415 (cse_insn): Likewise.
9416 * cselib.c (cselib_lookup_1): Likewise.
9417 * expmed.c (extract_bit_field_using_extv): Likewise.
9418 * function.c (assign_parm_setup_reg): Likewise.
9419 * ifcvt.c (noce_convert_multiple_sets): Likewise.
9420 * ira-build.c (create_insn_allocnos): Likewise.
9421 * lra-coalesce.c (merge_pseudos): Likewise.
9422 * lra-constraints.c (match_reload): Likewise.
9423 (simplify_operand_subreg): Likewise.
9424 (curr_insn_transform): Likewise.
9425 * lra-lives.c (process_bb_lives): Likewise.
9426 * lra.c (new_insn_reg): Likewise.
9427 (lra_substitute_pseudo): Likewise.
9428 * regcprop.c (mode_change_ok): Likewise.
9429 (maybe_mode_change): Likewise.
9430 (copyprop_hardreg_forward_1): Likewise.
9431 * reload.c (push_reload): Likewise.
9432 (find_reloads): Likewise.
9433 (find_reloads_subreg_address): Likewise.
9434 * reload1.c (alter_reg): Likewise.
9435 (eliminate_regs_1): Likewise.
9436 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9438 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
9440 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
9441 back to if statements, including unpack.
9443 2017-08-30 Martin Liska <mliska@suse.cz>
9446 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
9448 (func_checker::compare_asm_inputs_outputs): ... this function.
9449 (func_checker::compare_gimple_asm): Use the function to compare
9450 also ASM constrains.
9451 * ipa-icf-gimple.h: Rename the function.
9453 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9454 Alan Hayward <alan.hayward@arm.com>
9455 David Sherwood <david.sherwood@arm.com>
9457 * coretypes.h (complex_mode): New type.
9458 * gdbhooks.py (build_pretty_printer): Handle it.
9459 * machmode.h (complex_mode): New class.
9460 (complex_mode::includes_p): New function.
9461 (is_complex_int_mode): Likewise.
9462 (is_complex_float_mode): Likewise.
9463 * genmodes.c (get_mode_class): Handle complex mode classes.
9464 * function.c (expand_function_end): Use is_complex_int_mode.
9466 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9467 Alan Hayward <alan.hayward@arm.com>
9468 David Sherwood <david.sherwood@arm.com>
9470 * coretypes.h (scalar_mode_pod): New typedef.
9471 * gdbhooks.py (build_pretty_printer): Handle it.
9472 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
9473 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
9474 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
9475 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
9478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9479 Alan Hayward <alan.hayward@arm.com>
9480 David Sherwood <david.sherwood@arm.com>
9482 * machmode.h (mode_for_vector): Take a scalar_mode instead
9484 * stor-layout.c (mode_for_vector): Likewise.
9485 * explow.c (promote_mode): Use as_a <scalar_mode>.
9486 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
9488 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9489 Alan Hayward <alan.hayward@arm.com>
9490 David Sherwood <david.sherwood@arm.com>
9492 * target.def (preferred_simd_mode): Take a scalar_mode
9493 instead of a machine_mode.
9494 * targhooks.h (default_preferred_simd_mode): Likewise.
9495 * targhooks.c (default_preferred_simd_mode): Likewise.
9496 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
9497 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
9498 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
9499 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
9500 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
9501 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
9502 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
9503 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
9505 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
9506 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
9507 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
9508 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
9509 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
9510 * doc/tm.texi: Regenerate.
9511 * optabs-query.c (can_vec_mask_load_store_p): Return false for
9514 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9515 Alan Hayward <alan.hayward@arm.com>
9516 David Sherwood <david.sherwood@arm.com>
9518 * target.def (scalar_mode_supported_p): Take a scalar_mode
9519 instead of a machine_mode.
9520 * targhooks.h (default_scalar_mode_supported_p): Likewise.
9521 * targhooks.c (default_scalar_mode_supported_p): Likewise.
9522 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
9523 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9524 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
9525 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
9526 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
9527 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
9528 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
9529 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
9530 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
9531 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
9532 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
9533 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
9535 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
9536 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
9537 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
9538 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
9539 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
9541 * doc/tm.texi: Regenerate.
9543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9544 Alan Hayward <alan.hayward@arm.com>
9545 David Sherwood <david.sherwood@arm.com>
9547 * coretypes.h (opt_scalar_mode): New typedef.
9548 * gdbhooks.py (build_pretty_printers): Handle it.
9549 * machmode.h (mode_iterator::get_2xwider): Add overload for
9551 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
9553 * expr.c (convert_mode_scalar): Likewise.
9554 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9555 * optabs.c (expand_float): Likewise.
9556 (expand_fix): Likewise.
9557 * tree-vect-stmts.c (vectorizable_conversion): 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 * optabs.c (expand_float): Explicitly check for scalars before
9564 using a branching expansion.
9565 (expand_fix): Likewise.
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 * expr.c (convert_mode): Split scalar handling out into...
9572 (convert_mode_scalar): ...this new function. Treat the modes
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 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
9581 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
9583 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9584 Alan Hayward <alan.hayward@arm.com>
9585 David Sherwood <david.sherwood@arm.com>
9587 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
9588 rather than a machine_mode.
9589 (fixed_from_string): Likewise.
9590 (fixed_convert): Likewise.
9591 (fixed_convert_from_int): Likewise.
9592 (fixed_convert_from_real): Likewise.
9593 (real_convert_from_fixed): Likewise.
9594 * fixed-value.c (fixed_from_double_int): Likewise.
9595 (fixed_from_string): Likewise.
9596 (fixed_convert): Likewise.
9597 (fixed_convert_from_int): Likewise.
9598 (fixed_convert_from_real): Likewise.
9599 (real_convert_from_fixed): Likewise.
9600 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
9602 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9603 Alan Hayward <alan.hayward@arm.com>
9604 David Sherwood <david.sherwood@arm.com>
9606 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
9607 of separate mode class checks. Do not allow vector modes here.
9608 (immed_wide_int_const): Use as_a <scalar_mode>.
9609 * explow.c (trunc_int_for_mode): Likewise.
9610 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
9611 (wi::shwi): Likewise.
9612 (wi::min_value): Likewise.
9613 (wi::max_value): Likewise.
9614 * dwarf2out.c (loc_descriptor): Likewise.
9615 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
9618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9619 Alan Hayward <alan.hayward@arm.com>
9620 David Sherwood <david.sherwood@arm.com>
9622 * tree.h (SCALAR_TYPE_MODE): New macro.
9623 * expr.c (expand_expr_addr_expr_1): Use it.
9624 (expand_expr_real_2): Likewise.
9625 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
9626 (fold_convert_const_fixed_from_int): Likewise.
9627 (fold_convert_const_fixed_from_real): Likewise.
9628 (native_encode_fixed): Likewise
9629 (native_encode_complex): Likewise
9630 (native_encode_vector): Likewise.
9631 (native_interpret_fixed): Likewise.
9632 (native_interpret_real): Likewise.
9633 (native_interpret_complex): Likewise.
9634 (native_interpret_vector): Likewise.
9635 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
9636 (simd_clone_adjust_argument_types): Likewise.
9637 (simd_clone_init_simd_arrays): Likewise.
9638 (simd_clone_adjust): Likewise.
9639 * stor-layout.c (layout_type): Likewise.
9640 * tree.c (build_minus_one_cst): Likewise.
9641 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
9642 * tree-inline.c (estimate_move_cost): Likewise.
9643 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
9644 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
9645 (vectorizable_reduction): Likewise.
9646 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
9647 (vect_recog_mixed_size_cond_pattern): Likewise.
9648 (check_bool_pattern): Likewise.
9649 (adjust_bool_pattern): Likewise.
9650 (search_type_for_mask_1): Likewise.
9651 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
9652 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9653 (vectorizable_load): Likewise.
9654 (vectorizable_store): Likewise.
9655 * ubsan.c (ubsan_encode_value): Likewise.
9656 * varasm.c (output_constant): Likewise.
9658 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9659 Alan Hayward <alan.hayward@arm.com>
9660 David Sherwood <david.sherwood@arm.com>
9662 * coretypes.h (scalar_mode): New class.
9663 * machmode.h (scalar_mode): Likewise.
9664 (scalar_mode::includes_p): New function.
9665 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
9666 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
9667 * genmodes.c (get_mode_class): Handle remaining scalar modes.
9668 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
9669 * expmed.c (store_bit_field_1): Likewise.
9670 (extract_bit_field_1): Likewise.
9671 * expr.c (write_complex_part): Likewise.
9672 (read_complex_part): Likewise.
9673 (emit_move_complex_push): Likewise.
9674 (expand_expr_real_2): Likewise.
9675 * function.c (assign_parm_setup_reg): Likewise.
9676 (assign_parms_unsplit_complex): Likewise.
9677 * optabs.c (expand_binop): Likewise.
9678 * rtlanal.c (subreg_get_info): Likewise.
9679 * simplify-rtx.c (simplify_immed_subreg): Likewise.
9680 * varasm.c (output_constant_pool_2): Likewise.
9682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9683 Alan Hayward <alan.hayward@arm.com>
9684 David Sherwood <david.sherwood@arm.com>
9686 * expmed.c (extract_high_half): Use scalar_int_mode and remove
9688 (expmed_mult_highpart_optab): Likewise.
9689 (expmed_mult_highpart): Likewise.
9691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9692 Alan Hayward <alan.hayward@arm.com>
9693 David Sherwood <david.sherwood@arm.com>
9695 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
9696 instead of a machine_mode.
9697 (builtin_memset_read_str): Likewise.
9698 * builtins.c (c_readstr): Likewise.
9699 (builtin_memcpy_read_str): Likewise.
9700 (builtin_strncpy_read_str): Likewise.
9701 (builtin_memset_read_str): Likewise.
9702 (builtin_memset_gen_str): Likewise.
9703 (expand_builtin_signbit): Use scalar_int_mode for local variables.
9704 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
9705 instead of a machine_mode.
9706 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
9708 (make_extraction): Likewise.
9709 (try_widen_shift_mode): Take and return scalar_int_modes instead
9711 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
9712 a scalar_int_mode instead of a machine_mode.
9713 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
9714 (avr_addr_space_pointer_mode): Likewise.
9715 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
9716 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
9717 (msp430_unwind_word_mode): Likewise.
9718 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
9719 (spu_addr_space_pointer_mode): Likewise.
9720 (spu_addr_space_address_mode): Likewise.
9721 (spu_libgcc_cmp_return_mode): Likewise.
9722 (spu_libgcc_shift_count_mode): Likewise.
9723 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
9724 (rl78_addr_space_pointer_mode): Likewise.
9725 (fl78_unwind_word_mode): Likewise.
9726 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
9728 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
9729 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
9730 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
9731 (mips_valid_pointer_mode): Likewise.
9732 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
9733 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
9734 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
9736 (ft32_addr_space_address_mode): Likewise.
9737 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
9738 scalar_int_mode instead of a machine_mode.
9739 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
9741 (m32c_addr_space_address_mode): Likewise.
9742 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
9743 (rs6000_eh_return_filter_mode): Likewise.
9744 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
9745 (rs6000_eh_return_filter_mode): Likewise.
9746 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
9747 (s390_libgcc_shift_count_mode): Likewise.
9748 (s390_unwind_word_mode): Likewise.
9749 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
9751 * target.def (mode_rep_extended): Likewise.
9752 (valid_pointer_mode): Likewise.
9753 (addr_space.valid_pointer_mode): Likewise.
9754 (eh_return_filter_mode): Return a scalar_int_mode rather than
9756 (libgcc_cmp_return_mode): Likewise.
9757 (libgcc_shift_count_mode): Likewise.
9758 (unwind_word_mode): Likewise.
9759 (addr_space.pointer_mode): Likewise.
9760 (addr_space.address_mode): Likewise.
9761 * doc/tm.texi: Regenerate.
9762 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
9764 (do_jump): Use scalar_int_mode for local variables.
9765 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
9766 rather than a machine_mode.
9767 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
9768 (scompare_loc_descriptor_wide): Likewise.
9769 (scompare_loc_descriptor_narrow): Likewise.
9770 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
9772 * except.c (sjlj_emit_dispatch_table): Likewise.
9773 (expand_builtin_eh_copy_values): Likewise.
9774 * explow.c (convert_memory_address_addr_space_1): Likewise.
9775 Take a scalar_int_mode rather than a machine_mode.
9776 (convert_memory_address_addr_space): Take a scalar_int_mode rather
9777 than a machine_mode.
9778 (memory_address_addr_space): Use scalar_int_mode for local variables.
9779 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
9780 rather than a machine_mode.
9781 * expmed.c (mask_rtx): Likewise.
9782 (init_expmed_one_conv): Likewise.
9783 (expand_mult_highpart_adjust): Likewise.
9784 (extract_high_half): Likewise.
9785 (expmed_mult_highpart_optab): Likewise.
9786 (expmed_mult_highpart): Likewise.
9787 (expand_smod_pow2): Likewise.
9788 (expand_sdiv_pow2): Likewise.
9789 (emit_store_flag_int): Likewise.
9790 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
9792 (extract_low_bits): Likewise.
9793 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
9795 * expr.c (pieces_addr::adjust): Likewise.
9796 (can_store_by_pieces): Likewise.
9797 (store_by_pieces): Likewise.
9798 (clear_by_pieces_1): Likewise.
9799 (expand_expr_addr_expr_1): Likewise.
9800 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
9801 (expand_expr_real_1): Likewise.
9802 (try_casesi): Likewise.
9803 * final.c (shorten_branches): Likewise.
9804 * fold-const.c (fold_convert_const_int_from_fixed): Change the
9805 type of "mode" to machine_mode.
9806 * internal-fn.c (expand_arith_overflow_result_store): Take a
9807 scalar_int_mode rather than a machine_mode.
9808 (expand_mul_overflow): Use scalar_int_mode for local variables.
9809 * loop-doloop.c (doloop_modify): Likewise.
9810 (doloop_optimize): Likewise.
9811 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
9812 than a machine_mode.
9813 (expand_doubleword_shift_condmove): Likewise.
9814 (expand_doubleword_shift): Likewise.
9815 (expand_doubleword_clz): Likewise.
9816 (expand_doubleword_popcount): Likewise.
9817 (expand_doubleword_parity): Likewise.
9818 (expand_absneg_bit): Use scalar_int_mode for local variables.
9819 (prepare_float_lib_cmp): Likewise.
9820 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
9821 rather than a machine_mode.
9822 (convert_memory_address_addr_space): Likewise.
9823 (get_mode_bounds): Likewise.
9824 (get_address_mode): Return a scalar_int_mode rather than a
9826 * rtlanal.c (get_address_mode): Likewise.
9827 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
9828 than a machine_mode.
9829 * targhooks.c (default_mode_rep_extended): Likewise.
9830 (default_valid_pointer_mode): Likewise.
9831 (default_addr_space_valid_pointer_mode): Likewise.
9832 (default_eh_return_filter_mode): Return a scalar_int_mode rather
9833 than a machine_mode.
9834 (default_libgcc_cmp_return_mode): Likewise.
9835 (default_libgcc_shift_count_mode): Likewise.
9836 (default_unwind_word_mode): Likewise.
9837 (default_addr_space_pointer_mode): Likewise.
9838 (default_addr_space_address_mode): Likewise.
9839 * targhooks.h (default_eh_return_filter_mode): Likewise.
9840 (default_libgcc_cmp_return_mode): Likewise.
9841 (default_libgcc_shift_count_mode): Likewise.
9842 (default_unwind_word_mode): Likewise.
9843 (default_addr_space_pointer_mode): Likewise.
9844 (default_addr_space_address_mode): Likewise.
9845 (default_mode_rep_extended): Take a scalar_int_mode rather than
9847 (default_valid_pointer_mode): Likewise.
9848 (default_addr_space_valid_pointer_mode): Likewise.
9849 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
9851 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
9852 rather than a machine_mode.
9853 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
9854 for local variables.
9855 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9856 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
9857 than a machine_mode.
9859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9860 Alan Hayward <alan.hayward@arm.com>
9861 David Sherwood <david.sherwood@arm.com>
9863 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
9864 the mode argument to scalar_int_mode.
9865 (do_jump_by_parts_zero_rtx): Likewise.
9866 (do_jump_by_parts_equality_rtx): Likewise.
9867 (do_jump_by_parts_greater): Take a mode argument.
9868 (do_jump_by_parts_equality): Likewise.
9869 (do_jump_1): Update calls accordingly.
9871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9872 Alan Hayward <alan.hayward@arm.com>
9873 David Sherwood <david.sherwood@arm.com>
9875 * is-a.h (safe_dyn_cast): New function.
9876 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
9877 (jump_table_for_label): Likewise.
9878 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
9879 instead of an rtx_insn *.
9880 (shorten_branches): Use dyn_cast instead of LABEL_P and
9881 JUMP_TABLE_DATA_P. Use jump_table_for_label and
9882 rtx_jump_table_data::get_data_mode.
9883 (final_scan_insn): Likewise.
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 * combine.c (try_combine): Use is_a <scalar_int_mode> when
9890 trying to combine a full-register integer set with a subreg
9893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9894 Alan Hayward <alan.hayward@arm.com>
9895 David Sherwood <david.sherwood@arm.com>
9897 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
9898 that is always either address_mode or pointer_mode.
9900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9901 Alan Hayward <alan.hayward@arm.com>
9902 David Sherwood <david.sherwood@arm.com>
9904 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
9905 when the two are known to be equal.
9907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9908 Alan Hayward <alan.hayward@arm.com>
9909 David Sherwood <david.sherwood@arm.com>
9911 * simplify-rtx.c (simplify_const_unary_operation): Use
9912 is_a <scalar_int_mode> instead of checking for a nonzero
9913 precision. Forcibly convert op_mode to a scalar_int_mode
9914 in that case. More clearly differentiate the operand and
9915 result modes and use the former when deciding what the value
9916 of a count-bits operation should be. Use is_int_mode instead
9917 of checking for a MODE_INT. Remove redundant check for whether
9918 this mode has a zero precision.
9920 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9921 Alan Hayward <alan.hayward@arm.com>
9922 David Sherwood <david.sherwood@arm.com>
9924 * optabs.c (widen_leading): Change the type of the mode argument
9925 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
9926 (widen_bswap): Likewise.
9927 (expand_parity): Likewise.
9928 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
9929 (expand_ffs): Likewise.
9930 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
9939 (expand_expr_real_1): Update call accordingly.
9941 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9942 Alan Hayward <alan.hayward@arm.com>
9943 David Sherwood <david.sherwood@arm.com>
9945 * expmed.c (store_bit_field_using_insv): Add op0_mode and
9946 value_mode arguments. Use scalar_int_mode internally.
9947 (store_bit_field_1): Rename the new integer mode from imode
9948 to op0_mode and use it instead of GET_MODE (op0). Update calls
9949 to store_split_bit_field, store_bit_field_using_insv and
9950 store_fixed_bit_field.
9951 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
9952 Use scalar_int_mode internally. Use a bit count rather than a mode
9953 when calculating the largest bit size for get_best_mode.
9954 Update calls to store_split_bit_field and store_fixed_bit_field_1.
9955 (store_fixed_bit_field_1): Add mode and value_mode arguments.
9956 Remove assertion that OP0 has a scalar integer mode.
9957 (store_split_bit_field): Add op0_mode and value_mode arguments.
9958 Update calls to extract_fixed_bit_field.
9959 (extract_bit_field_using_extv): Add an op0_mode argument.
9960 Use scalar_int_mode internally.
9961 (extract_bit_field_1): Rename the new integer mode from imode to
9962 op0_mode and use it instead of GET_MODE (op0). Update calls to
9963 extract_split_bit_field, extract_bit_field_using_extv and
9964 extract_fixed_bit_field.
9965 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
9966 to extract_split_bit_field and extract_fixed_bit_field_1.
9967 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
9968 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
9970 (extract_split_bit_field): Add an op0_mode argument. Update call
9971 to extract_fixed_bit_field.
9973 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9974 Alan Hayward <alan.hayward@arm.com>
9975 David Sherwood <david.sherwood@arm.com>
9977 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
9978 * explow.c (hard_function_value): Likewise.
9979 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
9980 convert_to_mode call outside the loop.
9981 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
9982 for the mode iterator. Require the mode specified by max_pieces
9984 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
9986 (copy_blkmode_to_reg): Likewise.
9987 (set_storage_via_setmem): Likewise.
9988 * optabs.c (prepare_cmp_insn): Likewise.
9989 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9990 * stor-layout.c (finish_bitfield_representative): Likewise.
9992 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9993 Alan Hayward <alan.hayward@arm.com>
9994 David Sherwood <david.sherwood@arm.com>
9996 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
9997 * expr.c (convert_move): Use them.
9998 (convert_modes): Likewise.
9999 (store_expr_with_bounds): Likewise.
10001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10002 Alan Hayward <alan.hayward@arm.com>
10003 David Sherwood <david.sherwood@arm.com>
10005 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10006 parameter for the mode of "x". Remove the "known_x", "known_mode"
10007 and "known_ret" arguments. Change the type of the mode argument
10008 to scalar_int_mode.
10009 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10010 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10011 (reg_num_sign_bit_copies_for_combine): Likewise.
10012 * rtlanal.c (nonzero_bits1): Likewise.
10013 (num_sign_bit_copies1): Likewise.
10014 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10015 (reg_num_sign_bit_copies_general): Likewise.
10016 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10017 (reg_nonzero_bits_general): Likewise.
10019 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10020 Alan Hayward <alan.hayward@arm.com>
10021 David Sherwood <david.sherwood@arm.com>
10023 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10024 than in subroutines. Return 1 for non-integer modes.
10025 (cached_num_sign_bit_copies): Change the type of the mode parameter
10026 to scalar_int_mode.
10027 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10028 classes. Handle CONST_INT_P first and then check whether X also
10029 has a scalar integer mode. Check the same thing for inner registers
10030 of a SUBREG and for values that are being extended or truncated.
10032 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10033 Alan Hayward <alan.hayward@arm.com>
10034 David Sherwood <david.sherwood@arm.com>
10036 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10037 in subroutines. Return the mode mask for non-integer modes.
10038 (cached_nonzero_bits): Change the type of the mode parameter
10039 to scalar_int_mode.
10040 (nonzero_bits1): Likewise. Remove early exit for other mode
10041 classes. Handle CONST_INT_P first and then check whether X
10042 also has a scalar integer mode.
10044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10045 Alan Hayward <alan.hayward@arm.com>
10046 David Sherwood <david.sherwood@arm.com>
10048 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10049 Return a scalar_int_mode and assert that the size is greater than
10051 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10052 (op_by_pieces_d::op_by_pieces_d): Likewise.
10053 (op_by_pieces_d::run): Likewise.
10054 (can_store_by_pieces): Likewise.
10056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10057 Alan Hayward <alan.hayward@arm.com>
10058 David Sherwood <david.sherwood@arm.com>
10060 * combine.c (extract_left_shift): Add a mode argument and update
10062 (make_compound_operation_int): Change the type of the mode parameter
10063 to scalar_int_mode and update the call to extract_left_shift.
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_and_const_int): Change the type of the mode
10070 parameter to scalar_int_mode.
10071 (simplify_and_const_int_1): Likewise. Update recursive call.
10073 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10074 Alan Hayward <alan.hayward@arm.com>
10075 David Sherwood <david.sherwood@arm.com>
10077 * combine.c (simplify_compare_const): Check that the mode is a
10078 scalar_int_mode (rather than VOIDmode) before testing its
10080 (simplify_comparison): Move COMPARISON_P handling out of the
10081 loop and restrict the latter part of the loop to scalar_int_modes.
10082 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10083 and when considering SUBREG_REGs. Use is_int_mode instead of
10084 checking GET_MODE_CLASS against MODE_INT.
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 (try_widen_shift_mode): Move check for equal modes to...
10091 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10092 shift_unit_mode and for modes involved in scalar shifts.
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 * combine.c (force_int_to_mode): New function, split out from...
10099 (force_to_mode): ...here. Keep xmode up-to-date and use it
10100 instead of GET_MODE (x).
10102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10103 Alan Hayward <alan.hayward@arm.com>
10104 David Sherwood <david.sherwood@arm.com>
10106 * optabs-query.h (extraction_insn::struct_mode): Change type to
10107 opt_scalar_int_mode and update comment.
10108 (extraction_insn::field_mode): Change type to scalar_int_mode.
10109 (extraction_insn::pos_mode): Likewise.
10110 * combine.c (make_extraction): Update accordingly.
10111 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10112 (get_optab_extraction_insn): Likewise.
10113 * recog.c (simplify_while_replacing): Likewise.
10114 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10115 parameter to opt_scalar_int_mode.
10117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10118 Alan Hayward <alan.hayward@arm.com>
10119 David Sherwood <david.sherwood@arm.com>
10121 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10122 to a scalar_int_mode instead of a machine_mode.
10123 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10124 (get_best_mode): Return a boolean and use a pointer argument to store
10125 the selected mode. Replace the limit mode parameter with a bit limit.
10126 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10127 for the values returned by bit_field_mode_iterator::next_mode.
10128 (store_bit_field): Update call to get_best_mode.
10129 (store_fixed_bit_field): Likewise.
10130 (extract_fixed_bit_field): Likewise.
10131 * expr.c (optimize_bitfield_assignment_op): Likewise.
10132 * fold-const.c (optimize_bit_field_compare): Likewise.
10133 (fold_truth_andor_1): Likewise.
10134 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10135 Update for new type of m_mode.
10136 (get_best_mode): As above.
10138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10139 Alan Hayward <alan.hayward@arm.com>
10140 David Sherwood <david.sherwood@arm.com>
10142 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10143 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10144 (store_bit_field): Check is_a <scalar_int_mode> before calling
10145 strict_volatile_bitfield_p.
10146 (extract_bit_field): Likewise.
10148 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10149 Alan Hayward <alan.hayward@arm.com>
10150 David Sherwood <david.sherwood@arm.com>
10152 * target.def (cstore_mode): Return a scalar_int_mode.
10153 * doc/tm.texi: Regenerate.
10154 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10155 * targhooks.h (default_cstore_mode): Likewise.
10156 * targhooks.c (default_cstore_mode): Likewise, using a forced
10158 * expmed.c (emit_cstore): Expect the target of the cstore to be
10161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10162 Alan Hayward <alan.hayward@arm.com>
10163 David Sherwood <david.sherwood@arm.com>
10165 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10167 (niter_desc): Likewise mode.
10168 (iv_analyze): Add a mode parameter.
10170 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10171 and change its type to scalar_int_mode.
10172 * loop-iv.c: Update commentary at head of file.
10173 (iv_constant): Pass the mode paraeter before the rtx it describes
10174 and change its type to scalar_int_mode. Remove VOIDmode handling.
10175 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10176 (iv_extend): Likewise.
10177 (shorten_into_mode): Likewise.
10178 (iv_add): Use scalar_int_mode.
10179 (iv_mult): Likewise.
10180 (iv_shift): Likewise.
10181 (canonicalize_iv_subregs): Likewise.
10182 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10183 it describes and change its mode to scalar_int_mode. Also change
10184 the type of the returned inner_mode to scalar_int_mode.
10185 (get_biv_step): Likewise, turning outer_mode from a pointer
10186 into a direct parameter. Update call to get_biv_step_1.
10187 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10188 iv_constant and get_biv_step.
10189 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10190 and change its type to scalar_int_mode. Don't initialise iv->mode
10191 to VOIDmode and remove later checks for its still being VOIDmode.
10192 Update calls to iv_analyze_op and iv_analyze_expr. Check
10193 is_a <scalar_int_mode> when changing the mode under consideration.
10194 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10195 Update call to iv_analyze_expr.
10196 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10197 inner register is not also a scalar_int_mode. Update call to
10199 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10200 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10201 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10202 separate mode class checks. Update calls to iv_analyze. Remove
10203 fix-up of VOIDmodes after iv_analyze_biv.
10204 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10205 don't have a scalar_int_mode. Update call to biv_p.
10207 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10208 Alan Hayward <alan.hayward@arm.com>
10209 David Sherwood <david.sherwood@arm.com>
10211 * cfgexpand.c (convert_debug_memory_address): Use
10212 as_a <scalar_int_mode>.
10213 * combine.c (expand_compound_operation): Likewise.
10214 (make_extraction): Likewise.
10215 (change_zero_ext): Likewise.
10216 (simplify_comparison): Likewise.
10217 * cse.c (cse_insn): Likewise.
10218 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10219 (mem_loc_descriptor): Likewise.
10220 (loc_descriptor): Likewise.
10221 * expmed.c (init_expmed_one_mode): Likewise.
10222 (synth_mult): Likewise.
10223 (emit_store_flag_1): Likewise.
10224 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10225 of a comparison with size.
10226 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10227 (reduce_to_bit_field_precision): Likewise.
10228 * function.c (expand_function_end): Likewise.
10229 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10230 * loop-doloop.c (doloop_modify): Likewise.
10231 * optabs.c (expand_binop): Likewise.
10232 (expand_unop): Likewise.
10233 (expand_copysign_absneg): Likewise.
10234 (prepare_cmp_insn): Likewise.
10235 (maybe_legitimize_operand): Likewise.
10236 * recog.c (const_scalar_int_operand): Likewise.
10237 * rtlanal.c (get_address_mode): Likewise.
10238 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10239 (simplify_cond_clz_ctz): Likewise.
10240 * tree-nested.c (get_nl_goto_field): Likewise.
10241 * tree.c (build_vector_type_for_mode): Likewise.
10242 * var-tracking.c (use_narrower_mode): Likewise.
10244 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10245 Alan Hayward <alan.hayward@arm.com>
10246 David Sherwood <david.sherwood@arm.com>
10248 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10249 * builtins.c (expand_builtin_signbit): Use it.
10250 * cfgexpand.c (expand_debug_expr): Likewise.
10251 * dojump.c (do_jump): Likewise.
10252 (do_compare_and_jump): Likewise.
10253 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
10254 * expmed.c (make_tree): Likewise.
10255 * expr.c (expand_expr_real_2): Likewise.
10256 (expand_expr_real_1): Likewise.
10257 (try_casesi): Likewise.
10258 * fold-const-call.c (fold_const_call_ss): Likewise.
10259 * fold-const.c (unextend): Likewise.
10260 (extract_muldiv_1): Likewise.
10261 (fold_single_bit_test): Likewise.
10262 (native_encode_int): Likewise.
10263 (native_encode_string): Likewise.
10264 (native_interpret_int): Likewise.
10265 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
10266 * internal-fn.c (expand_addsub_overflow): Likewise.
10267 (expand_neg_overflow): Likewise.
10268 (expand_mul_overflow): Likewise.
10269 (expand_arith_overflow): Likewise.
10270 * match.pd: Likewise.
10271 * stor-layout.c (layout_type): Likewise.
10272 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10273 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
10274 * tree-ssanames.c (get_range_info): Likewise.
10275 * tree-switch-conversion.c (array_value_type) Likewise.
10276 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
10277 (vect_recog_divmod_pattern): Likewise.
10278 (vect_recog_mixed_size_cond_pattern): Likewise.
10279 * tree-vrp.c (extract_range_basic): Likewise.
10280 (simplify_float_conversion_using_ranges): Likewise.
10281 * tree.c (int_fits_type_p): Likewise.
10282 * ubsan.c (instrument_bool_enum_load): Likewise.
10283 * varasm.c (mergeable_string_section): Likewise.
10284 (narrowing_initializer_constant_valid_p): Likewise.
10285 (output_constant): Likewise.
10287 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10288 Alan Hayward <alan.hayward@arm.com>
10289 David Sherwood <david.sherwood@arm.com>
10291 * machmode.h (NARROWEST_INT_MODE): New macro.
10292 * expr.c (alignment_for_piecewise_move): Use it instead of
10293 GET_CLASS_NARROWEST_MODE (MODE_INT).
10294 (push_block): Likewise.
10295 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
10297 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10299 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10300 Alan Hayward <alan.hayward@arm.com>
10301 David Sherwood <david.sherwood@arm.com>
10303 * postreload.c (move2add_valid_value_p): Change the type of the
10304 mode parameter to scalar_int_mode.
10305 (move2add_use_add2_insn): Add a mode parameter and use it instead
10307 (move2add_use_add3_insn): Likewise.
10308 (reload_cse_move2add): Update accordingly.
10310 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10311 Alan Hayward <alan.hayward@arm.com>
10312 David Sherwood <david.sherwood@arm.com>
10314 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
10316 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
10317 * optabs.c (expand_unop): Likewise.
10319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10320 Alan Hayward <alan.hayward@arm.com>
10321 David Sherwood <david.sherwood@arm.com>
10323 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
10324 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
10325 (popcount_loc_descriptor): Likewise.
10326 (bswap_loc_descriptor): Likewise.
10327 (rotate_loc_descriptor): Likewise.
10328 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
10329 calling the functions above.
10331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10332 Alan Hayward <alan.hayward@arm.com>
10333 David Sherwood <david.sherwood@arm.com>
10335 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
10337 (try_combine): Likewise.
10338 (simplify_if_then_else): Likewise.
10339 * cse.c (cse_insn): Likewise.
10340 * dwarf2out.c (mem_loc_descriptor): Likewise.
10341 * emit-rtl.c (gen_lowpart_common): Likewise.
10342 * simplify-rtx.c (simplify_truncation): Likewise.
10343 (simplify_binary_operation_1): Likewise.
10344 (simplify_const_relational_operation): Likewise.
10345 (simplify_ternary_operation): Likewise.
10346 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): 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 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
10353 * reload.c (push_reload): Likewise.
10354 (find_reloads): Likewise.
10356 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10357 Alan Hayward <alan.hayward@arm.com>
10358 David Sherwood <david.sherwood@arm.com>
10360 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
10361 (make_compound_operation_int): Likewise.
10362 (change_zero_ext): Likewise.
10363 * expr.c (convert_move): Likewise.
10364 (convert_modes): Likewise.
10365 * fwprop.c (forward_propagate_subreg): Likewise.
10366 * loop-iv.c (get_biv_step_1): Likewise.
10367 * optabs.c (widen_operand): Likewise.
10368 * postreload.c (move2add_valid_value_p): Likewise.
10369 * recog.c (simplify_while_replacing): Likewise.
10370 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10371 (simplify_binary_operation_1): Likewise. Remove redundant
10372 mode equality check.
10374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10375 Alan Hayward <alan.hayward@arm.com>
10376 David Sherwood <david.sherwood@arm.com>
10378 * combine.c (combine_simplify_rtx): Add checks for
10379 is_a <scalar_int_mode>.
10380 (simplify_if_then_else): Likewise.
10381 (make_field_assignment): Likewise.
10382 (simplify_comparison): Likewise.
10383 * ifcvt.c (noce_try_bitop): Likewise.
10384 * loop-invariant.c (canonicalize_address_mult): Likewise.
10385 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10387 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10388 Alan Hayward <alan.hayward@arm.com>
10389 David Sherwood <david.sherwood@arm.com>
10391 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
10392 is_a <scalar_int_mode> instead of != BLKmode.
10394 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10395 Alan Hayward <alan.hayward@arm.com>
10396 David Sherwood <david.sherwood@arm.com>
10398 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
10399 instead of != VOIDmode.
10400 * combine.c (if_then_else_cond): Likewise.
10401 (change_zero_ext): Likewise.
10402 * dwarf2out.c (mem_loc_descriptor): Likewise.
10403 (loc_descriptor): Likewise.
10404 * rtlanal.c (canonicalize_condition): Likewise.
10405 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
10407 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10408 Alan Hayward <alan.hayward@arm.com>
10409 David Sherwood <david.sherwood@arm.com>
10411 * simplify-rtx.c (simplify_binary_operation_1): Use
10412 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
10414 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10415 Alan Hayward <alan.hayward@arm.com>
10416 David Sherwood <david.sherwood@arm.com>
10418 * wide-int.h (int_traits<unsigned char>) New class.
10419 (int_traits<unsigned short>) Likewise.
10420 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
10421 Use GET_MODE_UNIT_PRECISION and remove redundant test for
10423 * combine.c (set_nonzero_bits_and_sign_copies): Use
10424 is_a <scalar_int_mode>.
10425 (find_split_point): Likewise.
10426 (combine_simplify_rtx): Likewise.
10427 (simplify_logical): Likewise.
10428 (expand_compound_operation): Likewise.
10429 (expand_field_assignment): Likewise.
10430 (make_compound_operation): Likewise.
10431 (extended_count): Likewise.
10432 (change_zero_ext): Likewise.
10433 (simplify_comparison): Likewise.
10434 * dwarf2out.c (scompare_loc_descriptor): Likewise.
10435 (ucompare_loc_descriptor): Likewise.
10436 (minmax_loc_descriptor): Likewise.
10437 (mem_loc_descriptor): Likewise.
10438 (loc_descriptor): Likewise.
10439 * expmed.c (init_expmed_one_mode): Likewise.
10440 * lra-constraints.c (lra_constraint_offset): Likewise.
10441 * optabs.c (prepare_libcall_arg): Likewise.
10442 * postreload.c (move2add_note_store): Likewise.
10443 * reload.c (operands_match_p): Likewise.
10444 * rtl.h (load_extend_op): Likewise.
10445 * rtlhooks.c (gen_lowpart_general): Likewise.
10446 * simplify-rtx.c (simplify_truncation): Likewise.
10447 (simplify_unary_operation_1): Likewise.
10448 (simplify_binary_operation_1): Likewise.
10449 (simplify_const_binary_operation): Likewise.
10450 (simplify_const_relational_operation): Likewise.
10451 (simplify_subreg): Likewise.
10452 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10453 * var-tracking.c (adjust_mems): Likewise.
10454 (prepare_call_arguments): Likewise.
10456 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10457 Alan Hayward <alan.hayward@arm.com>
10458 David Sherwood <david.sherwood@arm.com>
10460 * machmode.h (is_int_mode): New fuction.
10461 * combine.c (find_split_point): Use it.
10462 (combine_simplify_rtx): Likewise.
10463 (simplify_if_then_else): Likewise.
10464 (simplify_set): Likewise.
10465 (simplify_shift_const_1): Likewise.
10466 (simplify_comparison): Likewise.
10467 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
10468 * cse.c (notreg_cost): Likewise.
10469 (cse_insn): Likewise.
10470 * cselib.c (cselib_lookup_1): Likewise.
10471 * dojump.c (do_jump_1): Likewise.
10472 (do_compare_rtx_and_jump): Likewise.
10473 * dse.c (get_call_args): Likewise.
10474 * dwarf2out.c (rtl_for_decl_init): Likewise.
10475 (native_encode_initializer): Likewise.
10476 * expmed.c (emit_store_flag_1): Likewise.
10477 (emit_store_flag): Likewise.
10478 * expr.c (convert_modes): Likewise.
10479 (store_field): Likewise.
10480 (expand_expr_real_1): Likewise.
10481 * fold-const.c (fold_read_from_constant_string): Likewise.
10482 * gimple-ssa-sprintf.c (get_format_string): Likewise.
10483 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
10484 * optabs.c (expand_binop): Likewise.
10485 (expand_unop): Likewise.
10486 (expand_abs_nojump): Likewise.
10487 (expand_one_cmpl_abs_nojump): Likewise.
10488 * simplify-rtx.c (mode_signbit_p): Likewise.
10489 (val_signbit_p): Likewise.
10490 (val_signbit_known_set_p): Likewise.
10491 (val_signbit_known_clear_p): Likewise.
10492 (simplify_relational_operation_1): Likewise.
10493 * tree.c (vector_type_mode): Likewise.
10495 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10496 Alan Hayward <alan.hayward@arm.com>
10497 David Sherwood <david.sherwood@arm.com>
10499 * machmode.h (smallest_mode_for_size): Fix formatting.
10500 (smallest_int_mode_for_size): New function.
10501 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
10502 instead of smallest_mode_for_size.
10503 * combine.c (make_extraction): Likewise.
10504 * config/arc/arc.c (arc_expand_movmem): Likewise.
10505 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
10506 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10507 * config/s390/s390.c (s390_expand_insv): Likewise.
10508 * config/sparc/sparc.c (assign_int_registers): Likewise.
10509 * config/spu/spu.c (spu_function_value): Likewise.
10510 (spu_function_arg): Likewise.
10511 * coverage.c (get_gcov_type): Likewise.
10512 (get_gcov_unsigned_t): Likewise.
10513 * dse.c (find_shift_sequence): Likewise.
10514 * expmed.c (store_bit_field_1): Likewise.
10515 * expr.c (convert_move): Likewise.
10516 (store_field): Likewise.
10517 * internal-fn.c (expand_arith_overflow): Likewise.
10518 * optabs-query.c (get_best_extraction_insn): Likewise.
10519 * optabs.c (expand_twoval_binop_libfunc): Likewise.
10520 * stor-layout.c (layout_type): Likewise.
10521 (initialize_sizetypes): Likewise.
10522 * targhooks.c (default_get_mask_mode): Likewise.
10523 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
10525 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10526 Alan Hayward <alan.hayward@arm.com>
10527 David Sherwood <david.sherwood@arm.com>
10529 * machmode.h (opt_mode::else_blk): New function.
10530 (int_mode_for_mode): Declare.
10531 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
10532 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
10534 * cfgexpand.c (expand_debug_expr): Likewise.
10535 * combine.c (gen_lowpart_or_truncate): Likewise.
10536 (gen_lowpart_for_combine): Likewise.
10537 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
10538 * config/avr/avr.c (avr_to_int_mode): Likewise.
10539 (avr_out_plus_1): Likewise.
10540 (avr_out_plus): Likewise.
10541 (avr_out_round): Likewise.
10542 * config/i386/i386.c (ix86_split_to_parts): Likewise.
10543 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10544 (s390_expand_vcond): Likewise.
10545 * config/spu/spu.c (spu_split_immediate): Likewise.
10546 (spu_expand_mov): Likewise.
10547 * dse.c (get_stored_val): Likewise.
10548 * expmed.c (store_bit_field_1): Likewise.
10549 (convert_extracted_bit_field): Use int_mode_for_mode instead of
10551 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
10552 (extract_low_bits): Likewise.
10553 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
10554 handling rather than repeating the check.
10555 (emit_group_store): Likewise.
10556 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
10557 * optabs.c (expand_absneg_bit): Likewise.
10558 (expand_copysign_absneg): Likewise.
10559 (expand_copysign_bit): Likewise.
10560 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
10561 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
10562 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
10563 * var-tracking.c (prepare_call_arguments): Likewise.
10564 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
10565 int_mode_for_mode instead of mode_for_size.
10566 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10568 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10569 Alan Hayward <alan.hayward@arm.com>
10570 David Sherwood <david.sherwood@arm.com>
10572 * machmode.h (int_mode_for_size): New function.
10573 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
10574 instead of mode_for_size.
10575 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
10577 * combine.c (expand_field_assignment): Use int_mode_for_size
10578 instead of mode_for_size.
10579 (make_extraction): Likewise.
10580 (simplify_shift_const_1): Likewise.
10581 (simplify_comparison): Likewise.
10582 * dojump.c (do_jump): Likewise.
10583 * dwarf2out.c (mem_loc_descriptor): Likewise.
10584 * emit-rtl.c (init_derived_machine_modes): Likewise.
10585 * expmed.c (flip_storage_order): Likewise.
10586 (convert_extracted_bit_field): Likewise.
10587 * expr.c (copy_blkmode_from_reg): Likewise.
10588 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
10589 * internal-fn.c (expand_mul_overflow): Likewise.
10590 * lower-subreg.c (simple_move): Likewise.
10591 * optabs-libfuncs.c (init_optabs): Likewise.
10592 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10593 * tree.c (vector_type_mode): Likewise.
10594 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
10595 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
10596 * tree-vect-generic.c (expand_vector_parallel): Likewise.
10597 * tree-vect-stmts.c (vectorizable_load): Likewise.
10598 (vectorizable_store): Likewise.
10600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10601 Alan Hayward <alan.hayward@arm.com>
10602 David Sherwood <david.sherwood@arm.com>
10604 * coretypes.h (pod_mode): New type.
10605 (scalar_int_mode_pod): New typedef.
10606 * machmode.h (pod_mode): New class.
10607 (int_n_data_t::m): Change type to scalar_int_mode_pod.
10608 * genmodes.c (emit_mode_int_n): Update accordingly.
10609 * lower-subreg.h (target_lower_subreg): Change type to
10610 scalar_int_mode_pod.
10611 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
10612 scalar_int_mode_pod.
10614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10615 Alan Hayward <alan.hayward@arm.com>
10616 David Sherwood <david.sherwood@arm.com>
10618 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
10619 machine_mode to scalar_int_mode.
10620 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
10621 (rs6000_option_override_internal): Remove cast to int.
10622 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
10623 machine_mode to scalar_int_mode.
10624 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
10625 (rs6000_option_override_internal): Remove cast to int.
10626 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
10627 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
10629 * config/s390/s390.c (s390_expand_builtin): Likewise.
10630 * coretypes.h (scalar_int_mode): New type.
10631 (opt_scalar_int_mode): New typedef.
10632 * machmode.h (scalar_int_mode): New class.
10633 (scalar_int_mode::includes_p): New function.
10634 (byte_mode): Change type to scalar_int_mode.
10635 (word_mode): Likewise.
10636 (ptr_mode): Likewise.
10637 * emit-rtl.c (byte_mode): Likewise.
10638 (word_mode): Likewise.
10639 (ptr_mode): Likewise.
10640 (init_derived_machine_modes): Update accordingly.
10641 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
10642 and MODE_PARTIAL_INT.
10643 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
10644 opt_scalar_int_mode.
10646 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10647 Alan Hayward <alan.hayward@arm.com>
10648 David Sherwood <david.sherwood@arm.com>
10650 * target.def (libgcc_floating_mode_supported_p): Take a
10652 * doc/tm.texi: Regenerate.
10653 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
10655 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
10656 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
10659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10660 Alan Hayward <alan.hayward@arm.com>
10661 David Sherwood <david.sherwood@arm.com>
10663 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
10664 * doc/tm.texi: Regenerate.
10665 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
10666 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
10667 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
10668 * targhooks.h (default_floatn_mode): Likewise.
10669 * targhooks.c (default_floatn_mode): Likewise.
10670 * tree.c (build_common_tree_nodes): Update accordingly.
10672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10673 Alan Hayward <alan.hayward@arm.com>
10674 David Sherwood <david.sherwood@arm.com>
10676 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
10677 (mode_iterator::iterate_p): Likewise.
10678 (mode_iterator::get_wider): Likewise.
10679 * expr.c (init_expr_target): Use opt_scalar_float_mode.
10681 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10682 Alan Hayward <alan.hayward@arm.com>
10683 David Sherwood <david.sherwood@arm.com>
10685 * coretypes.h (opt_scalar_float_mode): New typedef.
10686 * machmode.h (float_mode_for_size): New function.
10687 * emit-rtl.c (double_mode): Delete.
10688 (init_emit_once): Use float_mode_for_size.
10689 * stor-layout.c (layout_type): Likewise.
10690 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
10692 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10693 Alan Hayward <alan.hayward@arm.com>
10694 David Sherwood <david.sherwood@arm.com>
10696 * output.h (assemble_real): Take a scalar_float_mode.
10697 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
10698 * config/arm/arm.md (consttable_4): Likewise.
10699 (consttable_8): Likewise.
10700 (consttable_16): Likewise.
10701 * config/mips/mips.md (consttable_float): Likewise.
10702 * config/s390/s390.c (s390_output_pool_entry): Likewise.
10703 * varasm.c (assemble_real): Take a scalar_float_mode.
10704 (output_constant_pool_2): Update accordingly.
10705 (output_constant): Likewise.
10707 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10708 Alan Hayward <alan.hayward@arm.com>
10709 David Sherwood <david.sherwood@arm.com>
10711 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
10712 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
10713 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
10714 (native_encode_real): Likewise.
10715 (native_interpret_real): Likewise.
10716 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
10717 * tree-vrp.c (simplify_float_conversion_using_ranges): 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 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
10724 <scalar_float_mode>. Simplify.
10725 (gen_extend_conv_libfunc): Likewise.
10727 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10728 Alan Hayward <alan.hayward@arm.com>
10729 David Sherwood <david.sherwood@arm.com>
10731 * coretypes.h (scalar_float_mode): New type.
10732 * machmode.h (mode_traits::from_int): Use machine_mode if
10733 USE_ENUM_MODES is defined.
10734 (is_a): New function.
10736 (dyn_cast): Likewise.
10737 (scalar_float_mode): New class.
10738 (scalar_float_mode::includes_p): New function.
10739 (is_float_mode): Likewise.
10740 * gdbhooks.py (MachineModePrinter): New class.
10741 (build_pretty_printer): Use it for scalar_float_mode.
10742 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
10743 (format_helper::format_helper): Turn into a template.
10744 * genmodes.c (get_mode_class): New function.
10745 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
10746 or machine_mode if none.
10747 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
10748 as_a <scalar_float_mode>.
10749 * dwarf2out.c (mem_loc_descriptor): Likewise.
10750 (insert_float): Likewise.
10751 (add_const_value_attribute): Likewise.
10752 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10753 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
10754 (expand_unop): Update accordingly.
10755 (expand_abs_nojump): Likewise.
10756 (expand_copysign_absneg): Take a scalar_float_mode.
10757 (expand_copysign_bit): Likewise.
10758 (expand_copysign): Update accordingly.
10760 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10761 Alan Hayward <alan.hayward@arm.com>
10762 David Sherwood <david.sherwood@arm.com>
10764 * coretypes.h (opt_mode): New class.
10765 * machmode.h (opt_mode): Likewise.
10766 (opt_mode::else_void): New function.
10767 (opt_mode::require): Likewise.
10768 (opt_mode::exists): Likewise.
10769 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
10770 (GET_MODE_2XWIDER_MODE): Likewise.
10771 (mode_iterator::get_wider): Update accordingly.
10772 (mode_iterator::get_2xwider): Likewise.
10773 (mode_iterator::get_known_wider): Likewise, turning into a template.
10774 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
10775 forcing a wider mode to exist.
10776 * config/cr16/cr16.h (LONG_REG_P): Likewise.
10777 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10778 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
10779 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
10780 * lower-subreg.c (init_lower_subreg): Likewise.
10781 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
10782 on the final iteration.
10783 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
10784 a wider mode exists before asking for a move pattern.
10785 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
10786 forcing a wider mode to exist.
10787 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
10788 returning false if no such mode exists.
10789 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
10790 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
10791 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
10792 Avoid checking for a MODE_INT if we already know the mode is not a
10794 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
10795 forcing a wider mode to exist.
10796 (expmed_mult_highpart_optab): Likewise.
10797 (expmed_mult_highpart): Likewise.
10798 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
10800 * lto-streamer-in.c (lto_input_mode_table): Likewise.
10801 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
10802 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10803 * internal-fn.c (expand_mul_overflow): Update use of
10804 GET_MODE_2XWIDER_MODE.
10805 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10806 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
10807 GET_MODE_WIDER_MODE.
10808 (convert_plusminus_to_widen): Likewise.
10809 * tree-switch-conversion.c (array_value_type): Likewise.
10810 * var-tracking.c (emit_note_insn_var_location): Likewise.
10811 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10812 Return false inside rather than outside the loop if no wider mode
10814 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
10815 and GET_MODE_2XWIDER_MODE
10816 (can_compare_p): Use else_void.
10817 * gdbhooks.py (OptMachineModePrinter): New class.
10818 (build_pretty_printer): Use it for opt_mode.
10820 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10821 Alan Hayward <alan.hayward@arm.com>
10822 David Sherwood <david.sherwood@arm.com>
10824 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
10825 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
10827 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10828 Alan Hayward <alan.hayward@arm.com>
10829 David Sherwood <david.sherwood@arm.com>
10831 * machmode.h (mode_traits): New structure.
10832 (get_narrowest_mode): New function.
10833 (mode_iterator::start): Likewise.
10834 (mode_iterator::iterate_p): Likewise.
10835 (mode_iterator::get_wider): Likewise.
10836 (mode_iterator::get_known_wider): Likewise.
10837 (mode_iterator::get_2xwider): Likewise.
10838 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
10839 (FOR_EACH_MODE): Likewise.
10840 (FOR_EACH_MODE_FROM): Likewise.
10841 (FOR_EACH_MODE_UNTIL): Likewise.
10842 (FOR_EACH_WIDER_MODE): Likewise.
10843 (FOR_EACH_2XWIDER_MODE): Likewise.
10844 * builtins.c (expand_builtin_strlen): Use new mode iterators.
10845 * combine.c (simplify_comparison): Likewise
10846 * config/i386/i386.c (type_natural_mode): Likewise.
10847 * cse.c (cse_insn): Likewise.
10848 * dse.c (find_shift_sequence): Likewise.
10849 * emit-rtl.c (init_derived_machine_modes): Likewise.
10850 (init_emit_once): Likewise.
10851 * explow.c (hard_function_value): Likewise.
10852 * expmed.c (extract_fixed_bit_field_1): Likewise.
10853 (extract_bit_field_1): Likewise.
10854 (expand_divmod): Likewise.
10855 (emit_store_flag_1): Likewise.
10856 * expr.c (init_expr_target): Likewise.
10857 (convert_move): Likewise.
10858 (alignment_for_piecewise_move): Likewise.
10859 (widest_int_mode_for_size): Likewise.
10860 (emit_block_move_via_movmem): Likewise.
10861 (copy_blkmode_to_reg): Likewise.
10862 (set_storage_via_setmem): Likewise.
10863 (compress_float_constant): Likewise.
10864 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10865 * optabs-query.c (get_best_extraction_insn): Likewise.
10866 * optabs.c (expand_binop): Likewise.
10867 (expand_twoval_unop): Likewise.
10868 (expand_twoval_binop): Likewise.
10869 (widen_leading): Likewise.
10870 (widen_bswap): Likewise.
10871 (expand_parity): Likewise.
10872 (expand_unop): Likewise.
10873 (prepare_cmp_insn): Likewise.
10874 (prepare_float_lib_cmp): Likewise.
10875 (expand_float): Likewise.
10876 (expand_fix): Likewise.
10877 (expand_sfix_optab): Likewise.
10878 * postreload.c (move2add_use_add2_insn): Likewise.
10879 * reg-stack.c (reg_to_stack): Likewise.
10880 * reginfo.c (choose_hard_reg_mode): Likewise.
10881 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10882 * stor-layout.c (mode_for_size): Likewise.
10883 (smallest_mode_for_size): Likewise.
10884 (mode_for_vector): Likewise.
10885 (finish_bitfield_representative): Likewise.
10886 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
10887 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
10888 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10889 * var-tracking.c (prepare_call_arguments): Likewise.
10891 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10892 Alan Hayward <alan.hayward@arm.com>
10893 David Sherwood <david.sherwood@arm.com>
10895 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
10896 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
10897 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
10898 * machmode.h (mode_size): Move earlier in file.
10899 (mode_precision): Likewise.
10900 (mode_inner): Likewise.
10901 (mode_nunits): Likewise.
10902 (mode_unit_size): Likewise.
10903 (unit_unit_precision): Likewise.
10904 (mode_wider): Likewise.
10905 (mode_2xwider): Likewise.
10906 (machine_mode): New class.
10907 (mode_to_bytes): New function.
10908 (mode_to_bits): Likewise.
10909 (mode_to_precision): Likewise.
10910 (mode_to_inner): Likewise.
10911 (mode_to_unit_size): Likewise.
10912 (mode_to_unit_precision): Likewise.
10913 (mode_to_nunits): Likewise.
10914 (GET_MODE_SIZE): Use mode_to_bytes.
10915 (GET_MODE_BITSIZE): Use mode_to_bits.
10916 (GET_MODE_PRECISION): Use mode_to_precision.
10917 (GET_MODE_INNER): Use mode_to_inner.
10918 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
10919 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
10920 (GET_MODE_NUNITS): Use mode_to_nunits.
10921 * system.h (ALWAYS_INLINE): New macro.
10922 * config/powerpcspe/powerpcspe-c.c
10923 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
10924 int for arg1_mode and arg2_mode.
10926 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10927 Alan Hayward <alan.hayward@arm.com>
10928 David Sherwood <david.sherwood@arm.com>
10930 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10931 Prefix mode names with E_ in case statements.
10932 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10933 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
10934 (aarch64_split_simd_move): Likewise.
10935 (aarch64_gen_storewb_pair): Likewise.
10936 (aarch64_gen_loadwb_pair): Likewise.
10937 (aarch64_gen_store_pair): Likewise.
10938 (aarch64_gen_load_pair): Likewise.
10939 (aarch64_get_condition_code_1): Likewise.
10940 (aarch64_constant_pool_reload_icode): Likewise.
10941 (get_rsqrte_type): Likewise.
10942 (get_rsqrts_type): Likewise.
10943 (get_recpe_type): Likewise.
10944 (get_recps_type): Likewise.
10945 (aarch64_gimplify_va_arg_expr): Likewise.
10946 (aarch64_simd_container_mode): Likewise.
10947 (aarch64_emit_load_exclusive): Likewise.
10948 (aarch64_emit_store_exclusive): Likewise.
10949 (aarch64_expand_compare_and_swap): Likewise.
10950 (aarch64_gen_atomic_cas): Likewise.
10951 (aarch64_emit_bic): Likewise.
10952 (aarch64_emit_atomic_swap): Likewise.
10953 (aarch64_emit_atomic_load_op): Likewise.
10954 (aarch64_evpc_trn): Likewise.
10955 (aarch64_evpc_uzp): Likewise.
10956 (aarch64_evpc_zip): Likewise.
10957 (aarch64_evpc_ext): Likewise.
10958 (aarch64_evpc_rev): Likewise.
10959 (aarch64_evpc_dup): Likewise.
10960 (aarch64_gen_ccmp_first): Likewise.
10961 (aarch64_gen_ccmp_next): Likewise.
10962 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10963 (alpha_emit_xfloating_libcall): Likewise.
10964 (emit_insxl): Likewise.
10965 (alpha_arg_type): Likewise.
10966 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
10967 (arc_preferred_simd_mode): Likewise.
10968 (arc_secondary_reload): Likewise.
10969 (get_arc_condition_code): Likewise.
10970 (arc_print_operand): Likewise.
10971 (arc_legitimate_constant_p): Likewise.
10972 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10973 * config/arc/arc.md (casesi_load): Likewise.
10974 (casesi_compact_jump): Likewise.
10975 * config/arc/predicates.md (proper_comparison_operator): Likewise.
10976 (cc_use_register): Likewise.
10977 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10978 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10979 (arm_init_iwmmxt_builtins): Likewise.
10980 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
10981 (neon_expand_vector_init): Likewise.
10982 (arm_attr_length_move_neon): Likewise.
10983 (maybe_get_arm_condition_code): Likewise.
10984 (arm_emit_vector_const): Likewise.
10985 (arm_preferred_simd_mode): Likewise.
10986 (arm_output_iwmmxt_tinsr): Likewise.
10987 (thumb1_output_casesi): Likewise.
10988 (thumb2_output_casesi): Likewise.
10989 (arm_emit_load_exclusive): Likewise.
10990 (arm_emit_store_exclusive): Likewise.
10991 (arm_expand_compare_and_swap): Likewise.
10992 (arm_evpc_neon_vuzp): Likewise.
10993 (arm_evpc_neon_vzip): Likewise.
10994 (arm_evpc_neon_vrev): Likewise.
10995 (arm_evpc_neon_vtrn): Likewise.
10996 (arm_evpc_neon_vext): Likewise.
10997 (arm_validize_comparison): Likewise.
10998 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
10999 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11000 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11001 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11002 (c6x_preferred_simd_mode): Likewise.
11003 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11004 (epiphany_rtx_costs): Likewise.
11005 * config/epiphany/predicates.md (proper_comparison_operator):
11007 * config/frv/frv.c (condexec_memory_operand): Likewise.
11008 (frv_emit_move): Likewise.
11009 (output_move_single): Likewise.
11010 (output_condmove_single): Likewise.
11011 (frv_hard_regno_mode_ok): Likewise.
11012 (frv_matching_accg_mode): Likewise.
11013 * config/h8300/h8300.c (split_adds_subs): Likewise.
11014 (h8300_rtx_costs): Likewise.
11015 (h8300_print_operand): Likewise.
11016 (compute_mov_length): Likewise.
11017 (output_logical_op): Likewise.
11018 (compute_logical_op_length): Likewise.
11019 (compute_logical_op_cc): Likewise.
11020 (h8300_shift_needs_scratch_p): Likewise.
11021 (output_a_shift): Likewise.
11022 (compute_a_shift_length): Likewise.
11023 (compute_a_shift_cc): Likewise.
11024 (expand_a_rotate): Likewise.
11025 (output_a_rotate): Likewise.
11026 * config/i386/i386.c (classify_argument): Likewise.
11027 (function_arg_advance_32): Likewise.
11028 (function_arg_32): Likewise.
11029 (function_arg_64): Likewise.
11030 (function_value_64): Likewise.
11031 (ix86_gimplify_va_arg): Likewise.
11032 (ix86_legitimate_constant_p): Likewise.
11033 (put_condition_code): Likewise.
11034 (split_double_mode): Likewise.
11035 (ix86_avx256_split_vector_move_misalign): Likewise.
11036 (ix86_expand_vector_logical_operator): Likewise.
11037 (ix86_split_idivmod): Likewise.
11038 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11039 (ix86_build_const_vector): Likewise.
11040 (ix86_build_signbit_mask): Likewise.
11041 (ix86_match_ccmode): Likewise.
11042 (ix86_cc_modes_compatible): Likewise.
11043 (ix86_expand_branch): Likewise.
11044 (ix86_expand_sse_cmp): Likewise.
11045 (ix86_expand_sse_movcc): Likewise.
11046 (ix86_expand_int_sse_cmp): Likewise.
11047 (ix86_expand_vec_perm_vpermi2): Likewise.
11048 (ix86_expand_vec_perm): Likewise.
11049 (ix86_expand_sse_unpack): Likewise.
11050 (ix86_expand_int_addcc): Likewise.
11051 (ix86_split_to_parts): Likewise.
11052 (ix86_vectorize_builtin_gather): Likewise.
11053 (ix86_vectorize_builtin_scatter): Likewise.
11054 (avx_vpermilp_parallel): Likewise.
11055 (inline_memory_move_cost): Likewise.
11056 (ix86_tieable_integer_mode_p): Likewise.
11057 (x86_maybe_negate_const_int): Likewise.
11058 (ix86_expand_vector_init_duplicate): Likewise.
11059 (ix86_expand_vector_init_one_nonzero): Likewise.
11060 (ix86_expand_vector_init_one_var): Likewise.
11061 (ix86_expand_vector_init_concat): Likewise.
11062 (ix86_expand_vector_init_interleave): Likewise.
11063 (ix86_expand_vector_init_general): Likewise.
11064 (ix86_expand_vector_set): Likewise.
11065 (ix86_expand_vector_extract): Likewise.
11066 (emit_reduc_half): Likewise.
11067 (ix86_emit_i387_round): Likewise.
11068 (ix86_mangle_type): Likewise.
11069 (ix86_expand_round_sse4): Likewise.
11070 (expand_vec_perm_blend): Likewise.
11071 (canonicalize_vector_int_perm): Likewise.
11072 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11073 (expand_vec_perm_1): Likewise.
11074 (expand_vec_perm_interleave3): Likewise.
11075 (expand_vec_perm_even_odd_pack): Likewise.
11076 (expand_vec_perm_even_odd_1): Likewise.
11077 (expand_vec_perm_broadcast_1): Likewise.
11078 (ix86_vectorize_vec_perm_const_ok): Likewise.
11079 (ix86_expand_vecop_qihi): Likewise.
11080 (ix86_expand_mul_widen_hilo): Likewise.
11081 (ix86_expand_sse2_abs): Likewise.
11082 (ix86_expand_pextr): Likewise.
11083 (ix86_expand_pinsr): Likewise.
11084 (ix86_preferred_simd_mode): Likewise.
11085 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11086 * config/i386/sse.md (*andnot<mode>3): Likewise.
11087 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11088 (*<code><mode>3): Likewise.
11089 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11090 (ia64_expand_atomic_op): Likewise.
11091 (ia64_arg_type): Likewise.
11092 (ia64_mode_to_int): Likewise.
11093 (ia64_scalar_mode_supported_p): Likewise.
11094 (ia64_vector_mode_supported_p): Likewise.
11095 (expand_vec_perm_broadcast): Likewise.
11096 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11097 (iq2000_function_arg_advance): Likewise.
11098 (iq2000_function_arg): Likewise.
11099 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11100 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11101 (m68k_libcall_value): Likewise.
11102 (m68k_function_value): Likewise.
11103 (sched_attr_op_type): Likewise.
11104 * config/mcore/mcore.c (mcore_output_move): Likewise.
11105 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11107 (microblaze_function_arg): Likewise.
11108 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11109 (mips_print_operand): Likewise.
11110 (mips_mode_ok_for_mov_fmt_p): Likewise.
11111 (mips_vector_mode_supported_p): Likewise.
11112 (mips_preferred_simd_mode): Likewise.
11113 (mips_expand_vpc_loongson_even_odd): Likewise.
11114 (mips_expand_vec_unpack): Likewise.
11115 (mips_expand_vi_broadcast): Likewise.
11116 (mips_expand_vector_init): Likewise.
11117 (mips_expand_vec_reduc): Likewise.
11118 (mips_expand_msa_cmp): Likewise.
11119 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11120 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11121 (cc_flags_for_mode): Likewise.
11122 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11123 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11124 (nds32_output_casesi_pc_relative): Likewise.
11125 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11126 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11127 (nvptx_gen_unpack): Likewise.
11128 (nvptx_gen_pack): Likewise.
11129 (nvptx_gen_shuffle): Likewise.
11130 (nvptx_gen_wcast): Likewise.
11131 (nvptx_preferred_simd_mode): Likewise.
11132 * config/pa/pa.c (pa_secondary_reload): Likewise.
11133 * config/pa/predicates.md (base14_operand): Likewise.
11134 * config/powerpcspe/powerpcspe-c.c
11135 (altivec_resolve_overloaded_builtin): Likewise.
11136 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11138 (rs6000_preferred_simd_mode): Likewise.
11139 (output_vec_const_move): Likewise.
11140 (rs6000_expand_vector_extract): Likewise.
11141 (rs6000_split_vec_extract_var): Likewise.
11142 (reg_offset_addressing_ok_p): Likewise.
11143 (rs6000_legitimate_offset_address_p): Likewise.
11144 (rs6000_legitimize_address): Likewise.
11145 (rs6000_emit_set_const): Likewise.
11146 (rs6000_const_vec): Likewise.
11147 (rs6000_emit_move): Likewise.
11148 (spe_build_register_parallel): Likewise.
11149 (rs6000_darwin64_record_arg_recurse): Likewise.
11150 (swap_selector_for_mode): Likewise.
11151 (spe_init_builtins): Likewise.
11152 (paired_init_builtins): Likewise.
11153 (altivec_init_builtins): Likewise.
11154 (do_load_for_compare): Likewise.
11155 (rs6000_generate_compare): Likewise.
11156 (rs6000_expand_float128_convert): Likewise.
11157 (emit_load_locked): Likewise.
11158 (emit_store_conditional): Likewise.
11159 (rs6000_output_function_epilogue): Likewise.
11160 (rs6000_handle_altivec_attribute): Likewise.
11161 (rs6000_function_value): Likewise.
11162 (emit_fusion_gpr_load): Likewise.
11163 (emit_fusion_p9_load): Likewise.
11164 (emit_fusion_p9_store): Likewise.
11165 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11166 (fusion_gpr_mem_load): Likewise.
11167 (fusion_addis_mem_combo_load): Likewise.
11168 (fusion_addis_mem_combo_store): Likewise.
11169 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11170 (fusion_gpr_mem_load): Likewise.
11171 (fusion_addis_mem_combo_load): Likewise.
11172 (fusion_addis_mem_combo_store): Likewise.
11173 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11175 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11176 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11177 (rs6000_preferred_simd_mode): Likewise.
11178 (output_vec_const_move): Likewise.
11179 (rs6000_expand_vector_extract): Likewise.
11180 (rs6000_split_vec_extract_var): Likewise.
11181 (reg_offset_addressing_ok_p): Likewise.
11182 (rs6000_legitimate_offset_address_p): Likewise.
11183 (rs6000_legitimize_address): Likewise.
11184 (rs6000_emit_set_const): Likewise.
11185 (rs6000_const_vec): Likewise.
11186 (rs6000_emit_move): Likewise.
11187 (rs6000_darwin64_record_arg_recurse): Likewise.
11188 (swap_selector_for_mode): Likewise.
11189 (paired_init_builtins): Likewise.
11190 (altivec_init_builtins): Likewise.
11191 (rs6000_expand_float128_convert): Likewise.
11192 (emit_load_locked): Likewise.
11193 (emit_store_conditional): Likewise.
11194 (rs6000_output_function_epilogue): Likewise.
11195 (rs6000_handle_altivec_attribute): Likewise.
11196 (rs6000_function_value): Likewise.
11197 (emit_fusion_gpr_load): Likewise.
11198 (emit_fusion_p9_load): Likewise.
11199 (emit_fusion_p9_store): Likewise.
11200 * config/rx/rx.c (rx_gen_move_template): Likewise.
11201 (flags_from_mode): Likewise.
11202 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11203 (s390_slb_comparison): Likewise.
11204 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11205 (s390_vector_mode_supported_p): Likewise.
11206 (s390_cc_modes_compatible): Likewise.
11207 (s390_match_ccmode_set): Likewise.
11208 (s390_canonicalize_comparison): Likewise.
11209 (s390_emit_compare_and_swap): Likewise.
11210 (s390_branch_condition_mask): Likewise.
11211 (s390_rtx_costs): Likewise.
11212 (s390_secondary_reload): Likewise.
11213 (__SECONDARY_RELOAD_CASE): Likewise.
11214 (s390_expand_cs): Likewise.
11215 (s390_preferred_simd_mode): Likewise.
11216 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11217 * config/sh/sh.c (sh_print_operand): Likewise.
11218 (dump_table): Likewise.
11219 (sh_secondary_reload): Likewise.
11220 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11221 * config/sh/sh.md (casesi_worker_1): Likewise.
11222 (casesi_worker_2): Likewise.
11223 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11224 (fcc_comparison_operator): Likewise.
11225 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11226 (emit_soft_tfmode_cvt): Likewise.
11227 (sparc_preferred_simd_mode): Likewise.
11228 (output_cbranch): Likewise.
11229 (sparc_print_operand): Likewise.
11230 (sparc_expand_vec_perm_bmask): Likewise.
11231 (vector_init_bshuffle): Likewise.
11232 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11233 (spu_vector_mode_supported_p): Likewise.
11234 (spu_expand_insv): Likewise.
11235 (spu_emit_branch_or_set): Likewise.
11236 (spu_handle_vector_attribute): Likewise.
11237 (spu_builtin_splats): Likewise.
11238 (spu_builtin_extract): Likewise.
11239 (spu_builtin_promote): Likewise.
11240 (spu_expand_sign_extend): Likewise.
11241 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11242 (tilegx_simd_int): Likewise.
11243 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11244 (tilepro_simd_int): Likewise.
11245 * config/v850/v850.c (const_double_split): Likewise.
11246 (v850_print_operand): Likewise.
11247 (ep_memory_offset): Likewise.
11248 * config/vax/vax.c (vax_rtx_costs): Likewise.
11249 (vax_output_int_move): Likewise.
11250 (vax_output_int_add): Likewise.
11251 (vax_output_int_subtract): Likewise.
11252 * config/visium/predicates.md (visium_branch_operator): Likewise.
11253 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
11254 (visium_print_operand_address): Likewise.
11255 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11256 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
11257 (xtensa_expand_conditional_branch): Likewise.
11258 (xtensa_copy_incoming_a7): Likewise.
11259 (xtensa_output_literal): Likewise.
11260 * dfp.c (decimal_real_maxval): Likewise.
11261 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11264 Alan Hayward <alan.hayward@arm.com>
11265 David Sherwood <david.sherwood@arm.com>
11267 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
11268 (mode_nunits_inline): Likewise.
11269 (mode_inner_inline): Likewise.
11270 (mode_unit_size_inline): Likewise.
11271 (mode_unit_precision_inline): Likewise.
11272 (emit_insn_modes_h): Likewise. Also emit a #define of the
11274 (emit_mode_wider): Add an E_ prefix to mode names.
11275 (emit_mode_complex): Likewise.
11276 (emit_mode_inner): Likewise.
11277 (emit_mode_adjustments): Likewise.
11278 (emit_mode_int_n): Likewise.
11279 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
11280 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
11281 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
11282 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
11283 (CRC32_BUILTIN, ENTRY): Likewise.
11284 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
11285 (aarch64_pop_regs): Likewise.
11286 (aarch64_process_components): Likewise.
11287 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
11288 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
11289 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
11290 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
11291 * config/arm/arm.c (arm_init_libfuncs): Likewise.
11292 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
11294 * config/i386/i386-builtin.def (pcmpestr): Likewise.
11295 (pcmpistr): Likewise.
11296 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
11297 * config/mmix/mmix.c (mmix_output_condition): Likewise.
11298 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11300 * config/rl78/rl78.c (mduc_regs): Likewise.
11301 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11302 (htm_expand_builtin): Likewise.
11303 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
11304 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
11305 * config/xtensa/xtensa.c (print_operand): Likewise.
11306 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
11307 (NUM_MODE_VECTOR_INT): Likewise.
11308 * genoutput.c (null_operand): Likewise.
11309 (output_operand_data): Likewise.
11310 * genrecog.c (print_parameter_value): Likewise.
11311 * lra.c (debug_operand_data): Likewise.
11313 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11315 * dwarf2out.c (scompare_loc_descriptor_wide)
11316 (scompare_loc_descriptor_narrow): New functions, split out from...
11317 (scompare_loc_descriptor): ...here.
11318 * expmed.c (emit_store_flag_int): New function, split out from...
11319 (emit_store_flag): ...here.
11321 2017-08-30 Richard Biener <rguenther@suse.de>
11323 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
11324 (dwarf2out_early_finish): Move setting of AT_pubnames from
11325 early debug output to early finish.
11327 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11329 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
11330 and -mdata-region to the assembler.
11332 2017-08-30 Richard Biener <rguenther@suse.de>
11334 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
11336 (gen_subprogram_die): Add DW_AT_object_pointer only early.
11337 (dwarf2out_early_global_decl): Only generate a DIE for the
11338 abstract origin if it doesn't already exist or is a declaration DIE.
11339 (resolve_addr): Do not add the linkage name twice when
11340 generating a stub DIE for the DW_TAG_GNU_call_site target.
11342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11344 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11345 Use machine_mode rather than int for arg1_mode.
11347 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
11350 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
11351 that the second argument of the built-in functions to unpack
11352 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
11353 switch statement instead a lot of if statements.
11354 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
11355 Allow 64-bit values to be in Altivec registers as well as
11356 traditional floating point registers.
11357 (pack<mode>, FMOVE128_VSX iterator): Likewise.
11359 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
11361 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
11362 MAX_REGS_PER_ADDRESS == 1.
11364 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
11366 * config/i386/i386.opt (flag_fentry): Do not init to -1.
11367 * config/i386/i386.c (ix86_option_override_internal): Simplify
11368 setting of opts->x_flag_entry.
11370 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11371 Jakub Jelinek <jakub@redhat.com>
11372 Richard Biener <rguenther@suse.de>
11374 PR tree-optimization/81503
11375 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
11376 folded constant fits in the target type; reorder tests for clarity.
11378 2017-08-29 Martin Liska <mliska@suse.cz>
11380 * passes.def: Include pass_lower_switch.
11381 * stmt.c (dump_case_nodes): Remove and move to
11382 tree-switch-conversion.
11383 (case_values_threshold): Likewise.
11384 (expand_switch_as_decision_tree_p): Likewise.
11385 (emit_case_decision_tree): Likewise.
11386 (expand_case): Likewise.
11387 (balance_case_nodes): Likewise.
11388 (node_has_low_bound): Likewise.
11389 (node_has_high_bound): Likewise.
11390 (node_is_bounded): Likewise.
11391 (emit_case_nodes): Likewise.
11392 (struct simple_case_node): New struct.
11393 (add_case_node): Remove.
11394 (emit_case_dispatch_table): Use vector instead of case_list.
11395 (reset_out_edges_aux): Remove.
11396 (compute_cases_per_edge): Likewise.
11397 (expand_case): Build list of simple_case_node.
11398 (expand_sjlj_dispatch_table): Use it.
11399 * tree-switch-conversion.c (struct case_node): Moved from
11400 stmt.c and adjusted.
11401 (emit_case_nodes): Likewise.
11402 (node_has_low_bound): Likewise.
11403 (node_has_high_bound): Likewise.
11404 (node_is_bounded): Likewise.
11405 (case_values_threshold): Likewise.
11406 (reset_out_edges_aux): Likewise.
11407 (compute_cases_per_edge): Likewise.
11408 (add_case_node): Likewise.
11409 (dump_case_nodes): Likewise.
11410 (balance_case_nodes): Likewise.
11411 (expand_switch_as_decision_tree_p): Likewise.
11412 (emit_jump): Likewise.
11413 (emit_case_decision_tree): Likewise.
11414 (try_switch_expansion): Likewise.
11415 (do_jump_if_equal): Likewise.
11416 (emit_cmp_and_jump_insns): Likewise.
11417 (fix_phi_operands_for_edge): New function.
11418 (record_phi_operand_mapping): Likewise.
11419 (class pass_lower_switch): New pass.
11420 (pass_lower_switch::execute): New function.
11421 (make_pass_lower_switch): Likewise.
11422 (conditional_probability):
11423 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
11424 * tree-pass.h: Add make_pass_lower_switch.
11426 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11429 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
11432 2017-08-29 Richard Biener <rguenther@suse.de>
11434 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
11435 we do not add a DW_AT_inline attribute twice.
11436 (gen_subprogram_die): Remove code setting DW_AT_inline on
11437 DECL_ABSTRACT_P nodes.
11439 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
11441 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
11442 calls to internal functions.
11443 (gimplify_modify_expr): Likewise.
11444 * tree-call-cdce.c (use_internal_fn): Likewise.
11445 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
11446 (convert_to_divmod): Set the nothrow flag.
11447 * tree-if-conv.c (predicate_mem_writes): Likewise.
11448 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11449 (vectorizable_call): Likewise.
11450 (vectorizable_store): Likewise.
11451 (vectorizable_load): Likewise.
11452 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
11453 (vect_recog_mask_conversion_pattern): Likewise.
11455 2017-08-29 Martin Liska <mliska@suse.cz>
11458 * gcc.c (driver_handle_option): Add new argument.
11459 * opts-common.c (handle_option): Pass
11460 target_option_override_hook.
11461 * opts-global.c (lang_handle_option): Add new option.
11462 (set_default_handlers): Add new argument.
11463 (decode_options): Likewise.
11464 * opts.c (target_handle_option): Likewise.
11465 (common_handle_option): Call target_option_override_hook.
11466 * opts.h (struct cl_option_handler_func): Add hook for
11467 target option override.
11468 (struct cl_option_handlers): Likewise.
11469 (set_default_handlers): Add new argument.
11470 (decode_options): Likewise.
11471 (common_handle_option): Likewise.
11472 (target_handle_option): Likewise.
11473 * toplev.c (toplev::main): Pass targetm.target_option.override
11476 2017-08-29 Richard Biener <rguenther@suse.de>
11477 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
11479 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
11480 life to the active subtree.
11482 2017-08-28 Jeff Law <law@redhat.com>
11484 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
11485 derive_equivalences.
11486 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
11487 Code moved into....
11488 (edge_info::derive_equivalences): New private member function
11490 * tree-ssa-dom.c (class edge_info): Changed from a struct
11491 to a class. Add ctor/dtor, methods and data members.
11492 (edge_info::edge_info): Renamed from allocate_edge_info.
11493 Initialize additional members.
11494 (edge_info::~edge_info): New.
11495 (free_dom_edge_info): Delete the edge info.
11496 (record_edge_info): Use new class & associated member functions.
11497 Tighten forms for testing for edge equivalences.
11498 (record_temporary_equivalences): Iterate over the simple
11499 equivalences rather than assuming there's only one per edge.
11500 (cprop_into_successor_phis): Iterate over the simple
11501 equivalences rather than assuming there's only one per edge.
11502 (optimize_stmt): Use operand_equal_p rather than pointer
11503 equality for mini-DSE code.
11505 2017-08-28 Nathan Sidwell <nathan@acm.org>
11507 * gcc.c (execute): Fold SIGPIPE handling into switch
11508 statement. Adjust internal error message.
11510 2017-08-28 Richard Biener <rguenther@suse.de>
11513 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
11514 Do nothing for removed DIEs.
11516 2017-08-28 Richard Biener <rguenther@suse.de>
11518 PR tree-optimization/81977
11519 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
11522 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
11525 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
11526 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
11527 using targetm.gen_mem_thread_fence.
11529 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
11532 * config/i386/i386.md (*<btsc><mode>): Change operand 2
11533 predicate to register_operand. Reorder operands.
11534 (*btr<mode>): Ditto.
11535 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
11536 (*btr<mode>_mask): Ditto.
11538 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
11540 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
11541 * config/rs6000/xmmintrin.h: New file.
11542 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
11544 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11547 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
11548 parameter and_insn and return it.
11549 (recombine_lvx_pattern): Insert a copy to ensure availability of
11550 the base register of the copied masking operation at the point of
11551 the instruction replacement.
11552 (recombine_stvx_pattern): Likewise.
11554 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
11556 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
11557 undocumented switches.
11558 (-mpower9-dform-vector): Likewise.
11559 (-mpower9-dform): Likewise.
11560 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
11561 comments to delete references to -mpower9-dform* switches.
11562 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
11563 Delete reference to -mpower9-dform* switches, test for
11564 -mpower9-vector instead.
11565 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
11566 (OTHER_P9_VECTOR_MASKS): Likewise.
11567 (POWERPC_MASKS): Likewise.
11568 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
11569 tests against -mpower9-dform* to -mpower9-vector. Delete code
11570 that checked for -mpower9-dform* consistancy with other options.
11571 Add test for -mpower9-misc to enable other power9 switches.
11572 (rs6000_init_hard_regno_mode_ok): Likewise.
11573 (rs6000_option_override_internal): Likewise.
11574 (rs6000_emit_prologue): Likewise.
11575 (rs6000_emit_epilogue): Likewise.
11576 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
11577 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
11578 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
11580 (emit_fusion_p9_store): Likewise.
11581 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
11582 resetting these macros if the assembler does not support ISA 3.0
11584 (TARGET_P9_DFORM_VECTOR): Likewise.
11585 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
11586 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
11588 2017-08-25 Alan Modra <amodra@gmail.com>
11590 PR rtl-optimization/81747
11591 * cse.c (cse_extended_basic_block): Don't attempt to record
11592 equivalences for degenerate conditional branches that branch
11593 to their fall-through.
11595 2017-08-24 Martin Sebor <msebor@redhat.com>
11597 PR middle-end/81908
11598 * gimple-fold.c (size_must_be_zero_p): New function.
11599 (gimple_fold_builtin_memory_op): Call it.
11601 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
11603 * config/rs6000/mm_malloc.h: New file.
11605 2017-08-24 Bin Cheng <bin.cheng@arm.com>
11607 PR tree-optimization/81913
11608 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
11609 analysis when either IVs in condition can wrap.
11611 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
11613 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
11614 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
11616 2017-08-24 Richard Biener <rguenther@suse.de>
11619 * targhooks.c (default_target_can_inline_p): Properly
11620 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
11621 is present and always compare.
11622 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
11623 infer -mfpmath=sse from TARGET_SSE_P.
11624 (ix86_can_inline_p): Properly use target_option_default_node when
11625 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
11627 2017-08-24 Richard Biener <rguenther@suse.de>
11630 * dwarf2out.c (output_die): Handle flag_generate_offload like
11632 (output_comp_unit): Likewise.
11633 (gen_array_type_die): Likewise.
11634 (dwarf2out_early_finish): Likewise.
11635 (note_variable_value_in_expr): Likewise.
11636 (dwarf2out_finish): Likewise. Adjust assert.
11637 * cgraphunit.c (symbol_table::compile): Move setting of
11638 flag_generate_offload earlier ...
11639 (symbol_table::finalize_compilation_unit): ... here, before
11640 early debug finalization.
11642 2017-08-24 Richard Biener <rguenther@suse.de>
11644 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
11645 and ipa-fnsummary.h.
11646 (ix86_can_inline_p): When ix86_fpmath flags do not match
11647 check whether the callee uses FP math at all.
11649 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
11651 PR middle-end/81931
11652 * tree-ssanames.c (get_nonzero_bits): Use element_precision
11653 instead of TYPE_PRECISION.
11655 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
11656 Alan Hayward <alan.hayward@arm.com>
11657 David Sherwood <david.sherwood@arm.com>
11659 * combine.c (make_extraction): Use subreg_offset_from_lsb.
11661 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
11663 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
11665 (ix86_frame::stack_realign_allocate): New field.
11666 (struct machine_frame_state): Modify comments.
11667 (machine_frame_state::sp_realigned_fp_end): New field.
11668 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
11669 layout calculation.
11670 (sp_valid_at): Add assertion to assure no attempt to access invalid
11671 offset of a realigned stack.
11672 (fp_valid_at): Likewise.
11673 (choose_baseaddr): Modify comments.
11674 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
11675 ix86_expand_prologue.
11676 (ix86_expand_prologue): Modify stack realignment and allocation.
11677 (ix86_expand_epilogue): Modify comments.
11678 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
11679 avx2_runtime, avx512f, and avx512f_runtime.
11681 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
11683 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
11684 (mstackrealign): Do not init to -1.
11685 * config/i386/i386.c (ix86_option_override_internal):
11686 Check opts_set, not opts when setting default value of
11687 opts->x_ix86_force_align_arg_pointer.
11689 2017-08-23 Richard Biener <rguenther@suse.de>
11691 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
11692 lang_hooks.decl_printable_name.
11693 * print-rtl-function.c (print_rtx_function): Likewise.
11694 * tree-pretty-print.c (dump_function_header): Likewise.
11696 2017-08-23 Richard Biener <rguenther@suse.de>
11699 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
11700 -g0 at compile-time.
11702 2017-08-23 Tamar Christina <tamar.christina@arm.com>
11704 PR middle-end/19706
11705 * doc/sourcebuild.texi (Other hardware attributes):
11708 2017-08-23 Tamar Christina <tamar.christina@arm.com>
11710 PR middle-end/19706
11711 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
11712 Fix single-use check.
11714 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
11716 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
11718 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
11720 * doc/install.texi: Modify to add more details on running selected
11723 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
11725 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
11726 is combined with -mabi=ms.
11727 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
11730 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11732 PR tree-optimization/81488
11733 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
11734 and cached_basis fields.
11735 (MAX_SPREAD): New constant.
11736 (alloc_cand_and_find_basis): Initialize new fields.
11737 (clear_visited): New function.
11738 (create_phi_basis_1): Rename from create_phi_basis, set visited
11739 and cached_basis fields.
11740 (create_phi_basis): New wrapper function.
11741 (phi_add_costs_1): Rename from phi_add_costs, add spread
11742 parameter, set visited field, short-circuit when limits reached.
11743 (phi_add_costs): New wrapper function.
11744 (record_phi_increments_1): Rename from record_phi_increments, set
11746 (record_phi_increments): New wrapper function.
11747 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
11748 (phi_incr_cost): New wrapper function.
11749 (all_phi_incrs_profitable_1): Rename from
11750 all_phi_incrs_profitable, set visited field.
11751 (all_phi_incrs_profitable): New wrapper function.
11753 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
11754 Alan Hayward <alan.hayward@arm.com>
11755 David Sherwood <david.sherwood@arm.com>
11757 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
11758 that takes the outer and inner modes.
11759 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
11760 comparison as the canonical test for a paradoxical subreg.
11761 * combine.c (simplify_set): Use paradoxical_subreg_p.
11762 (make_extraction): Likewise.
11763 (force_to_mode): Likewise.
11764 (rtx_equal_for_field_assignment_p): Likewise.
11765 (gen_lowpart_for_combine): Likewise.
11766 (simplify_comparison): Likewise.
11767 * cse.c (equiv_constant): Likewise.
11768 * expmed.c (store_bit_field_1): Likewise.
11769 * final.c (alter_subreg): Likewise.
11770 * fwprop.c (propagate_rtx): Likewise.
11771 (forward_propagate_subreg): Likewise.
11772 * ira-conflicts.c (ira_build_conflicts): Likewise.
11773 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
11774 * lra-constraints.c (curr_insn_transform): Likewise.
11775 (split_reg): Likewise.
11776 * lra-eliminations.c (move_plus_up): Likewise.
11777 (lra_eliminate_regs_1): Likewise.
11778 * recog.c (general_operand): Likewise.
11779 * ree.c (combine_reaching_defs): Likewise.
11780 * reload.c (push_reload): Likewise.
11781 (find_reloads): Likewise.
11782 * reload1.c (elimination_effects): Likewise.
11783 (compute_reload_subreg_offset): Likewise.
11784 (choose_reload_regs): Likewise.
11785 * rtlanal.c (subreg_lsb_1): Likewise.
11786 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11787 (simplify_subreg): Likewise.
11788 * var-tracking.c (track_loc_p): Likewise.
11789 * emit-rtl.c (byte_lowpart_offset): Likewise.
11790 (paradoxical_subreg_p): Delete out-of-line definition.
11792 2017-08-22 Jeff Law <law@redhat.com>
11794 PR tree-optimization/81741
11795 PR tree-optimization/71947
11796 * tree-ssa-dom.c: Include tree-inline.h.
11797 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
11798 equivalences if one is more expensive to compute than the other.
11799 * tree-ssa-scopedtables.h (class const_or_copies): Make
11800 record_const_or_copy_raw method private.
11801 (class avail_exprs_stack): New method simplify_binary_operation.
11802 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
11803 avail_exprs_stack::simplify_binary_operation as needed.
11804 (avail_exprs_stack::simplify_binary_operation): New function.
11806 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
11808 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
11809 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
11810 (DOT_SYMBOLS): Likewise.
11811 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
11812 (RELOCATABLE_NEEDS_FIXUP): Likewise.
11813 (RS6000_ABI_NAME): Likewise.
11814 (TARGET_CMODEL): Likewise.
11815 (TOC_SECTION_ASM_OP): Likewise.
11816 (SET_CMODEL): New macro.
11817 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
11819 2017-08-22 Richard Biener <rguenther@suse.de>
11821 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
11822 to allow for free-lang-data replacements similar to verify_type_variant.
11824 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
11825 Alan Hayward <alan.hayward@arm.com>
11826 David Sherwood <david.sherwood@arm.com>
11828 * config/aarch64/aarch64.md (casesi): Use DImode rather than
11829 VOIDmode for the LABEL_REF.
11831 2017-08-22 Richard Biener <rguenther@suse.de>
11833 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
11835 2017-08-22 Richard Biener <rguenther@suse.de>
11837 * common.opt (feliminate-dwarf2-dups): Ignore.
11838 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
11839 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
11840 same_die_p_wrap, compute_section_prefix,
11841 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
11842 (comdat_symbol_id, comdat_symbol_number): Likewise.
11843 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
11845 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
11846 (output_die): Mark unreachable path unreachable.
11847 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
11848 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
11849 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
11850 (dwarf2out_early_finish): Likewise.
11852 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
11854 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
11856 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
11859 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
11860 not VAR_P. Filter attribute warnings with OPT_Wattributes.
11861 (avr_attribute_table) <io, io_low, address>: Initialize
11862 .decl_required with true.
11864 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
11866 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
11867 undocumented debugging options.
11868 (-mvsx-scalar-double): Likewise.
11869 (-mallow-df-permute): Likewise.
11870 (-mvectorize-builtins): Likewise.
11871 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11872 (rs6000_builtin_vectorized_function): Likewise.
11873 (rs6000_builtin_md_vectorized_function): Likewise.
11874 (rs6000_opt_vars): Likewise.
11876 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
11879 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
11880 (*btrq_imm): Rename from *btrq.
11881 (*btcq_imm): Rename from *btcq.
11882 (btsc): New code attribute.
11883 (*<btsc><mode>): New insn pattern.
11884 (*btr<mode>): Ditto.
11885 (*<btsc><mode>_mask): New insn_and_split pattern.
11886 (*btr<mode>_mask): Ditto.
11888 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11889 Alan Hayward <alan.hayward@arm.com>
11890 David Sherwood <david.sherwood@arm.com>
11892 * function.c (pad_below): Simplify padding calculation.
11894 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11895 Alan Hayward <alan.hayward@arm.com>
11896 David Sherwood <david.sherwood@arm.com>
11898 * target.def (function_prologue): Remove frame size argument.
11899 (function_epilogue): Likewise.
11900 * doc/tm.texi: Regenerate.
11901 * final.c (final_start_function): Update call to function_prologue.
11902 (final_end_function): Update call to function_epilogue.
11903 (default_function_pro_epilogue): Remove frame size argument.
11904 * output.h (default_function_pro_epilogue): Likewise.
11905 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
11906 (arm_output_function_prologue): Likewise.
11907 * config/frv/frv.c (frv_function_prologue): Likewise.
11908 (frv_function_epilogue): Likewise.
11909 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
11910 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
11911 (ia64_output_function_epilogue): Likewise.
11912 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
11913 (m32r_output_function_epilogue): Likewise.
11914 * config/microblaze/microblaze.c (microblaze_function_prologue)
11915 (microblaze_function_epilogue): Likewise.
11916 * config/mips/mips.c (mips_output_function_prologue): Likewise.
11917 (mips_output_function_epilogue): Likewise.
11918 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
11919 (mmix_target_asm_function_epilogue): Likewise.
11920 * config/msp430/msp430.c (msp430_start_function): Likewise.
11921 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
11922 (nds32_asm_function_epilogue): Likewise.
11923 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
11924 * config/pa/pa.c (pa_output_function_prologue): Likewise.
11925 (pa_output_function_epilogue): Likewise.
11926 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
11927 (rs6000_output_function_epilogue): Likewise.
11928 * config/rl78/rl78.c (rl78_start_function): Likewise.
11929 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
11930 (rs6000_output_function_epilogue): Likewise.
11931 * config/rx/rx.c (rx_output_function_prologue): Likewise.
11932 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
11933 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
11934 (sparc_asm_function_epilogue): Likewise.
11936 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11938 * tree.h (type_has_mode_precision_p): New function.
11939 * convert.c (convert_to_integer_1): Use it.
11940 * expr.c (expand_expr_real_2): Likewise.
11941 (expand_expr_real_1): Likewise.
11942 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
11943 * match.pd: Likewise.
11944 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
11945 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
11946 * tree-tailcall.c (process_assignment): Likewise.
11947 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11948 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
11949 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
11950 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11951 (vectorizable_assignment): Likewise.
11952 (vectorizable_shift): Likewise.
11953 (vectorizable_operation): Likewise.
11954 * tree-vrp.c (register_edge_assert_for_2): Likewise.
11956 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
11958 * match.pd: Add pow (C, x) simplification.
11960 2017-08-21 Richard Biener <rguenther@suse.de>
11962 PR tree-optimization/81900
11963 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
11964 for blocks with abnormal predecessors.
11965 (compute_antic): Do not set visited flag prematurely.
11967 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
11970 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
11972 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11974 * stor-layout.h (vector_type_mode): Move to...
11975 * tree.h (vector_type_mode): ...here.
11976 * stor-layout.c (vector_type_mode): Move to...
11977 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
11979 2017-08-21 Richard Biener <rguenther@suse.de>
11981 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
11982 register_external_die hooks.
11983 (debug_false_tree_charstarstar_uhwistar): Declare.
11984 (debug_nothing_tree_charstar_uhwi): Likewise.
11985 * debug.c (do_nothing_debug_hooks): Adjust.
11986 (debug_false_tree_charstarstar_uhwistar): New do nothing.
11987 (debug_nothing_tree_charstar_uhwi): Likewise.
11988 * dbxout.c (dbx_debug_hooks): Adjust.
11989 (xcoff_debug_hooks): Likewise.
11990 * sdbout.c (sdb_debug_hooks): Likewise.
11991 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11992 * dwarf2out.c (macinfo_label_base): New global.
11993 (dwarf2out_register_external_die): New function for the
11994 register_external_die hook.
11995 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
11996 (dwarf2_debug_hooks): Use them.
11997 (dwarf2_lineno_debug_hooks): Adjust.
11998 (struct die_struct): Add with_offset flag.
11999 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12000 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12001 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12002 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12003 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12004 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12005 defining section names for the early LTO debug variants.
12006 (reset_indirect_string): New helper.
12007 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12008 (print_dw_val): Add support for offsetted symbol references.
12009 (get_ultimate_context): Split out from is_cxx.
12010 (is_cxx): Use get_ultimate_context.
12011 (is_fortran): Add decl overload.
12012 (compute_comp_unit_symbol): Split out worker from
12013 compute_section_prefix.
12014 (compute_section_prefix): Call compute_comp_unit_symbol and
12015 set comdat_type_p here.
12016 (output_die): Skip DIE symbol output for the LTO added one.
12017 Handle DIE symbol references with offset.
12018 (output_comp_unit): Guard section name mangling properly.
12019 For LTO debug sections emit a symbol at the section beginning
12020 which we use to refer to its DIEs.
12021 (add_abstract_origin_attribute): For DIEs registered via
12022 dwarf2out_register_external_die directly refer to the early
12023 DIE rather than indirectly through the shadow one we created.
12024 Remove obsolete call to dwarf2out_abstract_function for
12025 non-function/block origins.
12026 (gen_array_type_die): When generating early LTO debug do
12027 not emit DW_AT_string_length.
12028 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12029 late when in LTO. As suggested place a gcc_unreachable for
12030 the DECL_ABSTRACT_P case.
12031 (gen_subprogram_die): Avoid another specification DIE
12032 for early built declarations/definitions for the late LTO case.
12033 (gen_variable_die): Add type references for late duplicated VLA dies
12035 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12036 we have the abstract instance already.
12037 (process_scope_var): Adjust decl DIE contexts in LTO which
12038 first puts them in limbo.
12039 (gen_decl_die): Do not generate type DIEs late apart from
12040 types for VLAs or for decls we do not yet have a DIE. Do not
12041 call dwarf2out_abstract_function late.
12042 (dwarf2out_early_global_decl): Make sure to create DIEs
12043 for abstract instances of a decl first.
12044 (dwarf2out_late_global_decl): Adjust comment.
12045 (output_macinfo_op): With multiple macro sections use
12046 macinfo_label_base to distinguish labels.
12047 (output_macinfo): Likewise. Update macinfo_label_base.
12048 Pass in the line info label.
12049 (note_variable_value_in_expr): When generating LTO resolve
12050 all variable values here by generating DIEs as needed.
12051 (init_sections_and_labels): Add early LTO debug flag parameter
12052 and generate different sections and names if set. Add generation
12053 counter for the labels so we can have multiple of them.
12054 (reset_dies): Helper to allow DIEs to be output multiple times.
12055 (dwarf2out_finish): When outputting DIEs to the fat part of an
12056 LTO object first reset DIEs.
12057 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12058 (modified_type_die): Check for decl_ultimate_origin being self
12060 (gen_type_die_with_usage): Likewise.
12061 (gen_typedef_die): Allow decl_ultimate_origin being self.
12062 (set_decl_abstract_flags): Remove.
12063 (set_block_abstract_flags): Likewise.
12064 (dwarf2out_abstract_function): Treat the early generated DIEs
12065 as the abstract copy and only add DW_AT_inline and
12066 DW_AT_artificial here and call set_decl_origin_self.
12067 If the DIE has an abstract origin don't do anything.
12068 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12069 if we have none yet (Go fails to build one, PR78628).
12070 (variably_modified_type_p): Prevent endless recursion for Ada
12071 cyclic pointer types.
12072 * lto-streamer-in.c: Include debug.h.
12073 (dref_queue): New global.
12074 (lto_read_tree_1): Stream in DIE references.
12075 (lto_input_tree): Register DIE references.
12076 (input_function): Stream DECL_DEBUG_ARGS.
12077 * lto-streamer-out.c: Include debug.h.
12078 (lto_write_tree_1): Output DIE references.
12079 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12080 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12081 (output_function): Stream DECL_DEBUG_ARGS.
12082 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12083 Stream DECL_ABSTRACT_ORIGIN.
12084 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12085 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12086 DECL_CONTEXT for file-scope decls.
12087 * lto-streamer.h (struct dref_entry): Declare.
12088 (dref_queue): Likewise.
12089 * cfgexpand.c (pass_expand::execute): Do not call the
12090 outlining_inline_function hook here.
12091 * lto-wrapper.c (debug_obj): New global.
12092 (tool_cleanup): Unlink it if required.
12093 (debug_objcopy): New function.
12094 (run_gcc): Handle early debug sections in the IL files by
12095 extracting them to separate files, partially linkin them and
12096 feeding the result back as result to the linker.
12097 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12098 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12099 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12100 sections into a separate segment.
12101 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12103 (darwin_asm_dwarf_section): Likewise.
12104 (darwin_asm_output_dwarf_offset): Likewise.
12105 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12107 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12108 Alan Hayward <alan.hayward@arm.com>
12109 David Sherwood <david.sherwood@arm.com>
12111 * read-md.h (md_reader::record_potential_iterator_use): Replace
12112 pointer argument with an rtx and an index.
12113 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12114 (apply_mode_iterator): Likewise.
12115 (apply_code_iterator): Likewise.
12116 (apply_int_iterator): Likewise.
12117 (apply_subst_iterator): Likewise.
12118 (record_iterator_use): Likewise.
12119 (record_attribute_use): Likewise.
12120 (md_reader::record_potential_iterator_use): Likewise. Update calls
12121 to record_iterator_use and apply_iterator.
12122 (iterator_use): Replace ptr with x and index.
12123 (attribute_use): Likewise.
12124 (apply_attribute_uses): Update calls to apply_iterator.
12125 (apply_iterators): Likewise. Update initialization of iterator_use.
12126 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12127 and record_potential_iterator_use.
12128 (rtx_reader::read_rtx_operand): Likewise.
12130 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12131 Alan Hayward <alan.hayward@arm.com>
12132 David Sherwood <david.sherwood@arm.com>
12134 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12137 2017-08-21 Richard Biener <rguenther@suse.de>
12139 PR middle-end/81884
12140 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12141 at struct end conservatively when comparing common bases.
12143 2017-08-21 Richard Biener <rguenther@suse.de>
12145 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12146 (mem_ref_in_stmt): Remove.
12147 (determine_max_movement): Use ref index to get at the reference.
12148 (invariantness_dom_walker::before_dom_children): Deal with
12149 lim data already initialized.
12150 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12152 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12154 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12155 (TARGET_ISA_ROUND): Ditto.
12156 (TARGET_ROUND): Ditto.
12157 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12158 * config/i386/i386.md: Ditto.
12159 * config/i386/sse.md: Ditto.
12160 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12161 with OPTION_MASK_ISA_SSE4_1.
12163 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12166 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12167 __builtin_ia32_lzcnt_u16.
12169 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12172 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12173 (rs6000_set_current_function): Rewrite function to use it.
12175 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12178 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12179 and TYPE_WARN_IF_NOT_ALIGN.
12180 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12181 (handle_warn_if_not_align): New.
12182 (place_union_field): Call handle_warn_if_not_align.
12183 (place_field): Call handle_warn_if_not_align.
12184 Copy TYPE_WARN_IF_NOT_ALIGN.
12185 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12186 (layout_type): Likewise.
12187 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12189 (tree_decl_common): Add warn_if_not_align.
12190 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12191 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12192 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12193 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12194 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12195 * doc/extend.texi: Document warn_if_not_aligned attribute.
12196 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12198 2017-08-17 Martin Liska <mliska@suse.cz>
12201 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12202 (get_data_dependence): Use it as pointer type.
12203 (distribute_loop): Likewise.
12205 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12207 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12208 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12209 (p8_vmrgow_<mode>_direct): New define_insn.
12210 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12211 handle endianness for vmrgew and vmrgow permute patterns.
12213 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12215 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12216 * config/rs6000/rs6000-cpus.def: Remove comment.
12217 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12218 (POWERPC_MASKS): Likewise.
12219 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12220 use of TARGET_VSX_TIMODE.
12221 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12222 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12223 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12224 (rs6000_option_override_internal): Remove dead code.
12225 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12226 (rs6000_legitimize_reload_address): Likewise.
12227 (rs6000_legitimate_address_p): Likewise.
12228 (rs6000_opt_masks): Delete "vsx-timode".
12229 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12230 from function comment.
12231 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12232 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12233 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12235 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12236 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12237 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12238 (VSX_TI): Likewise.
12240 (define_peephole2): Likewise.
12242 2017-08-17 Martin Sebor <msebor@redhat.com>
12245 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12246 past the end of an array.
12247 (test_pp_format): Add test cases.
12249 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
12251 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
12252 missing ECF_NOTHROW flags.
12254 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12257 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
12258 operands residing in integer registers.
12259 (*vsx_le_perm_load_<mode>): Likewise.
12260 (*vsx_le_perm_store_<mode>): Likewise.
12261 (define_peephole2): Add peepholes to optimize the above.
12263 2017-08-17 Marek Polacek <polacek@redhat.com>
12265 PR middle-end/81814
12266 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
12267 to mimic what shorten_compare did. Change the return type to bool.
12268 (fold_cond_expr_with_comparison): Update call to
12269 operand_equal_for_comparison_p.
12270 (fold_ternary_loc): Likewise.
12272 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
12274 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
12276 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
12278 2017-08-17 Richard Biener <rguenther@suse.de>
12280 * tree-ssa-structalias.c (solve_graph): When propagating
12281 to successors update the graphs succ edges and avoid duplicate work.
12283 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
12286 * config/i386/i386.c (ix86_option_override_internal): Save target
12287 specific options after ix86_stack_protector_guard_reg was changed.
12289 2017-08-17 Richard Biener <rguenther@suse.de>
12291 PR tree-optimization/81827
12292 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
12293 (new_var_info): Initialize it conservatively.
12294 (get_call_vi): Mark register vars.
12295 (new_scalar_tmp_constraint_exp): Likewise.
12296 (handle_rhs_call): Likewise.
12297 (handle_const_call): Likewise.
12298 (create_function_info_for): Likewise.
12299 (solve_constraints): Sort varinfos to separate register from
12300 non-register vars to pack points-to solution bitmaps during
12303 2017-08-17 Marek Polacek <polacek@redhat.com>
12305 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
12307 2017-08-17 Richard Biener <rguenther@suse.de>
12309 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
12310 to true when overflow is undefined and we saturated the result.
12312 2017-08-17 Alan Modra <amodra@gmail.com>
12315 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
12316 Don't use store multiple if only one reg needs saving.
12317 (interesting_frame_related_regno): New function.
12318 (rs6000_frame_related): Don't emit frame info for regs that
12320 (rs6000_emit_epilogue): Likewise.
12322 2017-08-16 Nathan Sidwell <nathan@acm.org>
12324 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
12325 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
12326 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
12327 * tree.c (free_lang_data_in_type): Use else-if chain. Always
12328 clear TYPE_LANG_1. Remove obsolete member-function stripping.
12329 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
12330 (verify_type): Adjust for TYPE_BINFO move.
12331 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
12332 process TYPE_BINFO directly.
12333 (hash_tree): Likewise.
12334 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12336 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12339 2017-08-16 David Malcolm <dmalcolm@redhat.com>
12341 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
12343 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
12346 * config/i386/i386.md (*anddi_1_btr): Change predicates of
12347 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
12348 Add ix86_binary_operator_ok to insn constraint.
12349 (*iordi_1_bts): Ditto.
12350 (*xordi_1_btc): Ditto.
12351 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
12352 Update corresponding peephole2 pattern.
12356 2017-08-16 Bin Cheng <bin.cheng@arm.com>
12358 PR tree-optimization/81832
12359 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
12360 copy loop header which has IFN_LOOP_DIST_ALIAS call.
12362 2017-08-16 Marek Polacek <polacek@redhat.com>
12365 * fold-const.c (fold_indirect_ref_1): Restore original behavior
12366 regarding size_zero_node.
12368 2017-08-16 Martin Liska <mliska@suse.cz>
12371 * config.gcc: Respect previously set extra_objs in case
12374 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
12376 PR tree-optimization/81835
12377 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
12378 the phi SSA_NAME. Check that the condition in a COND_EXPR does
12379 not depend on the phi.
12381 2017-08-16 Alan Modra <amodra@gmail.com>
12383 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
12386 2017-08-16 Alan Modra <amodra@gmail.com>
12388 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
12389 (save_reg_p): ..into this. Update all callers.
12390 (first_reg_to_save): Simplify.
12392 2017-08-16 Alan Modra <amodra@gmail.com>
12394 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
12397 2017-08-15 Joseph Myers <joseph@codesourcery.com>
12401 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
12402 constant count if that count is less than 32.
12404 2017-08-15 Nathan Sidwell <nathan@acm.org>
12406 * gcc.c (execute): Emit friendlier message if inferior is killed
12407 by an external cause.
12409 2017-08-15 Richard Biener <rguenther@suse.de>
12411 PR tree-optimization/81790
12412 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
12413 CONSTRUCTORs from simplifying and VN.
12415 2017-08-14 Martin Sebor <msebor@redhat.com>
12417 * builtin-attrs.def: Add comments.
12419 2017-08-14 Martin Sebor <msebor@redhat.com>
12422 * doc/extend.texi (attribute nonstring): Document new attribute.
12424 2017-08-14 Martin Sebor <msebor@redhat.com>
12427 * tree-diagnostic.c (default_tree_printer): Handle %G.
12428 * gimple-pretty-print.h (percent_G_format): Declare new function.
12429 * gimple-pretty-print.c (percent_G_format): Define.
12430 * tree-pretty-print.c (percent_K_format): Add argument.
12432 2017-08-14 Martin Sebor <msebor@redhat.com>
12434 PR translation/79998
12435 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
12436 Remove a stray space.
12438 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
12441 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
12442 (*iordi_1_bts): Ditto.
12443 (*xordi_1_btc): Ditto.
12445 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12448 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
12449 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12451 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
12452 quoted strings, and make more translator-friendly.
12453 (darwin_rs6000_override_options): Likewise.
12454 (rs6000_option_override_internal): Likewise.
12455 (rs6000_return_in_memory): Fix overlong line.
12456 (init_cmulative_args): Use quoted strings, and make more
12457 translator-friendly.
12458 (rs6000_pass_by_reference): Fix overlong line.
12459 (def_builtin): Use quoted strings.
12460 (altivec_expand_predicate_builtin): Use quoted strings, and make
12461 more translator-friendly.
12462 (htm_expand_builtin): Use quoted strings.
12463 (cpu_expand_builtin): Use quoted strings, and make more
12464 translator-friendly.
12465 (altivec_expand_builtin): Likewise.
12466 (paired_expand_predicate_builtin): Likewise.
12467 (rs6000_invalid_builtin): Likewise.
12468 (builtin_function_type): Use quoted strings.
12469 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
12470 more translator-friendly.
12471 (rs6000_trampoline_init): Likewise.
12472 (rs6000_handle_altivec_attribute): Likewise.
12473 (rs6000_inner_target_options): Use quoted strings.
12474 (rs6000_disable_incompatible_switches): Likewise.
12475 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
12476 strings, and make more translator-friendly.
12477 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
12479 2017-08-14 Bin Cheng <bin.cheng@arm.com>
12481 PR tree-optimization/81799
12482 * tree-loop-distribution.c (version_loop_by_alias_check): Force
12483 cond_expr to simple gimple operand.
12485 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
12487 PR middle-end/46932
12488 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
12490 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
12494 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
12495 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
12496 TARGET_GASISR_PROLOGUES.
12497 * config/avr/avr.c (avr_option_override): Same.
12498 (avr_pass_pre_proep::execute): Same.
12500 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
12503 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
12504 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
12506 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
12508 * config/i386/i386.md (*load_tp_<mode>): Redefine as
12509 define_insn_and_split. Split to a memory load from 0 in
12510 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
12511 using PTR mode iterator.
12512 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
12513 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
12514 (*add_tp_<mode>): Redefine as define_insn_and_split.
12515 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
12516 address space. Merge with *add_tp_x32 using PTR mode iterator.
12517 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
12518 Split to an add with a memory load from 0 in
12519 DEFAULT_TLS_SEG_REG address space.
12521 2017-08-12 Andrew Pinski <apinski@cavium.com>
12523 * config/aarch64/aarch64-option-extensions.def (rdma):
12524 Fix feature string to what Linux prints out in /proc/cpuinfo.
12526 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
12529 * dwarf2out.c (modified_type_die): For C typedef types that have
12530 an ultimate origin, process the ultimate origin instead of the
12532 (gen_typedef_die): Assert that input DECLs have no ultimate
12534 (gen_type_die_with_usage): For typedef variants that have an
12535 ultimate origin, just call gen_decl_die on the original DECL.
12536 (process_scope_var): Avoid creating DIEs for local typedefs and
12537 concrete static variables.
12539 2017-08-12 Alan Modra <amodra@gmail.com>
12543 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
12544 match gnu-user.h startfile.
12545 (ENDFILE_LINUX_SPEC): Similarly.
12547 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
12550 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
12552 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
12554 2017-08-11 Tamar Christina <tamar.christina@arm.com>
12555 * config/aarch64/aarch64.md (mov<mode>): Change.
12556 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12557 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
12558 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
12560 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
12562 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
12563 for storage order barriers.
12565 2017-08-11 Martin Liska <mliska@suse.cz>
12568 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
12569 variables of void type.
12571 2017-08-11 Martin Liska <mliska@suse.cz>
12573 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
12574 TARGET_SUPPORTS_ALIASES.
12575 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
12576 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
12577 (optimize_weakref): Likewise.
12578 * symtab.c (symtab_node::noninterposable_alias): Likewise.
12579 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
12580 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
12582 2017-08-11 Martin Liska <mliska@suse.cz>
12585 * config/i386/i386.c (make_resolver_func): Do complete
12586 refactoring of the function.
12588 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12591 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
12592 * config/i386/i386.c (ix86_stack_protect_guard): Use
12593 ix86_stack_protect_guard_symbol_str to generate varible declaration.
12594 * doc/invoke.texi (x86 Options): Document
12595 -mstack-protector-guard-symbol= option.
12597 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12599 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
12600 * config/i386/i386.c (ix86_split_stack_guard): New function.
12601 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
12602 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
12603 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
12604 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
12605 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
12606 (split_stack_space_check): Call ix86_split_stack_guard.
12608 2017-08-10 Martin Sebor <msebor@redhat.com>
12610 * print-tree.c (print_node): Print location using the established
12612 Replace spaces with colons.
12613 (debug_raw, debug): Ditto.
12615 2017-08-10 Martin Sebor <msebor@redhat.com>
12618 * pretty-print.c (pp_format): Correct the handling of %s precision.
12620 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
12623 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
12625 (ix86_finalize_stack_frame_flags): This. Also clear
12626 frame_pointer_needed if -fno-omit-frame-pointer is used without
12628 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
12629 with ix86_finalize_stack_frame_flags.
12630 (ix86_expand_epilogue): Likewise.
12631 (ix86_expand_split_stack_prologue): Likewise.
12632 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
12634 2017-08-10 Martin Liska <mliska@suse.cz>
12637 * c-attribs.c (handle_target_attribute):
12638 Report warning for an empty string argument of target attribute.
12640 2017-08-09 Jakub Jelinek <jakub@redhat.com>
12643 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
12645 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
12646 or DECL_NONLOCAL labels.
12647 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
12648 or DECL_NONLOCAL labels here.
12650 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
12652 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
12653 to indicate when early gimple folding has been disabled.
12654 (rs6000_gimple_fold_builtin): Add debug content.
12655 (rs6000_invalid_builtin): Fix whitespace.
12656 (rs6000_expand_builtin): Fix whitespace.
12657 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
12659 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
12662 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
12663 SAVE_MULTIPLE if not all the registers that saves, should be saved.
12665 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
12667 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
12668 (qdf24xx): Likewise.
12669 * config/aarch64/aarch64.md: Include falkor.md.
12670 * config/aarch64/falkor.md: New.
12672 2017-08-09 Marek Polacek <polacek@redhat.com>
12675 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
12676 * diagnostic.c (emit_diagnostic): Add a comment.
12677 (emit_diagnostic_valist): New function.
12679 2017-08-09 Marek Polacek <polacek@redhat.com>
12682 * input.c (make_location): New overload.
12683 * input.h (make_location): Declare.
12685 2017-08-08 Alan Modra <amodra@gmail.com>
12686 H.J. Lu <hongjiu.lu@intel.com>
12689 * gcc.c (NO_PIE_SPEC): Delete.
12690 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
12692 (LINK_PIE_SPEC): ..to here.
12693 (LINK_COMMAND_SPEC): Support -no-pie.
12694 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
12695 chain of crtbegin*.o selection, update for PIE_SPEC changes and
12697 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
12698 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
12699 (ENDFILE_CRTEND_SPEC): Similarly.
12701 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
12704 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
12705 (mstack-protector-guard-offset=): Ditto.
12706 * config/i386/i386.c (ix86_option_override): Handle
12707 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
12709 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
12710 ix86_stack_protect_guard_offset variables.
12711 (TARGET_STACK_PROTECT_GUARD): Always define.
12712 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
12713 and -mstack-protector-guard-offset= options.
12715 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12717 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
12718 boundary case for the last candidate.
12720 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12722 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
12723 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
12725 2017-08-08 Tamar Christina <tamar.christina@arm.com>
12727 PR middle-end/19706
12728 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
12729 * config/aarch64/aarch64-builtins.c
12730 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
12731 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
12732 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
12734 2017-08-08 Tamar Christina <tamar.christina@arm.com>
12735 Andrew Pinski <pinskia@gmail.com>
12737 PR middle-end/19706
12738 * internal-fn.def (XORSIGN): New.
12739 * optabs.def (xorsign_optab): New.
12740 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
12741 (convert_expand_mult_copysign): New.
12742 (pass_optimize_widening_mul::execute): Call
12743 convert_expand_mult_copysign.
12745 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12747 PR tree-optimization/81354
12748 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
12749 Insert on edges rather than explicitly creating landing pads.
12750 (analyze_candidates_and_replace): Commit edge inserts.
12752 2017-08-08 Richard Biener <rguenther@suse.de>
12754 PR middle-end/81719
12755 * tree-ssa-loop-niter.c: Include tree-dfa.h.
12756 (expand_simple_operations): Also look through ADDR_EXPRs with
12757 MEM_REF bases treating them as POINTER_PLUS_EXPR.
12759 2017-08-08 Richard Biener <rguenther@suse.de>
12761 PR tree-optimization/81723
12762 * tree-vect-slp.c (struct bst_traits): New hash traits.
12763 (bst_fail): New global.
12764 (vect_build_slp_tree_2): New worker, split out from ...
12765 (vect_build_slp_tree): ... this now wrapping it with using
12766 bst_fail set to cache SLP tree build fails. Properly handle
12768 (vect_analyze_slp_instance): Allocate and free bst_fail.
12770 2017-08-08 Martin Liska <mliska@suse.cz>
12773 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
12774 LABEL_DECLs that can be from a different function.
12776 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12778 PR tree-optimization/81744
12779 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
12780 loop's number of iterations.
12782 2017-08-08 Martin Liska <mliska@suse.cz>
12784 * asan.c: Include header files.
12785 * attribs.c (build_decl_attribute_variant): New function moved
12787 (build_type_attribute_qual_variant): Likewise.
12788 (cmp_attrib_identifiers): Likewise.
12789 (simple_cst_list_equal): Likewise.
12790 (omp_declare_simd_clauses_equal): Likewise.
12791 (attribute_value_equal): Likewise.
12792 (comp_type_attributes): Likewise.
12793 (build_type_attribute_variant): Likewise.
12794 (lookup_ident_attribute): Likewise.
12795 (remove_attribute): Likewise.
12796 (merge_attributes): Likewise.
12797 (merge_type_attributes): Likewise.
12798 (merge_decl_attributes): Likewise.
12799 (merge_dllimport_decl_attributes): Likewise.
12800 (handle_dll_attribute): Likewise.
12801 (attribute_list_equal): Likewise.
12802 (attribute_list_contained): Likewise.
12803 * attribs.h (lookup_attribute): New function moved from tree.[ch].
12804 (lookup_attribute_by_prefix): Likewise.
12805 * bb-reorder.c: Include header files.
12806 * builtins.c: Likewise.
12807 * calls.c: Likewise.
12808 * cfgexpand.c: Likewise.
12809 * cgraph.c: Likewise.
12810 * cgraphunit.c: Likewise.
12811 * convert.c: Likewise.
12812 * dwarf2out.c: Likewise.
12813 * final.c: Likewise.
12814 * fold-const.c: Likewise.
12815 * function.c: Likewise.
12816 * gimple-expr.c: Likewise.
12817 * gimple-fold.c: Likewise.
12818 * gimple-pretty-print.c: Likewise.
12819 * gimple.c: Likewise.
12820 * gimplify.c: Likewise.
12821 * hsa-common.c: Likewise.
12822 * hsa-gen.c: Likewise.
12823 * internal-fn.c: Likewise.
12824 * ipa-chkp.c: Likewise.
12825 * ipa-cp.c: Likewise.
12826 * ipa-devirt.c: Likewise.
12827 * ipa-fnsummary.c: Likewise.
12828 * ipa-inline.c: Likewise.
12829 * ipa-visibility.c: Likewise.
12831 * lto-cgraph.c: Likewise.
12832 * omp-expand.c: Likewise.
12833 * omp-general.c: Likewise.
12834 * omp-low.c: Likewise.
12835 * omp-offload.c: Likewise.
12836 * omp-simd-clone.c: Likewise.
12837 * opts-global.c: Likewise.
12838 * passes.c: Likewise.
12839 * predict.c: Likewise.
12840 * sancov.c: Likewise.
12841 * sanopt.c: Likewise.
12842 * symtab.c: Likewise.
12843 * toplev.c: Likewise.
12844 * trans-mem.c: Likewise.
12845 * tree-chkp.c: Likewise.
12846 * tree-eh.c: Likewise.
12847 * tree-into-ssa.c: Likewise.
12848 * tree-object-size.c: Likewise.
12849 * tree-parloops.c: Likewise.
12850 * tree-profile.c: Likewise.
12851 * tree-ssa-ccp.c: Likewise.
12852 * tree-ssa-live.c: Likewise.
12853 * tree-ssa-loop.c: Likewise.
12854 * tree-ssa-sccvn.c: Likewise.
12855 * tree-ssa-structalias.c: Likewise.
12856 * tree-ssa.c: Likewise.
12857 * tree-streamer-in.c: Likewise.
12858 * tree-vectorizer.c: Likewise.
12859 * tree-vrp.c: Likewise.
12860 * tsan.c: Likewise.
12861 * ubsan.c: Likewise.
12862 * varasm.c: Likewise.
12863 * varpool.c: Likewise.
12864 * tree.c: Remove functions moved to attribs.[ch].
12865 * tree.h: Likewise.
12866 * config/aarch64/aarch64.c: Add attrs.h header file.
12867 * config/alpha/alpha.c: Likewise.
12868 * config/arc/arc.c: Likewise.
12869 * config/arm/arm.c: Likewise.
12870 * config/avr/avr.c: Likewise.
12871 * config/bfin/bfin.c: Likewise.
12872 * config/c6x/c6x.c: Likewise.
12873 * config/cr16/cr16.c: Likewise.
12874 * config/cris/cris.c: Likewise.
12875 * config/darwin.c: Likewise.
12876 * config/epiphany/epiphany.c: Likewise.
12877 * config/fr30/fr30.c: Likewise.
12878 * config/frv/frv.c: Likewise.
12879 * config/ft32/ft32.c: Likewise.
12880 * config/h8300/h8300.c: Likewise.
12881 * config/i386/winnt.c: Likewise.
12882 * config/ia64/ia64.c: Likewise.
12883 * config/iq2000/iq2000.c: Likewise.
12884 * config/lm32/lm32.c: Likewise.
12885 * config/m32c/m32c.c: Likewise.
12886 * config/m32r/m32r.c: Likewise.
12887 * config/m68k/m68k.c: Likewise.
12888 * config/mcore/mcore.c: Likewise.
12889 * config/microblaze/microblaze.c: Likewise.
12890 * config/mips/mips.c: Likewise.
12891 * config/mmix/mmix.c: Likewise.
12892 * config/mn10300/mn10300.c: Likewise.
12893 * config/moxie/moxie.c: Likewise.
12894 * config/msp430/msp430.c: Likewise.
12895 * config/nds32/nds32-isr.c: Likewise.
12896 * config/nds32/nds32.c: Likewise.
12897 * config/nios2/nios2.c: Likewise.
12898 * config/nvptx/nvptx.c: Likewise.
12899 * config/pa/pa.c: Likewise.
12900 * config/pdp11/pdp11.c: Likewise.
12901 * config/powerpcspe/powerpcspe.c: Likewise.
12902 * config/riscv/riscv.c: Likewise.
12903 * config/rl78/rl78.c: Likewise.
12904 * config/rx/rx.c: Likewise.
12905 * config/s390/s390.c: Likewise.
12906 * config/sh/sh.c: Likewise.
12907 * config/sol2.c: Likewise.
12908 * config/sparc/sparc.c: Likewise.
12909 * config/spu/spu.c: Likewise.
12910 * config/stormy16/stormy16.c: Likewise.
12911 * config/tilegx/tilegx.c: Likewise.
12912 * config/tilepro/tilepro.c: Likewise.
12913 * config/v850/v850.c: Likewise.
12914 * config/vax/vax.c: Likewise.
12915 * config/visium/visium.c: Likewise.
12916 * config/xtensa/xtensa.c: Likewise.
12918 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12921 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
12922 constraints since the -mupper-regs-* switches have been
12924 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
12925 into a vector from a double word element that was extracted from
12926 another vector, and eliminate extra XXPERMDI instructions.
12927 (vsx_concat_<mode>_2): Likewise.
12928 (vsx_concat_<mode>_3): Likewise.
12929 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
12930 concat to allow optimizing inserts from previous extracts.
12932 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
12934 * config/i386/i386.c (ix86_stack_protect_guard): Generate
12935 memory reference to a SSP offset in TLS address space.
12936 (ix86_print_operand) <case '@'>: Remove.
12937 (ix86_print_operand_punct_valid_p): Remove '@' code.
12938 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
12939 UNSPEC_SP_TLS_TEST.
12940 (stack_tls_protect_set_<mode>): Remove.
12941 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
12942 (stack_tls_protect_test_<mode>): Remove.
12943 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
12945 2017-08-07 Olivier Hainque <hainque@adacore.com>
12948 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
12950 2017-08-07 Douglas Rupp <rupp@adacore.com>
12952 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
12953 variable was referenced as multidir in command.
12955 2017-08-07 Jakub Jelinek <jakub@redhat.com>
12958 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
12961 2017-08-07 Martin Liska <mliska@suse.cz>
12963 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
12964 * config/rl78/rl78.c: Add include of attribs.h.
12965 * config/sh/sh.c: Likewise.
12966 * config/v850/v850.c: Likewise.
12968 2017-08-07 Tom de Vries <tom@codesourcery.com>
12970 PR middle-end/78266
12971 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
12973 2017-08-07 Martin Liska <mliska@suse.cz>
12975 * config/mips/mips.c: Include attribs.h.
12977 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
12980 * doc/invoke.texi: Document change in behvaior for -Ofast for
12983 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
12985 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
12987 (aarch64_pop_regs): Likewise.
12988 (aarch64_gen_load_pair): Likewise.
12989 (aarch64_save_callee_saves): Likewise.
12990 (aarch64_restore_callee_saves): Likewise.
12992 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12994 * config/i386/i386.c: Revert the last change.
12996 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12999 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13001 (ix86_finalize_stack_frame_flags): This. Also clear
13002 frame_pointer_needed if -fno-omit-frame-pointer is used without
13004 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13005 with ix86_finalize_stack_frame_flags.
13006 (ix86_expand_epilogue): Likewise.
13007 (ix86_expand_split_stack_prologue): Likewise.
13009 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13012 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13013 to P_AES for Westmere.
13015 2017-08-07 Jonathan Yong <10walls@gmail.com>
13017 * config/i386/mingw.opt (fset-stack-executable): Removed.
13018 * config/i386/cygming.opt (fset-stack-executable): Moved
13020 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13022 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13024 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13026 2017-08-07 Marek Polacek <polacek@redhat.com>
13028 PR middle-end/81737
13029 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13031 2017-08-07 Martin Liska <mliska@suse.cz>
13033 * attribs.h (canonicalize_attr_name): New function.
13034 (cmp_attribs): Move from c-format.c and adjusted.
13035 (is_attribute_p): Moved from tree.h.
13036 * tree-inline.c: Add new includes.
13037 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13038 (private_is_attribute_p): Remove.
13039 (private_lookup_attribute): Likewise.
13040 (private_lookup_attribute_by_prefix): Simplify.
13041 (remove_attribute): Use is_attribute_p.
13042 * tree.h: Remove removed declarations.
13044 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13046 PR middle-end/81698
13047 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13048 instead of computing it in the function. Formatting fix.
13049 (expand_case): Don't rely on default_edge being the first edge,
13050 clear it if removing it, pass default_edge to
13051 emit_case_dispatch_table.
13052 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13055 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13057 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13058 insn in the function, emit NOP after the insn.
13060 2017-08-06 Tom de Vries <tom@codesourcery.com>
13062 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13065 2017-08-06 Tom de Vries <tom@codesourcery.com>
13067 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13070 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13072 PR tree-optimization/57371
13073 * match.pd: New pattern.
13075 2017-08-04 Marek Polacek <polacek@redhat.com>
13077 PR middle-end/81695
13078 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13079 perform the computation in offset_int.
13081 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13083 PR tree-optimization/81136
13084 * tree-vectorizer.h: Include tree-hash-traits.h.
13085 (vec_base_alignments): New typedef.
13086 (vec_info): Add a base_alignments field.
13087 (vect_record_base_alignments): Declare.
13088 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13090 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13091 (create_data_ref): Add an is_conditional_in_stmt argument.
13092 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13093 the is_conditional_in_stmt field.
13094 (data_ref_loc): Add an is_conditional_in_stmt field.
13095 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13096 (find_data_references_in_stmt): Update call to create_data_ref.
13097 (graphite_find_data_references_in_stmt): Likewise.
13098 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13099 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13100 (vect_record_base_alignment): New function.
13101 (vect_record_base_alignments): Likewise.
13102 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13103 for nested statements even if we fail to compute a misalignment.
13104 Use pooled base alignments for unconditional references.
13105 (vect_find_same_alignment_drs): Compare base addresses instead
13107 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13108 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13110 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13112 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13113 Add an explicit name for the enum. Use auto_vec for slp_instances
13114 and grouped_stores.
13115 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13117 (_bb_vec_info): Add a constructor and destructor.
13118 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13119 (destroy_loop_vec_info): Delete.
13120 (vect_destroy_datarefs): Likewise.
13121 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13122 (vec_info::vec_info): New function.
13123 (vec_info::~vec_info): Likewise.
13124 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13125 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13126 destroy_loop_vec_info.
13127 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13128 (_loop_vec_info::_loop_vec_info): ...this.
13129 (destroy_loop_vec_info): Replace with...
13130 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13131 the stmt_vec_infos. Leave handling of vec_info information to its
13132 destructor. Remove explicit vector releases.
13133 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13134 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13135 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13136 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13137 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13138 (destroy_bb_vec_info): Replace with...
13139 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13140 information to its destructor.
13141 (vect_slp_analyze_bb_1): Use new and delete instead of
13142 new_bb_vec_info and destroy_bb_vec_info.
13143 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13146 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13148 * tree-data-ref.h (subscript): Add access_fn field.
13149 (data_dependence_relation): Add could_be_independent_p.
13150 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13151 (same_access_functions): Move to tree-data-ref.c.
13152 * tree-data-ref.c (ref_contains_union_access_p): New function.
13153 (access_fn_component_p): Likewise.
13154 (access_fn_components_comparable_p): Likewise.
13155 (dr_analyze_indices): Add a reference to access_fn_component_p.
13156 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13158 (constant_access_functions): Likewise.
13159 (add_other_self_distances): Likewise.
13160 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13161 (initialize_data_dependence_relation): Use XCNEW and remove
13162 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13163 of access functions that have the same type. Allow the
13164 subsequence to end with different bases in some circumstances.
13165 Record the chosen access functions in SUB_ACCESS_FN.
13166 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13167 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13168 (subscript_dependence_tester_1): Likewise dra and drb.
13169 (build_classic_dist_vector): Update calls accordingly.
13170 (subscript_dependence_tester): Likewise.
13171 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13172 DDR_COULD_BE_INDEPENDENT_P.
13173 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13174 comp_alias_ddrs instead of may_alias_ddrs.
13175 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13177 (vect_analyze_data_ref_dependence): Use it if
13178 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13179 distance vectors if that fails.
13180 (dependence_distance_ge_vf): New function.
13181 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13182 LOOP_VINFO_MAY_ALIAS_DDRS.
13184 2017-08-04 Richard Biener <rguenther@suse.de>
13186 PR middle-end/81705
13187 * fold-const.c (fold_binary_loc): Properly restrict
13188 minus_var0 && minus_var1 case when associating undefined overflow
13191 2017-08-04 Tom de Vries <tom@codesourcery.com>
13193 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13195 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13197 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13198 Don't start diagnostic messages with a capital letter.
13199 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13201 (rs6000_invalid_builtin): Likewise.
13202 (rs6000_trampoline_init): Likewise.
13204 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13207 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13208 after setting changeable df flags.
13210 2017-08-03 Richard Biener <rguenther@suse.de>
13212 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13213 up if the use is in USE - X.
13215 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13217 * toplev.c (dumpfile.h): New include.
13218 (internal_error_reentered): New static function. Use it...
13219 (internal_error_function): ...here to handle reentered internal_error.
13221 2017-08-03 Richard Biener <rguenther@suse.de>
13223 PR middle-end/81148
13224 * fold-const.c (split_tree): Add minus_var and minus_con
13225 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13226 here but always use minus_*.
13227 (associate_trees): Assert we never associate with MINUS_EXPR
13228 and NULL first operand. Do not recurse for PLUS_EXPR operands
13229 when associating as MINUS_EXPR either.
13230 (fold_binary_loc): Track minus_var and minus_con.
13232 2017-08-03 Tom de Vries <tom@codesourcery.com>
13235 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13236 ACCEL_COMPILER, apply finish_options on
13237 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13239 2017-08-03 Tom de Vries <tom@codesourcery.com>
13242 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13243 function_entry_patch_area_size > 0.
13245 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13248 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13249 instead of 10??LU, perform unit multiplication in wide_int,
13250 don't change alloc_object_size_limit if the limit is larger
13253 PR tree-optimization/81655
13254 PR tree-optimization/81588
13255 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
13256 the case when ranges[i].low and high are 1 for unsigned type with
13259 PR middle-end/81052
13260 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
13261 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
13263 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13265 * tree-vrp.h: Add include guard.
13267 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
13270 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
13271 (ud2): New insn pattern.
13272 * config/i386/i386.c (ix86_expand_epilogue):
13273 For naked functions, generate ud2 instead of trap insn.
13275 2017-08-02 Marek Polacek <polacek@redhat.com>
13278 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
13280 2017-08-02 Tom de Vries <tom@codesourcery.com>
13281 Cesar Philippidis <cesar@codesourcery.com>
13283 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
13284 Add missing edge probabilities.
13286 2017-08-02 Tamar Christina <tamar.christina@arm.com>
13288 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
13289 Correct endianness.
13291 2017-08-02 Jakub Jelinek <jakub@redhat.com>
13293 PR middle-end/79499
13294 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
13295 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
13296 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
13297 prologue_seq sequences - if any.
13299 2017-08-02 Richard Biener <rguenther@suse.de>
13301 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
13302 via vectors if supported, integer extracts via punning if supported
13303 or otherwise vector extracts.
13305 2017-08-02 Richard Biener <rguenther@suse.de>
13307 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
13309 (bitmap_insert_into_set): ... this.
13311 2017-08-02 Richard Biener <rguenther@suse.de>
13313 PR tree-optimization/81633
13315 2015-08-17 Alan Hayward <alan.hayward@arm.com>
13317 PR tree-optimization/71752
13318 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
13320 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
13322 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
13323 (machine_function::call_ms2sysv_pad_out): Remove field.
13324 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
13325 (ix86_compute_frame_layout): Likewise.
13327 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
13330 * config/i386/i386.c (ix86_set_func_type): Disallow naked
13331 attribute with interrupt attribute.
13333 2017-08-01 Andrew Pinski <apinski@cavium.com>
13335 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
13336 BIT_INSERT_EXPR's operand 1
13337 to see if the types precision matches.
13339 2017-08-01 Martin Liska <mliska@suse.cz>
13341 PR middle-end/70140
13342 * builtins.c (expand_builtin_memcpy_args): Remove.
13343 (expand_builtin_memcpy): Call newly added function
13344 expand_builtin_memory_copy_args.
13345 (expand_builtin_memcpy_with_bounds): Likewise.
13346 (expand_builtin_mempcpy): Remove last argument.
13347 (expand_builtin_mempcpy_with_bounds): Likewise.
13348 (expand_builtin_memory_copy_args): New function created from
13349 expand_builtin_mempcpy_args with small modifications.
13350 (expand_builtin_mempcpy_args): Remove.
13351 (expand_builtin_stpcpy): Remove unused argument.
13352 (expand_builtin): Likewise.
13353 (expand_builtin_with_bounds): Likewise.
13355 2017-08-01 Martin Liska <mliska@suse.cz>
13358 Make mempcpy more optimal (PR middle-end/70140).
13360 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13363 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13364 __builtin_vec_cmpne verify both arguments are compatible vectors
13365 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
13366 verify arg1_type is a pointer or array type. For __builtin_vec_st,
13367 move computation of aligned to after checking the argument types.
13371 * config/rs6000/vsx.md (vextract_fp_from_shorth,
13372 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
13375 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
13376 Doug Rupp <rupp@adacore.com>
13377 Olivier Hainque <hainque@adacore.com>
13379 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
13380 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
13382 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
13383 for TARGET_OS_CPP_BUILTIN.
13384 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
13385 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
13387 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
13388 passing required abi options to the assembler for EABI configurations.
13389 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
13390 of .text.hot and .text.unlikely sections for kernel modules when
13391 using ARM style exceptions.
13392 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
13393 options. Add EXTRA_CC1_SPEC.
13394 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
13396 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
13398 (ARM_TARGET2_DWARF_FORMAT): Define.
13399 * config/arm/t-vxworks: Adjust multilib control to removal of the
13400 Diab command line options.
13402 2017-08-01 Martin Liska <mliska@suse.cz>
13404 PR gcov-profile/81561
13405 * gcov.c (unblock): Make unblocking safe as we need to preserve
13406 index correspondence of blocks and block_lists.
13408 2017-08-01 Richard Biener <rguenther@suse.de>
13410 PR tree-optimization/81181
13411 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
13412 (compute_antic): ... end of iteration here.
13414 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
13416 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
13417 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
13418 (ftree-slp-vectorize): Likewise.
13419 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
13420 can no longer be set independent of flag_tree_loop_vectorize.
13421 * omp-general.c (emp_max_vf): Likewise.
13422 * opts.c (enable_fdo_optimizations): Remove references to
13423 flag_tree_vectorize, these are now implicit.
13424 (common_handle_option): Remove handling for OPT_ftree_vectorize,
13425 and leave it for the options machinery.
13427 2017-08-01 Martin Liska <mliska@suse.cz>
13429 PR middle-end/70140
13430 * builtins.c (expand_builtin_memcpy_args): Remove.
13431 (expand_builtin_memcpy): Call newly added function
13432 expand_builtin_memory_copy_args.
13433 (expand_builtin_memcpy_with_bounds): Likewise.
13434 (expand_builtin_mempcpy): Remove last argument.
13435 (expand_builtin_mempcpy_with_bounds): Likewise.
13436 (expand_builtin_memory_copy_args): New function created from
13437 expand_builtin_mempcpy_args with small modifications.
13438 (expand_builtin_mempcpy_args): Remove.
13439 (expand_builtin_stpcpy): Remove unused argument.
13440 (expand_builtin): Likewise.
13441 (expand_builtin_with_bounds): Likewise.
13443 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13446 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
13447 print "ds:" only for immediates in generic address space.
13449 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13452 * config/i386/i386.c (ix86_funciton_naked): New prototype.
13453 (ix86_function_ok_for_sibcall): Return false for naked functions.
13455 2017-08-01 Richard Biener <rguenther@suse.de>
13457 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
13458 (compute_antic): Seed worklist with exit block predecessors.
13459 * cfganal.c (dfs_find_deadend): For a cycle return the source
13460 of the edge closing it.
13462 2017-08-01 Tamar Christina <tamar.christina@arm.com>
13464 * config/aarch64/aarch64.c
13465 (aarch64_can_const_movi_rtx_p): Move 0 check.
13467 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13469 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
13470 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
13472 * match.pd: Ditto in address comparison pattern.
13474 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13476 PR tree-optimization/81627
13477 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
13478 closed ssa form for store-store chain.
13480 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13482 PR tree-optimization/81620
13483 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
13484 for store-store chain.
13486 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13488 PR tree-optimization/81588
13489 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
13490 ranges[i].in_p, invert comparison code ccode. For >/>=,
13491 swap rhs1 and rhs2 and comparison code unconditionally,
13492 for </<= don't do that. Don't swap rhs1/rhs2 again if
13493 ranges[i].in_p, instead invert comparison code ccode if
13494 opcode or oe->rank is BIT_IOR_EXPR.
13497 * optabs.def (vec_extract_optab, vec_init_optab): Change from
13498 a direct optab to conversion optab.
13499 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
13500 with GET_MODE_INNER as last argument instead of optab_handler.
13501 * expmed.c (extract_bit_field_1): Likewise. Use vector from
13502 vector extraction if possible and optab is available.
13503 * expr.c (store_constructor): Use convert_optab_handler instead
13504 of optab_handler. Use vector initialization from smaller
13505 vectors if possible and optab is available.
13506 * tree-vect-stmts.c (vectorizable_load): Likewise.
13507 * doc/md.texi (vec_extract, vec_init): Document that the optabs
13508 now have two modes.
13509 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
13510 of vec_init from half-sized vectors with the same element mode.
13511 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
13512 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
13513 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
13514 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
13515 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
13516 after mode in gen_vec_extract* calls.
13517 (vec_extract<mode>): Renamed to ...
13518 (vec_extract<mode><ssescalarmodelower>): ... this.
13519 (vec_extract<mode><ssehalfvecmodelower>): New expander.
13520 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
13521 element mode after mode in gen_vec_init* calls.
13522 (VEC_INIT_HALF_MODE): New mode iterator.
13523 (vec_init<mode>): Renamed to ...
13524 (vec_init<mode><ssescalarmodelower>): ... this.
13525 (vec_init<mode><ssehalfvecmodelower>): New expander.
13526 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
13527 (vec_extractv2sfsf): ... this.
13528 (vec_initv2sf): Renamed to ...
13529 (vec_initv2sfsf): ... this.
13530 (vec_extractv2si): Renamed to ...
13531 (vec_extractv2sisi): ... this.
13532 (vec_initv2si): Renamed to ...
13533 (vec_initv2sisi): ... this.
13534 (vec_extractv4hi): Renamed to ...
13535 (vec_extractv4hihi): ... this.
13536 (vec_initv4hi): Renamed to ...
13537 (vec_initv4hihi): ... this.
13538 (vec_extractv8qi): Renamed to ...
13539 (vec_extractv8qiqi): ... this.
13540 (vec_initv8qi): Renamed to ...
13541 (vec_initv8qiqi): ... this.
13542 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
13543 (vec_init<mode>): Renamed to ...
13544 (vec_init<mode><VEC_base_l>): ... this.
13545 (vec_extract<mode>): Renamed to ...
13546 (vec_extract<mode><VEC_base_l>): ... this.
13547 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
13548 (vec_initv2sfsf): ... this.
13549 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
13550 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13551 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
13552 element mode after mode in gen_vec_init* calls.
13553 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
13554 (vec_init<mode><Vel>): ... this.
13555 (vec_extract<mode>): Renamed to ...
13556 (vec_extract<mode><Vel>): ... this.
13557 * config/aarch64/iterators.md (Vel): New mode attribute.
13558 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
13559 Add element mode after mode in gen_vec_extract* calls.
13560 * config/s390/vector.md (non_vec_l): New mode attribute.
13561 (vec_extract<mode>): Renamed to ...
13562 (vec_extract<mode><non_vec_l>): ... this.
13563 (vec_init<mode>): Renamed to ...
13564 (vec_init<mode><non_vec_l>): ... this.
13565 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
13566 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
13568 * config/arm/iterators.md (V_elem_l): New mode attribute.
13569 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
13570 (vec_extract<mode><V_elem_l>): ... this.
13571 (vec_extractv2di): Renamed to ...
13572 (vec_extractv2didi): ... this.
13573 (vec_init<mode>): Renamed to ...
13574 (vec_init<mode><V_elem_l>): ... this.
13575 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
13576 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
13577 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
13578 Add element mode after gen_vec_extract* calls.
13579 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
13580 (vec_init<mode><unitmode>): ... this.
13581 (vec_extract<mode>): Renamed to ...
13582 (vec_extract<mode><unitmode>): ... this.
13583 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
13584 (vec_init<mode><unitmode>): ... this.
13585 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
13586 (vec_initv2sfsf): ... this.
13587 (vec_extractv2sf): Renamed to ...
13588 (vec_extractv2sfsf): ... this.
13589 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
13590 Add element mode after gen_vec_extract* calls.
13591 * config/mips/mips.md (unitmode): New mode iterator.
13592 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
13593 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
13594 * config/spu/spu.md (inner_l): New mode attribute.
13595 (vec_init<mode>): Renamed to ...
13596 (vec_init<mode><inner_l>): ... this.
13597 (vec_extract<mode>): Renamed to ...
13598 (vec_extract<mode><inner_l>): ... this.
13599 * config/sparc/sparc.md (veltmode): New mode iterator.
13600 (vec_init<VMALL:mode>): Renamed to ...
13601 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
13602 * config/ia64/vect.md (vec_initv2si): Renamed to ...
13603 (vec_initv2sisi): ... this.
13604 (vec_initv2sf): Renamed to ...
13605 (vec_initv2sfsf): ... this.
13606 (vec_extractv2sf): Renamed to ...
13607 (vec_extractv2sfsf): ... this.
13608 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
13609 (vec_init<mode>): Renamed to ...
13610 (vec_init<mode><VEC_base_l>): ... this.
13611 (vec_extract<mode>): Renamed to ...
13612 (vec_extract<mode><VEC_base_l>): ... this.
13613 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
13614 (vec_initv2sfsf): ... this.
13615 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
13616 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13617 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
13618 gen_vec_init* calls.
13620 2017-08-01 Richard Biener <rguenther@suse.de>
13622 PR tree-optimization/81297
13623 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
13624 TREE_OVERFLOW from INTEGER_CSTs.
13626 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
13628 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
13630 2017-07-31 Carl Love <cel@us.ibm.com>
13632 * config/rs6000/rs6000-c: Add support for built-in functions
13633 vector signed char vec_xl_be (signed long long, signed char *);
13634 vector unsigned char vec_xl_be (signed long long, unsigned char *);
13635 vector signed int vec_xl_be (signed long long, signed int *);
13636 vector unsigned int vec_xl_be (signed long long, unsigned int *);
13637 vector signed long long vec_xl_be (signed long long, signed long long *);
13638 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
13639 vector signed short vec_xl_be (signed long long, signed short *);
13640 vector unsigned short vec_xl_be (signed long long, unsigned short *);
13641 vector double vec_xl_be (signed long long, double *);
13642 vector float vec_xl_be (signed long long, float *);
13643 * config/rs6000/altivec.h (vec_xl_be): Add #define.
13644 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
13645 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
13647 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
13648 (altivec_expand_builtin): Add switch statement to call
13649 altivec_expand_xl_be for each builtin.
13650 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
13651 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
13652 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
13653 __builtin_vsx_le_be_v16qi.
13654 * doc/extend.texi: Update the built-in documentation file for the
13655 new built-in functions.
13657 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
13660 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
13662 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
13664 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13666 * config.gcc: Add z14.
13667 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
13668 CPU model numbers for z13s and z14.
13669 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
13671 * config/s390/s390-opts.h (enum processor_type): Rename
13672 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
13673 * config/s390/s390.c (processor_table): Add field for CPU name to
13674 be passed to Binutils.
13675 (s390_asm_output_machine_for_arch): Use the new field in
13676 processor_table for Binutils.
13677 (s390_expand_builtin): Replace arch12 with z14.
13678 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
13679 (s390_get_sched_attrmask): Likewise.
13680 (s390_get_unit_mask): Likewise.
13681 * config/s390/s390.opt: Add z14 to processor_type enum.
13683 2017-07-31 Martin Jambor <mjambor@suse.cz>
13686 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
13687 regardless of optimization level.
13689 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
13690 Martin Liska <mliska@suse.cz>
13692 * predict.def: Remove old comment and adjust probability.
13693 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
13694 PREDICT statements.
13696 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
13699 * config/i386/i386.c (ix86_function_naked): New function.
13700 (ix86_can_use_return_insn_p): Return false for naked functions.
13701 (ix86_expand_prologue): Skip prologue for naked functions.
13702 (ix86_expand_epilogue): Skip epilogue for naked functions
13703 and emit trap instruction.
13704 (ix86_warn_func_return): New function.
13705 (ix86_attribute_table): Add "naked" attribute specification.
13706 (TARGET_WARN_FUNC_RETURN): Define.
13707 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
13709 2017-07-31 Martin Liska <mliska@suse.cz>
13711 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
13712 (dump_gimple_bb_header): Always dump BB info.
13713 (pp_cfg_jump): Do not append info about BB when dumping a jump.
13715 2017-07-31 Martin Liska <mliska@suse.cz>
13718 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
13719 also with current_function_decl non-null equality.
13721 2017-07-31 Jakub Jelinek <jakub@redhat.com>
13724 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
13725 change type to the element type, instead add eltype variable and
13726 use it where we are interested in the element type.
13728 PR tree-optimization/81603
13729 * ipa-polymorphic-call.c
13730 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
13731 offset arithmetic in offset_int, bail out if the resulting bit offset
13732 doesn't fit into shwi.
13734 2017-07-31 Martin Liska <mliska@suse.cz>
13736 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
13737 (gimplify_save_expr): Fix comment.
13739 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
13742 * config/i386/i386.c (ix86_function_arg): Update arguments for
13744 (ix86_compute_frame_layout): Set the initial stack offset to
13745 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
13746 INCOMING_FRAME_SP_OFFSET.
13747 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
13748 stack before exception handler returns.
13749 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
13750 the 'ERROR_CODE' for exception handler.
13752 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
13754 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
13755 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
13756 (ASM_OUTPUT_REG_POP): Ditto.
13757 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
13758 instead of asm_fprintf to output pure string.
13760 2017-07-29 Jakub Jelinek <jakub@redhat.com>
13762 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
13763 to imported_module_or_decl hook.
13764 (debug_nothing_tree_tree_tree_bool): Remove.
13765 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
13766 * debug.c (do_nothing_debug_hooks): Use
13767 debug_nothing_tree_tree_tree_bool_bool instead of
13768 debug_nothing_tree_tree_tree_bool.
13769 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13770 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
13771 * sdbout.c (sdb_debug_hooks): Likewise.
13772 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
13773 (gen_namespace_die): Add DW_AT_export_symbols attribute if
13775 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
13776 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
13777 attribute, don't add anything.
13779 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13781 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
13782 (fold_build2_stat_loc): Likewise.
13783 (fold_build3_stat_loc): Likewise.
13784 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
13785 (fold_build1_loc): Remove macro.
13786 (fold_build2_loc): Likewise.
13787 (fold_build3_loc): Likewise.
13789 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13791 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
13792 (gimple_build_debug_bind_source_stat): Likewise.
13793 * gimple.h (gimple_build_debug_bind): Remove macro.
13794 (gimple_build_debug_bind_source): Likewise.
13796 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13798 * bitmap.c (bitmap_alloc): Adjust.
13799 (bitmap_gc_alloc): Likewise.
13800 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
13802 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13804 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
13805 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
13806 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
13807 (bitmap_gc_alloc_stat): Likewise.
13808 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
13810 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13812 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
13813 * rtl.h (shallow_copy_rtx): Remove macro.
13815 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13817 * emit-rtl.c (gen_raw_REG): Adjust.
13818 * gengenrtl.c (gendef): Likewise.
13819 * rtl.c (rtx_alloc_stat): Remove _stat from name.
13820 * rtl.h (rtx_alloc): Remove macro.
13822 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13824 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
13825 (build_tree_list_stat): Likewise.
13826 * tree.h (build_tree_list): Remove macro.
13827 (build_tree_list_vec): Likewise.
13829 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13831 * tree.c (make_vector_stat): Remove _stat from name.
13832 (build_vector_stat): Likewise.
13833 * tree.h (make_vector_stat): Remove macro.
13834 (build_vector_stat): Likewise.
13836 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13838 * tree.h (build_var_debug_value): Remove prototype.
13840 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13842 * tree.c (tree_cons_stat): Remove _stat from name.
13843 * tree.h (tree_cons): Remove macro.
13845 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13847 * tree.c (build_vl_exp_stat): Remove _stat from name.
13848 * tree.h (build_vl_exp): Remove macro.
13850 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13852 * tree.c (build_decl_stat): Remove _stat from name.
13853 * tree.h (build_decl): Remove macro.
13855 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13857 * gimple.c (gimple_build_with_ops_stat): Adjust.
13858 (gimple_alloc_stat): Remove _stat from name.
13859 * gimple.h (gimple_alloc): Remove macro.
13861 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13863 * tree.c (make_tree_vec_stat): Remove _stat from name.
13864 (grow_tree_vec_stat): Likewise.
13865 * tree.h (make_tree_vec_stat): Adjust prototype.
13866 (grow_tree_vec_stat): Likewise.
13867 (make_tree_vec): Remove macro.
13868 (grow_tree_vec): Likewise.
13870 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13872 * fold-const.c (fold_build1_stat_loc): Adjust.
13873 (fold_build2_stat_loc): Likewise.
13874 (fold_build3_stat_loc): Likewise.
13875 * tree.c (build0_stat): Remove _stat from name.
13876 (build1_stat): Likewise.
13877 (build2_stat): Likewise.
13878 (build3_stat): Likewise.
13879 (build4_stat): Likewise.
13880 (build5_stat): Likewise.
13881 * tree.h (build1_loc): Remove macro, and rename _stat function
13883 (build2_loc): Likewise.
13884 (build3_loc): Likewise.
13885 (build4_loc): Likewise.
13886 (build5_loc): Likewise.
13888 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13890 * tree.c (make_int_cst_stat): Remove _stat from name.
13891 * tree.h (make_int_cst_stat): Adjust prototype.
13892 (make_int_cst): Remove macro.
13894 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13896 * tree.c (make_tre_binfo_stat): Remove _stat from name.
13897 * tree.h (make_tree_binfo_stat): Adjust prototype.
13898 (make_tree_binfo): Remove.
13900 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13902 * tree.c (copy_node_stat): Rename to copy_node.
13903 (build_distinct_type_copy): Adjust.
13904 * tree.h (copy_node_stat): Adjust prototype.
13905 (copy_node): Remove macro.
13907 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13909 * tree.c (make_node_stat): rename to make_node.
13910 (build_tree_list_stat): Adjust.
13911 (build0_stat): Likewise.
13912 (build2_stat): Likewise.
13913 (build3_stat): Likewise.
13914 (build4_stat): Likewise.
13915 (build5_stat): Likewise.
13916 (build_decl_stat): Likewise.
13917 * tree.h (make_node_stat): Adjust prototype.
13918 (make_node): remove macro.
13920 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
13922 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
13923 (PPC_FEATURE2_SCV): Likewise.
13924 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
13926 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13928 * config/aarch64/aarch64.c
13929 (aarch64_internal_mov_immediate): Add new special pattern.
13930 * config/aarch64/aarch64.md (*movdi_aarch64):
13931 Add reg/32bit const mov case.
13933 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13934 Richard Sandiford <richard.sandiford@linaro.org>
13936 * config/aarch64/aarch64.md (mov<mode>): Generalize.
13937 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13938 Add integer and movi cases.
13939 (movi-split-hf-df-sf split, fp16): New.
13940 (enabled): Added TARGET_FP_F16INST.
13941 * config/aarch64/iterators.md (GPF_HF): New.
13942 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
13944 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13946 * config/aarch64/aarch64.c
13947 (aarch64_simd_container_mode): Add prototype.
13948 (aarch64_expand_mov_immediate): Add HI support.
13949 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
13950 (aarch64_can_const_movi_rtx_p): New.
13951 (aarch64_preferred_reload_class):
13952 Remove restrictions of using FP registers for certain SIMD operations.
13953 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
13954 (aarch64_valid_floating_const): Add integer move validation.
13955 (aarch64_simd_imm_scalar_p): Remove.
13956 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
13957 (aarch64_legitimate_constant_p): Expand list of supported cases.
13958 * config/aarch64/aarch64-protos.h
13959 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
13960 (aarch64_reinterpret_float_as_int): New.
13961 (aarch64_simd_imm_scalar_p): Remove.
13962 * config/aarch64/constraints.md (Uvi): New.
13963 (Dd): Split into Ds and new Dd.
13964 * config/aarch64/aarch64.md (*movsi_aarch64):
13966 (*movdi_aarch64): Add SIMD mov case.
13968 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13970 * tree-predcom.c: (struct chain): Handle store-store chain in which
13971 stores for elimination only store loop invariant values.
13972 (execute_pred_commoning_chain): Ditto.
13973 (prepare_initializers_chain_store_elim): Ditto.
13974 (prepare_finalizers): Ditto.
13975 (is_inv_store_elimination_chain): New function.
13976 (initialize_root_vars_store_elim_1): New function.
13978 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13980 * tree-predcom.c: Revise general description of the pass.
13981 (enum chain_type): New enum type for store elimination.
13982 (struct chain): New field supporting store elimination.
13983 (struct component): Ditto.
13984 (dump_chain): Dump store-stores chain.
13985 (release_chain): Release resources.
13986 (split_data_refs_to_components): Compute and create component
13987 contains only stores for elimination.
13988 (get_chain_last_ref_at): New function.
13989 (make_invariant_chain): Initialization.
13990 (make_rooted_chain): Specify chain type in parameter and record it.
13991 (add_looparound_copies): Skip for store-stores chain.
13992 (determine_roots_comp): Compute type of chain and pass it to
13994 (initialize_root_vars_store_elim_2): New function.
13995 (finalize_eliminated_stores): New function.
13996 (remove_stmt): Handle store for elimination.
13997 (execute_pred_commoning_chain): Execute predictive commoning on
13998 store-store chains.
13999 (determine_unroll_factor): Skip unroll for store-stores chain.
14000 (prepare_initializers_chain_store_elim): New function.
14001 (prepare_initializers_chain): Hanlde store-store chain.
14002 (prepare_finalizers_chain, prepare_finalizers): New function.
14003 (tree_predictive_commoning_loop): Return integer value indicating
14004 if loop is unrolled or lcssa form is corrupted.
14005 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14007 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14009 * tree-predcom.c (initialize_root): Delete.
14010 (execute_pred_commoning_chain): Initialize root vars and replace
14011 reference of non-combined chain directly, rather than call above
14014 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14016 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14017 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14019 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14021 * tree-predcom.c (struct chain): New field init_seq.
14022 (release_chain): Release init_seq.
14023 (prepare_initializers_chain): Record intialization stmts in above
14025 (insert_init_seqs): New function.
14026 (tree_predictive_commoning_loop): Call insert_init_seqs.
14028 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14030 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14032 2017-07-28 Richard Biener <rguenther@suse.de>
14034 * match.pd: Remove superfluous :c.
14035 * genmatch.c (simplify::id): Add member.
14036 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14038 (current_id): New global.
14039 (dt_node::parent): Move from ...
14040 (dt_operand::parent): ... here. Add for_id member.
14041 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14042 (decision_tree::find_node): Relax order requirement when
14043 merging DT_TRUE nodes to ones inbetween the current simplify
14044 and the one we try to merge with. Add diagnostic whenever
14045 we need to enforce pattern order by not merging.
14046 (decision_tree::insert): Set current_id.
14047 (decision_tree::print_node): Dump parent node and for_id.
14048 (parser::last_id): Add member.
14049 (parser::push_simplify): Assign unique id.
14050 (parser::parser): Initialize last_id.
14052 2017-07-28 Martin Liska <mliska@suse.cz>
14055 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14056 gimple_build_debug_bind.
14058 2017-07-28 Richard Biener <rguenther@suse.de>
14060 PR tree-optimization/81502
14061 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14063 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14065 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14066 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14067 for BIT_FIELD_REF args.
14068 * fold-const.c (make_bit_field_ref): Likewise.
14069 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14071 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14074 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14075 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14076 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14077 Or it into SANITIZER_UNDEFINED.
14078 * ubsan.c: Include gimple-fold.h and varasm.h.
14079 (ubsan_expand_ptr_ifn): New function.
14080 (instrument_pointer_overflow): New function.
14081 (maybe_instrument_pointer_overflow): New function.
14082 (instrument_object_size): Formatting fix.
14083 (pass_ubsan::execute): Call instrument_pointer_overflow
14084 and maybe_instrument_pointer_overflow.
14085 * internal-fn.c (expand_UBSAN_PTR): New function.
14086 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14087 * sanitizer.def (__ubsan_handle_pointer_overflow,
14088 __ubsan_handle_pointer_overflow_abort): New builtins.
14089 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14090 * internal-fn.def (UBSAN_PTR): New internal function.
14091 * opts.c (sanitizer_opts): Add pointer-overflow.
14092 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14093 * fold-const.c (build_range_check): Compute pointer range check in
14094 integral type if pointer arithmetics would be needed. Formatting
14097 2017-07-28 Martin Liska <mliska@suse.cz>
14100 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14101 parameters that are of a variable-length.
14103 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14105 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14107 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14108 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14109 (CRT_CALL_STATIC_FUNCTION): Likewise.
14110 (ASM_DEFAULT_SPEC): New define.
14111 (ASM_SPEC32): Likewise.
14112 (ASM_SPEC64): Likewise.
14113 (ASM_SPEC_COMMON): Likewise.
14114 (ASM_SPEC): Likewise.
14115 (INVALID_64BIT): Likewise.
14116 (LINK_OS_DEFAULT_SPEC): Likewise.
14117 (LINK_OS_SPEC32): Likewise.
14118 (LINK_OS_SPEC64): Likewise.
14119 (POWERPC_LINUX): Likewise.
14120 (PTRDIFF_TYPE): Likewise.
14121 (RESTORE_FP_PREFIX): Likewise.
14122 (RESTORE_FP_SUFFIX): Likewise.
14123 (SAVE_FP_PREFIX): Likewise.
14124 (SAVE_FP_SUFFIX): Likewise.
14125 (SIZE_TYPE): Likewise.
14126 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14127 (TARGET_64BIT): Likewise.
14128 (TARGET_64BIT): Likewise.
14129 (TARGET_AIX): Likewise.
14130 (WCHAR_TYPE_SIZE): Likewise.
14131 (WCHAR_TYPE): Undefine.
14132 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14133 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14134 (CPP_OS_RTEMS_SPEC): Delete.
14135 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14136 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14138 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14140 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14142 PR tree-optimization/81578
14143 * tree-parloops.c (build_new_reduction): Bail out if
14144 reduction_code isn't one of the standard OpenMP reductions.
14145 Move the details printing after that decision.
14147 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14149 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14150 related to reload_in_progress.
14151 (splat_input_operand): Likewise.
14152 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14154 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14156 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14157 (TARGET_INSTANTIATE_DECLS): Likewise.
14158 (legitimate_indexed_address_p): Delete reload_in_progress code.
14159 (rs6000_debug_legitimate_address_p): Likewise.
14160 (rs6000_eliminate_indexed_memrefs): Likewise.
14161 (rs6000_emit_le_vsx_store): Likewise.
14162 (rs6000_emit_move_si_sf_subreg): Likewise.
14163 (rs6000_emit_move): Likewise.
14164 (register_to_reg_type): Likewise.
14165 (rs6000_pre_atomic_barrier): Likewise.
14166 (rs6000_machopic_legitimize_pic_address): Likewise.
14167 (rs6000_allocate_stack_temp): Likewise.
14168 (rs6000_address_for_fpconvert): Likewise.
14169 (rs6000_address_for_altivec): Likewise.
14170 (rs6000_secondary_memory_needed_rtx): Delete function.
14171 (rs6000_check_sdmode): Likewise.
14172 (rs6000_alloc_sdmode_stack_slot): Likewise.
14173 (rs6000_instantiate_decls): Likewise.
14174 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14175 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14176 Delete reload_in_progress.
14177 (*vec_reload_and_plus_<mptrsize>): Likewise.
14178 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14179 (vsx_div_v2di): Likewise.
14180 (vsx_udiv_v2di): Likewise.
14182 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14184 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14185 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14186 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14187 (rs6000_debug_reg_global): Delete print of LRA status.
14188 (rs6000_option_override_internal): Delete dead LRA related code.
14189 (rs6000_lra_p): Delete function.
14190 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14192 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14194 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14195 * config/riscv/rtems.h: New file.
14197 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14198 Sudakshina Das <sudi.das@arm.com>
14200 * config/aarch64/aarch64.md
14201 (define_split for and<mode>3nr_compare): Move
14202 non aarch64_logical_operand to a register.
14203 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14204 register immediate operand to a register.
14205 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14207 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14209 PR middle-end/81564
14210 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14212 2017-07-27 Richard Biener <rguenther@suse.de>
14214 PR tree-optimization/81573
14215 PR tree-optimization/81494
14216 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14217 multi defuse cycle case.
14219 2017-07-27 Richard Biener <rguenther@suse.de>
14221 PR tree-optimization/81571
14222 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14225 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14227 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14228 earlier and only if MASK_FPU is set. Adjust formatting.
14230 2017-07-27 Martin Liska <mliska@suse.cz>
14232 * opt-functions.awk: Add validation of value of Init.
14233 * optc-gen.awk: Pass new argument.
14235 2017-07-27 Martin Liska <mliska@suse.cz>
14237 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14238 Fix wrong condition.
14240 2017-07-27 Martin Liska <mliska@suse.cz>
14242 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14243 BBs and edges seen by autoFDO.
14245 2017-07-27 Richard Biener <rguenther@suse.de>
14247 PR tree-optimization/81502
14248 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14249 with incompatible but same sized type.
14250 (execute_update_addresses_taken): Likewise.
14252 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
14254 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
14255 flag_tree_loop_vectorize rather than flag_tree_vectorize.
14257 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14260 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
14261 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
14262 Change s_operand to memory_operand.
14264 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
14266 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
14267 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
14268 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
14269 Emit instructions rather than returning an expression. Handle TFmode
14270 and KFmode by casting to TImode.
14271 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
14272 (rs6000_emit_le_vsx_store): Likewise.
14273 * config/rs6000/vsx.md (VSX_TI): New iterator.
14274 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
14275 (*vsx_le_undo_permute_<mode>): Likewise.
14276 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
14277 emit the split sequence.
14278 (*vsx_le_perm_store_<mode>): Likewise.
14280 2017-07-27 Jakub Jelinek <jakub@redhat.com>
14282 PR tree-optimization/81555
14283 PR tree-optimization/81556
14284 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
14285 if true, force CHANGED for the recursive invocation.
14286 (reassociate_bb): Remember original length of ops array, pass
14287 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
14289 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
14290 attributes for noipa attribute. For naked attribute use
14291 lookup_attribute first before lookup_attribute_spec.
14292 * final.c (rest_of_handle_final): Disable IPA RA for functions with
14294 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
14295 for functions with noipa attribute.
14296 (cgraph_externally_visible_p): Return true for functions with noipa
14298 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
14299 for functions with noipa attribute.
14300 * doc/extend.texi: Document noipa function attribute.
14301 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
14302 also for functions with noipa attribute.
14303 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
14305 2017-07-26 Andrew Pinski <apinski@cavium.com>
14307 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
14308 vec_unalign_load_cost and vec_unalign_store_cost.
14310 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14312 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
14313 -mvsx-small-integer option.
14314 (ISA_3_0_MASKS_IEEE): Likewise.
14315 (OTHER_VSX_VECTOR_MASKS): Likewise.
14316 (POWERPC_MASKS): Likewise.
14317 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
14318 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
14319 code, only testing for DImode being allowed in non-VSX floating
14321 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
14322 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
14324 (rs6000_option_override_internal): Delete -mvsx-small-integer.
14325 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
14326 TARGET_P8_VECTOR test.
14327 (rs6000_secondary_reload_simple_move): Likewise.
14328 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
14329 since TARGET_P9_VECTOR was already tested.
14330 (rs6000_opt_masks): Remove -mvsx-small-integer.
14331 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
14332 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14334 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
14335 test for TARGET_VEXTRACTUB was used, and that uses
14337 (p9 extract splitter): Likewise.
14338 (vsx_extract_<mode>_di_p9): Likewise.
14339 (vsx_extract_<mode>_store_p9): Likewise.
14340 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
14341 for TARGET_P9_VECTOR was used. Delete code that is now dead with
14342 the elimination of TARGET_VSX_SMALL_INTEGER.
14343 (vsx_extract_<mode>_p8): Likewise.
14344 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
14345 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
14346 (vsx_set_<mode>_p9): Likewise.
14347 (vsx_set_v4sf_p9): Likewise.
14348 (vsx_set_v4sf_p9_zero): Likewise.
14349 (vsx_insert_extract_v4sf_p9): Likewise.
14350 (vsx_insert_extract_v4sf_p9_2): Likewise.
14351 * config/rs6000/rs6000.md (sign extend splitter): Change
14352 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
14353 (floatsi<mode>2_lfiwax_mem): Likewise.
14354 (floatunssi<mode>2_lfiwzx_mem): Likewise.
14355 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
14356 since a test for TARGET_P9_VECTOR was used.
14357 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14358 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
14359 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14360 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14361 TARGET_P8_VECTOR test.
14362 (fix_trunc<mode>si2_stfiwx): Likewise.
14363 (fix_trunc<mode>si2_internal): Likewise.
14364 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
14365 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14367 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14368 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14369 TARGET_P8_VECTOR test.
14370 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14371 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
14372 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14374 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14375 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
14376 since a test for TARGET_P9_VECTOR was used.
14377 (splitter for loading small constants): Likewise.
14379 2017-07-26 Andrew Pinski <apinski@cavium.com>
14381 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
14384 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
14387 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
14388 (fp_valid_at): Likewise.
14390 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14392 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
14393 (qdf24xx_addrcost_table): Likewise.
14394 (cortexa57_tunings): Update to use generic_branch_cost.
14395 (cortexa72_tunings): Likewise.
14396 (cortexa73_tunings): Likewise.
14397 (qdf24xx_tunings): Likewise.
14399 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14401 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
14402 (thunderx2t99_branch_cost): Likewise.
14403 (cortexa35_tunings): Update to use generic_branch_cost.
14404 (cortexa53_tunings): Likewise.
14405 (cortexa57_tunings): Likewise.
14406 (cortexa72_tunings): Likewise.
14407 (cortexa73_tunings): Likewise.
14408 (thunderx2t99_tunings): Likewise.
14410 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14412 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
14413 (sparc_option_override): Honour MASK_FSMULD.
14414 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
14415 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
14416 * config/sparc/sparc.opt (mfsmuld): New option.
14417 * doc/invoke.texi (mfsmuld): Document option.
14419 2017-07-26 Marek Polacek <polacek@redhat.com>
14421 PR middle-end/70992
14422 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
14424 2017-07-26 Richard Biener <rguenther@suse.de>
14426 * gimple-match-head.c (do_valueize): Return OP if valueize
14428 (get_def): New helper to get at the def stmt of a SSA name
14429 if valueize allows.
14430 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
14431 do_valueize to get at the def stmt.
14432 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
14434 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
14436 PR middle-end/46932
14437 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
14439 2017-07-26 Martin Liska <mliska@suse.cz>
14442 * function.c (expand_function_start): Make expansion of
14443 nonlocal_goto_save_area after parm_birth_insn.
14445 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14447 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
14448 from all CPU target flags enable members.
14450 2017-07-26 Richard Biener <rguenther@suse.de>
14452 * genmatch.c (dt_simplify::gen): Make iterator vars const.
14453 (decision_tree::gen): Make 'type' const.
14454 (write_predicate): Likewise.
14456 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14458 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
14459 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
14460 (rs6000_option_override_internal): Likewise.
14461 (rs6000_expand_vector_set): Likewise.
14462 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
14463 (TARGET_UPPER_REGS_SF): Likewise.
14464 (TARGET_UPPER_REGS_DI): Likewise.
14465 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
14466 (TARGET_DIRECT_MOVE_64BIT): Likewise.
14467 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
14468 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14469 (Splitters for DI constants in Altivec registers): Likewise.
14470 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
14471 (vsx_set_v4sf_p9): Likewise.
14472 (vsx_set_v4sf_p9_zero): Likewise.
14473 (vsx_insert_extract_v4sf_p9): Likewise.
14474 (vsx_insert_extract_v4sf_p9_2): Likewise.
14476 2017-07-25 Carl Love <cel@us.ibm.com>
14478 * doc/extend.texi: Update the built-in documentation file for the
14479 existing built-in functions
14480 vector signed char vec_cnttz (vector signed char);
14481 vector unsigned char vec_cnttz (vector unsigned char);
14482 vector signed short vec_cnttz (vector signed short);
14483 vector unsigned short vec_cnttz (vector unsigned short);
14484 vector signed int vec_cnttz (vector signed int);
14485 vector unsigned int vec_cnttz (vector unsigned int);
14486 vector signed long long vec_cnttz (vector signed long long);
14487 vector unsigned long long vec_cnttz (vector unsigned long long);
14489 2017-07-25 Andrew Pinski <apinski@cavium.com>
14491 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
14492 accesses where the use is for the first operand of a BIT_INSERT.
14494 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14497 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14498 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14500 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14502 * config/i386/gstabs.h: Delete.
14503 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
14505 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
14507 * config/i386/i386.c (ix86_decompose_address): Do not check for
14508 register RTX when looking at index_reg or base_reg.
14509 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
14511 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
14513 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
14514 to update EH info here.
14516 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14518 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
14520 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14522 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
14524 2017-07-25 Torsten Duwe <duwe@suse.de>
14526 * common.opt: Introduce -fpatchable-function-entry
14527 command line option, and its variables function_entry_patch_area_size
14528 and function_entry_patch_area_start.
14529 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
14530 including a two-value parser.
14531 * target.def (print_patchable_function_entry): New target hook.
14532 * targhooks.h (default_print_patchable_function_entry): New function.
14533 * targhooks.c (default_print_patchable_function_entry): Likewise.
14534 * toplev.c (process_options): Switch off IPA-RA if
14535 patchable function entries are being generated.
14536 * varasm.c (assemble_start_function): Look at the
14537 patchable-function-entry command line switch and current
14538 function attributes and maybe generate NOP instructions by
14539 calling the print_patchable_function_entry hook.
14540 * doc/extend.texi: Document patchable_function_entry attribute.
14541 * doc/invoke.texi: Document -fpatchable_function_entry
14542 command line option.
14543 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
14545 * doc/tm.texi: Re-generate.
14547 2017-07-25 Jakub Jelinek <jakub@redhat.com>
14550 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
14551 TARGET_AVX512DQ rather than TARGET_AVX512BW.
14553 2017-07-25 Tamar Christina <tamar.christina@arm.com>
14555 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
14557 2017-07-25 Richard Biener <rguenther@suse.de>
14559 PR tree-optimization/81455
14560 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
14561 not walk in cycles when looking for guards.
14563 2017-07-25 Richard Biener <rguenther@suse.de>
14565 PR tree-optimization/81529
14566 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
14567 when optimizing backedge uses.
14569 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
14571 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
14573 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
14574 to dl_section_ref. On AIX, append an expression to subtract
14575 the size of the section length to dl_section_ref.
14577 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
14579 * configure.ac: If any of the config.* scripts fail, exit 1.
14580 * configure: Regenerate.
14582 2017-07-25 Richard Biener <rguenther@suse.de>
14584 PR middle-end/81546
14585 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
14586 of immediate uses, be more verbose on errors.
14588 2017-07-25 Richard Biener <rguenther@suse.de>
14590 PR tree-optimization/81510
14591 * tree-vect-loop.c (vect_is_simple_reduction): When the
14592 reduction stmt is not inside the loop bail out.
14594 2017-07-25 Richard Biener <rguenther@suse.de>
14596 PR tree-optimization/81303
14597 * tree-vect-loop-manip.c (vect_loop_versioning): Build
14598 profitability check against LOOP_VINFO_NITERSM1.
14600 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14602 * domwalk.c (cmp_bb_postorder): Simplify.
14603 (sort_bbs_postorder): New function. Use it...
14604 (dom_walker::walk): ...here to optimize common cases.
14606 2017-07-25 Martin Liska <mliska@suse.cz>
14609 * ipa-visibility.c (function_and_variable_visibility): Make the
14610 redirection just on target that supports aliasing.
14611 Fix GNU coding style.
14613 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
14616 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
14618 * config/arm/rtems.h: Likewise.
14619 * config/bfin/rtems.h: Likewise.
14620 * config/i386/rtemself.h: Likewise.
14621 * config/lm32/rtems.h: Likewise.
14622 * config/m32c/rtems.h: Likewise.
14623 * config/m68k/rtemself.h: Likewise.
14624 * config/microblaze/rtems.h: Likewise.
14625 * config/mips/rtems.h: Likewise.
14626 * config/moxie/rtems.h: Likewise.
14627 * config/nios2/rtems.h: Likewise.
14628 * config/powerpcspe/rtems.h: Likewise.
14629 * config/rs6000/rtems.h: Likewise.
14630 * config/rtems.h: Likewise.
14631 * config/sh/rtems.h: Likewise.
14632 * config/sh/rtemself.h: Likewise.
14633 * config/sparc/rtemself.h: Likewise.
14635 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
14638 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
14639 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
14640 * tree-ssa-structalias.c (alias_get_name): Same.
14642 2017-07-25 Bin Cheng <bin.cheng@arm.com>
14645 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
14646 instructions if no du chain is found.
14648 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
14650 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
14652 2017-07-25 Richard Biener <rguenther@suse.de>
14654 PR middle-end/81505
14655 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
14658 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14660 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
14661 upper-regs options.
14662 (ISA_2_7_MASKS_SERVER): Likewise.
14663 (ISA_3_0_MASKS_IEEE): Likewise.
14664 (OTHER_P8_VECTOR_MASKS): Likewise.
14665 (OTHER_VSX_VECTOR_MASKS): Likewise.
14666 (POWERPC_MASKS): Likewise.
14667 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
14668 duplicate list of options.
14669 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
14670 explicit -mupper-regs options.
14671 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
14672 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
14673 alias for -mupper-regs-df.
14674 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
14675 (rs6000_init_hard_regno_mode_ok): Likewise.
14676 (rs6000_option_override_internal): Likewise.
14677 (rs6000_opt_masks): Likewise.
14678 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
14679 options in terms of whether -mvsx or -mpower8-vector was used.
14680 (TARGET_UPPER_REGS_DI): Likewise.
14681 (TARGET_UPPER_REGS_SF): Likewise.
14682 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
14683 -mupper-regs-* options.
14685 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
14687 * passes.c (emergency_dump_function): Print some empty lines and a
14688 header before the RTL dump.
14690 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
14692 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
14694 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
14697 * config/aarch64/aarch64.c (aarch64_classify_symbol):
14698 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
14700 2017-07-24 Carl Love <cel@us.ibm.com>
14702 * config/rs6000/rs6000-c.c: Add support for built-in functions
14703 vector float vec_extract_fp32_from_shorth (vector unsigned short);
14704 vector float vec_extract_fp32_from_shortl (vector unsigned short);
14705 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
14706 vec_extract_fp_from_shortl): Add defines for the two builtins.
14707 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
14708 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
14710 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
14711 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
14712 * doc/extend.texi: Update the built-in documentation file for the
14713 new built-in function.
14715 2017-07-24 Jakub Jelinek <jakub@redhat.com>
14718 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
14720 * doc/generic.texi: Likewise.
14721 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14722 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14724 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
14726 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
14727 (aarch64_mls_elt_merge<mode>): Likewise.
14729 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
14731 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
14732 having __cxa_atexit.
14734 2017-07-23 Michael Collison <michael.collison@arm.com>
14736 * config/arm/arm.c (arm_option_override): Deprecate
14737 use of -mstructure-size-boundary.
14738 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
14739 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
14741 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14744 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
14745 Reduce cost estimate for direct moves.
14747 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
14750 * config/i386/i386.c (ix86_option_override_internal): Disable
14751 BMI, BMI2 and TBM instructions for -m16.
14753 2017-07-21 Carl Love <cel@us.ibm.com>
14755 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14756 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14757 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14758 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14759 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14760 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14761 VMULOSW): New enum "unspec" values.
14762 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14763 altivec_vmulosw): New patterns.
14764 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14765 VMULOSW): Add definitions.
14767 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
14769 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
14770 (qdf24xx): Likewise.
14771 * config/aarch64/aarch64-options-extensions.def (rdma); New.
14772 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
14773 (AARCH64_FL_V8_1): Renumber.
14774 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
14775 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
14776 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
14777 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
14778 rdma to feature modifiers list.
14780 2017-07-21 Yury Gribov <tetra2005@gmail.com>
14782 PR middle-end/56727
14783 * ipa-visibility (function_and_variable_visibility): Convert
14784 recursive PLT call to direct call if appropriate.
14786 2017-07-21 Andrew Pinski <apinski@cavium.com>
14788 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
14789 operand 1 to see if the types precision matches.
14790 * fold-const.c (operand_equal_p): Likewise.
14792 2017-07-21 Richard Biener <rguenther@suse.de>
14794 PR tree-optimization/81303
14795 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
14796 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
14797 (vect_peeling_hash_get_lowest_cost): Adjust.
14798 (vect_enhance_data_refs_alignment): Likewise. Use
14799 vect_get_peeling_costs_all_drs to compute the penalty for no
14800 peeling to match up costs.
14802 2017-07-21 Richard Biener <rguenther@suse.de>
14804 PR tree-optimization/81500
14805 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
14806 we didn't identify a reduction path.
14808 2017-07-21 Tom de Vries <tom@codesourcery.com>
14809 Cesar Philippidis <cesar@codesourcery.com>
14811 PR gcov-profile/81442
14812 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
14815 2017-07-21 Tom de Vries <tom@codesourcery.com>
14818 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
14820 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
14821 nvptx_override_options_after_change.
14823 2017-07-21 Ulrich Drepper <drepper@redhat.com>
14825 * dwarf2out.c (output_file_names): Avoid double testing for
14826 dwarf_version >= 5.
14828 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
14830 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
14832 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
14834 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
14836 (try_crossjump_to_edge): Do not punt on partitioned functions.
14838 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
14840 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
14841 Put all BBs reachable only via paths crossing cold region to cold
14843 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
14845 2016-07-21 Richard Biener <rguenther@suse.de>
14847 PR tree-optimization/81303
14848 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
14849 into account prologue and epilogue iterations when raising
14850 min_profitable_iters to sth at least covering one vector iteration.
14852 2017-07-21 Tamar Christina <tamar.christina@arm.com>
14854 * config/arm/arm.c (arm_test_cpu_arch_dat):
14857 2017-07-20 Nathan Sidwell <nathan@acm.org>
14859 Remove TYPE_METHODS.
14860 * tree.h (TYPE_METHODS): Delete.
14861 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
14862 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
14863 (dbxout_type_methods): Scan TYPE_FIELDS.
14864 (dbxout_type): Don't check TYPE_METHODS here.
14865 * function.c (use_register_for_decl): Always ignore register for
14866 class types when not optimizing.
14867 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
14868 * tree.c (free_lang_data_in_type): Stitch out member functions and
14869 templates from TYPE_FIELDS.
14870 (build_distinct_type_copy, verify_type_variant,
14871 verify_type): Member fns are on TYPE_FIELDS.
14872 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
14873 * tree-pretty-print.c (dump_generic_node): Likewise.
14875 2017-07-20 Jakub Jelinek <jakub@redhat.com>
14878 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
14879 V2TImode and V4TImode.
14880 (ix86_expand_vector_extract): Likewise.
14881 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
14882 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
14883 (ssescalarmode): Handle V4TImode and V2TImode.
14884 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
14885 (*vec_extractv2ti, *vec_extractv4ti): New insns.
14886 (VEXTRACTI128_MODE): New mode iterator.
14887 (splitter for *vec_extractv?ti first element): New.
14888 (VEC_INIT_MODE): New mode iterator.
14889 (vec_init<mode>): Consolidate 3 expanders into one using
14890 VEC_INIT_MODE mode iterator.
14892 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
14894 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
14895 non_spilled_static_chain_regno_p.
14897 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
14899 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
14901 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
14903 * bb-reorder.c (connect_traces): Allow copying of blocks within
14906 2017-07-20 Richard Biener <rguenther@suse.de>
14908 * gimple.h (gimple_phi_result): Add gphi * overload.
14909 (gimple_phi_result_ptr): Likewise.
14910 (gimple_phi_arg): Likewise. Adjust index assert to only
14911 allow actual argument accesses rather than all slots available
14913 (gimple_phi_arg_def): Add gphi * overload.
14914 * tree-phinodes.c (make_phi_node): Initialize only actual
14916 (resize_phi_node): Clear memory not covered by old node,
14917 do not initialize excess argument slots.
14918 (reserve_phi_args_for_new_edge): Initialize new argument slot
14921 2017-07-20 Bin Cheng <bin.cheng@arm.com>
14923 PR tree-optimization/81388
14925 2016-07-21 Bin Cheng <bin.cheng@arm.com>
14927 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
14928 by removing computation of may_be_zero.
14930 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14931 Tom de Vries <tom@codesourcery.com>
14933 PR middle-end/81030
14934 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
14935 when gimple level profile disagrees with what RTL expander did.
14937 2017-07-20 Richard Biener <rguenther@suse.de>
14939 PR tree-optimization/61171
14940 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
14941 (vect_analyze_stmt): Add slp instance parameter.
14942 (vectorizable_reduction): Likewise.
14943 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
14944 (vect_is_simple_reduction): Deal with chains not detected
14945 as SLP reduction chain, specifically not properly associated
14946 chains containing a mix of plus/minus.
14947 (get_reduction_op): Remove.
14948 (get_initial_defs_for_reduction): Simplify, pass in whether
14949 this is a reduction chain, pass in the SLP node for the PHIs.
14950 (vect_create_epilog_for_reduction): Get the SLP instance as
14952 (vectorizable_reduction): Get the SLP instance as arg.
14953 During analysis remember the SLP node with the PHIs in the
14954 instance. Simplify getting at the vectorized reduction PHIs.
14955 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
14956 through SLP instance.
14957 (vect_slp_analyze_operations): Likewise.
14958 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
14959 (vect_transform_stmt): Likewise.
14961 2017-07-20 Tom de Vries <tom@codesourcery.com>
14963 PR tree-optimization/81489
14964 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
14965 read of phi arg location to before loop that modifies phi.
14967 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14969 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
14972 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14974 PR middle-end/81331
14975 * except.c (execute): Fix ordering issue.
14977 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
14979 PR rtl-optimization/81423
14980 * combine.c (make_compound_operation_int): Don't try to optimize
14981 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
14983 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
14985 PR rtl-optimization/81423
14986 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
14987 with a constant that is -1 in the truncated to mode.
14989 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14991 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
14992 (determine_unlikely_bbs): ... here.
14993 * predict.h (propagate_unlikely_bbs_forward): Declare.
14994 * cfgexpand.c (pass_expand::execute): Use it.
14995 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
14997 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
14998 propagate_unlikely_bbs_forward.
15000 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15002 PR middle-end/81331
15003 * except.c (maybe_add_nop_after_section_switch): New function.
15006 2017-07-19 Tom de Vries <tom@codesourcery.com>
15008 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15010 2017-07-19 Tom de Vries <tom@codesourcery.com>
15012 * gimple.h (gimple_phi_arg): Make assert more strict.
15014 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15016 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15017 * config/rs6000/mmintrin.h: New file.
15018 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15020 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15022 PR tree-optimization/81346
15023 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15025 2017-07-19 Tom de Vries <tom@codesourcery.com>
15027 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15029 2017-07-19 Tom de Vries <tom@codesourcery.com>
15031 * config/nvptx/nvptx-modes.def: Add V2DImode.
15032 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15033 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15034 (nvptx_output_mov_insn): Handle lack of mov.b128.
15035 (nvptx_print_operand): Handle 'H' and 'L' codes.
15036 (nvptx_vector_mode_supported): Allow V2DImode.
15037 (nvptx_preferred_simd_mode): New function.
15038 (nvptx_data_alignment): New function.
15039 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15040 nvptx_preferred_simd_mode.
15041 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15043 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15045 2017-07-19 Tom de Vries <tom@codesourcery.com>
15047 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15048 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15049 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15050 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15051 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15052 (mov<VECIM>_insn): New define_insn.
15053 (define_expand "mov<VECIM>): New define_expand.
15055 2017-07-19 Tom de Vries <tom@codesourcery.com>
15057 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15059 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15061 PR tree-optimization/81346
15062 * fold-const.h (fold_div_compare, range_check_type): Declare.
15063 * fold-const.c (range_check_type): New function.
15064 (build_range_check): Use range_check_type.
15065 (fold_div_compare): No longer static, rewritten into
15066 a match.pd helper function.
15067 (fold_comparison): Don't call fold_div_compare here.
15068 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15069 as helper function.
15071 2017-07-19 Nathan Sidwell <nathan@acm.org>
15073 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15074 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15075 * tree.c (find_decls_types_r, verify_type): Use
15076 TYPE_{MIN,MAX}_VALUE_RAW.
15077 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15078 (hash_tree): Likewise.
15079 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15081 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15084 2017-07-18 Tom de Vries <tom@codesourcery.com>
15086 PR middle-end/81464
15087 * omp-expand.c (expand_omp_for_static_chunk): Handle
15088 equal-argument loop exit phi.
15090 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15093 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15094 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15095 operand 2 predicate.
15096 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15097 operand 2 predicate.
15098 (ror,rol -> rorx splitters): Use const_int_operand as
15099 operand 2 predicate.
15101 2017-06-18 Richard Biener <rguenther@suse.de>
15103 PR tree-optimization/81410
15104 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15105 the gap in the ! slp_perm SLP case after each group.
15107 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15109 PR middle-end/81463
15110 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15113 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15115 PR middle-end/81462
15116 * predict.c (set_even_probabilities): Cleanup; do not affect
15117 probabilities that are already known.
15118 (combine_predictions_for_bb): Call even when count is set.
15120 2017-07-18 Nathan Sidwell <nathan@acm.org>
15122 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15125 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15128 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15129 optimization for loop niter analysis.
15131 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15134 * config/avr/avr.c (avr_optimize_casesi): Don't use
15135 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15137 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15139 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15140 body_cost_vec from _vect_peel_extended_info.
15141 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15142 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15145 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15147 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15149 2017-07-18 Richard Biener <rguenther@suse.de>
15151 PR tree-optimization/80620
15152 PR tree-optimization/81403
15153 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15154 info when re-using a VN table entry.
15156 2017-07-18 Richard Biener <rguenther@suse.de>
15158 PR tree-optimization/81418
15159 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15160 vectype_in. Verify that with lane-reducing reduction operations
15161 we have a single def-use cycle.
15163 2017-07-17 Carl Love <cel@us.ibm.com>
15165 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15167 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15168 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15169 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15170 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15171 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15172 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15173 VMULOSW): New enum "unspec" values.
15174 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15175 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15176 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15177 altivec_vmulosw): New patterns.
15178 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15179 VMULOSW): Add definitions.
15180 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15182 * config/alpha/alpha.c: Include predict.h.
15184 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15186 * tree-vrp.c (compare_assert_loc): Fix comparison function
15187 to return predictable results.
15189 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15191 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15193 (subdi3): Likewise.
15194 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15195 * doc/invoke.texi (mexpand-adddi): Update text.
15197 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15199 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15200 that also clobbers the CC register. The old expand code is moved
15202 (*arc_clzsi2): ... here.
15203 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15204 the CC register. The old expand code is moved to ...
15205 (arc_ctzsi2): ... here.
15207 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15209 * config/arc/arc.opt (mindexed-loads): Use initial value
15210 TARGET_INDEXED_LOADS_DEFAULT.
15211 (mauto-modify-reg): Use initial value
15212 TARGET_AUTO_MODIFY_REG_DEFAULT.
15213 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15214 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15215 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15216 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15218 2017-07-17 Martin Liska <mliska@suse.cz>
15221 * opts.c (finish_options): Do not allow -fgnu-tm
15222 w/ -fsanitize={kernel-,}address. Say sorry.
15224 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15227 * tree-loop-distribution.c (classify_partition): Only assert on
15228 numer of iterations.
15229 (merge_dep_scc_partitions): Delete prameter. Update function call.
15230 (distribute_loop): Remove code handling loop with unknown niters.
15231 (pass_loop_distribution::execute): Skip loop with unknown niters.
15233 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15236 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15237 function sort_partitions_by_post_order.
15239 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15241 PR tree-optimization/81374
15242 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15243 the max index of basic blocks, rather than number of basic blocks.
15245 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15247 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15249 (arc_legitimate_pic_operand_p): Likewise.
15250 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
15252 (arc_needs_pcl_p): Likewise.
15253 (arc_legitimate_pc_offset_p): Likewise.
15254 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
15255 function is also used in constrains.md.
15256 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
15257 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
15258 PLUS. Only return true/false in known cases, otherwise assert.
15259 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
15260 is already called in arc_legitimate_constant_p.
15261 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
15263 (LEGITIMATE_PIC_OPERAND_P): Use
15264 arc_raw_symbolic_reference_mentioned_p function.
15265 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
15270 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15271 Andrew Burgess <andrew.burgess@embecosm.com>
15273 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
15274 (arc_return_address_register): New function.
15275 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
15276 (arc_handle_fndecl_attribute): Add naked attribute.
15277 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15278 (TARGET_WARN_FUNC_RETURN): Likewise.
15279 (arc_allocate_stack_slots_for_args): New function.
15280 (arc_warn_func_return): Likewise.
15281 (machine_function): Change type fn_type.
15282 (arc_compute_function_type): Consider new naked function type,
15283 change function return type.
15284 (arc_must_save_register): Adapt to handle new
15285 arc_compute_function_type's return type.
15286 (arc_expand_prologue): Likewise.
15287 (arc_expand_epilogue): Likewise.
15288 (arc_return_address_regs): Delete.
15289 (arc_return_address_register): New function.
15290 (arc_epilogue_uses): Use above function.
15291 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
15292 (arc_function_type): Change encoding, add naked type.
15293 (ARC_INTERRUPT_P): Change to handle the new encoding.
15294 (ARC_FAST_INTERRUPT_P): Likewise.
15295 (ARC_NORMAL_P): Define.
15296 (ARC_NAKED_P): Likewise.
15297 (arc_compute_function_type): Delete prototype.
15298 * config/arc/arc.md (in_ret_delay_slot): Use
15299 arc_return_address_register function.
15300 (simple_return): Likewise.
15301 (p_return_i): Likewise.
15303 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15305 PR tree-optimization/81428
15306 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
15307 can't be built for those types.
15309 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15311 Remove stuff dead since r239246.
15313 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
15314 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
15315 (avr_inform_devices): Remove dead stuff.
15317 2017-07-17 Tamar Christina <tamar.christina@arm.com>
15319 * config/arm/arm_neon.h: Fix softp typo.
15321 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15323 PR tree-optimization/81365
15324 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
15325 aggregate moves onto bb predecessor edges, make sure there are no
15326 loads that could alias the lhs in between the start of bb and the
15329 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15332 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
15333 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
15334 [LSHIFTRT, outer_code = TRUNCATE]: Same.
15336 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15338 PR tree-optimization/81396
15339 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
15340 (init_symbolic_number): Initialize it to 1.
15341 (perform_symbolic_merge): Add n_ops from both operands into the new
15343 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
15344 without base_addr as useless if they need more than one operation.
15345 (bswap_replace): Handle !bswap case for NULL base_addr.
15347 2017-07-17 Tom de Vries <tom@codesourcery.com>
15350 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
15353 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
15355 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
15356 conditional builtin define __FIX_LEON3FT_B2BST.
15358 2017-07-17 Daniel Cederman <cederman@gaisler.com>
15360 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
15361 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
15364 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15366 PR rtl-optimization/81424
15367 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
15368 to remove potential trapping from operands if -fnon-call-exceptions.
15370 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15372 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
15373 profile_proability for scalling.
15374 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
15376 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15378 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
15380 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15382 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
15383 fixpoint arithmetics.
15385 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15387 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
15388 fixpoint arithmetics.
15390 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15392 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
15393 fixpoint arithmetics.
15395 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15397 * profile-count.h (profile_probability::from_reg_br_prob_note,
15398 profile_probability::to_reg_br_prob_note): New functions.
15399 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
15400 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
15401 * predict.c (probability_reliable_p): Update.
15402 (edge_probability_reliable_p): Update.
15403 (br_prob_note_reliable_p): Update.
15404 (invert_br_probabilities): Update.
15405 (add_reg_br_prob_note): New function.
15406 (combine_predictions_for_insn): Update.
15407 * asan.c (asan_clear_shadow): Update.
15408 * cfgbuild.c (compute_outgoing_frequencies): Update.
15409 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15410 (update_br_prob_note): Update.
15411 (rtl_verify_edges): Update.
15412 (purge_dead_edges): Update.
15413 (fixup_reorder_chain): Update.
15414 * emit-rtl.c (try_split): Update.
15415 * ifcvt.c (cond_exec_process_insns): Update.
15416 (cond_exec_process_if_block): Update.
15417 (dead_or_predicable): Update.
15418 * internal-fn.c (expand_addsub_overflow): Update.
15419 (expand_neg_overflow): Update.
15420 (expand_mul_overflow): Update.
15421 * loop-doloop.c (doloop_modify): Update.
15422 * loop-unroll.c (compare_and_jump_seq): Update.
15423 * optabs.c (emit_cmp_and_jump_insn_1): Update.
15424 * predict.h: Update.
15425 * reorg.c (mostly_true_jump): Update.
15427 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
15428 * config/alpha/alpha.c (emit_unlikely_jump): Update.
15429 * config/arc/arc.c: (emit_unlikely_jump): Update.
15430 * config/arm/arm.c: (emit_unlikely_jump): Update.
15431 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
15432 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
15433 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
15434 (ix86_print_operand): Update.
15435 (ix86_split_fp_branch): Update.
15436 (predict_jump): Update.
15437 * config/ia64/ia64.c (ia64_print_operand): Update.
15438 * config/mmix/mmix.c (mmix_print_operand): Update.
15439 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
15440 (rs6000_expand_split_stack_prologue): Update.
15441 * config/rs6000/rs6000.c: Update.
15442 * config/s390/s390.c (s390_expand_vec_strlen): Update.
15443 (s390_expand_vec_movstr): Update.
15444 (s390_expand_cs_tdsi): Update.
15445 (s390_expand_split_stack_prologue): Update.
15446 * config/sh/sh.c (sh_print_operand): Update.
15447 (expand_cbranchsi4): Update.
15448 (expand_cbranchdi4): Update.
15449 * config/sparc/sparc.c (output_v9branch): Update.
15450 * config/spu/spu.c (get_branch_target): Update.
15451 (ea_load_store_inline): Update.
15452 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
15453 * config/tilepro/tilepro.c: Update.
15455 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15457 * gimplify.c (mostly_copy_tree_r): Revert latest change.
15458 (gimplify_save_expr): Likewise.
15460 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15462 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
15464 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15466 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
15468 * timevar.def (TV_IPA_FNSUMMARY): Define.
15470 2017-07-16 Daniel Cederman <cederman@gaisler.com>
15472 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
15473 to back store errata sensitive sequence from being generated.
15474 (sqrtdf2_fix): Likewise.
15476 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15478 * tree-ssa-threadupdate.c (compute_path_counts,
15479 update_joiner_offpath_counts): Use profile_probability.
15481 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
15484 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15486 * config/arm/arm-c.c (arm_cpu_builtins): Define
15487 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15489 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
15491 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15492 array entries to represent __ieee128 versions of the
15493 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
15494 scalar_extract_sig, and scalar_insert_exp built-in functions.
15495 (altivec_resolve_overloaded_builtin): Add special case handling
15496 for the __builtin_scalar_insert_exp function, as represented by
15497 the P9V_BUILTIN_VEC_VSIEDP constant.
15498 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
15499 exponent support for __ieee128 argument.
15500 (VSESQP): Add scalar extract signature support for __ieee128
15502 (VSTDCNQP): Add scalar test negative support for __ieee128
15504 (VSIEQP): Add scalar insert exponent support for __int128 argument
15505 with __ieee128 result.
15506 (VSIEQPF): Add scalar insert exponent support for __ieee128
15507 argument with __ieee128 result.
15508 (VSTDCQP): Add scalar test data class support for __ieee128
15510 (VSTDCNQP): Add overload support for scalar test negative with
15511 __ieee128 argument.
15512 (VSTDCQP): Add overload support for scalar test data class
15513 __ieee128 argument.
15514 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
15515 UNSPEC_VSX_SXSIGDP.
15516 (UNSPEC_VSX_SIEXPQP): New constant.
15517 (xsxexpqp): New insn for VSX scalar extract exponent quad
15519 (xsxsigqp): New insn for VSX scalar extract significand quad
15521 (xsiexpqpf): New insn for VSX scalar insert exponent quad
15522 precision with floating point argument.
15523 (xststdcqp): New expand for VSX scalar test data class quad
15525 (xststdcnegqp): New expand for VSX scalar test negative quad
15527 (xststdcqp): New insn to match expansions for VSX scalar test data
15528 class quad precision and VSX scalar test negative quad precision.
15529 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
15530 special case operand checking to enforce that second operand of
15531 VSX scalar test data class with quad precision argument is a 7-bit
15533 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15534 prototypes and descriptions of __ieee128 versions of
15535 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
15536 scalar_test_data_class, and scalar_test_neg built-in functions.
15538 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15540 PR tree-optimization/81162
15541 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
15542 replace a negate with an add.
15544 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
15546 * doc/invoke.texi (arm/-mcpu): Document +crypto.
15548 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15550 * config/arm/arm-c.c (arm_cpu_builtins): Define
15551 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15553 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15555 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
15556 (armv8-r): Set ARM Cortex-R52 as default CPU.
15557 * config/arm/arm-tables.opt: Regenerate.
15558 * config/arm/arm-tune.md: Regenerate.
15559 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15561 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
15562 extension for -mcpu=cortex-r52.
15564 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15566 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
15567 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
15568 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
15569 (fp-armv8): Define it as FP_ARMv8 only.
15570 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
15571 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
15573 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
15574 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
15575 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
15576 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
15577 than TARGET_FPU_ARMV8.
15578 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
15579 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
15580 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
15582 * config/arm/neon.md (neon_vrint): Likewise.
15583 (neon_vcvt): Likewise.
15584 (neon_<fmaxmin_op><mode>): Likewise.
15585 (<fmaxmin><mode>3): Likewise.
15586 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
15587 * config/arm/predicates.md (arm_cond_move_operator): Check against
15588 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
15590 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
15592 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
15593 to top of function.
15595 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15597 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
15598 loop in comment with memset.
15600 2017-07-14 Martin Liska <mliska@suse.cz>
15602 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
15603 * dwarf2out.c (is_java): Remove the function.
15604 (output_pubname): Remove usage of the function.
15605 (lower_bound_default): Remove usage of DW_LANG_Java.
15606 (gen_compile_unit_die): Likewise.
15607 * gcc.c: Remove compiler defaults for .java and .zip files.
15608 * gimple-expr.c (remove_suffix): Change as there's no longer
15609 extension than 4-letter one.
15610 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
15611 (gimplify_save_expr): Likewise.
15612 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
15613 as it's possible even for other languages than Java.
15614 * langhooks.h (struct lang_hooks): Remove Java from a comment.
15615 * lto-opts.c (lto_write_options): Remove reference to Java.
15616 * opts.c (strip_off_ending): Update file extension handling.
15617 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
15618 * tree-eh.c (lower_resx): Likewise.
15619 * tree.c (free_lang_data_in_type): Remove dead code.
15620 (find_decls_types_r): Likewise.
15621 (build_common_builtin_nodes): Remove Java from a comment.
15622 (verify_type): Remove dead code.
15623 * varasm.c (assemble_external): Remove Java from a comment.
15625 2017-07-14 Martin Liska <mliska@suse.cz>
15627 * opts.c (finish_options): Add quotes.
15628 (common_handle_option): Likewise.
15630 2017-07-14 Martin Liska <mliska@suse.cz>
15632 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
15633 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
15634 Remove N_SO_PASCAL.
15635 * dwarf2out.c (lower_bound_default): Do not handle
15637 (gen_compile_unit_die): Likewise.
15638 * gcc.c: Remove default extension binding for GNU Pascal.
15639 * stmt.c: Remove Pascal language from a comment.
15640 * xcoffout.c: Likewise.
15642 2017-07-13 David Malcolm <dmalcolm@redhat.com>
15645 * diagnostic-show-locus.c (fixit_cmp): New function.
15646 (layout::layout): Sort m_fixit_hints.
15647 (column_range::column_range): Assert that the values are valid.
15648 (struct char_span): New struct.
15649 (correction::overwrite): New method.
15650 (struct source_line): New struct.
15651 (line_corrections::add_hint): Add assertions. Reimplement memcpy
15652 calls in terms of classes source_line and char_span, and
15653 correction::overwrite.
15654 (selftest::test_overlapped_fixit_printing_2): New function.
15655 (selftest::diagnostic_show_locus_c_tests): Call it.
15657 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
15659 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
15660 early if there is no lhs.
15662 2017-07-13 Martin Liska <mliska@suse.cz>
15664 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
15665 (gen_reference_type_die): Likewise.
15666 * stor-layout.c: Remove Pascal-related comment.
15668 2017-07-13 Martin Liska <mliska@suse.cz>
15670 * opts.c (finish_options): Add quotes to error messages.
15671 (parse_sanitizer_options): Likewise.
15673 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
15675 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
15677 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
15679 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
15681 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
15683 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
15685 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
15687 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
15690 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
15691 provides the hardware capability bits, define the macro
15692 __BUILTIN_CPU_SUPPORTS__.
15693 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
15694 if GLIBC does not provide the hardware capability bits. Add a
15695 gcc_unreachable call if the built-in cpu function is neither
15696 __builtin_cpu_is nor __builtin_cpu_supports.
15697 (rs6000_get_function_versions_dispatcher): Change the warning
15698 that an old GLIBC is used which does not export the capability
15699 bits to be an error.
15700 * doc/extend.texi (target_clones attribute): Document the
15701 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
15702 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
15703 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
15704 the macros defined by GCC if the newer GLIBC is available.
15706 2017-07-12 Jeff Law <law@redhat.com>
15708 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
15709 remaining includes slightly.
15710 * config/riscv/riscv-builtins.c: Include profile-count.h.
15712 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
15715 * config/avr/avr.c (avr_set_current_function): In diagnostic
15716 messages: Quote keywords and (parts of) identifiers.
15717 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
15720 2017-07-12 Carl Love <cel@us.ibm.com>
15722 * config/rs6000/rs6000-c.c: Add support for built-in functions
15723 vector bool char vec_revb (vector bool char);
15724 vector bool short vec_revb (vector short char);
15725 vector bool int vec_revb (vector bool int);
15726 vector bool long long vec_revb (vector bool long long);
15727 * doc/extend.texi: Update the built-in documentation file for the
15728 new built-in functions.
15730 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15732 * config/s390/s390.md: Remove movcc splitter.
15734 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15736 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
15737 load/store on condition.
15739 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
15742 * config/avr/avr.c (avr_encode_section_info)
15743 [progmem && !TREE_READONLY]: Error if progmem object needs
15746 2017-07-11 Michael Collison <michael.collison@arm.com>
15748 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
15751 2017-07-11 Carl Love <cel@us.ibm.com>
15753 * config/rs6000/rs6000-c.c: Add support for builtins
15754 vector unsigned int vec_parity_lsbb (vector signed int);
15755 vector unsigned int vec_parity_lsbb (vector unsigned int);
15756 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
15757 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
15758 vector unsigned long long vec_parity_lsbb (vector signed long long);
15759 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
15760 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
15761 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
15762 * doc/extend.texi: Update the built-in documentation file for the
15763 new built-in functions.
15765 2017-07-11 David Malcolm <dmalcolm@redhat.com>
15767 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
15768 (layout::m_primary_loc): New field.
15769 (layout::layout): Initialize new field. Move location filtering
15770 logic from here to...
15771 (layout::maybe_add_location_range): ...this new method. Add
15772 support for filtering to just the lines already specified by other
15774 (layout::will_show_line_p): New method.
15775 (gcc_rich_location::add_location_if_nearby): New method.
15776 (selftest::test_add_location_if_nearby): New test function.
15777 (selftest::diagnostic_show_locus_c_tests): Call it.
15778 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
15781 2017-07-11 Tom de Vries <tom@codesourcery.com>
15783 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
15784 (bb_first_real_insn): New function.
15785 (nvptx_single): Add extra initialization of broadcasted condition
15788 2017-07-11 Nathan Sidwell <nathan@acm.org>
15790 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
15792 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
15794 * doc/extend.texi (AVR Function Attributes): Remove weblink to
15795 Binutils doc as TEXI will mess them up.
15796 * doc/invoke.texi (AVR Options): Same here.
15798 2017-07-11 Daniel Cederman <cederman@gaisler.com>
15800 * config/sparc/sparc.opt (mfix-ut700): New option.
15801 (mfix-gr712rc): Likewise.
15802 (sparc_fix_b2bst): New variable.
15803 * doc/invoke.texi (SPARC options): Document them.
15804 (ARM options): Fix warnings.
15805 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
15806 instructions to prevent sequences that can trigger the store-store
15807 errata for certain LEON3FT processors.
15808 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
15809 (sparc_option_override): Set sparc_fix_b2bst appropriately.
15810 * config/sparc/sparc.md (fix_b2bst): New attribute.
15811 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
15813 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
15816 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
15818 (*rsqrtsf2_sse): Ditto.
15820 (div<mode>3): Macroize insn from divdf3 and divsf3
15821 using MODEF mode iterator.
15823 2017-07-10 Martin Sebor <msebor@redhat.com>
15825 PR tree-optimization/80397
15826 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
15827 instead of testing for equality to INTEGER_TYPE.
15829 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
15831 * config.gcc: Remove uclibc from arc target spec.
15833 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
15835 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
15837 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15840 * lto-wrapper.c (remove_option): New function.
15841 (merge_and_complain): Merge PIC/PIE options more realistically.
15843 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
15845 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
15849 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
15850 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
15851 * config.in: Regenerate.
15852 * configure: Regenerate.
15853 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
15854 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
15855 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
15856 (TARGET_GASISR_PROLOGUES): ...target mask.
15857 * common/config/avr/avr-common.c
15858 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
15859 Set -mgas-isr-prologues.
15860 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
15861 INSERT_PASS_BEFORE for it.
15862 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
15863 * config/avr/avr.c (avr_option_override)
15864 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
15865 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
15866 (avr_attribute_table) <no_gccisr>: Add new function attribute.
15867 (avr_set_current_function) <is_no_gccisr>: Init machine field.
15868 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
15870 (make_avr_pass_pre_proep): New function.
15871 (emit_push_sfr) <treg>: Add argument to function and use it
15872 instead of TMP_REG.
15873 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
15874 and set machine->gasisr.yes.
15875 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
15876 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
15877 __gcc_isr.n_pushed to .L__stack_usage.
15878 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
15879 (avr_asm_final_postscan_insn): ...this new static function.
15880 * config/avr/avr.h (machine_function)
15881 <is_no_gccisr, use_L__stack_usage>: New fields.
15882 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
15883 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
15884 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
15885 (gasisr, *gasisr): New expander and insn.
15886 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15887 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
15888 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
15890 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
15892 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
15895 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
15897 Move jump-tables out of .text again.
15900 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
15901 (ASM_OUTPUT_ADDR_VEC): New function.
15902 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
15903 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
15904 INSN_ADDRESSes as asm comment.
15905 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
15906 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
15907 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
15908 * config/avr/avr.md (*tablejump): Adjust comment.
15909 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
15910 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
15912 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
15913 (avr_output_addr_vec): New proto.
15914 (avr_log_t) <insn_addresses>: New field.
15916 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
15919 * config/i386/i386.c (ix86_function_arg_advance): Set
15920 outgoing_args_on_stack to true if there are outgoing arguments
15922 (ix86_function_arg): Likewise.
15923 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
15924 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
15925 * config/i386/i386.h (machine_function): Add
15926 outgoing_args_on_stack.
15928 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
15930 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15931 supporting pthreds.
15932 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
15934 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
15936 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
15937 (REAL_H): Remove $(MACHMODE_H).
15938 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
15940 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
15941 $(MACHMODE_H) and double-int.h.
15942 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
15944 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
15947 2017-07-07 Andrew Pinski <apinski@cavium.com>
15949 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
15950 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
15952 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15954 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
15955 Add warning if GCC was not configured to link against a GLIBC that
15956 exports the hardware capability bits.
15957 (make_resolver_func): Make resolver function private and not a
15958 COMDAT function. Create the name with clone_function_name instead
15959 of make_unique_name.
15962 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
15963 correct operand in doing the split.
15965 2017-07-07 Carl Love <cel@us.ibm.com>
15967 * config/rs6000/rs6000-c: Add support for built-in function
15968 vector unsigned short vec_pack_to_short_fp32 (vector float,
15970 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
15971 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
15972 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
15973 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
15974 (convert_4f32_8i16): Add define_expand.
15975 * doc/extend.texi: Update the built-in documentation file for the
15976 new built-in function.
15978 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15980 * config/sparc/m8.md: New file.
15981 * config/sparc/sparc.md: Include m8.md.
15983 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15985 * config/sparc/sparc.opt: New option -mvis4b.
15986 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
15987 (sparc_option_override): Handle VIS4B.
15988 (enum sparc_builtins): Define
15989 SPARC_BUILTIN_DICTUNPACK{8,16,32},
15990 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
15991 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
15992 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
15993 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
15994 (check_constant_argument): New function.
15995 (sparc_vis_init_builtins): Define builtins
15996 __builtin_vis_dictunpack{8,16,32},
15997 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
15998 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
15999 __builtin_vis_fpcmpde{8,16,32}shl and
16000 __builtin_vis_fpcmpur{8,16,32}shl.
16001 (sparc_expand_builtin): Check that the constant operands to
16002 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16003 constant and in range.
16004 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16006 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16007 (SPARC_IMM5_P): Likewise.
16008 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16009 (enabled): Handle vis4b.
16010 (UNSPEC_DICTUNPACK): New unspec.
16011 (UNSPEC_FPCMPSHL): Likewise.
16012 (UNSPEC_FPUCMPSHL): Likewise.
16013 (UNSPEC_FPCMPDESHL): Likewise.
16014 (UNSPEC_FPCMPURSHL): Likewise.
16015 (cpu_feature): New CPU feature `vis4b'.
16016 (dictunpack{8,16,32}): New insns.
16017 (FPCSMODE): New mode iterator.
16018 (fpcscond): New code iterator.
16019 (fpcsucond): Likewise.
16020 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16021 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16022 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16023 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16024 * config/sparc/constraints.md: Define constraints `q' for unsigned
16025 2-bit integer constants and `t' for unsigned 5-bit integer
16027 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16029 (imm5_operand_dictunpack16): Likewise.
16030 (imm5_operand_dictunpack32): Likewise.
16031 (imm2_operand): Likewise.
16032 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16033 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16034 ditunpack* and fpcmp*shl builtins.
16036 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16038 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16039 * config.in: Add HAVE_AS_SPARC6 define.
16040 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16042 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16044 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16045 (CPP_CPU_SPEC): Handle m8.
16046 (ASM_CPU_SPEC): Likewise.
16047 * config/sparc/sparc-opts.h (enum processor_type): Add
16049 * config/sparc/sparc.c (m8_costs): New struct.
16050 (sparc_option_override): Handle TARGET_CPU_m8.
16051 (sparc32_initialize_trampoline): Likewise.
16052 (sparc64_initialize_trampoline): Likewise.
16053 (sparc_issue_rate): Likewise.
16054 (sparc_register_move_cost): Likewise.
16055 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16056 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16057 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16058 (CPP_CPU_SPEC): Handle M8.
16059 (ASM_CPU_SPEC): Likewise.
16060 (AS_M8_FLAG): Define.
16061 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16062 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16063 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16065 * configure: Regenerate.
16066 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16069 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16071 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16073 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16074 ("*movdi_insn_sp32"): Do not set v3pipe.
16075 ("*movsi_insn"): Likewise.
16076 ("*movdi_insn_sp64"): Likewise.
16077 ("*movsf_insn"): Likewise.
16078 ("*movdf_insn_sp32"): Likewise.
16079 ("*movdf_insn_sp64"): Likewise.
16080 ("*zero_extendsidi2_insn_sp64"): Likewise.
16081 ("*sign_extendsidi2_insn"): Likewise.
16082 ("*mov<VM32:mode>_insn"): Likewise.
16083 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16084 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16085 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16086 ("<vlop:code><VL:mode>3"): Likewise.
16087 ("*not_<vlop:code><VL:mode>3"): Likewise.
16088 ("*nand<VL:mode>_vis"): Likewise.
16089 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16090 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16091 ("one_cmpl<VL:mode>2"): Likewise.
16092 ("faligndata<VM64:mode>_vis"): Likewise.
16093 ("alignaddrsi_vis"): Likewise.
16094 ("alignaddrdi_vis"): Likweise.
16095 ("alignaddrlsi_vis"): Likewise.
16096 ("alignaddrldi_vis"): Likewise.
16097 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16098 ("bmaskdi_vis"): Likewise.
16099 ("bmasksi_vis"): Likewise.
16100 ("bshuffle<VM64:mode>_vis"): Likewise.
16101 ("cmask8<P:mode>_vis"): Likewise.
16102 ("cmask16<P:mode>_vis"): Likewise.
16103 ("cmask32<P:mode>_vis"): Likewise.
16104 ("pdistn<P:mode>_vis"): Likewise.
16105 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16107 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16109 * config/sparc/sparc.md ("subtype"): New insn attribute.
16110 ("*wrgsr_sp64"): Set insn subtype.
16111 ("*rdgsr_sp64"): Likewise.
16112 ("alignaddrsi_vis"): Likewise.
16113 ("alignaddrdi_vis"): Likewise.
16114 ("alignaddrlsi_vis"): Likewise.
16115 ("alignaddrldi_vis"): Likewise.
16116 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16117 ("fexpand_vis"): Likewise.
16118 ("fpmerge_vis"): Likewise.
16119 ("faligndata<VM64:mode>_vis"): Likewise.
16120 ("bshuffle<VM64:mode>_vis"): Likewise.
16121 ("cmask8<P:mode>_vis"): Likewise.
16122 ("cmask16<P:mode>_vis"): Likewise.
16123 ("cmask32<P:mode>_vis"): Likewise.
16124 ("fchksm16_vis"): Likewise.
16125 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16126 ("fmean16_vis"): Likewise.
16127 ("fp<plusminus_insn>64_vis"): Likewise.
16128 ("<plusminus_insn>v8qi3"): Likewise.
16129 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16130 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16131 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16132 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16133 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16134 ("*movqi_insn"): Likewise.
16135 ("*movhi_insn"): Likewise.
16136 ("*movsi_insn"): Likewise.
16137 ("movsi_pic_gotdata_op"): Likewise.
16138 ("*movdi_insn_sp32"): Likewise.
16139 ("*movdi_insn_sp64"): Likewise.
16140 ("movdi_pic_gotdata_op"): Likewise.
16141 ("*movsf_insn"): Likewise.
16142 ("*movdf_insn_sp32"): Likewise.
16143 ("*movdf_insn_sp64"): Likewise.
16144 ("*zero_extendhisi2_insn"): Likewise.
16145 ("*zero_extendqihi2_insn"): Likewise.
16146 ("*zero_extendqisi2_insn"): Likewise.
16147 ("*zero_extendqidi2_insn"): Likewise.
16148 ("*zero_extendhidi2_insn"): Likewise.
16149 ("*zero_extendsidi2_insn_sp64"): Likewise.
16150 ("ldfsr"): Likewise.
16151 ("prefetch_64"): Likewise.
16152 ("prefetch_32"): Likewise.
16153 ("tie_ld32"): Likewise.
16154 ("tie_ld64"): Likewise.
16155 ("*tldo_ldub_sp32"): Likewise.
16156 ("*tldo_ldub1_sp32"): Likewise.
16157 ("*tldo_ldub2_sp32"): Likewise.
16158 ("*tldo_ldub_sp64"): Likewise.
16159 ("*tldo_ldub1_sp64"): Likewise.
16160 ("*tldo_ldub2_sp64"): Likewise.
16161 ("*tldo_ldub3_sp64"): Likewise.
16162 ("*tldo_lduh_sp32"): Likewise.
16163 ("*tldo_lduh1_sp32"): Likewise.
16164 ("*tldo_lduh_sp64"): Likewise.
16165 ("*tldo_lduh1_sp64"): Likewise.
16166 ("*tldo_lduh2_sp64"): Likewise.
16167 ("*tldo_lduw_sp32"): Likewise.
16168 ("*tldo_lduw_sp64"): Likewise.
16169 ("*tldo_lduw1_sp64"): Likewise.
16170 ("*tldo_ldx_sp64"): Likewise.
16171 ("*mov<VM32:mode>_insn"): Likewise.
16172 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16173 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16175 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16177 * config/sparc/sparc.md ("type"): New insn type viscmp.
16178 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16180 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16181 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16182 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16183 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16185 ("n7_vis_logical_11cycle"): Likewise.
16186 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16187 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16188 * config/sparc/niagara.md ("niag_vis"): Likewise.
16189 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16190 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16192 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16194 * config/sparc/sparc.md: New instruction type `bmask'.
16195 (bmaskdi_vis): Use the `bmask' type.
16196 (bmasksi_vis): Likewise.
16197 * config/sparc/ultra3.md (us3_array): Likewise.
16198 * config/sparc/niagara7.md (n7_array): Likewise.
16199 * config/sparc/niagara4.md (n4_array): Likewise.
16200 * config/sparc/niagara2.md (niag2_vis): Likewise.
16201 (niag3_vis): Likewise.
16202 * config/sparc/niagara.md (niag_vis): Likewise.
16204 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16206 * ipa-comdats.c: Remove optimize check from gate.
16207 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16208 for functions not optimized.
16209 (ipa_fn_summary_read): Skip optimize check.
16210 (ipa_fn_summary_write): Likewise.
16211 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16213 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16215 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16216 (check_callers): Give up on caller which is not optimized.
16217 (inline_small_functions): Likewise.
16218 (ipa_inline): Do not give up when not optimizing.
16219 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16220 away unoptimizes cdtors.
16221 (whole_program_function_and_variable_visibility): Do
16222 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16223 * ipa.c (process_references): Do not check optimize.
16224 (symbol_table::remove_unreachable_nodes): Update optimize check.
16225 (set_writeonly_bit): Update optimize check.
16226 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16227 (pass_ipa_single_use::gate): Remove.
16229 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16231 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16232 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16233 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16234 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16235 permute_load, permute_store, adjust_extract, adjust_splat,
16236 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16237 replace_swap_with_copy, dump_swap_insn_table,
16238 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16239 recombine_lvx_pattern, recombine_stvx_pattern,
16240 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16241 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16242 to file rs6000-p8swap.c.
16243 * config/rs6000/rs6000-p8swap.c: New file.
16244 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16245 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16246 and rs6000*-*-* targets.
16248 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16250 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
16252 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16254 * lto-wrapper.c (merge_and_complain): Do not merge
16255 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
16256 fsigned_zeros, ftrapping_math, fwrapv.
16257 (append_compiler_options): Do not track these options.
16258 (append_linker_options): Likewie
16260 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16262 * cgraphunit.c (cgraph_node::finalize_function): When
16263 !flag_toplevel_reorde set no_reorder flag.
16264 (varpool_node::finalize_decl): Likewise.
16265 (symbol_table::compile): Drop no toplevel reorder path.
16267 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16269 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
16270 edges; zero probability is not better than uninitialized.
16272 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16274 * asan.h (asan_sanitize_allocas_p): Declare.
16275 * asan.c (asan_sanitize_allocas_p): New function.
16276 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
16277 (handle_builtin_alloca): Likewise.
16278 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
16279 if !asan_sanitize_allocas_p.
16280 * params.def (asan-instrument-allocas): Add new option.
16281 * params.h (ASAN_PROTECT_ALLOCAS): Define.
16282 * opts.c (common_handle_option): Disable allocas sanitization for
16285 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16287 * asan.c: Include gimple-fold.h.
16288 (get_last_alloca_addr): New function.
16289 (handle_builtin_stackrestore): Likewise.
16290 (handle_builtin_alloca): Likewise.
16291 (asan_emit_allocas_unpoison): Likewise.
16292 (get_mem_refs_of_builtin_call): Add new parameter, remove const
16293 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
16294 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
16295 (instrument_builtin_call): Pass gimple iterator to
16296 get_mem_refs_of_builtin_call.
16297 (last_alloca_addr): New global.
16298 * asan.h (asan_emit_allocas_unpoison): Declare.
16299 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
16300 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
16301 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
16302 if function calls alloca.
16303 * gimple-fold.c (replace_call_with_value): Remove static keyword.
16304 * gimple-fold.h (replace_call_with_value): Declare.
16305 * internal-fn.c: Include asan.h.
16306 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
16307 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
16309 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16311 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
16312 (C_SELFTEST_FLAGS): New.
16313 (CPP_SELFTEST_FLAGS): New.
16314 (SELFTEST_DEPS): New, from deps of s-selftest.
16315 (C_SELFTEST_DEPS): New, from deps of s-selftest.
16316 (CPP_SELFTEST_DEPS): New.
16317 (selftest): Add dependency on s-selftest-c++.
16318 (s-selftest): Rename to...
16319 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
16320 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
16321 than SELFTEST_FLAGS.
16322 (selftest-gdb): Rename to...
16323 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
16325 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
16326 (selftest-valgrind): Rename to...
16327 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
16329 (selftest-valgrind): Reintroduce as an alias for
16330 selftest-c-valgrind.
16331 (s-selftest-c++): New.
16332 (selftest-c++-gdb): New.
16333 (selftest-c++-valgrind): New.
16335 2017-07-06 Olivier Hainque <hainque@adacore.com>
16337 * gcc.c (process_command): When deciding if undefined variables
16338 should be ignored when processing specs, accept "gcc -v" as well.
16340 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16342 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
16343 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
16345 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16347 * config/arm/arm-cpus.in (armv8-r): Add new entry.
16348 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
16349 * config/arm/arm-tables.opt: Regenerate.
16350 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
16352 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
16354 2017-07-06 Carl Love <cel@us.ibm.com>
16356 * ChangeLog: Clean up from mid air collision
16358 2017-07-06 Carl Love <cel@us.ibm.com>
16360 * config/rs6000/rs6000-c.c: Add support for built-in functions
16361 vector signed int vec_subc (vector signed int, vector signed int);
16362 vector signed __int128 vec_subc (vector signed __int128,
16363 vector signed __int128);
16364 vector unsigned __int128 vec_subc (vector unsigned __int128,
16365 vector unsigned __int128);
16366 vector signed int vec_sube (vector signed int, vector signed int,
16367 vector signed int);
16368 vector unsigned int vec_sube (vector unsigned int,
16369 vector unsigned int,
16370 vector unsigned int);
16371 vector signed __int128 vec_sube (vector signed __int128,
16372 vector signed __int128,
16373 vector signed__int128);
16374 vector unsigned __int128 vec_sube (vector unsigned __int128,
16375 vector unsigned __int128,
16376 vector unsigned __int128);
16377 vector signed int vec_subec (vector signed int, vector signed int,
16378 vector signed int);
16379 vector unsigned int vec_subec (vector unsigned int,
16380 vector unsigned int,
16381 vector unsigned int);
16382 vector signed __int128 vec_subec (vector signed __int128,
16383 vector signed __int128,
16384 vector signed__int128);
16385 vector unsigned __int128 vec_subec (vector unsigned __int128,
16386 vector unsigned __int128,
16387 vector unsigned __int128);
16388 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
16389 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
16390 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
16391 BU_ALTIVEC_OVERLOAD_X definitions.
16392 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
16393 * doc/extend.texi: Update the built-in documentation file for the new
16394 built-in functions.
16396 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16399 * diagnostic-show-locus.c (layout::layout): Use start and finish
16400 spelling location for the start and finish of each range.
16401 * genmatch.c (linemap_client_expand_location_to_spelling_point):
16402 Add unused aspect param.
16403 * input.c (expand_location_1): Add "aspect" param, and use it
16404 to access the correct part of the location.
16405 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
16407 (expand_location_to_spelling_point): Likewise.
16408 (linemap_client_expand_location_to_spelling_point): Add "aspect"
16409 param, and pass it to expand_location_1.
16411 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
16413 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
16414 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
16415 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
16416 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
16417 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
16418 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
16419 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
16420 _mm_maskz_getmant_ss): New intrinsics.
16421 (__builtin_ia32_getexpss128_mask): Changed to ...
16422 __builtin_ia32_getexpss128_round ... this.
16423 (__builtin_ia32_getexpsd128_mask): Changed to ...
16424 __builtin_ia32_getexpsd128_round ... this.
16425 * config/i386/i386-builtin-types.def
16426 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
16427 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
16428 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
16429 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
16430 __builtin_ia32_getmantss_mask_round): New builtins.
16431 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
16432 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
16433 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
16434 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
16435 * config/i386/sse.md
16436 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
16437 avx512f_sgetexp<mode><mask_scalar_name>
16438 <round_saeonly_scalar_name> ... this.
16439 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
16440 %0, %1, %2<round_saeonly_op3>}): Changed to ...
16441 vgetexp<ssescalarmodesuffix>
16442 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16443 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
16444 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
16445 avx512f_vgetmant<mode><mask_scalar_name>
16446 <round_saeonly_scalar_name> ... this.
16447 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
16448 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
16449 vgetmant<ssescalarmodesuffix>
16450 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
16451 %0<mask_scalar_operand4>, %1, %2
16452 <round_saeonly_scalar_mask_op4>, %3} ... this.
16453 * config/i386/subst.md (mask_scalar_operand4,
16454 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
16455 round_saeonly_scalar_nimm_predicate): New subst attributes.
16457 2017-07-06 Julia Koval <julia.koval@intel.com>
16459 * config/i386/i386.c (ix86_erase_embedded_rounding):
16460 Remove code for old rounding pattern.
16462 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
16464 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
16466 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
16468 * doc/sourcebuild.texi (Test Directives, Variants of
16469 dg-require-support): Add documentation for dg-require-stack-check.
16471 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
16473 * config/i386/subst.md (mask_scalar, round_scalar,
16474 round_saeonly_scalar): New meta-templates.
16475 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
16476 round_scalar_mask_operand3, round_scalar_mask_op3,
16477 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
16478 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
16479 round_saeonly_scalar_constraint,
16480 round_saeonly_scalar_prefix): New subst attribute.
16481 * config/i386/sse.md
16482 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
16483 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
16484 <round_scalar_name> ... this.
16485 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
16486 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
16487 <round_scalar_name> ... this.
16488 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
16489 <sse>_vm<code><mode>3<mask_scalar_name>
16490 <round_saeonly_scalar_name> ... this.
16491 (v<plusminus_mnemonic><ssescalarmodesuffix>
16492 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16493 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16494 v<plusminus_mnemonic><ssescalarmodesuffix>
16495 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16496 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16497 (v<multdiv_mnemonic><ssescalarmodesuffix>
16498 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16499 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16500 v<multdiv_mnemonic><ssescalarmodesuffix>
16501 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16502 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16503 (v<maxmin_float><ssescalarmodesuffix>
16504 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
16505 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
16506 v<maxmin_float><ssescalarmodesuffix>
16507 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16508 %0<mask_scalar_operand3>, %1, %<iptr>2
16509 <round_saeonly_scalar_mask_op3>} ... this.
16511 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
16513 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
16514 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
16516 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
16517 Alan Hayward <alan.hayward@arm.com>
16518 David Sherwood <david.sherwood@arm.com>
16520 * combine.c (simplify_if_then_else): Remove "enum" before
16522 * compare-elim.c (can_eliminate_compare): Likewise.
16523 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16525 (aarch64_lookup_simd_builtin_type): Likewise.
16526 (aarch64_simd_builtin_type): Likewise.
16527 (aarch64_init_simd_builtin_types): Likewise.
16528 (aarch64_simd_expand_args): Likewise.
16529 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
16531 (aarch64_reverse_mask): Likewise.
16532 (aarch64_simd_emit_reg_reg_move): Likewise.
16533 (aarch64_gen_adjusted_ldpstp): Likewise.
16534 (aarch64_ccmp_mode_to_code): Likewise.
16535 (aarch64_operands_ok_for_ldpstp): Likewise.
16536 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16537 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
16539 (aarch64_min_divisions_for_recip_mul): Likewise.
16540 (aarch64_reassociation_width): Likewise.
16541 (aarch64_get_condition_code_1): Likewise.
16542 (aarch64_simd_emit_reg_reg_move): Likewise.
16543 (aarch64_simd_attr_length_rglist): Likewise.
16544 (aarch64_reverse_mask): Likewise.
16545 (aarch64_operands_ok_for_ldpstp): Likewise.
16546 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16547 (aarch64_gen_adjusted_ldpstp): Likewise.
16548 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
16550 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
16551 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16552 (arm_lookup_simd_builtin_type): Likewise.
16553 (arm_simd_builtin_type): Likewise.
16554 (arm_init_simd_builtin_types): Likewise.
16555 (arm_expand_builtin_args): Likewise.
16556 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
16557 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
16558 (ft32_setup_incoming_varargs): Likewise.
16559 (ft32_function_arg): Likewise.
16560 (ft32_function_arg_advance): Likewise.
16561 (ft32_pass_by_reference): Likewise.
16562 (ft32_arg_partial_bytes): Likewise.
16563 (ft32_valid_pointer_mode): Likewise.
16564 (ft32_addr_space_pointer_mode): Likewise.
16565 (ft32_addr_space_legitimate_address_p): Likewise.
16566 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
16568 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
16569 (ix86_emit_outlined_ms2sysv_restore): Likewise.
16570 (iamcu_alignment): Likewise.
16571 (canonicalize_vector_int_perm): Likewise.
16572 (ix86_noce_conversion_profitable_p): Likewise.
16573 (ix86_mpx_bound_mode): Likewise.
16574 (ix86_operands_ok_for_move_multiple): Likewise.
16575 * config/microblaze/microblaze-protos.h
16576 (microblaze_expand_conditional_branch_reg): Likewise.
16577 * config/microblaze/microblaze.c
16578 (microblaze_expand_conditional_branch_reg): Likewise.
16579 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16581 (rs6000_reassociation_width): Likewise.
16582 (rs6000_invalid_binary_op): Likewise.
16583 (fusion_p9_p): Likewise.
16584 (emit_fusion_p9_load): Likewise.
16585 (emit_fusion_p9_store): Likewise.
16586 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
16588 (riscv_hard_regno_mode_ok_p): Likewise.
16589 (riscv_address_insns): Likewise.
16590 (riscv_split_symbol): Likewise.
16591 (riscv_legitimize_move): Likewise.
16592 (riscv_function_value): Likewise.
16593 (riscv_hard_regno_nregs): Likewise.
16594 (riscv_expand_builtin): Likewise.
16595 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
16596 (riscv_build_integer): Likewise.
16597 (riscv_split_integer): Likewise.
16598 (riscv_legitimate_constant_p): Likewise.
16599 (riscv_cannot_force_const_mem): Likewise.
16600 (riscv_regno_mode_ok_for_base_p): Likewise.
16601 (riscv_valid_base_register_p): Likewise.
16602 (riscv_valid_offset_p): Likewise.
16603 (riscv_valid_lo_sum_p): Likewise.
16604 (riscv_classify_address): Likewise.
16605 (riscv_legitimate_address_p): Likewise.
16606 (riscv_address_insns): Likewise.
16607 (riscv_load_store_insns): Likewise.
16608 (riscv_force_binary): Likewise.
16609 (riscv_split_symbol): Likewise.
16610 (riscv_force_address): Likewise.
16611 (riscv_legitimize_address): Likewise.
16612 (riscv_move_integer): Likewise.
16613 (riscv_legitimize_const_move): Likewise.
16614 (riscv_legitimize_move): Likewise.
16615 (riscv_address_cost): Likewise.
16616 (riscv_subword): Likewise.
16617 (riscv_output_move): Likewise.
16618 (riscv_canonicalize_int_order_test): Likewise.
16619 (riscv_emit_int_order_test): Likewise.
16620 (riscv_function_arg_boundary): Likewise.
16621 (riscv_pass_mode_in_fpr_p): Likewise.
16622 (riscv_pass_fpr_single): Likewise.
16623 (riscv_pass_fpr_pair): Likewise.
16624 (riscv_get_arg_info): Likewise.
16625 (riscv_function_arg): Likewise.
16626 (riscv_function_arg_advance): Likewise.
16627 (riscv_arg_partial_bytes): Likewise.
16628 (riscv_function_value): Likewise.
16629 (riscv_pass_by_reference): Likewise.
16630 (riscv_setup_incoming_varargs): Likewise.
16631 (riscv_print_operand): Likewise.
16632 (riscv_elf_select_rtx_section): Likewise.
16633 (riscv_save_restore_reg): Likewise.
16634 (riscv_for_each_saved_reg): Likewise.
16635 (riscv_register_move_cost): Likewise.
16636 (riscv_hard_regno_mode_ok_p): Likewise.
16637 (riscv_hard_regno_nregs): Likewise.
16638 (riscv_class_max_nregs): Likewise.
16639 (riscv_memory_move_cost): Likewise.
16640 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
16641 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
16642 (rl78_addr_space_address_mode): Likewise.
16643 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16645 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16646 (rs6000_reassociation_width): Likewise.
16647 (rs6000_invalid_binary_op): Likewise.
16648 (fusion_p9_p): Likewise.
16649 (emit_fusion_p9_load): Likewise.
16650 (emit_fusion_p9_store): Likewise.
16651 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
16652 (ok_for_simple_move_operands): Likewise.
16653 (ok_for_simple_move_strict_operands): Likewise.
16654 (ok_for_simple_arith_logic_operands): Likewise.
16655 (visium_legitimize_reload_address): Likewise.
16656 (visium_select_cc_mode): Likewise.
16657 (output_cbranch): Likewise.
16658 (visium_split_double_move): Likewise.
16659 (visium_expand_copysign): Likewise.
16660 (visium_expand_int_cstore): Likewise.
16661 (visium_expand_fp_cstore): Likewise.
16662 * config/visium/visium.c (visium_pass_by_reference): Likewise.
16663 (visium_function_arg): Likewise.
16664 (visium_function_arg_advance): Likewise.
16665 (visium_libcall_value): Likewise.
16666 (visium_setup_incoming_varargs): Likewise.
16667 (visium_legitimate_constant_p): Likewise.
16668 (visium_legitimate_address_p): Likewise.
16669 (visium_legitimize_address): Likewise.
16670 (visium_secondary_reload): Likewise.
16671 (visium_register_move_cost): Likewise.
16672 (visium_memory_move_cost): Likewise.
16673 (prepare_move_operands): Likewise.
16674 (ok_for_simple_move_operands): Likewise.
16675 (ok_for_simple_move_strict_operands): Likewise.
16676 (ok_for_simple_arith_logic_operands): Likewise.
16677 (visium_function_value_1): Likewise.
16678 (rtx_ok_for_offset_p): Likewise.
16679 (visium_legitimize_reload_address): Likewise.
16680 (visium_split_double_move): Likewise.
16681 (visium_expand_copysign): Likewise.
16682 (visium_expand_int_cstore): Likewise.
16683 (visium_expand_fp_cstore): Likewise.
16684 (visium_split_cstore): Likewise.
16685 (visium_select_cc_mode): Likewise.
16686 (visium_split_cbranch): Likewise.
16687 (output_cbranch): Likewise.
16688 (visium_print_operand_address): Likewise.
16689 * expmed.c (flip_storage_order): Likewise.
16690 * expmed.h (emit_cstore): Likewise.
16691 (flip_storage_order): Likewise.
16692 * genrecog.c (validate_pattern): Likewise.
16693 * hsa-gen.c (gen_hsa_addr): Likewise.
16694 * internal-fn.c (expand_arith_overflow): Likewise.
16695 * ira-color.c (allocno_copy_cost_saving): Likewise.
16696 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16697 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
16698 (process_invariant_for_inheritance): Likewise.
16699 * lra-eliminations.c (move_plus_up): Likewise.
16700 * omp-low.c (lower_oacc_reductions): Likewise.
16701 * simplify-rtx.c (simplify_subreg): Likewise.
16702 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
16703 (TARGET_CHKP_BOUND_MODE): Likewise..
16704 * targhooks.c (default_chkp_bound_mode): Likewise.
16705 (default_setup_incoming_vararg_bounds): Likewise.
16706 * targhooks.h (default_chkp_bound_mode): Likewise.
16707 (default_setup_incoming_vararg_bounds): Likewise.
16708 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
16709 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
16710 (have_whole_vector_shift): Likewise.
16711 * tree-vect-stmts.c (vectorizable_load): Likewise.
16712 * doc/tm.texi: Regenerate.
16714 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
16716 Graceful degrade if Binutils PR21472 is not available.
16719 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
16720 .rodata in flash test fails.
16721 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
16722 * confgure: Regenerate.
16723 * config.in: Regenerate.
16724 * config/avr/avr.c (avr_asm_named_section)
16725 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
16726 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
16727 (avr_asm_init_sections): Same.
16729 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16731 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
16732 (fma<VH:mode>4_intrinsic): Likewise.
16733 (*fmsub<VCVTF:mode>4): Likewise.
16734 (*fmsub<VH:mode>4_intrinsic): Likewise.
16736 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
16739 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
16740 Don't depend on "optimize > 0".
16741 (out_movhi_r_mr, out_movqi_mr_r): Same.
16742 (out_movhi_mr_r, out_movqi_r_mr): Same.
16743 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
16744 io_address_operand on "optimize > 0".
16746 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16748 * tree-loop-distribution.c: Add general explanantion on the pass.
16749 (generate_loops_for_partition): Mark distributed loop.
16750 (pg_add_dependence_edges): New parameter. Handle alias data
16751 dependence specially and record it in the parameter if asked.
16752 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
16753 (init_partition_graph_vertices, add_partition_graph_edge): New.
16754 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
16755 (free_partition_graph_vdata, build_partition_graph): New.
16756 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
16757 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
16758 (data_ref_segment_size, latch_dominated_by_data_ref): New.
16759 (compute_alias_check_pairs, version_loop_by_alias_check): New.
16760 (version_for_distribution_p, finalize_partitions): New.
16761 (distribute_loop): Handle alias data dependence specially. Factor
16762 out loop fusion code as functions and call these functions.
16764 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16766 * tree-loop-distribution.c (classify_partition): New parameter and
16767 better handle reduction statement.
16768 (rdg_build_partitions): Revise comment.
16769 (distribute_loop): Compute statements in all partitions and pass it
16770 to classify_partition.
16772 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16774 * tree-loop-distribution.c (enum partition_type): New.
16775 (struct partition): New field type.
16776 (partition_merge_into): Add parameter. Update partition type.
16777 (data_dep_in_cycle_p, update_type_for_merge): New functions.
16778 (build_rdg_partition_for_vertex): Compute partition type.
16779 (rdg_build_partitions): Dump partition type.
16780 (distribute_loop): Update calls to partition_merge_into.
16782 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16784 * tree-loop-distribution.c (struct ddr_hasher): New.
16785 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
16787 (classify_partition): Call get_data_dependence.
16788 (pg_add_dependence_edges): Ditto.
16789 (distribute_loop): Release data dependence hash table.
16791 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16793 * tree-loop-distribution.c (ref_base_address): Delete.
16794 (similar_memory_accesses): Rename ...
16795 (share_memory_accesses): ... to this. Check if partitions access
16796 the same memory reference.
16797 (distribute_loop): Call share_memory_accesses.
16799 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16801 * tree-loop-distribution.c (struct partition): New field recording
16802 its data reference.
16803 (partition_alloc, partition_free): Init and release data refs.
16804 (partition_merge_into): Merge data refs.
16805 (build_rdg_partition_for_vertex): Collect data refs for partition.
16806 (pg_add_dependence_edges): Change parameters from vector to bitmap.
16808 (distribute_loop): Remve data refs from vertice data of partition
16811 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16813 * tree-loop-distribution.c (params.h): Include header file.
16814 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
16815 (datarefs_vec): New global var.
16816 (create_rdg_vertices): Use datarefs_vec directly.
16817 (free_rdg): Don't free data references.
16818 (build_rdg): Update use. Don't free data references.
16819 (distribute_loop): Compute global variable for data references.
16820 Bail out if there are too many data references.
16822 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16824 * tree-loop-distribution.c (loop_nest): New global var.
16825 (build_rdg): Use loop directly, rather than loop nest.
16826 (pg_add_dependence_edges): Remove loop nest parameter. Use global
16828 (distribute_loop): Compute global variable loop nest. Update use.
16830 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16832 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
16833 (partition_merge_into): New parameter. Dump reason for fusion.
16834 (distribute_loop): Update use of partition_merge_into.
16836 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16838 * tree-loop-distribution.c (bb_top_order_index): New.
16839 (bb_top_order_index_size, bb_top_order_cmp): New.
16840 (stmts_from_loop): Use topological order.
16841 (pass_loop_distribution::execute): Compute and release topological
16842 order for basic blocks.
16844 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16846 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
16849 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16851 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
16852 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
16853 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
16854 * internal-fn.def (LOOP_DIST_ALIAS): New.
16855 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
16856 (fold_loop_internal_call): ... this.
16857 (vect_loop_dist_alias_call): New function.
16858 (set_uid_loop_bbs): Call fold_loop_internal_call.
16859 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
16862 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
16865 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
16866 Require dead FLAGS_REG at the beginning of a peephole.
16868 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
16871 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
16872 arguments in the call to __builtin_ia32_sbb_u32.
16873 (_subborrow_u64): Swap _X and _Y arguments in the call to
16874 __builtin_ia32_sbb_u64.
16876 2017-07-04 Jakub Jelinek <jakub@redhat.com>
16879 * tree-vrp.c (compare_assert_loc): Turn into a function template
16880 with stable template parameter. Only test if a->e is NULL,
16881 !a->e == !b->e has been verified already. Use e == NULL or
16882 e != NULL instead of e or ! e tests. If stable is true, don't use
16883 iterative_hash_expr, on the other side allow a or b or both NULL
16884 and sort the NULLs last.
16885 (process_assert_insertions): Sort using compare_assert_loc<false>
16886 instead of compare_assert_loc, later sort using
16887 compare_assert_loc<true> before calling process_assert_insertions_for
16888 in a loop. Use break instead of continue once seen NULL pointer.
16890 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16892 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16893 Cortex-R7 and Cortex-R8 processors.
16895 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16897 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
16898 uninitialized while src is not.
16900 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
16902 * common/config/arm/arm-common.c: Adjust include path for
16904 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
16905 (arm-cpu.h): Create in build directory. Adjust dependency rules.
16906 (arm-cpu-data.h): Likewise.
16907 (arm-cpu-cdata.h): Likewise.
16908 * config/arm/arm-cpu.h: Delete.
16909 * config/arm/arm-cpu-cdata.h: Delete.
16910 * config/arm/arm-cpu-data.h: Delete.
16912 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
16914 * config/arm/arm-cpus.in (cortex-a55): New.
16915 (cortex-a75): Likewise.
16916 (cortex-a75.cortex-a55): Likewise.
16917 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
16919 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
16920 * config/arm/arm-cpu-cdata.h: Regenerate.
16921 * config/arm/arm-cpu-data.h: Regenerate.
16922 * config/arm/arm-cpu.h: Regenerate.
16923 * config/arm/arm-tables.opt: Regenerate.
16924 * config/arm/arm-tune.md: Regenerate.
16926 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16928 * haifa-sched.c (sched_create_recovery_edges): Update profile.
16930 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16932 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
16935 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
16937 PR tree-optimization/81292
16938 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
16939 full_string_p, also call adjust_related_strinfos if the adjustment
16940 is simple, otherwise invalidate related strinfos.
16942 2017-07-04 Martin Liska <mliska@suse.cz>
16945 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
16946 newly created variable as DECL_IGNORED_P.
16948 2017-07-04 Martin Liska <mliska@suse.cz>
16951 * ipa-inline.c (inline_small_functions):
16952 Use xstrdup_for_dump.
16954 2017-07-04 Tom de Vries <tom@codesourcery.com>
16956 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
16958 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
16961 * config/darwin.c (darwin_function_switched_text_sections):
16964 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
16966 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
16968 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
16970 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
16972 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16974 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
16975 min_profitable_iters, and th as inclusive lower bounds.
16976 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
16977 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
16978 for min_profitable_iters and min_profitable_estimate.
16979 (vect_transform_loop): Treat th as an inclusive lower bound.
16980 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
16982 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
16985 * config/darwin.c (darwin_function_switched_text_sections):
16986 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
16987 in two pieces, and suppress the use of buf.
16989 2017-07-03 Nathan Sidwell <nathan@acm.org>
16991 * hash-table.h (hash_table_mod1): Fix indentation.
16993 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16995 PR middle-end/81290
16996 * predict.c (force_edge_cold): Be more careful about propagation
16998 * profile-count.h (profile_probability::guessed,
16999 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17001 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17003 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17005 * doc/invoke.texi (rcpc architecture extension): Document it.
17007 2017-07-03 Richard Biener <rguenther@suse.de>
17009 PR tree-optimization/60510
17010 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17011 the scalar reduction PHI and use it.
17012 (vectorizable_reduction): Properly guard the single_defuse_cycle
17013 path for non-SLP reduction chains where we cannot use it.
17014 Rework reduc_def/index and vector type deduction. Rework
17015 vector operand gathering during reduction op code-gen.
17016 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17017 chains dissolve the chain and leave it to non-SLP reduction
17020 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17022 * tree-data-ref.h (dr_alignment): Declare.
17023 * tree-data-ref.c (dr_alignment): New function.
17024 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17025 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17027 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17029 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17031 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17032 and base_misalignment fields.
17033 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17034 * tree-data-ref.c: Include builtins.h.
17035 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17036 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17037 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17038 * tree-vect-data-refs.c: Include tree-cfg.h.
17039 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17040 fields instead of calculating an alignment here.
17041 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17042 innermost_loop_behavior fields.
17044 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17046 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17048 (DR_STEP_ALIGNMENT): New macro.
17049 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17050 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17051 (create_data_ref): Print it.
17052 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17053 to tell whether the step preserves vector (mis)alignment.
17054 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17055 Move the check for an integer step and generalise to all INTEGER_CST.
17056 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17057 Print the outer step alignment.
17059 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17061 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17062 with offset_alignment.
17063 (DR_ALIGNED_TO): Delete.
17064 (DR_OFFSET_ALIGNMENT): New macro.
17065 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17066 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17067 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17068 (create_data_ref): Likewise.
17069 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17070 (vect_analyze_data_refs): Likewise.
17071 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17072 creating dummy innermost behavior.
17074 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17076 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17077 with a "innermost_loop_behavior *" and refeence tree.
17078 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17079 (create_data_ref): Update call accordingly.
17080 * tree-predcom.c (find_looparound_phi): Likewise.
17082 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17084 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17085 fields with dr_wrt_vec_loop.
17086 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17087 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17088 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17089 (vect_dr_behavior): New function.
17090 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17091 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17092 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17093 track whether the step preserves the misalignment.
17094 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17095 Use vect_dr_behavior.
17096 (vect_setup_realignment): Update call accordingly.
17097 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17098 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17099 call to vect_create_addr_base_for_vector_ref.
17100 (vect_create_cond_for_align_checks): Likewise.
17101 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17102 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17103 (vect_recog_mask_conversion_pattern): Likewise.
17104 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17105 (new_stmt_vec_info): Remove redundant zeroing.
17107 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17109 * common/config/arm/arm-common.c (arm_be8_option): New function.
17110 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17111 (ISA_ARMv6): Add isa_bit_be8.
17112 * config/arm/arm.h (arm_be8_option): Add prototype.
17113 (BE8_SPEC_FUNCTION): New define.
17114 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17115 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17116 (mlittle-endian): Similarly.
17117 (mbe8, mbe32): New options.
17118 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17119 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17121 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17123 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17125 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17127 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17128 (cleanup_tree_cfg_bb): Use it.
17129 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17131 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17133 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17136 * loop-doloop.c (add_test): Update profile.
17138 2017-07-03 Martin Liska <mliska@suse.cz>
17141 * sanopt.c (rewrite_usage_of_param): New function.
17142 (sanitize_rewrite_addressable_params): Likewise.
17143 (pass_sanopt::execute): Call rewrite_usage_of_param.
17145 2017-07-03 Richard Biener <rguenther@suse.de>
17147 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17148 back to using VIEW_CONVERT_EXPR.
17150 2017-07-03 Martin Liska <mliska@suse.cz>
17153 * doc/extend.texi: Document when a resolver function is
17154 generated for target_clones.
17156 2017-07-03 Martin Liska <mliska@suse.cz>
17158 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17159 and shadow memory of auto variables which are subject of
17160 use-after-scope sanitization.
17161 (asan_expand_mark_ifn): Add do set only when is_poison.
17163 2016-07-03 Richard Biener <rguenther@suse.de>
17165 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17167 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17168 (vectorizable_reduction): Perform reduction PHI creation when
17169 visiting a reduction PHI and adjust and simplify code generation
17170 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17171 (vect_transform_loop): Visit reduction PHIs.
17172 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17173 defs into the SLP tree.
17174 (vect_build_slp_tree): Reduction defs terminate the recursion.
17175 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17177 (vect_get_vec_defs_for_stmt_copy): Export.
17178 (vect_get_vec_defs): Likewise.
17179 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17181 (vect_get_vec_defs_for_stmt_copy): Declare.
17182 (vect_get_vec_defs): Likewise.
17184 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17186 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17187 parameter with a "loop" parameter and use it instead of the
17188 loop containing DR_STMT. Don't check simple_iv when doing
17189 BB analysis. Describe the two analysis modes in the comment.
17191 2017-07-03 Tom de Vries <tom@codesourcery.com>
17193 PR tree-optimization/69468
17194 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17195 (find_same_succ_bb): Handle ignore_edge_flags.
17197 2017-07-03 Tom de Vries <tom@codesourcery.com>
17199 PR tree-optimization/81192
17200 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17202 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17204 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17206 2017-07-03 Tom de Vries <tom@codesourcery.com>
17208 PR tree-optimization/81192
17209 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17210 BB_SAME_SUCC (bb) == NULL.
17212 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17214 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17217 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17219 * dumpfile.c: Include profile-count.h
17220 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17222 (insert_cond_bb): Update profile.
17223 * tree-cfg.h (insert_cond_bb): Update prototype.
17224 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17225 * tree-dump.c: Do not include tree-cfg.
17227 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17229 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17231 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17233 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17236 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17238 * tree-complex.c (expand_complex_div_wide): update profile.
17240 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17241 Alan Hayward <alan.hayward@arm.com>
17242 David Sherwood <david.sherwood@arm.com>
17244 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17245 (CORETYPES_H): New define.
17246 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17247 (insn-modes-inline.h, s-modes-inline-h): New rules.
17248 (generated_files): Add insn-modes-inline.h.
17249 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17250 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
17251 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
17252 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
17253 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
17254 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
17255 (build/gencodes.o, build/genconditions.o): Likewise.
17256 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
17257 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
17258 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
17259 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
17260 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
17261 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
17262 * coretypes.h: Include everything up to real.h for generators.
17263 Include insn-modes.h first. Include wide-int-print.h after
17264 wide-int.h. Include insn-modes-inline.h and then machmode.h.
17265 * machmode.h: Don't include insn-modes.h here.
17266 * function-tests.c: Remove includes of signop.h, machmode.h,
17267 double-int.h and wide-int.h.
17269 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
17271 * optc-save-gen.awk: Likewise.
17272 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
17273 * godump.c: Remove include of wide-int-print.h.
17274 * pretty-print.h: Likewise.
17275 * wide-int-print.cc: Likewise.
17276 * wide-int.cc: Likewise.
17277 * hash-map-tests.c: Remove include of signop.h.
17278 * hash-set-tests.c: Likewise.
17279 * rtl-tests.c: Likewise.
17280 * mkconfig.sh: Remove include of machmode.h.
17281 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
17283 (emit_insn_modes_inline_h): ...this new function. Emit the code
17284 into an insn-modes-inline.h header file, adding appropriate
17285 include guards and end comments.
17286 (emit_insn_modes_c_header): Remove include of machmode.h.
17287 (emit_min_insn_modes_c_header): Include coretypes.h rather than
17289 (main): Handle -i flag and call emit_insn_modes_inline_h when
17292 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17294 * tree-ssa-strlen.c (strinfo): Rename the length field to
17295 nonzero_chars. Add a full_string_p field.
17296 (compare_nonzero_chars, zero_length_string_p): New functions.
17297 (get_addr_stridx): Add an offset_out parameter.
17298 Use compare_nonzero_chars.
17299 (get_stridx): Update accordingly. Use compare_nonzero_chars.
17300 (new_strinfo): Update after above changes to strinfo.
17301 (set_endptr_and_length): Set full_string_p.
17302 (get_string_length): Update after above changes to strinfo.
17303 (unshare_strinfo): Update call to new_strinfo.
17304 (maybe_invalidate): Likewise.
17305 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
17306 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
17307 as a uhwi instead of an shwi. Update after above changes to
17308 strinfo and new_strinfo.
17309 (zero_length_string): Assert that chainsi contains full strings.
17310 Use zero_length_string_p. Update call to new_strinfo.
17311 (adjust_related_strinfos): Update after above changes to strinfo.
17312 Copy full_string_p from origsi.
17313 (adjust_last_stmt): Use zero_length_string_p.
17314 (handle_builtin_strlen): Update after above changes to strinfo and
17315 new_strinfo. Install the lhs as the string length if the previous
17316 entry didn't describe a full string.
17317 (handle_builtin_strchr): Update after above changes to strinfo
17319 (handle_builtin_strcpy): Likewise.
17320 (handle_builtin_strcat): Likewise.
17321 (handle_builtin_malloc): Likewise.
17322 (handle_pointer_plus): Likewise.
17323 (handle_builtin_memcpy): Likewise. Track nonzero characters
17324 that aren't necessarily followed by a nul terminator.
17325 (handle_char_store): Likewise.
17327 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17329 PR tree-optimization/80769
17330 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
17331 for malloc and calloc. Document the new invariant that all related
17332 strinfos have delayed lengths or none do.
17333 (verify_related_strinfos): Move earlier in file.
17334 (set_endptr_and_length): New function, split out from...
17335 (get_string_length): ...here. Also set the lengths of related
17337 (zero_length_string): Assert that chainsi has known (rather than
17339 (adjust_related_strinfos): Likewise.
17341 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17343 PR tree-optimization/81136
17344 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
17345 assert that two references with the same misalignment have the same
17346 compile-time misalignment if those compile-time misalignments
17349 2017-07-01 Andi Kleen <ak@linux.intel.com>
17351 * print-tree.c (print_node): Print all attributes.
17353 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17355 * cfg.c (scale_bbs_frequencies): New function.
17356 * cfg.h (scale_bbs_frequencies): Declare it.
17357 * cfgloopanal.c (single_likely_exit): Cleanup.
17358 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
17360 (scale_loop_profile): Likewise.
17361 (loop_version): Likewise.
17362 (create_empty_loop_on_edge): Update.
17363 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
17364 scale_loop_frequencies, scale_loop_profile, loopify,
17365 loop_version): Update prototypes.
17366 * modulo-sched.c (sms_schedule): Update.
17367 * predict.c (unlikely_executed_edge_p): Also check probability.
17368 (probably_never_executed_edge_p): Fix typo.
17369 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17370 * tree-parloops.c (gen_parallel_loop): Update.
17371 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
17372 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17373 * tree-ssa-loop-split.c (split_loop): Update.
17374 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17375 * tree-vect-loop-manip.c (vect_do_peeling): Update.
17376 (vect_loop_versioning): Update.
17377 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17379 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17381 * trans-mem.c (split_bb_make_tm_edge): Update profile.
17383 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17385 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
17386 to keep profile consistent.
17388 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17390 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
17391 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
17392 * profile-count.h (max_safe_multiplier): Make unsigned.
17393 (profile_count::guessed_zero): New.
17395 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17397 * bb-reorder.c (fix_up_crossing_landing_pad,
17398 fix_crossing_conditional_branches): Use make_single_succ_edge
17399 to keep profile consistent.
17401 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17403 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
17406 2017-07-01 Jakub Jelinek <jakub@redhat.com>
17409 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
17410 the right scopes, make sure cond_jump isn't preserved between multiple
17411 iterations. Search for fallthru edge whenever there are 3+ edges and
17412 use find_fallthru_edge for it.
17414 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17416 Patch by Alexander Monakov <amonakov@ispras.ru>
17417 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
17418 probabilities consistently.
17420 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17422 * pa.c (pa_expand_compare_and_swap_loop): Update call of
17423 emit_cmp_and_jump_insns.
17425 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17428 * tree-inline.c (expand_call_inline): Combine profile statuses.
17430 2017-06-30 Andrew Pinski <apinski@cavium.com>
17432 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
17433 fold_stmt returned true.
17435 2017-06-30 Nathan Sidwell <nathan@acm.org>
17437 * ggc.h (empty_string): Delete.
17438 * cfgexpand.c (expand_asm_stmt): Use plain "".
17439 * optabs.c (expand_asm_memory_barrier): Likewise.
17440 * stringpool.c (empty_string): Delete.
17441 (digit_vector, digit_string): Delete.
17442 (ggc_alloc_string): Use plain "", don't optimize single digit
17443 strings. Use ggc_alloc_atomic.
17445 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
17447 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
17448 comparison set and one other set, use the cost of the non-comparison
17451 2017-06-30 Nathan Sidwell <nathan@acm.org>
17453 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
17456 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
17458 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
17459 loops. Remove now unneeded calls to gimple_switch_set_label() that
17460 just set removed labels to NULL_TREE.
17462 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
17464 * tree-ssanames.c (set_range_info_raw): Abstract from ...
17465 (set_range_info): ...here. Only call set_range_info_raw if domain
17467 (set_nonzero_bits): Call set_range_info_raw.
17468 * tree-ssanames.h (set_range_info_raw): New.
17470 2017-06-30 Jakub Jelinek <jakub@redhat.com>
17473 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
17474 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
17475 of nonimmediate_operand and <store_mask_constraint> instead of m
17476 for the input operand. For V8FI iterator, always split if input
17477 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
17478 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
17479 <store_mask_predicate> instead of register_operand and
17480 <store_mask_constraint> instead of v for the input operand. Make
17481 sure both operands aren't MEMs for if not <mask_applied>.
17483 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
17485 * lto-wrapper.c (copy_file) Close both file descriptors before
17488 2017-06-30 Martin Liska <mliska@suse.cz>
17491 * multiple_target.c (create_dispatcher_calls): Make ifunc
17492 also for function that don't have calls or are not referenced.
17494 2017-06-30 Richard Biener <rguenther@suse.de>
17496 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
17497 analyze the first scalar stmt. Move vector type computation
17498 for the BB case here from ...
17499 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
17500 live operation processing in the SLP case properly.
17502 2017-06-30 Richard Biener <rguenther@suse.de>
17504 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
17506 2017-06-30 Martin Liska <mliska@suse.cz>
17509 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
17510 before BUILT_IN_UNWIND_RESUME when ASAN is used.
17512 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
17514 * doc/invoke.texi (AArch64): Add missing options and remove redundant
17517 2017-06-30 Richard Biener <rguenther@suse.de>
17519 PR tree-optimization/81249
17520 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
17521 condition reduction result to original scalar type.
17523 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17525 * profile-count.h (enum profile_quality): Fix typos and whitespace
17528 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17530 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
17531 type for branch probabilities.
17533 2017-06-29 Julian Brown <julian@codesourcery.com>
17534 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17536 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
17537 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
17538 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
17539 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
17541 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17543 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
17544 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
17545 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
17546 CC usage from generic code to here.
17547 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
17548 CC usage into the target macros.
17550 2017-06-29 Maya Rashish <coypu@sdf.org>
17552 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
17555 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17557 * arm/arm-builtins.c: Include profile-count.h
17558 * except.c (sjlj_emit_function_enter): Use
17559 profile_probability::unlikely.
17561 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17563 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
17564 and tocrel_offset be pointer args rather than implicitly using
17566 (legitimate_constant_pool_address_p, rs6000_emit_move,
17567 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
17568 tocrel_offset and use in toc_relative_expr_p call.
17569 (print_operand, print_operand_address): Use static tocrel_base_oac
17570 and tocrel_offset_oac.
17571 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
17574 2017-06-29 Maya Rashish <coypu@sdf.org>
17576 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
17578 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
17580 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
17581 objects, take into account only the alignment of 'op0' and 'mode1' if
17584 2017-06-29 Steve Ellcey <sellcey@cavium.com>
17586 * ccmp.c (ccmp_tree_comparison_p): New function.
17587 (ccmp_candidate_p): Update to use above function.
17588 (get_compare_parts): New function.
17589 (expand_ccmp_next): Update to use new functions.
17590 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
17592 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
17593 take mode as argument.
17594 * ccmp.h (expand_ccmp_expr): Add mode as argument.
17595 * expr.c (expand_expr_real_1): Pass mode as argument.
17597 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
17599 * combine.c (combine_instructions): Print insns to dump_file, together
17602 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17604 * asan.c (asan_emit_stack_protection): Update.
17605 (create_cond_insert_point): Update.
17606 * auto-profile.c (afdo_propagate_circuit): Update.
17607 * basic-block.h (struct edge_def): Turn probability to
17608 profile_probability.
17609 (EDGE_FREQUENCY): Update.
17610 * bb-reorder.c (find_traces_1_round): Update.
17611 (better_edge_p): Update.
17612 (sanitize_hot_paths): Update.
17613 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
17614 (make_single_succ_edge): Update.
17615 (check_bb_profile): Update.
17616 (dump_edge_info): Update.
17617 (update_bb_profile_for_threading): Update.
17618 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
17619 probabilitycount to 0.
17620 * cfgbuild.c (compute_outgoing_frequencies): Update.
17621 * cfgcleanup.c (try_forward_edges): Update.
17622 (outgoing_edges_match): Update.
17623 (try_crossjump_to_edge): Update.
17624 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
17625 (expand_gimple_tailcall): Update.
17626 (construct_init_block): Use make_single_succ_edge.
17627 (construct_exit_block): Use make_single_succ_edge.
17628 * cfghooks.c (verify_flow_info): Update.
17629 (redirect_edge_succ_nodup): Update.
17630 (split_edge): Update.
17631 (account_profile_record): Update.
17632 * cfgloopanal.c (single_likely_exit): Update.
17633 * cfgloopmanip.c (scale_loop_profile): Update.
17634 (set_zero_probability): Remove.
17635 (duplicate_loop_to_header_edge): Update.
17636 * cfgloopmanip.h (loop_version): Update prototype.
17637 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
17638 (force_nonfallthru_and_redirect): Update.
17639 (update_br_prob_note): Update.
17640 (rtl_verify_edges): Update.
17641 (purge_dead_edges): Update.
17642 (rtl_lv_add_condition_to_bb): Update.
17643 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
17644 * cgraphunit.c (init_lowered_empty_function): Update.
17645 (cgraph_node::expand_thunk): Update.
17646 * cilk-common.c: Include profile-count.h
17647 * dojump.c (inv): Remove.
17648 (jumpifnot): Update.
17649 (jumpifnot_1): Update.
17650 (do_jump_1): Update.
17652 (do_jump_by_parts_greater_rtx): Update.
17653 (do_compare_rtx_and_jump): Update.
17654 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
17655 do_jump_1. do_compare_rtx_and_jump): Update prototype.
17656 * dwarf2cfi.c: Include profile-count.h
17657 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
17658 (sjlj_emit_dispatch_table): Likewise.
17659 * explow.c: Include profile-count.h
17660 * expmed.c (emit_store_flag_force): Update.
17661 (do_cmp_and_jump): Update.
17662 * expr.c (compare_by_pieces_d::generate): Update.
17663 (compare_by_pieces_d::finish_mode): Update.
17664 (emit_block_move_via_loop): Update.
17665 (store_expr_with_bounds): Update.
17666 (store_constructor): Update.
17667 (expand_expr_real_2): Update.
17668 (expand_expr_real_1): Update.
17669 * expr.h (try_casesi, try_tablejump): Update prototypes.
17670 * gimple-pretty-print.c (dump_probability): Update.
17671 (dump_profile): New.
17672 (dump_gimple_label): Update.
17673 (dump_gimple_bb_header): Update.
17674 * graph.c (draw_cfg_node_succ_edges): Update.
17675 * hsa-gen.c (convert_switch_statements): Update.
17676 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
17677 (find_if_case_1): Update.
17678 (find_if_case_2): Update.
17679 * internal-fn.c (expand_arith_overflow_result_store): Update.
17680 (expand_addsub_overflow): Update.
17681 (expand_neg_overflow): Update.
17682 (expand_mul_overflow): Update.
17683 (expand_vector_ubsan_overflow): Update.
17684 * ipa-cp.c (good_cloning_opportunity_p): Update.
17685 * ipa-split.c (split_function): Use make_single_succ_edge.
17686 * ipa-utils.c (ipa_merge_profiles): Update.
17687 * loop-doloop.c (add_test): Update.
17688 (doloop_modify): Update.
17689 * loop-unroll.c (compare_and_jump_seq): Update.
17690 (unroll_loop_runtime_iterations): Update.
17691 * lra-constraints.c (lra_inheritance): Update.
17692 * lto-streamer-in.c (input_cfg): Update.
17693 * lto-streamer-out.c (output_cfg): Update.
17694 * mcf.c (adjust_cfg_counts): Update.
17695 * modulo-sched.c (sms_schedule): Update.
17696 * omp-expand.c (expand_omp_for_init_counts): Update.
17697 (extract_omp_for_update_vars): Update.
17698 (expand_omp_ordered_sink): Update.
17699 (expand_omp_for_ordered_loops): Update.
17700 (expand_omp_for_generic): Update.
17701 (expand_omp_for_static_nochunk): Update.
17702 (expand_omp_for_static_chunk): Update.
17703 (expand_cilk_for): Update.
17704 (expand_omp_simd): Update.
17705 (expand_omp_taskloop_for_outer): Update.
17706 (expand_omp_taskloop_for_inner): Update.
17707 * omp-simd-clone.c (simd_clone_adjust): Update.
17708 * optabs.c (expand_doubleword_shift): Update.
17709 (expand_abs): Update.
17710 (emit_cmp_and_jump_insn_1): Update.
17711 (expand_compare_and_swap_loop): Update.
17712 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
17713 * predict.c (predictable_edge_p): Update.
17714 (edge_probability_reliable_p): Update.
17715 (set_even_probabilities): Update.
17716 (combine_predictions_for_insn): Update.
17717 (combine_predictions_for_bb): Update.
17718 (propagate_freq): Update.
17719 (estimate_bb_frequencies): Update.
17720 (force_edge_cold): Update.
17721 * profile-count.c (profile_count::dump): Add missing space into dump.
17722 (profile_count::debug): Add newline.
17723 (profile_count::differs_from_p): Explicitly convert to unsigned.
17724 (profile_count::stream_in): Update.
17725 (profile_probability::dump): New member function.
17726 (profile_probability::debug): New member function.
17727 (profile_probability::differs_from_p): New member function.
17728 (profile_probability::differs_lot_from_p): New member function.
17729 (profile_probability::stream_in): New member function.
17730 (profile_probability::stream_out): New member function.
17731 * profile-count.h (profile_count_quality): Rename to ...
17732 (profile_quality): ... this one.
17733 (profile_probability): New.
17734 (profile_count): Update.
17735 * profile.c (compute_branch_probabilities): Update.
17736 * recog.c (peep2_attempt): Update.
17737 * sched-ebb.c (schedule_ebbs): Update.
17738 * sched-rgn.c (find_single_block_region): Update.
17739 (compute_dom_prob_ps): Update.
17740 (schedule_region): Update.
17741 * sel-sched-ir.c (compute_succs_info): Update.
17742 * stmt.c (struct case_node): Update.
17743 (do_jump_if_equal): Update.
17744 (get_outgoing_edge_probs): Update.
17745 (conditional_probability): Update.
17746 (emit_case_dispatch_table): Update.
17747 (expand_case): Update.
17748 (expand_sjlj_dispatch_table): Update.
17749 (emit_case_nodes): Update.
17750 * targhooks.c: Update.
17751 * tracer.c (better_p): Update.
17752 (find_best_successor): Update.
17753 * trans-mem.c (expand_transaction): Update.
17754 * tree-call-cdce.c: Update.
17755 * tree-cfg.c (gimple_split_edge): Upate.
17756 (move_sese_region_to_fn): Upate.
17757 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
17758 * tree-eh.c (lower_resx): Upate.
17759 (cleanup_empty_eh_move_lp): Upate.
17760 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17761 * tree-inline.c (copy_edges_for_bb): Update.
17762 (copy_cfg_body): Update.
17763 * tree-parloops.c (gen_parallel_loop): Update.
17764 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
17765 (gimple_gen_time_profiler): Update.
17766 * tree-ssa-dce.c (remove_dead_stmt): Update.
17767 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
17768 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
17769 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
17770 (unloop_loops): Update.
17771 (try_peel_loop): Update.
17772 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17773 * tree-ssa-loop-split.c (connect_loops): Update.
17774 (split_loop): Update.
17775 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17776 (hoist_guard): Update.
17777 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
17778 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
17779 (value_replacement): Update.
17780 * tree-ssa-reassoc.c (branch_fixup): Update.
17781 * tree-ssa-tail-merge.c (replace_block_by): Update.
17782 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
17783 (create_edge_and_update_destination_phis): Update.
17784 (compute_path_counts): Update.
17785 (recompute_probabilities): Update.
17786 (update_joiner_offpath_counts): Update.
17787 (freqs_to_counts_path): Update.
17788 (duplicate_thread_path): Update.
17789 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
17790 (struct switch_conv_info): Update.
17791 (gen_inbound_check): Update.
17792 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17793 (vect_do_peeling): Update.
17794 (vect_loop_versioning): Update.
17795 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17796 (optimize_mask_stores): Update.
17797 * ubsan.c (ubsan_expand_null_ifn): Update.
17798 * value-prof.c (gimple_divmod_fixed_value): Update.
17799 (gimple_divmod_fixed_value_transform): Update.
17800 (gimple_mod_pow2): Update.
17801 (gimple_mod_pow2_value_transform): Update.
17802 (gimple_mod_subtract): Update.
17803 (gimple_mod_subtract_transform): Update.
17804 (gimple_ic): Update.
17805 (gimple_stringop_fixed_value): Update.
17806 (gimple_stringops_transform): Update.
17807 * value-prof.h: Update.
17809 2017-06-29 Carl Love <cel@us.ibm.com>
17811 * config/rs6000/rs6000-c.c: Add support for built-in functions
17812 vector signed int vec_signed (vector float);
17813 vector signed long long vec_signed (vector double);
17814 vector signed int vec_signed2 (vector double, vector double);
17815 vector signed int vec_signede (vector double);
17816 vector signed int vec_signedo (vector double);
17817 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
17818 instruction generator.
17819 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
17820 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
17821 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
17823 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
17824 vunsignede_v2df): Add define_expands.
17825 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
17826 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
17827 VEC_UNSIGNEDO): Add definitions.
17828 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
17829 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
17830 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
17831 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
17832 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
17833 * config/rs6000/altivec.h (vec_signed, vec_signed2,
17834 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
17835 vec_unsignede, vec_unsignedo): Add builtin defines.
17836 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
17838 * doc/extend.texi: Update the built-in documentation file for the
17839 new built-in functions.
17841 2017-06-29 Richard Biener <rguenther@suse.de>
17843 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
17844 reduction chains to LOOP_VINFO_REDUCTIONS.
17845 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
17846 SLP reductions after processing reduction chains.
17848 2017-06-29 Nathan Sidwell <nathan@acm.org>
17850 * builtins.c (fold_builtin_FUNCTION): Use
17851 lang_hooks.decl_printable_name.
17853 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
17855 PR middle-end/81194
17856 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
17857 with only one label.
17858 * stmt.c (expand_case): Assert NCASES is greater than one.
17860 2017-06-29 Richard Biener <rguenther@suse.de>
17862 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
17864 (group_case_labels): Likewise.
17865 (find_taken_edge): Push sanity checking on val to workers...
17866 (find_taken_edge_cond_expr): ... here
17867 (find_taken_edge_switch_expr): ... and here, handle cases
17868 with just a default label.
17869 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
17870 (group_case_labels): Likewise.
17871 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
17872 group_case_labels does anything cleanup the CFG again.
17874 2017-06-29 Bin Cheng <bin.cheng@arm.com>
17876 PR tree-optimization/81196
17877 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
17878 exit condition comparing two IVs.
17880 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
17882 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
17883 profile to the dummy entry at the end of the list of architectures.
17884 * config/arm/arm-cpu-cdata.h: Regenerated.
17886 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17887 Michael Collison <michael.collison@arm.com>
17890 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
17892 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
17893 (*aarch64_reg_<mode>3_minus_mask): New pattern.
17894 (*aarch64_<optab>_reg_di3_mask2): New pattern.
17895 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
17896 of shift when the shift amount is masked with constant equal to
17897 the size of the mode.
17898 * config/aarch64/predicates.md (subreg_lowpart_operator): New
17901 2017-06-29 Martin Liska <mliska@suse.cz>
17903 * config/i386/i386.opt: Change range from [1,5] to [0,5].
17905 2017-06-29 Yury Gribov <tetra2005@gmail.com>
17908 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
17911 (edge_growth_cache_entry::edge_growth_cache_entry): New
17913 (reset_edge_growth_cache): Update to use constructor.
17915 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17917 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
17918 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
17919 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
17921 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
17923 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
17924 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
17926 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
17928 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
17929 (*-linux-uclibc*): Add t-uclibc tmake_file.
17930 * config/t-musl: New.
17931 * config/t-uclibc: New.
17933 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
17935 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
17937 (gen_comm_data): Emit architectural setting of arch_prof.
17938 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
17940 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
17941 (armv8-m.base, armv8-m.main): Likewise.
17942 * arm-protos.h (arm_build_target): Add profile field.
17943 (arch_option): Likewise.
17944 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
17946 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
17947 arm_active_target.profile.
17949 2017-06-28 Richard Biener <rguenther@suse.de>
17951 PR middle-end/81227
17952 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
17953 TYPE_OVERFLOW_WRAPS.
17954 * match.pd (negate_expr_p): Likewise.
17955 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
17956 fold_build2, not fold_binary.
17958 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
17960 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
17961 Convert memory address to Pmode.
17962 (aarch64_print_operand): Assert MEM operands are always Pmode.
17964 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
17967 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
17968 Remove redundant if.
17969 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
17970 * config/arm/aarch-common-protos.h
17971 (aarch_forward_to_shift_is_not_shifted_re): Remove.
17972 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
17974 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
17977 * multiple_target.c (create_dispatcher_calls): Set the default
17978 clone to be static, not public.
17980 2017-06-28 Richard Biener <rguenther@suse.de>
17982 * tree-vect-loop.c (vectorizable_reduction): Move special
17983 cond reduction IV var creation ...
17984 (vect_create_epilog_for_reduction): ... here. Remove induction_index
17985 parameter. Use STMT_VINFO_VECTYPE.
17986 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
17989 2017-06-28 Martin Liska <mliska@suse.cz>
17992 * ipa-visibility.c (non_local_p): Handle visibility.
17994 2017-06-28 Martin Liska <mliska@suse.cz>
17997 * common.opt: Add IntegerRange to various options.
17998 * opt-functions.awk (integer_range_info): New function.
17999 * optc-gen.awk: Add integer_range_info to cl_options struct.
18000 * opts-common.c (decode_cmdline_option): Handle
18001 CL_ERR_INT_RANGE_ARG.
18002 (cmdline_handle_error): Likewise.
18003 * opts.c (print_filtered_help): Show valid interval in
18004 when --help is provided.
18005 * opts.h (struct cl_option): Add range_min and range_max fields.
18006 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18008 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18010 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18011 (x * C EQ/NE y * C): New transformation.
18013 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18015 * genmultilib (combination_space): Accept '+' in option names.
18017 2017-06-28 Martin Liska <mliska@suse.cz>
18020 * asan.c (instrument_derefs): Bail out inner references
18021 that are hard register variables.
18023 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18026 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18027 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18029 2017-06-28 Richard Biener <rguenther@suse.de>
18031 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18032 (vect_get_slp_defs): Adjust.
18033 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18035 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18037 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18038 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18039 (vectorizable_reduction): Adjust.
18040 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18042 (vect_get_slp_defs): Likewise.
18043 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18044 (vectorizable_bswap): Adjust.
18045 (vectorizable_call): Likewise.
18046 (vectorizable_conversion): Likewise.
18047 (vectorizable_assignment): Likewise.
18048 (vectorizable_shift): Likewise.
18049 (vectorizable_operation): Likewise.
18050 (vectorizable_store): Likewise.
18051 (vectorizable_condition): Likewise.
18052 (vectorizable_comparison): Likewise.
18054 2017-06-28 Michael Collison <michael.collison@arm.com>
18057 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18059 (arm_gen_movmemqi): Removed unused variable 'i'.
18060 Convert 'for' loop into 'while' loop.
18061 (arm_expand_prologue): Remove last unnecessary set of insn.
18062 (thumb_pop): Remove unused variable 'pushed_words'.
18063 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18065 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18067 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18068 * config/s390/s390-protos.h: Add prototype of
18069 s390_rel_address_ok_p.
18070 * config/s390/s390.c (s390_got_symbol): New function.
18071 (s390_rel_address_ok_p): New function.
18072 (legitimize_pic_address): Use s390_rel_address_ok_p.
18073 (s390_load_got): Use s390_got_symbol.
18074 (s390_option_override): Issue error if
18075 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18076 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18078 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18080 2017-06-27 Andrew Pinski <apinski@cavium.com>
18082 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18083 (X * copysign (1.0, X)): New pattern.
18084 (X * copysign (1.0, -X)): New pattern.
18085 (copysign (-1.0, CST)): New pattern.
18087 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18089 * genmultilib (combination_space): Remove variable.
18090 Validate reuse rules against regular expression for any sequence
18091 of multilib options in any order.
18093 2017-06-27 Michael Collison <michael.collison@arm.com>
18095 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18096 call aarch64_split_simd_combine.
18097 * (aarch64_combine_internal<mode>): Delete pattern.
18098 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18099 Allow register and subreg operands.
18101 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18103 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18104 specific need, just fallback on defaults.
18105 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18107 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18108 Olivier Hainque <hainque@adacore.com>
18110 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18112 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18113 targets. Pick a default if no particular attempt applied.
18114 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18117 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18119 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18120 (x86_64-wrs-vxworks7): Likewise.
18122 2017-06-27 Marek Polacek <polacek@redhat.com>
18125 * ubsan.c (instrument_null): Check get_base_address's result for null.
18127 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18129 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18131 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18133 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18134 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18135 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18136 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18137 New function types.
18138 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18139 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18140 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18141 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18142 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18143 BUILT_IN_FEUPDATEENV): New builtins.
18144 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18145 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18146 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18147 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18149 (builtin_structptr_types): Adjust size.
18150 * tree.c (builtin_structptr_types): Add four entries.
18152 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18153 Olivier Hainque <hainque@adacore.com>
18155 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18156 (TLS_SYM): New local macro, forcing reference to __tls__ on
18157 link command lines for VxWorks 7 RTPs, triggering initialization
18159 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18160 OS features TLS support, true for RTPs on VxWorks 7.
18161 * config/vxworks.c (vxworks_override_options): Setup emutls
18164 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18166 * predict.c (test_prediction_value_range): Use -1U instead of -1
18167 to avoid narrowing conversion warning.
18168 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18169 to avoid narrowing conversion warning.
18170 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18172 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18174 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18176 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18177 64bit configurations.
18178 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18179 (SIZE_TYPE): Likewise.
18180 * config/vxworks.c (vxworks_emutls_var_fields): Use
18181 long_unsigned_type_node instead of unsigned_type_node as the offset
18182 field type, which is "pointer" mode in emutls.c.
18184 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18187 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18189 PR middle-end/81207
18190 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18191 gimple_vuse copying separately from gimple_vdef copying.
18193 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18195 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18198 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18199 Olivier Hainque <hainque@adacore.com>
18201 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18202 for all vxworks7 targets.
18203 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18204 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18205 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18206 variations for VX6/VX7 and 32/64bits later on in ...
18207 (VXWORKS_LIB_SPEC): Leverage new macros.
18208 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18209 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18211 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18213 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18214 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18216 2017-06-26 Carl Love <cel@us.ibm.com>
18218 * config/rs6000/rs6000-c.c: Add support for built-in functions
18219 vector bool char vec_reve (vector bool char);
18220 vector signed char vec_reve (vector signed char);
18221 vector unsigned char vec_reve (vector unsigned char);
18222 vector bool int vec_reve (vector bool int);
18223 vector signed int vec_reve (vector signed int);
18224 vector unsigned int vec_reve (vector unsigned int);
18225 vector bool long long vec_reve (vector bool long long);
18226 vector signed long long vec_reve (vector signed long long);
18227 vector unsigned long long vec_reve (vector unsigned long long);
18228 vector bool short vec_reve (vector bool short);
18229 vector signed short vec_reve (vector signed short);
18230 vector double vec_reve (vector double);
18231 vector float vec_reve (vector float);
18232 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18233 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18234 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18235 (altivec_vreve): New pattern.
18236 * config/rs6000/altivec.h (vec_reve): New define.
18237 * doc/extend.texi (vec_rev): Update the built-in documentation file
18238 for the new built-in functions.
18240 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18242 PR tree-optimization/71815
18243 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18245 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18247 (slsr_process_phi): Likewise.
18248 (replace_uncond_cands_and_profitable_phis): Don't replace a
18249 multiply candidate with a stride of 1 (copy or cast).
18250 (phi_incr_cost): Call uses_consumed_by_stmt rather than
18252 (lowest_cost_path): Likewise.
18253 (total_savings): Likewise.
18255 2017-06-26 Richard Biener <rguenther@suse.de>
18258 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
18259 Use def_builtin_pure for all gather builtins.
18261 2017-06-26 Richard Biener <rguenther@suse.de>
18263 PR tree-optimization/81203
18264 * tree-tailcall.c (find_tail_calls): Do not move stmts into
18265 non-dominating BBs.
18267 2017-06-26 Marek Polacek <polacek@redhat.com>
18270 * doc/invoke.texi: Document -Wmultistatement-macros.
18272 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
18274 * doc/sourcebuild.texi (ARM-specific attributes): Document new
18275 arm_neon_ok_no_float_abi effective target.
18277 2017-06-26 Richard Biener <rguenther@suse.de>
18279 PR tree-optimization/80928
18280 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
18281 (copy_bbs): Set BB_DUPLICATED flag early.
18282 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
18284 (execute_on_shrinking_pred): Likewise.
18285 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
18286 BB_DUPLICATED blocks.
18287 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
18288 iterate over all PHIs considering removal of *gsi.
18290 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
18292 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
18295 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18297 * config/rs6000/rs6000-string.c: (expand_block_clear,
18298 do_load_for_compare, select_block_compare_mode,
18299 compute_current_alignment, expand_block_compare,
18300 expand_strncmp_align_check, expand_strn_compare,
18301 expand_block_move, rs6000_output_load_multiple)
18302 Move functions related to string/block move/compare
18303 to a separate file.
18304 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
18305 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
18306 for this function which is now used in two files.
18307 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
18308 * config.gcc: Add rs6000-string.o to extra_objs for
18309 targets powerpc*-*-* and rs6000*-*-*.
18311 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
18314 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
18315 32-bit, since indexed is not valid for DImode.
18316 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
18317 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
18318 (define_peephole2 for Altivec d-form load): Add 32-bit support.
18319 (define_peephole2 for Altivec d-form store): Likewise.
18322 * multiple_target.c (create_dispatcher_calls): Only create the
18323 dispatcher call if the function is the default clone of a
18324 versioned function.
18326 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
18328 PR middle-end/80902
18329 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
18330 a call, force the call to not be a tail call.
18332 2017-06-23 Jeff Law <law@redhat.com>
18334 * doc/contrib.texi: Add entry for Steven Pemberton's work on
18337 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
18339 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
18340 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
18341 handling for early expansion of vector shifts (sl,sr,sra,rl).
18342 (builtin_function_type): Add vector shift right instructions
18343 to the unsigned argument list.
18345 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
18347 rtl-optimizatoin/79286
18348 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
18349 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
18350 trap. PIC register plus a const unspec without offset can never trap.
18352 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
18354 * tree.h (builtin_structptr_type): New type.
18355 (builtin_structptr_types): Declare new array.
18356 * tree.c (builtin_structptr_types): New array.
18357 (free_lang_data, build_common_tree_nodes): Use it.
18359 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
18362 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
18365 2017-06-22 Matt Turner <mattst88@gmail.com>
18367 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
18368 Lake models to skylake case. Assume skylake for unknown
18369 models with clflushopt.
18371 2017-06-22 Jeff Law <law@redhat.com>
18373 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
18374 frame sizes that do not satisfy aarch64_uimm12_shift.
18376 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
18378 * profile-count.h (apply_probability,
18379 apply_scale, probability_in): Fix checks for zero.
18381 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18383 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
18384 * doc/cppdiropts.texi (-I @var{dir}): Document it.
18386 2016-06-22 Richard Biener <rguenther@suse.de>
18388 * tree-vect-loop.c (vect_model_reduction_cost): Handle
18389 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
18390 REDUC_MAX_EXPR support.
18391 (vectorizable_reduction): Likewise.
18392 (vect_create_epilog_for_reduction): Likewise.
18394 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
18396 * match.pd (A / (1 << B) -> A >> B): New.
18397 * generic-match-head.c: Include optabs-tree.h.
18398 * gimple-match-head.c: Likewise.
18399 * optabs-tree.h (target_supports_op_p): New.
18400 * optabs-tree.c (target_supports_op_p): New.
18402 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18404 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
18405 $gcc_cv_ld --help output.
18406 (gcc_cv_ld_demangle): Likewise.
18407 (gcc_cv_ld_eh_frame_hdr): Likewise.
18408 (gcc_cv_ld_pie): Likewise.
18409 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
18410 (gcc_cv_ld_buildid): Likewise.
18411 (gcc_cv_ld_sysroot): Likewise.
18412 (ld_bndplt_support): Likewise.
18413 (ld_pushpopstate_support): Likewise.
18414 * configure: Regenerate.
18415 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
18417 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18420 * config/i386/sse.md (round<mode>2): Renumber match_dup and
18421 operands indexes to avoid gap between operands and match_dups.
18423 2017-06-21 Andrew Pinski <apinski@cavium.com>
18425 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
18426 Increment Arith_shift and Arith_shift_reg by 1.
18427 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
18429 * config/aarch64/aarch64.c (thunderx_tunings): Enable
18430 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
18431 (aarch64_strip_extend): Add new argument and test for it.
18432 (aarch64_cheap_mult_shift_p): New function.
18433 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
18434 add a cost if it is true.
18435 Update calls to aarch64_strip_extend.
18436 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
18438 2017-06-21 Andrew Pinski <apinski@cavium.com>
18440 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
18442 (thunderxt88): Likewise.
18443 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
18444 (thunderx_prefetch_tune): New variable.
18445 (thunderx2t99_prefetch_tune): Update for the correct values.
18446 (thunderxt88_tunings): New variable.
18447 (thunderx_tunings): Use thunderx_prefetch_tune instead of
18448 generic_prefetch_tune.
18449 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
18451 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18453 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
18454 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
18455 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
18456 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
18457 (aarch64_atomic_cas<mode>, GPI): Likewise.
18459 2017-06-21 Martin Liska <mliska@suse.cz>
18461 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
18462 statements on cold and hot labels.
18463 * predict.c (tree_estimate_probability_bb): Remove the
18464 prediction from this place.
18466 2017-06-21 Martin Liska <mliska@suse.cz>
18468 PR tree-optimization/79489
18469 * gimplify.c (maybe_add_early_return_predict_stmt): New
18471 (gimplify_return_expr): Call the function.
18472 * predict.c (tree_estimate_probability_bb): Remove handling
18474 * predict.def: Update comment about early return predictor.
18475 * gimple-predict.h (is_gimple_predict): New function.
18476 * predict.def: Change default value of early return to 66.
18477 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
18479 * passes.def: Put pass_strip_predict_hints to the beginning of
18482 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
18484 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
18485 FUNCTION_DECL declarations.
18486 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
18488 (dwaf2out_decl): Likewise.
18489 * godump.c (go_early_global_decl): Skip call to the real debug hook
18490 for FUNCTION_DECL declarations.
18491 * passes.c (rest_of_decl_compilation): Skip call to the
18492 early_global_decl debug hook for FUNCTION_DECL declarations, unless
18493 -fdump-go-spec is passed.
18495 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18497 * config/i386/i386.c (struct builtin_isa): New field pure_p.
18498 Reorder for compactness.
18499 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
18500 (def_builtin_pure, def_builtin_pure2): New functions.
18501 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
18503 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18505 * match.pd (nop_convert): New predicate.
18506 ((A +- CST1) +- CST2): Allow some NOP conversions.
18508 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18511 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
18512 with ctors/dtors if GOVD_SHARED is set.
18514 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18516 * config/aarch64/aarch64.md (movti_aarch64):
18517 Emit mov rather than orr.
18518 (movtf_aarch64): Likewise.
18519 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
18520 Emit mov rather than orr.
18522 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18524 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
18525 Swap alternatives, make integer dup more expensive.
18527 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18529 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
18530 Return true for non-tls symbols.
18532 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
18534 * config/aarch64/aarch64-cores.def (cortex-a55): New.
18535 (cortex-a75): Likewise.
18536 (cortex-a75.cortex-a55): Likewise.
18537 * config/aarch64/aarch64-tune.md: Regenerate.
18538 * doc/invoke.texi (-mtune): Document new values for -mtune.
18540 2017-06-21 Tom de Vries <tom@codesourcery.com>
18542 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
18543 stack_size feature.
18544 (Effective-Target Keywords, Other attributes): Suggest using
18545 dg-add-options stack_size feature to get stack limit in stack_size
18546 effective target documentation.
18548 2017-06-21 Julian Brown <julian@codesourcery.com>
18549 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18551 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
18552 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
18553 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
18555 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
18556 attribute type list for neon_multiply.
18557 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
18558 attribute type list for neon_multiply.
18559 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
18560 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
18561 attribute type list for neon_multiply.
18562 * config/arm/types.md (crypto_pmull): Add.
18563 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
18564 attribute type list.
18566 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
18568 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
18571 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18573 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
18574 to make sure not to dereference a NULL cost_classes_ptr pointer.
18576 2017-06-20 Carl Love <cel@us.ibm.com>
18578 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18579 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18580 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18581 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18582 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18583 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18584 VMULOSW): New enum "unspec" values.
18585 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
18586 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
18587 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18588 altivec_vmulosw): New patterns.
18589 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18590 VMULOSW): Add definitions.
18592 2017-06-20 Julia Koval <julia.koval@intel.com>
18594 * config/i386/i386.c: Fix rounding expand for new pattern.
18595 * config/i386/subst.md: Fix pattern (parallel -> unspec).
18597 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18599 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
18600 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
18602 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18604 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
18607 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18609 * config/aarch64/aarch64-cores.def: Rearrange to sort by
18610 architecture, then by implementer ID.
18611 * config/aarch64/aarch64-tune.md: Regenerate.
18613 2017-06-20 Richard Biener <rguenther@suse.de>
18615 PR middle-end/81097
18616 * fold-const.c (split_tree): Fold to type before negating.
18618 2017-06-20 David Malcolm <dmalcolm@redhat.com>
18620 * diagnostic-show-locus.c
18621 (selftest::test_fixit_deletion_affecting_newline): New function.
18622 (selftest::diagnostic_show_locus_c_tests): Call it.
18624 2017-06-20 Andreas Schwab <schwab@suse.de>
18627 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
18630 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18632 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
18633 __ARM_FEATURE_COPROC according to support.
18635 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18637 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
18638 Rewritten to avoid overflow for > 32-bit pointers.
18641 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
18642 by removing enum keyword.
18643 (ubsan_type_descriptor): Likewise. Formatting fix.
18646 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
18647 splitter): Require TARGET_SSE2 in the condition.
18649 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
18652 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
18653 for doing vector set of SFmode on ISA 3.0.
18654 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
18655 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
18657 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
18658 SFmode value into a V4SF variable that was extracted from another
18659 V4SF variable without converting the element to double precision
18660 and back to single precision vector format.
18661 (vsx_insert_extract_v4sf_p9_2): Likewise.
18663 2017-06-19 Jakub Jelinek <jakub@redhat.com>
18665 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
18666 in UWHI to avoid undefined overflow.
18669 * ubsan.h (enum ubsan_encode_value_phase): New.
18670 (ubsan_encode_value): Change second argument to
18671 enum ubsan_encode_value_phase with default value of
18672 UBSAN_ENCODE_VALUE_GENERIC.
18673 * ubsan.c (ubsan_encode_value): Change second argument to
18674 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
18675 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
18676 create_tmp_var_raw instead of create_tmp_var and use a
18678 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
18679 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
18680 ubsan_encode_value callers.
18683 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
18684 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
18685 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
18687 2017-06-19 Richard Biener <rguenther@suse.de>
18689 PR middle-end/81118
18690 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
18691 estimates if we changed anything.
18693 2017-06-19 Richard Biener <rguenther@suse.de>
18695 PR tree-optimization/80887
18696 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
18697 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
18698 simplified lookups, then reset mprts_hook.
18699 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
18701 (try_to_simplify): Likewise.
18703 2017-06-19 Martin Liska <mliska@suse.cz>
18706 * gimplify.c (gimplify_switch_expr):
18707 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
18709 2017-06-19 Martin Liska <mliska@suse.cz>
18711 * doc/install.texi: Document that PGO runs in 4 stages.
18713 2017-06-19 Martin Liska <mliska@suse.cz>
18716 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
18717 to dispatcher function name.
18718 * multiple_target.c (replace_function_decl): New function.
18719 (create_dispatcher_calls): Redirect both edges and references.
18721 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
18723 * profile-count.c (profile_count::dump): Dump quality.
18724 (profile_count::differs_from_p): Update for unsigned val.
18725 * profile-count.h (profile_count_quality): New enum.
18726 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
18728 2017-06-19 Richard Biener <rguenther@suse.de>
18730 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
18731 struct function as arg.
18732 (estimate_numbers_of_iterations): Export overload with loop arg.
18733 (free_numbers_of_iterations_estimates_loop): Use an overload of
18734 free_numbers_of_iterations_estimates instead.
18735 * tree-cfg.c (remove_bb): Adjust.
18736 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
18737 * tree-parloops.c (gen_parallel_loop): Likewise.
18738 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
18740 (tree_unroll_loops_completely): Likewise.
18741 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
18742 Use an overload instead and export.
18743 (estimated_loop_iterations): Adjust.
18744 (max_loop_iterations): Likewise.
18745 (likely_max_loop_iterations): Likewise.
18746 (estimate_numbers_of_iterations): Take struct function as arg
18748 (loop_exits_before_overflow): Adjust.
18749 (free_numbers_of_iterations_estimates_loop): Use an overload.
18750 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
18751 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
18753 2017-06-19 Richard Biener <rguenther@suse.de>
18756 * ipa-prop.c (find_constructor_constant_at_offset): Handle
18757 RANGE_EXPR conservatively.
18759 2017-06-16 Carl Love <cel@us.ibm.com>
18761 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18762 definitions for vec_float, vec_float2, vec_floato,
18763 vec_floate built-ins.
18764 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
18765 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
18767 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
18768 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
18769 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
18770 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
18771 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
18772 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
18773 vec_floato): Add builtin defines.
18774 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
18775 Update the built-in documentation file for the new built-in
18778 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18780 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
18781 (mthumb): Mark as the negative of -marm.
18783 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18785 * doc/invoke.texi (ARM Options, -mcpu): Document supported
18787 (ARM Options, -mtune): Document that this accepts the same
18788 extension options as -mcpu.
18789 (ARM Options, -mfpu): Document addition of -mfpu=auto.
18791 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18793 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
18794 permitted extensions.
18796 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18798 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
18799 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
18800 (armv8-m.main): Add option +nodsp.
18801 * config/arm/arm-cpu-cdata.h: Regenerated.
18803 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18805 * config/arm/t-fuchsia: New file.
18806 * config.gcc (arm*-*-fuchsia*): Use it.
18808 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18810 * config/arm/t-symbian: Rewrite for new option infrastructure.
18812 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18814 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
18815 (MULTILIB_REQUIRED): Likewise.
18817 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18819 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
18820 (MULTILIB_RESUE): Likewise.
18821 (MULTILIB_MATCHES): Likewise.
18822 (MULTLIB_REQUIRED): Likewise.
18824 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18826 * config/arm/t-rtems: Rewrite for new option framework.
18828 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18830 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
18831 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
18832 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
18833 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
18834 * config/arm/t-multilib: ... here.
18835 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
18836 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
18837 armv7-a and armv8*-a when A-profile libraries have not been built.
18838 * config/arm/t-rmprofile: Rewrite.
18840 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18842 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
18843 with a backslash. Remove the backslash after substituting unescaped
18845 * doc/fragments.texi (MULTILIB_REUSE): Document it.
18847 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18849 * config.gcc: (arm*-*-*): When building a-profile libraries, force
18850 the driver to pass through the default setting of -mfloat-abi.
18851 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
18853 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
18854 (all_feat_combs): New rule.
18855 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
18857 * config/arm/t-aprofile: Rewrite.
18859 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18861 * config/arm/arm.h (FPUTYPE_AUTO): Define.
18862 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
18863 fpu is not specified by the user/command-line.
18864 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
18865 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
18866 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
18867 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
18868 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
18869 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
18871 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18873 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
18874 * config/arm/t-arm-elf: Rewritten.
18876 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18878 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
18879 have some floating-point instructions.
18880 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
18881 (TARGET_MAYBE_HARD_FLOAT): New macro.
18882 * config/arm/arm-builtins.c (arm_init_builtins): Use
18883 TARGET_MAYBE_HARD_FLOAT.
18884 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
18886 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18888 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
18889 (configargs.h): Include it.
18890 (arm_print_hint_for_fpu_option): New function.
18891 (arm_parse_fpu_option): New function.
18892 (candidate_extension): New class.
18893 (arm_canon_for_multilib): New function.
18894 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
18895 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
18896 (ARCH_CANONICAL_SPECS): New macro.
18897 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
18899 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18901 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
18902 are set after handling multilib fragments. Set target_cpu_default2
18905 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18907 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
18909 (arm*-*-*): Set target_cpu_default2 to a quoted string.
18910 * config/arm/parsecpu.awk (check_cpu): Validate any extension
18912 (check_arch): Likewise.
18913 * config/arm/arm.c (arm_configure_build_target): Handle
18914 TARGET_CPU_DEFAULT being a string constant. Scan any feature
18915 options in the default.
18917 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18919 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
18920 when an option is an alias of another.
18921 * config/arm/parsecpu.awk (optalias): New parser token.
18922 (gen_comm_data): Mark non-alias options as such. Emit entries
18923 for extension aliases.
18924 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
18925 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
18926 (armv6kz, armv6zk, armv6t2): Likewise.
18927 (armv7): Make vfpv3-d16 an alias.
18928 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
18930 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
18931 Sort in canonical order.
18932 (armv8-a): Sort in canonical order.
18933 (armv8.1-a, armv8.2-a): Likewise.
18934 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
18936 (cortex-a9): Sort in canonical order.
18937 * config/arm/arm.c (selftests.h): Include it.
18938 (arm_test_cpu_arch_data): New function.
18939 (arm_run_self_tests): New function.
18940 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
18941 (targetm): Move declaration to the end of the file.
18942 * arm-cpu-cdata.h: Regenerated.
18944 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18946 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
18947 call to target_mode_check describing the type of option passed.
18948 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
18949 (arm_target_thumb_only): Use arm_parse_arch_option_name or
18950 arm_parse_cpu_option_name to match parameters against list of
18952 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
18953 arm_arch_core_flags data structure.
18954 * config/arm/arm-cpu_cdata.h: Regenerated.
18956 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18958 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
18960 (arm_print_hint_for_cpu_option): Likewise.
18961 (arm_print_hint_for_arch_option): Likewise.
18962 (arm_parse_cpu_option_name): Likewise.
18963 (arm_parse_arch_option_name): Likewise.
18964 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
18965 of entries in the all_fpus list.
18966 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
18967 (arm_parse_cpu_option_name): Declare.
18968 (arm_parse_arch_option_name): Declare.
18969 (arm_parse_option_features): Declare.
18970 (arm_intialize_isa): Declare.
18971 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
18973 (gen_comm_data): ... here. Make definitions non-static.
18974 * config/arm/arm-cpu-data.h: Regenerated.
18975 * config/arm/arm-cpu-cdata.h: Regenerated.
18977 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18979 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
18980 (cpu_arch_extension): New structure.
18981 (cpu_arch_option, arch_option, cpu_option): New structures.
18982 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
18983 architecture types.
18984 (gen_data): Generate new format data tables.
18985 * config/arm/arm.c (cpu_tune): New structure.
18986 (cpu_option, processors): Delete.
18987 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
18988 (arm_print_hint_for_cpu_option): ... this and ...
18989 (arm_print_hint_for_arch_option): ... this.
18990 (arm_parse_arch_cpu_name): Delete. Replace with ...
18991 (arm_parse_cpu_option_name): ... this and ...
18992 (arm_parse_arch_option_name): ... this.
18993 (arm_unrecognized_feature): Change type of target parameter to
18995 (arm_parse_arch_cpu_features): Delete. Replace with ...
18996 (arm_parse_option_features): ... this.
18997 (arm_configure_build_target): Rework to use new configuration data
18999 (arm_print_tune_info): Rework for new configuration data tables.
19000 * config/arm/arm-cpu-data.h: Regenerated.
19001 * config/arm/arm-cpu.h: Regenerated.
19003 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19005 * Makefile.in (OBJS): Move sbitmap.o from here ...
19006 (OBJS-libcommon): ... to here.
19008 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19010 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19011 (ISA_ALL_CRYPTO): New macro.
19012 (ISA_ALL_SIMD): New macro
19013 (ISA_ALL_FP): New macro.
19014 * config/arm/arm.c (fpu_bitlist): Update initializer.
19015 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19017 (arm9e): Add fpu. Add option for nofp
19018 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19019 (arm926ej-s, arm1026ej-s): Likewise.
19020 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19021 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19022 neon-fp16, neon-vfpv4, nofp and nosimd.
19023 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19024 (cortex-a8): Add fpu. Add option for nofp.
19025 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19026 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19027 (cortex-r4f): Add fpu.
19028 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19029 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19031 (cortex-r8): Likewise.
19032 (cortex-m4): Add fpu. Add option for nofp.
19033 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19034 (cortex-a17.cortex-a7): Likewise.
19035 (cortex-a32): Add fpu. Add options for crypto and nofp.
19036 (cortex-a35, cortex-a53): Likewise.
19037 (cortex-a57): Add fpu. Add option for crypto.
19038 (cortex-a72, cortex-a73): Likewise.
19039 (exynos-m1): Likewise.
19040 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19041 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19042 (cortex-m33): Add fpu. Add option for nofp.
19043 * config/arm/arm-cpu-cdata.h: Regenerated
19044 * config/arm/arm-cpu-data.h: Regenerated.
19046 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19048 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19049 (armv5te, armv5tej): Likewise.
19050 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19051 (armv7): Add options fp and vfpv3-d16.
19052 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19053 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19055 (armv7ve): Likewise.
19056 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19057 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19058 (armv8-a): Add nocrypto option.
19059 (armv8.1-a, armv8.2-a): Likewise.
19060 (armv8-m.main): add options fp, fp.dp and nofp.
19062 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19064 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19066 (armv8-a+crc): Delete.
19067 (armv8.1-a): Add options simd, crypto and nofp.
19068 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19069 (armv8.2-a+fp16): Delete.
19070 (armv8-m.main): Add option dsp.
19071 (armv8-m.main+dsp): Delete.
19072 (cortex-a8): Add fpu. Add nofp option.
19073 (cortex-a9): Add fpu. Add nofp and nosimd options.
19074 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19075 link to main cpu and architecture data structures.
19076 (gen_comm_data): Only put isa attributes from the main architecture
19078 (option): New statement for architecture and CPU entries.
19079 * arm.c (struct cpu_option): New structure.
19080 (struct processors): Add entry for options.
19081 (arm_unrecognized_feature): New function.
19082 (arm_parse_arch_cpu_name): Ignore any characters after the first
19084 (arm_parse_arch_cpu_feature): New function.
19085 (arm_configure_build_target): Separate out any CPU and architecture
19086 features and parse separately. Don't error out if -mfpu=auto is
19087 used with only an architecture string.
19088 (arm_print_asm_arch_directives): New function.
19089 (arm_file_start): Call it.
19090 * config/arm/arm-cpu-cdata.h: Regenerated.
19091 * config/arm/arm-cpu-data.h: Likewise.
19092 * config/arm/arm-tables.opt: Likewise.
19094 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19096 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19097 assembler when it is not -mfpu=auto.
19099 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19101 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19102 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19103 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19104 (ASM_CPU_SPEC): Rewrite.
19105 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19106 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19107 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19108 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19109 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19110 copied string is NUL-terminated. Also strip any characters prefixed
19112 (arm_rewrite_selected_arch): New function.
19113 (arm_rewrite_march): New function.
19115 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19117 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19118 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19119 (march, mcpu, mtune): Convert to string-based options.
19120 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19121 (arm_parse_arch_cpu_name): New function.
19122 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19123 identify selected architecture or CPU.
19124 (arm_option_save): New function.
19125 (TARGET_OPTION_SAVE): Redefine.
19126 (arm_option_restore): Restore string options.
19127 (arm_option_print): Print string options.
19129 2017-06-16 Martin Sebor <msebor@redhat.com>
19131 PR tree-optimization/80933
19132 PR tree-optimization/80934
19133 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19134 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19135 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19136 (gimple_fold_builtin): Call them.
19138 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19140 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19141 as unlikely; update profile.
19143 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19145 * predict.c (force_edge_cold): Handle declaring edges impossible
19148 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19150 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19152 (try_unroll_loop_completely): Fix reporting.
19154 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19156 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19158 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19161 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19162 if given a non-constant argument for an intrinsic which requires a
19165 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19167 * profile.c (compare_freqs): New function.
19168 (branch_prob): Sort edge list.
19169 (find_spanning_tree): Assume that the list is priority sorted.
19171 2017-06-16 Richard Biener <rguenther@suse.de>
19173 PR tree-optimization/81090
19174 * passes.def (pass_record_bounds): Remove.
19175 * tree-pass.h (make_pass_record_bounds): Likewise.
19176 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19177 make_pass_record_bounds): Likewise.
19178 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19179 not free niter estimates at the beginning but at the end.
19180 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19182 2017-06-16 Richard Biener <rguenther@suse.de>
19184 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19185 initializer to workaround ICE in host GCC 4.8.
19187 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19189 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19191 (clone_inlined_nodes): Update.
19193 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19195 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19196 prefetch settings, and enable prefetching by default at -O3.
19198 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19200 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19201 Set flag_prefetch_loop_arrays according to tuning data.
19203 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19205 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19206 New tune structure.
19207 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19208 [Unrelated to main purpose of the patch] Place the pointer field last
19209 to enable type checking errors when tune structure are wrongly merged.
19210 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19211 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19212 (thunderx2t99_prefetch_tune): New tune constants.
19213 (tune_params *_tunings): Update all tunings (no functional change).
19214 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19215 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19216 from tunings structures.
19218 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19221 * ubsan.c (instrument_null): Add T argument, use it instead
19222 of computing it based on IS_LHS.
19223 (instrument_object_size): Likewise.
19224 (pass_ubsan::execute): Adjust instrument_null and
19225 instrument_object_size callers to pass gimple_get_lhs or
19226 gimple_assign_rhs1 result to it. Use instrument_null instead of
19227 calling get_base_address and instrument_mem_ref. Handle
19228 aggregate call arguments for object-size sanitization.
19230 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19232 PR tree-optimization/81089
19233 * tree-vrp.c (is_masked_range_test): Validate operands of
19236 2017-06-15 Martin Sebor <msebor@redhat.com>
19239 * dumpfile.c (dump_register): Avoid calling memset to initialize
19240 a class with a default ctor.
19241 * gcc.c (struct compiler): Remove const qualification.
19242 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19243 * hash-table.h: Ditto.
19244 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19246 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19247 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19249 * params.h (struct param_info): Make struct members non-const.
19250 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
19251 with default initialization.
19252 * vec.h (vec_copy_construct, vec_default_construct): New helper
19254 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
19255 with vec_copy_construct.
19256 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
19257 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
19258 * doc/invoke.texi (-Wclass-memaccess): Document.
19260 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19262 * emit-rtl.h (is_leaf): Update comment about local
19263 register allocator.
19265 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19268 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
19269 for a variable to have a section before checking if the section has a
19271 Set section to.persistent if persistent attribute is set.
19272 Warn if .persistent attribute is used on an automatic variable.
19274 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
19276 PR rtl-optimization/80474
19277 * reorg.c (update_block): Do not ignore instructions in a delay slot.
19279 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
19281 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
19284 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
19286 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
19287 (casesi): Emit bounds checking as RTL.
19288 (casesi_internal_mips16_<mode>): Remove bounds checking.
19290 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
19292 * config/xtensa/xtensa.c (xtensa_option_override): Append
19293 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
19294 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
19295 xtensa_doloop_hooks): Define unconditionally.
19296 (xtensa_reorg_loops): Only call reorg_loops in the presence of
19298 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
19299 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
19300 for it in xtensa_option_override.
19301 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
19302 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
19304 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
19306 * doc/cppopts.texi: Document '-' special value to -MF.
19308 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
19310 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
19311 (cortex_a53_fconst): Likewise.
19312 (cortex_a53_fpmul): Likewise.
19313 (cortex_a53_f_load_64): Likewise.
19314 (cortex_a53_f_load_many): Likewise.
19315 (cortex_a53_advsimd_alu): Likewise.
19316 (cortex_a53_advsimd_alu_q): Likewise.
19317 (cortex_a53_advsimd_mul): Likewise.
19318 (cortex_a53_advsimd_mul_q): Likewise.
19319 (fpmac bypass): Add new bypass for fpmac-fpmac case.
19320 Add missing fmul, r2f_cvt and fconst cases.
19322 2017-06-14 Richard Biener <rguenther@suse.de>
19324 PR middle-end/81088
19325 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
19327 (fold_binary_loc): When associating do not treat pre-existing
19328 TREE_OVERFLOW on literal constants as a reason to allow
19329 TREE_OVERFLOW on associated literal constants.
19331 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19333 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
19334 (MASK_FEATURES): New macro.
19335 * config/sparc/sparc.c (sparc_option_override): Remove the special
19336 handling of -mfpu and generalize it to all MASK_FEATURES switches.
19338 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19340 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
19341 a division of 0 if non-call exceptions are enabled.
19343 2017-06-14 Andrew Pinski <apinski@cavium.com>
19344 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19347 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
19348 Improve vector initialization code gen for only variable case.
19350 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19352 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
19354 2017-06-14 Richard Biener <rguenther@suse.de>
19356 PR tree-optimization/81083
19357 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
19360 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19362 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
19363 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
19364 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
19365 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
19366 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
19367 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
19369 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19371 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
19372 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
19374 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19376 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
19378 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19380 * config/rs6000/t-rtems: Don't handle SPE.
19382 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19384 * config/rs6000/t-linux: Don't handle SPE.
19386 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19388 * config/rs6000/eabispe.h: Delete file.
19390 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19392 * config/rs6000/t-spe: Delete file.
19394 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19396 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
19397 (rs6000_legitimate_offset_address_p): Return false for anything in
19398 V2SImode or V2SFmode.
19400 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19402 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
19403 except V2SF and V2SI. Rearrange the vector modes, and add comments.
19404 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
19406 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
19407 (rs6000_legitimate_offset_address_p): Ditto.
19408 (rs6000_emit_move): Ditto.
19409 (rs6000_init_builtins): Remove V4HI_type_node.
19411 2017-06-13 Martin Liska <mliska@suse.cz>
19414 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
19415 (gate_asan): Likewise.
19416 * asan.h (asan_no_sanitize_address_p): Remove the function.
19417 (sanitize_flags_p): New function.
19418 * builtins.def: Fix coding style.
19419 * common.opt: Use renamed enum value.
19420 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
19421 * doc/extend.texi: Document no_sanitize attribute.
19422 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
19423 to SANITIZE_UNDEFINED_NONDEFAULT.
19424 * gcc.c (sanitize_spec_function): Use the renamed enum value.
19425 * gimple-fold.c (optimize_atomic_compare_exchange_p):
19426 Use sanitize_flags_p.
19427 * gimplify.c (gimplify_function_tree): Likewise.
19428 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
19429 * opts.c (parse_no_sanitize_attribute): New function.
19430 (common_handle_option): Use renamed enum value.
19431 * opts.h (parse_no_sanitize_attribute): Declare.
19432 * tree.c (sanitize_flags_p): New function.
19433 * tree.h: Declared here.
19434 * tsan.c: Use sanitize_flags_p.
19435 * ubsan.c (ubsan_expand_null_ifn): Likewise.
19436 (instrument_mem_ref): Likewise.
19437 (instrument_bool_enum_load): Likewise.
19438 (do_ubsan_in_current_function): Remove the function.
19439 (pass_ubsan::execute): Use sanitize_flags_p.
19440 * ubsan.h: Remove do_ubsan_in_current_function
19441 * tree-cfg.c (print_no_sanitize_attr_value): New function.
19442 (dump_function_to_file): Use it here.
19444 2017-06-13 Martin Jambor <mjambor@suse.cz>
19446 PR tree-optimization/80803
19447 PR tree-optimization/81063
19448 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
19449 (propagate_subaccesses_across_link): Enqueue subtree whenever
19450 necessary instead of relying on the caller.
19452 2017-06-13 Martin Jambor <mjambor@suse.cz>
19454 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
19455 that have a first_link.
19456 (sort_and_splice_var_accesses): Do not check first_link before
19458 (subtree_mark_written_and_enqueue): Likewise.
19459 (propagate_all_subaccesses): Likewise and do not stop at first
19460 parent with a first_link.
19462 2017-06-13 Martin Jambor <mjambor@suse.cz>
19464 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
19467 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19469 * match.pd: New pattern.
19471 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19473 * tree-vrp.c (is_masked_range_test): New function.
19474 (register_edge_assert_for): Determine ranges for
19477 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19479 PR tree-optimization/67328
19480 * fold-const.c (maskable_range_p): New function.
19481 (build_range_check): Generate bittests if possible.
19483 2017-06-13 Martin Liska <mliska@suse.cz>
19485 * gimple-pretty-print.c (dump_probability): Add new argument.
19486 (dump_edge_probability): Dump both probability and count.
19487 (dump_gimple_label): Likewise.
19488 (dump_gimple_bb_header): Likewise.
19490 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
19493 * config/avr/avr-devices.c: Fix indentation.
19494 * config/avr/gen-avr-mmcu-specs.c: Dito.
19496 2017-06-13 Richard Biener <rguenther@suse.de>
19498 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
19499 instead get vector type from stmt_info.
19500 (vectorizable_reduction): Adjust. Remove dead code.
19502 2017-06-13 Richard Biener <rguenther@suse.de>
19504 PR middle-end/81065
19505 * fold-const.c (extract_muldiv_1): Remove bogus distribution
19506 case of C * (x * C2 + C3).
19507 (fold_addr_of_array_ref_difference): Properly fold index difference.
19509 2017-06-12 David S. Miller <davem@davemloft.net>
19512 * config/sparc/sparc.md (return expander): Emit frame blockage if
19513 function uses alloca.
19515 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
19517 * combine.c (make_field_assignment): Check len rather than the mode
19518 precision when calling force_to_mode.
19520 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
19522 Support multilibs and devices that see flash in RAM address range.
19525 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
19526 (avr_mcu_t) <flash_pm_offset>: New field.
19527 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
19528 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
19529 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
19530 (AVR_TINY_PM_OFFSET): Remove macro.
19531 * config/avr/avr.opt (-mshort-calls): New option.
19532 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
19533 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
19534 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
19535 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
19536 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
19537 instead of avr_arch->have_jmp_call.
19538 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
19539 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
19540 avr_arch->flash_pm_offset to define.
19541 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
19542 new field flash_pm_offset. Add entry for avrxmega3.
19543 (avr_texinfo): Add entry for avrxmega3.
19544 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
19545 attiny212, attiny214,
19546 attiny412, attiny414, attiny416, attiny417,
19547 attiny814, attiny816, attiny817,
19548 attiny1614, attiny1616, attiny1617,
19549 attiny3214, attiny3216, attiny3217.
19550 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
19551 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
19552 (avr_print_operand_address) [AVR_TINY]: Same.
19553 (avr_asm_init_sections) <readonly_data_section>: Only patch
19554 callback if avr_arch->flash_pm_offset = 0.
19555 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
19556 for rodata if avr_arch->flash_pm_offset != 0.
19557 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
19558 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
19559 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
19560 (m_options): Append opt_rcall.
19561 (m_dirnames): Append dir_rcall.
19562 * config/avr/t-multilib: Regenerate.
19564 * configure.ac [target=avr]: Check whether avrxmega3 default
19565 linker description file works as needed.
19566 * configure: Regenerate.
19567 * doc/avr-mmcu.texi: Regenerate.
19568 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
19569 <__AVR_ARCH__>: Document avrxmega3 and 103.
19570 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
19571 <__AVR_SHORT_CALLS__>: Document it.
19572 <__AVR_PM_BASE_ADDRESS__>: Document it.
19573 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
19574 (AVR Variable Attributes) <progmem>: Document this is
19575 not needed for avrxmega3.
19576 (AVR Named Address Spaces) <__flash>: Dito.
19578 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
19580 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
19582 2017-06-12 Doug Rupp <rupp@adacore.com>
19584 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
19585 Append vxworks-stdint.h to the tm_file list.
19586 * config/vxworks-stdint.h: New file.
19588 2017-06-12 Martin Liska <mliska@suse.cz>
19590 PR tree-optimization/81041
19591 * tree-profile.c (gimple_gen_ic_func_profiler):
19592 Create an extra BB in profile-generate
19593 (gimple_gen_time_profiler): Likewise.
19595 2017-06-12 Jakub Jelinek <jakub@redhat.com>
19597 PR tree-optimization/81003
19598 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
19599 (update_range_test): Use it instead of force_gimple_operand_gsi.
19601 2017-06-12 Richard Biener <rguenther@suse.de>
19603 PR tree-optimization/81053
19604 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
19605 with backedge value not defined in loop. Simplify def stmt
19608 2017-06-11 Tom de Vries <tom@codesourcery.com>
19611 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
19613 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
19614 nvptx_cannot_force_const_mem.
19616 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19618 * opts.c (finish_options): Move test for flag_split_stack after
19619 it has been initialized.
19621 2017-06-11 Jason Merrill <jason@redhat.com>
19623 * tree.h (id_equal): New.
19624 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
19625 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
19626 instead of strcmp of IDENTIFIER_POINTER.
19628 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19630 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
19631 (mark_all_inlined_calls_cdtor): Fix formating.
19632 (inline_transform): Rescale profile before inlining.
19634 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19636 * cgraph.h (cgraph_edge::clone): Update prototype.
19637 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
19638 (cgraph_node::create_clone): Update.
19639 (cgraph_node::create_version_clone): Update.
19640 * tree-inline.c (copy_bb): Update.
19641 (expand_call_inline): Update.
19643 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
19645 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
19646 factored out from ...
19647 (rs6000_emit_prologue): ... here.
19649 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
19651 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
19652 factored out from ...
19653 (rs6000_emit_prologue): ... here.
19655 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19657 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
19659 (handle_missing_profiles): Fix computation of tp_first_run.
19660 (counts_to_freqs): Do not touch freqs when count is 0.
19662 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19664 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
19667 2017-06-10 Tom de Vries <tom@codesourcery.com>
19669 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
19670 attributes): Document signal effective target.
19672 2017-06-10 Tom de Vries <tom@codesourcery.com>
19674 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
19675 Document effective target stack_size.
19677 2017-06-09 David Malcolm <dmalcolm@redhat.com>
19679 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
19680 to the edit_context if they can be auto-applied.
19682 2017-06-9 Ian Lance Taylor <iant@golang.org>
19684 * opts.c (finish_options): If -fsplit-stack, disable implicit
19685 -forder-blocks-and-partition.
19686 * doc/invoke.texi (Optimize Options): Document that when using
19687 -fsplit-stack -forder-blocks-and-partition is not implicitly
19690 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19692 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
19693 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
19694 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
19695 * builtins.def (abort, trap, unreachable): Declare cold.
19696 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
19697 * tree-core.h (ECF_COLD): New.
19698 * tree.c (set_call_expr_flags): Handle ECF_COLD.
19699 (build_common_builtin_nodes): Mark unreachable and abort as cold.
19701 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19703 * predict.c (unlikely_executed_stmt_p): Cleanup.
19705 2017-06-09 Richard Biener <rguenther@suse.de>
19707 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
19708 model if the ref is always written to.
19710 2017-06-09 Tamar Christina <tamar.christina@arm.com>
19712 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
19714 2017-06-09 Tamar Christina <tamar.christina@arm.com>
19716 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
19719 2017-06-09 Tom de Vries <tom@codesourcery.com>
19722 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
19723 "target cannot support label values" when encountering LABEL_REF.
19725 2017-06-09 Martin Liska <mliska@suse.cz>
19727 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
19728 (gimple_gen_ic_func_profiler): Emit direct comparison
19729 of __gcov_indirect_call_callee with NULL.
19730 (gimple_gen_time_profiler): Change probability from
19731 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
19733 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19735 * profile.c (edge_gcov_counts): Turn to pointer.
19736 (compute_branch_probabilities, compute_branch_probabilities): Update.
19737 (branch_prob): Do not clear edge_gcov_count.
19738 * profile.h (edge_gcov_counts): Turn to pointer.
19739 (edge_gcov_count): Update.
19741 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19743 * gimple.h (gimple_check_failed): Mark cold.
19745 2017-06-09 Richard Biener <rguenther@suse.de>
19747 PR tree-optimization/66623
19748 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
19749 refactor check_reduction into two parts, properly computing
19750 whether we have to check reduction validity for outer loop
19753 2017-06-09 Richard Biener <rguenther@suse.de>
19755 PR tree-optimization/79483
19756 * graphite-scop-detection.c (order): New global.
19757 (get_order): Compute bb to order mapping that satisfies code
19758 generation constraints.
19759 (cmp_pbbs): New helper.
19760 (build_scops): Start domwalk at entry block, sort generated
19763 2017-06-09 Richard Biener <rguenther@suse.de>
19765 PR middle-end/81007
19766 * ipa-polymorphic-call.c
19767 (ipa_polymorphic_call_context::restrict_to_inner_class):
19768 Skip FIELD_DECLs with error_mark_node type.
19769 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
19772 2017-06-09 Martin Liska <mliska@suse.cz>
19774 * predict.c (struct branch_predictor): New struct.
19775 (test_prediction_value_range): New test.
19776 (predict_c_tests): New function.
19777 * selftest-run-tests.c (selftest::run_tests): Run the function.
19778 * selftest.h: Declare new tests.
19780 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
19783 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
19784 gen_add3_insn did not fail.
19785 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
19786 r0, construct that number in a temporary reg and add that reg to r0.
19787 If asked to put the result in r0 as well, fail.
19789 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
19791 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19792 for early expansion of vec_eqv.
19794 2017-06-08 Jakub Jelinek <jakub@redhat.com>
19796 PR middle-end/81005
19797 * ubsan.c (instrument_null): Avoid pointless code temporary.
19798 (pass_ubsan::execute): Instrument aggregate arguments of calls.
19800 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
19804 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
19807 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
19810 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19812 * predict.c (unlikely_executed_edge_p): Move ahead.
19813 (probably_never_executed_edge_p): Use it.
19815 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
19817 PR middle-end/79988
19818 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
19819 gimple_call_builtin_p call.
19821 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19823 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
19824 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
19825 rtl_check_failed_type2, rtl_check_failed_code1,
19826 rtl_check_failed_code2, rtl_check_failed_code_mode,
19827 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
19828 rtvec_check_failed_bounds, rtl_check_failed_flag,
19829 _fatal_insn_not_found, _fatal_insn): Likewise.
19830 * tree.h (tree_contains_struct_check_failed,
19831 tree_check_failed, tree_not_check_failed,
19832 tree_class_check_failed, tree_range_check_failed,
19833 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
19834 tree_vec_elt_check_failed, phi_node_elt_check_failed,
19835 tree_operand_check_failed, omp_clause_check_failed,
19836 omp_clause_operand_check_failed, omp_clause_range_check_failed):
19839 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19841 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
19842 flag_branch_probabilities.
19843 * ipa-inline.c (edge_badness): Likewise.
19844 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
19845 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
19846 * predict.c (maybe_hot_frequency_p): Likewise.
19847 (probably_never_executed): Likewise.
19848 * sched-ebb.c (schedule_ebbs): Likewise.
19849 * sched-rgn.c (find_single_block_region): Likewise.
19850 * tracer.c (tail_duplicate): Likewise.
19852 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19854 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
19855 longer requires x_flag_profile_use.
19857 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19859 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
19860 instead of flag_reorder_blocks_and_partition.
19861 * dbxout.c (dbxout_function_end): Likewise.
19862 * dwarf2out.c (gen_subprogram_die): Likewise.
19863 * haifa-sched.c (sched_create_recovery_edges): Likewise.
19864 * hw-doloop.c (reorg_loops): Likewise.
19865 * varasm.c (assemble_start_function,
19866 assemble_end_function): Likewise.
19867 (decide_function_section): Do not check for
19868 flag_reorder_blocks_and_partition.
19870 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
19872 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19874 (chkp_get_hard_register_fake_addr_expr): Ditto.
19875 (chkp_build_addr_expr): Add check for hard reg case.
19876 (chkp_parse_array_and_component_ref): Ditto.
19877 (chkp_find_bounds_1): Ditto.
19878 (chkp_process_stmt): Don't generate bounds store for
19881 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19883 * predict.c (maybe_hot_bb_p): Do not check profile status.
19884 (maybe_hot_edge_p): Likewise.
19885 (probably_never_executed): Check for zero counts even if profile
19887 (unlikely_executed_edge_p): New function.
19888 (unlikely_executed_stmt_p): New function.
19889 (unlikely_executed_bb_p): New function.
19890 (set_even_probabilities): Use unlikely predicates.
19891 (combine_predictions_for_bb): Likewise.
19892 (predict_paths_for_bb): Likewise.
19893 (predict_paths_leading_to_edge): Likewise.
19894 (determine_unlikely_bbs): New function.
19895 (estimate_bb_frequencies): Use it.
19896 (compute_function_frequency): Use zero counts even if profile is
19898 * profile-count.h: Fix typo.
19900 2017-08-08 Julia Koval <julia.koval@intel.com>
19902 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
19903 _mm512_mask_cvtsepi16_storeu_epi8,
19904 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
19905 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
19906 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
19907 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
19908 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
19909 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
19910 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
19911 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
19912 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
19913 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
19914 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
19915 __builtin_ia32_pmovuswb256mem_mask,
19916 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
19917 __builtin_ia32_pmovwb512mem_mask): New builtins.
19919 2017-08-08 Julia Koval <julia.koval@intel.com>
19921 PR target/73350,80862
19922 * config/i386/subst.md (round): Fix round pattern.
19923 * config/i386/i386.c (ix86_erase_embedded_rounding):
19924 Fix erasing rounding for the fixed pattern.
19926 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19928 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
19930 2017-06-08 Martin Liska <mliska@suse.cz>
19932 PR gcov-profile/80911
19933 * gcov.c (block_info::block_info): New constructor.
19935 2017-06-07 Carl Love <cel@us.ibm.com>
19937 * config/rs6000/rs6000-c: The return type of the following
19938 built-in functions was implemented as int not long long. Fix sign
19939 of return value for the unsigned version of vec_mulo and vec_mule.
19940 vector unsigned long long vec_bperm (vector unsigned long long,
19941 vector unsigned char)
19942 vector signed long long vec_mule (vector signed int,
19944 vector unsigned long long vec_mule (vector unsigned int,
19945 vector unsigned int)
19946 vector signed long long vec_mulo (vector signed int,
19948 vector unsigned long long vec_mulo (vector unsigned int,
19949 vector unsigned int)
19950 * doc/extend.texi: Fix the documentation for the built-in
19953 2017-06-07 Carl Love <cel@us.ibm.com>
19956 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
19959 2017-06-07 Carl Love <cel@us.ibm.com>
19961 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
19962 support, Generate doublehv for signed int/float for BE case only.
19964 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
19966 * doc/invoke.texi (mcx16): Rewrite.
19968 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19970 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
19971 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
19972 *mov<mode>_softfloat, and an anonymous splitter): Use
19973 nonimmediate_operand instead of rs6000_nonimmediate_operand.
19975 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19977 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
19979 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
19980 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
19981 (rs6000_debug_reg_global): Adjust.
19982 (rs6000_init_hard_regno_mode_ok): Adjust.
19983 (rs6000_dbx_register_number): Adjust.
19984 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
19985 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
19986 Remove SPE_ACC and SPEFSCR.
19987 (REG_ALLOC_ORDER): Ditto.
19988 (FRAME_POINTER_REGNUM): Change to 111.
19989 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
19990 (REG_CLASS_NAMES): Ditto.
19991 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
19992 (REGISTER_NAMES): Ditto.
19993 (ADDITIONAL_REG_NAMES): Ditto.
19994 (rs6000_reg_names): Ditto.
19995 * config/rs6000/rs6000.md: Renumber some register number
19998 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20000 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20002 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20003 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20005 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20006 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20007 Delete the SPE high registers.
20008 (REG_ALLOC_ORDER): Ditto.
20009 (enum reg_class): Remove SPE_HIGH_REGS.
20010 (REG_CLASS_NAMES): Ditto.
20011 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20012 (REGISTER_NAMES): Ditto.
20013 (rs6000_reg_names): Ditto.
20014 * doc/tm.texi.in: Remove SPE as example.
20015 * doc/tm.texi: Regenerate.
20017 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20019 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20020 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20021 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20022 * config/rs6000/rs6000.md (type): Remove "brinc".
20024 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20026 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20027 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20028 * config/rs6000/linuxspe.h: Delete file.
20029 * config/rs6000/rs6000.md: Don't include spe.md.
20030 * config/rs6000/spe.h: Delete file.
20031 * config/rs6000/spe.md: Delete file.
20032 * config/rs6000/t-rs6000: Remove spe.md.
20034 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20036 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20037 (reg_or_none500mem_operand): Delete.
20038 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20039 instead of reg_or_none500mem_operand.
20041 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20043 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20044 handling of SPE flags.
20045 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20047 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20049 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20051 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20052 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20053 paired_divv2sf3): Similar.
20054 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20055 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20056 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20058 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20059 Rename the paired_* instruction patterns.
20060 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20062 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20063 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20064 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20065 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20066 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20067 PAIRED_VECTOR_MODE.
20068 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20069 (spe_func_has_64bit_regs_p): Delete.
20070 (spe_expand_predicate_builtin): Delete.
20071 (spe_expand_evsel_builtin): Delete.
20072 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20073 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20074 (invalid_e500_subreg): Delete.
20075 (rs6000_legitimize_address): Always force_reg op2 as well, for
20076 paired single memory accesses.
20077 (rs6000_member_type_forces_blk): Delete.
20078 (rs6000_spe_function_arg): Delete.
20079 (rs6000_expand_unop_builtin): Delete SPE handling.
20080 (rs6000_expand_binop_builtin): Ditto.
20081 (spe_expand_stv_builtin): Delete.
20082 (bdesc_2arg_spe): Delete.
20083 (spe_expand_builtin): Delete.
20084 (spe_expand_predicate_builtin): Delete.
20085 (spe_expand_evsel_builtin): Delete.
20086 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20087 (spe_init_builtins): Delete.
20088 (spe_func_has_64bit_regs_p): Delete.
20089 (savres_routine_name): Delete "info" parameter. Adjust callers.
20090 (rs6000_emit_stack_reset): Ditto.
20091 (rs6000_dwarf_register_span): Delete.
20092 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20093 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20094 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20096 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20098 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20099 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20100 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20101 mulv2sf3, divv2sf3): Delete expanders.
20103 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20105 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20107 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20109 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20110 * config/rs6000/rs6000.c: Ditto.
20112 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20114 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20115 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20116 comparison_operator.
20118 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20120 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20121 * config/rs6000/rs6000.opt: Ditto.
20122 * config/rs6000/t-rtems: Ditto.
20124 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20126 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20127 TARGET_E500_SINGLE by 0, simplify.
20128 * config/rs6000/rs6000.c: Ditto.
20129 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20130 (spe_build_register_parallel): Delete.
20131 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20132 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20133 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20134 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20135 (E500_CONVERT): Delete.
20136 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20138 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20140 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20141 * config/rs6000/dfp.md: Ditto.
20142 (negdd2, *negdd2_fpr): Merge.
20143 (absdd2, *absdd2_fpr): Merge.
20144 (negtd2, *negtd2_fpr): Merge.
20145 (abstd2, *abstd2_fpr): Merge.
20146 * config/rs6000/e500.h: Delete file.
20147 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20148 TARGET_FPRS by 1 and simplify.
20149 * config/rs6000/rs6000-c.c: Ditto.
20150 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20151 TARGET_DF_SPE by 0.
20152 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20154 * config/rs6000/rs6000.md: Ditto.
20155 (floatdidf2, *floatdidf2_fpr): Merge.
20156 (move_from_CR_gt_bit): Delete.
20157 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20158 (E500_CR_IOR_COMPARE): Delete.
20159 (All patterns that require !TARGET_FPRS): Delete.
20160 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20162 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20164 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20166 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20168 * graphds.c (add_edge): Intitialize edge's attached data.
20169 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20170 pointer parameter. Call pointed function on each edge during
20171 graph traversing. Skip traversing the edge when the function
20173 (graphds_dfs, graphds_scc): Ditto.
20174 (for_each_edge): New parameter. Pass the new parameter to callback
20176 * graphds.h (skip_edge_callback): New function pointer type.
20177 (graphds_dfs, graphds_scc): New function pointer parameter.
20178 (graphds_edge_callback, for_each_edge): New parameter.
20180 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20182 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20183 out code checking if runtime alias check is possible to below ...
20184 Call the new function.
20185 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20186 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20188 2017-06-07 Marek Polacek <polacek@redhat.com>
20191 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20192 TYPE_OVERFLOW_WRAPS check.
20194 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20196 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20197 if versioning is required.
20198 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20199 peeling with the check for versioning.
20201 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20203 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20204 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20205 Set true to new parameter if new ssa variable is defined.
20206 (vect_gen_vector_loop_niters): Refactor. Set range information
20207 for the new vector loop bound variable.
20208 (vect_do_peeling): Ditto.
20210 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20212 * tree-affine.c (ssa.h): Include header file.
20213 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20214 has wrapping overflow behavior.
20216 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20218 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20220 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20222 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20223 (tree_to_aff_combination): ... here.
20225 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20227 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20228 reg_pressure model function.
20229 (ivopts_global_cost_for_size): Delete.
20230 (determine_set_costs, iv_ca_recount_cost): Call new model function
20231 ivopts_estimate_reg_pressure.
20233 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20235 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20236 expensive than udiv. Remove floating point cases from mod.
20238 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20240 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20241 Increase idiv cost.
20243 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20245 * config/aarch64/aarch64.md
20246 (copysignsf3): Fix mask generation.
20248 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20250 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
20252 (class dump_manager): Add register_dumps method.
20253 * dumpfile.c: Include langhooks.h.
20254 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
20255 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
20256 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
20257 (dump_manager::dump_register): Start with 512 entries instead of 32.
20258 (dump_manager::register_dumps): New method.
20259 * toplev.c (general_init): Instead of invoking register_dumps
20260 langhook, invoke register_dumps method on the dump manager.
20261 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
20264 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
20266 * doc/md.texi: Clarify the restrictions on a define_insn condition.
20267 Say that # requires an associated define_split to exist, and that
20268 the define_split must be suitable for use after register allocation.
20270 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20272 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
20273 (compute_outgoing_frequencies): Also initialize zero counts.
20274 (find_many_sub_basic_blocks): Do not produce uninitialized profile
20275 around loops; preserve more of profile when nothing changes.
20277 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
20279 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
20281 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
20282 * config/arm/arm-cpu-cdata.h: Regenerate.
20283 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
20284 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
20285 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
20286 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
20287 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
20289 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
20291 * config/arm/t-rmprofile: Likewise.
20292 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
20294 2017-06-06 David S. Miller <davem@davemloft.net>
20297 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
20298 blockage if function uses alloca.
20300 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20302 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
20303 New "uid" fields to hold pretty-print IDs of group and ref.
20304 Memory references are now identified as <group_id>:<ref_id>
20305 instead of using [random] addresses.
20306 (dump_mem_details): Simplify, no functional change.
20307 (dump_mem_ref): Simplify and make output more concise.
20308 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
20309 (find_or_create_group): Initialize group uid.
20310 (record_ref): Initialize ref uid. Improve debug output.
20311 (prune_group_by_reuse, should_issue_prefetch_p,)
20312 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
20313 (mark_nontemporal_store, determine_loop_nest_reuse):
20314 Improve debug output.
20316 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20318 * dbgcnt.def (prefetch): New debug counter.
20319 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
20320 (schedule_prefetches): Stop issueing prefetches if debug counter
20323 2017-06-06 Tom de Vries <tom@codesourcery.com>
20325 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
20326 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
20328 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20330 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
20331 Use aarch64_reg_or_zero predicate for operand 4.
20332 (aarch64_compare_and_swap<mode> define_insn_and_split):
20333 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
20334 (aarch64_store_exclusive<mode>): Likewise for operand 2.
20336 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20338 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
20339 (arm_compute_save_core_reg_mask): This.
20340 (thumb1_compute_save_reg_mask): Rename into ...
20341 (thumb1_compute_save_core_reg_mask): This.
20342 (arm_compute_save_reg0_reg12_mask): Adapt comment.
20343 (arm_compute_frame_layout): Likewise.
20345 2017-06-06 Richard Biener <rguenther@suse.de>
20347 PR tree-optimization/80974
20348 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
20349 keep or clear leaders SSA info.
20351 2017-06-06 Tom de Vries <tom@codesourcery.com>
20353 * config/nvptx/nvptx.c (split_mode_p): New function.
20354 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
20356 2017-06-06 Tom de Vries <tom@codesourcery.com>
20358 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
20360 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20363 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
20366 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20368 * shrink-wrap.c (handle_simple_exit): Update profile.
20369 (try_shrink_wrapping): Upate profile.
20371 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20373 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
20374 (tree_guess_outgoing_edge_probabilities): New.
20375 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
20376 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
20378 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20380 * ipa-split.c (split_function): Initialize return bb profile.
20382 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20384 * profile.c (compute_branch_probabilities): Also initialize
20385 EXIT_BLOCK profile.
20387 2017-06-06 Richard Biener <rguenther@suse.de>
20389 PR tree-optimization/80928
20390 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
20391 (vect_analyze_loop_operations): Properly guard analysis for
20393 (vect_transform_loop): Likewise.
20394 (vect_analyze_loop_2): Also reset SLP type on PHIs.
20395 (vect_model_induction_cost): Do not cost for pure SLP.
20396 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
20397 of induction in inner loop vectorization.
20398 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
20399 (vect_get_and_check_slp_defs): Handle vect_induction_def.
20400 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
20402 (vect_analyze_slp_cost_1): Cost induction.
20403 (vect_detect_hybrid_slp_stmts): Handle PHIs.
20404 (vect_get_slp_vect_defs): Likewise.
20405 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
20406 (vect_transform_stmt): Handle SLP reductions.
20407 * tree-vectorizer.h (vectorizable_induction): Adjust.
20409 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20411 * config/rs6000/rs6000.c (make_resolver_func): Update
20412 init_lowered_empty_function call.
20414 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
20416 * doc/invoke.texi: Document the -fprofile-abs-path option.
20417 * common.opt (fprofile-abs-path): New option.
20418 * gcov-io.h (gcov_write_filename): Declare.
20419 * gcov-io.c (gcov_write_filename): New function.
20420 * coverage.c (coverage_begin_function): Use gcov_write_filename.
20421 * profile.c (output_location): Likewise.
20423 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20425 * shring-wrap.c: Revert accidental commit.
20427 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
20429 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
20431 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20433 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
20435 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
20436 profile in callgraph edge.
20437 * profile-count.h (apply_probability): If THIS is 0, then result is 0
20438 (apply_scale): Likewise.
20439 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
20440 Also scale profile when inlining function with zero profile.
20441 (initialize_cfun): Update exit block profile even when it is zero.
20442 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
20443 when profile is read.
20445 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20447 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
20448 (CLONE_*): New constants to define the processors we can generate
20449 code for with the target_clone attribute.
20450 (rs6000_clone_map): New array to identify which clone processors
20451 the current program is running on.
20452 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
20453 target_clone attribute.
20454 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
20455 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
20456 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
20457 (cpu_expand_builtin): Add support for target_clone attribute.
20458 (rs6000_valid_attribute_p): Allow "default" attribute.
20459 (get_decl_name): New debug function to simplify printing the
20460 current function name in debugging statements.
20461 (rs6000_clone_priority): New functions to support the target_clone
20462 attribute, and be able to generate code to switch between ISA 2.05
20463 through ISA 3.0 (power6 through power9).
20464 (rs6000_compare_version_priority): Likewise.
20465 (rs6000_get_function_versions_dispatcher): Likewise.
20466 (make_resolver_func): Likewise.
20467 (add_condition_to_bb): Likewise.
20468 (dispatch_function_versions): Likewise.
20469 (rs6000_generate_version_dispatcher_body): Likewise.
20470 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
20471 (fusion_gpr_load_p): Fix a spacing issue.
20472 * doc/extend.texi (Common Function Attributes): Document that the
20473 PowerPC supports the target_clone attribute.
20475 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20477 * config/arm/arm.h: explain F symbol found in description of ARM
20478 register allocation in its legend.
20480 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20482 * config/mips/frame-header-opt.c: Include profile-count.h.
20483 * config/riscv/riscv.c: Include profile-count.h
20485 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20487 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
20489 (sm_set_flag_if_changed): Add bbs field.
20490 (execute_sm_if_changed_flag_set): Pass BBS.
20491 (execute_sm): Update.
20493 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20495 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
20498 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20500 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
20501 (peephole2): New peephole2 to emit the above.
20502 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
20504 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20506 * config/aarch64/aarch64.c (define_peephole2 above
20507 *sub_<shift>_<mode>): New peephole.
20509 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20511 * config/i386/i386.c (make_resolver_func): Update.
20512 * Makefile.in: Add profile-count.h and profile-count.o
20513 * auto-profile.c (afdo_indirect_call): Update to new API.
20514 (afdo_set_bb_count): Update.
20515 (afdo_propagate_edge): Update.
20516 (afdo_propagate_circuit): Update.
20517 (afdo_calculate_branch_prob): Update.
20518 (afdo_annotate_cfg): Update.
20519 * basic-block.h: Include profile-count.h
20520 (struct edge_def): Turn count to profile_count.
20521 (struct basic_block_def): Likewie.
20522 (REG_BR_PROB_BASE): Move to profile-count.h
20523 (RDIV): Move to profile-count.h
20524 * bb-reorder.c (max_entry_count): Turn to profile_count.
20525 (find_traces): Update.
20526 (rotate_loop):Update.
20527 (connect_traces):Update.
20528 (sanitize_hot_paths):Update.
20529 * bt-load.c (migrate_btr_defs): Update.
20530 * cfg.c (RDIV): Remove.
20531 (init_flow): Use alloc_block.
20532 (alloc_block): Uninitialize count.
20533 (unchecked_make_edge): Uninitialize count.
20534 (check_bb_profile): Update.
20535 (dump_edge_info): Update.
20536 (dump_bb_info): Update.
20537 (update_bb_profile_for_threading): Update.
20538 (scale_bbs_frequencies_int): Update.
20539 (scale_bbs_frequencies_gcov_type): Update.
20540 (scale_bbs_frequencies_profile_count): New.
20541 * cfg.h (update_bb_profile_for_threading): Update.
20542 (scale_bbs_frequencies_profile_count): Declare.
20543 * cfgbuild.c (compute_outgoing_frequencies): Update.
20544 (find_many_sub_basic_blocks): Update.
20545 * cfgcleanup.c (try_forward_edges): Update.
20546 (try_crossjump_to_edge): Update.
20547 * cfgexpand.c (expand_gimple_tailcall): Update.
20548 (construct_exit_block): Update.
20549 * cfghooks.c (verify_flow_info): Update.
20550 (dump_bb_for_graph): Update.
20551 (split_edge): Update.
20552 (make_forwarder_block): Update.
20553 (duplicate_block): Update.
20554 (account_profile_record): Update.
20555 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
20556 (get_estimated_loop_iterations): Update.
20557 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
20558 (single_likely_exit): Update.
20559 * cfgloopmanip.c (scale_loop_profile): Update.
20561 (set_zero_probability): Update.
20562 (lv_adjust_loop_entry_edge): Update.
20563 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20564 (purge_dead_edges): Update.
20565 (rtl_account_profile_record): Update.
20566 * cgraph.c (cgraph_node::create): Uninitialize count.
20567 (symbol_table::create_edge): Uninitialize count.
20568 (cgraph_update_edges_for_call_stmt_node): Update.
20569 (cgraph_edge::dump_edge_flags): Update.
20570 (cgraph_node::dump): Update.
20571 (cgraph_edge::maybe_hot_p): Update.
20572 * cgraph.h: Include profile-count.h
20573 (create_clone), create_edge, create_indirect_edge): Update.
20574 (cgraph_node): Turn count to profile_count.
20575 (cgraph_edge0: Likewise.
20576 (make_speculative, clone): Update.
20577 (create_edge): Update.
20578 (init_lowered_empty_function): Update.
20579 * cgraphclones.c (cgraph_edge::clone): Update.
20580 (duplicate_thunk_for_node): Update.
20581 (cgraph_node::create_clone): Update.
20582 * cgraphunit.c (cgraph_node::analyze): Update.
20583 (cgraph_node::expand_thunk): Update.
20584 * final.c (dump_basic_block_info): Update.
20585 * gimple-streamer-in.c (input_bb): Update.
20586 * gimple-streamer-out.c (output_bb): Update.
20587 * graphite.c (print_global_statistics): Update.
20588 (print_graphite_scop_statistics): Update.
20589 * hsa-brig.c: Include basic-block.h.
20590 * hsa-dump.c: Include basic-block.h.
20591 * hsa-gen.c (T sum_slice): Update.
20592 (convert_switch_statements):Update.
20593 * hsa-regalloc.c: Include basic-block.h.
20594 * ipa-chkp.c (chkp_produce_thunks): Update.
20595 * ipa-cp.c (struct caller_statistics): Update.
20596 (init_caller_stats): Update.
20597 (gather_caller_stats): Update.
20598 (ipcp_cloning_candidate_p): Update.
20599 (good_cloning_opportunity_p): Update.
20600 (get_info_about_necessary_edges): Update.
20601 (dump_profile_updates): Update.
20602 (update_profiling_info): Update.
20603 (update_specialized_profile): Update.
20604 (perhaps_add_new_callers): Update.
20605 (decide_about_value): Update.
20606 (ipa_cp_c_finalize): Update.
20607 * ipa-devirt.c (struct odr_type_warn_count): Update.
20608 (struct decl_warn_count): Update.
20609 (struct final_warning_record): Update.
20610 (possible_polymorphic_call_targets): Update.
20611 (ipa_devirt): Update.
20612 * ipa-fnsummary.c (redirect_to_unreachable): Update.
20613 * ipa-icf.c (sem_function::merge): Update.
20614 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
20615 * ipa-inline.c (compute_uninlined_call_time): Update.
20616 (compute_inlined_call_time): Update.
20617 (want_inline_small_function_p): Update.
20618 (want_inline_self_recursive_call_p): Update.
20619 (edge_badness): Update.
20620 (lookup_recursive_calls): Update.
20621 (recursive_inlining): Update.
20622 (inline_small_functions): Update.
20623 (dump_overall_stats): Update.
20624 (dump_inline_stats): Update.
20625 * ipa-profile.c (ipa_profile_generate_summary): Update.
20626 (ipa_propagate_frequency): Update.
20627 (ipa_profile): Update.
20628 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20629 * ipa-utils.c (ipa_merge_profiles): Update.
20630 * loop-doloop.c (doloop_modify): Update.
20631 * loop-unroll.c (report_unroll): Update.
20632 (unroll_loop_runtime_iterations): Update.
20633 * lto-cgraph.c (lto_output_edge): Update.
20634 (lto_output_node): Update.
20635 (input_node): Update.
20636 (input_edge): Update.
20637 (merge_profile_summaries): Update.
20638 * lto-streamer-in.c (input_cfg): Update.
20639 * lto-streamer-out.c (output_cfg): Update.
20640 * mcf.c (create_fixup_graph): Update.
20641 (adjust_cfg_counts): Update.
20642 (sum_edge_counts): Update.
20643 * modulo-sched.c (sms_schedule): Update.
20644 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
20645 * predict.c (maybe_hot_count_p): Update.
20646 (probably_never_executed): Update.
20647 (dump_prediction): Update.
20648 (combine_predictions_for_bb): Update.
20649 (propagate_freq): Update.
20650 (handle_missing_profiles): Update.
20651 (counts_to_freqs): Update.
20652 (rebuild_frequencies): Update.
20653 (force_edge_cold): Update.
20654 * predict.h: Include profile-count.h
20655 (maybe_hot_count_p, counts_to_freqs): UPdate.
20656 * print-rtl-function.c: Do not include cfg.h
20657 * print-rtl.c: Include basic-block.h
20658 * profile-count.c: New file.
20659 * profile-count.h: New file.
20660 * profile.c (is_edge_inconsistent): Update.
20661 (correct_negative_edge_counts): Update.
20662 (is_inconsistent): Update.
20663 (set_bb_counts): Update.
20664 (read_profile_edge_counts): Update.
20665 (compute_frequency_overlap): Update.
20666 (compute_branch_probabilities): Update; Initialize and deinitialize
20668 (branch_prob): Update.
20669 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
20670 (edge_gcov_count): New.
20671 (bb_gcov_count): New.
20672 * shrink-wrap.c (try_shrink_wrapping): Update.
20673 * tracer.c (better_p): Update.
20674 * trans-mem.c (expand_transaction): Update.
20675 (ipa_tm_insert_irr_call): Update.
20676 (ipa_tm_insert_gettmclone_call): Update.
20677 * tree-call-cdce.c: Update.
20678 * tree-cfg.c (gimple_duplicate_sese_region): Update.
20679 (gimple_duplicate_sese_tail): Update.
20680 (gimple_account_profile_record): Update.
20681 (execute_fixup_cfg): Update.
20682 * tree-inline.c (copy_bb): Update.
20683 (copy_edges_for_bb): Update.
20684 (initialize_cfun): Update.
20685 (freqs_to_counts): Update.
20686 (copy_cfg_body): Update.
20687 (expand_call_inline): Update.
20688 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20689 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
20690 (try_unroll_loop_completely): Update.
20691 (try_peel_loop): Update.
20692 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20693 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
20694 * tree-ssa-loop-split.c (connect_loops): Update.
20695 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
20696 * tree-ssa-reassoc.c (branch_fixup): Update.
20697 * tree-ssa-tail-merge.c (replace_block_by): Update.
20698 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
20699 (compute_path_counts): Update.
20700 (update_profile): Update.
20701 (recompute_probabilities): Update.
20702 (update_joiner_offpath_counts): Update.
20703 (estimated_freqs_path): Update.
20704 (freqs_to_counts_path): Update.
20705 (clear_counts_path): Update.
20706 (ssa_fix_duplicate_block_edges): Update.
20707 (duplicate_thread_path): Update.
20708 * tree-switch-conversion.c (case_bit_test_cmp): Update.
20709 (struct switch_conv_info): Update.
20710 * tree-tailcall.c (decrease_profile): Update.
20711 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20712 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20713 * value-prof.c (check_counter): Update.
20714 (gimple_divmod_fixed_value): Update.
20715 (gimple_mod_pow2): Update.
20716 (gimple_mod_subtract): Update.
20717 (gimple_ic_transform): Update.
20718 (gimple_stringop_fixed_value): Update.
20719 * value-prof.h (gimple_ic): Update.
20721 2017-06-02 Carl Love <cel@us.ibm.com>
20723 * config/rs6000/rs6000-c: Add support for built-in functions
20724 vector double vec_doublee (vector signed int);
20725 vector double vec_doublee (vector unsigned int);
20726 vector double vec_doublee (vector float);
20727 vector double vec_doubleh (vector signed int);
20728 vector double vec_doubleh (vector unsigned int);
20729 vector double vec_doubleh (vector float);
20730 vector double vec_doublel (vector signed int);
20731 vector double vec_doublel (vector unsigned int);
20732 vector double vec_doublel (vector float);
20733 vector double vec_doubleo (vector signed int);
20734 vector double vec_doubleo (vector unsigned int);
20735 vector double vec_doubleo (vector float);.
20736 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
20737 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
20739 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
20740 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
20741 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
20743 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
20744 vec_doublel, vec_doubleh.
20745 * doc/extend.texi: Update the built-in documentation file for the
20746 new built-in functions.
20748 2017-06-02 David Malcolm <dmalcolm@redhat.com>
20751 * ipa-inline-analysis.c (free_growth_caches): Set
20752 edge_removal_hook_holder to NULL after removing it.
20754 2017-06-02 Sudakshina Das <sudi.das@arm.com>
20756 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
20757 comparision with zero.
20759 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
20760 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20761 for early expansion of vec_min and vec_max builtins.
20762 (builtin_function_type): Add min/max unsigned variants to those
20763 identified as having unsigned arguments.
20765 2017-06-02 Olivier Hainque <hainque@adacore.com>
20767 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
20769 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20771 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
20772 Use VALL_F16 iterator rather than VALL.
20774 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20776 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
20777 Emit CBNZ inside loop when doing a strong exchange and comparing
20778 against zero. Generate the CC flags after the loop.
20780 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
20782 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
20783 (dl_section_ref): New.
20784 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
20785 On AIX, append an expression to subtract the size of the
20786 section length to dl_section_ref.
20788 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
20790 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20791 for early expansion of vector absolute builtins.
20793 2017-06-02 Richard Biener <rguenther@suse.de>
20795 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
20796 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
20798 2017-06-02 Richard Biener <rguenther@suse.de>
20800 PR tree-optimization/80948
20801 * tree-tailcall.c (find_tail_calls): Track stmts to move in
20802 stmt order as well.
20804 2017-06-02 Richard Biener <rguenther@suse.de>
20806 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
20808 * tree-vect-stmts.c (process_use): Do not mark backedge defs
20809 for inductions as relevant.
20811 2017-06-02 Richard Biener <rguenther@suse.de>
20813 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
20814 (vectorizable_induction): ... this. Remove dead code.
20816 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
20818 * builtins. (expand_builtin_alloca): Remove second parameter and
20819 infer its value from the first parameter instead.
20820 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
20822 2017-06-02 Jakub Jelinek <jakub@redhat.com>
20824 PR rtl-optimization/80903
20825 * loop-doloop.c (add_test): Unshare sequence.
20827 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
20829 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
20831 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
20833 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
20835 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
20836 xlogue_layout::get_instance, logue_layout::xlogue_layout,
20837 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
20838 (xlogue_layout::get_stub_rtx): Make static.
20839 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
20840 (xlogue_layout::compute_stub_managed_regs): Rename to...
20841 (xlogue_layout::count_stub_managed_regs): ...this.
20842 (xlogue_layout::is_stub_managed_reg): New function.
20843 (xlogue_layout::m_stub_names): Rename to...
20844 (xlogue_layout::s_stub_names): ...this, make static.
20845 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
20846 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
20847 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
20848 xlogue_layout::s_stub_names): Instantiate statics.
20849 (stub_managed_regs): Remove.
20850 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
20851 (disable_call_ms2sysv_xlogues): Rename to...
20852 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
20853 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
20855 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
20856 change after reload_completed.
20857 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
20859 (ix86_expand_prologue): Likewise.
20860 (ix86_expand_epilogue): Likewise.
20861 (ix86_expand_split_stack_prologue): Likewise.
20862 (ix86_compute_frame_layout): Remove frame parameter ...
20863 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
20864 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
20866 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
20867 (ix86_frame): Move from here ...
20868 * config/i386/i386.h (ix86_frame): ... to here.
20869 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
20870 complete ix86_frame data structure instead. Remove some_ld_name.
20872 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
20874 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
20875 symbols that hold a DECL_VALUE_EXPR.
20877 2017-06-01 Martin Jambor <mjambor@suse.cz>
20879 PR tree-optimization/80898
20880 * tree-sra.c (process_subtree_disqualification): Removed.
20881 (disqualify_candidate): Do not acll
20882 process_subtree_disqualification.
20883 (subtree_mark_written_and_enqueue): New function.
20884 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
20885 RHS has been disqualified and re-queue LHS if necessary. Apart
20886 from that, ignore disqualified RHS.
20888 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20890 * config/s390/s390.c (s390_emit_epilogue): Disable early return
20891 address fetch for z10 or later.
20893 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20895 * config/arc/arc.md (tst_movb): Add guard when splitting.
20897 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20899 * config/arc/arc.c (arc_can_eliminate): Test against
20900 arc_frame_pointer_needed.
20902 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20904 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
20905 to prevent store reordering.
20906 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
20907 (type): Add block type.
20908 (stack_tie): Define special instruction to be used in
20911 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20913 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
20914 constraint. It is not valid for the pattern.
20915 (noncommutative_binary_comparison): Likewise.
20917 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20919 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
20922 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20924 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
20925 be used by the reg-alloc.
20927 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20929 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
20930 reg-alloc when having mul64 or mul32x16 instructions.
20931 (mulsidi3): Likewise.
20932 (umulsidi3): Likewise.
20933 (mulsi32x16): New pattern.
20934 (mulsi64): Likewise.
20935 (mulsidi64): Likewise.
20936 (umulsidi64): Likewise.
20937 (MUL32x16_REG): Define.
20938 (mul64_600): Use MUL32x16_REG.
20939 (mac64_600): Likewise.
20940 (umul64_600): Likewise.
20941 (umac64_600): Likewise.
20943 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20945 * config/arc/arc.md (mulsi3_700): Make it commutative.
20947 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
20949 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
20951 (*sign_extendsidi2_insn): Likewise for movstosw.
20953 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
20955 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
20956 the type of the input discriminant value. Convert the
20957 discriminant value of signedness vary.
20959 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
20961 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
20962 Add to -Wall section.
20964 2017-06-01 Richard Biener <rguenther@suse.de>
20966 PR middle-end/66313
20967 * fold-const.c (fold_plusminus_mult_expr): If the factored
20968 factor may be zero use a wrapping type for the inner operation.
20969 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
20970 and handle moved defs.
20971 (process_assignment): Properly guard the unary op case. Return a
20972 tri-state indicating that moving the stmt before the call may allow
20973 to continue. Pass through to_move.
20974 (find_tail_calls): Handle moving unrelated defs before
20977 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
20980 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
20981 splitter result in the canonical way.
20983 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
20985 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
20986 also for 32bit target. Update insn attributes.
20987 (zero-extendsidi2 splitter): Allow all registers for operand 1.
20989 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
20991 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
20992 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
20993 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
20994 (_mm_maskz_min_ss): New intrinsics.
20996 2017-05-31 Martin Liska <mliska@suse.cz>
20998 * tree-vect-loop.c (vect_create_epilog_for_reduction):
20999 Change comment style to one we normally use.
21000 (vectorizable_reduction): Likewise.
21001 (vectorizable_induction): Likewise.
21002 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21003 (vectorizable_call): Likewise.
21004 (vectorizable_simd_clone_call): Likewise.
21005 (vectorizable_conversion): Likewise.
21006 (vectorizable_assignment): Likewise.
21007 (vectorizable_shift): Likewise.
21008 (vectorizable_operation): Likewise.
21009 (vectorizable_store): Likewise.
21010 (vectorizable_load): Likewise.
21011 * tree-vectorizer.h: Likewise.
21013 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21015 * passes.c (emergency_dump_function): New.
21016 * tree-pass.h (emergency_dump_function): Declare.
21017 * plugin.c (plugins_internal_error_function): Remove.
21018 * plugin.h (plugins_internal_error_function): Remove declaration.
21019 * toplev.c (internal_error_function): New static function. Use it...
21020 (general_init): ...here.
21022 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21024 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21025 (arc_rtx_costs): Add costs for new patterns.
21026 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21027 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21029 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21031 * tree-ssa-strlen.c (get_next_strinfo): New function.
21032 (get_stridx_plus_constant): Use it.
21033 (zero_length_string): Likewise.
21034 (adjust_related_strinfos): Likewise.
21035 (adjust_last_stmt): Likewise.
21037 2017-05-31 Richard Biener <rguenther@suse.de>
21040 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21041 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21043 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21045 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21046 loop_vinfo argument and use of dependence distance vectors.
21047 Check instead whether the two references differ only in their
21048 initial value and assume that they have the same alignment if the
21049 difference is a multiple of the vector alignment.
21050 (vect_analyze_data_refs_alignment): Update call accordingly.
21052 2017-05-31 Martin Liska <mliska@suse.cz>
21055 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21057 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21059 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21060 (create_intersect_range_checks): Move from ...
21061 * tree-data-ref.c (create_intersect_range_checks_index)
21062 (create_intersect_range_checks): ... to here.
21063 (create_runtime_alias_checks): New function factored from ...
21064 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21065 here. Call above function.
21066 * tree-data-ref.h (create_runtime_alias_checks): New function.
21068 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21070 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21071 segment length for dr_b and compute it in wide_int.
21073 2017-05-31 Richard Biener <rguenther@suse.de>
21075 PR tree-optimization/80906
21076 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21077 and pass through iv_map.
21078 (copy_bb_and_scalar_dependences): Adjust.
21079 (translate_pending_phi_nodes): Likewise.
21080 (copy_loop_close_phi_args): Handle code-generating IVs instead
21083 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21085 * diagnostic-color.c (color_dict): Add "type-diff".
21086 (parse_gcc_colors): Update comment.
21087 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21088 -fdiagnostics-show-template-tree and -fno-elide-type.
21089 (GCC_COLORS): Add type-diff to example.
21091 (-fdiagnostics-show-template-tree): New.
21092 (-fno-elide-type): New.
21093 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21094 the pp_format_decoder callback. Call any m_format_postprocessor's
21096 (pretty_printer::pretty_printer): Initialize
21097 m_format_postprocessor.
21098 (pretty_printer::~pretty_printer): Delete any
21099 m_format_postprocessor.
21100 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21101 (class format_postprocessor): New class.
21102 (struct pretty_printer::format_decoder): Document the new parameters.
21103 (struct pretty_printer::m_format_postprocessor): New field.
21104 * tree-diagnostic.c (default_tree_printer): Update for new
21105 bool and const char ** params.
21106 * tree-diagnostic.h (default_tree_printer): Likewise.
21108 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21110 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21111 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21112 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21113 rs6000_gen_cell_microcode code.
21114 (rs6000_final_prescan_insn): Delete.
21115 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21116 "warn-cell-microcode" entries.
21117 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21118 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21119 throughout. Change cc_reg_not_micro_cr0_operand to
21120 cc_reg_not_cr0_operand throughout.
21121 (*extendhi<mode>2_noload): Delete.
21122 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21123 (mwarn-cell-microcode): Delete.
21124 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21125 -mgen-cell-microcode and -mwarn-cell-microcode.
21127 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21130 * config/i386/constraints.md (Yd): New constraint.
21132 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21133 and (?Yd, r) alternatives. Update insn attributes.
21134 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21135 and (?*Yd, r) alternatives. Update insn attributes.
21136 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21138 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21140 * gimplify.c (gimplify_modify_expr): Don't create a
21141 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21144 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21146 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21148 2017-05-30 Richard Biener <rguenther@suse.de>
21150 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21151 and reduc_def fields.
21152 (STMT_VINFO_REDUC_TYPE): New define.
21153 (STMT_VINFO_REDUC_DEF): Likewise.
21154 (vect_force_simple_reduction): Adjust prototype.
21155 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21156 (vect_is_simple_reduction): Remove check_reduction argument.
21157 (vect_force_simple_reduction): Adjust and set
21158 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21159 (vectorizable_reduction): Do not re-do reduction analysis
21160 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21161 * tree-parloops.c (gather_scalar_reductions): Adjust.
21163 2017-05-30 Richard Biener <rguenther@suse.de>
21165 PR middle-end/80901
21166 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21169 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21171 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21172 Introduce unknown_misalignment parameter and remove vf.
21173 (vect_peeling_hash_get_lowest_cost):
21174 Pass unknown_misalignment parameter.
21175 (vect_enhance_data_refs_alignment):
21176 Fix unsupportable data ref treatment.
21178 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21180 * tree-vect-data-refs.c (vect_get_data_access_cost):
21181 Workaround for SLP handling.
21182 (vect_enhance_data_refs_alignment):
21183 Compute costs for doing no peeling at all, compare to the best
21184 peeling costs so far and avoid peeling if cheaper.
21186 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21188 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21189 Return peeling info and set costs to zero for unlimited cost
21191 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21192 with unknown misalignment. Compute and costs for unknown
21193 misalignment, compare them to the costs for known misalignment
21194 and choose the cheapest for peeling.
21196 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21198 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21199 (vect_get_peeling_costs_all_drs): Create function.
21200 (vect_peeling_hash_get_lowest_cost):
21201 Use vect_get_peeling_costs_all_drs.
21202 (vect_peeling_supportable): Create function.
21203 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21205 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21207 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21208 DR_HAS_NEGATIVE_STEP.
21209 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21210 (vect_enhance_data_refs_alignment): Use.
21211 (vect_duplicate_ssa_name_ptr_info): Use.
21212 * tree-vectorizer.h (dr_misalignment): Use.
21213 (known_alignment_for_access_p): Use.
21215 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21218 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21220 (has_section_name): New function.
21222 2017-05-30 Martin Liska <mliska@suse.cz>
21225 * auto-profile.c (get_function_decl_from_block): Fix
21228 2017-05-30 Richard Biener <rguenther@suse.de>
21230 PR middle-end/80876
21231 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21233 2017-05-30 Martin Liska <mliska@suse.cz>
21235 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21236 * dumpfile.h (struct dump_file_info): Remove ctors.
21238 2017-05-30 Martin Liska <mliska@suse.cz>
21240 * predict.def: Fix GNU coding style.
21242 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21244 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21245 Mark 'to' argument with ATTRIBUTE_UNUSED.
21247 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21249 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21250 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
21251 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
21254 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
21256 * doc/install.texi (Options specification): Restore entry of
21257 --enable-sjlj-exceptions.
21259 2017-05-27 Michael Eager <eager@eagercon.com>
21262 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
21264 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
21266 * config/microblaze/microblaze.h
21267 (FIXED_REGISTERS): Update in macro.
21268 (CALL_USED_REGISTERS): Update in macro.
21270 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
21272 * doc/install.texi: Add links to macOS binary distributions.
21274 2017-05-27 Jakub Jelinek <jakub@redhat.com>
21278 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21280 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21282 2017-05-26 Martin Liska <mliska@suse.cz>
21284 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
21286 2017-05-26 Martin Liska <mliska@suse.cz>
21288 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
21289 always leading ';; '.
21290 (dump_bb_info): Likewise.
21291 (brief_dump_cfg): Likewise.
21292 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
21293 * dumpfile.c: Remove usage of TDF_VERBOSE.
21294 * dumpfile.h (enum dump_kind): Likewise.
21295 (dump_gimple_bb_header): Do not use TDF_COMMENT.
21296 * print-tree.c (debug_verbose): Remove.
21297 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
21298 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
21299 * tree-diagnostic.c (default_tree_printer): Replace
21300 TDF_DIAGNOSTIC with TDF_SLIM.
21302 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21304 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
21305 in parameter loop, rather than loop_vinfo.
21306 (create_intersect_range_checks): Ditto.
21307 (vect_create_cond_for_alias_checks): Update call to above functions.
21309 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21311 PR tree-optimization/80815
21312 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
21313 for merging runtime alias checks. Handle negative DR_STEPs.
21315 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21317 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
21319 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
21320 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
21321 out code pruning runtime alias checks.
21322 * tree-data-ref.c (prune_runtime_alias_test_list): New function
21323 factored out from above.
21324 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21326 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21328 (prune_runtime_alias_test_list): New decalaration.
21330 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21332 * tree-vect-data-refs.c (compare_tree): Rename and move ...
21333 * tree-data-ref.c (data_ref_compare_tree): ... to here.
21334 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
21335 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
21336 (operator==, comp_dr_with_seg_len_pair): Ditto.
21337 (vect_prune_runtime_alias_test_list): Ditto.
21339 2017-05-26 Martin Liska <mliska@suse.cz>
21342 * params.def: Bound partial-inlining-entry-probability param.
21344 2017-05-26 Marek Polacek <polacek@redhat.com>
21347 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
21350 2017-05-26 Richard Biener <rguenther@suse.de>
21352 PR tree-optimization/80842
21353 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
21356 2017-05-26 Richard Biener <rguenther@suse.de>
21358 PR tree-optimization/80844
21359 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
21361 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
21363 * doc/md.texi (Machine Constraints): Update x86 family
21364 machine constraints section to match 'config/i386/constraints.md'.
21366 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
21368 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
21370 2017-05-25 Nathan Sidwell <nathan@acm.org>
21372 * doc/invoke.texi (--enable-languages): Update documentation.
21374 2017-05-25 Martin Liska <mliska@suse.cz>
21376 * dumpfile.c: Add TDF_FOLDING.
21377 * dumpfile.h (enum dump_kind): Likewise.
21378 * genmatch.c (dt_simplify::gen_1): Use it.
21380 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21382 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
21384 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21386 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21387 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
21389 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21391 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
21392 * match.pd (X == C): Rewrite it here.
21393 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
21394 with_certain_nonzero_bits2): New predicates.
21395 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
21397 2017-05-24 Nathan Sidwell <nathan@acm.org>
21399 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
21402 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
21405 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21407 * config/powerpcspe: New port. Files are copied from the rs6000
21408 port, with "rs6000" in filenames replaced by "powerpcspe".
21410 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
21412 PR rtl-optimization/80754
21413 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
21415 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
21417 * config/sparc/sparc.md (length): Return the correct value for -mflat
21418 sibcalls to match output_sibcall.
21420 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21424 * config/rs6000/rs6000.c (struct machine_function): Add new field
21426 (rs6000_get_separate_components): Init that field, use it.
21427 (rs6000_components_for_bb): Use the field.
21429 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
21431 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
21433 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
21435 PR middle-end/80823
21436 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
21438 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21441 * config/s390/s390.c (s390_check_qrst_address): Check incoming
21442 address against address_operand predicate.
21443 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
21445 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
21447 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
21448 parameters passed indirectly.
21450 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
21452 * config/i386/i386.md (*movdi_internal): Remove SSE4
21453 alternative 18 (?r, *v). Update insn attributes.
21454 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
21455 Update insn attributes.
21456 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
21457 Update insn attributes.
21458 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
21459 alternative 1 (r, v). Remove isa attribute.
21460 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
21461 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
21462 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
21464 2017-05-23 Tom de Vries <tom@codesourcery.com>
21466 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
21469 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21471 * cgraphunit.c (symbol_table::process_new_functions): Update.
21472 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
21473 (inline_generate_summary): Rename to ...
21474 (ipa_fn_summary_generate): ... this one.
21475 (inline_read_summary): Rename to ...
21476 (ipa_fn_summary_read): ... this one.
21477 (inline_write_summary): Rename to ...
21478 (ipa_fn_summary_write): ... this one.
21479 (inline_free_summary): Rename to ...
21480 (ipa_free_fn_summary): ... this one.
21481 (pass_data_local_fn_summary, pass_local_fn_summary,
21482 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
21483 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
21484 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
21485 make_pass_ipa_fn_summary): New.
21486 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
21487 inline_write_summary, inline_free_summary): Remove.
21488 (ipa_free_fn_summary) : New.
21489 * ipa-inline.c (ipa_inline): Update.
21490 (pass_ipa_inline): Do not generate summaries.
21491 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
21493 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
21494 and add pass_ipa_fn_summary.
21495 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
21497 (make_pass_inline_parameters): Remove.
21499 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
21501 * omp-low.c (struct omp_context): Remove "default_kind" member.
21504 * omp-offload.c (execute_oacc_device_lower): Remove the
21505 parallelism dimensions function attributes for unparallelized
21506 OpenACC kernels constructs.
21508 2017-05-23 Martin Liska <mliska@suse.cz>
21510 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
21512 (cgraph_edge::make_speculative): Likewise.
21513 (cgraph_edge::resolve_speculation): Likewise.
21514 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21515 (cgraph_node::dump): Likewise.
21516 * cgraph.h: Likewise.
21517 * cgraphunit.c (analyze_functions): Likewise.
21518 (symbol_table::compile): Likewise.
21519 * ipa-cp.c (print_all_lattices): Likewise.
21520 (determine_versionability): Likewise.
21521 (initialize_node_lattices): Likewise.
21522 (ipcp_verify_propagated_values): Likewise.
21523 (estimate_local_effects): Likewise.
21524 (update_profiling_info): Likewise.
21525 (create_specialized_node): Likewise.
21526 (perhaps_add_new_callers): Likewise.
21527 (decide_about_value): Likewise.
21528 (decide_whether_version_node): Likewise.
21529 (identify_dead_nodes): Likewise.
21530 (ipcp_store_bits_results): Likewise.
21531 * ipa-devirt.c (dump_targets): Likewise.
21532 (ipa_devirt): Likewise.
21533 * ipa-icf.c (sem_item::dump): Likewise.
21534 (sem_function::equals): Likewise.
21535 (sem_variable::equals): Likewise.
21536 (sem_item_optimizer::read_section): Likewise.
21537 (sem_item_optimizer::execute): Likewise.
21538 (congruence_class::dump): Likewise.
21539 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
21540 (dump_inline_summary): Likewise.
21541 (estimate_node_size_and_time): Likewise.
21542 (inline_analyze_function): Likewise.
21543 * ipa-inline-transform.c (inline_call): Likewise.
21544 * ipa-inline.c (report_inline_failed_reason): Likewise.
21545 (want_early_inline_function_p): Likewise.
21546 (edge_badness): Likewise.
21547 (update_edge_key): Likewise.
21548 (inline_small_functions): Likewise.
21549 * ipa-profile.c (ipa_profile): Likewise.
21550 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
21551 (ipa_make_edge_direct_to_target): Likewise.
21552 (remove_described_reference): Likewise.
21553 (ipa_impossible_devirt_target): Likewise.
21554 (propagate_controlled_uses): Likewise.
21555 (ipa_print_node_params): Likewise.
21556 (ipcp_transform_function): Likewise.
21557 * ipa-pure-const.c (pure_const_read_summary): Likewise.
21558 (propagate_pure_const): Likewise.
21559 * ipa-reference.c (generate_summary): Likewise.
21560 (read_write_all_from_decl): Likewise.
21561 (propagate): Likewise.
21562 (ipa_reference_read_optimization_summary): Likewise.
21563 * ipa-utils.c (ipa_merge_profiles): Likewise.
21564 * ipa.c (walk_polymorphic_call_targets): Likewise.
21565 (symbol_table::remove_unreachable_nodes): Likewise.
21566 (ipa_single_use): Likewise.
21567 * passes.c (execute_todo): Likewise.
21568 * predict.c (drop_profile): Likewise.
21569 * symtab.c (symtab_node::get_dump_name): New function.
21570 (symtab_node::dump_name): Likewise.
21571 (symtab_node::dump_asm_name): Likewise.
21572 (symtab_node::dump_references): Likewise.
21573 (symtab_node::dump_referring): Likewise.
21574 (symtab_node::dump_base): Likewise.
21575 (symtab_node::debug_symtab): Likewise.
21576 * tree-sra.c (convert_callers_for_node): Likewise.
21577 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
21578 * value-prof.c (init_node_map): Likewise.
21580 2017-05-23 Martin Liska <mliska@suse.cz>
21582 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
21583 and symtab_node::debug_symtab to symbol_table::debug.
21584 * cgraphunit.c (analyze_functions): Use the renamed function.
21585 (symbol_table::compile): Likewise.
21586 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
21587 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
21588 * passes.c (execute_todo): Likewise.
21589 * symtab.c (symbol_table::dump): New function.
21590 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
21592 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21594 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
21595 that nonconst implies exec.
21597 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21599 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21600 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
21601 (inline_edge_summary_vec): Turn into ...
21602 (ipa_call_summaries): ... this one.
21603 (redirect_to_unreachable, edge_set_predicate,
21604 evaluate_properties_for_edge, inline_summary_alloc,
21605 reset_ipa_call_summary, reset_inline_summary,
21606 inline_summary_t::duplicate): Update.
21607 (inline_edge_duplication_hook): Turn to ...
21608 (ipa_call_summary_t::duplicate): ... this one.
21609 (inline_edge_removal_hook): Turn to ...
21610 (ipa_call_summary_t::remove): ... this one.
21611 (dump_inline_edge_summary): Turn to ...
21612 (dump_ipa_call_summary): ... this one.
21613 (estimate_function_body_sizes): Update.
21614 (inline_update_callee_summaries): Update.
21615 (remap_edge_change_prob): Update.
21616 (remap_edge_summaries): Update.
21617 (inline_merge_summary): Update.
21618 (do_estimate_edge_time): Update.
21619 (inline_generate_summary): Update.
21620 (inline_read_section): Update.
21621 (inline_read_summary): Update.
21622 (inline_free_summary): Update.
21623 * ipa-inline.c (can_inline_edge_p): Update.
21624 (compute_inlined_call_time): Update.
21625 (want_inline_small_function_p): Update.
21626 (edge_badness): Update.
21627 (early_inliner): Update.
21628 * ipa-inline.h (inline_edge_summary): Turn to ...
21629 (ipa_call_summary): ... this one.
21630 (ipa_call_summary_t): New class.
21631 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
21632 (ipa_call_summaries): New.
21633 (inline_edge_summary): Remove.
21634 (estimate_edge_growth): Update.
21635 * ipa-profile.c (ipa_propagate_frequency_1): Update.
21636 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21637 * ipa-split.c (execute_split_functions): Update.
21638 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
21640 2017-05-23 Tom de Vries <tom@codesourcery.com>
21642 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
21643 attributes): Document rdrand effective target.
21645 2017-05-23 Tom de Vries <tom@codesourcery.com>
21647 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
21648 attributes): Sort alphabetically.
21650 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
21652 * config/avr/genmultilib.awk: Use gsub instead of gensub.
21654 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
21657 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
21658 V2DF/V2DI splat into two separate patterns, one that handles
21659 registers, and the other that only handles memory. Drop support
21660 for splatting from a GPR on ISA 2.07 and then splitting the
21661 splat into direct move and splat.
21662 (vsx_splat_<mode>_reg): Likewise.
21663 (vsx_splat_<mode>_mem): Likewise.
21665 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
21667 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
21669 2017-05-22 Jakub Jelinek <jakub@redhat.com>
21671 PR middle-end/80809
21672 * omp-low.c (finish_taskreg_remap): New function.
21673 (finish_taskreg_scan): If unit size of ctx->record_type
21674 is non-constant, unshare the size expression and replace
21675 decls in it with possible outer var refs.
21677 PR middle-end/80809
21678 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
21679 GOVD_SHARED rather than GOVD_PRIVATE with it.
21680 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
21681 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
21683 PR middle-end/80853
21684 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
21685 as last argument to build_outer_var_ref for pointer bases of array
21686 section reductions.
21688 2017-05-19 Martin Sebor <msebor@redhat.com>
21690 * print-tree.c (print_node): Print DECL_READ_P flag.
21692 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21694 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
21695 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
21696 * cgraph.c: Likewise.
21697 * cgraphunit.c: Likewise.
21698 * gengtype.c: Likewise.
21699 * ipa-cp.c: Likewise.
21700 * ipa-devirt.c: Likewise.
21701 * ipa-icf.c: Likewise.
21702 * ipa-predicate.c: Likewise.
21703 * ipa-profile.c: Likewise.
21704 * ipa-prop.c: Likewise.
21705 * ipa-split.c: Likewise.
21707 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
21708 edge_predicate_pool, dump_inline_hints,
21709 inline_summary::account_size_time, redirect_to_unreachable,
21710 edge_set_predicate, set_hint_predicate,
21711 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
21712 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
21713 inline_summary_t::remove, remap_hint_predicate_after_duplication,
21714 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
21715 ipa_call_summary_t::remove, initialize_growth_caches,
21716 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
21717 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
21718 mark_modified, unmodified_parm_1, unmodified_parm,
21719 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
21720 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
21721 compute_bb_predicates, will_be_nonconstant_expr_predicate,
21722 will_be_nonconstant_predicate, record_modified_bb_info,
21723 get_minimal_bb, record_modified, param_change_prob,
21724 phi_result_unknown_predicate, predicate_for_phi_result,
21725 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
21726 estimate_function_body_sizes, compute_inline_parameters,
21727 compute_inline_parameters_for_curren, pass_data_inline_parameters,
21728 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
21729 inline_update_callee_summaries, remap_edge_change_prob,
21730 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
21731 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
21732 inline_analyze_function, inline_summary_t::insert,
21733 inline_generate_summary, read_ipa_call_summary, inline_read_section,
21734 inline_read_summary, write_ipa_call_summary, inline_write_summary,
21735 inline_free_summary): Move to ipa-fnsummary.h
21736 (predicate_t): Remove.
21737 * ipa-fnsummary.c: New file.
21738 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
21739 (enum inline_hints_vals, inline_hints, agg_position_info,
21740 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
21741 inline_summaries, ipa_call_summary, ipa_call_summary_t,
21742 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
21743 dump_inline_summary, dump_inline_hints, inline_generate_summary,
21744 inline_read_summary, inline_write_summary, inline_free_summary,
21745 inline_analyze_function, initialize_inline_failed,
21746 inline_merge_summary, inline_update_overall_summary,
21747 compute_inline_parameters): Move to ipa-fnsummary.h
21748 * ipa-fnsummary.h: New file.
21749 * ipa-inline-transform.h: Include ipa-inline.h.
21750 * ipa-inline.c: LIkewise.
21752 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21754 * ipa-inline.c (edge_badness): Use inlined_time instead of
21755 inline_summaries->get.
21757 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21759 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
21761 2017-05-22 Nathan Sidwell <nathan@acm.org>
21763 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
21764 (fdump-lang): Document 'raw' option.
21765 * dumpfile.h (TDI_tu): Delete.
21766 * dumpfile.c (dump_files): Remove translation-unit.
21767 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
21769 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
21771 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
21772 command option from $(AWK) call.
21773 * config/avr/genmultilib.awk: Simplify and rewrite so that it
21774 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
21775 [FORMAT]: Remove handling of variable.
21776 * config/avr/t-multilib: Regenerate.
21778 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21780 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
21782 (dump_inline_summary): Do not print self_time.
21783 (estimate_function_body_sizes): Do not set self_time.
21784 (compute_inline_parameters): Likewise.
21785 (inline_read_section, inline_write_summary): Do not stream self_time.
21786 * ipa-inline.h (inline_summary): Drop self_time.
21788 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21790 * ipa-inline-analysis.c (account_size_time): Rename to ...
21791 (inline_summary::account_size_time): ... this one.
21792 (reset_ipa_call_summary): Turn to ...
21793 (ipa_call_summary::reset): ... this one.
21794 (reset_inline_summary): Turn to ...
21795 (inline_summary::reset): ... this one.
21796 (inline_summary_t::remove): Update.
21797 (inline_summary_t::duplicate): Update.
21798 (ipa_call_summary_t::remove): Update.
21799 (dump_inline_summary): Update.
21800 (estimate_function_body_sizes): Update.
21801 (compute_inline_parameters): Update.
21802 (estimate_node_size_and_time): Update.
21803 (inline_merge_summary): Update.
21804 (inline_update_overall_summary): Update.
21805 (inline_read_section): Update.
21806 (inline_write_summary): Update.
21807 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
21808 add account_size_time and reset member functions.
21809 (ipa_call_summary): Add reset function.
21810 * ipa-predicate.h (predicate::operator &): Constify.
21812 2017-05-22 Richard Biener <rguenther@suse.de>
21814 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
21816 2017-05-19 Jason Merrill <jason@redhat.com>
21818 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
21820 2017-05-19 Marek Polacek <polacek@redhat.com>
21823 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
21824 TYPE_OVERFLOW_WRAPS checks.
21826 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
21828 * tree-core.h (enum omp_clause_default_kind): Add
21829 "OMP_CLAUSE_DEFAULT_PRESENT".
21830 * tree-pretty-print.c (dump_omp_clause): Handle it.
21831 * gimplify.c (enum gimplify_omp_var_data): Add
21832 "GOVD_MAP_FORCE_PRESENT".
21833 (gimplify_adjust_omp_clauses_1): Map it to
21834 "GOMP_MAP_FORCE_PRESENT".
21835 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
21837 * gimplify.c (oacc_default_clause): Clarify.
21839 2017-05-19 Nathan Sidwell <nathan@acm.org>
21841 LANG_HOOK_REGISTER_DUMPS
21842 * toplev.c (general_init): Call register dump lang hook.
21843 * doc/invoke.texi: Document -fdump-lang option family.
21844 * dumpfile.c (dump_files): Remove class dump here.
21845 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
21846 * dumpfile.h (tree_dump_index): Remove TDI_class.
21847 * langhooks-def.h (lhd_register_dumps): Declare.
21848 (LANG_HOOKS_REGISTER_DUMPS): Define.
21849 (LANG_HOOKS_INITIALIZER): Add it.
21850 * langhooks.c (lhd_register_dumps): Define.
21851 * langhooks.h (struct lang_hooks): Add register_dumps.
21853 2017-05-19 Nathan Sidwell <nathan@acm.org>
21855 * context.h (context::set_passes): New.
21856 * context.c (context::context): Do not create pass manager.
21857 * toplev.c (general_init): Create pass manager here.
21859 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
21861 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
21862 use this splitter if two add or or instructions would also work for
21863 the constant we want to generate.
21865 2017-05-19 Richard Biener <rguenther@suse.de>
21868 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
21869 predicate evaluation.
21871 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
21873 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
21875 * ipa-inline.c (want_inline_small_function_p): Do not cast to
21878 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
21880 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21881 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
21882 (inline_edge_summary_vec): Turn into ...
21883 (ipa_call_summaries): ... this one.
21884 (redirect_to_unreachable, edge_set_predicate,
21885 evaluate_properties_for_edge, inline_summary_alloc,
21886 reset_ipa_call_summary, reset_inline_summary,
21887 inline_summary_t::duplicate): Update.
21888 (inline_edge_duplication_hook): Turn to ...
21889 (ipa_call_summary_t::duplicate): ... this one.
21890 (inline_edge_removal_hook): Turn to ...
21891 (ipa_call_summary_t::remove): ... this one.
21892 (dump_inline_edge_summary): Turn to ...
21893 (dump_ipa_call_summary): ... this one.
21894 (estimate_function_body_sizes): Update.
21895 (inline_update_callee_summaries): Update.
21896 (remap_edge_change_prob): Update.
21897 (remap_edge_summaries): Update.
21898 (inline_merge_summary): Update.
21899 (do_estimate_edge_time): Update.
21900 (inline_generate_summary): Update.
21901 (inline_read_section): Update.
21902 (inline_read_summary): Update.
21903 (inline_free_summary): Update.
21904 * ipa-inline.c (can_inline_edge_p): Update.
21905 (compute_inlined_call_time): Update.
21906 (want_inline_small_function_p): Update.
21907 (edge_badness): Update.
21908 (early_inliner): Update.
21909 * ipa-inline.h (inline_edge_summary): Turn to ...
21910 (ipa_call_summary): ... this one.
21911 (ipa_call_summary_t): New class.
21912 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
21913 (ipa_call_summaries): New.
21914 (inline_edge_summary): Remove.
21915 (estimate_edge_growth): Update.
21916 * ipa-profile.c (ipa_propagate_frequency_1): Update.
21917 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21918 * ipa-split.c (execute_split_functions): Update.
21919 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
21921 2017-05-19 Richard Biener <rguenther@suse.de>
21923 PR middle-end/80764
21924 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
21926 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
21928 * config/rs6000/rs6000.c (struct machine_function): Add field
21929 fpr_is_wrapped_separately.
21930 (rs6000_get_separate_components): Use 64 components. Handle the
21931 new FPR components.
21932 (rs6000_components_for_bb): Handle the FPR components.
21933 (rs6000_emit_prologue_components): Handle the FPR components.
21934 (rs6000_emit_epilogue_components): Handle the FPR components.
21935 (rs6000_set_handled_components): Handle the FPR components.
21936 (rs6000_emit_prologue): Don't output prologue code for those FPRs
21937 that are already separately shrink-wrapped.
21938 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
21939 that are already separately shrink-wrapped.
21941 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21944 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
21947 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
21948 (define_peephole2 for Altivec d-form load): Add peepholes to catch
21949 cases where the register allocator uses a move and an offsettable
21950 memory operation to/from a FPR register on ISA 2.06/2.07.
21951 (define_peephole2 for Altivec d-form store): Likewise.
21953 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
21956 * config/i386/mmx.md (*mov<mode>_internal): Enable
21957 alternatives 11, 12, 13 and 14 also for 32bit targets.
21958 Remove alternatives 15, 16, 17 and 18.
21959 * config/i386/sse.md (vec_concatv2di): Change
21960 alternative (!x, *y) to (x, ?!*Yn).
21962 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
21964 * dumpfile.h (enum dump_kind): Remove stray comma.
21966 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
21968 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
21969 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
21970 predicate::num_conditions
21971 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
21972 (CHANGED): turn into predicate::changed.
21973 (agg_position_info): Move to ipa-predicate.h
21974 (add_condition, predicate::add_clause, predicate::operator &=,
21975 predicate::or_with, predicate::evaluate, predicate::probability,
21976 dump_condition, dump_clause, predicate::dump,
21977 predicate::remap_after_duplication, predicate::remap_after_inlining,
21978 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
21979 (evaluate_conditions_for_known_args): Update.
21980 (set_cond_stmt_execution_predicate): Update.
21981 * ipa-inline.h: Include ipa-predicate.h
21982 (condition, inline_param_summary, conditions, agg_position_info,
21983 predicate): Move to ipa-predicate.h
21984 * ipa-predicate.c: New file.
21985 * ipa-predicate.h: New file.
21987 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
21989 * final.c (leaf_function_p): Check we are not in a sequence.
21991 2017-05-18 Martin Liska <mliska@suse.cz>
21993 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
21994 * dumpfile.c (dump_register): Use new enum dump_kind.
21995 (get_dump_file_name): Likewise.
21996 (dump_enable_all): Likewise.
21997 (dump_switch_p_1): Likewise.
21998 (enable_rtl_dump_file): Remove usage of TDF_RTL.
21999 * dumpfile.h (enum dump_kind): New enum type.
22000 (struct dump_file_info): Create constructor and
22001 format fields and comments.
22002 * passes.c (pass_manager::register_one_dump_file):
22004 * statistics.c (statistics_early_init): Likewise.
22005 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22006 TDF_TREE with TDF_SLIM.
22007 (gather_memory_references_ref): Likewise.
22009 2017-05-18 Martin Liska <mliska@suse.cz>
22011 * vec.h (struct vnull): Use it.
22013 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22015 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22016 (true_predicate, false_predicate, true_predicate_p,
22017 false_predicate_p): Remove.
22018 (single_cond_predicate, not_inlined_predicate): Turn to member function
22020 (add_condition): Update.
22021 (add_clause): Turn to...
22022 (predicate::add_clause): ... this one; update; allow passing NULL
22024 (and_predicates): Turn to ...
22025 (predicate::operator &=): ... this one.
22026 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22027 (or_predicates): Turn to ...
22028 (predicate::or_with): ... this one.
22029 (evaluate_predicate): Turn to ...
22030 (predicate::evaluate): ... this one.
22031 (predicate_probability): Turn to ...
22032 (predicate::probability): ... this one.
22033 (dump_condition): Update.
22034 (dump_predicate): Turn to ...
22035 (predicate::dump): ... this one.
22036 (account_size_time): Update.
22037 (edge_set_predicate): Update.
22038 (set_hint_predicate): UPdate.
22039 (evaluate_conditions_for_known_args): Update.
22040 (evaluate_properties_for_edge): Update.
22041 (remap_predicate_after_duplication): Turn to...
22042 (predicate::remap_after_duplication): ... this one.
22043 (remap_hint_predicate_after_duplication): Update.
22044 (inline_summary_t::duplicate): UPdate.
22045 (dump_inline_edge_summary): Update.
22046 (dump_inline_summary): Update.
22047 (set_cond_stmt_execution_predicate): Update.
22048 (set_switch_stmt_execution_predicate): Update.
22049 (compute_bb_predicates): Update.
22050 (will_be_nonconstant_expr_predicate): Update.
22051 (will_be_nonconstant_predicate): Update.
22052 (phi_result_unknown_predicate): Update.
22053 (predicate_for_phi_result): Update.
22054 (array_index_predicate): Update.
22055 (estimate_function_body_sizes): Update.
22056 (estimate_node_size_and_time): Update.
22057 (estimate_ipcp_clone_size_and_time): Update.
22058 (remap_predicate): Rename to ...
22059 (predicate::remap_after_inlining): ... this one.
22060 (remap_hint_predicate): Update.
22061 (inline_merge_summary): Update.
22062 (inline_update_overall_summary): Update.
22063 (estimate_size_after_inlining): Update.
22064 (read_predicate): Rename to ...
22065 (predicate::stream_in): ... this one.
22066 (read_inline_edge_summary): Update.
22067 (write_predicate): Rename to ...
22068 (predicate::stream_out): ... this one.
22069 (write_inline_edge_summary): Update.
22070 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22071 (clause_t): Turn to uint32_t
22072 (predicate): Turn to class; implement constructor and operators
22074 (size_time_entry): Update.
22075 (inline_summary): Update.
22076 (inline_edge_summary): Update.
22078 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22080 * fold-const.c (fold_binary_loc): Move transformation...
22081 * match.pd (C - X CMP X): ... here.
22083 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22085 * config/sparc/sparc.c (sparc_option_override): Set function
22086 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22087 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22089 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22091 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22093 2017-05-18 Marek Polacek <polacek@redhat.com>
22096 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22097 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22098 gimple_assign_load_p.
22100 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22102 PR middle-end/80692
22103 * real.c (do_compare): Give decimal_do_compare preference over
22104 comparing just the signs.
22106 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22108 * doc/md.texi (Canonicalization of Instructions): Describe the
22109 canonical form of instructions that inherently set a condition
22112 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22114 PR middle-end/80775
22115 * tree-cfg.c: Move deletion of unreachable case statements to after
22116 the merging of consecutive case labels.
22118 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22120 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22121 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22122 restoring of callee-saved registers.
22124 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22126 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22127 * config/visium/visium.c (single_set_and_flags): Likewise.
22128 * config/visium/visium.md (Substitutions): Likewise.
22130 2017-05-17 Martin Liska <mliska@suse.cz>
22132 * cfg.c: Introduce dump_flags_t type and
22133 use it instead of int type.
22135 * cfghooks.c: Likewise.
22136 * cfghooks.h (struct cfg_hooks): Likewise.
22137 * cfgrtl.c: Likewise.
22138 * cfgrtl.h: Likewise.
22139 * cgraph.c (cgraph_node::get_body): Likewise.
22140 * coretypes.h: Likewise.
22141 * domwalk.c: Likewise.
22142 * domwalk.h: Likewise.
22143 * dumpfile.c (struct dump_option_value_info): Likewise.
22144 (dump_enable_all): Likewise.
22145 (dump_switch_p_1): Likewise.
22146 (opt_info_switch_p): Likewise.
22147 * dumpfile.h (enum tree_dump_index): Likewise.
22148 (struct dump_file_info): Likewise.
22149 * genemit.c: Likewise.
22150 * generic-match-head.c: Likewise.
22151 * gengtype.c (open_base_files): Likewise.
22152 * gimple-pretty-print.c: Likewise.
22153 * gimple-pretty-print.h: Likewise.
22154 * graph.c (print_graph_cfg): Likewise.
22155 * graphite-scop-detection.c (dot_all_sese): Likewise.
22156 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22157 * loop-unroll.c (report_unroll): Likewise.
22158 * passes.c (pass_manager::register_one_dump_file): Likewise.
22159 * print-tree.c: Likewise.
22160 * statistics.c: Likewise.
22161 * tree-cfg.c: Likewise.
22162 * tree-cfg.h: Likewise.
22163 * tree-dfa.c: Likewise.
22164 * tree-dfa.h: Likewise.
22165 * tree-dump.c (dump_function): Likewise.
22166 * tree-dump.h (struct dump_info): Likewise.
22167 * tree-pretty-print.c: Likewise.
22168 * tree-pretty-print.h: Likewise.
22169 * tree-ssa-live.c: Likewise.
22170 * tree-ssa-live.h: Likewise.
22171 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22172 * tree-vect-loop.c: Likewise.
22173 * tree-vect-slp.c: Likewise.
22175 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22176 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22178 PR tree-optimization/80457
22179 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22180 of all arguments to a statement as scalar_to_vec operations.
22181 (vectorizable_call): Adjust call to vect_model_simple_cost for
22183 (vectorizable_conversion): Likewise.
22184 (vectorizable_assignment): Likewise.
22185 (vectorizable_shift): Likewise.
22186 (vectorizable_operation): Likewise.
22187 (vectorizable_comparison): Likewise.
22188 (vect_is_simple_cond): Record the def types for operands.
22189 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22190 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22191 for statement argument count.
22193 2017-05-16 Carl Love <cel@us.ibm.com>
22195 * config/rs6000/rs6000-c: Add support for built-in functions
22196 vector unsigned long long vec_bperm (vector unsigned long long,
22197 vector unsigned char)
22198 vector signed long long vec_mule (vector signed int,
22200 vector unsigned long long vec_mule (vector unsigned int,
22201 vector unsigned int)
22202 vector signed long long vec_mulo (vector signed int,
22204 vector unsigned long long vec_mulo (vector unsigned int,
22205 vector unsigned int)
22206 vector signed char vec_sldw (vector signed char,
22207 vector signed char,
22209 vector unsigned char vec_sldw (vector unsigned char,
22210 vector unsigned char,
22212 vector signed short vec_sldw (vector signed short,
22213 vector signed short,
22215 vector unsigned short vec_sldw (vector unsigned short,
22216 vector unsigned short,
22218 vector signed int vec_sldw (vector signed int,
22221 vector unsigned int vec_sldw (vector unsigned int,
22222 vector unsigned int,
22224 vector signed long long vec_sldw (vector signed long long,
22225 vector signed long long,
22227 vector unsigned long long vec_sldw (vector unsigned long long,
22228 vector unsigned long long,
22230 * config/rs6000/rs6000-c: Add support for built-in functions
22231 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22232 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22233 * doc/extend.texi: Update the built-in documentation for the
22234 new built-in functions.
22236 2017-05-16 Marek Polacek <polacek@redhat.com>
22240 * tree.c (save_expr): Don't fold the expression.
22242 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22244 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22245 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22246 and (?*y,m). Update insn attributes.
22248 2017-05-16 Martin Liska <mliska@suse.cz>
22250 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
22251 flags argument of print_gimple_stmt, print_gimple_expr,
22252 print_generic_stmt and print_generic_expr.
22253 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
22254 * coretypes.h: Likewise.
22255 * except.c (dump_eh_tree): Likewise.
22256 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
22257 * gimple-pretty-print.h: Likewise.
22258 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
22259 (backprop::push_to_worklist): Likewise.
22260 (backprop::pop_from_worklist): Likewise.
22261 (backprop::process_use): Likewise.
22262 (backprop::intersect_uses): Likewise.
22263 (note_replacement): Likewise.
22264 * gimple-ssa-store-merging.c
22265 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
22266 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
22267 (pass_store_merging::execute): Likewise.
22268 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
22269 (ssa_base_cand_dump_callback): Likewise.
22270 (dump_incr_vec): Likewise.
22271 (replace_refs): Likewise.
22272 (replace_mult_candidate): Likewise.
22273 (create_add_on_incoming_edge): Likewise.
22274 (create_phi_basis): Likewise.
22275 (insert_initializers): Likewise.
22276 (all_phi_incrs_profitable): Likewise.
22277 (introduce_cast_before_cand): Likewise.
22278 (replace_one_candidate): Likewise.
22279 * gimplify.c (gimplify_expr): Likewise.
22280 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
22281 (set_rename): Likewise.
22282 (rename_uses): Likewise.
22283 (copy_loop_phi_nodes): Likewise.
22284 (add_close_phis_to_merge_points): Likewise.
22285 (copy_loop_close_phi_args): Likewise.
22286 (copy_cond_phi_args): Likewise.
22287 (graphite_copy_stmts_from_block): Likewise.
22288 (translate_pending_phi_nodes): Likewise.
22289 * graphite-poly.c (print_pdr): Likewise.
22290 (dump_gbb_cases): Likewise.
22291 (dump_gbb_conditions): Likewise.
22292 (print_scop_params): Likewise.
22293 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
22294 (build_cross_bb_scalars_use): Likewise.
22295 (gather_bbs::before_dom_children): Likewise.
22296 * hsa-dump.c (dump_hsa_immed): Likewise.
22297 * ipa-cp.c (print_ipcp_constant_value): Likewise.
22298 (get_replacement_map): Likewise.
22299 * ipa-inline-analysis.c (dump_condition): Likewise.
22300 (estimate_function_body_sizes): Likewise.
22301 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
22302 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
22303 * ipa-prop.c (ipa_dump_param): Likewise.
22304 (ipa_print_node_jump_functions_for_edge): Likewise.
22305 (ipa_modify_call_arguments): Likewise.
22306 (ipa_modify_expr): Likewise.
22307 (ipa_dump_param_adjustments): Likewise.
22308 (ipa_dump_agg_replacement_values): Likewise.
22309 (ipcp_modif_dom_walker::before_dom_children): Likewise.
22310 * ipa-pure-const.c (check_stmt): Likewise.
22311 (pass_nothrow::execute): Likewise.
22312 * ipa-split.c (execute_split_functions): Likewise.
22313 * omp-offload.c (dump_oacc_loop_part): Likewise.
22314 (dump_oacc_loop): Likewise.
22315 * trans-mem.c (tm_log_emit): Likewise.
22316 (tm_memopt_accumulate_memops): Likewise.
22317 (dump_tm_memopt_set): Likewise.
22318 (dump_tm_memopt_transform): Likewise.
22319 * tree-cfg.c (gimple_verify_flow_info): Likewise.
22320 (print_loop): Likewise.
22321 * tree-chkp-opt.c (chkp_print_addr): Likewise.
22322 (chkp_gather_checks_info): Likewise.
22323 (chkp_get_check_result): Likewise.
22324 (chkp_remove_check_if_pass): Likewise.
22325 (chkp_use_outer_bounds_if_possible): Likewise.
22326 (chkp_reduce_bounds_lifetime): Likewise.
22327 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
22328 (chkp_mark_completed_bounds): Likewise.
22329 (chkp_register_incomplete_bounds): Likewise.
22330 (chkp_mark_invalid_bounds): Likewise.
22331 (chkp_maybe_copy_and_register_bounds): Likewise.
22332 (chkp_build_returned_bound): Likewise.
22333 (chkp_get_bound_for_parm): Likewise.
22334 (chkp_build_bndldx): Likewise.
22335 (chkp_get_bounds_by_definition): Likewise.
22336 (chkp_generate_extern_var_bounds): Likewise.
22337 (chkp_get_bounds_for_decl_addr): Likewise.
22338 * tree-chrec.c (chrec_apply): Likewise.
22339 * tree-data-ref.c (dump_data_reference): Likewise.
22340 (dump_subscript): Likewise.
22341 (dump_data_dependence_relation): Likewise.
22342 (analyze_overlapping_iterations): Likewise.
22343 * tree-inline.c (expand_call_inline): Likewise.
22344 (tree_function_versioning): Likewise.
22345 * tree-into-ssa.c (dump_defs_stack): Likewise.
22346 (dump_currdefs): Likewise.
22347 (dump_names_replaced_by): Likewise.
22348 (dump_update_ssa): Likewise.
22349 (update_ssa): Likewise.
22350 * tree-object-size.c (pass_object_sizes::execute): Likewise.
22351 * tree-parloops.c (build_new_reduction): Likewise.
22352 (try_create_reduction_list): Likewise.
22353 (ref_conflicts_with_region): Likewise.
22354 (oacc_entry_exit_ok_1): Likewise.
22355 (oacc_entry_exit_single_gang): Likewise.
22356 * tree-pretty-print.h: Likewise.
22357 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
22358 (get_scalar_evolution): Likewise.
22359 (add_to_evolution): Likewise.
22360 (get_loop_exit_condition): Likewise.
22361 (analyze_evolution_in_loop): Likewise.
22362 (analyze_initial_condition): Likewise.
22363 (analyze_scalar_evolution): Likewise.
22364 (instantiate_scev): Likewise.
22365 (number_of_latch_executions): Likewise.
22366 (gather_chrec_stats): Likewise.
22367 (final_value_replacement_loop): Likewise.
22368 (scev_const_prop): Likewise.
22369 * tree-sra.c (dump_access): Likewise.
22370 (disqualify_candidate): Likewise.
22371 (create_access): Likewise.
22372 (reject): Likewise.
22373 (maybe_add_sra_candidate): Likewise.
22374 (create_access_replacement): Likewise.
22375 (analyze_access_subtree): Likewise.
22376 (analyze_all_variable_accesses): Likewise.
22377 (sra_modify_assign): Likewise.
22378 (initialize_constant_pool_replacements): Likewise.
22379 (find_param_candidates): Likewise.
22380 (decide_one_param_reduction): Likewise.
22381 (replace_removed_params_ssa_names): Likewise.
22382 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
22383 * tree-ssa-copy.c (dump_copy_of): Likewise.
22384 (copy_prop_visit_cond_stmt): Likewise.
22385 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
22386 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
22387 (record_equivalences_from_stmt): Likewise.
22388 * tree-ssa-dse.c (compute_trims): Likewise.
22389 (delete_dead_call): Likewise.
22390 (delete_dead_assignment): Likewise.
22391 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
22392 (forward_propagate_into_cond): Likewise.
22393 (pass_forwprop::execute): Likewise.
22394 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
22395 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
22397 (move_computations_worker): Likewise.
22398 (execute_sm): Likewise.
22399 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
22400 (remove_exits_and_undefined_stmts): Likewise.
22401 (remove_redundant_iv_tests): Likewise.
22402 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
22403 (adjust_iv_update_pos): Likewise.
22404 * tree-ssa-math-opts.c (bswap_replace): Likewise.
22405 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
22406 (value_replacement): Likewise.
22407 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
22408 * tree-ssa-pre.c (print_pre_expr): Likewise.
22409 (get_representative_for): Likewise.
22410 (create_expression_by_pieces): Likewise.
22411 (insert_into_preds_of_block): Likewise.
22412 (eliminate_insert): Likewise.
22413 (eliminate_dom_walker::before_dom_children): Likewise.
22414 (eliminate): Likewise.
22415 (remove_dead_inserted_code): Likewise.
22416 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
22417 * tree-ssa-reassoc.c (get_rank): Likewise.
22418 (eliminate_duplicate_pair): Likewise.
22419 (eliminate_plus_minus_pair): Likewise.
22420 (eliminate_not_pairs): Likewise.
22421 (undistribute_ops_list): Likewise.
22422 (eliminate_redundant_comparison): Likewise.
22423 (update_range_test): Likewise.
22424 (optimize_range_tests_var_bound): Likewise.
22425 (optimize_vec_cond_expr): Likewise.
22426 (rewrite_expr_tree): Likewise.
22427 (rewrite_expr_tree_parallel): Likewise.
22428 (linearize_expr): Likewise.
22429 (break_up_subtract): Likewise.
22430 (linearize_expr_tree): Likewise.
22431 (attempt_builtin_powi): Likewise.
22432 (attempt_builtin_copysign): Likewise.
22433 (transform_stmt_to_copy): Likewise.
22434 (transform_stmt_to_multiply): Likewise.
22435 (dump_ops_vector): Likewise.
22436 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
22437 (print_scc): Likewise.
22438 (set_ssa_val_to): Likewise.
22439 (visit_reference_op_store): Likewise.
22440 (visit_use): Likewise.
22441 (sccvn_dom_walker::before_dom_children): Likewise.
22442 (run_scc_vn): Likewise.
22443 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
22445 (expr_hash_elt::print): Likewise.
22446 (const_and_copies::pop_to_marker): Likewise.
22447 (const_and_copies::record_const_or_copy_raw): Likewise.
22448 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
22449 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
22450 (dump_predicates): Likewise.
22451 (find_uninit_use): Likewise.
22452 (warn_uninitialized_phi): Likewise.
22453 (pass_late_warn_uninitialized::execute): Likewise.
22454 * tree-ssa.c (verify_vssa): Likewise.
22455 (verify_ssa): Likewise.
22456 (maybe_optimize_var): Likewise.
22457 * tree-vrp.c (dump_value_range): Likewise.
22458 (dump_all_value_ranges): Likewise.
22459 (dump_asserts_for): Likewise.
22460 (register_edge_assert_for_2): Likewise.
22461 (vrp_visit_cond_stmt): Likewise.
22462 (vrp_visit_switch_stmt): Likewise.
22463 (vrp_visit_stmt): Likewise.
22464 (vrp_visit_phi_node): Likewise.
22465 (simplify_cond_using_ranges_1): Likewise.
22466 (fold_predicate_in): Likewise.
22467 (evrp_dom_walker::before_dom_children): Likewise.
22468 (evrp_dom_walker::push_value_range): Likewise.
22469 (evrp_dom_walker::pop_value_range): Likewise.
22470 (execute_early_vrp): Likewise.
22472 2017-05-16 Richard Biener <rguenther@suse.de>
22474 * dwarf2out.c (loc_list_from_tree_1): Do not create
22475 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
22477 2017-05-16 Richard Biener <rguenther@suse.de>
22479 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
22481 (note_variable_value_in_expr): If we resolved the decl ref
22482 do not push to the stack.
22484 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
22486 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
22487 operations in fast-math mode.
22488 (vaddq_f16): Likewise.
22489 (vmul_f16): Likewise.
22490 (vmulq_f16): Likewise.
22491 (vsub_f16): Likewise.
22492 (vsubq_f16): Likewise.
22493 * config/arm/neon.md (add<mode>3): New.
22495 (fma:<VH:mode>3): New. Also remove outdated comment.
22498 2017-05-16 Martin Liska <mliska@suse.cz>
22502 * ipa-devirt.c (warn_types_mismatch): Fix typo.
22503 (odr_types_equivalent_p): Likewise.
22505 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
22507 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
22509 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
22512 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
22513 non-interunit SSE move alternatives with '?'.
22514 (zero-extendsidi peephole2): New peephole to skip intermediate
22515 general register in SSE zero-extend sequence.
22517 2017-05-15 Jeff Law <law@redhat.com>
22519 * reorg.c (relax_delay_slots): Create a new variable to hold
22520 the temporary target rather than clobbering TARGET_LABEL.
22522 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
22523 missing argument to extract_bit_field call.
22524 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
22526 2017-05-15 Martin Liska <mliska@suse.cz>
22529 * gcc.c (process_command): Do not allow empty argument of -o option.
22531 2017-05-15 Renlin Li <renlin.li@arm.com>
22533 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
22534 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
22535 * config/aarch64/constraints.md (Usf): Add long call check.
22536 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
22537 (call_value): Likewise.
22538 (sibcall): Likewise.
22539 (sibcall_value): Likewise.
22541 (call_value_insn): New.
22542 (sibcall_insn): Update rtx pattern.
22543 (sibcall_value_insn): Likewise.
22544 (call_internal): Remove.
22545 (call_value_internal): Likewise.
22546 (sibcall_internal): Likewise.
22547 (sibcall_value_internal): Likewise.
22548 (call_reg): Likewise.
22549 (call_symbol): Likewise.
22550 (call_value_reg): Likewise.
22551 (call_value_symbol): Likewise.
22553 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
22556 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
22558 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
22560 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
22561 compatible with CCGOCmode and with CCZmode.
22563 2017-05-14 Martin Sebor <msebor@redhat.com>
22565 PR middle-end/77671
22566 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
22567 (gimple_fold_builtin_snprintf): Same.
22568 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
22569 (gimple_fold_builtin_snprintf): Same.
22570 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
22571 of character types.
22572 (is_call_safe): New function.
22573 (try_substitute_return_value): Call it.
22574 (try_simplify_call): New function.
22575 (pass_sprintf_length::handle_gimple_call): Call it.
22577 2017-05-14 Martin Sebor <msebor@redhat.com>
22579 PR middle-end/80669
22580 * builtins.c (expand_builtin_stpncpy): Simplify.
22582 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
22584 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
22585 * config/i386/i386.h
22586 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
22587 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
22588 (struct machine_function): Add new members call_ms2sysv,
22589 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
22590 (struct machine_frame_state): New fields sp_realigned and
22591 sp_realigned_offset.
22592 * config/i386/i386.c
22593 (enum xlogue_stub): New enum.
22594 (enum xlogue_stub_sets): New enum.
22595 (class xlogue_layout): New class.
22596 (struct ix86_frame): New fields stack_realign_allocate_offset,
22597 stack_realign_offset and outlined_save_offset. Modify comments to
22598 detail stack layout when using out-of-line stubs.
22599 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
22600 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
22601 -mcall-ms2sysv-xlogues.
22602 (stub_managed_regs): New static variable.
22603 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
22604 registers managed by out-of-line stub.
22605 (disable_call_ms2sysv_xlogues): New function.
22606 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
22607 m->call_ms2sysv when appropriate and compute frame layout for
22609 (sp_valid_at, fp_valid_at): New inline functions.
22610 (choose_basereg): New function.
22611 (choose_baseaddr): Add align parameter, use choose_basereg and modify
22613 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
22614 Use align parameter of choose_baseaddr to generated aligned SSE movs
22616 (pro_epilogue_adjust_stack): Modify to track
22617 machine_frame_state::sp_realigned.
22618 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
22619 (ix86_nsaved_sseregs): Likewise.
22620 (ix86_emit_save_regs): Likewise.
22621 (ix86_emit_save_regs_using_mov): Likewise.
22622 (ix86_emit_save_sse_regs_using_mov): Likewise.
22623 (get_scratch_register_on_entry): Likewise.
22624 (gen_frame_set): New function.
22625 (gen_frame_load): Likewise.
22626 (gen_frame_store): Likewise.
22627 (emit_outlined_ms2sysv_save): Likewise.
22628 (emit_outlined_ms2sysv_restore): Likewise.
22629 (ix86_expand_prologue): Modify stack re-alignment code and call
22630 emit_outlined_ms2sysv_save when appropriate.
22631 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
22632 parameter rtx_insn *insn, which allows the function to be used to only
22633 generate the notes.
22634 (ix86_expand_epilogue): Modify validity checks of frame and stack
22635 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
22636 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
22637 * config/i386/predicates.md
22638 (save_multiple): New predicate.
22639 (restore_multiple): Likewise.
22640 * config/i386/sse.md
22641 (save_multiple<mode>): New pattern.
22642 (save_multiple_realign<mode>): Likewise.
22643 (restore_multiple<mode>): Likewise.
22644 (restore_multiple_and_return<mode>): Likewise.
22645 (restore_multiple_leave_return<mode>): Likewise.
22646 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
22648 2017-05-14 Julia Koval <julia.koval@intel.com>
22650 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
22651 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
22652 (__builtin_ia32_xsetbv): New builtins.
22653 * config/i386/i386.c (ix86_expand_special_args_builtin):
22655 (ix86_expand_builtin): Special expand for new intrinsics.
22656 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
22657 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
22658 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
22660 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22662 * cfganal.c (inverted_post_order_compute): Change argument type
22664 * cfganal.h (inverted_post_order_compute): Adjust prototype.
22665 * df-core.c (rest_of_handle_df_initialize): Adjust.
22666 (rest_of_handle_df_finish): Likewise.
22667 (df_analyze_1): Likewise.
22668 (df_analyze): Likewise.
22669 (loop_inverted_post_order_compute): Change argument to be a vec *.
22670 (df_analyze_loop): Adjust.
22671 (df_get_n_blocks): Likewise.
22672 (df_get_postorder): Likewise.
22673 * df.h (struct df_d): Change field to be a vec.
22674 * lcm.c (compute_laterin): Adjust.
22675 (compute_available): Likewise.
22676 * lra-lives.c (lra_create_live_ranges_1): Likewise.
22677 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
22678 * tree-ssa-pre.c (compute_antic): Likewise.
22680 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22682 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
22683 (depth_first_search::depth_first_search): Change structure init
22684 function to this constructor.
22685 (depth_first_search::add_bb): Rename function to this member.
22686 (depth_first_search::execute): Likewise.
22687 (flow_dfs_compute_reverse_finish): Adjust.
22689 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22691 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
22692 (longest_simple_path): Likewise.
22693 * shrink-wrap.c (spread_components): Likewise.
22694 (disqualify_problematic_components): Likewise.
22695 (emit_common_heads_for_components): Likewise.
22696 (emit_common_tails_for_components): Likewise.
22697 (insert_prologue_epilogue_for_components): Likewise.
22699 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22701 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
22704 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22706 * df-core.c (df_set_blocks): Start using auto_bitmap.
22707 (df_compact_blocks): Likewise.
22708 * df-problems.c (df_rd_confluence_n): Likewise.
22709 * df-scan.c (df_insn_rescan_all): Likewise.
22710 (df_process_deferred_rescans): Likewise.
22711 (df_update_entry_block_defs): Likewise.
22712 (df_update_exit_block_uses): Likewise.
22713 (df_entry_block_bitmap_verify): Likewise.
22714 (df_exit_block_bitmap_verify): Likewise.
22715 (df_scan_verify): Likewise.
22716 * lra-constraints.c (lra_constraints): Likewise.
22717 (undo_optional_reloads): Likewise.
22718 (lra_undo_inheritance): Likewise.
22719 * lra-remat.c (calculate_gen_cands): Likewise.
22720 (do_remat): Likewise.
22721 * lra-spills.c (assign_spill_hard_regs): Likewise.
22722 (spill_pseudos): Likewise.
22723 * tree-ssa-pre.c (bitmap_set_and): Likewise.
22724 (bitmap_set_subtract_values): Likewise.
22726 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22728 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
22729 management with auto_bitmap.
22730 (fix_inter_tick): Likewise.
22731 (fix_recovery_deps): Likewise.
22732 * ira.c (add_store_equivs): Likewise.
22733 (find_moveable_pseudos): Likewise.
22734 (split_live_ranges_for_shrink_wrap): Likewise.
22735 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
22736 (rtx_reuse_manager::seen_def_p): Likewise.
22737 (rtx_reuse_manager::set_seen_def): Likewise.
22738 * print-rtl.h (class rtx_reuse_manager): Likewise.
22740 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22742 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
22744 (migrate_btr_def): Likewise.
22745 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
22746 * df-core.c (loop_post_order_compute): Likewise.
22747 (loop_inverted_post_order_compute): Likewise.
22748 * hsa-common.h: Likewise.
22749 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
22750 * init-regs.c (initialize_uninitialized_regs): Likewise.
22751 * ipa-inline.c (resolve_noninline_speculation): Likewise.
22752 (inline_small_functions): Likewise.
22753 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
22754 * ira.c (combine_and_move_insns): Likewise.
22755 (build_insn_chain): Likewise.
22756 * loop-invariant.c (find_invariants): Likewise.
22757 * lower-subreg.c (propagate_pseudo_copies): Likewise.
22758 * predict.c (tree_predict_by_opcode): Likewise.
22759 (predict_paths_leading_to): Likewise.
22760 (predict_paths_leading_to_edge): Likewise.
22761 (estimate_loops_at_level): Likewise.
22762 (estimate_loops): Likewise.
22763 * shrink-wrap.c (try_shrink_wrapping): Likewise.
22764 (spread_components): Likewise.
22765 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
22766 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
22767 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
22768 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
22769 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
22770 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
22771 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
22772 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
22773 (mark_threaded_blocks): Likewise.
22774 (thread_through_all_blocks): Likewise.
22775 * tree-ssa.c (verify_ssa): Likewise.
22776 (execute_update_addresses_taken): Likewise.
22777 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
22779 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22781 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
22783 (post_order_compute): Likewise.
22784 (inverted_post_order_compute): Likewise.
22785 (pre_and_rev_post_order_compute_fn): Likewise.
22787 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22789 * genrecog.c (int_set::int_set): Explicitly construct our
22790 auto_vec base class.
22791 * vec.h (auto_vec::auto_vec): New constructor.
22793 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22795 * bitmap.h (class auto_bitmap): New constructor taking
22796 bitmap_obstack * argument.
22798 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22800 * bitmap.h (class auto_bitmap): Change type of m_bits to
22801 bitmap_head, and adjust ctor / dtor and member operators.
22803 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
22805 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
22806 when returned register mode doesn't match original mode.
22808 2017-05-12 Jeff Law <law@redhat.com>
22809 Jakub Jelinek <jakub@redhat.com>
22811 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
22812 we look for cc setter after the compare-elim changes.
22813 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
22814 within the vector to match what compare-elim now expects.
22815 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
22816 (xorsi3_flags, one_cmplsi2_flags): Likewise.
22818 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
22819 after the compare-elim changes.
22820 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
22821 the vector to match what compare-elim now expects.
22822 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
22823 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
22824 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
22825 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
22826 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
22828 * config/visium/visium.c (single_set_and_flags): Fix where
22829 we look for cc setter after the compare-elim changes.
22830 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
22831 with the vector to match what compare-elim now expects.
22832 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
22833 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
22834 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
22835 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
22836 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
22837 (neg<mode>2_insn_set_overflow): Likewise.
22839 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
22841 PR middle-end/79794
22842 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
22843 maybe_expand_insn call, set ops[0].target. If still set after call,
22844 set alt_rtl. Add extra arg to recursive calls.
22845 (extract_bit_field): Add alt_rtl argument. Pass to
22847 * expmed.h (extract_bit_field): Fix prototype.
22848 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
22849 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
22850 to extract_bit_field_calls.
22851 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
22852 Pass alt_rtl to extract_bit_field calls.
22853 * calls.c (store_unaligned_arguments_into_psuedos)
22854 load_register_parameters): Pass extra NULL to extract_bit_field calls.
22855 * optabs.c (maybe_legitimize_operand): Clear op->target when call
22857 * optabs.h (struct expand_operand): Add target bitfield.
22859 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
22861 * compare-elim.c (try_eliminate_compare): Canonicalize
22862 operation with embedded compare to
22863 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
22864 (set (reg) (operation)].
22866 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
22868 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
22871 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
22872 cost of adding a carry flag for ADC instruction.
22873 [case MINUS]: Ignore the cost of subtracting a carry flag
22874 for SBB instruction.
22876 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
22878 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
22880 * config/rs6000/bmiintrin.h: New file.
22881 * config/rs6000/bmi2intrin.h: New file.
22882 * config/rs6000/x86intrin.h: New file.
22884 2017-05-12 Jeff Law <law@redhat.com>
22886 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
22889 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
22891 PR middle-end/80707
22892 * tree-cfg.c: Remove cfg edges of unreachable case statements.
22894 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22896 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
22897 early expansion of vector divide builtins.
22898 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
22899 builtins identified as having unsigned arguments.
22901 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22903 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
22904 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
22905 expansion of vector logical operations (and, andc, or, xor,
22908 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22910 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
22911 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
22913 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22915 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
22916 early GIMPLE expansion of vector multiplies.
22918 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22920 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
22921 TARGET_HAVE_MOVT conditional.
22922 (movt splitter): Likewise.
22924 2017-05-12 Richard Biener <rguenther@suse.de>
22926 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
22927 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
22928 Fold all stmts not inplace.
22930 2017-05-12 Richard Biener <rguenther@suse.de>
22932 PR tree-optimization/80713
22933 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
22934 inserted_exprs bit for not removed stmts.
22936 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
22938 PR middle-end/69921
22939 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
22940 parallelized" attribute for parallelized OpenACC kernels.
22941 * omp-offload.c (execute_oacc_device_lower): Use it.
22943 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
22944 Set "oacc kernels" attribute.
22945 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
22946 parameter. Adjust all users.
22947 (oacc_fn_attrib_kernels_p): Remove function.
22948 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
22949 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
22950 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
22951 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
22952 assert "oacc kernels" attribute is set.
22954 2017-05-11 Carl Love <cel@us.ibm.com>
22956 * config/rs6000/rs6000-c: Add support for built-in functions
22957 vector unsigned char vec_popcnt (vector signed char)
22958 vector unsigned char vec_popcnt (vector unsigned char)
22959 vector unsigned short vec_popcnt (vector signed short)
22960 vector unsigned short vec_popcnt (vector unsigned short)
22961 vector unsigned int vec_popcnt (vector signed int)
22962 vector unsigned int vec_popcnt (vector unsigned int)
22963 vector unsigned long long vec_popcnt (vector signed long long)
22964 vector unsigned long long vec_popcnt (vector unsigned long long)
22965 vector signed long long vec_slo (vector signed long long,
22966 vector signed char)
22967 vector signed long long vec_slo (vector signed long long,
22968 vector unsigned char)
22969 vector unsigned long long vec_slo (vector unsigned long long,
22970 vector signed char)
22971 vector unsigned long long vec_slo (vector unsigned long long,
22972 vector unsigned char)
22973 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
22974 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
22975 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
22976 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
22977 * doc/extend.texi: Update the built-in documentation file for the
22978 new built-in functions.
22980 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
22982 * attribs.h (sorted_attr_string): Move machine independent
22983 functions for target clone support from the i386 port to common
22984 code. Rename ix86_function_versions to common_function_versions.
22985 Rename make_name to make_unique_name.
22986 (common_function_versions): Likewise.
22987 (make_unique_name): Likewise.
22988 (make_dispatcher_decl): Likewise.
22989 (is_function_default_version): Likewise.
22990 * attribs.c (attr_strcmp): Likewise.
22991 (sorted_attr_string): Likewise.
22992 (common_function_versions): Likewise.
22993 (make_unique_name): Likewise.
22994 (make_dispatcher_decl): Likewise.
22995 (is_function_default_version): Likewise.
22996 * config/i386/i386.c (attr_strcmp): Likewise.
22997 (sorted_attr_string): Likewise.
22998 (ix86_function_versions): Likewise.
22999 (make_name): Likewise.
23000 (make_dispatcher_decl): Likewise.
23001 (is_function_default_version): Likewise.
23002 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23004 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23007 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23008 Account for direct move costs for vec_construct of integer
23011 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23014 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23015 (UNSPEC_STX_ATOMIC): Ditto.
23016 (loaddi_via_sse): New insn.
23017 (storedi_via_sse): Ditto.
23018 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23019 Update corresponding peephole2 patterns.
23020 (atomic_storedi_fpu): Ditto.
23022 2017-05-11 Julia Koval <julia.koval@intel.com>
23024 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23025 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23027 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23028 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23029 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23031 2017-05-11 Nathan Sidwell <nathan@acm.org>
23033 * graphite-poly.c: Include dumpfile.h.
23035 * dumpfle.h (dump_function): Declare here ...
23036 * tree-dump.h (dump_function): ... not here.
23037 * dumpfile.c: #include tree-cfg.h.
23038 (dump_function): Move here from ...
23039 * tree-dump.c (dump_function): ... here.
23040 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23041 * graphite-poly.c: Don't include tree-dump.h.
23042 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23043 * print-tree.c: Likewise.
23044 * stor-layout.c: Likewise.
23045 * tree-nested.c: Likewise.
23047 * dumpfile.c (dump_start): Use TDF_FLAGS.
23048 (dump_enable_all): Fix TDF_KIND check thinko.
23050 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23052 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23053 array entries to represent two legal parameterizations of the
23054 overloaded __builtin_cmpb function, as represented by the
23055 P6_OV_BUILTIN_CMPB constant.
23056 (altivec_resolve_overloaded_builtin): Add special case handling
23057 for the __builtin_cmpb function, as represented by the
23058 P6_OV_BUILTIN_CMPB constant.
23059 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23060 (BU_P6_64BIT_2): New macro.
23061 (BU_P6_OVERLOAD_2): New macro
23062 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23063 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23064 (CMPB): Add overload support to represent both 32-bit and 64-bit
23065 compare-bytes function.
23066 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23067 support for TARGET_CMPB.
23068 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23069 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23070 documentation of the __builtin_cmpb overloaded built-in function.
23072 2017-05-11 Richard Biener <rguenther@suse.de>
23074 PR tree-optimization/80705
23075 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23076 bases are not vectorizable.
23078 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23080 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23081 when counting register pressure.
23083 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23085 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23086 (iv_ca_more_deps): Renamed to ...
23087 (iv_ca_compare_deps): ... this.
23088 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23090 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23092 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23094 (determine_group_iv_costs): ... here.
23095 (find_inv_vars_cb): Record inv var if it's not recorded before.
23097 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23099 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23100 (get_shiftadd_cost): Ditto.
23102 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23104 * tree-ssa-address.c: Include header file.
23105 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23107 (add_to_parts): Refactor.
23108 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23109 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23112 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23114 PR tree-optimization/53090
23115 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23117 (extract_cond_operands): Detect condition with IV on both sides
23118 and return COMP_IV_EXPR_2.
23119 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23120 (rewrite_use_compare): Simplify by removing call to function
23121 extract_cond_operands.
23123 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23125 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23126 (extract_cond_operands): Detect condition comparing against non-
23127 invariant bound and return appropriate enum value.
23128 (find_interesting_uses_cond): Update use of extract_cond_operands.
23129 Handle its return value accordingly.
23130 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23132 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23134 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23135 nonlinear iv_use computation in loop invariant sensitive way.
23137 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23139 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23140 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23142 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23144 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23145 (dump_cand): Support iv_cand.inv_exprs.
23146 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23148 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23151 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23153 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23155 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23156 as local function. Include necessary header files.
23157 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23159 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23161 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23163 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23165 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23166 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23167 RSHIFT_EXPR and BIT_NOT_EXPR.
23169 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23171 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23172 (adjust_setup_cost): New parameter supporting round up adjustment.
23173 (struct address_cost_data): Delete.
23174 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23175 (split_address_cost, ptr_difference_cost): Delete.
23176 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23177 (struct ainc_cost_data): New struct.
23178 (get_address_cost_ainc): New function.
23179 (get_address_cost, get_computation_cost): Reimplement.
23180 (determine_group_iv_cost_address): Record inv_expr for all uses of
23182 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23183 (iv_ca_has_deps): Reimplemented to ...
23184 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23186 (iv_ca_extend): Call iv_ca_more_deps.
23188 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23190 * tree-ssa-address.c (struct mem_address): Move to header file.
23191 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23192 * tree-ssa-address.h (struct mem_address): Move from C file.
23193 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23195 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23197 * tree-affine.h (aff_combination_type): New interface.
23198 (aff_combination_zero_p): Remove static.
23199 (aff_combination_const_p): New interface.
23200 (aff_combination_singleton_var_p): New interfaces.
23202 2017-05-11 Richard Biener <rguenther@suse.de>
23204 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23205 Skip unreachable blocks and destinations.
23206 (eliminate): Move stmt removal and fixup ...
23207 (fini_eliminate): ... here. Skip inserted exprs.
23208 (pass_pre::execute): Move fini_pre after fini_eliminate.
23209 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23210 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23211 PRE to get rid of dead code that has invalid SSA form and
23212 split critical edges again.
23214 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23216 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23218 2017-05-11 Richard Biener <rguenther@suse.de>
23220 * passes.c (execute_function_todo): Verify loops if they are
23221 said to be up-to-date.
23222 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23223 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23225 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23228 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23229 handle calling assemble_external ourself.
23232 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23233 modes with zero size. Enhance comment.
23235 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23237 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23238 built-ins for vec_xl and vec_xst with short and char pointer
23241 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23243 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23244 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23245 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23246 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23247 (_mm_maskz_min_round_ss): New intrinsics.
23248 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23249 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23250 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
23251 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
23252 (__builtin_ia32_minss_mask_round): New builtins.
23253 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23254 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23255 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
23257 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
23258 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
23260 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
23263 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23265 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
23266 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
23267 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
23268 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
23269 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
23270 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
23271 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
23272 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23273 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23274 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
23275 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
23276 (__builtin_ia32_mulss_mask_round): New builtins.
23277 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23278 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23279 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
23281 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
23282 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23284 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23287 2017-05-10 Julia Koval <julia.koval@intel.com>
23289 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
23290 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
23291 (_mm256_setr_m128i): New intrinsics.
23293 2017-05-10 Julia Koval <julia.koval@intel.com>
23295 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
23296 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
23297 (_mm_maskz_rcp14_ss): New intrinsics.
23298 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
23299 (__builtin_ia32_rcp14ss_mask): New builtins.
23300 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
23302 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
23304 PR tree-optimization/51513
23305 * tree-cfg.c (gimple_seq_unreachable_p): New function.
23306 (assert_unreachable_fallthru_edge_p): Use it.
23307 (group_case_labels_stmt): Likewise.
23308 * tree-cfg.h: Prototype it.
23309 * stmt.c: Include cfghooks.h and tree-cfg.h.
23310 (emit_case_dispatch_table) <gap_label>: New local variable.
23311 Use it to fill dispatch table gaps.
23312 Test for default_label before updating probabilities.
23313 (expand_case) <default_label>: Remove unneeded initialization.
23314 Test for unreachable default case statement and remove its edge.
23315 Set default_label accordingly.
23316 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
23318 2017-05-10 Carl Love <cel@us.ibm.com>
23320 * config/rs6000/rs6000-c: Add support for built-in functions
23321 vector signed char vec_neg (vector signed char)
23322 vector signed short int vec_neg (vector short int)
23323 vector signed int vec_neg (vector signed int)
23324 vector signed long long vec_neg (vector signed long long)
23325 vector float vec_neg (vector float)
23326 vector double vec_neg (vector double)
23327 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
23329 * config/rs6000/altivec.h: Add define for vec_neg
23330 * doc/extend.texi: Update the built-in documentation for the
23331 new built-in functions.
23333 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23335 PR tree-optimization/77644
23336 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
23338 2017-05-10 Nathan Sidwell <nathan@acm.org>
23340 * dumpfile.h (TDI_lang_all): New.
23341 (TDF_KIND): New. Renumber others
23342 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
23344 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
23346 (get_dump_file_name): Adjust suffix generation.
23347 (dump_enable_all): Use TDF_KIND.
23348 * doc/invoke.texi (-fdump-lang-all): Document.
23350 * dumpfile.h: Tabify.
23352 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
23355 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
23356 Move member access before delete.
23358 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
23360 * tree-inline.c (expand_call_inline): Split block at stmt
23363 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
23366 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
23367 are now unused after splitting mov{sf,sd}_hardfloat.
23368 (f32_lr2): Likewise.
23369 (f32_lm): Likewise.
23370 (f32_lm2): Likewise.
23371 (f32_li): Likewise.
23372 (f32_li2): Likewise.
23373 (f32_lv): Likewise.
23374 (f32_sr): Likewise.
23375 (f32_sr2): Likewise.
23376 (f32_sm): Likewise.
23377 (f32_sm2): Likewise.
23378 (f32_si): Likewise.
23379 (f32_si2): Likewise.
23380 (f32_sv): Likewise.
23381 (f32_dm): Likewise.
23382 (f32_vsx): Likewise.
23383 (f32_av): Likewise.
23384 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
23385 For movsf, order stores so the VSX stores occur before the GPR
23386 store which encourages the register allocator to use a traditional
23387 FPR instead of a GPR. For movsd, order the stores so that the GPR
23388 store comes before the VSX stores to allow the power6 to work.
23389 This is due to the power6 not having a 32-bit integer store
23390 instruction from a FPR.
23391 (movsf_hardfloat): Likewise.
23392 (movsd_hardfloat): Likewise.
23394 2017-05-09 Martin Sebor <msebor@redhat.com>
23396 PR translation/80280
23397 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
23400 PR translation/80280
23401 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
23402 data member added in r247778.
23403 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
23405 2017-05-09 Nathan Sidwell <nathan@acm.org>
23407 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
23409 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
23412 2017-05-09 Marek Polacek <polacek@redhat.com>
23414 * doc/invoke.texi: Fix typo.
23416 2017-05-09 Richard Biener <rguenther@suse.de>
23418 * tree-vrp.c (vrp_val_is_max): Adjust comment.
23419 (vrp_val_is_min): Likewise.
23420 (set_value_range_to_value): Likewise.
23421 (set_value_range_to_nonnegative): Likewise.
23422 (gimple_assign_nonzero_p): Likewise.
23423 (gimple_stmt_nonzero_p): Likewise.
23424 (vrp_int_const_binop): Likewise. Remove unreachable case.
23425 (adjust_range_with_scev): Adjust comments.
23426 (compare_range_with_value): Likewise.
23427 (extract_range_from_phi_node): Likewise.
23428 (test_for_singularity): Likewise.
23430 2017-05-09 Richard Biener <rguenther@suse.de>
23432 * tree-vrp.c (get_single_symbol): Add assert that we don't
23433 get overflowed constants as invariant part.
23434 (compare_values_warnv): Add comment before the TREE_NO_WARNING
23435 checks. Use wi::cmp instead of recursing for integer constants.
23436 (compare_values): Just ignore whether we assumed undefined
23437 overflow instead of failing the compare.
23438 (extract_range_for_var_from_comparison_expr): Add comment before the
23439 TREE_NO_WARNING sets.
23440 (test_for_singularity): Likewise.
23441 (extract_range_from_comparison): Do not disable optimization
23442 when we assumed undefined overflow.
23443 (extract_range_basic): Remove init of unused var.
23445 2017-05-09 Richard Biener <rguenther@suse.de>
23447 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
23448 (extract_range_from_multiplicative_op_1): Adjust.
23449 (extract_range_from_binary_expr_1): Use int_const_binop.
23451 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
23454 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
23455 rs6000_store_data_bypass_p in seven define_bypass directives and
23456 in several comments.
23457 * config/rs6000/rs6000-protos.h: Add prototype for
23458 rs6000_store_data_bypass_p function.
23459 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
23460 function implements slightly different (rs6000-specific) semantics
23461 than store_data_bypass_p, returning false rather than aborting
23462 with assertion error when arguments do not satisfy the
23463 requirements of store data bypass.
23464 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
23465 rs6000_store_data_bypass_p.
23467 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
23469 * config/xtensa/xtensa-protos.h
23470 (xtensa_initial_elimination_offset): New declaration.
23471 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23472 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
23473 macro definition, add case for FRAME_POINTER_REGNUM when
23474 FRAME_GROWS_DOWNWARD.
23475 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
23476 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
23477 xtensa_initial_elimination_offset.
23479 2017-05-08 Nathan Sidwell <nathan@acm.org>
23481 * doc/invoke.texi: Alphabetize -fdump options.
23483 2017-05-08 Martin Sebor <msebor@redhat.com>
23485 PR translation/80280
23486 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
23488 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
23490 * target.def (compute_frame_layout): New optional target hook.
23491 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
23492 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
23493 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
23495 * reload1.c (verify_initial_elim_offsets): Likewise.
23496 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
23497 (use_simple_return_p): Call arm_compute_frame_layout if needed.
23498 (arm_get_frame_offsets): Split up into this ...
23499 (arm_compute_frame_layout): ... and this function.
23501 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
23503 * config/aarch64/constraints.md (Usa): New constraint.
23504 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
23506 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
23508 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
23509 with_multilib_list after it has been checked.
23511 2017-05-08 Richard Biener <rguenther@suse.de>
23513 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
23514 (bitmap_set_subtract_values): Likewise.
23516 2017-05-08 Richard Biener <rguenther@suse.de>
23518 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
23519 (gimple_assign_nonzero): ... this and remove strict_overflow_p
23521 (gimple_stmt_nonzero_warnv_p): Rename to ...
23522 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
23524 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
23525 (extract_range_basic): Adjust, do not disable propagation on
23526 strict overflow sensitive simplification.
23527 (vrp_visit_cond_stmt): Likewise.
23529 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
23531 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
23532 body size unconditionally.
23534 2017-05-07 Jeff Law <law@redhat.com>
23537 2017-05-06 Jeff Law <law@redhat.com>
23538 PR tree-optimization/78496
23539 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23542 PR tree-optimization/78496
23543 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23544 (simplify_stmt_using_ranges): Call it.
23545 (vrp_dom_walker::before_dom_children): Extract equivalences
23546 from an ASSERT_EXPR with an equality comparison against a
23549 2017-05-06 Jeff Law <law@redhat.com>
23551 PR tree-optimization/78496
23552 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23555 PR tree-optimization/78496
23556 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23557 (simplify_stmt_using_ranges): Call it.
23558 (vrp_dom_walker::before_dom_children): Extract equivalences
23559 from an ASSERT_EXPR with an equality comparison against a
23562 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23564 * lra-constraints.c (lra_copy_reg_equiv): New function.
23565 (split_reg): Use it to copy equivalence information from the
23566 original register to the spill register.
23568 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23570 PR rtl-optimization/75964
23571 * simplify-rtx.c (simplify_const_relational_operation): Remove
23572 invalid handling of comparisons of integer ABS.
23574 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
23576 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
23577 initialize to zero.
23578 (init_regs): Remove declaration.
23579 (function_arg_advance_32): Initialize error_p as boolean variable.
23581 2017-05-05 Nathan Sidwell <nathan@acm.org>
23583 * store-motion.c (remove_reachable_equiv_notes): Reformat long
23584 lines. Use for (;;).
23586 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23588 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
23589 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
23590 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
23591 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
23592 VF=2 that require versioning.
23594 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23596 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
23599 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23601 * diagnostic.h (diagnostic_override_option_index): Convert from
23602 macro to inline function.
23604 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23606 * diagnostic.c (last_module_changed_p): New function.
23607 (set_last_module): New function.
23608 (diagnostic_report_current_module): Convert macro usage to
23609 the above functions.
23610 * diagnostic.h (diagnostic_context::last_module): Strengthen
23611 from const line_map * to const line_map_ordinary *.
23612 (diagnostic_last_module_changed): Delete macro.
23613 (diagnostic_set_last_module): Delete macro.
23615 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23617 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
23618 with diagnostic_report_diagnostic.
23619 (diagnostic_n_impl_richloc): Likewise.
23620 * diagnostic.h (report_diagnostic): Delete macro.
23621 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
23622 with diagnostic_report_diagnostic.
23623 * substring-locations.c (format_warning_va): Likewise.
23625 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23627 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
23628 save/restor of format_spec. Move option-printing code to...
23629 (print_option_information): ...this new function, and
23630 reimplement by simply printing to the pretty_printer,
23631 rather than appending to the format string.
23633 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23635 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
23636 handling logic into...
23637 (update_effective_level_from_pragmas): ...this new function.
23639 2017-05-04 Andrew Waterman <andrew@sifive.com>
23641 * config/riscv/riscv.opt (mstrict-align): New option.
23642 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
23643 (SLOW_UNALIGNED_ACCESS): Define.
23644 (riscv_slow_unaligned_access): Declare.
23645 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
23647 (riscv_slow_unaligned_access): New variable.
23648 (rocket_tune_info): Set slow_unaligned_access to true.
23649 (optimize_size_tune_info): Set slow_unaligned_access to false.
23650 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
23651 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
23652 (riscv_option_override): Set riscv_slow_unaligned_access.
23653 * doc/invoke.texi: Add -mstrict-align to RISC-V.
23655 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
23657 * config/riscv/riscv.md: Unify indentation.
23659 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23664 * config/rs6000/rs6000.md (u code attribute): Add FIX and
23666 (extendsi<mode>2): Add support for doing sign extension via
23667 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
23668 don't have ISA 3.0 instructions.
23669 (extendsi<mode>2 splitter): Likewise.
23670 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
23671 generate the normal insns since SImode can now go in vector
23672 registers. Disallow the special UNSPECs needed for previous
23673 machines to hide SImode being used. Add new insns
23674 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
23675 (fix_trunc<mode>si2_stfiwx): Likewise.
23676 (fix_trunc<mode>si2_internal): Likewise.
23677 (fixuns_trunc<mode>si2): Likewise.
23678 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23679 (fctiw<u>z_<mode>_smallint): Likewise.
23680 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
23681 of floating point to 32-bit integer from doing a direct move to
23682 the GPR registers to do a store.
23683 (fctiwz_<mode>): Break long line.
23685 2017-05-05 Bin Cheng <bin.cheng@arm.com>
23687 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
23688 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
23689 (addr_list, addr_offset_valid_p): New.
23690 (split_address_groups): Check offset validity with above function.
23691 (gt-tree-ssa-loop-ivopts.h): Include header file.
23693 2017-05-05 Nathan Sidwell <nathan@acm.org>
23695 * config.gcc (arm*-*-*): Add missing 'fi'.
23697 2017-05-05 Steve Ellcey <sellcey@cavium.com>
23699 * doc/invoke.texi (-fopt-info): Explicitly say order of options
23700 included in -fopt-info does not matter.
23701 * doc/optinfo.texi (-fopt-info): Fix description of default
23702 behavour. Explicitly say order of options included in -fopt-info
23705 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
23707 * config.gcc: Allow combinations of aprofile and rmprofile values for
23708 --with-multilib-list.
23709 * config/arm/t-multilib: New file.
23710 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
23711 variables. Remove setting of ISA and floating-point ABI in
23712 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
23713 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
23714 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
23715 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
23717 * config/arm/t-rmprofile: Likewise except for the matches changes.
23718 * doc/install.texi (--with-multilib-list): Document the combination of
23719 aprofile and rmprofile values and warn about pitfalls in doing that.
23721 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
23723 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
23724 (movdi_aarch64): Likewise.
23726 2017-05-05 Jakub Jelinek <jakub@redhat.com>
23728 PR tree-optimization/80632
23729 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
23731 (build_arrays): Initialize it for virtual phis.
23732 (fix_phi_nodes): Use it for virtual phis.
23734 PR tree-optimization/80558
23735 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
23736 [x, y] op z into [x op, y op z] for op & or | if conditions
23739 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
23740 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23743 * config/arm/arm.md (use_literal_pool): Remove.
23744 (64-bit immediate split): No longer takes cost into consideration
23745 if arm_disable_literal_pool is enabled.
23746 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
23747 used when arm_disable_literal_pool is enabled.
23748 (arm_max_const_double_inline_cost): Remove use of
23749 arm_disable_literal_pool.
23750 (push_minipool_fix): Add assert.
23751 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
23752 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
23753 (no_literal_pool_sf_immediate): New.
23755 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23757 PR tree-optimization/80613
23758 * tree-ssa-dce.c (propagate_necessity): Remove cases for
23759 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
23761 2017-05-05 Richard Biener <rguenther@suse.de>
23763 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
23765 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
23767 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
23768 of this flag from insn conditions due to removal from r247495.
23770 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
23772 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
23774 (arm_early_store_addr_dep_ptr): Likewise.
23775 * config/arm/aarch-common-protos.h
23776 (arm_early_load_addr_dep_ptr): Add prototype.
23777 (arm_early_store_addr_dep_ptr): Likewise.
23778 * config/arm/cortex-a53.md: Add new bypasses.
23780 2017-05-05 Jakub Jelinek <jakub@redhat.com>
23782 * tree.c (next_type_uid): Change type to unsigned.
23783 (type_hash_canon): Decrement back next_type_uid if
23784 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
23785 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
23788 2017-05-04 Martin Sebor <msebor@redhat.com>
23790 * builtins.c: Fix a trivial typo in a comment.
23792 PR middle-end/79234
23793 * builtins.c (check_sizes): Adjust to handle reading past the end.
23794 Avoid printing excessive upper bound of ranges. Use %E to print
23795 tree nodes instead of converting them to %wu.
23796 (expand_builtin_memchr): New function.
23797 (compute_dest_size): Rename...
23798 (compute_objsize): ...to this.
23799 (expand_builtin_memcpy): Adjust.
23800 (expand_builtin_mempcpy): Adjust.
23801 (expand_builtin_strcat): Adjust.
23802 (expand_builtin_strcpy): Adjust.
23803 (check_strncat_sizes): Adjust.
23804 (expand_builtin_strncat): Adjust.
23805 (expand_builtin_strncpy): Adjust and simplify.
23806 (expand_builtin_memset): Adjust.
23807 (expand_builtin_bzero): Adjust.
23808 (expand_builtin_memcmp): Adjust.
23809 (expand_builtin): Handle memcmp.
23810 (maybe_emit_chk_warning): Check strncat just once.
23812 2017-05-04 Martin Sebor <msebor@redhat.com>
23814 PR preprocessor/79214
23815 PR middle-end/79222
23816 PR middle-end/79223
23817 * builtins.c (check_sizes): Add inlining context and issue
23818 warnings even when -Wno-system-headers is set.
23819 (check_strncat_sizes): Same.
23820 (expand_builtin_strncat): Same.
23821 (expand_builtin_memmove): New function.
23822 (expand_builtin_stpncpy): Same.
23823 (expand_builtin): Handle memmove and stpncpy.
23825 2017-05-04 Bin Cheng <bin.cheng@arm.com>
23827 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
23828 which is not used any more.
23830 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23832 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
23834 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23836 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
23837 (cortexa53_tunings): Likewise.
23838 (cortexa57_tunings): Likewise.
23839 (cortexa72_tunings): Likewise.
23840 (cortexa73_tunings): Likewise.
23842 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23844 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
23845 Set loop alignment to 8.
23847 2017-05-04 Martin Sebor <msebor@redhat.com>
23849 PR translation/80280
23850 * builtins.c (expand_builtin_object_size): Add missing quoting to
23851 %D and like directives.
23852 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
23853 (hsa_type_for_tree_type): Same.
23854 (verify_function_arguments): Same.
23855 * symtab.c (symbol_table::change_decl_assembler_name): Same.
23856 * varasm.c (get_section): Same.
23859 2017-05-04 Martin Sebor <msebor@redhat.com>
23861 PR translation/80280
23862 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
23864 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23866 * config/aarch64/aarch64.c (generic_addrcost_table):
23867 Change HI/TI mode setting.
23869 2017-05-04 Martin Jambor <mjambor@suse.cz>
23871 PR tree-optimization/80622
23872 * tree-sra.c (comes_initialized_p): New function.
23873 (build_accesses_from_assign): Only set write lazily when
23874 comes_initialized_p is false.
23875 (analyze_access_subtree): Use comes_initialized_p.
23876 (propagate_subaccesses_across_link): Assert !comes_initialized_p
23877 instead of testing for PARM_DECL.
23879 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23881 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
23882 constraint on operand 0 to allow more general addressing modes.
23883 Adjust output template.
23884 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
23886 * config/aarch64/aarch64-protos.h
23887 (aarch64_address_valid_for_prefetch_p): Declare prototype.
23888 * config/aarch64/constraints.md (Dp): New address constraint.
23889 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
23892 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
23894 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
23895 update use of estimate_ipcp_clone_size_and_time.
23896 (estimate_local_effects): Update use of
23897 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
23898 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
23899 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
23900 Return nonspecialized time.
23902 2017-05-04 Richard Biener <rguenther@suse.de>
23904 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
23905 for the last VUSE which def dominates the PHI. Directly call
23907 (get_continuation_for_phi_1): Remove.
23909 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
23911 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
23912 to explain the use of truncating division. Cap the number of
23913 iterations to the maximum given by nb_iterations_upper_bound,
23916 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23918 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
23919 * configure: Regenerate.
23920 * config.in: Regenerate.
23921 * config/i386/driver-mingw32.c: new file.
23922 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
23923 * config.host: Link driver-mingw32.o on MinGW host.
23924 * doc/install.texi: Document new --enable-mingw-wildcard configure
23927 2017-05-04 Marek Polacek <polacek@redhat.com>
23929 PR tree-optimization/80612
23930 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
23932 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23933 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
23935 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
23936 (movt splitter): Likewise.
23937 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
23938 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
23939 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
23940 block for Thumb-1 with MOVT.
23941 (thumb2_legitimate_address_p): Move code block ...
23942 (can_avoid_literal_pool_for_label_p): ... into this new function.
23943 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
23945 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
23946 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
23947 "M-profile targets with the MOVT instruction".
23949 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23951 * config/arm/arm-builtins.c (arm_init_builtins): Rename
23952 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
23953 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
23955 2017-05-04 Martin Liska <mliska@suse.cz>
23957 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
23958 variable cond_code.
23960 2017-05-04 Richard Biener <rguenther@suse.de>
23962 * tree.c (array_at_struct_end_p): Handle arrays at struct
23963 end with flexarrays more conservatively. Refactor and treat
23964 arrays of arrays or aggregates more strict. Fix
23965 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
23966 * tree.c (array_at_struct_end_p): Adjust prototype.
23967 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
23968 * gimple-fold.c (get_range_strlen): Likewise.
23969 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
23971 2017-05-04 Richard Biener <rguenther@suse.de>
23973 PR tree-optimization/31130
23974 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
23976 (supports_overflow_infinity): Likewise.
23977 (is_negative_overflow_infinity): Likewise.
23978 (is_positive_overflow_infinity): Likewise.
23979 (is_overflow_infinity): Likewise.
23980 (stmt_overflow_infinity): Likewise.
23981 (overflow_infinity_range_p): Likewise.
23982 (usable_range_p): Remove as always returning true.
23983 (make_overflow_infinity): Remove.
23984 (negative_overflow_infinity): Likewise.
23985 (positive_overflow_infinity): Likewise.
23986 (avoid_overflow_infinity): Likewise.
23987 (set_value_range): Adjust accordingly.
23988 (set_value_range_to_nonnegative): Likewise, remove now unused
23989 overflow_infinity arg.
23990 (vrp_operand_equal_p): Adjust.
23991 (update_value_range): Likewise.
23992 (range_int_cst_singleton_p): Likewise.
23993 (operand_less_p): Likewise.
23994 (compare_values_warnv): Likewise.
23995 (extract_range_for_var_from_comparison_expr): Likewise.
23996 (vrp_int_const_binop): Likewise.
23997 (zero_nonzero_bits_from_vr): Likewise.
23998 (extract_range_from_multiplicative_op_1): Likewise.
23999 (extract_range_from_binary_expr_1): Likewise.
24000 (extract_range_from_unary_expr): Likewise.
24001 (extract_range_from_comparison): Likewise.
24002 (extract_range_basic): Likewise.
24003 (adjust_range_with_scev): Likewise.
24004 (compare_ranges): Likewise.
24005 (compare_range_with_value): Likewise.
24006 (dump_value_range): Likewise.
24007 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24009 (simplify_cond_using_ranges): Adjust.
24011 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24013 * brig-builtins.def: Added a builtin for class_f64.
24014 * builtin-types.def: Added a builtin type needed by class_f64.
24016 2017-05-03 Jason Merrill <jason@redhat.com>
24018 * timevar.def: Add TV_CONSTEXPR.
24020 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24022 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24024 2017-05-03 Martin Jambor <mjambor@suse.cz>
24026 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24027 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24028 * ipa-cp.c (ipcp_propagate_stage): Do not call
24029 ipa_update_after_lto_read.
24030 * ipa-inline.c (ipa_inline): Likewise.
24032 2017-05-03 Martin Jambor <mjambor@suse.cz>
24034 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24035 tag. Added a default constructor and a destructor.
24036 (ipa_edge_args_sum_t): New class;
24037 (ipa_edge_args_sum): Declare.
24038 (ipa_edge_args_vector): Remove declaration.
24039 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24040 (ipa_free_edge_args_substructures): Remove declaration.
24041 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24042 (ipa_edge_args_info_available_for_edge_p): Likewise.
24043 * ipa-prop.c (ipa_edge_args_vector): Removed.
24044 (edge_removal_hook_holder): Likewise.
24045 (edge_duplication_hook_holder): Likewise.
24046 (ipa_edge_args_sum): New variable.
24047 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24048 ipa_edge_args_vector.
24049 (ipa_free_edge_args_substructures): Likewise.
24050 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24051 ipa_edge_args_vector.
24052 (ipa_edge_removal_hook): Turned into method
24053 ipa_edge_args_sum_t::remove.
24054 (ipa_edge_duplication_hook): Turned into method
24055 ipa_edge_args_sum_t::duplicate.
24056 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24057 registering edge hooks.
24058 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24059 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24060 ipa_edge_args_sum instead of ipa_edge_args_vector.
24061 * ipa-profile.c (ipa_profile): Likewise.
24063 2017-05-03 Martin Jambor <mjambor@suse.cz>
24065 * symbol-summary.h (function_summary): New method exists.
24066 (function_summary::symtab_removal): Deallocate through release.
24067 (call_summary): New class.
24068 (gt_ggc_mx): New overload.
24069 (gt_pch_nx): Likewise.
24070 (gt_pch_nx): Likewise.
24072 2017-05-03 Jeff Law <law@redhat.com>
24074 PR tree-optimization/78496
24075 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24076 from simplify_cond_using_ranges. Split off code to walk
24077 backwards through casts into ...
24078 (simplify_cond_using_ranges_2): New function.
24079 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24080 (execute_vrp): After identifying jump threads, call
24081 simplify_cond_using_ranges_2.
24083 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24086 * ipa-inline.h (inline_summary): Add ctor.
24087 (create_ggc): Do not use ggc_cleared_alloc.
24089 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24090 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24092 * gcc.c (handle_braces): Support escaping in switch matching
24094 * doc/invoke.texi (Spec Files): Document it.
24095 Remove superfluous @code markup in items.
24097 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24099 * diagnostic-show-locus.c (struct column_range): New struct.
24100 (get_affected_columns): New function.
24101 (get_printed_columns): New function.
24102 (struct correction): New struct.
24103 (correction::ensure_capacity): New function.
24104 (correction::ensure_terminated): New function.
24105 (struct line_corrections): New struct.
24106 (line_corrections::~line_corrections): New dtor.
24107 (line_corrections::add_hint): New function.
24108 (layout::print_trailing_fixits): Reimplement in terms of the new
24110 (selftest::test_overlapped_fixit_printing): New function.
24111 (selftest::diagnostic_show_locus_c_tests): Call it.
24113 2017-05-03 Nathan Sidwell <nathan@acm.org>
24115 Canonicalize canonical type hashing
24116 * tree.h (type_hash_canon_hash): Declare.
24117 * tree.c (type_hash_list, attribute_hash_list): Move into
24118 type_hash_canon_hash.
24119 (build_type_attribute_qual_variant): Break out hash code calc into
24120 type_hash_canon_hash.
24121 (type_hash_canon_hash): New. Generic type hash computation.
24122 (build_range_type_1, build_array_type_1, build_function_type,
24123 build_method_type_directly, build_offset_type, build_complex_type,
24124 make_vector_type): Call it.
24126 2017-05-03 Richard Biener <rguenther@suse.de>
24128 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24129 When all DRs have unknown misaligned do not always peel
24130 when there is a store but apply the same costing model as if
24131 there were only loads.
24133 2017-05-03 Richard Biener <rguenther@suse.de>
24136 PR tree-optimization/80492
24137 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24138 compare_base_decls returning dont-know properly.
24140 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24142 * config/arm/iterators.md (CCSI): New mode iterator.
24143 (arch): New mode attribute.
24144 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24145 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24146 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24147 code iterator for success result mode.
24148 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24149 the corresponding new insn generators.
24151 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24154 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24155 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24157 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24159 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24160 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24161 (DDR_A): Wrap DDR argument in brackets.
24162 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24163 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24164 (DDR_REVERSED_P): Likewise.
24166 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24168 PR tree-optimization/79472
24169 * tree-switch-conversion.c (struct switch_conv_info): Add
24170 contiguous_range and default_case_nonstandard fields.
24171 (collect_switch_conv_info): Compute contiguous_range and
24172 default_case_nonstandard fields, don't clear final_bb if
24173 contiguous_range and only the default case doesn't have the required
24175 (check_all_empty_except_final): Set default_case_nonstandard instead
24176 of failing if contiguous_range and the default case doesn't have empty
24178 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24179 and only the default case doesn't have the required constants. Skip
24181 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24182 if default_case_nonstandard.
24183 (build_constructors): Build constant 1 just once. Assert that default
24184 values aren't inserted in between cases if contiguous_range. Skip
24186 (build_arrays): Skip virtual phis.
24187 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24188 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24189 Handle virtual phis.
24190 (gen_inbound_check): Handle default_case_nonstandard case.
24191 (process_switch): Adjust check_final_bb caller. Call
24192 gather_default_values with the first non-default case instead of
24193 default case if default_case_nonstandard.
24195 2017-05-02 Nathan Sidwell <nathan@acm.org>
24197 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24198 check. Fix formatting.
24200 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24202 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24203 errors when comparing specialized and unspecialized times.
24205 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24207 * diagnostic-show-locus.c
24208 (layout::should_print_annotation_line_p): Make private.
24209 (layout::print_annotation_line): Make private.
24210 (layout::annotation_line_showed_range_p): Make private.
24211 (layout::show_ruler): Make private.
24212 (layout::print_source_line): Make private. Pass in line and
24213 line_width, rather than calling location_get_source_line. Drop
24215 (layout::print_leading_fixits): New method.
24216 (layout::print_any_fixits): Rename to...
24217 (layout::print_trailing_fixits): ...this, and make private.
24218 Don't print newline fixits.
24219 (diagnostic_show_locus): Move logic for printing one row into...
24220 (layout::print_line): ...this new function. Move the
24221 location_get_source_line call and error-handling from
24222 print_source_line to here. Call print_leading_fixits, and rename
24223 print_any_fixits to print_trailing_fixits.
24224 (selftest::test_fixit_insert_containing_newline): Update now that
24225 newlines are partially supported.
24226 (selftest::test_fixit_insert_containing_newline_2): New test.
24227 (selftest::test_fixit_replace_containing_newline): Update comments.
24228 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24229 * edit-context.c (class added_line): New class.
24230 (class edited_line): Describe newline handling in comment.
24231 (edited_line::actually_edited_p): New method.
24232 (edited_line::print_content): Delete redundant decl.
24233 (edited_line::m_predecessors): New field.
24234 (edited_file::print_content): Call edited_line::print_content.
24235 (edited_file::print_diff): Update to support newlines.
24236 (edited_file::print_diff_hunk): Likewise.
24237 (edited_file::print_run_of_changed_lines): New function.
24238 (edited_file::print_diff_line): Convert to...
24239 (print_diff_line): ...this.
24240 (edited_file::get_effective_line_count): New function.
24241 (edited_line::edited_line): Initialize new field m_predecessors.
24242 (edited_line::~edited_line): Clean up m_predecessors.
24243 (edited_line::apply_fixit): Handle newlines.
24244 (edited_line::get_effective_line_count): New function.
24245 (edited_line::print_content): New function.
24246 (edited_line::print_diff_lines): New function.
24247 (selftest::test_applying_fixits_insert_containing_newline): New
24249 (selftest::test_applying_fixits_replace_containing_newline): New
24251 (selftest::insert_line): New function.
24252 (selftest::test_applying_fixits_multiple_lines): Add example of
24254 (selftest::edit_context_c_tests): Call the new tests.
24256 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24258 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
24259 parameter cand. Update dump information.
24260 (get_computation_cost): Update uses.
24262 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24264 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
24265 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
24266 (get_computation_at, rewrite_use_address): Update use of
24267 get_computation_aff.
24269 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24271 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
24272 (get_computation): Delete.
24273 (get_computation_cost): Implement like get_computation_cost_at.
24274 Use get_computation_at.
24275 (get_computation_cost_at): Delete.
24276 (rewrite_use_nonlinear_expr): Use get_computation_at.
24277 (rewrite_use_compare, remove_unused_ivs): Ditto.
24279 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24281 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
24283 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24285 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
24286 (ivopts_global_cost_for_size): Rename parameter and update uses.
24287 (iv_ca_recount_cost): Update uses.
24288 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
24289 candidates seperately in n_invs and n_cands.
24290 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
24292 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24294 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
24295 (find_inv_vars_cb): New.
24296 (find_depends): Renamed to ...
24297 (find_inv_vars): ... this.
24298 (add_candidate_1, force_var_cost): Call find_inv_vars.
24299 (split_address_cost, determine_group_iv_cost_cond): Ditto.
24301 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24303 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
24304 inv_vars. Add inv_exprs.
24305 (struct iv_cand): Rename depends_on to inv_vars.
24306 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
24307 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
24308 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
24309 (dump_cand): Dump inv_vars.
24310 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
24311 (record_invariant, find_depends, add_candidate_1): Ditto.
24312 (set_group_iv_cost, force_var_cost): Ditto.
24313 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
24314 (get_computation_cost_at, get_computation_cost): Ditto.
24315 (determine_group_iv_cost_generic): Ditto.
24316 (determine_group_iv_cost_address): Ditto.
24317 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
24318 (determine_group_iv_costs): Ditto.
24319 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
24320 (iv_ca_set_remove_invariants): Renamed to ...
24321 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
24322 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
24323 (iv_ca_set_add_invariants): Renamed to ...
24324 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
24325 (iv_ca_set_cp): Use iv_ca_set_add_invs.
24326 (iv_ca_has_deps): Support inv_vars and inv_exprs.
24327 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
24328 (create_new_ivs): Remove useless dump.
24330 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24332 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
24334 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
24335 (iv_ca_set_no_cp, create_new_iv): Ditto.
24337 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24339 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24341 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24343 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
24344 function tree_check2.
24346 2017-05-02 Martin Liska <mliska@suse.cz>
24348 * doc/gcov.texi: Add missing preposition.
24349 * gcov.c (function_info::function_info): Properly fill up
24350 all member variables.
24352 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24354 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
24356 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24358 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
24360 2017-05-02 Martin Liska <mliska@suse.cz>
24363 * lto-streamer-in.c (lto_read_tree_1): Remove
24364 LTO_STREAMER_DEBUG.
24365 * lto-streamer.c (struct tree_hash_entry): Likewise.
24366 (struct tree_entry_hasher): Likewise.
24367 (tree_entry_hasher::hash): Likewise.
24368 (tree_entry_hasher::equal): Likewise.
24369 (lto_streamer_init): Likewise.
24370 (lto_orig_address_map): Likewise.
24371 (lto_orig_address_get): Likewise.
24372 (lto_orig_address_remove): Likewise.
24373 * lto-streamer.h: Likewise.
24374 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
24375 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
24377 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
24379 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
24380 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
24381 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
24382 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
24383 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
24384 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
24385 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
24386 (mm_maskz_sub_ss): New intrinsics.
24387 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24388 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24389 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
24390 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
24391 (__builtin_ia32_subss_mask_round): New builtins.
24392 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24393 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24394 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
24396 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
24397 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24399 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24402 2017-05-02 Martin Jambor <mjambor@suse.cz>
24404 PR tree-optimization/78687
24405 * tree-sra.c (access): New field parent.
24406 (process_subtree_disqualification): New function.
24407 (disqualify_candidate): Call it.
24408 (build_accesses_from_assign): Reset write flag if creating an
24410 (build_access_subtree): Fill in parent field and also prpagate
24411 down grp_write flag.
24412 (create_artificial_child_access): New parameter set_grp_write, set
24413 grp_write to its value.
24414 (propagate_subaccesses_across_link): Also propagate grp_write flag
24416 (propagate_all_subaccesses): Push the closest parent back to work
24417 queue if add_access_to_work_queue returned true.
24419 2017-05-02 Richard Biener <rguenther@suse.de>
24421 * common.opt (fstrict-overflow): Alias negative to fwrapv.
24422 * doc/invoke.texi (fstrict-overflow): Remove all traces of
24423 -fstrict-overflow documentation.
24424 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
24425 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
24426 flag_strict_overflow.
24427 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
24428 * lto-opts.c (lto_write_options): Do not stream it.
24429 * lto-wrapper.c (merge_and_complain): Do not handle it.
24430 * opts.c (default_options_table): Do not set -fstrict-overflow.
24431 (finish_options): Likewise do not clear it when sanitizing.
24432 * simplify-rtx.c (simplify_const_relational_operation): Do not
24433 test flag_strict_overflow.
24435 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24437 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
24438 using enabled attribute.
24439 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
24440 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
24441 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
24442 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
24443 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
24444 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
24445 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
24446 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
24447 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
24448 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
24450 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24452 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
24454 2017-05-02 Richard Biener <rguenther@suse.de>
24456 PR tree-optimization/80591
24458 2017-04-10 Richard Biener <rguenther@suse.de>
24460 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24463 2017-05-02 Richard Biener <rguenther@suse.de>
24465 PR tree-optimization/80549
24466 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
24467 (cleanup_tree_cfg_noloop): Create forwarders to known loop
24468 headers if they do not have a preheader.
24470 2017-05-02 Martin Liska <mliska@suse.cz>
24473 * common.opt: Fix typo.
24474 * doc/invoke.texi: Likewise.
24476 2017-05-01 Jan Beulich <jbeulich@suse.com>
24478 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
24479 swapping, add (x,x,m,x,n) alternative.
24481 2017-05-01 Nathan Sidwell <nathan@acm.org>
24483 * calls.c (combine_pending_stack_adjustment_and_call): Remove
24484 unnecessary unadjusted_alignment check.
24486 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24489 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
24491 * gimplify.c (gimplify_cilk_detach): New function.
24492 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
24493 * tree-core.h: Document EXPR_CILK_SPAWN.
24494 * tree.h (EXPR_CILK_SPAWN): Define.
24496 2017-05-01 David Malcolm <dmalcolm@redhat.com>
24498 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
24499 to use new fixit_hint representation, using the "replace" logic.
24500 (get_line_span_for_fixit_hint): Likewise.
24501 (layout::print_any_fixits): Likewise.
24502 (selftest::test_one_liner_many_fixits): Rename to...
24503 (selftest::test_one_liner_many_fixits_1): ...this, and update
24504 comment and expected output to reflect that the multiple fix-it
24505 hints are now consolidated into one insertion.
24506 (selftest::test_one_liner_many_fixits_2): New test.
24507 (selftest::test_diagnostic_show_locus_one_liner): Update for
24509 (selftest::test_fixit_consolidation): Update for fix-it API
24511 * diagnostic.c (print_parseable_fixits): Likewise.
24512 * edit-context.c (edited_line::m_line_events): Convert from
24513 auto_vec <line_event *> to auto_vec <line_event>.
24514 (class line_event): Convert from abstract base class to a concrete
24515 class, taking over the role of replace_event.
24516 (class insert_event): Delete.
24517 (class replace_event): Rename to class line_event. Convert to
24519 (edit_context::add_fixits): Reimplement.
24520 (edit_context::apply_insert): Delete.
24521 (edit_context::apply_replace): Rename to...
24522 (edit_context::apply_fixit): ...this. Convert to half-open range.
24523 (edited_file::apply_insert): Delete.
24524 (edited_file::apply_replace): Rename to...
24525 (edited_file::apply_fixit): ...this.
24526 (edited_line::~edited_line): Drop deletion of events.
24527 (edited_line::apply_insert): Delete.
24528 (edited_line::apply_replace): Rename to...
24529 (edited_line::apply_fixit): ...this. Convert to half-open range.
24530 Update for change to type of m_line_events.
24531 * edit-context.h (edit_context::apply_insert): Delete.
24532 (edit_context::apply_replace): Rename to...
24533 (edit_context::apply_fixit): ...this.
24535 2017-05-01 Martin Sebor <msebor@redhat.com>
24537 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
24540 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
24543 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
24544 __get_cpuid_max returns 0.
24545 (__get_cpuid_count): Ditto.
24547 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
24549 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
24550 replacement expression is another instance of one of its arguments.
24552 2017-05-01 Jakub Jelinek <jakub@redhat.com>
24555 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
24556 check for stack push/pop autoinc.
24557 * config/i386/i386.c (ix86_agi_dependent): Return false
24558 if the only reason why modified_in_p returned true is that
24559 addr is SP based and set_insn is a push or pop.
24561 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24563 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
24566 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24569 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
24570 (account_size_time): Use two predicates - exec_pred and
24572 (evaluate_conditions_for_known_args): Compute both clause and
24574 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
24575 (inline_summary_t::duplicate): Update.
24576 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
24578 (compute_inline_parameters): Likewise.
24579 (estimate_edge_size_and_time): Update caluclation of time.
24580 (estimate_node_size_and_time): Compute both time and nonspecialized
24582 (estimate_ipcp_clone_size_and_time): Update.
24583 (inline_merge_summary): Update.
24584 (do_estimate_edge_time): Update.
24585 (do_estimate_edge_size): Update.
24586 (do_estimate_edge_hints): Update.
24587 (inline_read_section, inline_write_summary): Stream both new predicates.
24588 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
24590 (compute_inlined_call_time): Cleanup.
24591 (big_speedup_p): Update.
24592 (edge_badness): Update.
24593 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
24594 (size_time_entry): Replace predicate by exec_predicate and
24595 nonconst_predicate.
24596 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
24597 (estimate_edge_time): Return also nonspec_time.
24598 (reset_edge_growth_cache): Update.
24600 2017-04-29 Jakub Jelinek <jakub@redhat.com>
24602 PR rtl-optimization/80491
24603 * ifcvt.c (noce_process_if_block): When looking for x setter
24604 with missing else_bb, don't check only the insn right before
24605 cond_earliest, but look for the last insn that x is modified in
24606 within the same bb.
24608 PR rtl-optimization/80491
24609 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
24611 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
24613 PR tree-optimization/80487
24614 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
24616 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24618 PR tree-optimization/79697
24619 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
24620 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
24621 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
24623 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
24624 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
24626 2017-04-28 Martin Sebor <msebor@redhat.com>
24628 PR tree-optimization/80523
24629 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
24630 (init_target_to_host_charmap, target_to_host, target_strtol10): New
24632 (maybe_warn, format_directive, parse_directive): Use new functions.
24633 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
24635 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
24637 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
24639 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
24641 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
24642 target_header_dir): Set correctly.
24643 * configure: Regenerated.
24644 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24645 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24646 instead of SYSTEM_HEADER_DIR.
24648 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
24650 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
24651 (estimate_local_effects): Likewise.
24652 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
24653 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
24654 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
24655 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
24656 do_estimate_edge_time, estimate_edge_time): Likewise.
24657 * ipa-inline-analysis.c (estimate_node_size_and_time,
24658 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
24659 (estimate_time_after_inlining): Remove.
24661 2017-04-28 Martin Liska <mliska@suse.cz>
24663 * doc/gcov.texi: Enhance documentation of gcov.
24665 2017-04-28 Martin Liska <mliska@suse.cz>
24667 * doc/gcov.texi: Sort options in alphabetic order.
24668 * doc/gcov-dump.texi: Likewise.
24669 * doc/gcov-tool.texi: Likewise.
24670 * gcov.c (print_usage): Likewise.
24671 * gcov-dump.c (print_usage): Likewise.
24672 * gcov-tool.c (print_merge_usage_message): Likewise.
24673 (print_rewrite_usage_message): Likewise.
24674 (print_overlap_usage_message): Likewise.
24676 2017-04-28 Martin Liska <mliska@suse.cz>
24678 PR gcov-profile/53915
24679 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
24681 2017-04-28 Martin Liska <mliska@suse.cz>
24683 PR gcov-profile/79891
24684 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
24685 is marked by compiler as living on a line.
24686 (get_cycles_count): Remove usage of the union.
24687 (output_intermediate_file): Likewise.
24688 (find_source): Fix GNU coding style.
24689 (accumulate_line_counts): Remove old non-all block mode.
24690 (output_lines): Remove usage of the union.
24691 * profile.c (output_location): Include all BBs, even if
24692 belonging to a same line (and file) as a previous BB.
24694 2017-04-28 Martin Liska <mliska@suse.cz>
24696 * gcov.c (process_args): Handle new argument 'w'.
24697 (read_graph_file): Assign ID to BBs.
24698 (output_branch_count): Display BB # if verbose flag is set.
24699 (output_lines): Likewise for arcs.
24700 (print_usage): Add '--verbose' option help.
24701 * doc/gcov.texi: Document --verbose (-w) option.
24703 2017-04-28 Martin Liska <mliska@suse.cz>
24705 * gcov.c (struct block_location_info): New struct.
24706 (process_file): Fill up the new structure.
24707 (read_graph_file): Replace usage of encoding by the newly added
24709 (add_line_counts): Likewise.
24710 (accumulate_line_counts): Remove usage of the union.
24711 (function_info::function_info): New function.
24712 (function_info::~function_info): Likewise.
24713 (process_file): Call delete instead of release_function.
24714 (release_function): Release the function.
24715 (release_structures): Call delete instead of release_function.
24716 (solve_flow_graph): Replace usage of num_blocks.
24717 (find_exception_blocks): Likewise.
24718 (output_lines): Fix GNU coding style.
24720 2017-04-28 Martin Liska <mliska@suse.cz>
24723 * coverage.c (coverage_remove_note_file): New function.
24724 * coverage.h: Declare the function.
24725 * toplev.c (finalize): Clean if an error has been seen.
24727 2017-04-28 Martin Liska <mliska@suse.cz>
24729 PR gcov-profile/80031
24730 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
24731 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
24732 * gcov.c (read_graph_file): Read just number of blocks.
24733 * profile.c (branch_prob): Do not stream 0 flags per a basic
24736 2017-04-28 Martin Liska <mliska@suse.cz>
24738 * gcov-dump.c (tag_*): Add new argument to declarations.
24739 (dump_gcov_file): Likewise.
24740 (tag_blocks): Add and use new argument depth.
24741 (tag_arcs): Likewise.
24742 (tag_lines): Likewise.
24743 (tag_counters): Likewise.
24744 (tag_summary): Likewise.
24745 (dump_working_sets): Use depth to do a proper indentation.
24747 2017-04-28 Jakub Jelinek <jakub@redhat.com>
24750 * cgraph.h (symtab_node::debug_symtab): No longer inline.
24751 * symtab.c (symtab_node::debug_symtab): Move definition here.
24753 2017-04-28 Richard Biener <rguenther@suse.de>
24755 * lto-streamer.h (LTO_major_version): Bump to 7.
24757 2017-04-28 Richard Biener <rguenther@suse.de>
24759 * tree-vrp.c (assert_info): New struct.
24760 (add_assert_info): New helper.
24761 (register_edge_assert_for_2): Refactor to add asserts to a vector
24763 (register_edge_assert_for_1): Likewise.
24764 (register_edge_assert_for): Likewise.
24765 (finish_register_edge_assert_for): New helper actually registering
24766 asserts where live on edge.
24767 (find_conditional_asserts): Adjust.
24768 (find_switch_asserts): Likewise.
24769 (evrp_dom_walker::try_find_new_range): Generalize.
24770 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
24772 2017-04-27 Marek Polacek <polacek@redhat.com>
24775 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
24776 arg10 and arg11 to itype.
24778 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
24780 * doc/extend.texi (Object Size Checking): Improve grammar.
24782 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
24785 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
24786 that the logic for permitting reciprocal estimates matches that
24789 2017-04-27 Jakub Jelinek <jakub@redhat.com>
24792 * tree.c (type_cache_hasher::equal): Only compare
24793 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
24794 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
24795 non-aggregate element types.
24796 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
24797 about the flag on ARRAY_TYPEs in the comment, formatting fix.
24799 2017-04-27 Richard Biener <rguenther@suse.de>
24801 PR middle-end/80533
24802 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
24803 stripping ARRAY_REFs from MEM_EXPR make sure we're not
24804 keeping a reference to a trailing array.
24806 2017-04-27 Richard Biener <rguenther@suse.de>
24808 PR middle-end/80539
24809 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
24810 being in loop-closed SSA form conservatively.
24811 (chrec_fold_multiply_poly_poly): Likewise.
24813 2017-04-27 Tamar Christina <tamar.christina@arm.com>
24815 PR middle-end/79665
24816 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
24817 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
24819 2017-04-27 Jakub Jelinek <jakub@redhat.com>
24822 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
24823 (aarch64_function_arg_alignment): Return unsigned int again, but still
24824 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
24825 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24826 Don't emit -Wpsabi note.
24827 (aarch64_function_arg_boundary): Likewise.
24828 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24831 2017-04-26 Nathan Sidwell <nathan@acm.org>
24833 * tree.h (crc32_unsigned_n): Declare.
24834 (crc32_unsigned, crc32_unsigned): Make inline.
24835 * tree.c (crc32_unsigned_bits): Replace with ...
24836 (crc32_unsigned_n): ... this.
24837 (crc32_unsigned, crc32_byte): Remove.
24838 (crc32_string): Remove unnecessary braces.
24840 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
24842 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
24843 * ipa-inline-analysis.c (MAX_TIME): Remove.
24844 (account_size_time): Use sreal for time.
24845 (dump_inline_summary): Update.
24846 (estimate_function_body_sizes): Update.
24847 (estimate_edge_size_and_time): Update.
24848 (estimate_calls_size_and_time): Update.
24849 (estimate_node_size_and_time): Update.
24850 (inline_merge_summary): Update.
24851 (inline_update_overall_summary): Update.
24852 (estimate_time_after_inlining): Update.
24853 (inline_read_section): Update.
24854 (inline_write_summary): Update.
24855 * ipa-inline.c (compute_uninlined_call_time): Update.
24856 (compute_inlined_call_time): Update.
24857 (recursive_inlining): Update.
24858 (inline_small_functions): Update.
24859 (dump_overall_stats): Update.
24860 * ipa-inline.h: Include sreal.h.
24861 (size_time_entry): Turn time to sreal.
24862 (inline_summary): Turn self_time nad time to sreal.
24864 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
24866 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
24868 (sreal::stream_out, sreal::stream_in): New.
24869 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
24871 2017-04-25 Jakub Jelinek <jakub@redhat.com>
24873 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
24876 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
24879 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
24881 (dimode_scalar_chain::compute_convert_gain): Ditto.
24882 (dimode_scalar_chain::make_vector_copies): Ditto.
24883 (dimode_scalar_chain::convert_reg): Ditto.
24884 (dimode_scalar_chain::convert_insn): Ditto.
24885 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
24886 (VI248_AVX512BW_1): New mode iterator.
24887 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
24888 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
24891 2017-04-25 Martin Sebor <msebor@redhat.com>
24893 PR tree-optimization/80497
24894 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
24895 constants are representable in HOST_WIDE_INT.
24896 (parse_directive): Ditto.
24898 2017-04-25 Martin Sebor <msebor@redhat.com>
24901 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
24902 (new_zero_array): Adjust signature.
24903 (dom_info::dom_init): Used unsigned rather that size_t.
24904 (dom_info::dom_info): Same.
24906 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24907 Jakub Jelinek <jakub@redhat.com>
24910 * config/arm/arm.c: Include gimple.h.
24911 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
24912 returns negative, increment ncrn only if it returned positive.
24913 (arm_needs_doubleword_align): Return int instead of bool,
24914 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
24915 members, but if there is any such non-FIELD_DECL
24916 > PARM_BOUNDARY aligned decl, return -1 instead of false.
24917 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
24918 returns negative, increment nregs only if it returned positive.
24919 (arm_setup_incoming_varargs): Likewise.
24920 (arm_function_arg_boundary): Emit -Wpsabi note if
24921 arm_needs_doubleword_align returns negative, return
24922 DOUBLEWORD_ALIGNMENT only if it returned positive.
24924 2017-04-25 Marek Polacek <polacek@redhat.com>
24927 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
24928 first argument to type.
24930 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
24933 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
24934 type checks to test for compatibility instead of equality.
24936 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24937 Jakub Jelinek <jakub@redhat.com>
24940 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
24942 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
24943 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
24944 the alignment computation, but return their maximum in warn_alignment.
24945 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24946 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
24948 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
24949 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24952 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24954 * config/arc/simdext.md (dmpyh): Fix typo.
24956 2017-04-25 Richard Biener <rguenther@suse.de>
24958 PR tree-optimization/80492
24959 * alias.c (compare_base_decls): Handle registers with asm
24960 specification conservatively.
24961 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24962 compare_base_decls returning dont-know properly.
24964 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24966 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
24967 (legitimate_offset_address_p): New function.
24968 (arc_legitimate_address_p): Use above function.
24970 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24972 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
24974 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24976 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
24977 ACCH registers whenever they are available.
24979 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24981 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
24982 double regs fix when not used.
24984 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24986 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
24988 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
24989 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
24991 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24993 * config/arc/arc.c (arc_output_addsi): Check for h-register class
24994 when emitting short ADD instructions.
24996 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24998 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25000 (cmpsi_cc_c_insn): Likewise.
25001 (cbranchsi4_scratch): Compute proper instruction length using
25002 compact_hreg_operand.
25003 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25005 2017-04-25 Richard Biener <rguenther@suse.de>
25007 PR middle-end/80509
25008 * passes.c (pass_manager::pass_manager): Initialize
25009 m_name_to_pass_map.
25011 2017-04-25 Richard Biener <rguenther@suse.de>
25013 PR tree-optimization/79201
25014 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25016 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25019 * config/s390/vector.md: Split MEM->GPR vector moves for
25020 non-s_operand addresses.
25022 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25025 * config/s390/predicates.md (reload_const_wide_int_operand): New
25027 * config/s390/s390.md ("movti"): Remove d/P alternative.
25028 ("movti_bigconst"): New pattern definition.
25030 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25033 * s390-protos.h (s390_expand_cs_hqi): Removed.
25034 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25035 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25036 modes as well as CCZ1mode and CCZmode.
25037 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25038 signature of s390_emit_compare_and_swap.
25039 (s390_expand_cs_hqi): Likewise, make static.
25040 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25041 compare-and-swap, in some cases.
25042 (s390_expand_cs): Wrapper function.
25043 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25045 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25046 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25047 patterns for small and large integers. Forbid symref memory operands.
25048 Move expander to s390.c. Require cc register.
25049 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25050 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25051 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25052 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25053 symref memory operands. Remove CC mode and call s390_match_ccmode
25055 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25057 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25059 * config/s390/s390.md (define_peephole2): New peephole to help
25060 combining the load-and-test pattern with volatile memory.
25062 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25064 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25065 with CCZmode for TARGET_Z196.
25067 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25069 PR rtl-optimization/80501
25070 * combine.c (make_compound_operation_int): Set subreg_code to SET
25071 even for AND with mask of the sign bit of mode.
25073 PR rtl-optimization/80500
25074 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25075 sum's initial value.
25077 2017-04-25 Julian Brown <julian@codesourcery.com>
25078 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25080 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25082 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25084 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25086 2017-04-25 Julian Brown <julian@codesourcery.com>
25087 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25089 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25090 (thunderx2t99_sha): New Reservation.
25092 2017-04-25 Julian Brown <julian@codesourcery.com>
25093 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25095 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25096 type for 1-element load.
25098 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25100 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25102 2017-04-24 Martin Jambor <mjambor@suse.cz>
25104 PR tree-optimization/80293
25105 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25106 char arrays not totally scalarizable if it is false.
25107 (analyze_all_variable_accesses): Pass correct value in the new
25108 parameter. Add a statistics counter.
25110 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25112 PR middle-end/79931
25113 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25115 2017-04-24 Richard Biener <rguenther@suse.de>
25117 PR tree-optimization/80494
25118 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25119 out for complex types.
25121 2017-04-24 Richard Biener <rguenther@suse.de>
25123 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25124 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25125 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25126 (DFS): Adjust and never fail.
25127 (sccvn_dom_walker::fail): Remove.
25128 (sccvn_dom_walker::before_dom_children): Adjust.
25129 (run_scc_vn): Likewise and never fail.
25130 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25131 (pass_fre::execute): Likewise.
25133 2017-04-24 Richard Biener <rguenther@suse.de>
25135 PR tree-optimization/79725
25136 * tree-ssa-sink.c (statement_sink_location): Return whether
25137 failure reason was zero uses. Move that check later.
25138 (sink_code_in_bb): Deal with zero uses by removing the stmt
25141 2017-04-24 Richard Biener <rguenther@suse.de>
25144 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25145 pointer-based references.
25147 2017-04-24 Richard Biener <rguenther@suse.de>
25150 * pass_manager.h (pass_manager::operator new): Remove.
25151 (pass_manager::operator delete): Likewise.
25152 * passes.c (pass_manager::operator new): Remove.
25153 (pass_manager::operator delete): Likewise.
25154 (pass_manager::pass_manager): Zero individual pass members.
25156 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25159 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25160 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25161 Check "XEXP (src, 1)" operand here.
25162 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25163 Check "XEXP (src, 1)" operand here.
25164 (dimode_scalar_chain::make_vector_copies): Detect count register
25165 of a shift instruction. Zero extend count register from QImode
25166 to DImode to satisfy vector shift pattern count operand predicate.
25167 Substitute vector shift count operand with a DImode copy.
25168 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25171 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25173 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25174 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25175 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25176 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25177 (UNSPEC_NOREX_MEM): Remove definition.
25179 2017-04-21 Richard Biener <rguenther@suse.de>
25181 PR tree-optimization/79547
25182 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25183 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25184 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25185 without any constraints.
25187 2017-04-21 Richard Biener <rguenther@suse.de>
25189 PR tree-optimization/78847
25190 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25192 2017-04-21 Richard Biener <rguenther@suse.de>
25194 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25195 (build_distinct_type_copy): Likewise.
25196 (build_variant_type_copy): Likewise.
25197 * tree.c (build_qualified_type): Pass down mem-stat info.
25198 (build_distinct_type_copy): Likewise.
25199 (build_variant_type_copy): Likewise.
25201 2017-04-21 Richard Biener <rguenther@suse.de>
25203 PR tree-optimization/80237
25204 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25206 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25207 for a simplified result.
25209 2016-04-21 Richard Biener <rguenther@suse.de>
25211 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25212 sth as strict as a simple_iv but a chrec without symbols and an
25213 operand defined in the loop we are peeling (and not some subloop).
25214 (propagate_constants_for_unrolling): Propagate all constants.
25216 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25219 * config/i386/i386.c (print_reg): Remove assert for disalowed
25220 regno values, call output_operand_lossage instead.
25222 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25225 * config/i386/constraints.md (Yc): New register constraint.
25226 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25227 Use Yc constraint for alternative 2 of operand 0. Remove
25228 preferred_for_speed attribute.
25230 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25232 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25233 lastprivate clauses in SIMT case.
25235 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25237 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25239 2017-04-20 Richard Biener <rguenther@suse.de>
25241 PR tree-optimization/57796
25242 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25243 as N scalar stores.
25244 (vect_model_load_cost): Cost gathers as N scalar loads.
25246 2017-04-20 Richard Biener <rguenther@suse.de>
25248 * ggc-page.c (ggc_allocated_p): Rename to ...
25249 (safe_lookup_page_table_entry): ... this and return the lookup
25251 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
25253 2017-04-20 Richard Biener <rguenther@suse.de>
25255 PR tree-optimization/80453
25256 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
25257 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
25258 from the conditions.
25259 (vn_phi_eq): Pass them down.
25260 (vn_phi_lookup): Record them.
25261 (vn_phi_insert): Likewise.
25263 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
25265 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
25266 uninitialized variable warning to avoid buffer overrun.
25268 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25271 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
25272 is suppressed for '{ 0 }' in C.
25274 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25276 * BASE-VER: Set to 8.0.0.
25278 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
25280 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
25281 priority .init_array and .fini_array section with SECTION_NOTYPE
25284 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25286 PR middle-end/80423
25287 * tree.h (build_array_type): Add typeless_storage default argument.
25288 * tree.c (type_cache_hasher::equal): Also compare
25289 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
25290 (build_array_type): Add typeless_storage argument, set
25291 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
25293 (build_nonshared_array_type): Adjust build_array_type_1 caller.
25294 (build_array_type): Likewise. Add typeless_storage argument.
25296 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25297 Jakub Jelinek <jakub@redhat.com>
25299 PR tree-optimization/80426
25300 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
25301 operation on symbolic operands, also compute the overflow for the
25302 invariant part when the operation degenerates into a negation.
25304 2017-04-19 Jakub Jelinek <jakub@redhat.com>
25307 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
25308 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
25311 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
25313 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
25316 * config/avr/avr.c (tree.h): Include it.
25317 (cgraph.h): Include it.
25318 (avr_encode_section_info): Don't warn for uninitialized progmem
25319 variable if it's just an alias.
25321 2017-04-19 Richard Biener <rguenther@suse.de>
25324 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
25325 when needed by AutoPGO.
25327 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
25330 * doc/lto.texi: Remove an extra 'that'.
25332 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
25334 PR rtl-optimization/80429
25335 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
25336 are only used in debug insns.
25338 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25339 Vladimir Makarov <vmakarov@redhat.com>
25341 * config/sparc/predicates.md (input_operand): Add comment. Return
25342 true for any memory operand when LRA is in progress.
25343 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
25345 2017-04-18 Jeff Law <law@redhat.com>
25348 * mips.md ({return,simple_return}_internal): Do not overwrite
25351 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25353 PR tree-optimization/80443
25354 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
25355 instead of adding 1, subtract -1 and similarly instead of subtracting
25358 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
25360 PR rtl-optimization/80357
25361 * haifa-sched.c (tmp_bitmap): New variable.
25362 (model_recompute): Handle duplicate use records.
25363 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
25364 (free_global_sched_pressure_data): Free it.
25366 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
25369 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25370 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25371 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25372 instead of SYSTEM_HEADER_DIR.
25374 2017-04-18 Jeff Law <law@redhat.com>
25376 PR middle-end/80422
25377 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
25378 predecessors after walking up the insn chain.
25380 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25383 * dwarf2out.c (modified_type_die): Try harder not to emit internal
25384 sizetype type into debug info.
25386 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25389 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
25390 unneeded test for TARGET_UPPER_REGS_SF.
25391 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
25393 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25396 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
25397 instead of gsi_after_labels.
25399 2017-04-18 Jeff Law <law@redhat.com>
25401 * regcprop.c (maybe_mode_change): Avoid creating copies of the
25405 2017-04-13 Jeff Law <law@redhat.com>
25406 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25407 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25409 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
25412 * config/avr/avr.c (intl.h): Include it.
25413 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
25415 2017-04-18 Martin Liska <mliska@suse.cz>
25417 PR gcov-profile/78783
25418 * gcov-tool.c (gcov_output_files): Validate that destination
25419 file is either removed by the tool or by a user.
25421 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25422 Guy Benyei <guybe@mellanox.com>
25424 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
25425 block, and do not negate it, the stored id is already negative.
25427 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25429 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
25431 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25434 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
25435 masks of options that should be turned off if the VSX vector
25436 options are turned off.
25437 (OTHER_P8_VECTOR_MASKS): Likewise.
25438 (OTHER_VSX_VECTOR_MASKS): Likewise.
25439 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
25440 rs6000_disable_incompatible_switches to validate no type switches
25442 (rs6000_incompatible_switch): New function to disallow turning on
25443 other vector options if -mno-vsx, -mno-power8-vector, or
25444 -mno-power9-vector are specified.
25446 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25448 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
25450 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25452 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
25453 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
25454 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
25455 (ARG_POINTER_CFA_OFFSET): Likewise.
25457 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25459 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
25460 conditions to take advantage of various optimizations.
25462 2017-04-13 Jeff Law <law@redhat.com>
25464 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25465 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25466 (zero_extendsidi2_dext): Likewise.
25468 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25471 * fold-const.c (fold_ternary_loc): Revert
25472 use op0 instead of fold_convert_loc (loc, type, arg0) part of
25475 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
25477 PR rtl-optimization/80343
25478 * lra-remat.c (update_scratch_ops): Assign original hard reg to
25479 new scratch pseudo.
25481 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
25484 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
25485 to ubsan_encode_value.
25487 2017-04-13 Jeff Law <law@redhat.com>
25489 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
25490 appearing in DEBUG_INSNs.
25492 2017-04-13 Martin Liska <mliska@suse.cz>
25494 PR gcov-profile/80413
25495 * gcov-io.c (gcov_write_string): Copy to buffer just when
25496 allocated size is greater than zero.
25498 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25501 * dwarf2out.c (decls_for_scope): Ignore declarations of
25502 current_function_decl in BLOCK_NONLOCALIZED_VARS.
25504 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
25507 * ipa-visibility.c (non_local_p): Fix typos.
25508 (localize_node): When localizing symbol in same comdat group,
25509 dissolve the group only when we know external symbols are going
25511 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
25513 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25515 PR tree-optimization/79390
25516 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
25517 order does not result in usable sequence, retry with reversed operand
25523 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
25524 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
25525 op0 instead of fold_convert_loc (loc, type, arg0).
25527 2017-04-12 Jeff Law <law@redhat.com>
25529 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
25530 has a delay slot in the generated code.
25532 * config/cris/cris.md (cris_preferred_reload_class): Return
25533 GENNONACR_REGS rather than GENERAL_REGS.
25535 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25538 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
25539 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
25540 signedness of the result type.
25542 2017-04-12 Richard Biener <rguenther@suse.de>
25543 Jeff Law <law@redhat.com>
25545 PR tree-optimization/80359
25546 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
25547 trim stores to TARGET_MEM_REFs.
25549 2017-04-12 Richard Biener <rguenther@suse.de>
25551 PR tree-optimization/79390
25552 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
25553 threading case even more.
25555 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
25558 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
25559 for quad_address_p for TImode, instead of just not indexed_address.
25561 2017-04-12 Richard Biener <rguenther@suse.de>
25562 Bernd Edlinger <bernd.edlinger@hotmail.de>
25564 PR middle-end/79671
25565 * alias.c (component_uses_parent_alias_set_from): Handle
25566 TYPE_TYPELESS_STORAGE.
25567 (get_alias_set): Likewise.
25568 * tree-core.h (tree_type_common): Add typeless_storage flag.
25569 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
25570 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
25571 for types containing members with TYPE_TYPELESS_STORAGE.
25572 (place_field): Likewise.
25573 (layout_type): Likewise for ARRAY_TYPE.
25574 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
25575 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
25576 TYPE_TYPELESS_STORAGE.
25577 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
25579 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25582 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25583 first argument to type.
25585 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25589 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
25590 CONST0_RTX (mode) rather than const0_rtx where appropriate.
25591 (rs6000_expand_binop_builtin): Likewise.
25592 (rs6000_expand_ternop_builtin): Likewise; also add missing
25593 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
25594 vshasigma built-ins.
25595 * doc/extend.texi: Document that vec_xxpermdi's third argument
25596 must be a constant.
25598 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
25600 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
25601 Use shift_const cost parameter when calculating gain of STV shifts.
25603 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
25605 PR rtl-optimization/70478
25606 * lra-constraints.c (process_alt_operands): Check memory for
25607 disfavoring memory insn operand.
25609 2017-04-11 Jakub Jelinek <jakub@redhat.com>
25611 PR middle-end/80100
25612 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
25613 left shift in unsigned HOST_WIDE_INT type.
25615 PR rtl-optimization/80385
25616 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
25617 (not (neg X)) into (plus X -1) for complex or non-integral modes.
25620 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
25621 if they have any depend clauses.
25623 2017-04-11 Martin Liska <mliska@suse.cz>
25626 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
25627 * ipa-split.c (split_function): Create a local comdat symbol
25628 if caller is in a comdat group.
25630 2017-04-11 Martin Liska <mliska@suse.cz>
25633 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
25636 2017-04-11 Martin Sebor <msebor@redhat.com>
25638 PR middle-end/80364
25639 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
25640 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
25642 (directive::set_width, directive::set_precision, format_character):
25644 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
25647 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
25650 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
25651 conflict, set target->arch_name instead of target->cpu_name.
25653 2017-04-11 Richard Biener <rguenther@suse.de>
25655 PR tree-optimization/80374
25656 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
25657 build_zero_cst, remove fold_convertible_p check again.
25659 2017-04-11 Martin Liska <mliska@suse.cz>
25662 * ubsan.c (instrument_object_size): Do not instrument register
25665 2017-04-11 Jakub Jelinek <jakub@redhat.com>
25668 * config/i386/i386-builtin-types.def
25669 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
25670 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
25671 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
25672 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
25673 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
25674 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
25675 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
25676 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
25677 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
25678 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
25679 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
25680 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
25681 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
25682 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
25683 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
25684 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
25685 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
25686 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
25687 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
25688 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
25689 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
25690 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
25691 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
25692 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
25693 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
25694 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
25695 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
25696 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
25697 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
25698 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
25699 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
25700 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
25701 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
25702 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
25703 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
25704 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
25705 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
25706 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
25707 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
25708 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
25709 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
25710 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
25711 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
25712 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
25713 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
25714 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
25716 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
25717 flag to second_arg_count, handle 4 argument function type _COUNT
25718 aliases, handle second_arg_count on second argument rather than last.
25720 2017-04-10 Jeff Law <law@redhat.com>
25722 PR tree-optimization/80374
25723 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
25724 record anything if we can not convert integer_zero_node to the
25727 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25730 * config/rs6000/rs6000.c (rs6000_option_override_internal):
25731 Enhance special handling given to the TARGET_P9_MINMAX option in
25732 relation to certain other options.
25734 2017-04-10 Bin Cheng <bin.cheng@arm.com>
25736 PR tree-optimization/80153
25737 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
25738 remove POINTER_PLUS_EXPR's base part directly, rather than through
25741 2017-04-10 Richard Biener <rguenther@suse.de>
25742 Bin Cheng <bin.cheng@arm.com>
25744 PR tree-optimization/80153
25745 * tree-affine.c (aff_combination_to_tree): Get base pointer from
25746 the first element of pointer type aff_tree. Build result expr in
25748 (add_elt_to_tree): Convert to type unconditionally. Remove other
25749 fold_convert calls.
25750 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
25751 (rewrite_use_nonlinear_expr): Check invariant using iv information.
25753 2017-04-10 Richard Biener <rguenther@suse.de>
25755 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25758 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
25760 PR rtl-optimization/70478
25761 * lra-constraints.c (curr_small_class_check): New.
25762 (update_and_check_small_class_inputs): New.
25763 (process_alt_operands): Update curr_small_class_check. Disfavor
25764 alternative insn memory operands. Check available regs for small
25767 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
25770 * config/mips/mips.opt (-mvirt): Update description.
25771 * doc/invoke.texi (-mvirt): Likewise.
25773 2017-04-10 Richard Biener <rguenther@suse.de>
25775 PR middle-end/80362
25776 * fold-const.c (fold_binary_loc): Look at unstripped ops when
25777 looking for NEGATE_EXPR in -A / -B to A / B folding.
25779 2017-04-10 Martin Liska <mliska@suse.cz>
25781 PR gcov-profile/80224
25782 * gcov.c (print_usage): Fix usage string.
25783 (get_gcov_intermediate_filename): Remove.
25784 (output_gcov_file): Use both for normal and intermediate format.
25785 (generate_results): Do not initialize special file for
25786 intermediate format.
25788 2017-04-10 Richard Biener <rguenther@suse.de>
25790 PR tree-optimization/80304
25791 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
25794 2017-04-10 Nathan Sidwell <nathan@acm.org>
25797 * config/rs6000/rs6000.c (rs6000_vector_type): New.
25798 (rs6000_init_builtins): Use it.
25800 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25802 * config/arm/arm.md (<mrc>): Add mode to SET source.
25803 (<mrrc>): Likewise.
25805 2017-04-10 Richard Biener <rguenther@suse.de>
25807 PR middle-end/80344
25808 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
25810 2017-04-10 Jakub Jelinek <jakub@redhat.com>
25813 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
25814 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
25815 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
25816 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
25817 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
25818 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
25819 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
25820 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
25821 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
25822 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
25823 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
25824 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
25825 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
25826 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
25827 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
25828 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
25829 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
25830 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
25831 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
25832 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
25833 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
25834 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
25835 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
25837 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
25839 PR rtl-optimization/70478
25840 * lra-constraints.c: Reverse the last patch.
25842 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
25844 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
25845 Add comment for WCHAR_T.
25847 2017-04-08 Martin Liska <mliska@suse.cz>
25850 2017-04-07 Martin Liska <mliska@suse.cz>
25853 * ipa-split.c (split_function): Add function part to a same comdat
25856 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25859 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
25861 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
25863 * rs6000/rs6000.c (vec_load_pendulum): Rename...
25864 (vec_pairing): ...to this.
25865 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
25866 (rs6000_sched_init): Adjust for name change.
25867 (struct rs6000_sched_context): Likewise.
25868 (rs6000_init_sched_context): Likewise.
25869 (rs6000_set_sched_context): Likewise.
25871 2017-04-07 Jakub Jelinek <jakub@redhat.com>
25877 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
25879 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
25880 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
25881 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
25883 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
25885 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
25887 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
25889 PR rtl-optimization/70703
25890 * ira-color.c (update_conflict_hard_regno_costs): Use
25891 int64_t instead of HOST_WIDE_INT.
25893 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
25895 PR rtl-optimization/70478
25896 * lra-constraints.c (process_alt_operands): Disfavor alternative
25897 insn memory operands.
25899 2017-04-07 Jeff Law <law@redhat.com>
25901 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
25902 CALL and NOTE_INSN_CALL_ARG_LOCATION.
25904 2017-04-07 Martin Liska <mliska@suse.cz>
25907 * config/aarch64/aarch64.c (aarch64_process_target_attr):
25908 Show error message instead of an ICE.
25910 2017-04-07 Martin Liska <mliska@suse.cz>
25913 * ipa-split.c (split_function): Add function part to a same comdat
25916 2017-04-07 Richard Biener <rguenther@suse.de>
25918 PR middle-end/80341
25919 * tree.c (get_unwidened): Also handle ! for_type case for
25921 * convert.c (do_narrow): Split out from ...
25922 (convert_to_integer_1): ... here. Do not pass final truncation
25923 type to get_unwidened for TRUNC_DIV_EXPR.
25925 2017-04-07 Richard Biener <rguenther@suse.de>
25927 * tree-affine.c (wide_int_ext_for_comb): Take type rather
25929 (aff_combination_const): Adjust.
25930 (aff_combination_scale): Likewise.
25931 (aff_combination_add_elt): Likewise.
25932 (aff_combination_add_cst): Likewise.
25933 (aff_combination_convert): Likewise.
25934 (add_elt_to_tree): Likewise. Remove unused argument.
25935 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
25937 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
25939 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
25941 * config/arm/arm.c (arm_default_short_enums): Use
25942 ARM_DEFAULT_SHORT_ENUMS.
25943 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
25945 2017-04-06 Jakub Jelinek <jakub@redhat.com>
25948 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
25949 members with redundant out-of-class redeclaration.
25951 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25954 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
25955 * config/i386/i386.md (*zero_extendsidi2):
25956 Add (?*x,*x) and (?*v,*v) alternatives.
25958 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25961 * config/i386/i386.c (ix86_expand_builtin)
25962 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
25963 mode from insn data. Convert operands to insn operand mode.
25964 Copy operands that don't satisfy insn predicate to a register.
25966 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
25968 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
25971 2017-04-06 Richard Biener <rguenther@suse.de>
25973 PR tree-optimization/80334
25974 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
25975 preserve alignment of accesses.
25977 2017-04-06 Richard Biener <rguenther@suse.de>
25979 PR tree-optimization/80262
25980 * tree-sra.c (build_ref_for_offset): Preserve address-space
25982 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
25983 Drop useless address-space information on MEM_REF offsets.
25985 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
25987 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
25989 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
25991 PR rtl-optimization/70703
25992 * ira-color.c (update_conflict_hard_regno_costs): Use
25993 HOST_WIDE_INT instead of long.
25995 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
25998 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
25999 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26001 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26002 for x86_64 target. Handle -m3dnowa option.
26004 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26006 PR rtl-optimization/70703
26007 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26008 (update_conflict_hard_regno_costs): Use long instead of unsigned
26009 arithmetic for cost calculation.
26011 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26012 Bernd Edlinger <bernd.edlinger@hotmail.de>
26015 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26018 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26021 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26022 ptr_mode with Pmode throughout.
26023 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26024 into probe_stack_range and use DImode.
26026 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26029 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26030 call_eh_return is true.
26032 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26034 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26035 Initialize last_match_fntype_index.
26037 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26040 * tree-nvr.c: Include internal-fn.h.
26041 (pass_return_slot::execute): Ignore internal calls without
26044 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26045 Richard Biener <rguenther@suse.de>
26048 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26049 captures used multiple times, except for the last use.
26050 * generic-match-head.c: Include gimplify.h.
26052 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26054 PR tree-optimization/79390
26055 * target.h (struct noce_if_info): Declare.
26056 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26057 * target.def (noce_conversion_profitable_p): New target hook.
26058 * ifcvt.h (struct noce_if_info): New type, moved from ...
26059 * ifcvt.c (struct noce_if_info): ... here.
26060 (noce_conversion_profitable_p): Renamed to ...
26061 (default_noce_conversion_profitable_p): ... this. No longer
26063 (noce_try_store_flag_constants, noce_try_addcc,
26064 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26065 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26066 instead of noce_conversion_profitable_p.
26067 * config/i386/i386.c: Include ifcvt.h.
26068 (ix86_option_override_internal): Don't override
26069 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26070 (ix86_noce_conversion_profitable_p): New function.
26071 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26072 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26073 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26074 * doc/tm.texi: Regenerated.
26076 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26078 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26081 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26084 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26085 instructions for small multiply cores.
26087 2017-04-04 Jeff Law <law@redhat.com>
26089 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26091 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26092 that are not set by the loop over the elements.
26094 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26097 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26098 int mode, convert_modes it to mode as unsigned, otherwise use
26099 lowpart_subreg to mode rather than SImode.
26100 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26101 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26102 Use DImode instead of SImode for the shift count operand.
26103 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26106 2017-04-04 Richard Biener <rguenther@suse.de>
26108 PR middle-end/80281
26109 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26110 arithmetic done for the negate or the plus. Simplify.
26111 (A - (-B) -> A + B): Likewise.
26112 * fold-const.c (split_tree): Make sure to not negate pointers.
26114 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26116 PR rtl-optimization/60818
26117 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26118 a compare of comparisons with the thing compared if this results
26119 in a different machine mode.
26121 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26123 * alias.c (base_alias_check): Fix typo in comment.
26124 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26125 * cgraphunit.c (symbol_table::compile): Likewise.
26126 * collect2.c (maybe_run_lto_and_relink): Likewise.
26127 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26128 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26129 * config/avr/avr.c (avr_map_op_t): Likewise.
26130 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26131 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26132 * config/epiphany/epiphany.md (movcc): Likewise.
26133 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26134 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26136 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26137 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26138 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26139 * config/sh/sh.c (sh_rtx_costs): Likewise.
26140 * fold-const.c (fold_truth_andor): Likewise.
26141 * genautomata.c (collapse_flag): Likewise.
26142 * gengtype.h (struct type::u::s): Likewise.
26143 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26144 * input.c (FORMAT_AMOUNT): Likewise.
26145 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26146 (known_aggs_to_agg_replacement_list): Likewise.
26147 * ipa-inline-analysis.c: Likewise.
26148 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26149 * ipa-polymorphic-call.c
26150 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26151 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26152 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26154 * modulo-sched.c (apply_reg_moves): Likewise.
26155 * omp-expand.c (build_omp_regions_1): Likewise.
26156 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26157 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26158 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26159 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26160 * value-prof.c: Likewise.
26161 * var-tracking.c (val_reset): Likewise.
26163 2017-04-03 Richard Biener <rguenther@suse.de>
26165 PR tree-optimization/80275
26166 * fold-const.c (split_address_to_core_and_offset): Handle
26169 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26171 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26172 descriptors is at least equal to that of functions.
26174 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26176 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26178 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26181 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26182 (mov<IMOD4:mode>): New expander.
26183 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26185 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26187 PR rtl-optimization/79405
26188 * fwprop.c (propagations_left): New variable.
26189 (forward_propagate_into): Decrement it.
26190 (fwprop_init): Initialize it.
26191 (fw_prop): If the variable has reached zero, stop propagating.
26192 (fwprop_addr): Ditto.
26194 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26197 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26198 a FUNCTION_DECL, pass it as decl instead of origin to
26201 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26203 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26206 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26209 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26210 TARGET_VSX_SMALL_INTEGER.
26212 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26214 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26215 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26217 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26219 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26220 extraction from odd-numbered MSA register.
26222 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26224 PR middle-end/80173
26225 * expmed.c (store_bit_field_1): Don't attempt to create
26226 a word subreg out of hard registers wider than word if they
26227 have HARD_REGNO_NREGS of 1 for their mode.
26229 PR middle-end/80163
26230 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26231 conversions to integer types wider than word and pointer.
26234 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26235 (rtx_equal_for_cselib_p): Pass 0 to it.
26236 * cselib.c (cselib_hasher::equal): Likewise.
26237 (rtx_equal_for_cselib_1): Add depth argument. If depth
26238 is 128, don't look up VALUE locs and punt. Increment
26239 depth in recursive calls when walking VALUE locs.
26241 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26243 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26244 (make_gcov_file_name): Use the canonical path name for generating
26246 (read_line): Fix handling of files with ascii null bytes.
26248 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26250 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
26251 to initialise a vector register instead
26252 of using a const_int.
26254 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26256 PR translation/80189
26257 * gimplify.c (omp_default_clause): Use %qs instead of %s in
26258 diagnostic messages.
26260 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
26263 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
26264 (dfp_diex_<mode>): Update mode of operand 1.
26265 * doc/extend.texi (dxex, dxexq): Document change to return type.
26266 (diex, diexq): Document change to argument type.
26268 2017-03-30 Martin Jambor <mjambor@suse.cz>
26271 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
26272 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
26273 it reflects the signature changes performed at the callee side.
26274 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
26275 to cgraph_build_function_type_skip_args.
26276 (build_function_decl_skip_args): Adjust call to the above function.
26278 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26281 * config/i386/sse.md
26282 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
26283 register as dest whenever it is a MEM not rtx_equal_p to the
26284 corresponding dup operand, and when forcing into reg move the
26285 reg into the memory afterwards.
26286 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
26287 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
26288 for the force_reg mode.
26289 (avx512vl_vextractf128<mode>): Use register as dest either
26290 always when a MEM, or when it is a MEM not rtx_equal_p to the
26291 corresponding dup operand, or even not when it is a CONST_VECTOR
26292 depending on the mode and lo vs. hi.
26293 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
26295 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
26296 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
26297 Likewise. Require that operands[2] is even.
26298 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
26299 Remove extraneous parens. Require that operands[2] is a multiple
26301 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
26302 operands[0] is a MEM if <mask_applied>, the predicates/constraints
26303 disallow memory then.
26305 2017-03-30 Richard Biener <rguenther@suse.de>
26307 PR tree-optimization/77498
26308 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
26309 to non-constants over backedges.
26311 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
26313 PR rtl-optimization/80233
26314 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
26315 as last_combined_insn. Do not test for BARRIER_P separately.
26317 2017-03-29 Andreas Schwab <schwab@suse.de>
26320 * calls.c (prepare_call_address): Convert funexp to Pmode before
26321 copying to temp reg.
26323 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26325 PR tree-optimization/80158
26326 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
26327 Handle possible future case of more than one alternate
26329 (replace_rhs_if_not_dup): Likewise.
26330 (replace_one_candidate): Likewise.
26332 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
26334 PR rtl-optimization/80193
26335 * ira.c (ira): Do not check allocation for LRA.
26337 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
26339 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
26340 (nvptx_output_simt_exit): Declare.
26341 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
26342 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
26343 (init_softstack_frame): Move initialization of crtl->is_leaf to...
26344 (nvptx_declare_function_name): ...here. Emit declaration of local
26345 memory space buffer for omp_simt_enter insn.
26346 (nvptx_output_unisimt_switch): New.
26347 (nvptx_output_softstack_switch): New.
26348 (nvptx_output_simt_enter): New.
26349 (nvptx_output_simt_exit): New.
26350 * config/nvptx/nvptx.h (struct machine_function): New fields
26351 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
26352 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
26353 (UNSPECV_SIMT_EXIT): Ditto.
26354 (omp_simt_enter_insn): New insn.
26355 (omp_simt_enter): New expansion.
26356 (omp_simt_exit): New insn.
26357 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
26359 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
26360 (expand_GOMP_SIMT_ENTER_ALLOC): New.
26361 (expand_GOMP_SIMT_EXIT): New.
26362 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
26363 (GOMP_SIMT_ENTER_ALLOC): Ditto.
26364 (GOMP_SIMT_EXIT): Ditto.
26365 * target-insns.def (omp_simt_enter): New insn.
26366 (omp_simt_exit): Ditto.
26367 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
26369 (lower_rec_simd_input_clauses): Implement SIMT privatization.
26370 (lower_rec_input_clauses): Likewise.
26371 (lower_lastprivate_clauses): Handle SIMT privatization.
26373 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
26374 (ompdevlow_adjust_simt_enter): New.
26375 (find_simtpriv_var_op): New.
26376 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
26377 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
26379 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
26380 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
26381 (copy_decl_for_dup_finish): Ditto.
26383 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
26385 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
26388 * config/i386/i386.c (ix86_option_override_internal): Always
26389 allow -mpreferred-stack-boundary=3 for 64-bit targets.
26391 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26393 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
26395 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26397 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
26398 mark new edge's irreducible flag accordign to it.
26399 (vect_do_peeling): Check loop preheader edge's irreducible flag
26400 and pass it to function slpeel_add_loop_guard.
26402 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
26404 PR tree-optimization/80218
26405 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26406 Update block frequencies and counts.
26408 2017-03-28 Richard Biener <rguenther@suse.de>
26410 PR tree-optimization/78644
26411 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
26412 of a simplification result we may not use it at all.
26414 2017-03-28 Richard Biener <rguenther@suse.de>
26417 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
26418 without arguments, generate default definition of a SSA name.
26420 2017-03-28 Richard Biener <rguenther@suse.de>
26422 PR middle-end/80222
26423 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
26424 TYPE_REF_CAN_ALIAS_ALL references.
26425 * fold-const.c (fold_indirect_ref_1): Likewise.
26427 2017-03-28 Martin Liska <mliska@suse.cz>
26430 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
26431 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
26433 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26434 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
26436 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
26437 (EXTRA_SPECS): Define.
26438 (SUBTARGET_EXTRA_SPECS): Likewise.
26439 (SUBTARGET_CPP_SPEC): Likewise.
26440 * config/arc/elf.h (EXTRA_SPECS): Renamed to
26441 SUBTARGET_EXTRA_SPECS.
26442 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
26444 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26446 * config/arc/simdext.md (vst64_insn): Update pattern.
26447 (vld32wh_insn): Likewise.
26448 (vld32wl_insn): Likewise.
26449 (vld64_insn): Likewise.
26450 (vld32_insn): Likewise.
26452 2017-03-28 Marek Polacek <polacek@redhat.com>
26455 * fold-const.c (fold_comparison): Use protected_set_expr_location
26456 instead of SET_EXPR_LOCATION.
26458 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
26460 * tree.c (add_expr): Avoid name lookup warning.
26462 2017-03-27 Jeff Law <law@redhat.com>
26464 PR tree-optimization/80216
26465 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
26466 function name. Limit recursion depth.
26467 (record_temporary_equivalences): Corresponding changes.
26469 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
26471 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
26474 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26477 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
26479 * cfgcleanup.c (reg_note_cfa_p): New array.
26480 (insns_have_identical_cfa_notes): New function.
26481 (old_insns_match_p): Don't cross-jump in between /f
26482 and non-/f instructions. If both i1 and i2 are frame related,
26483 verify all CFA notes, their order and content.
26485 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
26488 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
26489 HImode and SImode with zero extend to DImode to one insn.
26490 (bswap<mode>2_extenddi): Likewise.
26491 (bswapsi2_extenddi): Likewise.
26492 (bswaphi2_extendsi): Likewise.
26493 (bswaphi2): Combine bswap HImode and SImode into one insn.
26494 Separate memory insns from swapping register.
26495 (bswapsi2): Likewise.
26496 (bswap<mode>2): Likewise.
26497 (bswaphi2_internal): Delete, no longer used.
26498 (bswapsi2_internal): Likewise.
26499 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
26500 store, and gpr<-gpr swap insns.
26501 (bswap<mode>2_store): Likewise.
26502 (bswaphi2_reg): Register only splitter, combine with the splitter.
26503 (bswaphi2 splitter): Likewise.
26504 (bswapsi2_reg): Likewise.
26505 (bswapsi2 splitter): Likewise.
26506 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
26507 the insns into load, store, and register/register insns.
26508 (bswapdi2_ldbrx): Likewise.
26509 (bswapdi2_load): Likewise.
26510 (bswapdi2_store): Likewise.
26511 (bswapdi2_reg): Likewise.
26513 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
26515 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
26516 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
26518 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
26521 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
26523 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26524 special handling for target option conflicts between dform
26525 options (-mpower9-dform, -mpower9-dform-vector,
26526 -mpower9-dform-scalar) and -mno-direct-move.
26528 2017-03-27 Richard Biener <rguenther@suse.de>
26530 PR tree-optimization/80181
26531 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
26533 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26535 * config/arc/predicates.md (move_double_src_operand): Replace the
26536 call to move_double_src_operand with a call to address_operand.
26538 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26540 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
26541 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
26542 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
26544 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26546 * config/arc/predicates.md (long_immediate_loadstore_operand):
26547 Consider scaled addresses cases.
26549 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26551 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
26552 restored when in interrupt.
26553 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
26554 doesn't have delay slot.
26556 2017-03-27 Richard Biener <rguenther@suse.de>
26559 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
26560 inlined thunk clones.
26562 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26565 * asan.c (instrument_derefs): Copy over last operand from
26566 original COMPONENT_REF to the new COMPONENT_REF with
26567 DECL_BIT_FIELD_REPRESENTATIVE.
26568 * ubsan.c (instrument_object_size): Likewise.
26570 2017-03-27 Richard Biener <rguenther@suse.de>
26572 PR tree-optimization/80170
26573 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
26574 sure DR/SCEV didnt fold in constants we do not see when looking
26575 at the reference base alignment.
26577 2017-03-27 Richard Biener <rguenther@suse.de>
26579 PR middle-end/80171
26580 * gimple-fold.c (fold_ctor_reference): Properly guard against
26581 NULL return value from canonicalize_constructor_val.
26583 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
26586 * config/i386/i386.c (ix86_expand_builtin)
26587 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
26588 flags reg setting and flags reg using instructions.
26589 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
26590 clobbering instructions to zero extend op2.
26592 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
26594 * doc/install.texi (Configuration) <--with-aix-soname>:
26595 Update link to AIX ld.
26597 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
26599 PR rtl-optimization/80160
26600 PR rtl-optimization/80159
26601 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
26602 reg_alternate_class into account.
26604 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
26607 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
26608 to consider in curr_insn_transform.
26610 2017-03-24 Jakub Jelinek <jakub@redhat.com>
26612 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
26613 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
26614 and emit_mode_inner.
26616 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26618 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
26619 argument to the overloaded builtin variants. Use the new flag to
26620 deprecate certain builtin variants.
26621 * config/s390/s390-builtin-types.def: Add new builtin types.
26622 * config/s390/s390-builtins.h: Support new flags field for
26623 overloaded builtins.
26624 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
26625 (s390_macro_to_expand): Enable vector float data type.
26626 (s390_cpu_cpp_builtins_internal): Indicate support of the new
26627 builtins by incrementing the __VEC__ version number.
26628 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
26630 (s390_resolve_overloaded_builtin): Emit error messages depending
26631 on the builtin flags.
26632 * config/s390/s390.c (s390_expand_builtin): Support additional
26633 flags argument. Change error message to match the messages
26634 emitted in s390-c.c.
26635 * config/s390/s390.md: New UNSPEC_* constants.
26636 (op_type): Add new instruction types.
26637 * config/s390/vecintrin.h: Add new builtins and test data class
26639 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
26640 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
26641 (VEC_INEXACT, VEC_NOINEXACT): New constants.
26642 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
26643 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
26644 ("vec_mergel<mode>"): V_HW -> VEC_HW.
26646 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
26647 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
26648 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
26649 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
26651 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
26652 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
26653 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
26654 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
26656 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
26657 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
26658 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
26659 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
26660 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
26661 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
26662 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
26664 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
26665 ("vec_scatter_element<V_HW_4:mode>_DI")
26666 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
26667 ("vec_fpint<mode>", "vflls")
26668 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
26669 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
26670 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
26671 ("*vec_cmphe<mode>_cc"): ... these.
26673 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
26674 mode constant instead of magic value.
26676 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26678 * config/s390/s390.c (s390_expand_vec_compare): Support other
26679 vector floating point modes than just V2DF.
26680 (s390_expand_vcond): Likewise.
26681 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
26682 (s390_cannot_change_mode_class): Prevent mode changes between TF
26683 and V1TF in vector registers.
26684 * config/s390/s390.md (DF, SF): New mode attributes.
26685 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
26686 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
26687 SFmode support for VRs.
26688 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
26690 (VFT, VF_HW): New mode iterators.
26691 (vw, sdx): New mode attributes.
26692 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
26693 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
26694 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
26695 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
26696 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
26697 also the new vector floating point modes. Renaming to ...
26699 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
26700 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
26701 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
26702 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
26703 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
26704 ("vec_unordered<mode>"): ... these.
26706 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
26707 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
26708 ("*vec_extendv2df"): New insn definitions.
26710 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26712 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
26713 ("mulditi3_2", "*muldi3_sign"): New patterns.
26714 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
26715 rename the pattern definition.
26717 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26719 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
26721 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
26723 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26725 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
26726 instruction if possible.
26727 * config/s390/vector.md (vec_halfnumelts): New mode
26729 ("*vec_vllezlf<mode>"): New pattern.
26731 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26733 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
26734 ("popcountv4si2", "popcountv2di2"): Rename to ...
26735 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
26736 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
26738 ("popcount<mode>2_vxe"): New pattern.
26740 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26742 * common/config/s390/s390-common.c (processor_flags_table): Add
26744 * config.gcc: Add arch12.
26745 * config/s390/driver-native.c (s390_host_detect_local_cpu):
26746 Default to arch12 for unknown CPU model numbers.
26747 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
26748 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
26749 PROCESSOR_max sanity check.
26750 * config/s390/s390-opts.h (enum processor_type): Add
26752 * config/s390/s390.c (processor_table): Add arch12.
26753 (s390_expand_builtin): Add check for B_VXE flag.
26754 (s390_issue_rate): Add PROCESSOR_ARCH12.
26755 (s390_get_sched_attrmask): Likewise.
26756 (s390_get_unit_mask): Likewise.
26757 (s390_sched_score): Enable z13 scheduling for arch12.
26758 (s390_sched_reorder): Likewise.
26759 (s390_sched_variable_issue): Likewise.
26760 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
26762 (s390_tune_attr): Use z13 scheduling also for arch12.
26763 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
26764 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
26765 (TARGET_VXE_P): New macros.
26766 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
26767 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
26768 * config/s390/s390.opt: Add arch12 as processor_type.
26770 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26772 * config/s390/s390.md
26773 ("fixuns_truncdddi2", "fixuns_trunctddi2")
26774 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
26775 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
26777 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
26778 Rename expanders to ...
26780 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
26781 ("fixuns_truncdddi2_emu"): ... these.
26783 ("fixuns_trunc<mode>si2_emu"): New expander.
26785 ("*fixuns_truncdfdi2_z13"): Rename to ...
26786 ("*fixuns_truncdfdi2_vx"): ... this.
26788 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26790 * config/s390/2964.md: Remove the single element vector compare
26791 instructions which are no longer used.
26792 * config/s390/s390.c (s390_select_ccmode): Remove handling of
26794 (s390_canonicalize_comparison): Remove handling of DFmode
26796 (s390_expand_vec_compare_scalar): Remove function.
26797 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
26798 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
26800 ("*cmp<mode>_ccs"): Add wfcdb instruction.
26802 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26804 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
26806 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
26807 will anyway by matched by mov<mode>_64dfp.
26809 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26811 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
26812 vlef/vstef. Add missing operand to vleif.
26814 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26816 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
26817 pair for all vector types with 64 bit elements.
26818 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
26819 * config/s390/vector.md (V_HW_64): ... here.
26820 (V_128_NOSINGLE): New mode iterator.
26821 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
26822 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
26823 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
26824 ("*vec_load_pairv2di"): Change to ...
26825 ("*vec_load_pair<mode>"): ... this one.
26827 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26829 * config/s390/constraints.md: Add comments.
26830 (jKK): Reject element sizes > 8 bytes.
26831 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
26833 * config/s390/s390.md: Add the s_operand checks formerly in
26834 s390_split_ok_p to various splitters where they are still
26836 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
26837 for 128 bit vectors. Plus two splitters.
26839 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26841 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
26844 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26847 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
26848 error if the boundary argument is not constant.
26850 2017-03-24 Jakub Jelinek <jakub@redhat.com>
26852 PR rtl-optimization/80112
26853 * loop-doloop.c (doloop_condition_get): Don't check condition
26854 if cmp isn't SET with IF_THEN_ELSE src.
26856 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26858 PR tree-optimization/80158
26859 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
26860 replacing a candidate statement, also replace it for the
26861 candidate's alternate interpretation.
26862 (replace_rhs_if_not_dup): Likewise.
26863 (replace_one_candidate): Likewise.
26865 2017-03-24 Richard Biener <rguenther@suse.de>
26867 PR tree-optimization/80167
26868 * graphite-isl-ast-to-gimple.c
26869 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
26871 (translate_isl_ast_to_gimple::get_rename): Likewise.
26873 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26875 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
26876 handling of certain combinations of target options, including the
26877 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
26878 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
26880 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26883 * config/arm/arm.md (*load_multiple): Add reload_completed to
26884 matching condition.
26886 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26887 Richard Biener <rguenth@suse.de>
26889 PR tree-optimization/79908
26890 PR tree-optimization/80136
26891 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
26892 been cast away, gimplify_and_add suffices.
26894 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
26896 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
26898 2017-03-23 Richard Biener <rguenther@suse.de>
26900 PR tree-optimization/80032
26901 * gimplify.c (gimple_push_cleanup): Forced unconditional
26902 cleanups still have to go to the conditional_cleanups
26905 2017-03-22 Jakub Jelinek <jakub@redhat.com>
26907 PR tree-optimization/80072
26908 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
26910 (next_operand_entry_id): Change type to unsigned int.
26911 (sort_by_operand_rank): Make sure to return the right return value
26912 even if unsigned fields are bigger than INT_MAX.
26913 (struct oecount): Change cnt and id type to unsigned int.
26914 (oecount_hasher::equal): Formatting fix.
26915 (oecount_cmp): Make sure to return the right return value
26916 even if unsigned fields are bigger than INT_MAX.
26917 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
26920 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
26921 TREE_READONLY on result if writing it more than once.
26924 * doc/invoke.texi (-fsanitize=thread): Document that with
26925 -fnon-call-exceptions atomics are not able to throw
26929 * tsan.c: Include tree-eh.h.
26930 (instrument_builtin_call): Call maybe_clean_eh_stmt or
26931 maybe_clean_or_replace_eh_stmt where needed.
26932 (instrument_memory_accesses): Add cfg_changed argument.
26933 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
26934 if it returned true.
26935 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
26937 PR rtl-optimization/63191
26938 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
26939 wrapper function, moved the whole old content into ...
26940 (ix86_delegitimize_address_1): ... this. New inline function.
26941 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
26942 true as last argument instead of ix86_delegitimize_address.
26944 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26946 * config/aarch64/aarch64.c (generic_branch_cost): Copy
26947 cortexa57_branch_cost.
26949 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26951 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
26953 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26956 * doc/md.texi (Constraints): Document wA constraint.
26957 * config/rs6000/constraints.md (wA): New.
26958 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
26959 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
26960 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
26961 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
26963 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
26966 * gimplify.c (is_oacc_declared): New function.
26967 (oacc_default_clause): Use it to set default flags for acc declared
26968 variables inside parallel regions.
26969 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
26970 declared variables.
26971 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
26972 declare attribute to any decl as necessary.
26974 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
26977 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
26978 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
26979 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
26980 (arm_arch_lpae): This.
26981 * config/arm/arm.c (arm_arch7ve): Rename into ...
26982 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
26983 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
26986 2017-03-22 Martin Liska <mliska@suse.cz>
26989 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
26990 error message instead of an ICE.
26992 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26994 * doc/extend.texi (6.11 Additional Floating Types): Revise.
26996 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
26998 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27000 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27003 2017-03-21 Martin Sebor <msebor@redhat.com>
27005 * doc/extend.texi: Use "cannot" instead of "can't."
27006 * doc/hostconfig.texi: Same.
27007 * doc/install.texi: Same.
27008 * doc/invoke.texi: Same.
27009 * doc/loop.texi: Same.
27010 * doc/md.texi: Same.
27011 * doc/objc.texi: Same.
27012 * doc/rtl.texi: Same.
27013 * doc/tm.texi: Same.
27014 * doc/tm.texi.in: Same.
27015 * doc/trouble.texi: Same.
27017 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27020 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27021 (collect_checksum_attributes): Set it.
27022 (die_checksum_ordered): Use it.
27024 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27026 PR tree-optimization/79908
27027 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27028 change: For a VA_ARG whose LHS has been cast away, use
27029 force_gimple_operand to construct the side effects.
27031 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27033 PR translation/80001
27034 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27035 more amenable to translation.
27036 (oacc_loop_auto_partitions): Likewise.
27038 2017-03-21 Marek Polacek <polacek@redhat.com>
27039 Martin Sebor <msebor@redhat.com>
27041 PR tree-optimization/80109
27042 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27043 on INTEGRAL_TYPE_P.
27045 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27046 Segher Boessenkool <segher@kernel.crashing.org>
27049 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27050 check reg_used_between_p between insn and one of succ or succ2
27051 depending on if succ is artificial insn not inserted into insn
27054 2017-03-21 Martin Liska <mliska@suse.cz>
27056 PR gcov-profile/80081
27057 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27058 * doc/gcc.texi: Include gcov-dump stuff.
27059 * doc/gcov-dump.texi: New file.
27061 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27063 PR rtl-optimization/79150
27064 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27065 conditional jump, if the jump is the last insn of the loop.
27067 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27068 Richard Biener <rguenth@suse.de>
27070 PR tree-optimization/79908
27071 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27072 been cast away, use force_gimple_operand to construct the side
27075 2017-03-21 Martin Liska <mliska@suse.cz>
27077 PR libfortran/79956
27078 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27081 2017-03-21 Brad Spengler <spender@grsecurity.net>
27084 * plugin.c (htab_hash_plugin): New function.
27085 (add_new_plugin): Use it and adjust.
27086 (parse_plugin_arg_opt): Adjust.
27087 (init_one_plugin): Likewise.
27089 2017-03-21 Richard Biener <rguenther@suse.de>
27091 PR tree-optimization/80032
27092 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27093 if set force the cleanup to happen unconditionally.
27094 (gimplify_target_expr): Push inserted clobbers with force_uncond
27095 to avoid them being removed by control-dependent DCE.
27097 2017-03-21 Richard Biener <rguenther@suse.de>
27099 PR tree-optimization/80122
27100 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27101 va_arg_pack_len when the inlined call stmt requires pack
27103 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27105 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27108 * tsan.c (instrument_builtin_call): If the memory model argument
27109 is not a constant, assume it is valid.
27112 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27115 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27117 PR rtl-optimization/79910
27118 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27119 if its dest is used by an insn before I2 (other than the combined
27120 insns themselves, which are properly handled already).
27122 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27125 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27127 * combine.c (record_used_regs): New static function.
27128 (try_combine): Handle situations where there is an additional
27129 instruction between I2 and I3 which needs to have a LOG_LINK
27133 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27135 * combine.c (try_combine): Delete redundant i1 test. Call
27136 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27138 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27141 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27142 alternatives 13/14.
27144 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27146 PR tree-optimization/80054
27147 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27148 the optimization if a PHI or any of its arguments is not dominated
27149 by the candidate's basis. Use gphi* rather than gimple* as
27151 (replace_profitable_candidates): Clean up a gimple* variable that
27152 should be a gphi* variable.
27154 2017-03-20 Martin Sebor <msebor@redhat.com>
27157 * doc/extend.texi (attribute constructor): Document present limitation.
27159 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27162 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27163 __POWER9_VECTOR__ #ifdef control, change template definition to
27164 use Power9-specific built-in function.
27165 (vec_any_eq): Likewise.
27166 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27167 to control outcomes from this test.
27168 (vector_ae_<mode>p): For VEC_F modes, likewise.
27170 2017-03-20 Ian Lance Taylor <iant@google.com>
27172 * config/i386/i386.c (ix86_function_regparm): Save an extra
27173 register for -fsplit-stack with DECL_STATIC_CHAIN.
27175 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27178 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27179 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27181 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27183 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27185 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27188 2017-03-20 Marek Polacek <polacek@redhat.com>
27191 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27193 2017-03-20 Richard Biener <rguenther@suse.de>
27195 PR tree-optimization/80113
27196 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27197 allocate extra SSA name for PHI def.
27198 (add_close_phis_to_outer_loops): Likewise.
27199 (add_close_phis_to_merge_points): Likewise.
27200 (copy_loop_close_phi_args): Likewise.
27201 (copy_cond_phi_nodes): Likewise.
27203 2017-03-20 Martin Liska <mliska@suse.cz>
27205 PR middle-end/79753
27206 * tree-chkp.c (chkp_build_returned_bound): Do not build
27207 returned bounds for a LHS that's not a BOUNDED_P type.
27209 2017-03-20 Martin Liska <mliska@suse.cz>
27213 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27214 COMPLEX_CST and VECTOR_CST.
27216 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27219 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27220 target operand. A new splitter adds the clobber statement in case
27221 the target operand is dead anyway.
27223 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27225 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27226 to age-old versions of binutils and glibc.
27228 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27230 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27232 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27234 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27236 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27238 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27239 requirement for binutils 2.13.
27241 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27243 * combine.c (try_combine): Delete redundant i1 test. Call
27244 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27246 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27248 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27249 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
27251 <riscv64-*-elf>: Re-arrange section
27252 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
27253 <riscv32-*-linux>: Likewise.
27254 <riscv64-*-elf>: Likewise
27255 <riscv64-*-linux>: Likewise.
27257 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
27260 * aarch64.opt(verbose-cost-dump): Fix typo.
27262 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
27265 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
27266 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
27268 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27270 * reload.c (find_reloads): When reloading a nonoffsettable address,
27271 use RELOAD_OTHER for it and its address reloads.
27273 PR rtl-optimization/79910
27274 * combine.c (record_used_regs): New static function.
27275 (try_combine): Handle situations where there is an additional
27276 instruction between I2 and I3 which needs to have a LOG_LINK
27279 2017-03-17 Jeff Law <law@redhat.com>
27281 PR tree-optimization/71437
27282 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
27283 conditional in the hash table first.
27284 (vrp_dom_walker::before_dom_children): Extract condition from
27285 ASSERT_EXPR. Record condition, its inverion and any implied
27286 conditions as well.
27288 2017-03-17 Marek Polacek <polacek@redhat.com>
27289 Markus Trippelsdorf <markus@trippelsdorf.de>
27291 PR tree-optimization/80079
27292 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
27295 2017-03-17 Richard Biener <rguenther@suse.de>
27297 PR middle-end/80075
27298 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
27299 Properly verify the LHS before the RHS possibly claims to be
27301 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
27304 2017-03-17 Martin Jambor <mjambor@suse.cz>
27306 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
27307 (List of -O2 options): Likewise.
27308 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
27311 2017-03-17 Tom de Vries <tom@codesourcery.com>
27313 * gcov-dump.c (print_usage): Print bug_report_url.
27315 2017-03-17 Richard Biener <rguenther@suse.de>
27317 PR middle-end/80050
27318 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
27319 (parser::peek): Likewise.
27321 2017-03-17 Richard Biener <rguenther@suse.de>
27323 PR tree-optimization/80048
27324 * sese.c (free_sese_info): Properly release rename_map and
27325 copied_bb_map elements.
27327 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
27329 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
27330 Add linked-list forward and backlinks. Insert on
27331 construction, remove on destruction.
27332 (class pass_store_merging): Add m_stores_head field.
27333 (pass_store_merging::terminate_and_process_all_chains):
27334 Iterate over m_stores_head list.
27335 (pass_store_merging::terminate_all_aliasing_chains):
27337 (pass_store_merging::execute): Check for debug stmts first.
27338 Push new chains onto the m_stores_head stack.
27340 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
27343 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
27344 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
27345 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
27347 2017-03-16 Jeff Law <law@redhat.com>
27349 PR tree-optimization/71437
27350 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
27351 member function. Implementation moved into after_dom_children
27352 member function and into the threader's thread_outgoing_edges
27354 (dom_opt_dom_walker::after_dom_children): Simplify by moving
27355 some code into new thread_outgoing_edges.
27356 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
27357 definition. Simplify marker handling (do it here). Assume we always
27358 have the available expression and the const/copies tables.
27359 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
27361 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
27362 * tree-vrp.c (equiv_stack): No longer file scoped.
27363 (vrp_dom_walker): New class.
27364 (vrp_dom_walker::before_dom_children): New member function.
27365 (vrp_dom_walker::after_dom_children): Likewise.
27366 (identify_jump_threads): Setup domwalker. Use it rather than
27367 walking edges in a random order by hand. Simplify setup/finalization.
27368 (finalize_jump_threads): Remove.
27369 (vrp_finalize): Do not call identify_jump_threads here.
27370 (execute_vrp): Do it here instead and call thread_through_all_blocks
27373 PR tree-optimization/71437
27374 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
27376 (simplify_stmt_for_jump_threading): Add basic_block argument. All
27378 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
27379 (dom_opt_dom_walker::thread_across_edge): Remove
27380 handle_dominating_asserts argument. All callers changed.
27381 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
27382 changes. Remove calls to lhs_of_dominating_assert. Other
27383 uses of handle_dominating_asserts turn into unconditional code
27384 (simplify_control_stmt_condition_1): Likewise.
27385 (simplify_control_stmt_condition): Likewise.
27386 (thread_through_normal_block, thread_across_edge): Likewise.
27387 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
27388 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
27389 object if it is not an SSA_NAME.
27390 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
27391 before calling into the VRP specific simplifiers.
27392 (identify_jump_threads): Remove handle_dominating_asserts
27395 2017-03-16 Jakub Jelinek <jakub@redhat.com>
27398 * tree-diagnostic.c (default_tree_printer): No longer static.
27399 * tree-diagnostic.h (default_tree_printer): New prototype.
27401 2017-03-16 Tamar Christina <tamar.christina@arm.com>
27403 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
27404 Change ins into fmov.
27406 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27408 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
27409 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
27410 Use h_con constraint for operand 1.
27411 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
27412 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
27414 2017-03-15 Jeff Law <law@redhat.com>
27416 PR tree-optimization/71437
27417 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
27418 (record_temporary_equivalences): Use it.
27420 PR tree-optimization/71437
27421 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
27422 tree-ssa-scopedtables.
27423 (lookup_avail_expr, build_and_record_new_cond): Likewise.
27424 (record_conditions, record_cond, vuse_eq): Likewise.
27425 (record_edge_info): Adjust to API tweak of record_conditions.
27426 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
27427 (record_temporary_equivalences, optimize_stmt): Likewise.
27428 (eliminate_redundant_computations): Likewise.
27429 (record_equivalences_from_stmt): Likewise.
27430 * tree-ssa-scopedtables.c: Include options.h and params.h.
27431 (vuse_eq): New function, moved from tree-ssa-dom.c
27432 (build_and_record_new_cond): Likewise.
27433 (record_conditions): Likewise. Accept vector of conditions rather
27434 than edge_equivalence structure for first argument.
27435 for the first argument.
27436 (avail_exprs_stack::lookup_avail_expr): New member function, moved
27437 from tree-ssa-dom.c.
27438 (avail_exprs_stack::record_cond): Likewise.
27439 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
27440 from tree-ssa-dom.c.
27441 (avail_exprs_stack): Add new member functions lookup_avail_expr
27443 (record_conditions): Declare.
27445 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
27448 * lra-constraints.c (process_alt_operands): Increase reject for
27449 reloading an input/output operand.
27451 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
27454 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
27455 insns to convert from signed/unsigned char/short to IEEE 128-bit
27457 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
27459 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
27462 * config/i386/i386.c (ix86_vector_duplicate_value): Create
27463 subreg of inner mode for values already in registers.
27465 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
27467 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
27468 iteration reg is used after the loop.
27470 2017-03-14 Martin Sebor <msebor@redhat.com>
27472 PR tree-optimization/79800
27473 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
27474 precision in negative-positive range.
27475 (format_floating): Call non-const overload with adjusted precision.
27477 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27480 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
27483 2017-03-14 Martin Sebor <msebor@redhat.com>
27485 PR middle-end/80020
27486 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
27487 * builtins.def (aligned_alloc): Use it.
27490 * Makefile.in (GTFILES): Add calls.c.
27491 * calls.c: Include "gt-calls.h".
27493 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
27495 PR rtl-optimization/79728
27496 * regs.h (struct target_regs): New field
27497 x_contains_allocatable_regs_of_mode.
27498 (contains_allocatable_regs_of_mode): New macro.
27499 * reginfo.c (init_reg_sets_1): Initialize it, and change
27500 contains_reg_of_mode so it includes global regs as well.
27501 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
27502 rather than contains_regs_of_mode.
27504 2017-03-14 Martin Liska <mliska@suse.cz>
27506 * doc/invoke.texi: Document options that can't be combined with
27507 -fcheck-pointer-bounds.
27509 2017-03-14 Martin Liska <mliska@suse.cz>
27511 PR middle-end/79831
27512 * doc/invoke.texi (-Wchkp): Document the option.
27514 2017-03-14 Martin Liska <mliska@suse.cz>
27516 * Makefile.in: Install gcov-dump.
27518 2017-03-14 Martin Liska <mliska@suse.cz>
27520 * multiple_target.c (expand_target_clones): Bail out for
27521 an invalid attribute.
27523 2017-03-14 Richard Biener <rguenther@suse.de>
27525 * alias.c (struct alias_set_entry): Pack properly.
27526 * cfgloop.h (struct loop): Likewise.
27527 * cse.c (struct set): Likewise.
27528 * ipa-utils.c (struct searchc_env): Likewise.
27529 * loop-invariant.c (struct invariant): Likewise.
27530 * lra-remat.c (struct cand): Likewise.
27531 * recog.c (struct change_t): Likewise.
27532 * rtl.h (struct address_info): Likewise.
27533 * symbol-summary.h (function_summary): Likewise.
27534 * tree-loop-distribution.c (struct partition): Likewise.
27535 * tree-object-size.c (struct object_size_info): Likewise.
27536 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
27537 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
27538 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
27539 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
27540 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
27541 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
27542 (struct _stmt_vec_info): Likewise.
27544 2017-03-14 Martin Liska <mliska@suse.cz>
27547 * multiple_target.c (create_dispatcher_calls): Check that
27548 a target can create a function dispatcher.
27550 2017-03-14 Martin Liska <mliska@suse.cz>
27553 * multiple_target.c (expand_target_clones): Drop local.local
27554 flag for default implementation.
27556 2017-03-14 Richard Biener <rguenther@suse.de>
27558 PR tree-optimization/80030
27559 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
27561 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
27563 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
27564 gcc_fallthrough() instead of __attribute__((fallthrough));
27566 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27568 * doc/gcc.texi: Remove "up" link to (DIR).
27569 * doc/gccint.texi: Ditto.
27571 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27573 * doc/install.texi (Specific) <avr>: Remove reference to
27576 2017-03-13 Jeff Law <law@redhat.com>
27578 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
27579 attribute rather than comments.
27581 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
27582 match_scratch operand is highest.
27584 2017-03-13 Martin Liska <mliska@suse.cz>
27586 PR middle-end/78339
27587 * ipa-pure-const.c (warn_function_noreturn): If the declarations
27588 is a CHKP clone, use original declaration.
27590 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27592 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
27593 (arc_conditional_register_usage): Use a different allocation order
27594 when optimizing for size.
27595 * common/config/arc/arc-common.c (arc_option_optimization_table):
27596 Section anchors default on when optimizing for size.
27598 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27600 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
27602 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27604 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
27605 * config/arc/arc.md (cpu_facility): Add cd variant.
27606 (*movqi_insn): Add code density variant.
27607 (*movhi_insn): Likewise.
27608 (*movqi_insn): Likewise.
27609 (*addsi3_mixed): Likewise.
27610 (subsi3_insn): Likewise.
27612 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27614 * config/arc/arc.md (movsi_cond_exec): Update constraint.
27616 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27618 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
27619 expressions with MINUS and UNARY ops.
27621 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27624 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
27626 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
27627 between vec_select and vector argument.
27628 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
27629 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
27630 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
27631 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
27632 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
27633 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
27635 2017-03-13 Richard Biener <rguenther@suse.de>
27638 * params.def (vect-max-peeling-for-alignment): Fix typo.
27640 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27642 * doc/install.texi (Specific) <mips-*-*>: Remove description of
27643 issue that only occurred with binutils below 2.18.
27645 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27647 * doc/install.texi (Specific) <cris-axis-elf>: No longer
27648 refer to binutils 2.11/2.12 minimum.
27650 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27652 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
27653 ftp.kernel.org and simplify binutils requirement.
27655 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
27657 * doc/invoke.texi (Warning Options): Fix spelling of link-time
27659 (Optimize Options): Ditto. Also remove redundancy.
27661 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27663 PR translation/79848
27664 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
27666 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
27667 to G_ to avoid double translation.
27669 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27671 PR translation/79923
27672 * auto-profile.c (get_combined_location): Convert leading
27673 character of diagnostics to lower case and remove trailing period.
27674 (read_profile): Likewise for various diagnostics.
27675 * config/arm/arm.c (arm_option_override): Remove trailing period
27676 from various diagnostics.
27677 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
27678 (msp430_expand_delay_cycles): Likewise.
27680 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27683 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
27684 full command-line argument, rather than just "str".
27685 (aarch64_validate_march): Likewise.
27686 (aarch64_validate_mtune): Likewise.
27688 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
27690 PR rtl-optimization/78911
27691 * lra-assigns.c (must_not_spill_p): New function.
27692 (spill_for): Use it.
27694 2017-03-10 Jakub Jelinek <jakub@redhat.com>
27696 PR tree-optimization/79981
27697 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
27698 ATOMIC_COMPARE_EXCHANGE ifn result.
27699 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
27700 IFN_ATOMIC_COMPARE_EXCHANGE.
27702 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27705 * opts.c (parse_sanitizer_options): Add missing question mark to
27706 "did you mean" message.
27708 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27710 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
27712 (VMULEUH_UNS): Likewise.
27713 (VMULOUB_UNS): Likewise.
27714 (VMULOUH_UNS): Likewise.
27715 * config/rs6000/rs6000.c (builtin_function_type): Remove
27716 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
27718 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27721 * read-rtl-function.c (function_reader::read_rtx_operand): Update
27722 x with result of extra_parsing_for_operand_code_0.
27723 (function_reader::extra_parsing_for_operand_code_0): Convert
27724 return type from void to rtx, returning x. When reading
27725 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
27726 larger size containing struct block_symbol.
27728 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
27730 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
27731 -mfloat128-hardware without -m64.
27733 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
27736 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
27737 entries to the case statement that marks unsigned arguments to
27738 overloaded functions.
27740 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27742 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
27743 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
27745 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
27748 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
27749 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
27751 2017-03-10 Martin Liska <mliska@suse.cz>
27755 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
27756 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
27759 2017-03-10 Olivier Hainque <hainque@adacore.com>
27761 * tree-switch-conversion (array_value_type): Start by resetting
27762 candidate type to it's main variant.
27764 2017-03-10 Jakub Jelinek <jakub@redhat.com>
27766 PR rtl-optimization/79909
27767 * combine.c (try_combine): Use simplify_replace_rtx on individual
27768 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
27769 of the whole CALL_INSN_FUNCTION_USAGE.
27771 PR tree-optimization/79972
27772 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
27773 get_range_info on SSA_NAMEs. Formatting fixes.
27775 2017-03-10 Richard Biener <rguenther@suse.de>
27776 Jakub Jelinek <jakub@redhat.com>
27778 PR tree-optimization/77975
27779 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
27780 edge to be constant.
27781 (get_val_for): For constant x return it. Formatting fix.
27782 (loop_niter_by_eval): Avoid pointless looping if the next iteration
27783 would use the same bases as the current one.
27785 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27787 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
27788 instead of vec_select for V1TImode.
27789 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
27791 (VSX_LE_128): Add V1TI to this mode iterator.
27792 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
27793 (*vsx_le_perm_store_<mode>): Likewise.
27794 (pre-reload splitter for VSX stores): Likewise.
27795 (post-reload splitter for VSX stores): Likewise.
27796 (*vsx_xxpermdi2_le_<mode>): Likewise.
27797 (*vsx_lxvd2x2_le_<mode>): Likewise.
27798 (*vsx_stxvd2x2_le_<mode>): Likewise.
27800 2017-03-09 Michael Eager <eager@eagercon.com>
27802 Correct failures with --enable-checking=yes,rtl.
27804 * config/microblaze/microblaze.c (microblaze_expand_shift):
27805 Replace GET_CODE test with CONST_INT_P and INTVAL test with
27806 test for const0_rtx.
27807 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
27808 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
27810 2017-03-09 Richard Biener <rguenther@suse.de>
27812 PR tree-optimization/79977
27813 * graphite-scop-detection.c (scop_detection::merge_sese):
27814 Handle the case of extra exits to blocks dominating the entry.
27816 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
27818 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27821 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
27823 PR rtl-optimization/79949
27824 * lra-constraints.c (process_alt_operands): Check memory when
27825 trying to predict a cycle. Print about the overall increase.
27827 2017-03-09 Richard Biener <rguenther@suse.de>
27829 PR middle-end/79971
27830 * gimple-expr.c (useless_type_conversion_p): Preserve
27831 TYPE_SATURATING for fixed-point types.
27833 2017-03-09 Richard Biener <rguenther@suse.de>
27836 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
27837 alignment of BLKmode params.
27839 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27842 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
27843 (VALL_NO_V2Q): Likewise.
27845 * config/aarch64/aarch64-simd.md
27846 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
27848 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
27849 VALL_NO_V2Q mode iterator.
27850 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
27852 2017-03-09 Martin Liska <mliska@suse.cz>
27854 PR tree-optimization/79631
27855 * tree-chkp-opt.c (chkp_is_constant_addr): Call
27856 tree_int_cst_sign_bit just for INTEGER constants.
27858 2017-03-09 Martin Liska <mliska@suse.cz>
27862 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
27865 2017-03-09 Marek Polacek <polacek@redhat.com>
27868 * tree.c (inchash::add_expr): Handle TREE_VEC.
27870 2017-03-09 Martin Liska <mliska@suse.cz>
27873 (chkp_narrow_size_and_offset): New function.
27874 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
27875 (void chkp_parse_bit_field_ref): New function.
27876 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
27877 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
27879 2017-03-09 Martin Liska <mliska@suse.cz>
27882 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
27883 (chkp_find_bounds_1): Remove gcc_unreachable.
27885 2017-03-09 Jakub Jelinek <jakub@redhat.com>
27888 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
27889 BUILT_IN_SYNC*, determine the access type from the size suffix and
27890 always build a MEM_REF with that type. Handle forgotten
27891 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
27894 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
27895 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
27896 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
27897 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
27898 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
27899 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
27900 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
27901 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
27902 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
27903 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
27904 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
27905 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
27906 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
27907 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
27908 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
27909 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
27910 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
27911 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
27912 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
27913 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
27914 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
27915 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
27916 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
27917 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
27918 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
27919 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
27920 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
27921 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
27922 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
27923 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
27924 definitions outside of __OPTIMIZE__ guarded section.
27927 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
27928 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
27929 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
27930 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
27933 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27935 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
27936 ("vfenez<mode>"): Add missing constraints.
27938 2017-03-08 Martin Sebor <msebor@redhat.com>
27941 * config/nds32/nds32.c (nds32_option_override):
27942 Fix misspelled diagnostic.
27944 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27947 * gimplify.c (gimplify_omp_for): Replace index var in outer
27948 taskloop statement with an artificial variable and add
27949 OMP_CLAUSE_PRIVATE clause for it.
27951 2017-03-08 Richard Biener <rguenther@suse.de>
27953 PR tree-optimization/79955
27954 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
27955 for accesses that are completely outside of the variable.
27957 2017-03-08 Andrew Haley <aph@redhat.com>
27959 PR tree-optimization/79943
27960 * tree-ssa-loop-split.c (compute_new_first_bound): When
27961 calculating the new upper bound, (END-BEG) should be added, not
27964 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27966 * config/avr/avr.md (setmemhi): Make sure match_dup
27967 operand number comes before match_scratch.
27969 2017-03-08 Richard Biener <rguenther@suse.de>
27971 PR tree-optimization/79920
27972 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
27973 with ncopies == 1 to ...
27974 (vect_transform_slp_perm_load): ... here. Properly compute
27975 all element loads by iterating VF times over the group. Do
27976 not handle ncopies (computed in a broken way) in
27977 vect_create_mask_and_perm.
27979 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27982 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
27983 is a uniform vector, use uniform_vector_p return value instead of
27984 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
27986 2017-03-07 Marek Polacek <polacek@redhat.com>
27988 PR middle-end/79809
27989 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
27990 (alloca_call_type): Likewise.
27992 2017-03-07 Martin Liska <mliska@suse.cz>
27994 * gcov.c (process_args): Put comment to correct location.
27996 2017-03-07 Martin Liska <mliska@suse.cz>
27998 PR middle-end/68270
27999 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28000 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28001 (chkp_narrow_bounds_for_field): Likewise.
28002 (chkp_parse_array_and_component_ref): Pass one more argument to
28005 2017-03-07 Richard Biener <rguenther@suse.de>
28007 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28010 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28012 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28013 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28015 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28018 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28019 to end of description.
28020 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28022 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28024 PR rtl-optimization/79901
28025 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28027 (*avx512f_<code><mode>3<mask_name>): ... this.
28028 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28029 iterator instead of VI8_AVX2_AVX512BW.
28031 PR rtl-optimization/79901
28032 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28033 min/max expander, expand it using expand_vec_cond_expr.
28036 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28039 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28042 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28043 to void * for PCH reasons.
28044 * dwarf2out.c (output_loc_operands, output_die): Cast
28045 v.val_vec.array to unsigned char *.
28047 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28050 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28053 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28055 PR rtl-optimization/79571
28056 * lra-constraints.c (process_alt_operands): Calculate static
28057 reject and subtract it from overall when only addresses will be
28060 2017-03-06 Julia Koval <julia.koval@intel.com>
28063 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28064 incoming stack boundary to 128 for 64-bit targets.
28066 2017-03-06 Richard Biener <rguenther@suse.de>
28068 PR tree-optimization/79894
28069 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28070 to NULL after folding it.
28072 2017-03-06 Richard Biener <rguenther@suse.de>
28074 PR tree-optimization/79824
28075 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28076 check disabling peeling for gaps.
28078 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28080 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28081 attributes): Document gettimeofday.
28083 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28085 * config/s390/s390.c (s390_option_override_internal): Set
28086 PARAM_MIN_VECT_LOOP_BOUND
28088 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28090 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28091 * config/s390/s390.md: Likewise.
28093 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28096 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28097 (<avx2_avx512>_perm<mode>): Rename to ...
28098 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28100 (<avx512>_perm<mode>_mask): Rename to ...
28101 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28103 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28104 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28105 instead of VI8F_256_512.
28106 (avx512f_perm<mode>): New define_expand.
28107 (avx512f_perm<mode>_mask): Likewise.
28108 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28109 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28111 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28113 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28114 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28116 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28118 2017-03-06 Martin Liska <mliska@suse.cz>
28121 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28122 when having a SSA NAME w/o VAR_DECL assigned to it.
28124 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28126 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28127 msa_dpsub_<su>_d): Fix MODE for vec_select.
28129 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28131 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28133 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28135 2017-03-06 Richard Biener <rguenther@suse.de>
28137 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28138 * plugin.c (register_plugin_info): Likewise.
28139 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28141 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28143 * config/i386/sse.md (sse_storehps, sse_storelps,
28144 avx_<castmode><avxsizesuffix>_<castmode>,
28145 avx512f_<castmode><avxsizesuffix>_<castmode>,
28146 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28147 in condition that at least one operand is not a MEM.
28149 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28151 PR middle-end/79805
28152 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28153 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28155 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28156 gimple_call_nothrow_p flag based on whether original builtin can throw.
28157 If it can, emit following stmts on the fallthrough edge.
28158 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28159 don't create new bb if inserting just debug stmts on the edge, try to
28160 insert them on the fallthru bb or just reset debug stmts.
28162 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28165 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28166 restore recog_data (including the operand rtxes inside it) around
28167 the call to get_insn_template.
28169 2017-03-03 Martin Sebor <msebor@redhat.com>
28171 PR tree-optimization/79699
28172 * context.c (context::~context): Free MPFR caches to avoid
28173 a memory leak on program exit.
28175 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28177 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28178 Use wide_int::ulow () instead of .elt (0).
28180 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28182 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28183 (*pushxf): Limit oF constraint to 32bit targets and add oC
28184 constraint for 64bit targets.
28185 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28186 (*pushdf): Change rmF constraint to rmC.
28188 2017-03-03 Martin Liska <mliska@suse.cz>
28190 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28191 Remove unused variable.
28193 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28196 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28197 is a memory operand, increase num_memory.
28198 (ix86_expand_args_builtin): Likewise.
28200 2017-03-03 Jan Hubicka <jh@suse.cz>
28203 * ipa-devirt.c (maybe_record_node): Properly handle
28204 __cxa_pure_virtual visibility.
28206 2017-03-03 Martin Liska <mliska@suse.cz>
28208 PR tree-optimization/79803
28209 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28211 (pass_loop_prefetch::execute): Disabled optimization if an
28212 assumption about L1 cache size is not met.
28214 2017-03-03 Martin Liska <mliska@suse.cz>
28216 PR rtl-optimization/79574
28217 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28218 (hash_scan_set): Likewise.
28219 (dump_hash_table): Likewise.
28220 (hoist_code): Likewise.
28222 2017-03-03 Richard Biener <rguenther@suse.de>
28224 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28225 in place of uhigh ().
28226 (fixed_convert_from_real): Likewise.
28228 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28231 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28233 2017-03-03 Richard Biener <rguenther@suse.de>
28235 PR middle-end/79818
28236 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28237 TYPE_OVERFLOW_UNDEFINED check.
28239 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28241 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28243 (vector_ae_<mode>_p): Likewise.
28244 (vector_nez_<mode>_p): Likewise.
28245 (vector_ne_v2di_p): Likewise.
28246 (vector_ae_v2di_p): Likewise.
28247 (vector_ne_<mode>_p): Likewise.
28248 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28250 (vsx_tsqrt<mode>2_fe): Likewise.
28252 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
28255 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
28257 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28259 PR rtl-optimization/79780
28260 * cprop.c (one_cprop_pass): When second and further conditional trap
28261 in a single basic block is turned into an unconditional trap, turn it
28262 into a deleted note to avoid RTL verification failures.
28264 2017-03-02 Richard Biener <rguenther@suse.de>
28266 * fold-const.c (const_binop): Use ulow () instead of elt (0).
28268 2017-03-02 Richard Biener <rguenther@suse.de>
28270 PR tree-optimization/79345
28272 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
28273 param and abort the walk, returning -1 if it is hit.
28274 (walk_aliased_vdefs): Take a limit param and pass it on.
28275 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
28276 defaulting to 0 and return a signed int.
28277 * tree-ssa-uninit.c (struct check_defs_data): New struct.
28278 (check_defs): New helper.
28279 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
28280 about uninitialized memory.
28281 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
28282 bogus uninitialized warning.
28283 (fixed_convert_from_real): Likewise.
28285 2017-03-02 Bin Cheng <bin.cheng@arm.com>
28287 PR tree-optimization/66768
28288 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
28289 iv_use if base object can't be determined.
28291 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28293 PR tree-optimization/79345
28294 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
28295 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
28296 (get_pattern_stats): Initialize it.
28297 * genemit.c (gen_expand): Verify match_scratch numbers come after
28298 match_operand/match_dup numbers.
28299 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
28300 match_scratch numbers.
28301 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
28303 * config/s390/s390.md (trunctdsd2): Likewise.
28305 2017-03-02 Richard Biener <rguenther@suse.de>
28307 * wide-int.h (wide_int_storage::operator=): Implement in terms
28310 2017-03-02 Richard Biener <rguenther@suse.de>
28312 PR tree-optimization/79777
28313 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
28314 the to insert expression to sth existing.
28316 2017-03-01 Martin Sebor <msebor@redhat.com>
28318 PR middle-end/79692
28319 * gimple-ssa-sprintf.c
28320 (directive::known_width_and_precision): New function.
28321 (format_integer): Use it.
28322 (get_mpfr_format_length): Consider the full range of precision
28323 when computing %g output with the # flag. Set the likely byte
28324 count to 3 rather than 1 when precision is indeterminate.
28325 (format_floating): Correct the lower bound of precision.
28327 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28329 * doc/invoke.texi: Document default code model for 64-bit Linux.
28331 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28334 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
28335 udiv rather than div since input pattern is unsigned.
28337 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
28339 * config/i386/i386.c (print_reg): Warn for values of
28340 unsupported size in integer register.
28342 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
28345 * config/rs6000/predicates.md (current_file_function_operand): Do
28346 not allow self calls to be local if the function is replaceable.
28348 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
28351 * config/rs6000/altivec.h (vec_ctz and others): Change the
28352 preprocessor macro that controls conditional compilation from
28353 _ARCH_PWR9 to __POWER9_VECTOR__.
28354 (vec_all_ne): Change parameterization of __altivec_scalar_pred
28355 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
28356 control (instead of _ARCH_PWR9 control) so that template
28357 definition uses power9-specific function.
28358 (vec_any_eq): Likewise.
28359 (vec_all_ne): Change macro definition to use a power9-specific
28360 expansion under #ifdef __POWER9_VECTOR__ control (instead of
28361 _ARCH_PWR9 control).
28362 (vec_any_eq) Likewise.
28363 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
28364 expansion for CMPNEF to remove support for xvcmpnesp instruction.
28365 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
28366 support for xvcmpnedp instruction.
28367 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
28368 macro expansion so that Power9 implementation of vec_all_ne does
28369 not use the AltiVec predicate framework.
28370 (VCMPNEH_P): Likewise.
28371 (VCMPNEW_P): Likewise.
28372 (VCMPNED_P): Likewise.
28373 (VCMPNEFP_P): Likewise.
28374 (VCMPNEDP_P): Likewise.
28375 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
28376 implementation of vec_any_eq to not use AltiVec predicate
28378 (VCMPAEH_P): Likewise.
28379 (VCMPAEW_P): Likewise.
28380 (VCMPAED_P): Likewise.
28381 (VCMPAEFP_P): Likewise.
28382 (VCMPAEDP_P): Likewise.
28383 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
28384 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
28385 not use the AltiVec predicate framework.
28386 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
28387 of vec_any_eq to not use AltiVec predicate framework.
28388 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28389 support for predefined __POWER9_VECTOR__ macro to indicate that
28390 Power9 instruction selection is enabled.
28391 (altivec_overloaded_builtins): Remove extraneous
28392 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
28393 function argument types RS6000_BTI_bool_V16QI and
28394 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
28395 entry for overloaded function argument types RS6000_BTI_bool_V4SI
28396 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
28397 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
28398 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
28399 Power9 for implementations of vec_cmpne. Change the signature for
28400 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
28401 (representing vec_all_ne) to remove the previously described first
28402 argument of type RS6000_BTI_INTSI, as this was an artifact of
28403 reliance on the AltiVec predicate framework, which is no longer
28404 used in the implementation of these functions. Add
28405 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
28406 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
28407 since, unlike the AltiVec predicate framework implementation, we
28408 do not share function descriptors between vec_alle and vec_anyeq.
28409 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
28410 set of modes that receive special treatment even when
28411 TARGET_P9_VECTOR is true. The special treatment emits code that
28412 does not depend on Power9 instructions.
28413 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
28414 define_expand to not rely on AltiVec predicate framework.
28415 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28417 (vector_ne_v2di_p): Change this define_expand to not rely on
28418 AltiVec predicate framework.
28419 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
28421 (vector_ne_<mode>_p): Change this define_expand to not rely on
28422 AltiVec predicate framework.
28423 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28425 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
28426 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
28427 define_insn pattern.
28428 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
28429 define_insn pattern because the xvcmpne<VSs>. instruction is not
28431 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
28432 instruction is not supported.
28434 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28436 * config/nvptx/nvptx.c: Include intl.h.
28438 2017-03-01 Martin Jambor <mjambor@suse.cz>
28441 * ipa-prop.h (ipa_bits): Removed field known.
28442 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
28443 to pointers. Adjusted their comments to warn about their sharing.
28444 (ipcp_transformation_summary): Change bits to a vector of pointers.
28445 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
28446 (ipa_get_ipa_bits_for_value): Declare.
28447 * tree-vrp.h (value_range): Mark as GTY((for_user)).
28448 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
28449 (ipa_bits_hash_table): Likewise.
28450 (ipa_vr_ggc_hash_traits): Likewise.
28451 (ipa_vr_hash_table): Likewise.
28452 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
28453 being pointers and vr_known being removed.
28454 (ipa_set_jf_unknown): Likewise.
28455 (ipa_get_ipa_bits_for_value): New function.
28456 (ipa_set_jfunc_bits): Likewise.
28457 (ipa_get_value_range): New overloaded functions.
28458 (ipa_set_jfunc_vr): Likewise.
28459 (ipa_compute_jump_functions_for_edge): Use the above functions to
28460 construct bits and vr parts of jump functions.
28461 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
28462 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28464 (ipcp_grow_transformations_if_necessary): Also allocate
28465 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28467 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
28468 them. Fix too long lines.
28469 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
28470 vr_known being removed.
28471 (ipa_read_jump_function): Use new setter functions to construct bits
28472 and vr parts of jump functions or set them to NULL.
28473 (write_ipcp_transformation_info): Adjust for bits being pointers.
28474 (read_ipcp_transformation_info): Likewise.
28475 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
28477 Include gt-ipa-prop.h.
28478 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
28480 (ipcp_store_bits_results): Likewise.
28481 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
28482 Do not write to existing jump functions but use a temporary instead.
28484 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28487 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
28488 attempt to use its first operand as BIT_FIELD_REF base.
28490 2017-03-01 Richard Biener <rguenther@suse.de>
28492 PR middle-end/79721
28493 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
28494 interpolating formula in wrapping arithmetic.
28495 (chrec_apply): Convert chrec_evaluate return value to wanted type.
28497 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28499 PR tree-optimization/79734
28500 * tree-vect-generic.c (expand_vector_condition): Optimize
28501 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
28502 Handle VEC_COND_EXPR where comparison has different inner width from
28503 type's inner width.
28505 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
28507 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
28508 markup, and similar issues. Remove @opindex entries for things
28509 that aren't options. Add missing -mmpy-option entries.
28511 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28513 PR tree-optimization/79737
28514 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
28515 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
28516 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
28517 instead of byte_size. Formatting fix.
28518 (shift_bytes_in_array_right): Formatting fix.
28520 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
28523 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
28524 condition on optimize for the leaf function test.
28526 2017-02-28 Martin Liska <mliska@suse.cz>
28529 * read-rtl-function.c (function_reader::handle_unknown_directive):
28530 Bail out when one uses -flto.
28532 2017-02-28 Martin Liska <mliska@suse.cz>
28534 * common.opt: Replace space with tabular for options of <number>
28536 * config/i386/i386.opt: Show <number> value for
28537 -mlarge-data-threshold.
28538 * opts.c (print_filtered_help): Do not display number in hexadecimal
28541 2017-02-28 Martin Liska <mliska@suse.cz>
28543 * common.opt: Fix --help=option -Q for options which are of
28546 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
28548 * config/i386/i386.c (print_reg): Error out for values
28549 of 8-bit size in invalid integer register.
28551 2017-02-28 Martin Sebor <msebor@redhat.com>
28553 PR tree-optimization/79691
28554 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
28556 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28559 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
28560 gcc_unreachable with output_operand_lossage.
28562 2017-02-28 Richard Biener <rguenther@suse.de>
28564 PR tree-optimization/79740
28565 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
28567 (visit_nary_op): Insert the nary into the hashtable if we
28568 pattern-matched sth.
28569 * tree-ssa-pre.c (eliminate_insert): Robustify.
28571 2017-02-28 Richard Biener <rguenther@suse.de>
28573 PR middle-end/79731
28574 * fold-const.c (decode_field_reference): Reject out-of-bound
28577 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28579 * config/i386/i386.c: Include intl.h.
28580 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
28581 instead of just cond ? "..." : "...".
28582 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
28583 * coverage.c (read_counts_file): Likewise.
28584 * omp-offload.c: Include intl.h.
28585 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
28586 of just cond ? "..." : "...".
28587 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
28588 of just cond ? "..." : "...".
28590 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
28593 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
28595 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
28596 'tune for' CPU name.
28597 * config/arm/arm-cpu-data.h: Regenerated.
28599 2017-02-28 Richard Biener <rguenther@suse.de>
28601 PR tree-optimization/79732
28602 * tree-inline.c (expand_call_inline): Do not shadow var.
28604 2017-02-28 Richard Biener <rguenther@suse.de>
28606 PR tree-optimization/79723
28607 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
28608 address-space properly.
28610 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
28612 * doc/optinfo.texi (Optimization groups): Fix option used for
28614 * doc/invoke.texi (-fopt-info): Document "omp".
28615 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
28616 (OPTGROUP_ALL): Add OPTGROUP_OMP.
28617 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
28618 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
28619 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
28621 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
28623 * dumpfile.c (optgroup_options): Instead of "openmp", associate
28624 OPTGROUP_OMP with "omp".
28626 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
28629 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
28630 for arithmetic shift of unsigned V2DI.
28632 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
28634 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
28635 arc/linux.h headers.
28636 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
28637 (LINK_SPEC): Likewise.
28638 (ARC_TLS_EXTRA_START_SPEC): Likewise.
28639 (EXTRA_SPECS): Likewise.
28640 (STARTFILE_SPEC): Likewise.
28641 (ENDFILE_SPEC): Likewise.
28642 (LIB_SPEC): Likewise.
28643 (TARGET_SDATA_DEFAULT): Likewise.
28644 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
28645 (MULTILIB_DEFAULTS): Likewise.
28646 (DWARF2_UNWIND_INFO): Likewise.
28647 * config/arc/big.h: New file.
28648 * config/arc/elf.h: Likewise.
28649 * config/arc/linux.h: Likewise.
28650 * config/arc/t-uClibc: Remove.
28652 2017-02-27 Bin Cheng <bin.cheng@arm.com>
28654 PR tree-optimization/77536
28655 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
28656 (tree_transform_and_unroll_loop): Use above function to compute the
28657 estimated niter of unrolled loop and use it when scaling profile.
28658 Also use count info rather than frequency if it's non-zero.
28659 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
28660 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
28661 (vect_transform_loop): Call above function.
28663 2017-02-27 Richard Biener <rguenther@suse.de>
28665 PR tree-optimization/45397
28666 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
28667 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
28668 (visit_nary_op): Add pattern matching for CSEing sign-changed
28669 or truncated operations with wider ones.
28671 2017-02-27 Richard Biener <rguenther@suse.de>
28673 PR tree-optimization/79690
28674 * tree-vect-stmts.c (vectorizable_store): Use vector type
28675 built from the DR with address-space.
28677 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
28679 * doc/invoke.texi (Optimize Options): Refine the description
28680 of asan-use-after-return.
28682 2017-02-25 Alan Modra <amodra@gmail.com>
28684 PR rtl-optimization/79584
28685 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
28686 base, not ad->base_term, the reg within base. Remove assertion
28687 that ad->base == ad->base_term. Replace gen_int_mode using
28688 bogus mode with const0_rtx.
28690 2017-02-25 Jakub Jelinek <jakub@redhat.com>
28692 PR middle-end/79396
28693 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
28694 FMA_EXPR like tcc_binary or tcc_unary.
28696 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
28699 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
28701 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
28702 (output_loc_operands): Handle DW_OP_call_ref and
28703 DW_OP_GNU_variable_value.
28704 (struct variable_value_struct): New type.
28705 (struct variable_value_hasher): Likewise.
28706 (variable_value_hash): New variable.
28707 (string_types): Remove.
28708 (copy_loc_descr): New function.
28709 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
28710 (prepend_loc_descr_to_each): New function.
28711 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
28712 instead of add_loc_descr_to_each if the first argument is single
28713 location list and the second has multiple.
28714 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
28715 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
28716 when looking for variable value which doesn't have other location info.
28717 (loc_list_from_tree): Formatting fix.
28718 (gen_array_type_die): Simplify DW_AT_string_length handling.
28719 (adjust_string_types): Remove.
28720 (gen_subprogram_die): Don't call adjust_string_types nor test/set
28721 string_types. Call resolve_variable_values.
28722 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
28723 (resolve_addr_in_expr): Likewise. Add A argument.
28724 (copy_deref_exprloc): Remove deref argument. Adjust for the
28725 original expression being DW_OP_GNU_variable_value with optionally
28726 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
28727 optionally after it.
28728 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
28729 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
28730 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
28731 (variable_value_hasher::hash, variable_value_hasher::equal): New
28733 (resolve_variable_value_in_expr, resolve_variable_value,
28734 resolve_variable_values, note_variable_value_in_expr,
28735 note_variable_value): New functions.
28736 (dwarf2out_early_finish): Call note_variable_value on all toplevel
28739 2017-02-24 Jakub Jelinek <jakub@redhat.com>
28742 * opts.h (handle_generated_option): Add GENERATED_P argument.
28743 * opts-common.c (handle_option): Adjust function comment.
28744 (handle_generated_option): Add GENERATED_P argument, pass it to
28746 (control_warning_option): Pass false to handle_generated_option
28748 * opts.c (maybe_default_option): Pass true to handle_generated_option
28750 * optc-gen.awk: Likewise.
28752 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28754 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
28755 a REG, look at the REG it is a SUBREG of.
28756 (splitter for cmpeqsi_t): Ditto.
28758 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28760 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
28761 the special USEs with the pattern of the insn, not the insn itself.
28763 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
28766 * doc/invoke.texi: Document -mload-store-pairs.
28768 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28769 Sandra Loosemore <sandra@codesourcery.com>
28771 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
28772 argument isn't a CONST_INT.
28773 (nios2_alternate_compare_const): Assert op is a CONST_INT.
28774 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
28775 (nios2_validate_compare): Bypass alternate compare logic if *op2
28776 is not a CONST_INT.
28777 (ldstwm_operation_p): Return false if first_base is not a REG or
28778 if first_offset is not a CONST_INT.
28780 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28782 * config/cris/cris.md: Use correct operand in a define_peephole2.
28784 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28786 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
28788 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28790 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
28791 this_insn if it is an INSN or JUMP_INSN.
28792 (force_offsettable): Look at base, not at addr.
28793 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
28794 on things that aren't necessarily CONST_INTs.
28796 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
28798 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
28799 -mfpmath=sse is the default also for x86-32 targets with SSE2
28800 instruction set when @option{-ffast-math} is enabled
28802 2017-02-24 Jeff Law <law@redhat.com>
28804 PR rtl-optimizatoin/79286
28805 * ira.c (update_equiv_regs): Drop may_trap_p exception to
28808 2017-02-24 Richard Biener <rguenther@suse.de>
28810 PR tree-optimization/79389
28811 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
28814 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
28816 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
28817 function comment to reflect reality.
28818 (loop_exits_before_overflow): Fix typo in function description.
28820 2017-02-24 Richard Biener <rguenther@suse.de>
28822 PR tree-optimization/79389
28823 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
28824 properly that a threading opportunity exists. Detect conditional
28825 copy/constant propagation opportunities.
28827 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
28829 * config/visium/visium.md (type): Add trap.
28830 (b): New mode attribute.
28831 (*btst): Rename into...
28832 (*btst<mode>): ...this and adjust.
28833 (*cbranchsi4_btst_insn): Rename into...
28834 (*cbranch<mode>4_btst_insn): ...this and adjust.
28835 (trap): New define_insn.
28837 2017-02-23 Jakub Jelinek <jakub@redhat.com>
28839 PR tree-optimization/79389
28840 * ifcvt.c (struct noce_if_info): Add rev_cond field.
28841 (noce_reversed_cond_code): New function.
28842 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
28843 reversed_comparison_code. Formatting fix.
28844 (noce_try_store_flag): Test rev_cond != NULL in addition to
28845 reversed_comparison_code.
28846 (noce_try_store_flag_constants): Likewise.
28847 (noce_try_store_flag_mask): Likewise.
28848 (noce_try_addcc): Use rev_cond if non-NULL instead of
28849 reversed_comparison_code.
28850 (noce_try_cmove_arith): Likewise. Formatting fixes.
28851 (noce_try_minmax, noce_try_abs): Clear rev_cond.
28852 (noce_find_if_block): Initialize rev_cond.
28853 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
28854 instead of false as last argument never attempt to reverse it
28857 2017-02-23 Bin Cheng <bin.cheng@arm.com>
28859 PR tree-optimization/79663
28860 * tree-predcom.c (combine_chains): Process refs in reverse order
28861 only for ZERO length chains, and add explaining comment.
28863 2017-02-23 Jeff Law <law@redhat.com>
28865 PR tree-optimization/79578
28866 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
28867 in call to operand_equal_p.
28869 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
28872 * config/i386/cpuid.h: Fix another undefined behavior.
28874 2017-02-23 Richard Biener <rguenther@suse.de>
28876 PR tree-optimization/79683
28877 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
28878 vector types for data-refs.
28880 2017-02-23 Martin Liska <mliska@suse.cz>
28882 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
28884 2017-02-23 Jakub Jelinek <jakub@redhat.com>
28886 PR middle-end/79665
28887 * internal-fn.c (get_range_pos_neg): Moved to ...
28888 * tree.c (get_range_pos_neg): ... here. No longer static.
28889 * tree.h (get_range_pos_neg): New prototype.
28890 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
28891 are known to be in between 0 and signed maximum inclusive, try to
28892 expand both unsigned and signed divmod and use the cheaper one from
28895 2017-02-22 Jeff Law <law@redhat.com>
28897 PR tree-optimization/79578
28898 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
28899 to compare base operands.
28901 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
28904 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
28905 gpc_reg_operand instead of fpr_reg_operand.
28907 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
28909 * config/mips/mips.c (mips_return_in_memory): Force FP
28910 vector types to be returned in memory for o32 ABI.
28912 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28914 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
28915 instead of DW_TAG_member for static data member declarations and don't
28916 set no_linkage_name for static inline data members.
28917 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
28920 2017-02-22 Martin Liska <mliska@suse.cz>
28922 * doc/invoke.texi: Replace inequality signs with square brackets
28925 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
28928 * lra-constraints.c (simplify_operand_subreg): Handle
28929 WORD_REGISTER_OPERATIONS targets.
28931 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28934 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
28935 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
28936 elimination by swapping fld*.
28938 2017-02-22 Richard Biener <rguenther@suse.de>
28940 PR tree-optimization/79673
28941 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
28942 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
28943 irrelevant address-space qualifiers and avoiding a
28944 ADDR_SPACE_CONVERT_EXPR from fold_convert.
28946 2017-02-22 Richard Biener <rguenther@suse.de>
28948 PR tree-optimization/79666
28949 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
28950 to not symbolically negate if that may introduce undefined
28953 2017-02-22 Martin Liska <mliska@suse.cz>
28956 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
28957 * data-streamer-out.c (streamer_write_gcov_count_stream):
28959 * value-prof.c (stream_out_histogram_value): Make assert more
28960 precise based on type of counter.
28962 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
28965 * config/i386/i386.md (standard_x87sse_constant_load splitter):
28966 Use nonimmediate_operand instead of memory_operand for operand 1.
28967 (float-extend standard_x87sse_constant_load splitter): Ditto.
28969 2017-02-21 Jeff Law <law@redhat.com>
28971 PR tree-optimization/79621
28972 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
28973 blocks with edges to themselves.
28975 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28978 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
28979 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
28980 Use gimple_call_builtin_p.
28983 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
28984 on temporarily removed DEBUG_INSNs.
28986 PR tree-optimization/79649
28987 * tree-loop-distribution.c (classify_partition): Give up on
28988 non-generic address space loads/stores.
28990 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
28992 * doc/loop.texi (Loop manipulation): Remove nonexistent
28993 tree_ssa_loop_version from the documentation.
28994 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
28996 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28999 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29000 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29001 * config/rs6000/rs6000.c: Include except.h.
29002 (rs6000_expand_split_stack_prologue): Call
29003 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29005 2017-02-21 Martin Jambor <mjambor@suse.cz>
29008 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29009 have been analyzed.
29011 2017-02-21 Martin Jambor <mjambor@suse.cz>
29013 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29014 for backward compatibility only.
29015 * doc/invoke.texi (Option Summary): Remove all references to
29016 -fipa-cp-alignment.
29018 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29022 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29024 * lra-constraints.c (curr_insn_transform): Handle
29025 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29027 2017-02-21 Martin Liska <mliska@suse.cz>
29029 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29031 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29033 PR translation/79638
29034 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29036 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29039 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29040 (arm_function_ok_for_sibcall): Return false for an indirect call by
29041 descriptor if all the argument registers are used.
29042 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29043 alignment of the function.
29045 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29047 PR tree-optimization/61441
29048 * simplify-rtx.c (simplify_const_unary_operation): For
29049 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29050 the sNaN unmodified.
29052 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29054 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29055 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29056 instead of SYSTEM_HEADER_DIR.
29058 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29059 Martin Liška <mliska@suse.cz>
29061 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29062 Fix typos and grammar, use active voice, and clarify.
29064 2017-02-20 Marek Polacek <polacek@redhat.com>
29066 PR middle-end/79537
29067 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29070 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29072 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29075 * config/i386/i386.c (ix86_expand_builtin): Handle
29076 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29077 ix86_builtins_isa[fcode].isa as a requirement of those
29078 flags and any other flag in the bitmask.
29079 (ix86_init_mmx_sse_builtins): Use 0 instead of
29080 ~OPTION_MASK_ISA_64BIT as mask.
29081 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29082 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29083 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29084 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29086 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29089 * lra-constraints.c (split_reg): Check requested split mode
29090 is supported by the register.
29092 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29094 * lra-constraints.c (simplify_operand_subreg): Remove early
29097 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29100 * lra-constraints.c (curr_insn_transform): Tighten condition
29101 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29103 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29106 * lra-constraints.c (curr_insn_transform): Handle
29107 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29109 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29112 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29114 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29116 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29119 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29122 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29124 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29127 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29129 * final.c (last_columnnum, override_columnnum): New variables.
29130 (final_start_function): Set last_columnnum, pass it to begin_prologue
29131 hook and pass 0 to dwarf2out_begin_prologue.
29132 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29133 to source_line debug hook.
29134 (notice_source_line): Compute last_columnnum and for debug_column_info
29135 return true on column changes.
29136 * debug.h (struct gcc_debug_hooks): Add column argument to
29137 source_line and begin_prologue hooks.
29138 (debug_nothing_int_charstar_int_bool): Remove prototype.
29139 (debug_nothing_int_int_charstar,
29140 debug_nothing_int_int_charstar_int_bool): New prototypes.
29141 (dwarf2out_begin_prologue): Add column argument.
29142 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29143 begin_prologue hooks.
29144 (debug_nothing_int_charstar_int_bool): Remove.
29145 (debug_nothing_int_int_charstar,
29146 debug_nothing_int_int_charstar_int_bool): New functions.
29147 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29148 through to dwarf2out_source_line.
29149 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29150 (dwarf2out_source_line): Add column argument, emit it if requested.
29151 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29153 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29154 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29155 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29156 through to dwarf2out_begin_prologue.
29157 (vmsdbgout_source_line): Add column argument, pass it through to
29158 dwarf2out_source_line.
29159 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29160 dbxout_source_line caller.
29161 (dbxout_source_line): Add column argument.
29163 * common.opt (gno-column-info, gcolumn-info): New options.
29164 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29165 (check_die): Also test for multiple DW_AT_decl_column attributes.
29166 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29167 DW_AT_decl_column if requested.
29168 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29170 (gen_variable_die): Likewise.
29171 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29172 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29175 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29176 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29177 (ix86_handle_option): Handle OPT_m3dnowa.
29178 * doc/invoke.texi (-m3dnowa): Document.
29179 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29180 -m3dnowa instead of -m3dnow -march=athlon.
29183 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29184 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29186 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29189 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29190 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29191 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29192 generator for vsx_xxpermdi_<mode>_be.
29193 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29194 force big-endian semantics.
29195 (vsx_xxpermdi_<mode>_be): New define_expand with same
29196 implementation as previous version of vsx_xxpermdi_<mode>.
29198 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29200 PR tree-optimization/79327
29201 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29202 variable, its initialization and use.
29204 2017-02-17 Julia Koval <julia.koval@intel.com>
29206 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29207 (OPTION_MASK_ISA_PKU_UNSET): New.
29208 (ix86_handle_option): Handle -mrdpid.
29209 * config/i386/cpuid.h (bit_RDPID): New.
29210 * config/i386/driver-i386.c (host_detect_local_cpu):
29211 Detect RDPID feature.
29212 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29213 * config/i386/i386-c.c (ix86_target_macros_internal):
29215 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29216 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29217 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29218 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29219 * config/i386/i386.md (define_insn "rdpid"): New.
29220 * config/i386/i386.opt Add -mrdpid.
29221 * config/i386/immintrin.h (_rdpid_u32): New.
29223 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29225 PR rtl-optimization/79541
29226 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29227 instead of transforming it into USE.
29229 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29231 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29232 If HONOR_SNANS (SFmode) force the input to a register.
29233 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29234 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29235 an frsp or similar insn.
29237 2017-02-17 Martin Liska <mliska@suse.cz>
29239 PR rtl-optimization/79577
29240 * params.def (selsched-max-sched-times): Increase minimum to 1.
29242 2017-02-17 Martin Liska <mliska@suse.cz>
29244 PR rtl-optimization/79574
29245 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29247 2017-02-17 Martin Liska <mliska@suse.cz>
29249 PR tree-optimization/79529
29250 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
29251 ssa_defined_default_def_p to handle cases which are implicitly
29253 * tree-ssa.c (ssa_defined_default_def_p): New function.
29254 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
29255 which are implicitly defined.
29256 * tree-ssa.h (ssa_defined_default_def_p): Declare.
29258 2017-02-17 Richard Biener <rguenther@suse.de>
29260 PR middle-end/79576
29261 * params.def (max-ssa-name-query-depth): Limit to 10.
29263 2017-02-17 Richard Biener <rguenther@suse.de>
29265 PR tree-optimization/79552
29266 * tree-ssa-structalias.c (visit_loadstore): Properly verify
29269 2017-02-17 Richard Biener <rguenther@suse.de>
29272 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
29274 2017-02-17 Marek Polacek <polacek@redhat.com>
29276 PR middle-end/79536
29277 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
29278 (fold_negate_expr): New wrapper.
29280 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
29282 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
29283 Correct terminology and de-emphasize pre-standard behavior.
29285 2017-02-16 Alan Modra <amodra@gmail.com>
29287 PR rtl-optimization/79286
29288 * ira.c (def_dominates_uses): New function.
29289 (update_equiv_regs): Don't create an equivalence for insns that
29290 may trap where the register def does not dominate the use.
29292 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
29294 PR rtl-optimization/78127
29295 * lra.c (lra): Call lra_eliminate before finish the loop after
29298 2017-02-16 Richard Biener <rguenther@suse.de>
29300 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
29302 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
29303 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
29304 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
29305 (isl_val_int_from_wi): New function.
29306 (extract_affine_gmp): Rename to ...
29307 (extract_affine_wi): ... this, take a widest_int.
29308 (extract_affine_int): Just wrap extract_affine_wi.
29309 (add_param_constraints): Use isl_val_int_from_wi.
29310 (add_loop_constraints): Likewise, and extract_affine_wi.
29312 2017-02-15 Jeff Law <law@redhat.com>
29314 PR middle-end/79521
29315 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
29316 ira_init_register_move_cost_if_necessary.
29318 2017-02-15 Martin Sebor <msebor@redhat.com>
29320 PR middle-end/32003
29321 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
29322 removed in a prior commit.
29324 2017-02-15 Bin Cheng <bin.cheng@arm.com>
29326 PR tree-optimization/79347
29327 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
29328 counters during peeling.
29330 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
29332 * Makefile.in (site.exp): Remove "set ISLVER".
29334 2017-02-15 Jakub Jelinek <jakub@redhat.com>
29337 * real.c (real_from_integer): Call real_convert even for decimal.
29339 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
29342 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
29344 2017-02-14 Andrew Pinski <apinski@cavium.com>
29346 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
29347 cores and change the partno/implementer to be correct.
29348 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
29349 the 'B" as the implementer.
29350 * config/aarch64/aarch64-tune.md: Regenerate.
29352 2017-02-14 Carl Love <cel@us.ibm.com>
29354 * config/rs6000/rs6000.c: Add case statement entry to make the
29355 xvcvuxdsp built-in argument unsigned.
29356 * config/rs6000/vsx.md: Fix the source and return operand types so they
29357 match the instruction definitions from the ISA document. Fix typo
29358 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
29361 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
29364 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
29365 member early_clobber_alts.
29366 * lra-lives.c (reg_early_clobber_p): New.
29367 (process_bb_lives): Use it.
29368 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
29369 (debug_operand_data): Initialize early_clobber_alts.
29370 (setup_operand_alternative): Set up early_clobber_alts.
29371 (collect_non_operand_hard_regs): Ditto. Pass early clobber
29372 alternatives to new_insn_reg.
29373 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
29375 (lra_update_insn_regno_info): Pass the new arg.
29377 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29379 PR middle-end/79505
29380 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
29381 (new_oacc_loop_raw): Don't clear already cleared fields.
29384 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
29385 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
29386 _mm512_prefetch_i64gather_ps): New inline functions and macros.
29388 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
29391 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
29393 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
29396 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
29397 the extra instruction to the right place to store 128-bit constant
29400 2017-02-14 Martin Sebor <msebor@redhat.com>
29402 PR middle-end/79448
29403 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
29404 warning for strings of unknown length.
29406 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
29408 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
29410 2017-02-14 Jeff Law <law@redhat.com>
29413 * ira-costs.c (scan_one_insn): Initialize register move costs
29414 for pseudos seen in USE/CLOBBER insns.
29416 PR tree-optimization/79095
29417 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
29418 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
29419 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
29420 if the operands are known to be not equal, then the resulting range
29422 (intersect_ranges): If the new range is ~[0,0] and the old range is
29423 wide, then prefer ~[0,0].
29424 * tree-vrp.c (overflow_comparison_p_1): New function.
29425 (overflow_comparison_p): New function.
29426 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
29427 if NAME is used in an overflow test.
29428 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
29429 overflow check that can be expressed as an equality test, then adjust
29430 ops to be that equality test.
29432 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29434 * config/s390/s390-builtin-types.def: Remove flags argument.
29435 * config/s390/s390.c (s390_init_builtins): Likewise.
29437 2017-02-14 Martin Liska <mliska@suse.cz>
29439 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
29440 vector. Fix trailing white spaces.
29442 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
29444 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
29447 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29449 PR rtl-optimization/68664
29450 * config/arm/arm.c (arm_sched_can_speculate_insn):
29451 New function. Declare prototype.
29452 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29454 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29456 PR rtl-optimization/68664
29457 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
29459 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29461 2017-02-14 Amit Pawar <amit.pawar@amd.com>
29463 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
29464 max skip bytes for function, loop and jump.
29466 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29468 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
29469 ABS_EXPR for gimple dump.
29471 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29474 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
29476 PR tree-optimization/79408
29477 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
29478 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
29479 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
29480 also if rhs1 is INTEGER_CST.
29482 2017-02-14 Richard Biener <rguenther@suse.de>
29484 PR middle-end/79432
29485 * tree-into-ssa.c (insert_phi_nodes): When the function can
29486 have abnormal edges rewrite SSA names with broken use-def
29487 dominance out of SSA and register them for PHI insertion.
29489 2017-02-13 Martin Sebor <msebor@redhat.com>
29491 PR middle-end/79496
29492 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
29493 clearing info.nowrite flag when snprintf size argument is a range.
29495 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29497 * cprop.c (cprop_jump): Add missing space in string literal.
29498 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
29499 (get_constraint_for_component_ref): Likewise.
29500 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
29501 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
29502 * lra-constraints.c (process_alt_operands): Likewise.
29503 * ipa-inline.c (inline_small_functions): Likewise.
29504 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
29505 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29506 * trans-mem.c (diagnose_tm_1_op): Likewise.
29507 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
29508 (grid_parallel_clauses_gridifiable): Likewise.
29510 * config/nvptx/mkoffload.c (process): Add space in between
29513 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
29514 "MOD4_SSE_REGS" and "ALL_REGS".
29516 * spellcheck.c (test_data): Add , in between "foo" and "food".
29518 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29521 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
29522 boundary crossing check and subsequent code generation agree.
29524 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29526 * config/aarch64/aarch64.c (has_memory_op): Delete.
29527 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
29530 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29532 PR rtl-optimization/79388
29533 PR rtl-optimization/79450
29534 * combine.c (distribute_notes): When removing TEM_INSN for which
29535 corresponding dest has last value recorded, invalidate that last
29538 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29540 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
29541 of explicit '@'. Add missing assembly comment marker on branch costs
29544 2017-02-13 Nathan Sidwell <nathan@acm.org>
29546 * gengtype-lex.l (<in_struct>): Add '/'.
29548 2017-02-13 Martin Liska <mliska@suse.cz>
29551 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
29553 2017-02-13 Richard Biener <rguenther@suse.de>
29555 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
29557 * configure: Re-generate.
29558 * config.in: Likewise.
29559 * graphite-dependences.c: Simplify as if
29560 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
29561 * graphite-isl-ast-to-gimple.c: Likewise.
29562 * graphite-optimize-isl.c: Likewise.
29563 * graphite-poly.c: Likewise.
29564 * graphite-sese-to-poly.c: Likewise.
29565 * graphite.h: Likewise.
29566 * toplev.c: Include isl/version.h and use isl_version () for
29567 printing the ISL version.
29568 * doc/install.texi: Update ISL requirement.
29570 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29572 * doc/standards.texi (Standards): Update reference to
29575 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29577 * doc/extend.texi (Named Address Spaces): sourceware.org now
29579 * doc/install.texi (Binaries): Ditto.
29582 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29584 * doc/cpp.texi: Replace "stringify"/"stringification" with C
29585 standard terminology "stringize"/"stringizing" throughout.
29586 * doc/cppinternals.texi: Likewise.
29588 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29590 * doc/extend.texi: Fix some spelling mistakes and typos.
29591 * doc/invoke.texi: Likewise.
29593 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29596 * params.def (inline-min-speedup) Change from 10 to 8.
29598 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29600 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
29603 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29606 * ipa-inline-analysis.c (get_minimal_bb): New function.
29607 (record_modified): Use it.
29608 (remap_edge_change_prob): Handle also ancestor functions.
29610 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
29612 * doc/contrib.texi (Contributors): Remove broken link into
29613 the Mauve CVS repository.
29615 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29617 PR middle-end/79454
29618 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
29619 result computation whenever lhs doesn't have vector mode, not
29620 just when it has BLKmode.
29622 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
29624 * doc/makefile.texi (profiledbootstrap): Refer to the
29625 installation instructions only in textual form.
29627 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29630 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
29632 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
29634 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
29635 (Specific): Update mingw-w64 reference.
29637 (Specific): Remove broken link to Renesas RX processor.
29639 2017-02-10 Richard Biener <rguenther@suse.de>
29641 * toplev.c (process_options): Do not mention obsolete graphite
29642 options when printing sorry message about missing graphite support.
29643 Mention -floop-nest-optimize.
29645 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
29647 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
29648 (vtst_p16): Likewise.
29649 (vtstq_p8): Likewise.
29650 (vtstq_p16): Likewise.
29652 (vtstq_p64): Likewise.
29653 * config/arm/arm_neon.h (vgetq_lane_p64): New.
29654 (vset_lane_p64): New.
29655 (vsetq_lane_p64): New.
29657 2017-02-10 Jakub Jelinek <jakub@redhat.com>
29659 PR tree-optimization/79411
29660 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
29661 stmt operands are SSA_NAMEs used in abnormal phis.
29662 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
29665 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
29668 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
29671 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
29673 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
29675 2017-02-09 Jakub Jelinek <jakub@redhat.com>
29677 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
29681 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
29682 not arbitrary TREE_CONSTANT.
29685 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
29686 "omp declare target link" attribute unless is_global_var.
29687 * omp-offload.c (find_link_var_op): Likewise.
29689 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
29690 Chung-Lin Tang <cltang@codesourcery.com>
29692 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
29694 (gimplify_adjust_omp_clauses): Don't delete TILE.
29695 (gimplify_omp_for): Deal with TILE.
29696 * internal-fn.c (expand_GOACC_TILE): New function.
29697 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
29699 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
29700 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
29702 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
29703 avoid DIV for outermost collapse var.
29704 (expand_oacc_for): Insert tile element loop as needed. Adjust.
29705 Remove out of date comments, fix whitespace.
29706 * omp-general.c (omp_extract_for_data): Deal with tiling.
29707 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
29708 adjust OLF_DIM_BASE value.
29709 (struct omp_for_data): Add tiling field.
29710 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
29711 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
29712 for auto loops. Remove default auto determining, moved to
29713 oacc_loop_fixed_partitions.
29714 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
29715 stmts, add e_mask field.
29716 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
29717 (oacc_thread_numbers): Use oacc_dim_call.
29718 (oacc_xform_tile): New.
29719 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
29720 (finish_oacc_loop): Adjust for ifns vector.
29721 (oacc_loop_discover_walk): Append loop abstraction sites to list,
29722 add case for GOACC_TILE fns.
29723 (oacc_loop_xform_loop): Delete.
29724 (oacc_loop_process): Iterate over call list directly, and add
29725 handling for GOACC_TILE fns.
29726 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
29728 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
29729 vector partitioning to outer loops. Assign 2 partitions to loops
29730 when available. Add TILE handling.
29731 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
29732 (execite_oacc_device_lower): Process GOACC_TILE fns,
29733 ignore unknown specs.
29734 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
29735 * tree.c (omp_clause_num_ops): Adjust TILE ops.
29736 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
29738 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
29740 * configure.ac (ACX_BUGURL): Update.
29741 * configure: Regenerate.
29743 2017-02-09 Richard Biener <rguenther@suse.de>
29745 PR tree-optimization/69823
29746 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
29747 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
29749 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
29751 * config/arc/arc-c.def: Add __NPS400__ definition.
29752 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
29753 (TARGET_NPS400): Define.
29755 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
29757 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
29759 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
29760 pointer, arch_info.
29761 (arc_cpu_types): Fill the arch_info field with a pointer into the
29762 arc_arch_types table.
29763 (arc_selected_cpu): Declare.
29764 * config/arc/arc.c (arc_selected_cpu): Make global.
29765 (arc_selected_arch): Delete.
29766 (arc_base_cpu): Delete.
29767 (arc_override_options): Remove references to deleted variables,
29768 update access to arch information.
29769 (ARC_OPT): Update access to arch information.
29770 (ARC_OPTX): Likewise.
29771 * config/arc/arc.h (arc_base_cpu): Remove declaration.
29772 (TARGET_ARC600): Update access to arch information.
29773 (TARGET_ARC601): Likewise.
29774 (TARGET_ARC700): Likewise.
29775 (TARGET_EM): Likewise.
29776 (TARGET_HS): Likewise.
29777 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
29780 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
29783 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
29784 condition/operands for integer GE/LE/GEU/LEU operations.
29786 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
29788 PR translation/79397
29789 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
29792 2017-02-08 Martin Jambor <mjambor@suse.cz>
29795 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
29796 whether allocation happened.
29797 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
29798 nothing was allocated.
29800 2017-02-08 Jakub Jelinek <jakub@redhat.com>
29802 PR tree-optimization/79408
29803 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
29804 constant, but SSA_NAME with a known integer range, use the minimum
29805 of that range instead of op1 to determine if modulo can be replaced
29806 with its first operand.
29808 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29810 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
29812 2017-02-08 Richard Biener <rguenther@suse.de>
29814 PR tree-optimization/71824
29815 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
29816 Check all loops contained in the merged region.
29818 2017-02-07 Andrew Pinski <apinski@cavium.com>
29820 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
29822 2017-02-07 Andrew Pinski <apinski@cavium.com>
29824 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
29825 (thunderxt88): Likewise.
29826 (thunderxt81): Disable LSE and change v8.1 to v8.
29827 (thunderxt83): Likewise.
29829 2017-02-07 Jakub Jelinek <jakub@redhat.com>
29830 Richard Biener <rguenther@suse.de>
29832 PR middle-end/79399
29833 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
29834 type from int to size_t.
29835 * ira-costs.c (struct_costs_size): Change type from int to size_t.
29837 2017-02-07 Jakub Jelinek <jakub@redhat.com>
29839 PR rtl-optimization/79386
29840 * cprop.c (bypass_conditional_jumps): Initialize
29841 bypass_last_basic_block already before splitting bbs after
29842 unconditional traps...
29843 (bypass_conditional_jumps): ... rather than here.
29846 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
29847 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
29848 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
29849 fix -masm=intel patterns.
29851 2017-02-07 Richard Biener <rguenther@suse.de>
29853 PR tree-optimization/79256
29854 PR middle-end/79278
29855 * builtins.c (get_object_alignment_2): Use min_align_of_type
29856 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
29857 and ADJUST_FIELD_ALIGN.
29859 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
29861 * doc/tm.texi: Regenerate.
29862 * stor-layout.c (layout_decl): Adjust.
29863 (update_alignment_for_field): Likewise.
29864 (place_field): Likewise.
29865 (min_align_of_type): Likewise.
29866 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
29867 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
29868 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
29869 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
29870 * config/frv/frv.c (frv_adjust_field_align): Likewise.
29871 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
29872 * config/i386/i386.c (x86_field_alignment): Likewise.
29873 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
29874 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
29875 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
29876 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
29877 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
29878 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
29882 2017-01-30 Richard Biener <rguenther@suse.de>
29884 PR tree-optimization/79256
29885 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
29886 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
29889 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
29891 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
29892 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
29893 builtins to SImode and emit a zero-extend, if necessary.
29895 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29897 * docs/invoke.texi (RISC-V Options): Alphabetize.
29899 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29901 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
29904 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29906 * config/riscv/riscv.c: New file.
29907 * common/config/riscv/riscv-common.c: Likewise.
29908 * config.gcc: Likewise.
29909 * config/riscv/constraints.md: Likewise.
29910 * config/riscv/elf.h: Likewise.
29911 * config/riscv/generic.md: Likewise.
29912 * config/riscv/linux.h: Likewise.
29913 * config/riscv/multilib-generator: Likewise.
29914 * config/riscv/peephole.md: Likewise.
29915 * config/riscv/pic.md: Likewise.
29916 * config/riscv/predicates.md: Likewise.
29917 * config/riscv/riscv-builtins.c: Likewise.
29918 * config/riscv/riscv-c.c: Likewise.
29919 * config/riscv/riscv-ftypes.def: Likewise.
29920 * config/riscv/riscv-modes.def: Likewise.
29921 * config/riscv/riscv-opts.h: Likewise.
29922 * config/riscv/riscv-protos.h: Likewise.
29923 * config/riscv/riscv.h: Likewise.
29924 * config/riscv/riscv.md: Likewise.
29925 * config/riscv/riscv.opt: Likewise.
29926 * config/riscv/sync.md: Likewise.
29927 * config/riscv/t-elf-multilib: Likewise.
29928 * config/riscv/t-linux: Likewise.
29929 * config/riscv/t-linux-multilib: Likewise.
29930 * config/riscv/t-riscv: Likewise.
29931 * configure.ac: Likewise.
29932 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
29933 Waterman as RISC-V maintainers.
29934 * doc/install.texi: Add RISC-V entries.
29935 * doc/invoke.texi: Add RISC-V options section.
29936 * doc/md.texi: Add RISC-V constraints section.
29937 * configure: Regenerated.
29939 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
29942 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
29943 false values to be constant vectors with all 0 or all 1 bits set.
29944 (vcondu<mode><mode>): Likewise.
29945 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
29947 (fpmask_comparison_operator): Update comment.
29948 (vecint_comparison_operator): New predicate.
29949 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
29950 vector conditionals when the true and false values are constant
29951 vectors with all 0 bits or all 1 bits set.
29953 2017-02-06 Martin Sebor <msebor@redhat.com>
29955 PR tree-optimization/79376
29956 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
29958 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
29960 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
29961 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
29962 to simplify split condition.
29964 2017-02-06 Jakub Jelinek <jakub@redhat.com>
29966 * omp-expand.c (oxpand_omp_atomic_fetch_op,
29967 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
29970 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
29972 PR rtl-optimization/68664
29973 * target.def (can_speculate_insn): New hook.
29974 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
29975 * doc/tm.texi: Regenerate.
29976 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
29977 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
29978 (rs6000_sched_can_speculate_insn): New function.
29980 2017-02-06 Jakub Jelinek <jakub@redhat.com>
29982 PR tree-optimization/79284
29983 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
29984 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
29985 vectorizable_mask_load_store, vectorizable_operation,
29986 vect_is_simple_cond, get_same_sized_vectype): Use it instead
29987 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
29988 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
29989 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
29990 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
29991 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
29992 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
29993 is_gimple_assign (stmt). Replace another such test with
29994 is_gimple_assign (stmt).
29996 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
29999 * config/avr/avr.c (rtl-iter.h): Include it.
30000 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30001 (avr_legitimate_combined_insn): ...and implementation.
30003 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30005 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30006 * config/s390/s390.c (s390_const_operand_ok)
30007 (s390_canonicalize_comparison, s390_extract_part)
30008 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30009 (s390_contiguous_bitmask_p, s390_rtx_costs)
30010 (legitimize_pic_address): Likewise.
30011 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30012 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30013 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30014 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30015 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30017 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30019 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30020 REGNO($0) == REGNO($1).
30022 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30024 * config/s390/linux.h(SIZE_TYPE): Add comment.
30026 2017-02-06 Julian Brown <julian@codesourcery.com>
30027 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30028 Virendra Pathak <virendra.pathak@broadcom.com>
30030 * config/aarch64/aarch64-cores.def: Change the scheduler
30032 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30033 * config/aarch64/thunderx2t99.md: New file.
30035 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30037 * doc/standards.texi (Go Language): Update link to language
30040 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30042 * tree-eh.c (lower_resx): Sanitize profile.
30043 (cleanup_empty_eh_move_lp): Likewise.
30045 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30048 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30050 * cfgloopmanip.h (loop_version): Update prototype.
30051 * modulo-sched.c (sms_schedule): Update call of loop_version.
30052 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30053 * tree-parloops.c (gen_parallel_loop): Likewise.
30054 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30055 * tree-ssa-loop-split.c (split_loop): Likewise.
30056 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30057 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30059 2017-02-05 Martin Liska <mliska@suse.cz>
30062 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30064 (print_operand_address): Initialize a struct to zero.
30066 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30068 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30069 garbage collector only in textual form.
30071 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30073 * doc/extend.texi (x86 specific memory model extensions for
30074 transactional memory): Simplify a phrase.
30076 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30079 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30080 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30081 (atomic_storedi_1): Likewise.
30083 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30085 PR tree-optimization/79338
30086 * tree-parloops.c (gather_scalar_reductions): Don't call
30087 vect_analyze_loop_form for loop->inner before destroying loop's
30090 2017-02-03 Martin Sebor <msebor@redhat.com>
30092 PR tree-optimization/79327
30093 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30094 when precision has resulted in leading zeros.
30095 (format_integer): Adjust the likely counter to assume an unknown
30096 argument that may be zero is non-zero.
30098 2017-02-03 Jason Merrill <jason@redhat.com>
30101 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30102 avoid copying non-taken branch.
30104 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30106 PR tree-optimization/79340
30107 * tree-vect-loop.c (vectorizable_reduction): Release
30108 vec_defs elements after safe_splicing them into other vectors.
30111 PR tree-optimization/79327
30112 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30113 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30115 (format_integer): Use wide_int_to_tree instead of build_int_cst
30116 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30117 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30118 of shortest and longest sequence.
30120 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30122 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30123 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30125 2017-02-03 Walter Lee <walt@tilera.com>
30128 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30129 after initial stackframe link reg save.
30130 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30132 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30135 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30136 wu for stxssp alternative.
30138 2017-02-03 Martin Sebor <msebor@redhat.com>
30140 PR tree-optimization/79352
30141 * gimple-fold.c (get_range_strlen): Add argument.
30142 (get_range_strlen): Change return type to bool.
30143 (get_maxval_strlen): Pass in a dummy argument.
30144 * gimple-fold.h (get_range_strlen): Change return type to bool.
30145 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30146 * tree.h (array_at_struct_end_p): Add argument.
30147 * tree.c (array_at_struct_end_p): Handle it.
30149 2017-02-03 Martin Liska <mliska@suse.cz>
30152 * multiple_target.c (create_dispatcher_calls): Redirect edge
30153 from a caller of a dispatcher.
30154 (expand_target_clones): Make the clones local.
30155 (ipa_target_clone): Do both target clones and resolvers.
30156 (ipa_dispatcher_calls): Remove the pass.
30157 (pass_dispatcher_calls::gate): Likewise.
30158 (make_pass_dispatcher_calls): Likewise.
30159 * passes.def (pass_target_clone): Put as very first IPA early
30162 2017-02-03 Martin Liska <mliska@suse.cz>
30164 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30165 in case of a function with ifunc attribute.
30167 2017-02-03 Martin Liska <mliska@suse.cz>
30169 * cgraph.c (cgraph_node::dump): Dump function version info.
30170 * symtab.c (symtab_node::dump_base): Add missing new line.
30172 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30174 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30175 (ifcombine_ifandif): Use it.
30177 2017-02-03 Martin Liska <mliska@suse.cz>
30179 * doc/invoke.texi: Document default value for
30180 use-after-scope-direct-emission-threshold.
30182 2017-02-03 Martin Liska <mliska@suse.cz>
30184 PR tree-optimization/79339
30185 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30186 (format_floating): Likewise.
30188 2017-02-03 Martin Liska <mliska@suse.cz>
30191 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30193 (ipa_node_params_t::remove): Likewise.
30194 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30195 initialization from removed ipa_node_params_t::insert.
30196 (ipa_node_params::~ipa_node_params): Move from removed
30197 ipa_node_params_t::release.
30198 * symbol-summary.h (symbol_summary::m_released): New member.
30199 Do not release a summary twice. Do not allow to call finalizer
30200 for types of a summary that live in GGC memory.
30202 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30204 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30207 2017-02-02 Martin Sebor <msebor@redhat.com>
30209 PR middle-end/79275
30210 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30211 (format_string): Tighten up the range of output for non-constant
30212 strings and correct the expected range for wide non-constant strings.
30214 2017-02-02 Martin Sebor <msebor@redhat.com>
30216 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30218 PR middle-end/32003
30219 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30221 (-fdump-tree-@var): Add to index and document how to come up
30222 with pass-specific option and dump file names.
30223 (-fdump-passes): Clarify where to look for output.
30225 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30227 PR middle-end/77445
30228 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30229 statistics of the analyzed path; allow threading for speed when
30230 any of BBs along the path are optimized for speed.
30232 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30234 PR middle-end/78468
30235 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30236 settings of the virtual registers.
30239 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30241 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30242 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30245 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30247 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30248 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30250 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30252 * config/s390/s390.md: Add missing comments with the expanded
30254 * config/s390/vector.md: Likewise.
30255 * config/s390/vx-builtins.md: Likewise.
30257 2017-02-02 Jakub Jelinek <jakub@redhat.com>
30260 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
30261 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
30262 conditions on a single line.
30264 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30266 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30267 __S390_VX__ to __VX__.
30269 2017-02-01 Andrew Pinski <apinski@cavium.com>
30271 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
30272 stmt_info to record_stmt_cost.
30273 (vect_get_known_peeling_cost): Pass stmt_info if known to
30275 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
30276 cpu_vector_cost field into
30277 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
30278 field into vec_int_stmt_cost and vec_fp_stmt_cost.
30279 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
30280 splitting of scalar_stmt_cost and vec_stmt_cost.
30281 (thunderx_vector_cost): Likewise.
30282 (cortexa57_vector_cost): LIkewise.
30283 (exynosm1_vector_cost): Likewise.
30284 (xgene1_vector_cost): Likewise.
30285 (thunderx2t99_vector_cost): Improve after the splitting of the two
30287 (aarch64_builtin_vectorization_cost): Update for the splitting of
30288 scalar_stmt_cost and vec_stmt_cost.
30290 2017-02-01 Torvald Riegel <triegel@redhat.com>
30291 Richard Henderson <rth@redhat.com>
30293 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
30294 conditional on existance of a fast atomic load.
30295 * optabs-query.c (can_atomic_load_p): New function.
30296 * optabs-query.h (can_atomic_load_p): Declare it.
30297 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
30298 no fast atomic load is available for the particular size of access.
30299 (expand_atomic_compare_and_swap): Likewise.
30300 (expand_atomic_load): Likewise.
30301 (expand_atomic_store): Likewise.
30302 (expand_atomic_fetch_op): Likewise.
30303 * testsuite/lib/target-supports.exp
30304 (check_effective_target_sync_int_128): Remove x86 because it provides
30305 no fast atomic load.
30306 (check_effective_target_sync_int_128_runtime): Likewise.
30308 2017-02-01 Richard Biener <rguenther@suse.de>
30310 * graphite.c: Include tree-vectorizer.h for find_loop_location.
30311 (graphite_transform_loops): Provide opt-info for optimized nests.
30312 * tree-parloop.c (parallelize_loops): Provide opt-info for
30313 parallelized loops.
30315 2017-02-01 Richard Biener <rguenther@suse.de>
30317 PR middle-end/79315
30318 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
30319 was not set before.
30321 2017-02-01 Richard Biener <rguenther@suse.de>
30323 PR tree-optimization/71824
30324 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30325 Verify the loops are valid in the merged SESE region.
30326 (scop_detection::can_represent_loop_1): Check analyzing the
30327 evolution of the number of iterations in the region succeeds.
30329 2017-01-31 Ian Lance Taylor <iant@golang.org>
30331 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
30332 REG_ARGS_SIZE note to 32-bit push insns and call insn.
30334 2017-01-31 David Malcolm <dmalcolm@redhat.com>
30336 PR preprocessor/79210
30337 * input.c (get_substring_ranges_for_loc): Replace line_width
30338 assertion with error-handling.
30340 2017-01-31 Richard Biener <rguenther@suse.de>
30342 PR tree-optimization/77318
30343 * graphite-sese-to-poly.c (extract_affine): Fix assert.
30344 (create_pw_aff_from_tree): Take loop parameter.
30345 (add_condition_to_pbb): Pass loop of the condition to
30346 create_pw_aff_from_tree.
30348 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30350 * config/s390/s390.c (s390_asan_shadow_offset): New function.
30351 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
30353 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
30357 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
30359 (convert_int_to_float128): Likewise.
30360 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
30361 (convert_int_to_float128): Likewise.
30362 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
30363 (UNSPEC_IEEE128_CONVERT): Likewise.
30364 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
30365 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
30366 Use local variables for IBM extended format.
30367 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
30368 (fix_trunc<mode>si2_fprs): Likewise.
30369 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
30370 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
30371 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
30372 to know that we can now have integers of all sizes in vector
30374 (fix<uns>_<mode>di2_hw): Likewise.
30375 (float<uns>_<mode>si2_hw): Likewise.
30376 (fix_<mode>si2_hw): Likewise.
30377 (fixuns_<mode>si2_hw): Likewise.
30378 (float<uns>_<mode>di2_hw): Likewise.
30379 (float_<mode>di2_hw): Likewise.
30380 (float_<mode>si2_hw): Likewise.
30381 (floatuns_<mode>di2_hw): Likewise.
30382 (floatuns_<mode>si2_hw): Likewise.
30383 (xscvqp<su>wz_<mode>): Delete, no longer used.
30384 (xscvqp<su>dz_<mode>): Likewise.
30385 (xscv<su>dqp_<mode>): Likewise.
30386 (ieee128_mfvsrd_64bit): Likewise.
30387 (ieee128_mfvsrd_32bit): Likewise.
30388 (ieee128_mfvsrwz): Likewise.
30389 (ieee128_mtvsrw): Likewise.
30390 (ieee128_mtvsrd_64bit): Likewise.
30391 (ieee128_mtvsrd_32bit): Likewise.
30393 2017-01-31 Martin Liska <mliska@suse.cz>
30396 * ipa-prop.c (ipa_free_all_node_params): Call release method
30397 instead of ~sumbol_summary to not to trigger double times
30400 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
30402 PR tree-optimization/71691
30403 * bitmap.h (class auto_bitmap): New.
30404 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
30405 is_maybe_undefined instead of ssa_undefined_value_p.
30407 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30409 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30410 __S390_ARCH_LEVEL__ to __ARCH__.
30412 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30414 PR tree-optimization/79267
30415 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
30416 if should_remove_lhs_p is true.
30418 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
30421 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
30422 (add_alignment_attribute): New.
30423 (base_type_die): Add alignment attribute.
30424 (subrange_type_die): Likewise.
30425 (modified_type_die): Likewise.
30426 (gen_array_type_die): Likewise.
30427 (gen_descr_array_type_die: Likewise.
30428 (gen_enumeration_type_die): Likewise.
30429 (gen_subprogram_die): Likewise.
30430 (gen_variable_die): Likewise.
30431 (gen_field_die): Likewise.
30432 (gen_ptr_to_mbr_type_die): Likewise.
30433 (gen_struct_or_union_type_die): Likewise.
30434 (gen_subroutine_type_die): Likewise.
30435 (gen_typedef_die): Likewise.
30436 (base_type_cmp): Compare alignment attribute.
30438 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30441 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
30442 (setb_unsigned) New pattern for setb with CCUNS.
30443 * config/rs6000/rs6000.c (expand_block_compare): Use a different
30444 subfc./subfe sequence to avoid overflow problems. Generate a
30445 shorter sequence with cmpld/setb for power9.
30446 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
30447 for generating subfc. instruction.
30448 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
30449 now uses this instruction.
30451 2017-01-30 Ian Lance Taylor <iant@google.com>
30454 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
30455 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
30457 2017-01-30 Martin Sebor <msebor@redhat.com>
30459 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
30460 Move constant to the right of a relational operator.
30461 (get_mpfr_format_length, format_character, format_string): Ditto.
30462 (should_warn_p, maybe_warn): Same.
30464 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
30466 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
30469 * asan.c (get_translation_unit_decl): Remove function.
30470 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
30472 2017-01-30 Martin Liska <mliska@suse.cz>
30474 PR gcov-profile/79259
30475 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
30476 -fprofile-generate.
30478 2017-01-30 Martin Liska <mliska@suse.cz>
30481 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
30482 Initialize variables with NULL value.
30484 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
30487 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
30489 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
30491 2017-01-30 Richard Biener <rguenther@suse.de>
30493 PR tree-optimization/79276
30494 * tree-vrp.c (process_assert_insertions): Properly adjust common
30495 when removing a duplicate.
30497 2017-01-30 Richard Biener <rguenther@suse.de>
30499 PR tree-optimization/79256
30500 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30501 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30503 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
30505 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
30508 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
30509 ("*r<noxa>sbg_<mode>_sll_bitmask")
30510 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
30511 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
30512 Use contiguous_bitmask_nowrap_operand.
30514 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30517 * config/rs6000/altivec.h (vec_xl): Revise #define.
30518 (vec_xst): Likewise.
30520 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
30522 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
30524 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
30526 PR rtl-optimization/79194
30527 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
30528 traps before call to bypass_conditional_jumps.
30530 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30532 PR tree-optimization/71374
30533 * lra-constraints.c (check_conflict_input_operands): New.
30534 (match_reload): Use it.
30536 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30539 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
30540 account to calculate conflict_set.
30542 2017-01-27 Bin Cheng <bin.cheng@arm.com>
30544 PR rtl-optimization/78559
30545 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
30546 other_insn in combine.
30548 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
30550 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
30551 uint16_type_node for BT_UINT16.
30553 2017-01-27 David Malcolm <dmalcolm@redhat.com>
30555 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
30556 "RTL Tests" to menu.
30557 (GIMPLE Tests): New node.
30558 (RTL Tests): New node.
30560 2017-01-27 Richard Biener <rguenther@suse.de>
30562 PR tree-optimization/79245
30563 * tree-loop-distribution.c (distribute_loop): Apply cost
30564 modeling also to detected patterns.
30566 2017-01-27 Richard Biener <rguenther@suse.de>
30568 PR tree-optimization/71433
30569 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
30570 (compare_assert_loc): New function.
30571 (process_assert_insertions): Sort and optimize assert locations
30572 to remove duplicates and push down identical assertions on
30573 edges to their destination block.
30575 2017-01-27 Richard Biener <rguenther@suse.de>
30577 PR tree-optimization/79244
30578 * tree-vrp.c (remove_range_assertions): Forcefully propagate
30579 out SSA names even if abnormal.
30581 2017-01-27 Jakub Jelinek <jakub@redhat.com>
30583 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
30584 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
30585 instead of MPFR_RNDN.
30587 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
30590 * arm.c (arm_option_override): Don't call build_target_option_node
30591 until after doing all option overrides.
30592 (arm_valid_target_attribute_tree): Likewise.
30594 2017-01-27 Martin Liska <mliska@suse.cz>
30596 * doc/invoke.texi (-fprofile-arcs): Document profiling support
30597 for {cd}tors and C++ {cd}tors.
30599 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
30601 * config/s390/s390.md ("*setmem_long_and")
30602 ("*setmem_long_and_31z"): Use zero_extend instead of and.
30604 2017-01-26 Martin Sebor <msebor@redhat.com>
30606 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
30609 2017-01-26 Martin Sebor <msebor@redhat.com>
30611 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
30612 HAVE_DFmode before using XFmode or DFmode.
30613 (parse_directive): Avoid using the z length modifier to avoid
30614 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
30616 PR middle-end/78703
30617 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
30618 to accept adjustment as an array.
30619 (get_int_range): New function.
30620 (struct directive): Make width and prec arrays.
30621 (directive::set_width, directive::set_precision): Call get_int_range.
30622 (format_integer, format_floating): Handle width and precision ranges.
30623 (format_string, parse_directive): Same.
30625 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30628 * dwarf2out.c (generate_skeleton_bottom_up): For children with
30629 comdat_type_p set, just clone them, but keep the children in the
30633 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
30634 which have direct callers with -fvar-tracking-assignments enabled
30636 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
30637 inside of type units.
30639 2017-01-26 Martin Sebor <msebor@redhat.com>
30641 PR middle-end/78703
30642 * gimple-ssa-sprintf.c (struct result_range): Add likely and
30644 (struct format_result): Replace number_chars, number_chars_min,
30645 and number_chars_max with a single member of struct result_range.
30647 (format_result::operator+=): Adjust.
30648 (struct fmtresult): Remove bounded. Handle likely and unlikely
30650 (fmtresult::adjust_for_width_or_precision): New function.
30651 (fmtresult:type_max_digits): New function.
30652 (bytes_remaining): Handle likely and unlikely counters.
30653 (min_bytes_remaining): Remove.
30654 (format_percent): Simplify.
30655 (format_integer, format_floating): Set likely and unlikely counters.
30656 (get_string_length, format_character, format_string): Same.
30657 (format_plain, should_warn_p): New function.
30658 (maybe_warn): Call should_warn_p. Update diagnostic messages
30659 and handle those for all directives, including plain strings.
30660 (format_directive): Handle likely and unlikely counters.
30661 Remove unnecessary quoting from diagnostics. Add an informational
30663 (add_bytes): Remove.
30664 (pass_sprintf_length::compute_format_length): Simplify.
30665 (try_substitute_return_value): Handle likely and unlikely counters.
30667 2017-01-26 Carl Love <cel@us.ibm.com>
30669 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
30670 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
30672 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
30675 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
30676 endianess for subregs into account.
30677 * lra-constraints.c (lra_constraints): Do risky transformations
30678 always on the first iteration.
30679 * lra-lives.c (check_pseudos_live_through_calls): Add arg
30680 last_call_used_reg_set.
30681 (process_bb_lives): Define and use last_call_used_reg_set.
30682 * lra.c (lra): Always continue after lra_constraints on the first
30685 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
30687 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
30689 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
30691 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30693 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
30694 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
30695 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
30696 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
30697 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
30698 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
30699 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
30700 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
30701 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
30703 2017-01-26 Marek Polacek <polacek@redhat.com>
30706 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
30707 for the third operand.
30709 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30711 PR middle-end/79236
30712 * omp-low.c (struct omp_context): Add simt_stmt field.
30713 (scan_omp_for): Return omp_context *.
30714 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
30715 context to the _simt_ SIMD stmt.
30716 (lower_omp_for): For combined SIMD with sibling _simt_
30717 SIMD, make sure to use the same decls in _looptemp_
30718 clauses as in the sibling.
30720 2017-01-26 David Sherwood <david.sherwood@arm.com>
30722 PR middle-end/79212
30723 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
30726 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30729 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
30730 emit fld b; fld a; if possible.
30732 * brig-builtins.def: Update copyright years.
30733 * config/arm/arm_acle_builtins.def: Update copyright years.
30735 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
30738 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
30739 constraint instead of o for the stxsd instruction.
30741 2017-01-25 Carl Love <cel@us.ibm.com>
30743 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
30744 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
30746 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
30748 * doc/invoke.texi (C++ Dialect Options): Fix typo.
30750 2017-01-25 Richard Biener <rguenther@suse.de>
30752 PR tree-optimization/69264
30753 * target.def (vector_alignment_reachable): Improve documentation.
30754 * doc/tm.texi: Regenerate.
30755 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
30757 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
30758 earlier changes with respect to TYPE_USER_ALIGN.
30759 (vector_alignment_reachable_p): Likewise. Improve dumping.
30761 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30764 * config/arm/arm.md (xordi3): Force constant operand into a register
30767 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30769 * doc/invoke.texi (-fstore-merging): Correct default optimization
30770 levels at which it is enabled.
30771 (-O): Move -fstore-merging from list to...
30774 2017-01-25 Richard Biener <rguenther@suse.de>
30777 * omp-expand.c: Include debug.h.
30778 (expand_omp_taskreg): Make sure to generate early debug before
30779 outlining anything from a function.
30780 (expand_omp_target): Likewise.
30781 (grid_expand_target_grid_body): Likewise.
30783 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
30786 * asan.c (get_translation_unit_decl): New function.
30787 (asan_add_global): Extract modules file name from globals
30788 TRANSLATION_UNIT_DECL name.
30790 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
30793 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
30794 for long calls with APCS frame and VFP.
30796 2017-01-24 David Malcolm <dmalcolm@redhat.com>
30798 * cfg.c (original_copy_tables_initialized_p): New function.
30799 * cfg.h (original_copy_tables_initialized_p): New decl.
30800 * cfgrtl.c (relink_block_chain): Guard the call to
30801 free_original_copy_tables with a call to
30802 original_copy_tables_initialized_p.
30803 * cgraph.h (symtab_node::native_rtl_p): New decl.
30804 * cgraphunit.c (symtab_node::native_rtl_p): New function.
30805 (symtab_node::needed_p): Don't assert for early assembly output
30806 for __RTL functions.
30807 (cgraph_node::finalize_function): Set "force_output" for __RTL
30809 (cgraph_node::analyze): Bail out early for __RTL functions.
30810 (analyze_functions): Update assertion to support __RTL functions.
30811 (cgraph_node::expand): Bail out early for __RTL functions.
30812 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
30814 * function.h (struct function): Update comment for field
30816 * gimple-expr.c: Include "tree-pass.h".
30817 (gimple_has_body_p): Return false for __RTL functions.
30818 * Makefile.in (OBJS): Add run-rtl-passes.o.
30819 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
30821 (gcc::pass_manager::get_clean_slate): New accessor.
30822 * passes.c: Include "insn-addr.h".
30823 (should_skip_pass_p): Add logging. Update logic for running
30824 "expand" to be compatible with both __GIMPLE and __RTL. Guard
30825 property-provider override so it is only done for gimple passes.
30827 (skip_pass): New function.
30828 (execute_one_pass): Call skip_pass when skipping passes.
30829 * read-md.c (md_reader::read_char): Support filtering
30830 the input to a subset of line numbers.
30831 (md_reader::md_reader): Initialize fields
30832 m_first_line and m_last_line.
30833 (md_reader::read_file_fragment): New function.
30834 * read-md.h (md_reader::read_file_fragment): New decl.
30835 (md_reader::m_first_line): New field.
30836 (md_reader::m_last_line): New field.
30837 * read-rtl-function.c (function_reader::create_function): Only
30838 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
30839 curr_properties. Set DECL_INITIAL to a dummy block.
30840 (read_rtl_function_body_from_file_range): New function.
30841 * read-rtl-function.h (read_rtl_function_body_from_file_range):
30843 * run-rtl-passes.c: New file.
30844 * run-rtl-passes.h: New file.
30846 2017-01-24 Jeff Law <law@redhat.com>
30848 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
30851 2017-01-24 Bin Cheng <bin.cheng@arm.com>
30853 PR tree-optimization/79159
30854 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
30855 (record_nonwrapping_iv): Improve boundary using above function if no
30856 value range information.
30858 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
30859 Martin Jambor <mjambor@suse.cz>
30861 * brig-builtins.def: New file.
30862 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
30863 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
30864 (DEF_HSAIL_SAT_BUILTIN): Likewise.
30865 (DEF_HSAIL_INTR_BUILTIN): Likewise.
30866 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
30867 * builtin-types.def (BT_INT8): New.
30868 (BT_INT16): Likewise.
30869 (BT_UINT8): Likewise.
30870 (BT_UINT16): Likewise.
30871 (BT_FN_ULONG): Likewise.
30872 (BT_FN_UINT_INT): Likewise.
30873 (BT_FN_UINT_ULONG): Likewise.
30874 (BT_FN_UINT_LONG): Likewise.
30875 (BT_FN_UINT_PTR): Likewise.
30876 (BT_FN_ULONG_PTR): Likewise.
30877 (BT_FN_INT8_FLOAT): Likewise.
30878 (BT_FN_INT16_FLOAT): Likewise.
30879 (BT_FN_UINT32_FLOAT): Likewise.
30880 (BT_FN_UINT16_FLOAT): Likewise.
30881 (BT_FN_UINT8_FLOAT): Likewise.
30882 (BT_FN_UINT64_FLOAT): Likewise.
30883 (BT_FN_UINT16_UINT32): Likewise.
30884 (BT_FN_UINT32_UINT16): Likewise.
30885 (BT_FN_UINT16_UINT16_UINT16): Likewise.
30886 (BT_FN_INT_PTR_INT): Likewise.
30887 (BT_FN_UINT_PTR_UINT): Likewise.
30888 (BT_FN_LONG_PTR_LONG): Likewise.
30889 (BT_FN_ULONG_PTR_ULONG): Likewise.
30890 (BT_FN_VOID_UINT64_UINT64): Likewise.
30891 (BT_FN_UINT8_UINT8_UINT8): Likewise.
30892 (BT_FN_INT8_INT8_INT8): Likewise.
30893 (BT_FN_INT16_INT16_INT16): Likewise.
30894 (BT_FN_INT_INT_INT): Likewise.
30895 (BT_FN_UINT_FLOAT_UINT): Likewise.
30896 (BT_FN_FLOAT_UINT_UINT): Likewise.
30897 (BT_FN_ULONG_UINT_UINT): Likewise.
30898 (BT_FN_ULONG_UINT_PTR): Likewise.
30899 (BT_FN_ULONG_ULONG_ULONG): Likewise.
30900 (BT_FN_UINT_UINT_UINT): Likewise.
30901 (BT_FN_VOID_UINT_PTR): Likewise.
30902 (BT_FN_UINT_UINT_PTR: Likewise.
30903 (BT_FN_UINT32_UINT64_PTR): Likewise.
30904 (BT_FN_INT_INT_UINT_UINT): Likewise.
30905 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
30906 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
30907 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
30908 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
30909 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
30910 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
30911 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
30912 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
30913 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
30914 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
30915 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
30916 * doc/frontends.texi: List BRIG FE.
30917 * doc/install.texi (Testing): Add BRIG tesring requirements.
30918 * doc/invoke.texi (Overall Options): Mention BRIG.
30919 * doc/standards.texi (Standards): Doucment BRIG HSA version.
30921 2017-01-24 Richard Biener <rguenther@suse.de>
30923 PR translation/79208
30924 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
30926 2017-01-24 Martin Jambor <mjambor@suse.cz>
30929 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
30930 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
30931 and known_contexts.
30933 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
30935 PR middle-end/79123
30936 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
30937 casts from signed to unsigned really don't have a range.
30939 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
30941 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
30942 GMP_RNDx for compatiblity.
30944 2017-01-24 Martin Liska <mliska@suse.cz>
30947 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
30948 that would prevent us to call alloca with -1 as argument.
30950 2017-01-24 Jakub Jelinek <jakub@redhat.com>
30952 * dwarf2out.c (output_compilation_unit_header, output_file_names):
30953 Avoid -Wformat-security warning.
30955 2017-01-23 Andrew Pinski <apinski@cavium.com>
30957 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
30960 2017-01-23 Martin Sebor <msebor@redhat.com>
30962 PR middle-end/78703
30963 * gimple-ssa-sprintf.c (warn_level): New global.
30964 (format_integer): Use it here and throughout the rest of the file.
30965 Use the same switch to compute sign as base.
30966 (maybe_warn): New function.
30967 (format_directive): Factor out warnings into maybe_warn.
30968 Add debugging output. Use warn_level.
30969 (add_bytes): Use warn_level.
30970 (pass_sprintf_length::compute_format_length): Add debugging output.
30971 (try_substitute_return_value): Same.
30972 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
30974 PR middle-end/78703
30975 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
30976 (struct fmtresult, format_integer, format_floating): Adjust.
30977 (fmtresult::fmtresult): Set max correctly in two argument ctor.
30978 (get_string_length, format_string,format_directive): Same.
30979 (pass_sprintf_length::compute_format_length): Same.
30980 (try_substitute_return_value): Simplify slightly.
30982 PR middle-end/78703
30983 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
30984 (fmtresult::operator+=): Outlined.
30985 (struct fmtresult): Add ctors.
30986 (struct conversion_spec): Rename...
30987 (struct directive): ...to this. Add and remove data members.
30988 (directive::set_width, directive::set_precision): New functions.
30989 (format_percent): Use fmtresult ctor.
30990 (get_width_and_precision): Remove.
30991 (format_integer): Make naming changes. Avoid computing width and
30993 (format_floating): Same. Adjust indentation.
30994 (format_character, format_none): New functions.
30995 (format_string): Moved character handling to format_character.
30996 (format_directive): Remove arguments, change return type.
30997 (parse_directive): New function.
30998 (pass_sprintf_length::compute_format_length): Move directive
30999 parsing to parse_directive.
31001 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31003 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31004 (assign_assembler_name_if_needed): ... this.
31005 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31006 (assign_assembler_name_if_needed): ... this.
31007 (free_lang_data_in_cgraph): Adjust callers.
31008 * cgraphunit.c (cgraph_node::analyze): Likewise.
31009 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31012 2017-01-23 Richard Biener <rguenther@suse.de>
31014 PR tree-optimization/79088
31015 PR tree-optimization/79188
31016 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31017 resetting loop bounds after last path deletion. Reset loop
31018 bounds of the target loop, make code match the comments.
31019 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31020 Make sure loops need no fixups.
31022 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31024 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31025 exponent support with double type for first argument.
31026 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31027 type returned by __builtin_vec_extract_sig,
31028 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31029 functions from "vector int" to "vector unsigned int" or from
31030 "vector long long int" to "vector unsigned long long int".
31031 Changed type returned by __builtin_vec_extract_exp,
31032 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31033 functions from "vector int" to "vector unsigned int" or from
31034 "vector long long int" to "vector unsigned long long int".
31035 Changed return type of __builtin_vec_test_data_class,
31036 __builtin_vec_test_data_class_sp, and
31037 __builtin_vec_test_data_class_dp from "vector int" to
31038 "vector bool int" or from "vector long long int" to "vector bool
31039 long long int" and changed second argument type from "unsigned
31040 int" to "int". Added new overloaded function forms "vector float
31041 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31042 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31043 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31044 double, vector unsigned long long int)" and "vector double
31045 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31046 long int)". Changed return type of
31047 __builtin_scalar_test_data_class and
31048 __builtin_scalar_test_data_class_sp and
31049 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31050 int" and changed second argument from "unsigned int" to "int".
31051 Changed type returned by __builtin_scalar_test_neg,
31052 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31053 from "int" to "bool int". Added new overloaded function form
31054 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31055 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31056 exponent double-precision with floating point first argument.
31057 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31058 documentation of scalar_test_data_class, scalar_test_neg,
31059 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31060 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31061 vec_test_data_class built-in functions to reflect refinements in
31062 their type signatures.
31064 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31066 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31068 (aarch64_elf_asm_destructor): Likewise.
31070 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31072 PR rtl-optimization/78634
31073 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31074 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31075 * ifcvt.c (noce_try_cmove): Add missing cost check.
31077 PR rtl-optimization/71724
31078 * combine.c (if_then_else_cond): Look for situations where it is
31079 beneficial to undo the work of one of the recursive calls.
31081 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31083 PR tree-optimization/70754
31084 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31085 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31086 combined stmt before it if not NULL.
31087 (combine_chains): Process refs reversely and compute dominance point
31090 2017-01-23 Martin Liska <mliska@suse.cz>
31092 PR tree-optimization/79196
31093 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31094 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31096 (strlen_optimize_stmt): Call the renamed function.
31098 2017-01-23 Michael Matz <matz@suse.de>
31100 PR tree-optimization/78384
31101 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31103 2017-01-23 Richard Biener <rguenther@suse.de>
31105 PR tree-optimization/79186
31106 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31107 both incoming edges before moving an assert.
31109 2017-01-23 Martin Jambor <mjambor@suse.cz>
31111 * ipa-prop.c (load_from_param_1): Removed.
31112 (load_from_unmodified_param): Bits from load_from_param_1 put back
31114 (load_from_param): Removed.
31115 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31116 with stmt. Reverted back to use of load_from_unmodified_param.
31118 2017-01-23 Martin Jambor <mjambor@suse.cz>
31121 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31122 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31123 field a pointer to garbage collected vector, mark lattices and
31124 ipcp_orig_node with GTY((skip)).
31125 (ipa_get_param_count): Adjust to descriptors being a pointer.
31126 (ipa_get_param): Likewise.
31127 (ipa_get_type): Likewise.
31128 (ipa_get_param_move_cost): Likewise.
31129 (ipa_set_param_used): Likewise.
31130 (ipa_get_controlled_uses): Likewise.
31131 (ipa_set_controlled_uses): Likewise.
31132 (ipa_is_param_used): Likewise.
31133 (ipa_node_params_t): Move into garbage collector. New methods insert
31135 (ipa_node_params_sum): Annotate wth GTY(()).
31136 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31138 (ipa_load_from_parm_agg): Adjust declaration.
31139 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31140 * ipa-profile.c (ipa_profile): Likewise.
31141 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31142 (ipa_populate_param_decls): Make descriptors parameter garbage
31144 (ipa_dump_param): Adjust to descriptors being a pointer.
31145 (ipa_alloc_node_params): Likewise.
31146 (ipa_initialize_node_params): Likewise.
31147 (load_from_param_1): Make descriptors parameter garbage collected.
31148 (load_from_unmodified_param): Likewise.
31149 (load_from_param): Likewise.
31150 (ipa_load_from_parm_agg): Likewise.
31151 (ipa_node_params::~ipa_node_params): Removed.
31152 (ipa_free_all_node_params): Remove call to delete operator.
31153 (ipa_node_params_t::insert): New.
31154 (ipa_node_params_t::remove): Likewise.
31155 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31156 copy known_csts and known_contexts vectors.
31157 (ipa_read_node_info): Adjust to descriptors being a pointer.
31158 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31160 (ipcp_transform_function): Make descriptors variable garbage
31163 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31165 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31166 * config/i386/avx512dqintrin.h: Ditto.
31167 * config/i386/avx512fintrin.h: Ditto.
31168 * config/i386/i386.c: Handle new builtins.
31169 * config/i386/i386-builtin.def: Add new builtins.
31170 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31171 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31173 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31174 Martin Liska <mliska@suse.cz>
31176 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31177 * asan.c (asan_expand_poison_ifn): Support stores and use
31178 appropriate ASAN report function.
31179 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31180 * internal-fn.def (ASAN_POISON_USE): Declare.
31181 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31182 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31183 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31184 ASAN_POISON calls w/o LHS.
31185 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31186 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31187 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31188 * gimplify.c (asan_poison_variables): Add attribute
31189 use_after_scope_memory to variables that really needs to live
31191 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31192 having the attribute.
31194 2017-01-23 Martin Liska <mliska@suse.cz>
31196 * asan.c (create_asan_shadow_var): New function.
31197 (asan_expand_poison_ifn): Likewise.
31198 * asan.h (asan_expand_poison_ifn): New declaration.
31199 * internal-fn.c (expand_ASAN_POISON): Likewise.
31200 * internal-fn.def (ASAN_POISON): New builtin.
31201 * sanopt.c (pass_sanopt::execute): Expand
31202 asan_expand_poison_ifn.
31203 * tree-inline.c (copy_decl_for_dup_finish): Make function
31205 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31206 * tree-ssa.c (is_asan_mark_p): New function.
31207 (execute_update_addresses_taken): Rewrite local variables
31208 (identified just by use-after-scope as addressable) into SSA.
31210 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31212 * doc/install.texi (Specific): opensource.apple.com uses https
31213 now. Remove trailing slash.
31215 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31217 * README.Portability: Remove note on an Irix compatibility issue.
31219 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31221 * gcov.c (INCLUDE_ALGORITHM): Define.
31222 (INCLUDE_VECTOR): Define.
31223 No longer include <vector> and <algorithm> directly.
31225 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31227 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31229 * doc/invoke.texi (Code Gen Options): Ditto.
31231 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31234 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31236 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31238 rtl-optimization/79125
31239 * cprop.c (local_cprop_pass): Handle cases where we make an
31240 unconditional trap.
31242 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31246 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31247 read from, for big endian.
31249 2017-01-20 Jiong Wang <jiong.wang@arm.com>
31251 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
31252 register pauth builtins for LP64 only.
31254 2017-01-20 Marek Polacek <polacek@redhat.com>
31257 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
31260 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31262 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
31264 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
31265 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
31266 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
31268 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31271 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
31272 in big-endian mode when they are not a single duplicated value.
31274 2017-01-20 Richard Biener <rguenther@suse.de>
31276 * BASE-VER: Bump to 7.0.1.
31278 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31280 * omp-low.c (omplow_simd_context): New struct. Use it...
31281 (lower_rec_simd_input_clauses): ...here and...
31282 (lower_rec_input_clauses): ...here to hold common data. Adjust all
31283 references to idx, lane, max_vf, is_simt.
31285 2017-01-20 Graham Markall <graham.markall@embecosm.com>
31287 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
31290 2017-01-20 Martin Jambor <mjambor@suse.cz>
31292 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
31293 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
31295 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
31296 (GTFILES): Rename hsa.c to hsa-common.c.
31297 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
31298 * hsa-dump.c: Likewise.
31299 * hsa-gen.c: Likewise.
31300 * hsa-regalloc.c: Likewise.
31301 * ipa-hsa.c: Likewise.
31302 * omp-expand.c: Likewise.
31303 * omp-low.c: Likewise.
31304 * toplev.c: Likewise.
31306 2017-01-20 Marek Polacek <polacek@redhat.com>
31309 * doc/invoke.texi: Document -Wduplicated-branches.
31310 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
31311 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
31312 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
31313 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
31314 return 0 only when not OEP_LEXICOGRAPHIC.
31315 (fold_build_cleanup_point_expr): Use the expression
31316 location when building CLEANUP_POINT_EXPR.
31317 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
31318 * tree.c (add_expr): Handle error_mark_node.
31320 2017-01-20 Martin Liska <mliska@suse.cz>
31323 * tree-profile.c (init_ic_make_global_vars): Do not call
31325 (gimple_init_gcov_profiler): Likewise.
31327 2017-01-20 Martin Liska <mliska@suse.cz>
31330 * cgraph.h (maybe_create_reference): Remove argument and
31332 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
31334 * ipa-cp.c (create_specialized_node): Likewise.
31335 * symtab.c (symtab_node::maybe_create_reference): Handle
31336 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
31338 2017-01-20 Martin Liska <mliska@suse.cz>
31340 * read-rtl-function.c (function_reader::create_function): Use
31341 build_decl instread of build_decl_stat.
31343 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
31345 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
31346 * config/i386/avx512dqintrin.h: Ditto.
31347 * config/i386/avx512fintrin.h: Ditto.
31348 * config/i386/i386-builtin-types.def: Add new types.
31349 * config/i386/i386.c: Handle new types.
31350 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
31351 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
31352 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
31353 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
31354 (__builtin_ia32_kshiftridi): New.
31355 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
31357 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
31361 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
31362 define to rs6000_init_stack_protect_guard.
31363 (rs6000_init_stack_protect_guard): New function.
31365 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31366 Yunqiang Su <yunqiang.su@imgtec.com>
31368 * config.gcc (supported_defaults): Add madd4.
31369 (with_madd4): Add validation.
31370 (all_defaults): Add madd4.
31371 * config/mips/mips.opt (mmadd4): New option.
31372 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31374 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31376 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
31377 (ISA_HAS_FUSED_MADD4): Likewise.
31378 * doc/invoke.texi (-mmadd4): Document the new option.
31379 * doc/install.texi (--with-madd4): Document the new option.
31381 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31383 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
31384 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
31385 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
31386 (aarch64_init_pauth_hint_builtins): New.
31387 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
31388 (aarch64_expand_builtin): Expand new builtins.
31390 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31392 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
31393 * combine-stack-adj.c (no_unhandled_cfa): Handle
31394 REG_CFA_TOGGLE_RA_MANGLE.
31395 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
31396 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
31397 info for return address signing.
31398 (aarch64_expand_epilogue): Likewise.
31400 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31402 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
31403 * config/aarch64/aarch64-protos.h
31404 (aarch64_return_address_signing_enabled): New declaration.
31405 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
31407 (aarch64_expand_prologue): Sign return address before it's pushed onto
31409 (aarch64_expand_epilogue): Authenticate return address fetched from
31411 (aarch64_override_options): Sanity check for ILP32 and ISA level.
31412 (aarch64_attributes): New function attributes for "sign-return-address".
31413 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
31414 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
31415 ("*do_return"): Generate combined instructions according to key index.
31416 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
31417 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
31419 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
31420 * config/aarch64/aarch64.opt (msign-return-address=): New.
31421 * doc/extend.texi (AArch64 Function Attributes): Documents
31422 "sign-return-address=".
31423 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
31425 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31427 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
31428 overall option summary.
31430 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31432 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
31433 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
31434 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
31435 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
31437 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
31439 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
31440 -mpower9-minmax by default for -mcpu=power9.
31441 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
31442 128-bit floating point.
31444 2017-01-20 Alan Modra <amodra@gmail.com>
31446 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
31447 optimizing for size.
31449 2017-01-20 Alan Modra <amodra@gmail.com>
31452 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
31453 for strcmp and strncmp from corresponding builtin decl.
31455 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31457 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
31458 instead of i386/rtems-64.h.
31459 * config/i386/rtems-64.h: Remove.
31461 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31465 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
31467 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
31469 2017-01-19 Tamar Christina <tamar.christina@arm.com>
31471 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
31472 Change int to HOST_WIDE_INT.
31473 * config/aarch64/aarch64-protos.h
31474 (aarch64_simd_gen_const_vector_dup): Likewise.
31475 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
31477 2017-01-19 David Malcolm <dmalcolm@redhat.com>
31479 * langhooks-def.h (lhd_type_for_size): New decl.
31480 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
31481 * langhooks.c (lhd_type_for_size): New function, taken from
31484 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
31486 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
31487 define_bypass for CR latency.
31488 (power9-cracked-alu): Update bypass latency and remove power9-branch.
31489 (power9-alu2): Add define_bypass for CR latency.
31491 (power9-mul): Update insn latency.
31492 (power9-mul-compare): Update insn latency, bypass latency and remove
31495 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31497 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
31499 * config/aarch64/aarch64.md
31500 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
31501 aarch64_nopcrelative_literal_loads.
31502 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
31504 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
31506 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
31507 TARGET_LOONGSON_3A.
31508 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
31510 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
31513 * config.gcc (supported_defaults): Add lxc1-sxc1.
31514 (with_lxc1_sxc1): Add validation.
31515 (all_defaults): Add lxc1-sxc1.
31516 * config/mips/mips.opt (mlxc1-sxc1): New option.
31517 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31519 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31520 __mips_no_lxc1_sxc1.
31521 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
31522 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
31523 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
31525 2017-01-19 Richard Biener <rguenther@suse.de>
31527 PR tree-optimization/72488
31528 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
31529 sure to restore SSA info.
31530 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
31532 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
31534 PR rtl-optimization/79121
31535 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
31536 of the inner type when shifting an extended value.
31538 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31541 * symtab.c (symtab_node::equal_address_to): Fix comparing of
31542 interposable aliases.
31544 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
31547 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
31548 Use the evmergelohi instruction.
31549 (mov_si<mode>_e500_subreg4_2_le): Likewise.
31550 (mov_sitf_e500_subreg8_2_be): Likewise.
31551 (mov_sitf_e500_subreg12_2_le): Likewise.
31552 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
31553 (mov_si<mode>_e500_subreg4_2_be): Likewise.
31554 (mov_sitf_e500_subreg8_2_le): Likewise.
31555 (mov_sitf_e500_subreg12_2_be): Likewise.
31557 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31559 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
31560 attribute from vecsimple to vecperm.
31561 (altivec_vbpermq2): Likewise.
31563 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31566 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
31568 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31569 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
31570 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
31571 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
31572 case where N arg is SIZE_MAX.
31573 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
31574 (cmpstrsi): Add pattern.
31576 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31578 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31579 __builtin_vec_revb builtins.
31580 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
31581 built-in functions to support generation of the ISA 3.0 XXBR<x>
31582 vector byte reverse instructions.
31583 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
31584 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
31585 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
31586 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
31587 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
31588 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
31589 (P9V_BUILTIN_VEC_REVB): Likewise.
31590 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
31591 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
31592 (p9_xxbrq_v16qi): Likewise.
31593 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
31594 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
31595 (p9_xxbrh_v8hi): Likewise.
31596 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
31597 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
31598 vec_revb built-in functions.
31600 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
31602 PR rtl-optimization/78952
31603 * config/i386/i386.md (any_extract): New code iterator.
31604 (*insvqi_2): Use any_extract for source operand.
31605 (*insvqi_3): Use any_shiftrt for source operand.
31607 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
31609 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
31611 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
31613 2017-01-18 Matthias Klose <doko@ubuntu.com>
31615 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
31617 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31619 * config/rs6000/altivec.h (vec_bperm): Change #define.
31620 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
31621 (altivec_vbpermq2): New define_insn.
31622 (altivec_vbpermd): Likewise.
31623 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
31624 function interface.
31625 (VBPERMD): Likewise.
31626 (VBPERM): New polymorphic function interface.
31627 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
31628 Add entries for P9V_BUILTIN_VEC_VBPERM.
31629 * doc/extend.texi: Add interfaces for vec_bperm.
31631 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31633 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
31634 first letter of error messages.
31635 (s390_resolve_overloaded_builtin): Likewise.
31636 * config/s390/s390.c (s390_expand_builtin): Likewise.
31637 (s390_invalid_arg_for_unprototyped_fn): Likewise.
31638 (s390_valid_target_attribute_inner_p): Likewise.
31639 * config/s390/s390.md ("tabort"): Likewise.
31641 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
31643 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
31644 (ISA_AVOID_DIV_HILO): New macro.
31645 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
31646 (ISA_HAS_DDIV): Likewise.
31648 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31650 * doc/invoke.texi (fabi-version): Correct number of occurrences.
31652 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31654 * doc/invoke.texi (fabi-version): Spelling fix.
31656 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31659 * doc/invoke.texi (fabi-version): Mention mangling fix for
31662 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31665 * doc/invoke.texi (fabi-version): Document discriminator mangling.
31667 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
31670 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
31671 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
31673 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
31675 (stack_protect_test): Ditto.
31676 * config/rs6000/rs6000.opt (mstack-protector-guard=,
31677 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
31679 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
31680 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
31681 -mstack-protector-guard-offset=.
31682 (RS/6000 and PowerPC Options): Ditto.
31684 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
31686 * config/i386/i386.h (MASK_CLASS_P): New define.
31687 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
31688 there are no registers from different register sets also when
31689 mask registers are used. Update function comment.
31690 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
31691 to (*k/*r) and (*k/*km) alternatives.
31693 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
31695 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
31696 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
31697 (EH_RETURN_HANDLER_RTX): New define.
31698 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
31699 Force frame pointer in EH return functions.
31700 (aarch64_expand_epilogue): Add barrier for eh_return.
31701 (aarch64_final_eh_return_addr): Remove.
31702 (aarch64_eh_return_handler_rtx): New function.
31703 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
31705 (aarch64_eh_return_handler_rtx): New prototype.
31707 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31709 * config/rs6000/altivec.h (vec_rlmi): New #define.
31710 (vec_vrlnm): Likewise.
31711 (vec_rlnm): Likewise.
31712 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
31713 (UNSPEC_VRLNM): Likewise.
31714 (VIlong): New mode iterator.
31715 (altivec_vrl<VI_char>mi): New define_insn.
31716 (altivec_vrl<VI_char>nm): Likewise.
31717 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
31719 (VRLDNM): Likewise.
31720 (RLNM): New polymorphic function entry.
31721 (VRLWMI): New monomorphic function entry.
31722 (VRLDMI): Likewise.
31723 (RLMI): New polymorphic function entry.
31724 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
31725 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
31726 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
31729 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31732 * dwarf2out.c (field_byte_offset): Restore the
31733 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
31734 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
31735 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
31738 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
31741 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
31743 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31746 * dwarf2out.c (add_data_member_location_attribute): For constant
31747 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
31748 instead of DW_AT_data_member_location, DW_AT_bit_offset and
31749 DW_AT_byte_size attributes.
31751 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
31753 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
31754 after forcing to constant memory when the code model is medium.
31756 2017-01-17 Julia Koval <julia.koval@intel.com>
31759 * config/i386/avx512fintrin.h
31760 (_mm512_i32gather_ps): Change __addr type to void const*.
31761 (_mm512_mask_i32gather_ps): Ditto.
31762 (_mm512_i32gather_pd): Ditto.
31763 (_mm512_mask_i32gather_pd): Ditto.
31764 (_mm512_i64gather_ps): Ditto.
31765 (_mm512_mask_i64gather_ps): Ditto.
31766 (_mm512_i64gather_pd): Ditto.
31767 (_mm512_mask_i64gather_pd): Ditto.
31768 (_mm512_i32gather_epi32): Ditto.
31769 (_mm512_mask_i32gather_epi32): Ditto.
31770 (_mm512_i32gather_epi64): Ditto.
31771 (_mm512_mask_i32gather_epi64): Ditto.
31772 (_mm512_i64gather_epi32): Ditto.
31773 (_mm512_mask_i64gather_epi32): Ditto.
31774 (_mm512_i64gather_epi64): Ditto.
31775 (_mm512_mask_i64gather_epi64): Ditto.
31776 (_mm512_i32scatter_ps): Change __addr type to void*.
31777 (_mm512_mask_i32scatter_ps): Ditto.
31778 (_mm512_i32scatter_pd): Ditto.
31779 (_mm512_mask_i32scatter_pd): Ditto.
31780 (_mm512_i64scatter_ps): Ditto.
31781 (_mm512_mask_i64scatter_ps): Ditto.
31782 (_mm512_i64scatter_pd): Ditto.
31783 (_mm512_mask_i64scatter_pd): Ditto.
31784 (_mm512_i32scatter_epi32): Ditto.
31785 (_mm512_mask_i32scatter_epi32): Ditto.
31786 (_mm512_i32scatter_epi64): Ditto.
31787 (_mm512_mask_i32scatter_epi64): Ditto.
31788 (_mm512_i64scatter_epi32): Ditto.
31789 (_mm512_mask_i64scatter_epi32): Ditto.
31790 (_mm512_i64scatter_epi64): Ditto.
31791 (_mm512_mask_i64scatter_epi64): Ditto.
31792 * config/i386/avx512pfintrin.h
31793 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
31794 (_mm512_mask_prefetch_i32gather_ps): Ditto.
31795 (_mm512_mask_prefetch_i64gather_pd): Ditto.
31796 (_mm512_mask_prefetch_i64gather_ps): Ditto.
31797 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
31798 (_mm512_prefetch_i32scatter_ps): Ditto.
31799 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
31800 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
31801 (_mm512_prefetch_i64scatter_pd): Ditto.
31802 (_mm512_prefetch_i64scatter_ps): Ditto.
31803 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
31804 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
31805 * config/i386/avx512vlintrin.h
31806 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
31807 (_mm_mmask_i32gather_ps): Ditto.
31808 (_mm256_mmask_i32gather_pd): Ditto.
31809 (_mm_mmask_i32gather_pd): Ditto.
31810 (_mm256_mmask_i64gather_ps): Ditto.
31811 (_mm_mmask_i64gather_ps): Ditto.
31812 (_mm256_mmask_i64gather_pd): Ditto.
31813 (_mm_mmask_i64gather_pd): Ditto.
31814 (_mm256_mmask_i32gather_epi32): Ditto.
31815 (_mm_mmask_i32gather_epi32): Ditto.
31816 (_mm256_mmask_i32gather_epi64): Ditto.
31817 (_mm_mmask_i32gather_epi64): Ditto.
31818 (_mm256_mmask_i64gather_epi32): Ditto.
31819 (_mm_mmask_i64gather_epi32): Ditto.
31820 (_mm256_mmask_i64gather_epi64): Ditto.
31821 (_mm_mmask_i64gather_epi64): Ditto.
31822 (_mm256_i32scatter_ps): Change __addr type to void*.
31823 (_mm256_mask_i32scatter_ps): Ditto.
31824 (_mm_i32scatter_ps): Ditto.
31825 (_mm_mask_i32scatter_ps): Ditto.
31826 (_mm256_i32scatter_pd): Ditto.
31827 (_mm256_mask_i32scatter_pd): Ditto.
31828 (_mm_i32scatter_pd): Ditto.
31829 (_mm_mask_i32scatter_pd): Ditto.
31830 (_mm256_i64scatter_ps): Ditto.
31831 (_mm256_mask_i64scatter_ps): Ditto.
31832 (_mm_i64scatter_ps): Ditto.
31833 (_mm_mask_i64scatter_ps): Ditto.
31834 (_mm256_i64scatter_pd): Ditto.
31835 (_mm256_mask_i64scatter_pd): Ditto.
31836 (_mm_i64scatter_pd): Ditto.
31837 (_mm_mask_i64scatter_pd): Ditto.
31838 (_mm256_i32scatter_epi32): Ditto.
31839 (_mm256_mask_i32scatter_epi32): Ditto.
31840 (_mm_i32scatter_epi32): Ditto.
31841 (_mm_mask_i32scatter_epi32): Ditto.
31842 (_mm256_i32scatter_epi64): Ditto.
31843 (_mm256_mask_i32scatter_epi64): Ditto.
31844 (_mm_i32scatter_epi64): Ditto.
31845 (_mm_mask_i32scatter_epi64): Ditto.
31846 (_mm256_i64scatter_epi32): Ditto.
31847 (_mm256_mask_i64scatter_epi32): Ditto.
31848 (_mm_i64scatter_epi32): Ditto.
31849 (_mm_mask_i64scatter_epi32): Ditto.
31850 (_mm256_i64scatter_epi64): Ditto.
31851 (_mm256_mask_i64scatter_epi64): Ditto.
31852 (_mm_i64scatter_epi64): Ditto.
31853 (_mm_mask_i64scatter_epi64): Ditto.
31854 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
31855 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
31856 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
31857 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
31858 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
31859 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
31860 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
31861 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
31862 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
31863 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
31864 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
31865 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
31866 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
31867 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
31868 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
31869 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
31870 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
31871 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
31872 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
31873 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
31874 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
31875 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
31876 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
31877 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
31878 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
31879 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
31880 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
31881 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
31882 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
31883 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
31884 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
31885 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
31886 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
31887 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
31888 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
31889 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
31890 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
31891 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
31892 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
31893 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
31894 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
31895 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
31896 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
31897 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
31898 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
31899 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
31900 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
31901 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
31902 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
31903 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
31904 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
31905 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
31906 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
31907 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
31908 definitions accordingly.
31910 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
31911 Kuan-Lin Chen <kuanlinchentw@gmail.com>
31914 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
31917 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
31920 * ira-conflicts.c (ira_build_conflicts): Update total conflict
31921 hard regs for inner regno.
31923 2017-01-17 Martin Liska <mliska@suse.cz>
31926 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
31927 assumption and add comment.
31929 2017-01-17 Nathan Sidwell <nathan@acm.org>
31931 * ipa-visibility.c (localize_node): New function, broken out of ...
31932 (function_and_variable_visibility): ... here. Call it.
31934 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31936 PR middle-end/77445
31937 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
31938 correctly set frequency of oudgoing edge.
31939 (duplicate_thread_path): Fix profile updating.
31941 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31944 * configure.ac: Add GCC_BASE_VER.
31945 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
31946 version from BASE-VER file.
31947 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
31948 (gcc.o): Depend on $(BASEVER).
31949 * common.opt (dumpfullversion): New option.
31950 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
31951 * doc/invoke.texi: Document -dumpfullversion.
31952 * doc/install.texi: Document --with-gcc-major-version-only.
31953 * configure: Regenerated.
31955 2017-01-17 Richard Biener <rguenther@suse.de>
31957 PR tree-optimization/71433
31958 * tree-vrp.c (register_new_assert_for): Merge same asserts
31959 on all incoming edges.
31960 (process_assert_insertions_for): Handle insertions at the
31963 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
31965 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
31966 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
31968 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
31971 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
31974 2017-01-17 Alan Modra <amodra@gmail.com>
31977 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
31978 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
31979 symbolic stack limit when pic.
31981 2017-01-16 Martin Sebor <msebor@redhat.com>
31983 PR tree-optimization/78608
31984 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
31986 2017-01-16 Jeff Law <law@redhat.com>
31989 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
31990 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
31991 for several include directories that may be relative to sysroot.
31992 * config/i386/x-mingw32 (gplus_includedir): Define.
31993 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
31994 (native_system_includedir): Likewise.
31995 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
31996 override if TARGET_SYSTEM_ROOT is defined.
31997 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
31999 PR tree-optimization/79090
32000 PR tree-optimization/33562
32001 PR tree-optimization/61912
32002 PR tree-optimization/77485
32003 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32004 and computed trims into the dump file.
32006 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32008 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32010 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32013 * gimplify.c (gimplify_init_constructor): If want_value and
32014 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32018 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32019 sequence. Formatting fixes.
32020 (doloop_optimize): Formatting fixes.
32023 * gcc.c (debug_level_greater_than_spec_func): New function.
32024 (static_spec_functions): Add debug-level-gt spec function.
32025 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32027 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32028 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32029 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32030 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32031 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32032 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32034 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32036 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32037 QImode fixups to general and mask registers only.
32039 2017-01-16 Carl Love <cel@us.ibm.com>
32041 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32042 for built-in functions
32043 vector signed char vec_nabs (vector signed char)
32044 vector signed short vec_nabs (vector signed short)
32045 vector signed int vec_nabs (vector signed int)
32046 vector signed long long vec_nabs (vector signed long long)
32047 vector float vec_nabs (vector float)
32048 vector double vec_nabs (vector double)
32049 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32051 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32052 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32053 * doc/extend.texi: Update the documentation file for the new built-in
32056 2017-01-16 Martin Sebor <msebor@redhat.com>
32058 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32061 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32063 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32064 UNSPEC_VSX__XXSPLTD to require special splat handling.
32066 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32069 * system.h: Poison strndup.
32071 2017-01-16 Alan Modra <amodra@gmail.com>
32074 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32077 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32079 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32081 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32083 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32084 call recog here. Assert that INSN_CODE (insn) is non-negative.
32086 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32089 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32091 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32092 in the currently scheduled RTL fragment.
32094 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32096 PR rtl-optimization/78751
32097 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32100 2017-01-14 Jeff Law <law@redhat.com>
32102 PR tree-optimization/79090
32103 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32104 variable length stores.
32105 (compute_trims): Delete dead assignment to *trim_tail.
32106 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32109 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32111 PR rtl-optimization/78626
32112 PR rtl-optimization/78727
32113 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32114 of a block, and split such blocks after everything else is finished.
32116 2017-01-14 Alan Modra <amodra@gmail.com>
32119 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32120 target legitimate_combined_insn.
32121 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32122 (rs6000_legitimate_combined_insn): New function.
32123 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32125 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32126 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32127 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32129 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32131 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32133 2017-01-13 Jeff Law <law@redhat.com>
32135 PR tree-optimization/33562
32136 PR tree-optimization/61912
32137 PR tree-optimization/77485
32138 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32140 (delete_dead_assignment): Likewise.
32141 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32142 statement to delete_dead_call and delete_dead_assignment.
32144 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32147 * substring-locations.c (format_warning_va): Strengthen case 1 so
32148 that both endpoints of the substring must be within the format
32149 range for just the substring to be printed.
32151 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32153 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32154 * config/i386/i386.c (ix86_target_string): Add missing options
32155 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32156 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32157 flags_other and ix86_target_other to flags2_other. Display unknown
32159 (ix86_valid_target_attribute_inner_p): Add missing options and
32160 reorder options by implied ISAs, as in ix86_target_string.
32162 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32164 * hash-table.h (hash_table::too_empty_p): New function.
32165 (hash_table::expand): Use it.
32166 (hash_table::traverse): Likewise.
32167 (hash_table::empty_slot): Use sizeof (value_type) instead of
32168 sizeof (PTR) to convert bytes to elements. Shrink the table
32169 if the current size is excessive for the current number of
32172 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32174 * ira-costs.c (record_reg_classes): Break from the inner loop
32175 early once alt_fail is known to be true. Update outer loop
32176 handling accordingly.
32178 2017-01-13 Jeff Law <law@redhat.com>
32180 * tree-ssa-dse.c (decrement_count): New function.
32181 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32182 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32183 when we know the partially dead statement is a mem* function.
32185 PR tree-optimization/61912
32186 PR tree-optimization/77485
32187 * tree-ssa-dse.c: Include expr.h.
32188 (maybe_trim_constructor_store): New function.
32189 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32191 PR tree-optimization/33562
32192 PR tree-optimization/61912
32193 PR tree-optimization/77485
32194 * doc/invoke.texi: Document new dse-max-object-size param.
32195 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32196 * tree-ssa-dse.c: Include params.h.
32197 (dse_store_status): New enum.
32198 (initialize_ao_ref_for_dse): New, partially extracted from
32200 (valid_ao_ref_for_dse, normalize_ref): New.
32201 (setup_live_bytes_from_ref, compute_trims): Likewise.
32202 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32203 (maybe_trim_partially_dead_store): Likewise.
32204 (maybe_trim_complex_store): Likewise.
32205 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32206 Track what bytes live from the original store. Return tri-state
32207 for dead, partially dead or live.
32208 (dse_dom_walker): Add constructor, destructor and new private members.
32209 (delete_dead_call, delete_dead_assignment): New extracted from
32211 (dse_optimize_stmt): Make a member of dse_dom_walker.
32212 Use initialize_ao_ref_for_dse.
32214 PR tree-optimization/33562
32215 PR tree-optimization/61912
32216 PR tree-optimization/77485
32217 * sbitmap.h (bitmap_count_bits): Prototype.
32218 (bitmap_clear_range, bitmap_set_range): Likewise.
32219 * sbitmap.c (bitmap_clear_range): New function.
32220 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32222 2017-01-13 Martin Liska <mliska@suse.cz>
32225 * function.c (set_cfun): Add new argument force.
32226 * function.h (set_cfun): Likewise.
32227 * ipa-inline-transform.c (inline_call): Use the function when
32228 strict alising from is dropped for function we inline to.
32230 2017-01-13 Richard Biener <rguenther@suse.de>
32232 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32233 for dumping GIMPLE INTEGER_CSTs.
32235 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32237 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32238 to 201112L since C++17.
32240 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32243 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32244 if -fsanitize=kernel-address is present.
32246 2017-01-13 Richard Biener <rguenther@suse.de>
32248 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32249 as _Literal ( type ) number in case usual suffixes do not
32250 preserve all information.
32252 2017-01-13 Richard Biener <rguenther@suse.de>
32254 PR tree-optimization/77283
32255 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
32256 and ssa-iterators.h.
32257 (is_feasible_trace): Implement a cost model based on joiner
32260 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
32263 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
32264 char or short to __float128/_Float128 directly.
32266 2017-01-12 Martin Sebor <msebor@redhat.com>
32268 to -Wformat-overflow.
32269 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
32270 (min_bytes_remaining): Same.
32271 (get_string_length): Same.
32272 (format_string): Same.
32273 (format_directive): Same.
32275 (pass_sprintf_length::handle_gimple_call): Same.
32277 2017-01-12 Jakub Jelinek <jakub@redhat.com>
32279 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
32280 info.nowrite calls with no lhs that can't throw. Return bool
32281 whether gsi_remove has been called or not.
32282 (pass_sprintf_length::handle_gimple_call): Return bool whether
32283 try_substitute_return_value called gsi_remove. Formatting fix.
32284 (pass_sprintf_length::execute): Don't use gsi_remove if
32285 handle_gimple_call returned true.
32288 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
32289 be removed due to side-effects, don't remove following barrier nor
32290 turn the successor edge into fallthru edge.
32292 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32295 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
32296 element-reversing loads and stores as not swappable.
32298 2017-01-12 Nathan Sidwell <nathan@acm.org>
32299 Nicolai Stange <nicstange@gmail.com>
32301 * combine.c (try_combine): Don't ignore result of overlap checking
32302 loop. Combine overlap & asm check into single loop.
32304 2017-01-12 Richard Biener <rguenther@suse.de>
32306 * tree-pretty-print.c (dump_generic_node): Provide -gimple
32307 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
32309 2017-01-12 Richard Biener <rguenther@suse.de>
32311 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
32312 and TS_TARGET_OPTION directly derive from TS_BASE.
32313 * tree-core.h (tree_optimization_option): Derive from tree_base.
32314 (tree_target_option): Likewise.
32316 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
32318 * config/i386/i386.c (memory_address_length): Increase len
32319 only when rip_relative_addr_p returns false.
32321 2017-01-11 Julia Koval <julia.koval@intel.com>
32323 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
32324 (OPTION_MASK_ISA_SGX_SET): New.
32325 (ix86_handle_option): Handle OPT_msgx.
32326 * config.gcc: Added sgxintrin.h.
32327 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
32328 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
32329 * config/i386/i386.c (ix86_target_string): Add -msgx.
32331 (ix86_option_override_internal): Handle new options.
32332 (ix86_valid_target_attribute_inner_p): Add sgx.
32333 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
32334 * config/i386/i386.opt: Add msgx.
32335 * config/i386/sgxintrin.h: New file.
32336 * config/i386/x86intrin.h: Add sgxintrin.h.
32338 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32341 * fold-const.c (maybe_nonzero_address): Return 1 for function
32343 (tree_single_nonzero_warnv_p): Don't handle function local objects
32347 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
32350 2017-01-11 David Malcolm <dmalcolm@redhat.com>
32353 * opts.c: Include "spellcheck.h"
32354 (struct string_fragment): New struct.
32355 (struct edit_distance_traits<const string_fragment &>): New
32357 (get_closest_sanitizer_option): New function.
32358 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
32360 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32362 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
32364 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
32365 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
32366 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
32367 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
32368 for initial die_offset if dwarf_split_debug_info.
32369 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
32370 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
32372 (output_skeleton_debug_sections): Formatting fix. Use
32373 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
32374 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
32376 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
32378 * config/arm/cortex-a53.md: Add bypasses for
32379 cortex_a53_r2f_cvt.
32380 (cortex_a53_r2f): Only use for transfers.
32381 (cortex_a53_f2r): Likewise.
32382 (cortex_a53_r2f_cvt): Add reservation for conversions.
32383 (cortex_a53_f2r_cvt): Likewise.
32385 2017-01-11 Tamar Christina <tamar.christina@arm.com>
32387 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
32388 to all inlined functions, change static to extern.
32390 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
32393 * config/arm/arm.c (legitimize_pic_address): Handle reference to
32395 (arm_assemble_integer): Likewise.
32397 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
32399 * config.gcc: Use new awk script to check CPU, FPU and architecture
32400 parameters for --with-... options.
32401 * config/arm/parsecpu.awk: New file
32402 * config/arm/arm-cpus.in: New file.
32403 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
32405 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
32407 * config/arm/t-arm: Update dependency rules.
32408 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
32409 of processing .def files.
32410 * config/arm/genopt.sh: Deleted.
32411 * config/arm/gentune.sh: Deleted.
32412 * config/arm/arm-cores.def: Deleted.
32413 * config/arm/arm-arches.def: Deleted.
32414 * config/arm/arm-fpus.def: Deleted.
32415 * config/arm/arm-tune.md: Regenerated.
32416 * config/arm/arm-tables.opt: Regenerated.
32417 * config/arm/arm-cpu.h: New generated file.
32418 * config/arm/arm-cpu-data.h: New generated file.
32419 * config/arm/arm-cpu-cdata.h: New generated file.
32421 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
32424 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
32426 (input_varpool_node): Unpack dynamically_initialized bit.
32428 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
32430 PR rtl-optimization/79032
32431 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
32432 the alignment of the adjusted memory reference against that of MODE,
32433 instead of the alignment of the original memory reference.
32435 2017-01-11 Martin Jambor <mjambor@suse.cz>
32437 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
32439 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
32440 decorated functions.
32442 2017-01-11 Richard Biener <rguenther@suse.de>
32444 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
32445 set range/nonnull info for PHI results. Do not set it on
32446 stmts marked for removal.
32448 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
32450 * expr.c (store_field): In the bitfield case, fetch the return value
32451 from the registers before applying a single big-endian adjustment.
32452 Always do a final load for a BLKmode value not larger than a word.
32454 2017-01-10 David Malcolm <dmalcolm@redhat.com>
32457 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32458 that we correctly handle column numbers greater than
32459 LINE_MAP_MAX_COLUMN_NUMBER.
32461 2017-01-10 Martin Sebor <msebor@redhat.com>
32463 PR middle-end/78245
32464 * gimple-ssa-sprintf.c (get_destination_size): Call
32465 {init,fini}object_sizes.
32466 * tree-object-size.c (addr_object_size): Adjust.
32467 (pass_through_call): Adjust.
32468 (pass_object_sizes::execute): Adjust.
32469 * tree-object-size.h (fini_object_sizes): Declare.
32471 2017-01-10 Martin Sebor <msebor@redhat.com>
32473 PR tree-optimization/78775
32474 * builtins.c (get_size_range): Move...
32475 * calls.c: ...to here.
32476 (alloc_max_size): Accept zero argument.
32477 (operand_signed_p): Remove.
32478 (maybe_warn_alloc_args_overflow): Call get_size_range.
32479 * calls.h (get_size_range): Declare.
32481 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
32483 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
32484 from TI's devices.csv file as of September 2016.
32485 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
32487 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
32489 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
32490 * doc/invoke.texi: Likewise.
32491 * doc/md.texi: Likewise.
32492 * doc/objc.texi: Likewise.
32494 2017-01-10 Joshua Conner <joshconner@google.com>
32496 * config/arm/fuchsia-elf.h: New file.
32497 * config/fuchsia.h: New file.
32498 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
32499 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
32501 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
32503 2016-01-10 Richard Biener <rguenther@suse.de>
32505 PR tree-optimization/79034
32506 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32507 Propagate out degenerate PHIs in the joiner.
32509 2017-01-10 Martin Liska <mliska@suse.cz>
32511 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
32512 (sort_congruence_classes_by_decl_uid): Likewise.
32513 (sort_congruence_class_groups_by_decl_uid): Likewise.
32514 (sem_item_optimizer::merge_classes): Sort class, groups in these
32515 classes and members in the groups by DECL_UID of declarations.
32516 This would make merge operations stable.
32518 2017-01-10 Martin Liska <mliska@suse.cz>
32520 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
32521 usage of m_classes_vec.
32522 (sem_item_optimizer::~sem_item_optimizer): Likewise.
32523 (sem_item_optimizer::get_group_by_hash): Likewise.
32524 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
32525 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
32526 (sem_item_optimizer::verify_classes): Likewise.
32527 (sem_item_optimizer::process_cong_reduction): Likewise.
32528 (sem_item_optimizer::dump_cong_classes): Likewise.
32529 (sem_item_optimizer::merge_classes): Likewise.
32530 * ipa-icf.h (congruence_class_hash): Rename from
32531 congruence_class_group_hash. Remove declaration of m_classes_vec.
32533 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
32535 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
32536 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
32537 * config.gcc: Add avx512vpopcntdqintrin.h.
32538 * config/i386/avx512vpopcntdqintrin.h: New.
32539 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
32540 * config/i386/i386-builtin-types.def: Add new types.
32541 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
32542 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
32543 __builtin_ia32_vpopcountq_v8di_mask): New.
32544 * config/i386/i386-c.c (ix86_target_macros_internal): Define
32545 __AVX512VPOPCNTDQ__.
32546 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
32547 (PTA_AVX512VPOPCNTDQ): Define.
32548 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
32549 TARGET_AVX512VPOPCNTDQ_P): Define.
32550 * config/i386/i386.opt: Add mavx512vpopcntdq.
32551 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
32552 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
32554 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32556 PR middle-end/77484
32557 * predict.def (PRED_CALL): Set to 67.
32559 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
32561 * expr.c (store_field): In the bitfield case, if the value comes from
32562 a function call and is of an aggregate type returned in registers, do
32563 not modify the field mode; extract the value in all cases if the mode
32564 is BLKmode and the size is not larger than a word.
32566 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
32569 * config/i386/cpuid.h: Fix undefined behavior.
32571 2017-01-04 Jeff Law <law@redhat.com>
32573 PR tree-optimization/79007
32574 PR tree-optimization/67955
32575 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
32576 conservative for pt.null when flag_non_call_exceptions is on.
32578 2017-01-09 Jakub Jelinek <jakub@redhat.com>
32580 PR translation/79019
32581 PR translation/79020
32582 * params.def (PARAM_INLINE_MIN_SPEEDUP,
32583 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
32584 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
32586 * config/avr/avr.opt (maccumulate-args): Likewise.
32587 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
32588 * common.opt (freport-bug): Likewise.
32589 * cif-code.def (CIF_FINAL_ERROR): Likewise.
32590 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
32591 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
32592 translatable string.
32593 * config/i386/i386.c (function_value_32): Likewise.
32594 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
32595 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
32597 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
32598 * common/config/msp430/msp430-common.c (msp430_handle_option):
32600 * symtab.c (symtab_node::verify_base): Likewise.
32601 * opts.c (set_debug_level): Likewise.
32602 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
32603 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
32604 missing whitespace to translatable strings.
32605 * config/avr/avr.md (bswapsi2): Fix typo in comment.
32606 * config/sh/superh.h: Likewise.
32607 * config/i386/xopintrin.h: Likewise.
32608 * config/i386/znver1.md: Likewise.
32609 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
32610 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
32611 * double-int.h (struct double_int): Likewise.
32612 * double-int.c (div_and_round_double): Likewise.
32613 * wide-int.cc: Likewise.
32614 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
32615 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
32616 * cfgcleanup.c (crossjumps_occured): Renamed to ...
32617 (crossjumps_occurred): ... this.
32618 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
32621 PR tree-optimization/78899
32622 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
32623 returning bool return struct loop *, NULL for failure and the new
32625 (versionable_outer_loop_p): Don't version outer loop if it has
32626 dont_vectorized bit set.
32627 (tree_if_conversion): When versioning outer loop, ensure
32628 tree_if_conversion is performed also on the inner loop of the
32629 non-vectorizable outer loop copy.
32630 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
32631 LOOP_VECTORIZED in inner loop of the scalar outer loop and
32632 prevent vectorization of it.
32633 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
32634 the outer loop vectorization of the non-scalar version is attempted
32635 before vectorization of the inner loop in scalar version. If
32636 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
32637 vectorization of its inner loop.
32638 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
32639 has 2 inner loops, rename also on edges from bb whose single pred
32640 is outer_loop->header. Fix typo in function comment.
32642 2017-01-09 Martin Sebor <msebor@redhat.com>
32645 * asan.c (asan_emit_stack_protection): Increase local buffer size
32646 to avoid snprintf truncation warning.
32648 2017-01-09 Andrew Pinski <apinski@cavium.com>
32650 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
32651 to reference thunderx2t99 for the tuning structure
32652 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
32654 (thunderx2t99_extra_costs): This.
32655 * config/aarch64/aarch64-tune.md: Regenerate.
32656 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
32657 (vulcan_addrcost_table): This.
32658 (vulcan_regmove_cost): Rename to ...
32659 (thunderx2t99_regmove_cost): This.
32660 (vulcan_vector_cost): Rename to ...
32661 (thunderx2t99_vector_cost): this.
32662 (vulcan_branch_cost): Rename to ...
32663 (thunderx2t99_branch_cost): This.
32664 (vulcan_tunings): Rename to ...
32665 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
32666 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
32668 2017-01-09 Martin Jambor <mjambor@suse.cz>
32672 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
32673 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
32674 (propagate_vr_accross_jump_function): Use the above function for all
32675 value range computations for pass-through jump functions and type
32676 converasion from explicit value range values.
32677 (ipcp_propagate_stage): Do not attempt to deduce types of formal
32678 parameters from TYPE_ARG_TYPES.
32679 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
32680 (ipa_write_node_info): Stream type of the actual argument.
32681 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
32683 2017-01-09 Martin Liska <mliska@suse.cz>
32686 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
32687 (lookup_compiler): Do not show error message with have_E.
32689 2017-01-09 Jakub Jelinek <jakub@redhat.com>
32691 PR tree-optimization/78938
32692 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
32693 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
32694 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
32695 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
32698 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32700 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
32703 2017-01-09 Richard Biener <rguenther@suse.de>
32705 PR tree-optimization/78997
32706 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
32707 name condition properly.
32709 2017-01-09 Richard Biener <rguenther@suse.de>
32712 * dwarf2out.c (is_cxx): New overload with context.
32713 (is_naming_typedef_decl): Use it.
32715 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32717 * invoke.texi (Option Summary): Correct spacing in option lists
32718 and add line breaks to fix over-long lines.
32720 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32722 PR middle-end/17660
32724 * extend.texi (Common Variable Attributes): Add xref to GCC
32725 Internals manual to explain mode attribute keywords.
32727 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32730 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
32731 and Preprocessor Options.
32732 (Options for Linking): Document -pthread here....
32733 (RS/6000 and PowerPC Options): ...not here.
32734 (Solaris 2 Options): ...or here.
32735 * doc/cppopts.texi: Document -pthread.
32737 2017-01-08 Martin Sebor <msebor@redhat.com>
32739 PR middle-end/77708
32740 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
32741 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
32742 New member functions.
32743 (format_directive): Used them.
32745 (pass_sprintf_length::handle_gimple_call): Same.
32746 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
32747 to avoid truncation for any argument.
32748 (extract_affine_mul): Same.
32749 * tree.c (get_file_function_name): Same.
32751 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32753 PR middle-end/77484
32754 * predict.def (PRED_INDIR_CALL): Set to 86.
32756 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32758 PR preprocessor/54124
32759 * doc/cppopts.texi: Reformat -d subtable to list the full name
32760 of the options. Add cross-reference to the docs for the general
32761 compiler -d options.
32762 * doc/invoke.texi (Developer Options): Add cross-reference to the
32763 preprocessor-specific -d option documentation.
32765 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32767 PR preprocessor/13498
32768 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
32769 redudant material, and reflect new command-line options.
32770 (System Headers): Likewise.
32772 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32774 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
32775 -isystem, and -idirafter. Copy-edit.
32776 * doc/cppopts.texi: Copy-edit. Remove contradiction about
32777 default for -ftrack-macro-expansion. Delete obsolete and
32778 badly-formatted implementation details about -fdebug-cpp output.
32779 * doc/cppwarnopts.texi: Copy-edit.
32781 2017-01-07 David Malcolm <dmalcolm@redhat.com>
32784 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32785 that the transition from a max line width >= 1<<10 to narrower
32786 lines works correctly.
32788 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
32790 * doc/options.texi (PerFunction): New.
32791 * opt-functions.awk (switch_flags): Map both Optimization and
32792 PerFunction to CL_OPTIMIZATION.
32793 * opth-gen.awk: Test for PerFunction flag along with
32795 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
32796 it only when the latter is present. Skip those that don't in
32797 the hash function generator.
32798 * common.opt (fvar-tracking): Mark as PerFunction instead of
32800 (fvar-tracking-assignments): Likewise.
32801 (fvar-tracking-assignments-toggle): Likewise.
32802 (fvar-tracking-uninit): Likewise.
32804 2017-01-07 Jakub Jelinek <jakub@redhat.com>
32806 PR translation/79018
32807 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
32810 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
32813 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
32814 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
32815 TARGET_LONG_JUMP_TABLE_OFFSETS.
32816 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
32817 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
32818 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
32819 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
32820 * config/m68k/m68k.md (tablejump expander): Likewise.
32821 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
32822 TARGET_LONG_JUMP_TABLE_OFFSETS.
32823 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
32824 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
32826 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
32827 David Holsgrove <david.holsgrove@xilinx.com>
32829 * common/config/microblaze/microblaze-common.c
32830 (TARGET_EXCEPT_UNWIND_INFO): Remove.
32831 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
32833 * config/microblaze/microblaze.c (microblaze_must_save_register)
32834 (microblaze_expand_epilogue, microblaze_return_addr): Handle
32836 (microblaze_eh_return): New function.
32837 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
32838 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
32839 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
32840 * config/microblaze/microblaze.md (eh_return): New pattern.
32842 2017-01-06 Jakub Jelinek <jakub@redhat.com>
32844 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
32845 GCC_DIAGNOSTIC_STRINGIFY): Define.
32847 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
32849 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32851 * config/arm/arm.md (<mcrr>): New.
32853 * config/arm/arm.c (arm_arch5te): New.
32854 (arm_option_override): Set arm_arch5te.
32855 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
32857 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
32858 (arm_mcrr_qualifiers): ... this. New.
32859 (MRRC_QUALIFIERS): Define to...
32860 (arm_mrrc_qualifiers): ... this. New.
32861 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
32863 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
32864 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
32865 (MRRCI, mrrc, MRRC): New.
32866 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
32867 VUNSPEC_MRRC2): New.
32869 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32871 * config/arm/arm.md (<mcr>): New.
32873 * config/arm/arm.c (arm_coproc_builtin_available): Add
32874 support for mcr, mrc, mcr2 and mrc2.
32875 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
32876 (arm_mcr_qualifiers): ... this. New.
32877 (MRC_QUALIFIERS): Define to ...
32878 (arm_mrc_qualifiers): ... this. New.
32879 (MCR_QUALIFIERS): Define to ...
32880 (arm_mcr_qualifiers): ... this. New.
32881 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
32883 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
32884 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
32885 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
32886 VUNSPEC_MRC2): New.
32888 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32890 * config/arm/arm.md (*ldc): New.
32894 * config/arm/arm.c (arm_coproc_builtin_available): Add
32895 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
32896 (arm_coproc_ldc_stc_legitimate_address): New.
32897 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
32898 'qualifier_const_pointer'.
32899 (LDC_QUALIFIERS): Define to...
32900 (arm_ldc_qualifiers): ... this. New.
32901 (STC_QUALIFIERS): Define to...
32902 (arm_stc_qualifiers): ... this. New.
32903 * config/arm/arm-protos.h
32904 (arm_coproc_ldc_stc_legitimate_address): New.
32905 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
32906 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
32907 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
32908 stc2, stcl, stc2l): New.
32909 * config/arm/constraints.md (Uz): New.
32910 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
32911 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
32912 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
32913 VUNSPEC_STC2L): New.
32915 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32917 * config/arm/arm.md (<cdp>): New.
32918 * config/arm/arm.c (neon_const_bounds): Rename this ...
32919 (arm_const_bounds): ... this.
32920 (arm_coproc_builtin_available): New.
32921 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
32922 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
32923 (CDP_QUALIFIERS): Define to...
32924 (arm_cdp_qualifiers): ... this. New.
32926 (arm_expand_builtin_args): Add case for 6 arguments.
32927 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
32928 (arm_const_bounds): ... this.
32929 (arm_coproc_builtin_available): New.
32930 * config/arm/arm_acle.h (__arm_cdp): New.
32932 * config/arm/arm_acle_builtins.def (cdp): New.
32934 * config/arm/iterators.md (CDPI,CDP,cdp): New.
32935 * config/arm/neon.md: Rename all 'neon_const_bounds' to
32936 'arm_const_bounds'.
32937 * config/arm/types.md (coproc): New.
32938 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
32939 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
32940 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
32941 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
32943 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32945 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
32946 (UBINOP_QUALIFIERS): New.
32948 (acle_builtin_data): New. Change comment.
32949 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
32950 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
32951 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
32952 arm_acle_builtins.def.
32953 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
32954 (arm_init_acle_builtins): New.
32955 (CRC32_BUILTIN): Remove.
32956 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
32957 crc32cb, crc32ch and crc32cw.
32958 (arm_init_crc32_builtins): Remove.
32959 (arm_init_builtins): Use arm_init_acle_builtins rather
32960 than arm_init_crc32_builtins.
32961 (arm_expand_acle_builtin): New.
32962 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
32963 * config/arm/arm_acle_builtins.def: New.
32965 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32967 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
32968 (arm_builtin_datum): ... this.
32969 (arm_init_neon_builtin): Rename to ...
32970 (arm_init_builtin): ... this. Add a new parameters PREFIX
32971 and USE_SIG_IN_NAME.
32972 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
32973 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
32974 'arm_builtin_datum'.
32975 (arm_init_vfp_builtins): Likewise.
32976 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
32977 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
32978 (arm_expand_neon_args): Rename to ...
32979 (arm_expand_builtin_args): ... this. Rename builtin_arg
32980 enum values and differentiate between ARG_BUILTIN_MEMORY
32981 and ARG_BUILTIN_NEON_MEMORY.
32982 (arm_expand_neon_builtin_1): Rename to ...
32983 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
32984 values, arm_expand_builtin_args and add bool parameter NEON.
32985 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
32986 (arm_expand_vfp_builtin): Likewise.
32987 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
32989 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32991 PR middle-end/77484
32992 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
32993 * predict.c (tree_estimate_probability_bb): Reverse direction of
32994 polymorphic call predictor.
32996 2017-01-06 David Malcolm <dmalcolm@redhat.com>
32998 * passes.c (execute_one_pass): Split out pass-skipping logic into...
32999 (determine_pass_name_match): ...this new function and...
33000 (should_skip_pass_p): ...this new function.
33002 2017-01-06 Nathan Sidwell <nathan@acm.org>
33004 * ipa-visibility.c (function_and_variable_visibility): Reformat
33005 comments and long lines. Remove extrneous if.
33006 * symtab.c (symtab_node::make_decl_local): Fix code format.
33007 (symtab_node::set_section_for_node): Fix comment typo.
33009 2017-01-06 Martin Liska <mliska@suse.cz>
33012 * lra-constraints.c: Rename invariant to lra_invariant.
33013 * predict.c (set_even_probabilities): Initialize e to NULL.
33015 2017-01-05 Martin Sebor <msebor@redhat.com>
33017 PR tree-optimization/78910
33018 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33019 (format_integer): Correct off-by-one error in the handling
33020 of precision with negative numbers in signed conversions..
33022 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33024 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33026 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33028 PR tree-optimization/71016
33029 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33030 factor_out_conditional_conversion. Formatting fix.
33031 (factor_out_conditional_conversion): Add cond_stmt argument.
33032 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33033 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33036 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33038 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33039 read-rtl-function.o, and selftest-rtl.o.
33040 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33041 (selftest::aarch64_test_loading_full_dump): New function.
33042 (selftest::aarch64_run_selftests): New function.
33043 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33044 selftest::aarch64_run_selftests.
33045 * config/i386/i386.c
33046 (selftest::ix86_test_loading_dump_fragment_1): New function.
33047 (selftest::ix86_test_loading_call_insn): New function.
33048 (selftest::ix86_test_loading_full_dump): New function.
33049 (selftest::ix86_test_loading_unspec): New function.
33050 (selftest::ix86_run_selftests): Call the new functions.
33051 * emit-rtl.c (maybe_set_max_label_num): New function.
33052 * emit-rtl.h (maybe_set_max_label_num): New decl.
33053 * function.c (instantiate_decls): Guard call to
33054 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33055 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33057 * gensupport.c (gen_reader::gen_reader): Pass "false"
33058 for new "compact" param of rtx_reader.
33059 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33060 rather than an empty string for NULL strings.
33061 * read-md.c: Potentially include config.h rather than bconfig.h.
33062 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33063 (have_error): New global, copied from errors.c.
33064 (md_reader::read_name): Rename to...
33065 (md_reader::read_name_1): ...this, adding "out_loc" param,
33066 and converting "missing name or number" to returning false, rather
33068 (md_reader::read_name): Reimplement in terms of read_name_1.
33069 (md_reader::read_name_or_nil): New function.
33070 (md_reader::read_string): Handle "(nil)" by returning NULL.
33071 (md_reader::md_reader): Add new param "compact".
33072 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33073 (md_reader::read_file): New method.
33074 * read-md.h (md_reader::md_reader): Add new param "compact".
33075 (md_reader::read_file): New method.
33076 (md_reader::is_compact): New accessor.
33077 (md_reader::read_name): Convert return type from void to file_location.
33078 (md_reader::read_name_or_nil): New decl.
33079 (md_reader::read_name_1): New decl.
33080 (md_reader::m_compact): New field.
33081 (noop_reader::noop_reader): Pass "false" for new "compact" param
33083 (rtx_reader::rtx_reader): Add new "compact" param.
33084 (rtx_reader::read_rtx_operand): Make virtual and convert return
33085 type from void to rtx.
33086 (rtx_reader::read_until): New decl.
33087 (rtx_reader::handle_any_trailing_information): New virtual function.
33088 (rtx_reader::postprocess): New virtual function.
33089 (rtx_reader::finalize_string): New virtual function.
33090 (rtx_reader::m_in_call_function_usage): New field.
33091 (rtx_reader::m_reuse_rtx_by_id): New field.
33092 * read-rtl-function.c: New file.
33093 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33094 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33095 (selftest::verify_three_block_rtl_cfg): New decl.
33096 * read-rtl-function.h: New file.
33097 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33098 For host, include function.h, memmodel.h, and emit-rtl.h.
33099 (one_time_initialization): New function.
33100 (struct compact_insn_name): New struct.
33101 (compact_insn_names): New array.
33102 (find_code): Handle insn codes in compact dumps.
33103 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33104 (bind_subst_iter_and_attr): Likewise.
33105 (add_condition_to_string): Likewise.
33106 (add_condition_to_rtx): Likewise.
33107 (apply_attribute_uses): Likewise.
33108 (add_current_iterators): Likewise.
33109 (apply_iterators): Likewise.
33110 (initialize_iterators): Guard usage of apply_subst_iterator with
33111 #ifdef GENERATOR_FILE.
33112 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33113 (md_reader::read_mapping): Likewise.
33114 (add_define_attr_for_define_subst): Likewise.
33115 (add_define_subst_attr): Likewise.
33116 (read_subst_mapping): Likewise.
33117 (check_code_iterator): Likewise.
33118 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33120 (one_time_initialization): New function.
33121 (rtx_reader::read_until): New method.
33122 (read_flags): New function.
33123 (parse_reg_note_name): New function.
33124 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33125 Handle reuse_rtx ids.
33126 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33127 Add parsing support for RTL dumps, mirroring the special-cases in
33128 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33129 values, and calling handle_any_trailing_information.
33130 (rtx_reader::read_rtx_operand): Convert return type from void
33131 to rtx, returning return_rtx. Handle case 'e'. Call
33132 finalize_string on XSTR and XTMPL fields.
33133 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33134 "(nil)" values were omitted. Call the postprocess vfunc on the
33136 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33137 class ctor. Initialize m_in_call_function_usage. Call
33138 one_time_initialization.
33139 * rtl-tests.c (selftest::test_uncond_jump): Call
33140 set_new_first_and_last_insn.
33141 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33142 * selftest-rtl.c: New file.
33143 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33144 (selftest::get_insn_by_uid): New decl.
33145 * selftest-run-tests.c (selftest::run_tests): Call
33146 read_rtl_function_c_tests.
33147 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33148 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33151 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33153 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33154 operands in a special way. Assert that pos+len <= mode precision.
33156 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33158 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33159 3 argument Alias with unlimited for the negative form.
33160 (fno-vect-cost-model): Removed.
33162 2017-01-05 Martin Liska <mliska@suse.cz>
33164 * hsa-gen.c (gen_hsa_divmod): New function.
33165 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33167 2017-01-05 Martin Liska <mliska@suse.cz>
33170 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33173 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33175 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33176 small constant length operands.
33178 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33180 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33181 between loop iterations.
33183 2017-01-05 Martin Liska <mliska@suse.cz>
33186 * gimplify.c (gimplify_decl_expr): Compare to
33187 asan_poisoned_variables instread of checking flags.
33188 (gimplify_target_expr): Likewise.
33189 (gimplify_expr): Likewise.
33190 (gimplify_function_tree): Conditionally initialize
33191 asan_poisoned_variables.
33193 2017-01-04 Jeff Law <law@redhat.com>
33195 PR tree-optimizatin/78812
33196 * rtl.h (contains_mem_rtx_p): Prototype.
33197 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33198 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33199 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33200 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33201 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33203 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33205 * input.c (assert_char_at_range): Default-initialize actual_range.
33207 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33209 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33210 to match the caller.
33212 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33214 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33215 insns after final jump in test to emit dummy move.
33217 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33219 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33220 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33222 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33224 * multiple_target.c (create_dispatcher_calls): Init e_next.
33225 * tree-ssa-loop-split.c (split_loop): Init border.
33226 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33229 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33234 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33235 (altivec_register_operand): Do not return true if the operand
33236 contains a SUBREG mixing SImode and SFmode.
33237 (vsx_register_operand): Likewise.
33238 (vsx_reg_sfsubreg_ok): New predicate.
33239 (vfloat_operand): Do not return true if the operand contains a
33240 SUBREG mixing SImode and SFmode.
33241 (vint_operand): Likewise.
33242 (vlogical_operand): Likewise.
33243 (gpc_reg_operand): Likewise.
33244 (int_reg_operand): Likewise.
33245 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33246 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33247 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33249 (rs6000_emit_move_si_sf_subreg): New helper function.
33250 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
33251 fixup SUBREGs involving SImode and SFmode.
33252 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
33253 numbers for the new peephole2 optimization.
33254 (peephole2 for SFmode unions): New peephole2 to optimize cases in
33255 the GLIBC math library that do AND/IOR/XOR operations on single
33256 precision floating point.
33257 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
33258 target macros to say whether we need to avoid SUBREGs mixing
33260 (TARGET_ALLOW_SF_SUBREG): Likewise.
33261 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
33262 (UNSPEC_SI_FROM_SF): Likewise.
33263 (iorxor): Change spacing.
33264 (and_ior_xor): New iterator for AND, IOR, and XOR.
33265 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
33266 (movdi_from_sf_zero_ext): Likewise.
33267 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
33268 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
33269 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
33270 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
33271 (fms<mode>4): Likewise.
33272 (fnma<mode>4): Likewise.
33273 (fnms<mode>4): Likewise.
33274 (nfma<mode>4): Likewise.
33275 (nfms<mode>4): Likewise.
33277 2017-01-04 Marek Polacek <polacek@redhat.com>
33280 * doc/invoke.texi: Document -Wpointer-compare.
33282 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33284 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
33287 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
33288 descriptions for -gdwarf-5 and emit them as uleb128 instead of
33291 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
33294 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
33295 documentation of the powerpc_popcntb_ok attribute.
33296 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33297 code to issue warning messages if a requested CPU configuration is
33298 not supported by the binary (assembler and loader) toolchain.
33299 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
33300 made to define a built-in function that has been disabled.
33301 (paired_init_builtins): Add assertion to prevent ICE if attempt is
33302 made to define a built-in function that has been disabled.
33303 (altivec_init_builtins): Add comment explaining why definition
33304 of the DST built-in functions is not preceded by an assertion
33305 check. Add assertions to prevent ICE if attempts are made to
33306 define an altivec predicate or an abs* built-in function that has
33308 (htm_init_builtins): Add comment explaining why definition of the
33309 htm built-in functions is not preceded by an assertion check.
33311 2017-01-04 Jeff Law <law@redhat.com>
33313 PR tree-optimizatin/67955
33314 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
33315 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
33316 the points-to solution does not include pt_null. Use DECL_PT_UID
33319 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
33321 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
33322 Use gen_int_mode instead of gen_lopwart for const_int operands.
33324 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33326 PR tree-optimization/71563
33327 * match.pd: Simplify X << Y into X if Y is known to be 0 or
33328 out of range value - has low bits known to be zero.
33330 2017-01-04 Alan Modra <amodra@gmail.com>
33332 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
33333 * configure: Regenerate.
33334 * config.in: Regenerate.
33336 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33339 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
33340 a substring of the message, but strcmp with the whole message. Ifdef
33341 ENABLE_NLS, translate the message first using dgettext.
33343 2017-01-03 Jeff Law <law@redhat.com>
33345 PR tree-optimizatin/78856
33346 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
33347 (mark_threaded_blocks): Remove code to truncate thread paths that
33348 cross multiple loop headers. Instead invalidate the cached loop
33349 iteration information and handle case of a thread path walking
33350 into an irreducible region.
33352 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
33355 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
33356 assertions. Add support for doing the signbit if the IEEE 128-bit
33357 floating point value is in a GPR.
33358 * config/rs6000/rs6000.md (Fsignbit): Delete.
33359 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
33360 Update the length attribute if the value is in a GPR.
33361 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
33362 the sign or zero extension instruction, since the value is always 0/1.
33363 (signbit<mode>2_dm2): Delete using <Fsignbit>.
33366 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
33367 extracting SImode to a GPR register so that we can generate a
33368 store, limit the vector to be in a traditional Altivec register
33369 for the vextuwrx instruction.
33371 2017-01-03 Ian Lance Taylor <iant@google.com>
33373 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
33375 2017-01-03 Martin Sebor <msebor@redhat.com>
33377 PR tree-optimization/78696
33378 * gimple-ssa-sprintf.c (format_floating): Correct handling of
33379 precision. Use MPFR for %f for greater fidelity. Correct handling
33381 (pass_sprintf_length::compute_format_length): Set width and precision
33382 specified by asrerisk to void_node for vararg functions.
33383 (try_substitute_return_value): Adjust dump output.
33385 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
33387 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
33389 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
33391 * doc/invoke.texi (SPARC options): Document -mlra as the default.
33392 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
33393 -mlra/-mno-lra was passed to the compiler.
33395 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
33397 PR rtl-optimization/65618
33398 * emit-rtl.c (try_split): Move initialization of "before" and
33399 "after" to just before the call to emit_insn_after_setloc.
33401 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
33403 * doc/md.texi (Standard Names): Remove reference to Java frontend.
33405 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
33407 * dwarf2out.c (gen_enumeration_type_die): When
33408 -gno-strict-dwarf, add a DW_AT_encoding attribute.
33410 2017-01-03 Jakub Jelinek <jakub@redhat.com>
33412 PR tree-optimization/78965
33413 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
33414 Change first argument from const call_info & to call_info &. For %n
33415 set info.nowrite to false.
33417 PR middle-end/78901
33418 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
33419 possibly throwing calls.
33421 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
33422 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
33423 and fns handling, rather than in a separate case SSA_NAME.
33425 2017-01-02 Jeff Law <law@redhat.com>
33427 * config/darwin-driver.c (darwin_driver_init): Const-correctness
33428 fixes for first_period and second_period variables.
33430 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
33433 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
33434 (*insvqi_1): New insn pattern.
33435 (*insvqi_1_mem_rex64): Ditto.
33436 (*insvqi_2): Ditto.
33437 (*insvqi_3): Rename from *insvqi.
33439 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
33441 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
33443 * doc/cfg.texi (Edges): Remove reference to Java.
33444 (Maintaining the CFG): Ditto.
33446 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33448 PR middle-end/77674
33449 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
33450 transparent aliases.
33452 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33454 PR middle-end/77484
33455 * predict.def (PRED_CALL): Update hitrate.
33456 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
33457 * predict.c (tree_estimate_probability_bb): Split CALL predictor
33458 into direct/indirect/polymorphic variants.
33460 2017-01-01 Jakub Jelinek <jakub@redhat.com>
33462 Update copyright years.
33464 * gcc.c (process_command): Update copyright notice dates.
33465 * gcov-dump.c (print_version): Ditto.
33466 * gcov.c (print_version): Ditto.
33467 * gcov-tool.c (print_version): Ditto.
33468 * gengtype.c (create_file): Ditto.
33469 * doc/cpp.texi: Bump @copying's copyright year.
33470 * doc/cppinternals.texi: Ditto.
33471 * doc/gcc.texi: Ditto.
33472 * doc/gccint.texi: Ditto.
33473 * doc/gcov.texi: Ditto.
33474 * doc/install.texi: Ditto.
33475 * doc/invoke.texi: Ditto.
33477 Copyright (C) 2017 Free Software Foundation, Inc.
33479 Copying and distribution of this file, with or without modification,
33480 are permitted in any medium without royalty provided the copyright
33481 notice and this notice are preserved.