1 2017-06-30 Nathan Sidwell <nathan@acm.org>
3 * ggc.h (empty_string): Delete.
4 * cfgexpand.c (expand_asm_stmt): Use plain "".
5 * optabs.c (expand_asm_memory_barrier): Likewise.
6 * stringpool.c (empty_string): Delete.
7 (digit_vector, digit_string): Delete.
8 (ggc_alloc_string): Use plain "", don't optimize single digit
9 strings. Use ggc_alloc_atomic.
11 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
13 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
14 comparison set and one other set, use the cost of the non-comparison
17 2017-06-30 Nathan Sidwell <nathan@acm.org>
19 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
22 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
24 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25 loops. Remove now unneeded calls to gimple_switch_set_label() that
26 just set removed labels to NULL_TREE.
28 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
30 * tree-ssanames.c (set_range_info_raw): Abstract from ...
31 (set_range_info): ...here. Only call set_range_info_raw if domain
33 (set_nonzero_bits): Call set_range_info_raw.
34 * tree-ssanames.h (set_range_info_raw): New.
36 2017-06-30 Jakub Jelinek <jakub@redhat.com>
39 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
40 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
41 of nonimmediate_operand and <store_mask_constraint> instead of m
42 for the input operand. For V8FI iterator, always split if input
43 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
44 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
45 <store_mask_predicate> instead of register_operand and
46 <store_mask_constraint> instead of v for the input operand. Make
47 sure both operands aren't MEMs for if not <mask_applied>.
49 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
51 * lto-wrapper.c (copy_file) Close both file descriptors before
54 2017-06-30 Martin Liska <mliska@suse.cz>
57 * multiple_target.c (create_dispatcher_calls): Make ifunc
58 also for function that don't have calls or are not referenced.
60 2017-06-30 Richard Biener <rguenther@suse.de>
62 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
63 analyze the first scalar stmt. Move vector type computation
64 for the BB case here from ...
65 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
66 live operation processing in the SLP case properly.
68 2017-06-30 Richard Biener <rguenther@suse.de>
70 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
72 2017-06-30 Martin Liska <mliska@suse.cz>
75 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
76 before BUILT_IN_UNWIND_RESUME when ASAN is used.
78 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
80 * doc/invoke.texi (AArch64): Add missing options and remove redundant
83 2017-06-30 Richard Biener <rguenther@suse.de>
85 PR tree-optimization/81249
86 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
87 condition reduction result to original scalar type.
89 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
91 * profile-count.h (enum profile_quality): Fix typos and whitespace
94 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
96 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
97 type for branch probabilities.
99 2017-06-29 Julian Brown <julian@codesourcery.com>
100 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
102 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
103 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
104 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
105 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
107 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
109 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
110 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
111 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
112 CC usage from generic code to here.
113 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
114 CC usage into the target macros.
116 2017-06-29 Maya Rashish <coypu@sdf.org>
118 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
121 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
123 * arm/arm-builtins.c: Include profile-count.h
124 * except.c (sjlj_emit_function_enter): Use
125 profile_probability::unlikely.
127 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
129 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
130 and tocrel_offset be pointer args rather than implicitly using
132 (legitimate_constant_pool_address_p, rs6000_emit_move,
133 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
134 tocrel_offset and use in toc_relative_expr_p call.
135 (print_operand, print_operand_address): Use static tocrel_base_oac
136 and tocrel_offset_oac.
137 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
140 2017-06-29 Maya Rashish <coypu@sdf.org>
142 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
144 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
146 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
147 objects, take into account only the alignment of 'op0' and 'mode1' if
150 2017-06-29 Steve Ellcey <sellcey@cavium.com>
152 * ccmp.c (ccmp_tree_comparison_p): New function.
153 (ccmp_candidate_p): Update to use above function.
154 (get_compare_parts): New function.
155 (expand_ccmp_next): Update to use new functions.
156 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
158 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
159 take mode as argument.
160 * ccmp.h (expand_ccmp_expr): Add mode as argument.
161 * expr.c (expand_expr_real_1): Pass mode as argument.
163 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
165 * combine.c (combine_instructions): Print insns to dump_file, together
168 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
170 * asan.c (asan_emit_stack_protection): Update.
171 (create_cond_insert_point): Update.
172 * auto-profile.c (afdo_propagate_circuit): Update.
173 * basic-block.h (struct edge_def): Turn probability to
175 (EDGE_FREQUENCY): Update.
176 * bb-reorder.c (find_traces_1_round): Update.
177 (better_edge_p): Update.
178 (sanitize_hot_paths): Update.
179 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
180 (make_single_succ_edge): Update.
181 (check_bb_profile): Update.
182 (dump_edge_info): Update.
183 (update_bb_profile_for_threading): Update.
184 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
185 probabilitycount to 0.
186 * cfgbuild.c (compute_outgoing_frequencies): Update.
187 * cfgcleanup.c (try_forward_edges): Update.
188 (outgoing_edges_match): Update.
189 (try_crossjump_to_edge): Update.
190 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
191 (expand_gimple_tailcall): Update.
192 (construct_init_block): Use make_single_succ_edge.
193 (construct_exit_block): Use make_single_succ_edge.
194 * cfghooks.c (verify_flow_info): Update.
195 (redirect_edge_succ_nodup): Update.
196 (split_edge): Update.
197 (account_profile_record): Update.
198 * cfgloopanal.c (single_likely_exit): Update.
199 * cfgloopmanip.c (scale_loop_profile): Update.
200 (set_zero_probability): Remove.
201 (duplicate_loop_to_header_edge): Update.
202 * cfgloopmanip.h (loop_version): Update prototype.
203 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
204 (force_nonfallthru_and_redirect): Update.
205 (update_br_prob_note): Update.
206 (rtl_verify_edges): Update.
207 (purge_dead_edges): Update.
208 (rtl_lv_add_condition_to_bb): Update.
209 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
210 * cgraphunit.c (init_lowered_empty_function): Update.
211 (cgraph_node::expand_thunk): Update.
212 * cilk-common.c: Include profile-count.h
213 * dojump.c (inv): Remove.
215 (jumpifnot_1): Update.
218 (do_jump_by_parts_greater_rtx): Update.
219 (do_compare_rtx_and_jump): Update.
220 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
221 do_jump_1. do_compare_rtx_and_jump): Update prototype.
222 * dwarf2cfi.c: Include profile-count.h
223 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
224 (sjlj_emit_dispatch_table): Likewise.
225 * explow.c: Include profile-count.h
226 * expmed.c (emit_store_flag_force): Update.
227 (do_cmp_and_jump): Update.
228 * expr.c (compare_by_pieces_d::generate): Update.
229 (compare_by_pieces_d::finish_mode): Update.
230 (emit_block_move_via_loop): Update.
231 (store_expr_with_bounds): Update.
232 (store_constructor): Update.
233 (expand_expr_real_2): Update.
234 (expand_expr_real_1): Update.
235 * expr.h (try_casesi, try_tablejump): Update prototypes.
236 * gimple-pretty-print.c (dump_probability): Update.
238 (dump_gimple_label): Update.
239 (dump_gimple_bb_header): Update.
240 * graph.c (draw_cfg_node_succ_edges): Update.
241 * hsa-gen.c (convert_switch_statements): Update.
242 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
243 (find_if_case_1): Update.
244 (find_if_case_2): Update.
245 * internal-fn.c (expand_arith_overflow_result_store): Update.
246 (expand_addsub_overflow): Update.
247 (expand_neg_overflow): Update.
248 (expand_mul_overflow): Update.
249 (expand_vector_ubsan_overflow): Update.
250 * ipa-cp.c (good_cloning_opportunity_p): Update.
251 * ipa-split.c (split_function): Use make_single_succ_edge.
252 * ipa-utils.c (ipa_merge_profiles): Update.
253 * loop-doloop.c (add_test): Update.
254 (doloop_modify): Update.
255 * loop-unroll.c (compare_and_jump_seq): Update.
256 (unroll_loop_runtime_iterations): Update.
257 * lra-constraints.c (lra_inheritance): Update.
258 * lto-streamer-in.c (input_cfg): Update.
259 * lto-streamer-out.c (output_cfg): Update.
260 * mcf.c (adjust_cfg_counts): Update.
261 * modulo-sched.c (sms_schedule): Update.
262 * omp-expand.c (expand_omp_for_init_counts): Update.
263 (extract_omp_for_update_vars): Update.
264 (expand_omp_ordered_sink): Update.
265 (expand_omp_for_ordered_loops): Update.
266 (expand_omp_for_generic): Update.
267 (expand_omp_for_static_nochunk): Update.
268 (expand_omp_for_static_chunk): Update.
269 (expand_cilk_for): Update.
270 (expand_omp_simd): Update.
271 (expand_omp_taskloop_for_outer): Update.
272 (expand_omp_taskloop_for_inner): Update.
273 * omp-simd-clone.c (simd_clone_adjust): Update.
274 * optabs.c (expand_doubleword_shift): Update.
275 (expand_abs): Update.
276 (emit_cmp_and_jump_insn_1): Update.
277 (expand_compare_and_swap_loop): Update.
278 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
279 * predict.c (predictable_edge_p): Update.
280 (edge_probability_reliable_p): Update.
281 (set_even_probabilities): Update.
282 (combine_predictions_for_insn): Update.
283 (combine_predictions_for_bb): Update.
284 (propagate_freq): Update.
285 (estimate_bb_frequencies): Update.
286 (force_edge_cold): Update.
287 * profile-count.c (profile_count::dump): Add missing space into dump.
288 (profile_count::debug): Add newline.
289 (profile_count::differs_from_p): Explicitly convert to unsigned.
290 (profile_count::stream_in): Update.
291 (profile_probability::dump): New member function.
292 (profile_probability::debug): New member function.
293 (profile_probability::differs_from_p): New member function.
294 (profile_probability::differs_lot_from_p): New member function.
295 (profile_probability::stream_in): New member function.
296 (profile_probability::stream_out): New member function.
297 * profile-count.h (profile_count_quality): Rename to ...
298 (profile_quality): ... this one.
299 (profile_probability): New.
300 (profile_count): Update.
301 * profile.c (compute_branch_probabilities): Update.
302 * recog.c (peep2_attempt): Update.
303 * sched-ebb.c (schedule_ebbs): Update.
304 * sched-rgn.c (find_single_block_region): Update.
305 (compute_dom_prob_ps): Update.
306 (schedule_region): Update.
307 * sel-sched-ir.c (compute_succs_info): Update.
308 * stmt.c (struct case_node): Update.
309 (do_jump_if_equal): Update.
310 (get_outgoing_edge_probs): Update.
311 (conditional_probability): Update.
312 (emit_case_dispatch_table): Update.
313 (expand_case): Update.
314 (expand_sjlj_dispatch_table): Update.
315 (emit_case_nodes): Update.
316 * targhooks.c: Update.
317 * tracer.c (better_p): Update.
318 (find_best_successor): Update.
319 * trans-mem.c (expand_transaction): Update.
320 * tree-call-cdce.c: Update.
321 * tree-cfg.c (gimple_split_edge): Upate.
322 (move_sese_region_to_fn): Upate.
323 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
324 * tree-eh.c (lower_resx): Upate.
325 (cleanup_empty_eh_move_lp): Upate.
326 * tree-if-conv.c (version_loop_for_if_conversion): Update.
327 * tree-inline.c (copy_edges_for_bb): Update.
328 (copy_cfg_body): Update.
329 * tree-parloops.c (gen_parallel_loop): Update.
330 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
331 (gimple_gen_time_profiler): Update.
332 * tree-ssa-dce.c (remove_dead_stmt): Update.
333 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
334 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
335 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
336 (unloop_loops): Update.
337 (try_peel_loop): Update.
338 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
339 * tree-ssa-loop-split.c (connect_loops): Update.
340 (split_loop): Update.
341 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
342 (hoist_guard): Update.
343 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
344 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
345 (value_replacement): Update.
346 * tree-ssa-reassoc.c (branch_fixup): Update.
347 * tree-ssa-tail-merge.c (replace_block_by): Update.
348 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
349 (create_edge_and_update_destination_phis): Update.
350 (compute_path_counts): Update.
351 (recompute_probabilities): Update.
352 (update_joiner_offpath_counts): Update.
353 (freqs_to_counts_path): Update.
354 (duplicate_thread_path): Update.
355 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
356 (struct switch_conv_info): Update.
357 (gen_inbound_check): Update.
358 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
359 (vect_do_peeling): Update.
360 (vect_loop_versioning): Update.
361 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
362 (optimize_mask_stores): Update.
363 * ubsan.c (ubsan_expand_null_ifn): Update.
364 * value-prof.c (gimple_divmod_fixed_value): Update.
365 (gimple_divmod_fixed_value_transform): Update.
366 (gimple_mod_pow2): Update.
367 (gimple_mod_pow2_value_transform): Update.
368 (gimple_mod_subtract): Update.
369 (gimple_mod_subtract_transform): Update.
371 (gimple_stringop_fixed_value): Update.
372 (gimple_stringops_transform): Update.
373 * value-prof.h: Update.
375 2017-06-29 Carl Love <cel@us.ibm.com>
377 * config/rs6000/rs6000-c.c: Add support for built-in functions
378 vector signed int vec_signed (vector float);
379 vector signed long long vec_signed (vector double);
380 vector signed int vec_signed2 (vector double, vector double);
381 vector signed int vec_signede (vector double);
382 vector signed int vec_signedo (vector double);
383 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
384 instruction generator.
385 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
386 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
387 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
389 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
390 vunsignede_v2df): Add define_expands.
391 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
392 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
393 VEC_UNSIGNEDO): Add definitions.
394 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
395 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
396 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
397 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
398 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
399 * config/rs6000/altivec.h (vec_signed, vec_signed2,
400 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
401 vec_unsignede, vec_unsignedo): Add builtin defines.
402 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
404 * doc/extend.texi: Update the built-in documentation file for the
405 new built-in functions.
407 2017-06-29 Richard Biener <rguenther@suse.de>
409 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
410 reduction chains to LOOP_VINFO_REDUCTIONS.
411 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
412 SLP reductions after processing reduction chains.
414 2017-06-29 Nathan Sidwell <nathan@acm.org>
416 * builtins.c (fold_builtin_FUNCTION): Use
417 lang_hooks.decl_printable_name.
419 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
422 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
424 * stmt.c (expand_case): Assert NCASES is greater than one.
426 2017-06-29 Richard Biener <rguenther@suse.de>
428 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
430 (group_case_labels): Likewise.
431 (find_taken_edge): Push sanity checking on val to workers...
432 (find_taken_edge_cond_expr): ... here
433 (find_taken_edge_switch_expr): ... and here, handle cases
434 with just a default label.
435 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
436 (group_case_labels): Likewise.
437 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
438 group_case_labels does anything cleanup the CFG again.
440 2017-06-29 Bin Cheng <bin.cheng@arm.com>
442 PR tree-optimization/81196
443 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
444 exit condition comparing two IVs.
446 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
448 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
449 profile to the dummy entry at the end of the list of architectures.
450 * config/arm/arm-cpu-cdata.h: Regenerated.
452 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
453 Michael Collison <michael.collison@arm.com>
456 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
458 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
459 (*aarch64_reg_<mode>3_minus_mask): New pattern.
460 (*aarch64_<optab>_reg_di3_mask2): New pattern.
461 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
462 of shift when the shift amount is masked with constant equal to
463 the size of the mode.
464 * config/aarch64/predicates.md (subreg_lowpart_operator): New
467 2017-06-29 Martin Liska <mliska@suse.cz>
469 * config/i386/i386.opt: Change range from [1,5] to [0,5].
471 2017-06-29 Yury Gribov <tetra2005@gmail.com>
474 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
477 (edge_growth_cache_entry::edge_growth_cache_entry): New
479 (reset_edge_growth_cache): Update to use constructor.
481 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
483 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
484 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
485 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
487 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
489 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
490 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
492 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
494 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
495 (*-linux-uclibc*): Add t-uclibc tmake_file.
496 * config/t-musl: New.
497 * config/t-uclibc: New.
499 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
501 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
503 (gen_comm_data): Emit architectural setting of arch_prof.
504 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
506 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
507 (armv8-m.base, armv8-m.main): Likewise.
508 * arm-protos.h (arm_build_target): Add profile field.
509 (arch_option): Likewise.
510 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
512 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
513 arm_active_target.profile.
515 2017-06-28 Richard Biener <rguenther@suse.de>
518 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
520 * match.pd (negate_expr_p): Likewise.
521 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
522 fold_build2, not fold_binary.
524 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
526 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
527 Convert memory address to Pmode.
528 (aarch64_print_operand): Assert MEM operands are always Pmode.
530 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
533 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
535 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
536 * config/arm/aarch-common-protos.h
537 (aarch_forward_to_shift_is_not_shifted_re): Remove.
538 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
540 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
543 * multiple_target.c (create_dispatcher_calls): Set the default
544 clone to be static, not public.
546 2017-06-28 Richard Biener <rguenther@suse.de>
548 * tree-vect-loop.c (vectorizable_reduction): Move special
549 cond reduction IV var creation ...
550 (vect_create_epilog_for_reduction): ... here. Remove induction_index
551 parameter. Use STMT_VINFO_VECTYPE.
552 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
555 2017-06-28 Martin Liska <mliska@suse.cz>
558 * ipa-visibility.c (non_local_p): Handle visibility.
560 2017-06-28 Martin Liska <mliska@suse.cz>
563 * common.opt: Add IntegerRange to various options.
564 * opt-functions.awk (integer_range_info): New function.
565 * optc-gen.awk: Add integer_range_info to cl_options struct.
566 * opts-common.c (decode_cmdline_option): Handle
567 CL_ERR_INT_RANGE_ARG.
568 (cmdline_handle_error): Likewise.
569 * opts.c (print_filtered_help): Show valid interval in
570 when --help is provided.
571 * opts.h (struct cl_option): Add range_min and range_max fields.
572 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
574 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
576 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
577 (x * C EQ/NE y * C): New transformation.
579 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
581 * genmultilib (combination_space): Accept '+' in option names.
583 2017-06-28 Martin Liska <mliska@suse.cz>
586 * asan.c (instrument_derefs): Bail out inner references
587 that are hard register variables.
589 2017-06-28 Jakub Jelinek <jakub@redhat.com>
592 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
593 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
595 2017-06-28 Richard Biener <rguenther@suse.de>
597 * tree-vectorizer.h (vect_get_vec_defs): Remove.
598 (vect_get_slp_defs): Adjust.
599 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
601 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
603 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
604 get_initial_defs_for_reduction instead of vect_get_vec_defs.
605 (vectorizable_reduction): Adjust.
606 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
608 (vect_get_slp_defs): Likewise.
609 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
610 (vectorizable_bswap): Adjust.
611 (vectorizable_call): Likewise.
612 (vectorizable_conversion): Likewise.
613 (vectorizable_assignment): Likewise.
614 (vectorizable_shift): Likewise.
615 (vectorizable_operation): Likewise.
616 (vectorizable_store): Likewise.
617 (vectorizable_condition): Likewise.
618 (vectorizable_comparison): Likewise.
620 2017-06-28 Michael Collison <michael.collison@arm.com>
623 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
625 (arm_gen_movmemqi): Removed unused variable 'i'.
626 Convert 'for' loop into 'while' loop.
627 (arm_expand_prologue): Remove last unnecessary set of insn.
628 (thumb_pop): Remove unused variable 'pushed_words'.
629 (thumb_exit): Remove last unnecessary set of regs_to_pop.
631 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
633 * config/s390/predicates.md: Use s390_rel_address_ok_p.
634 * config/s390/s390-protos.h: Add prototype of
635 s390_rel_address_ok_p.
636 * config/s390/s390.c (s390_got_symbol): New function.
637 (s390_rel_address_ok_p): New function.
638 (legitimize_pic_address): Use s390_rel_address_ok_p.
639 (s390_load_got): Use s390_got_symbol.
640 (s390_option_override): Issue error if
641 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
642 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
644 * config/s390/s390.opt: New option mpic-data-is-text-relative.
646 2017-06-27 Andrew Pinski <apinski@cavium.com>
648 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
649 (X * copysign (1.0, X)): New pattern.
650 (X * copysign (1.0, -X)): New pattern.
651 (copysign (-1.0, CST)): New pattern.
653 2017-06-27 Joseph Myers <joseph@codesourcery.com>
655 * genmultilib (combination_space): Remove variable.
656 Validate reuse rules against regular expression for any sequence
657 of multilib options in any order.
659 2017-06-27 Michael Collison <michael.collison@arm.com>
661 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
662 call aarch64_split_simd_combine.
663 * (aarch64_combine_internal<mode>): Delete pattern.
664 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
665 Allow register and subreg operands.
667 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
669 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
670 specific need, just fallback on defaults.
671 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
673 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
674 Olivier Hainque <hainque@adacore.com>
676 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
678 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
679 targets. Pick a default if no particular attempt applied.
680 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
683 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
685 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
686 (x86_64-wrs-vxworks7): Likewise.
688 2017-06-27 Marek Polacek <polacek@redhat.com>
691 * ubsan.c (instrument_null): Check get_base_address's result for null.
693 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
695 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
697 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
699 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
700 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
701 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
702 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
704 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
705 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
706 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
707 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
708 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
709 BUILT_IN_FEUPDATEENV): New builtins.
710 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
711 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
712 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
713 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
715 (builtin_structptr_types): Adjust size.
716 * tree.c (builtin_structptr_types): Add four entries.
718 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
719 Olivier Hainque <hainque@adacore.com>
721 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
722 (TLS_SYM): New local macro, forcing reference to __tls__ on
723 link command lines for VxWorks 7 RTPs, triggering initialization
725 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
726 OS features TLS support, true for RTPs on VxWorks 7.
727 * config/vxworks.c (vxworks_override_options): Setup emutls
730 2017-06-27 Jakub Jelinek <jakub@redhat.com>
732 * predict.c (test_prediction_value_range): Use -1U instead of -1
733 to avoid narrowing conversion warning.
734 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
735 to avoid narrowing conversion warning.
736 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
738 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
740 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
742 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
743 64bit configurations.
744 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
745 (SIZE_TYPE): Likewise.
746 * config/vxworks.c (vxworks_emutls_var_fields): Use
747 long_unsigned_type_node instead of unsigned_type_node as the offset
748 field type, which is "pointer" mode in emutls.c.
750 2017-06-27 Jakub Jelinek <jakub@redhat.com>
753 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
756 * gimple-fold.c (replace_call_with_call_and_fold): Handle
757 gimple_vuse copying separately from gimple_vdef copying.
759 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
761 * value-prof.c (free_hist): Remove call to memset and the enclosing if
764 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
765 Olivier Hainque <hainque@adacore.com>
767 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
768 for all vxworks7 targets.
769 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
770 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
771 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
772 variations for VX6/VX7 and 32/64bits later on in ...
773 (VXWORKS_LIB_SPEC): Leverage new macros.
774 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
775 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
777 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
779 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
780 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
782 2017-06-26 Carl Love <cel@us.ibm.com>
784 * config/rs6000/rs6000-c.c: Add support for built-in functions
785 vector bool char vec_reve (vector bool char);
786 vector signed char vec_reve (vector signed char);
787 vector unsigned char vec_reve (vector unsigned char);
788 vector bool int vec_reve (vector bool int);
789 vector signed int vec_reve (vector signed int);
790 vector unsigned int vec_reve (vector unsigned int);
791 vector bool long long vec_reve (vector bool long long);
792 vector signed long long vec_reve (vector signed long long);
793 vector unsigned long long vec_reve (vector unsigned long long);
794 vector bool short vec_reve (vector bool short);
795 vector signed short vec_reve (vector signed short);
796 vector double vec_reve (vector double);
797 vector float vec_reve (vector float);
798 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
799 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
800 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
801 (altivec_vreve): New pattern.
802 * config/rs6000/altivec.h (vec_reve): New define.
803 * doc/extend.texi (vec_rev): Update the built-in documentation file
804 for the new built-in functions.
806 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
808 PR tree-optimization/71815
809 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
811 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
813 (slsr_process_phi): Likewise.
814 (replace_uncond_cands_and_profitable_phis): Don't replace a
815 multiply candidate with a stride of 1 (copy or cast).
816 (phi_incr_cost): Call uses_consumed_by_stmt rather than
818 (lowest_cost_path): Likewise.
819 (total_savings): Likewise.
821 2017-06-26 Richard Biener <rguenther@suse.de>
824 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
825 Use def_builtin_pure for all gather builtins.
827 2017-06-26 Richard Biener <rguenther@suse.de>
829 PR tree-optimization/81203
830 * tree-tailcall.c (find_tail_calls): Do not move stmts into
833 2017-06-26 Marek Polacek <polacek@redhat.com>
836 * doc/invoke.texi: Document -Wmultistatement-macros.
838 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
840 * doc/sourcebuild.texi (ARM-specific attributes): Document new
841 arm_neon_ok_no_float_abi effective target.
843 2017-06-26 Richard Biener <rguenther@suse.de>
845 PR tree-optimization/80928
846 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
847 (copy_bbs): Set BB_DUPLICATED flag early.
848 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
850 (execute_on_shrinking_pred): Likewise.
851 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
852 BB_DUPLICATED blocks.
853 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
854 iterate over all PHIs considering removal of *gsi.
856 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
858 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
861 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
863 * config/rs6000/rs6000-string.c: (expand_block_clear,
864 do_load_for_compare, select_block_compare_mode,
865 compute_current_alignment, expand_block_compare,
866 expand_strncmp_align_check, expand_strn_compare,
867 expand_block_move, rs6000_output_load_multiple)
868 Move functions related to string/block move/compare
870 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
871 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
872 for this function which is now used in two files.
873 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
874 * config.gcc: Add rs6000-string.o to extra_objs for
875 targets powerpc*-*-* and rs6000*-*-*.
877 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
880 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
881 32-bit, since indexed is not valid for DImode.
882 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
883 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
884 (define_peephole2 for Altivec d-form load): Add 32-bit support.
885 (define_peephole2 for Altivec d-form store): Likewise.
888 * multiple_target.c (create_dispatcher_calls): Only create the
889 dispatcher call if the function is the default clone of a
892 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
895 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
896 a call, force the call to not be a tail call.
898 2017-06-23 Jeff Law <law@redhat.com>
900 * doc/contrib.texi: Add entry for Steven Pemberton's work on
903 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
905 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
906 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
907 handling for early expansion of vector shifts (sl,sr,sra,rl).
908 (builtin_function_type): Add vector shift right instructions
909 to the unsigned argument list.
911 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
913 rtl-optimizatoin/79286
914 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
915 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
916 trap. PIC register plus a const unspec without offset can never trap.
918 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
920 * tree.h (builtin_structptr_type): New type.
921 (builtin_structptr_types): Declare new array.
922 * tree.c (builtin_structptr_types): New array.
923 (free_lang_data, build_common_tree_nodes): Use it.
925 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
928 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
931 2017-06-22 Matt Turner <mattst88@gmail.com>
933 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
934 Lake models to skylake case. Assume skylake for unknown
935 models with clflushopt.
937 2017-06-22 Jeff Law <law@redhat.com>
939 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
940 frame sizes that do not satisfy aarch64_uimm12_shift.
942 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
944 * profile-count.h (apply_probability,
945 apply_scale, probability_in): Fix checks for zero.
947 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
949 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
950 * doc/cppdiropts.texi (-I @var{dir}): Document it.
952 2016-06-22 Richard Biener <rguenther@suse.de>
954 * tree-vect-loop.c (vect_model_reduction_cost): Handle
955 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
956 REDUC_MAX_EXPR support.
957 (vectorizable_reduction): Likewise.
958 (vect_create_epilog_for_reduction): Likewise.
960 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
962 * match.pd (A / (1 << B) -> A >> B): New.
963 * generic-match-head.c: Include optabs-tree.h.
964 * gimple-match-head.c: Likewise.
965 * optabs-tree.h (target_supports_op_p): New.
966 * optabs-tree.c (target_supports_op_p): New.
968 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
970 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
971 $gcc_cv_ld --help output.
972 (gcc_cv_ld_demangle): Likewise.
973 (gcc_cv_ld_eh_frame_hdr): Likewise.
974 (gcc_cv_ld_pie): Likewise.
975 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
976 (gcc_cv_ld_buildid): Likewise.
977 (gcc_cv_ld_sysroot): Likewise.
978 (ld_bndplt_support): Likewise.
979 (ld_pushpopstate_support): Likewise.
980 * configure: Regenerate.
981 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
983 2017-06-21 Jakub Jelinek <jakub@redhat.com>
986 * config/i386/sse.md (round<mode>2): Renumber match_dup and
987 operands indexes to avoid gap between operands and match_dups.
989 2017-06-21 Andrew Pinski <apinski@cavium.com>
991 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
992 Increment Arith_shift and Arith_shift_reg by 1.
993 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
995 * config/aarch64/aarch64.c (thunderx_tunings): Enable
996 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
997 (aarch64_strip_extend): Add new argument and test for it.
998 (aarch64_cheap_mult_shift_p): New function.
999 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
1000 add a cost if it is true.
1001 Update calls to aarch64_strip_extend.
1002 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
1004 2017-06-21 Andrew Pinski <apinski@cavium.com>
1006 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
1008 (thunderxt88): Likewise.
1009 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
1010 (thunderx_prefetch_tune): New variable.
1011 (thunderx2t99_prefetch_tune): Update for the correct values.
1012 (thunderxt88_tunings): New variable.
1013 (thunderx_tunings): Use thunderx_prefetch_tune instead of
1014 generic_prefetch_tune.
1015 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
1017 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1019 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
1020 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
1021 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
1022 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
1023 (aarch64_atomic_cas<mode>, GPI): Likewise.
1025 2017-06-21 Martin Liska <mliska@suse.cz>
1027 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
1028 statements on cold and hot labels.
1029 * predict.c (tree_estimate_probability_bb): Remove the
1030 prediction from this place.
1032 2017-06-21 Martin Liska <mliska@suse.cz>
1034 PR tree-optimization/79489
1035 * gimplify.c (maybe_add_early_return_predict_stmt): New
1037 (gimplify_return_expr): Call the function.
1038 * predict.c (tree_estimate_probability_bb): Remove handling
1040 * predict.def: Update comment about early return predictor.
1041 * gimple-predict.h (is_gimple_predict): New function.
1042 * predict.def: Change default value of early return to 66.
1043 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
1045 * passes.def: Put pass_strip_predict_hints to the beginning of
1048 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
1050 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
1051 FUNCTION_DECL declarations.
1052 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
1054 (dwaf2out_decl): Likewise.
1055 * godump.c (go_early_global_decl): Skip call to the real debug hook
1056 for FUNCTION_DECL declarations.
1057 * passes.c (rest_of_decl_compilation): Skip call to the
1058 early_global_decl debug hook for FUNCTION_DECL declarations, unless
1059 -fdump-go-spec is passed.
1061 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
1063 * config/i386/i386.c (struct builtin_isa): New field pure_p.
1064 Reorder for compactness.
1065 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
1066 (def_builtin_pure, def_builtin_pure2): New functions.
1067 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
1069 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
1071 * match.pd (nop_convert): New predicate.
1072 ((A +- CST1) +- CST2): Allow some NOP conversions.
1074 2017-06-21 Jakub Jelinek <jakub@redhat.com>
1077 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
1078 with ctors/dtors if GOVD_SHARED is set.
1080 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
1082 * config/aarch64/aarch64.md (movti_aarch64):
1083 Emit mov rather than orr.
1084 (movtf_aarch64): Likewise.
1085 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
1086 Emit mov rather than orr.
1088 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
1090 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
1091 Swap alternatives, make integer dup more expensive.
1093 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
1095 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
1096 Return true for non-tls symbols.
1098 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
1100 * config/aarch64/aarch64-cores.def (cortex-a55): New.
1101 (cortex-a75): Likewise.
1102 (cortex-a75.cortex-a55): Likewise.
1103 * config/aarch64/aarch64-tune.md: Regenerate.
1104 * doc/invoke.texi (-mtune): Document new values for -mtune.
1106 2017-06-21 Tom de Vries <tom@codesourcery.com>
1108 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
1110 (Effective-Target Keywords, Other attributes): Suggest using
1111 dg-add-options stack_size feature to get stack limit in stack_size
1112 effective target documentation.
1114 2017-06-21 Julian Brown <julian@codesourcery.com>
1115 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
1117 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
1118 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
1119 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
1121 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
1122 attribute type list for neon_multiply.
1123 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
1124 attribute type list for neon_multiply.
1125 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
1126 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
1127 attribute type list for neon_multiply.
1128 * config/arm/types.md (crypto_pmull): Add.
1129 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
1130 attribute type list.
1132 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
1134 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
1137 2017-06-20 Jakub Jelinek <jakub@redhat.com>
1139 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
1140 to make sure not to dereference a NULL cost_classes_ptr pointer.
1142 2017-06-20 Carl Love <cel@us.ibm.com>
1144 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1145 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1146 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1147 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1148 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1149 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
1150 VMULOSW): New enum "unspec" values.
1151 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
1152 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
1153 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
1154 altivec_vmulosw): New patterns.
1155 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1156 VMULOSW): Add definitions.
1158 2017-06-20 Julia Koval <julia.koval@intel.com>
1160 * config/i386/i386.c: Fix rounding expand for new pattern.
1161 * config/i386/subst.md: Fix pattern (parallel -> unspec).
1163 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
1165 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
1166 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
1168 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
1170 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
1173 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
1175 * config/aarch64/aarch64-cores.def: Rearrange to sort by
1176 architecture, then by implementer ID.
1177 * config/aarch64/aarch64-tune.md: Regenerate.
1179 2017-06-20 Richard Biener <rguenther@suse.de>
1182 * fold-const.c (split_tree): Fold to type before negating.
1184 2017-06-20 David Malcolm <dmalcolm@redhat.com>
1186 * diagnostic-show-locus.c
1187 (selftest::test_fixit_deletion_affecting_newline): New function.
1188 (selftest::diagnostic_show_locus_c_tests): Call it.
1190 2017-06-20 Andreas Schwab <schwab@suse.de>
1193 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
1196 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
1198 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
1199 __ARM_FEATURE_COPROC according to support.
1201 2017-06-20 Jakub Jelinek <jakub@redhat.com>
1203 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
1204 Rewritten to avoid overflow for > 32-bit pointers.
1207 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
1208 by removing enum keyword.
1209 (ubsan_type_descriptor): Likewise. Formatting fix.
1212 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
1213 splitter): Require TARGET_SSE2 in the condition.
1215 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
1218 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
1219 for doing vector set of SFmode on ISA 3.0.
1220 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
1221 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
1223 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
1224 SFmode value into a V4SF variable that was extracted from another
1225 V4SF variable without converting the element to double precision
1226 and back to single precision vector format.
1227 (vsx_insert_extract_v4sf_p9_2): Likewise.
1229 2017-06-19 Jakub Jelinek <jakub@redhat.com>
1231 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
1232 in UWHI to avoid undefined overflow.
1235 * ubsan.h (enum ubsan_encode_value_phase): New.
1236 (ubsan_encode_value): Change second argument to
1237 enum ubsan_encode_value_phase with default value of
1238 UBSAN_ENCODE_VALUE_GENERIC.
1239 * ubsan.c (ubsan_encode_value): Change second argument to
1240 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
1241 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
1242 create_tmp_var_raw instead of create_tmp_var and use a
1244 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
1245 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
1246 ubsan_encode_value callers.
1249 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
1250 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
1251 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
1253 2017-06-19 Richard Biener <rguenther@suse.de>
1256 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
1257 estimates if we changed anything.
1259 2017-06-19 Richard Biener <rguenther@suse.de>
1261 PR tree-optimization/80887
1262 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
1263 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
1264 simplified lookups, then reset mprts_hook.
1265 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
1267 (try_to_simplify): Likewise.
1269 2017-06-19 Martin Liska <mliska@suse.cz>
1272 * gimplify.c (gimplify_switch_expr):
1273 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
1275 2017-06-19 Martin Liska <mliska@suse.cz>
1277 * doc/install.texi: Document that PGO runs in 4 stages.
1279 2017-06-19 Martin Liska <mliska@suse.cz>
1282 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
1283 to dispatcher function name.
1284 * multiple_target.c (replace_function_decl): New function.
1285 (create_dispatcher_calls): Redirect both edges and references.
1287 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
1289 * profile-count.c (profile_count::dump): Dump quality.
1290 (profile_count::differs_from_p): Update for unsigned val.
1291 * profile-count.h (profile_count_quality): New enum.
1292 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
1294 2017-06-19 Richard Biener <rguenther@suse.de>
1296 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
1297 struct function as arg.
1298 (estimate_numbers_of_iterations): Export overload with loop arg.
1299 (free_numbers_of_iterations_estimates_loop): Use an overload of
1300 free_numbers_of_iterations_estimates instead.
1301 * tree-cfg.c (remove_bb): Adjust.
1302 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
1303 * tree-parloops.c (gen_parallel_loop): Likewise.
1304 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
1306 (tree_unroll_loops_completely): Likewise.
1307 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
1308 Use an overload instead and export.
1309 (estimated_loop_iterations): Adjust.
1310 (max_loop_iterations): Likewise.
1311 (likely_max_loop_iterations): Likewise.
1312 (estimate_numbers_of_iterations): Take struct function as arg
1314 (loop_exits_before_overflow): Adjust.
1315 (free_numbers_of_iterations_estimates_loop): Use an overload.
1316 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
1317 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
1319 2017-06-19 Richard Biener <rguenther@suse.de>
1322 * ipa-prop.c (find_constructor_constant_at_offset): Handle
1323 RANGE_EXPR conservatively.
1325 2017-06-16 Carl Love <cel@us.ibm.com>
1327 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1328 definitions for vec_float, vec_float2, vec_floato,
1329 vec_floate built-ins.
1330 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
1331 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
1333 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
1334 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
1335 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
1336 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
1337 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
1338 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
1339 vec_floato): Add builtin defines.
1340 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
1341 Update the built-in documentation file for the new built-in
1344 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1346 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
1347 (mthumb): Mark as the negative of -marm.
1349 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1351 * doc/invoke.texi (ARM Options, -mcpu): Document supported
1353 (ARM Options, -mtune): Document that this accepts the same
1354 extension options as -mcpu.
1355 (ARM Options, -mfpu): Document addition of -mfpu=auto.
1357 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1359 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
1360 permitted extensions.
1362 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1364 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
1365 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
1366 (armv8-m.main): Add option +nodsp.
1367 * config/arm/arm-cpu-cdata.h: Regenerated.
1369 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1371 * config/arm/t-fuchsia: New file.
1372 * config.gcc (arm*-*-fuchsia*): Use it.
1374 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1376 * config/arm/t-symbian: Rewrite for new option infrastructure.
1378 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1380 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
1381 (MULTILIB_REQUIRED): Likewise.
1383 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1385 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
1386 (MULTILIB_RESUE): Likewise.
1387 (MULTILIB_MATCHES): Likewise.
1388 (MULTLIB_REQUIRED): Likewise.
1390 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1392 * config/arm/t-rtems: Rewrite for new option framework.
1394 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1396 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
1397 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
1398 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
1399 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
1400 * config/arm/t-multilib: ... here.
1401 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
1402 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
1403 armv7-a and armv8*-a when A-profile libraries have not been built.
1404 * config/arm/t-rmprofile: Rewrite.
1406 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1408 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
1409 with a backslash. Remove the backslash after substituting unescaped
1411 * doc/fragments.texi (MULTILIB_REUSE): Document it.
1413 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1415 * config.gcc: (arm*-*-*): When building a-profile libraries, force
1416 the driver to pass through the default setting of -mfloat-abi.
1417 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
1419 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
1420 (all_feat_combs): New rule.
1421 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
1423 * config/arm/t-aprofile: Rewrite.
1425 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1427 * config/arm/arm.h (FPUTYPE_AUTO): Define.
1428 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
1429 fpu is not specified by the user/command-line.
1430 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
1431 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
1432 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
1433 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
1434 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
1435 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
1437 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1439 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
1440 * config/arm/t-arm-elf: Rewritten.
1442 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1444 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
1445 have some floating-point instructions.
1446 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
1447 (TARGET_MAYBE_HARD_FLOAT): New macro.
1448 * config/arm/arm-builtins.c (arm_init_builtins): Use
1449 TARGET_MAYBE_HARD_FLOAT.
1450 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
1452 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1454 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
1455 (configargs.h): Include it.
1456 (arm_print_hint_for_fpu_option): New function.
1457 (arm_parse_fpu_option): New function.
1458 (candidate_extension): New class.
1459 (arm_canon_for_multilib): New function.
1460 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
1461 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
1462 (ARCH_CANONICAL_SPECS): New macro.
1463 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
1465 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1467 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
1468 are set after handling multilib fragments. Set target_cpu_default2
1471 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1473 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
1475 (arm*-*-*): Set target_cpu_default2 to a quoted string.
1476 * config/arm/parsecpu.awk (check_cpu): Validate any extension
1478 (check_arch): Likewise.
1479 * config/arm/arm.c (arm_configure_build_target): Handle
1480 TARGET_CPU_DEFAULT being a string constant. Scan any feature
1481 options in the default.
1483 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1485 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
1486 when an option is an alias of another.
1487 * config/arm/parsecpu.awk (optalias): New parser token.
1488 (gen_comm_data): Mark non-alias options as such. Emit entries
1489 for extension aliases.
1490 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
1491 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
1492 (armv6kz, armv6zk, armv6t2): Likewise.
1493 (armv7): Make vfpv3-d16 an alias.
1494 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
1496 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
1497 Sort in canonical order.
1498 (armv8-a): Sort in canonical order.
1499 (armv8.1-a, armv8.2-a): Likewise.
1500 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
1502 (cortex-a9): Sort in canonical order.
1503 * config/arm/arm.c (selftests.h): Include it.
1504 (arm_test_cpu_arch_data): New function.
1505 (arm_run_self_tests): New function.
1506 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
1507 (targetm): Move declaration to the end of the file.
1508 * arm-cpu-cdata.h: Regenerated.
1510 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1512 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
1513 call to target_mode_check describing the type of option passed.
1514 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
1515 (arm_target_thumb_only): Use arm_parse_arch_option_name or
1516 arm_parse_cpu_option_name to match parameters against list of
1518 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
1519 arm_arch_core_flags data structure.
1520 * config/arm/arm-cpu_cdata.h: Regenerated.
1522 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1524 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
1526 (arm_print_hint_for_cpu_option): Likewise.
1527 (arm_print_hint_for_arch_option): Likewise.
1528 (arm_parse_cpu_option_name): Likewise.
1529 (arm_parse_arch_option_name): Likewise.
1530 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
1531 of entries in the all_fpus list.
1532 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
1533 (arm_parse_cpu_option_name): Declare.
1534 (arm_parse_arch_option_name): Declare.
1535 (arm_parse_option_features): Declare.
1536 (arm_intialize_isa): Declare.
1537 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
1539 (gen_comm_data): ... here. Make definitions non-static.
1540 * config/arm/arm-cpu-data.h: Regenerated.
1541 * config/arm/arm-cpu-cdata.h: Regenerated.
1543 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1545 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
1546 (cpu_arch_extension): New structure.
1547 (cpu_arch_option, arch_option, cpu_option): New structures.
1548 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
1550 (gen_data): Generate new format data tables.
1551 * config/arm/arm.c (cpu_tune): New structure.
1552 (cpu_option, processors): Delete.
1553 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
1554 (arm_print_hint_for_cpu_option): ... this and ...
1555 (arm_print_hint_for_arch_option): ... this.
1556 (arm_parse_arch_cpu_name): Delete. Replace with ...
1557 (arm_parse_cpu_option_name): ... this and ...
1558 (arm_parse_arch_option_name): ... this.
1559 (arm_unrecognized_feature): Change type of target parameter to
1561 (arm_parse_arch_cpu_features): Delete. Replace with ...
1562 (arm_parse_option_features): ... this.
1563 (arm_configure_build_target): Rework to use new configuration data
1565 (arm_print_tune_info): Rework for new configuration data tables.
1566 * config/arm/arm-cpu-data.h: Regenerated.
1567 * config/arm/arm-cpu.h: Regenerated.
1569 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1571 * Makefile.in (OBJS): Move sbitmap.o from here ...
1572 (OBJS-libcommon): ... to here.
1574 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1576 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
1577 (ISA_ALL_CRYPTO): New macro.
1578 (ISA_ALL_SIMD): New macro
1579 (ISA_ALL_FP): New macro.
1580 * config/arm/arm.c (fpu_bitlist): Update initializer.
1581 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
1583 (arm9e): Add fpu. Add option for nofp
1584 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
1585 (arm926ej-s, arm1026ej-s): Likewise.
1586 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
1587 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
1588 neon-fp16, neon-vfpv4, nofp and nosimd.
1589 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
1590 (cortex-a8): Add fpu. Add option for nofp.
1591 (cortex-a9): Add fpu. Add options for nosimd and nofp.
1592 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
1593 (cortex-r4f): Add fpu.
1594 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
1595 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
1597 (cortex-r8): Likewise.
1598 (cortex-m4): Add fpu. Add option for nofp.
1599 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
1600 (cortex-a17.cortex-a7): Likewise.
1601 (cortex-a32): Add fpu. Add options for crypto and nofp.
1602 (cortex-a35, cortex-a53): Likewise.
1603 (cortex-a57): Add fpu. Add option for crypto.
1604 (cortex-a72, cortex-a73): Likewise.
1605 (exynos-m1): Likewise.
1606 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
1607 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
1608 (cortex-m33): Add fpu. Add option for nofp.
1609 * config/arm/arm-cpu-cdata.h: Regenerated
1610 * config/arm/arm-cpu-data.h: Regenerated.
1612 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1614 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
1615 (armv5te, armv5tej): Likewise.
1616 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
1617 (armv7): Add options fp and vfpv3-d16.
1618 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
1619 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
1621 (armv7ve): Likewise.
1622 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
1623 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
1624 (armv8-a): Add nocrypto option.
1625 (armv8.1-a, armv8.2-a): Likewise.
1626 (armv8-m.main): add options fp, fp.dp and nofp.
1628 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1630 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
1632 (armv8-a+crc): Delete.
1633 (armv8.1-a): Add options simd, crypto and nofp.
1634 (armv8.2-a): Add options fp16, simd, crypto and nofp.
1635 (armv8.2-a+fp16): Delete.
1636 (armv8-m.main): Add option dsp.
1637 (armv8-m.main+dsp): Delete.
1638 (cortex-a8): Add fpu. Add nofp option.
1639 (cortex-a9): Add fpu. Add nofp and nosimd options.
1640 * config/arm/parsecpu.awk (gen_data): Generate option tables and
1641 link to main cpu and architecture data structures.
1642 (gen_comm_data): Only put isa attributes from the main architecture
1644 (option): New statement for architecture and CPU entries.
1645 * arm.c (struct cpu_option): New structure.
1646 (struct processors): Add entry for options.
1647 (arm_unrecognized_feature): New function.
1648 (arm_parse_arch_cpu_name): Ignore any characters after the first
1650 (arm_parse_arch_cpu_feature): New function.
1651 (arm_configure_build_target): Separate out any CPU and architecture
1652 features and parse separately. Don't error out if -mfpu=auto is
1653 used with only an architecture string.
1654 (arm_print_asm_arch_directives): New function.
1655 (arm_file_start): Call it.
1656 * config/arm/arm-cpu-cdata.h: Regenerated.
1657 * config/arm/arm-cpu-data.h: Likewise.
1658 * config/arm/arm-tables.opt: Likewise.
1660 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1662 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
1663 assembler when it is not -mfpu=auto.
1665 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1667 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
1668 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
1669 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
1670 (ASM_CPU_SPEC): Rewrite.
1671 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
1672 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
1673 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
1674 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
1675 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
1676 copied string is NUL-terminated. Also strip any characters prefixed
1678 (arm_rewrite_selected_arch): New function.
1679 (arm_rewrite_march): New function.
1681 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
1683 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
1684 (x_arm_cpu_string, x_arm_tune_string): Likewise.
1685 (march, mcpu, mtune): Convert to string-based options.
1686 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
1687 (arm_parse_arch_cpu_name): New function.
1688 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
1689 identify selected architecture or CPU.
1690 (arm_option_save): New function.
1691 (TARGET_OPTION_SAVE): Redefine.
1692 (arm_option_restore): Restore string options.
1693 (arm_option_print): Print string options.
1695 2017-06-16 Martin Sebor <msebor@redhat.com>
1697 PR tree-optimization/80933
1698 PR tree-optimization/80934
1699 * builtins.c (fold_builtin_3): Do not handle bcmp here.
1700 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
1701 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
1702 (gimple_fold_builtin): Call them.
1704 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
1706 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
1707 as unlikely; update profile.
1709 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
1711 * predict.c (force_edge_cold): Handle declaring edges impossible
1714 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
1716 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
1718 (try_unroll_loop_completely): Fix reporting.
1720 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
1722 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
1724 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
1727 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
1728 if given a non-constant argument for an intrinsic which requires a
1731 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
1733 * profile.c (compare_freqs): New function.
1734 (branch_prob): Sort edge list.
1735 (find_spanning_tree): Assume that the list is priority sorted.
1737 2017-06-16 Richard Biener <rguenther@suse.de>
1739 PR tree-optimization/81090
1740 * passes.def (pass_record_bounds): Remove.
1741 * tree-pass.h (make_pass_record_bounds): Likewise.
1742 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
1743 make_pass_record_bounds): Likewise.
1744 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
1745 not free niter estimates at the beginning but at the end.
1746 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
1748 2017-06-16 Richard Biener <rguenther@suse.de>
1750 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
1751 initializer to workaround ICE in host GCC 4.8.
1753 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
1755 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
1757 (clone_inlined_nodes): Update.
1759 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
1761 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
1762 prefetch settings, and enable prefetching by default at -O3.
1764 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
1766 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1767 Set flag_prefetch_loop_arrays according to tuning data.
1769 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
1771 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
1773 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
1774 [Unrelated to main purpose of the patch] Place the pointer field last
1775 to enable type checking errors when tune structure are wrongly merged.
1776 * config/aarch64/aarch64.c (generic_prefetch_tune,)
1777 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
1778 (thunderx2t99_prefetch_tune): New tune constants.
1779 (tune_params *_tunings): Update all tunings (no functional change).
1780 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
1781 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
1782 from tunings structures.
1784 2017-06-16 Jakub Jelinek <jakub@redhat.com>
1787 * ubsan.c (instrument_null): Add T argument, use it instead
1788 of computing it based on IS_LHS.
1789 (instrument_object_size): Likewise.
1790 (pass_ubsan::execute): Adjust instrument_null and
1791 instrument_object_size callers to pass gimple_get_lhs or
1792 gimple_assign_rhs1 result to it. Use instrument_null instead of
1793 calling get_base_address and instrument_mem_ref. Handle
1794 aggregate call arguments for object-size sanitization.
1796 2017-06-16 Yury Gribov <tetra2005@gmail.com>
1798 PR tree-optimization/81089
1799 * tree-vrp.c (is_masked_range_test): Validate operands of
1802 2017-06-15 Martin Sebor <msebor@redhat.com>
1805 * dumpfile.c (dump_register): Avoid calling memset to initialize
1806 a class with a default ctor.
1807 * gcc.c (struct compiler): Remove const qualification.
1808 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
1809 * hash-table.h: Ditto.
1810 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
1812 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
1813 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
1815 * params.h (struct param_info): Make struct members non-const.
1816 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
1817 with default initialization.
1818 * vec.h (vec_copy_construct, vec_default_construct): New helper
1820 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
1821 with vec_copy_construct.
1822 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
1823 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
1824 * doc/invoke.texi (-Wclass-memaccess): Document.
1826 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1828 * emit-rtl.h (is_leaf): Update comment about local
1831 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
1834 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
1835 for a variable to have a section before checking if the section has a
1837 Set section to.persistent if persistent attribute is set.
1838 Warn if .persistent attribute is used on an automatic variable.
1840 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
1842 PR rtl-optimization/80474
1843 * reorg.c (update_block): Do not ignore instructions in a delay slot.
1845 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
1847 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
1850 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
1852 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
1853 (casesi): Emit bounds checking as RTL.
1854 (casesi_internal_mips16_<mode>): Remove bounds checking.
1856 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
1858 * config/xtensa/xtensa.c (xtensa_option_override): Append
1859 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
1860 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
1861 xtensa_doloop_hooks): Define unconditionally.
1862 (xtensa_reorg_loops): Only call reorg_loops in the presence of
1864 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
1865 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
1866 for it in xtensa_option_override.
1867 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
1868 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
1870 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
1872 * doc/cppopts.texi: Document '-' special value to -MF.
1874 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
1876 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
1877 (cortex_a53_fconst): Likewise.
1878 (cortex_a53_fpmul): Likewise.
1879 (cortex_a53_f_load_64): Likewise.
1880 (cortex_a53_f_load_many): Likewise.
1881 (cortex_a53_advsimd_alu): Likewise.
1882 (cortex_a53_advsimd_alu_q): Likewise.
1883 (cortex_a53_advsimd_mul): Likewise.
1884 (cortex_a53_advsimd_mul_q): Likewise.
1885 (fpmac bypass): Add new bypass for fpmac-fpmac case.
1886 Add missing fmul, r2f_cvt and fconst cases.
1888 2017-06-14 Richard Biener <rguenther@suse.de>
1891 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
1893 (fold_binary_loc): When associating do not treat pre-existing
1894 TREE_OVERFLOW on literal constants as a reason to allow
1895 TREE_OVERFLOW on associated literal constants.
1897 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
1899 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
1900 (MASK_FEATURES): New macro.
1901 * config/sparc/sparc.c (sparc_option_override): Remove the special
1902 handling of -mfpu and generalize it to all MASK_FEATURES switches.
1904 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
1906 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
1907 a division of 0 if non-call exceptions are enabled.
1909 2017-06-14 Andrew Pinski <apinski@cavium.com>
1910 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
1913 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
1914 Improve vector initialization code gen for only variable case.
1916 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
1918 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
1920 2017-06-14 Richard Biener <rguenther@suse.de>
1922 PR tree-optimization/81083
1923 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
1926 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
1928 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
1929 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
1930 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
1931 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
1932 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
1933 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
1935 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
1937 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
1938 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
1940 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
1942 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
1944 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
1946 * config/rs6000/t-rtems: Don't handle SPE.
1948 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
1950 * config/rs6000/t-linux: Don't handle SPE.
1952 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
1954 * config/rs6000/eabispe.h: Delete file.
1956 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
1958 * config/rs6000/t-spe: Delete file.
1960 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
1962 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
1963 (rs6000_legitimate_offset_address_p): Return false for anything in
1964 V2SImode or V2SFmode.
1966 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
1968 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
1969 except V2SF and V2SI. Rearrange the vector modes, and add comments.
1970 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
1972 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
1973 (rs6000_legitimate_offset_address_p): Ditto.
1974 (rs6000_emit_move): Ditto.
1975 (rs6000_init_builtins): Remove V4HI_type_node.
1977 2017-06-13 Martin Liska <mliska@suse.cz>
1980 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
1981 (gate_asan): Likewise.
1982 * asan.h (asan_no_sanitize_address_p): Remove the function.
1983 (sanitize_flags_p): New function.
1984 * builtins.def: Fix coding style.
1985 * common.opt: Use renamed enum value.
1986 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
1987 * doc/extend.texi: Document no_sanitize attribute.
1988 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
1989 to SANITIZE_UNDEFINED_NONDEFAULT.
1990 * gcc.c (sanitize_spec_function): Use the renamed enum value.
1991 * gimple-fold.c (optimize_atomic_compare_exchange_p):
1992 Use sanitize_flags_p.
1993 * gimplify.c (gimplify_function_tree): Likewise.
1994 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
1995 * opts.c (parse_no_sanitize_attribute): New function.
1996 (common_handle_option): Use renamed enum value.
1997 * opts.h (parse_no_sanitize_attribute): Declare.
1998 * tree.c (sanitize_flags_p): New function.
1999 * tree.h: Declared here.
2000 * tsan.c: Use sanitize_flags_p.
2001 * ubsan.c (ubsan_expand_null_ifn): Likewise.
2002 (instrument_mem_ref): Likewise.
2003 (instrument_bool_enum_load): Likewise.
2004 (do_ubsan_in_current_function): Remove the function.
2005 (pass_ubsan::execute): Use sanitize_flags_p.
2006 * ubsan.h: Remove do_ubsan_in_current_function
2007 * tree-cfg.c (print_no_sanitize_attr_value): New function.
2008 (dump_function_to_file): Use it here.
2010 2017-06-13 Martin Jambor <mjambor@suse.cz>
2012 PR tree-optimization/80803
2013 PR tree-optimization/81063
2014 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
2015 (propagate_subaccesses_across_link): Enqueue subtree whenever
2016 necessary instead of relying on the caller.
2018 2017-06-13 Martin Jambor <mjambor@suse.cz>
2020 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
2021 that have a first_link.
2022 (sort_and_splice_var_accesses): Do not check first_link before
2024 (subtree_mark_written_and_enqueue): Likewise.
2025 (propagate_all_subaccesses): Likewise and do not stop at first
2026 parent with a first_link.
2028 2017-06-13 Martin Jambor <mjambor@suse.cz>
2030 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
2033 2017-06-13 Yury Gribov <tetra2005@gmail.com>
2035 * match.pd: New pattern.
2037 2017-06-13 Yury Gribov <tetra2005@gmail.com>
2039 * tree-vrp.c (is_masked_range_test): New function.
2040 (register_edge_assert_for): Determine ranges for
2043 2017-06-13 Yury Gribov <tetra2005@gmail.com>
2045 PR tree-optimization/67328
2046 * fold-const.c (maskable_range_p): New function.
2047 (build_range_check): Generate bittests if possible.
2049 2017-06-13 Martin Liska <mliska@suse.cz>
2051 * gimple-pretty-print.c (dump_probability): Add new argument.
2052 (dump_edge_probability): Dump both probability and count.
2053 (dump_gimple_label): Likewise.
2054 (dump_gimple_bb_header): Likewise.
2056 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
2059 * config/avr/avr-devices.c: Fix indentation.
2060 * config/avr/gen-avr-mmcu-specs.c: Dito.
2062 2017-06-13 Richard Biener <rguenther@suse.de>
2064 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
2065 instead get vector type from stmt_info.
2066 (vectorizable_reduction): Adjust. Remove dead code.
2068 2017-06-13 Richard Biener <rguenther@suse.de>
2071 * fold-const.c (extract_muldiv_1): Remove bogus distribution
2072 case of C * (x * C2 + C3).
2073 (fold_addr_of_array_ref_difference): Properly fold index difference.
2075 2017-06-12 David S. Miller <davem@davemloft.net>
2078 * config/sparc/sparc.md (return expander): Emit frame blockage if
2079 function uses alloca.
2081 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
2083 * combine.c (make_field_assignment): Check len rather than the mode
2084 precision when calling force_to_mode.
2086 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
2088 Support multilibs and devices that see flash in RAM address range.
2091 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
2092 (avr_mcu_t) <flash_pm_offset>: New field.
2093 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
2094 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
2095 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
2096 (AVR_TINY_PM_OFFSET): Remove macro.
2097 * config/avr/avr.opt (-mshort-calls): New option.
2098 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
2099 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
2100 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
2101 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
2102 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
2103 instead of avr_arch->have_jmp_call.
2104 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
2105 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
2106 avr_arch->flash_pm_offset to define.
2107 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
2108 new field flash_pm_offset. Add entry for avrxmega3.
2109 (avr_texinfo): Add entry for avrxmega3.
2110 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
2111 attiny212, attiny214,
2112 attiny412, attiny414, attiny416, attiny417,
2113 attiny814, attiny816, attiny817,
2114 attiny1614, attiny1616, attiny1617,
2115 attiny3214, attiny3216, attiny3217.
2116 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
2117 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
2118 (avr_print_operand_address) [AVR_TINY]: Same.
2119 (avr_asm_init_sections) <readonly_data_section>: Only patch
2120 callback if avr_arch->flash_pm_offset = 0.
2121 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
2122 for rodata if avr_arch->flash_pm_offset != 0.
2123 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
2124 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
2125 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
2126 (m_options): Append opt_rcall.
2127 (m_dirnames): Append dir_rcall.
2128 * config/avr/t-multilib: Regenerate.
2130 * configure.ac [target=avr]: Check whether avrxmega3 default
2131 linker description file works as needed.
2132 * configure: Regenerate.
2133 * doc/avr-mmcu.texi: Regenerate.
2134 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
2135 <__AVR_ARCH__>: Document avrxmega3 and 103.
2136 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
2137 <__AVR_SHORT_CALLS__>: Document it.
2138 <__AVR_PM_BASE_ADDRESS__>: Document it.
2139 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
2140 (AVR Variable Attributes) <progmem>: Document this is
2141 not needed for avrxmega3.
2142 (AVR Named Address Spaces) <__flash>: Dito.
2144 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
2146 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
2148 2017-06-12 Doug Rupp <rupp@adacore.com>
2150 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
2151 Append vxworks-stdint.h to the tm_file list.
2152 * config/vxworks-stdint.h: New file.
2154 2017-06-12 Martin Liska <mliska@suse.cz>
2156 PR tree-optimization/81041
2157 * tree-profile.c (gimple_gen_ic_func_profiler):
2158 Create an extra BB in profile-generate
2159 (gimple_gen_time_profiler): Likewise.
2161 2017-06-12 Jakub Jelinek <jakub@redhat.com>
2163 PR tree-optimization/81003
2164 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
2165 (update_range_test): Use it instead of force_gimple_operand_gsi.
2167 2017-06-12 Richard Biener <rguenther@suse.de>
2169 PR tree-optimization/81053
2170 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
2171 with backedge value not defined in loop. Simplify def stmt
2174 2017-06-11 Tom de Vries <tom@codesourcery.com>
2177 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
2179 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
2180 nvptx_cannot_force_const_mem.
2182 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
2184 * opts.c (finish_options): Move test for flag_split_stack after
2185 it has been initialized.
2187 2017-06-11 Jason Merrill <jason@redhat.com>
2189 * tree.h (id_equal): New.
2190 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
2191 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
2192 instead of strcmp of IDENTIFIER_POINTER.
2194 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
2196 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
2197 (mark_all_inlined_calls_cdtor): Fix formating.
2198 (inline_transform): Rescale profile before inlining.
2200 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
2202 * cgraph.h (cgraph_edge::clone): Update prototype.
2203 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
2204 (cgraph_node::create_clone): Update.
2205 (cgraph_node::create_version_clone): Update.
2206 * tree-inline.c (copy_bb): Update.
2207 (expand_call_inline): Update.
2209 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
2211 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
2212 factored out from ...
2213 (rs6000_emit_prologue): ... here.
2215 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
2217 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
2218 factored out from ...
2219 (rs6000_emit_prologue): ... here.
2221 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
2223 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
2225 (handle_missing_profiles): Fix computation of tp_first_run.
2226 (counts_to_freqs): Do not touch freqs when count is 0.
2228 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
2230 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
2233 2017-06-10 Tom de Vries <tom@codesourcery.com>
2235 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
2236 attributes): Document signal effective target.
2238 2017-06-10 Tom de Vries <tom@codesourcery.com>
2240 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2241 Document effective target stack_size.
2243 2017-06-09 David Malcolm <dmalcolm@redhat.com>
2245 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
2246 to the edit_context if they can be auto-applied.
2248 2017-06-9 Ian Lance Taylor <iant@golang.org>
2250 * opts.c (finish_options): If -fsplit-stack, disable implicit
2251 -forder-blocks-and-partition.
2252 * doc/invoke.texi (Optimize Options): Document that when using
2253 -fsplit-stack -forder-blocks-and-partition is not implicitly
2256 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
2258 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
2259 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
2260 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
2261 * builtins.def (abort, trap, unreachable): Declare cold.
2262 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
2263 * tree-core.h (ECF_COLD): New.
2264 * tree.c (set_call_expr_flags): Handle ECF_COLD.
2265 (build_common_builtin_nodes): Mark unreachable and abort as cold.
2267 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
2269 * predict.c (unlikely_executed_stmt_p): Cleanup.
2271 2017-06-09 Richard Biener <rguenther@suse.de>
2273 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
2274 model if the ref is always written to.
2276 2017-06-09 Tamar Christina <tamar.christina@arm.com>
2278 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
2280 2017-06-09 Tamar Christina <tamar.christina@arm.com>
2282 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
2285 2017-06-09 Tom de Vries <tom@codesourcery.com>
2288 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
2289 "target cannot support label values" when encountering LABEL_REF.
2291 2017-06-09 Martin Liska <mliska@suse.cz>
2293 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
2294 (gimple_gen_ic_func_profiler): Emit direct comparison
2295 of __gcov_indirect_call_callee with NULL.
2296 (gimple_gen_time_profiler): Change probability from
2297 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
2299 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
2301 * profile.c (edge_gcov_counts): Turn to pointer.
2302 (compute_branch_probabilities, compute_branch_probabilities): Update.
2303 (branch_prob): Do not clear edge_gcov_count.
2304 * profile.h (edge_gcov_counts): Turn to pointer.
2305 (edge_gcov_count): Update.
2307 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
2309 * gimple.h (gimple_check_failed): Mark cold.
2311 2017-06-09 Richard Biener <rguenther@suse.de>
2313 PR tree-optimization/66623
2314 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
2315 refactor check_reduction into two parts, properly computing
2316 whether we have to check reduction validity for outer loop
2319 2017-06-09 Richard Biener <rguenther@suse.de>
2321 PR tree-optimization/79483
2322 * graphite-scop-detection.c (order): New global.
2323 (get_order): Compute bb to order mapping that satisfies code
2324 generation constraints.
2325 (cmp_pbbs): New helper.
2326 (build_scops): Start domwalk at entry block, sort generated
2329 2017-06-09 Richard Biener <rguenther@suse.de>
2332 * ipa-polymorphic-call.c
2333 (ipa_polymorphic_call_context::restrict_to_inner_class):
2334 Skip FIELD_DECLs with error_mark_node type.
2335 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
2338 2017-06-09 Martin Liska <mliska@suse.cz>
2340 * predict.c (struct branch_predictor): New struct.
2341 (test_prediction_value_range): New test.
2342 (predict_c_tests): New function.
2343 * selftest-run-tests.c (selftest::run_tests): Run the function.
2344 * selftest.h: Declare new tests.
2346 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
2349 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
2350 gen_add3_insn did not fail.
2351 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
2352 r0, construct that number in a temporary reg and add that reg to r0.
2353 If asked to put the result in r0 as well, fail.
2355 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
2357 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
2358 for early expansion of vec_eqv.
2360 2017-06-08 Jakub Jelinek <jakub@redhat.com>
2363 * ubsan.c (instrument_null): Avoid pointless code temporary.
2364 (pass_ubsan::execute): Instrument aggregate arguments of calls.
2366 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
2370 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
2373 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
2376 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
2378 * predict.c (unlikely_executed_edge_p): Move ahead.
2379 (probably_never_executed_edge_p): Use it.
2381 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
2384 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
2385 gimple_call_builtin_p call.
2387 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
2389 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
2390 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
2391 rtl_check_failed_type2, rtl_check_failed_code1,
2392 rtl_check_failed_code2, rtl_check_failed_code_mode,
2393 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
2394 rtvec_check_failed_bounds, rtl_check_failed_flag,
2395 _fatal_insn_not_found, _fatal_insn): Likewise.
2396 * tree.h (tree_contains_struct_check_failed,
2397 tree_check_failed, tree_not_check_failed,
2398 tree_class_check_failed, tree_range_check_failed,
2399 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
2400 tree_vec_elt_check_failed, phi_node_elt_check_failed,
2401 tree_operand_check_failed, omp_clause_check_failed,
2402 omp_clause_operand_check_failed, omp_clause_range_check_failed):
2405 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
2407 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
2408 flag_branch_probabilities.
2409 * ipa-inline.c (edge_badness): Likewise.
2410 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
2411 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
2412 * predict.c (maybe_hot_frequency_p): Likewise.
2413 (probably_never_executed): Likewise.
2414 * sched-ebb.c (schedule_ebbs): Likewise.
2415 * sched-rgn.c (find_single_block_region): Likewise.
2416 * tracer.c (tail_duplicate): Likewise.
2418 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
2420 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
2421 longer requires x_flag_profile_use.
2423 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
2425 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
2426 instead of flag_reorder_blocks_and_partition.
2427 * dbxout.c (dbxout_function_end): Likewise.
2428 * dwarf2out.c (gen_subprogram_die): Likewise.
2429 * haifa-sched.c (sched_create_recovery_edges): Likewise.
2430 * hw-doloop.c (reorg_loops): Likewise.
2431 * varasm.c (assemble_start_function,
2432 assemble_end_function): Likewise.
2433 (decide_function_section): Do not check for
2434 flag_reorder_blocks_and_partition.
2436 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
2438 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
2440 (chkp_get_hard_register_fake_addr_expr): Ditto.
2441 (chkp_build_addr_expr): Add check for hard reg case.
2442 (chkp_parse_array_and_component_ref): Ditto.
2443 (chkp_find_bounds_1): Ditto.
2444 (chkp_process_stmt): Don't generate bounds store for
2447 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
2449 * predict.c (maybe_hot_bb_p): Do not check profile status.
2450 (maybe_hot_edge_p): Likewise.
2451 (probably_never_executed): Check for zero counts even if profile
2453 (unlikely_executed_edge_p): New function.
2454 (unlikely_executed_stmt_p): New function.
2455 (unlikely_executed_bb_p): New function.
2456 (set_even_probabilities): Use unlikely predicates.
2457 (combine_predictions_for_bb): Likewise.
2458 (predict_paths_for_bb): Likewise.
2459 (predict_paths_leading_to_edge): Likewise.
2460 (determine_unlikely_bbs): New function.
2461 (estimate_bb_frequencies): Use it.
2462 (compute_function_frequency): Use zero counts even if profile is
2464 * profile-count.h: Fix typo.
2466 2017-08-08 Julia Koval <julia.koval@intel.com>
2468 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
2469 _mm512_mask_cvtsepi16_storeu_epi8,
2470 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
2471 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
2472 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
2473 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
2474 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
2475 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
2476 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
2477 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
2478 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
2479 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
2480 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
2481 __builtin_ia32_pmovuswb256mem_mask,
2482 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
2483 __builtin_ia32_pmovwb512mem_mask): New builtins.
2485 2017-08-08 Julia Koval <julia.koval@intel.com>
2487 PR target/73350,80862
2488 * config/i386/subst.md (round): Fix round pattern.
2489 * config/i386/i386.c (ix86_erase_embedded_rounding):
2490 Fix erasing rounding for the fixed pattern.
2492 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
2494 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
2496 2017-06-08 Martin Liska <mliska@suse.cz>
2498 PR gcov-profile/80911
2499 * gcov.c (block_info::block_info): New constructor.
2501 2017-06-07 Carl Love <cel@us.ibm.com>
2503 * config/rs6000/rs6000-c: The return type of the following
2504 built-in functions was implemented as int not long long. Fix sign
2505 of return value for the unsigned version of vec_mulo and vec_mule.
2506 vector unsigned long long vec_bperm (vector unsigned long long,
2507 vector unsigned char)
2508 vector signed long long vec_mule (vector signed int,
2510 vector unsigned long long vec_mule (vector unsigned int,
2511 vector unsigned int)
2512 vector signed long long vec_mulo (vector signed int,
2514 vector unsigned long long vec_mulo (vector unsigned int,
2515 vector unsigned int)
2516 * doc/extend.texi: Fix the documentation for the built-in
2519 2017-06-07 Carl Love <cel@us.ibm.com>
2522 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
2525 2017-06-07 Carl Love <cel@us.ibm.com>
2527 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
2528 support, Generate doublehv for signed int/float for BE case only.
2530 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
2532 * doc/invoke.texi (mcx16): Rewrite.
2534 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
2536 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
2537 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
2538 *mov<mode>_softfloat, and an anonymous splitter): Use
2539 nonimmediate_operand instead of rs6000_nonimmediate_operand.
2541 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
2543 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
2545 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
2546 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
2547 (rs6000_debug_reg_global): Adjust.
2548 (rs6000_init_hard_regno_mode_ok): Adjust.
2549 (rs6000_dbx_register_number): Adjust.
2550 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
2551 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
2552 Remove SPE_ACC and SPEFSCR.
2553 (REG_ALLOC_ORDER): Ditto.
2554 (FRAME_POINTER_REGNUM): Change to 111.
2555 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
2556 (REG_CLASS_NAMES): Ditto.
2557 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
2558 (REGISTER_NAMES): Ditto.
2559 (ADDITIONAL_REG_NAMES): Ditto.
2560 (rs6000_reg_names): Ditto.
2561 * config/rs6000/rs6000.md: Renumber some register number
2564 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
2566 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
2568 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
2569 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
2571 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
2572 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
2573 Delete the SPE high registers.
2574 (REG_ALLOC_ORDER): Ditto.
2575 (enum reg_class): Remove SPE_HIGH_REGS.
2576 (REG_CLASS_NAMES): Ditto.
2577 (REG_CLASS_CONTENTS): Delete the SPE high registers.
2578 (REGISTER_NAMES): Ditto.
2579 (rs6000_reg_names): Ditto.
2580 * doc/tm.texi.in: Remove SPE as example.
2581 * doc/tm.texi: Regenerate.
2583 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
2585 * config/rs6000/8540.md (ppc8540_brinc): Delete.
2586 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
2587 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
2588 * config/rs6000/rs6000.md (type): Remove "brinc".
2590 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
2592 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
2593 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
2594 * config/rs6000/linuxspe.h: Delete file.
2595 * config/rs6000/rs6000.md: Don't include spe.md.
2596 * config/rs6000/spe.h: Delete file.
2597 * config/rs6000/spe.md: Delete file.
2598 * config/rs6000/t-rs6000: Remove spe.md.
2600 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
2602 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
2603 (reg_or_none500mem_operand): Delete.
2604 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
2605 instead of reg_or_none500mem_operand.
2607 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
2609 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
2610 handling of SPE flags.
2611 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
2613 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
2615 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
2617 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
2618 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
2619 paired_divv2sf3): Similar.
2620 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
2621 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
2622 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
2624 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
2625 Rename the paired_* instruction patterns.
2626 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
2628 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
2629 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
2630 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
2631 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
2632 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
2634 (struct machine_function): Delete field spe_insn_chain_scanned_p.
2635 (spe_func_has_64bit_regs_p): Delete.
2636 (spe_expand_predicate_builtin): Delete.
2637 (spe_expand_evsel_builtin): Delete.
2638 (TARGET_DWARF_REGISTER_SPAN): Do not define.
2639 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
2640 (invalid_e500_subreg): Delete.
2641 (rs6000_legitimize_address): Always force_reg op2 as well, for
2642 paired single memory accesses.
2643 (rs6000_member_type_forces_blk): Delete.
2644 (rs6000_spe_function_arg): Delete.
2645 (rs6000_expand_unop_builtin): Delete SPE handling.
2646 (rs6000_expand_binop_builtin): Ditto.
2647 (spe_expand_stv_builtin): Delete.
2648 (bdesc_2arg_spe): Delete.
2649 (spe_expand_builtin): Delete.
2650 (spe_expand_predicate_builtin): Delete.
2651 (spe_expand_evsel_builtin): Delete.
2652 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
2653 (spe_init_builtins): Delete.
2654 (spe_func_has_64bit_regs_p): Delete.
2655 (savres_routine_name): Delete "info" parameter. Adjust callers.
2656 (rs6000_emit_stack_reset): Ditto.
2657 (rs6000_dwarf_register_span): Delete.
2658 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
2659 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
2660 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
2662 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
2664 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
2665 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
2666 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
2667 mulv2sf3, divv2sf3): Delete expanders.
2669 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
2671 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
2673 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
2675 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
2676 * config/rs6000/rs6000.c: Ditto.
2678 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
2680 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
2681 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
2682 comparison_operator.
2684 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
2686 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
2687 * config/rs6000/rs6000.opt: Ditto.
2688 * config/rs6000/t-rtems: Ditto.
2690 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
2692 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
2693 TARGET_E500_SINGLE by 0, simplify.
2694 * config/rs6000/rs6000.c: Ditto.
2695 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
2696 (spe_build_register_parallel): Delete.
2697 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
2698 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
2699 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
2700 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
2701 (E500_CONVERT): Delete.
2702 * config/rs6000/spe.md: Remove many patterns and all define_constants.
2704 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
2706 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
2707 * config/rs6000/dfp.md: Ditto.
2708 (negdd2, *negdd2_fpr): Merge.
2709 (absdd2, *absdd2_fpr): Merge.
2710 (negtd2, *negtd2_fpr): Merge.
2711 (abstd2, *abstd2_fpr): Merge.
2712 * config/rs6000/e500.h: Delete file.
2713 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
2714 TARGET_FPRS by 1 and simplify.
2715 * config/rs6000/rs6000-c.c: Ditto.
2716 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
2718 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
2720 * config/rs6000/rs6000.md: Ditto.
2721 (floatdidf2, *floatdidf2_fpr): Merge.
2722 (move_from_CR_gt_bit): Delete.
2723 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
2724 (E500_CR_IOR_COMPARE): Delete.
2725 (All patterns that require !TARGET_FPRS): Delete.
2726 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
2728 2017-06-07 Bin Cheng <bin.cheng@arm.com>
2730 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
2732 2017-06-07 Bin Cheng <bin.cheng@arm.com>
2734 * graphds.c (add_edge): Intitialize edge's attached data.
2735 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
2736 pointer parameter. Call pointed function on each edge during
2737 graph traversing. Skip traversing the edge when the function
2739 (graphds_dfs, graphds_scc): Ditto.
2740 (for_each_edge): New parameter. Pass the new parameter to callback
2742 * graphds.h (skip_edge_callback): New function pointer type.
2743 (graphds_dfs, graphds_scc): New function pointer parameter.
2744 (graphds_edge_callback, for_each_edge): New parameter.
2746 2017-06-07 Bin Cheng <bin.cheng@arm.com>
2748 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
2749 out code checking if runtime alias check is possible to below ...
2750 Call the new function.
2751 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
2752 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
2754 2017-06-07 Marek Polacek <polacek@redhat.com>
2757 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
2758 TYPE_OVERFLOW_WRAPS check.
2760 2017-06-07 Bin Cheng <bin.cheng@arm.com>
2762 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
2763 if versioning is required.
2764 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
2765 peeling with the check for versioning.
2767 2017-06-07 Bin Cheng <bin.cheng@arm.com>
2769 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
2770 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
2771 Set true to new parameter if new ssa variable is defined.
2772 (vect_gen_vector_loop_niters): Refactor. Set range information
2773 for the new vector loop bound variable.
2774 (vect_do_peeling): Ditto.
2776 2017-06-07 Bin Cheng <bin.cheng@arm.com>
2778 * tree-affine.c (ssa.h): Include header file.
2779 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
2780 has wrapping overflow behavior.
2782 2017-06-07 Bin Cheng <bin.cheng@arm.com>
2784 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
2786 2017-06-07 Bin Cheng <bin.cheng@arm.com>
2788 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
2789 (tree_to_aff_combination): ... here.
2791 2017-06-07 Bin Cheng <bin.cheng@arm.com>
2793 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
2794 reg_pressure model function.
2795 (ivopts_global_cost_for_size): Delete.
2796 (determine_set_costs, iv_ca_recount_cost): Call new model function
2797 ivopts_estimate_reg_pressure.
2799 2017-06-07 Tamar Christina <tamar.christina@arm.com>
2801 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more expensive than udiv.
2802 Remove floating point cases from mod.
2804 2017-06-07 Tamar Christina <tamar.christina@arm.com>
2806 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost): Increase idiv cost.
2808 2017-06-07 Tamar Christina <tamar.christina@arm.com>
2810 * config/aarch64/aarch64.md
2811 (copysignsf3): Fix mask generation.
2813 2017-06-07 Jakub Jelinek <jakub@redhat.com>
2815 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
2817 (class dump_manager): Add register_dumps method.
2818 * dumpfile.c: Include langhooks.h.
2819 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
2820 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
2821 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
2822 (dump_manager::dump_register): Start with 512 entries instead of 32.
2823 (dump_manager::register_dumps): New method.
2824 * toplev.c (general_init): Instead of invoking register_dumps
2825 langhook, invoke register_dumps method on the dump manager.
2826 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
2829 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
2831 * doc/md.texi: Clarify the restrictions on a define_insn condition.
2832 Say that # requires an associated define_split to exist, and that
2833 the define_split must be suitable for use after register allocation.
2835 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
2837 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
2838 (compute_outgoing_frequencies): Also initialize zero counts.
2839 (find_many_sub_basic_blocks): Do not produce uninitialized profile
2840 around loops; preserve more of profile when nothing changes.
2842 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
2844 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
2846 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
2847 * config/arm/arm-cpu-cdata.h: Regenerate.
2848 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
2849 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
2850 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
2851 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
2852 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
2854 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
2856 * config/arm/t-rmprofile: Likewise.
2857 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
2859 2017-06-06 David S. Miller <davem@davemloft.net>
2862 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
2863 blockage if function uses alloca.
2865 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
2867 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
2868 New "uid" fields to hold pretty-print IDs of group and ref.
2869 Memory references are now identified as <group_id>:<ref_id>
2870 instead of using [random] addresses.
2871 (dump_mem_details): Simplify, no functional change.
2872 (dump_mem_ref): Simplify and make output more concise.
2873 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
2874 (find_or_create_group): Initialize group uid.
2875 (record_ref): Initialize ref uid. Improve debug output.
2876 (prune_group_by_reuse, should_issue_prefetch_p,)
2877 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
2878 (mark_nontemporal_store, determine_loop_nest_reuse):
2879 Improve debug output.
2881 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
2883 * dbgcnt.def (prefetch): New debug counter.
2884 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
2885 (schedule_prefetches): Stop issueing prefetches if debug counter
2888 2017-06-06 Tom de Vries <tom@codesourcery.com>
2890 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
2891 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
2893 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2895 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
2896 Use aarch64_reg_or_zero predicate for operand 4.
2897 (aarch64_compare_and_swap<mode> define_insn_and_split):
2898 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
2899 (aarch64_store_exclusive<mode>): Likewise for operand 2.
2901 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
2903 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
2904 (arm_compute_save_core_reg_mask): This.
2905 (thumb1_compute_save_reg_mask): Rename into ...
2906 (thumb1_compute_save_core_reg_mask): This.
2907 (arm_compute_save_reg0_reg12_mask): Adapt comment.
2908 (arm_compute_frame_layout): Likewise.
2910 2017-06-06 Richard Biener <rguenther@suse.de>
2912 PR tree-optimization/80974
2913 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
2914 keep or clear leaders SSA info.
2916 2017-06-06 Tom de Vries <tom@codesourcery.com>
2918 * config/nvptx/nvptx.c (split_mode_p): New function.
2919 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
2921 2017-06-06 Tom de Vries <tom@codesourcery.com>
2923 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
2925 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
2928 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
2931 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
2933 * shrink-wrap.c (handle_simple_exit): Update profile.
2934 (try_shrink_wrapping): Upate profile.
2936 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
2938 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
2939 (tree_guess_outgoing_edge_probabilities): New.
2940 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
2941 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
2943 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
2945 * ipa-split.c (split_function): Initialize return bb profile.
2947 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
2949 * profile.c (compute_branch_probabilities): Also initialize
2952 2017-06-06 Richard Biener <rguenther@suse.de>
2954 PR tree-optimization/80928
2955 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
2956 (vect_analyze_loop_operations): Properly guard analysis for
2958 (vect_transform_loop): Likewise.
2959 (vect_analyze_loop_2): Also reset SLP type on PHIs.
2960 (vect_model_induction_cost): Do not cost for pure SLP.
2961 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
2962 of induction in inner loop vectorization.
2963 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
2964 (vect_get_and_check_slp_defs): Handle vect_induction_def.
2965 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
2967 (vect_analyze_slp_cost_1): Cost induction.
2968 (vect_detect_hybrid_slp_stmts): Handle PHIs.
2969 (vect_get_slp_vect_defs): Likewise.
2970 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
2971 (vect_transform_stmt): Handle SLP reductions.
2972 * tree-vectorizer.h (vectorizable_induction): Adjust.
2974 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
2976 * config/rs6000/rs6000.c (make_resolver_func): Update
2977 init_lowered_empty_function call.
2979 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
2981 * doc/invoke.texi: Document the -fprofile-abs-path option.
2982 * common.opt (fprofile-abs-path): New option.
2983 * gcov-io.h (gcov_write_filename): Declare.
2984 * gcov-io.c (gcov_write_filename): New function.
2985 * coverage.c (coverage_begin_function): Use gcov_write_filename.
2986 * profile.c (output_location): Likewise.
2988 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
2990 * shring-wrap.c: Revert accidental commit.
2992 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
2994 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
2996 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
2998 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
3000 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
3001 profile in callgraph edge.
3002 * profile-count.h (apply_probability): If THIS is 0, then result is 0
3003 (apply_scale): Likewise.
3004 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
3005 Also scale profile when inlining function with zero profile.
3006 (initialize_cfun): Update exit block profile even when it is zero.
3007 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
3008 when profile is read.
3010 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
3012 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
3013 (CLONE_*): New constants to define the processors we can generate
3014 code for with the target_clone attribute.
3015 (rs6000_clone_map): New array to identify which clone processors
3016 the current program is running on.
3017 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
3018 target_clone attribute.
3019 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
3020 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
3021 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
3022 (cpu_expand_builtin): Add support for target_clone attribute.
3023 (rs6000_valid_attribute_p): Allow "default" attribute.
3024 (get_decl_name): New debug function to simplify printing the
3025 current function name in debugging statements.
3026 (rs6000_clone_priority): New functions to support the target_clone
3027 attribute, and be able to generate code to switch between ISA 2.05
3028 through ISA 3.0 (power6 through power9).
3029 (rs6000_compare_version_priority): Likewise.
3030 (rs6000_get_function_versions_dispatcher): Likewise.
3031 (make_resolver_func): Likewise.
3032 (add_condition_to_bb): Likewise.
3033 (dispatch_function_versions): Likewise.
3034 (rs6000_generate_version_dispatcher_body): Likewise.
3035 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
3036 (fusion_gpr_load_p): Fix a spacing issue.
3037 * doc/extend.texi (Common Function Attributes): Document that the
3038 PowerPC supports the target_clone attribute.
3040 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
3042 * config/arm/arm.h: explain F symbol found in description of ARM
3043 register allocation in its legend.
3045 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
3047 * config/mips/frame-header-opt.c: Include profile-count.h.
3048 * config/riscv/riscv.c: Include profile-count.h
3050 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
3052 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
3054 (sm_set_flag_if_changed): Add bbs field.
3055 (execute_sm_if_changed_flag_set): Pass BBS.
3056 (execute_sm): Update.
3058 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3060 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
3063 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3065 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
3066 (peephole2): New peephole2 to emit the above.
3067 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
3069 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3071 * config/aarch64/aarch64.c (define_peephole2 above
3072 *sub_<shift>_<mode>): New peephole.
3074 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
3076 * config/i386/i386.c (make_resolver_func): Update.
3077 * Makefile.in: Add profile-count.h and profile-count.o
3078 * auto-profile.c (afdo_indirect_call): Update to new API.
3079 (afdo_set_bb_count): Update.
3080 (afdo_propagate_edge): Update.
3081 (afdo_propagate_circuit): Update.
3082 (afdo_calculate_branch_prob): Update.
3083 (afdo_annotate_cfg): Update.
3084 * basic-block.h: Include profile-count.h
3085 (struct edge_def): Turn count to profile_count.
3086 (struct basic_block_def): Likewie.
3087 (REG_BR_PROB_BASE): Move to profile-count.h
3088 (RDIV): Move to profile-count.h
3089 * bb-reorder.c (max_entry_count): Turn to profile_count.
3090 (find_traces): Update.
3091 (rotate_loop):Update.
3092 (connect_traces):Update.
3093 (sanitize_hot_paths):Update.
3094 * bt-load.c (migrate_btr_defs): Update.
3095 * cfg.c (RDIV): Remove.
3096 (init_flow): Use alloc_block.
3097 (alloc_block): Uninitialize count.
3098 (unchecked_make_edge): Uninitialize count.
3099 (check_bb_profile): Update.
3100 (dump_edge_info): Update.
3101 (dump_bb_info): Update.
3102 (update_bb_profile_for_threading): Update.
3103 (scale_bbs_frequencies_int): Update.
3104 (scale_bbs_frequencies_gcov_type): Update.
3105 (scale_bbs_frequencies_profile_count): New.
3106 * cfg.h (update_bb_profile_for_threading): Update.
3107 (scale_bbs_frequencies_profile_count): Declare.
3108 * cfgbuild.c (compute_outgoing_frequencies): Update.
3109 (find_many_sub_basic_blocks): Update.
3110 * cfgcleanup.c (try_forward_edges): Update.
3111 (try_crossjump_to_edge): Update.
3112 * cfgexpand.c (expand_gimple_tailcall): Update.
3113 (construct_exit_block): Update.
3114 * cfghooks.c (verify_flow_info): Update.
3115 (dump_bb_for_graph): Update.
3116 (split_edge): Update.
3117 (make_forwarder_block): Update.
3118 (duplicate_block): Update.
3119 (account_profile_record): Update.
3120 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3121 (get_estimated_loop_iterations): Update.
3122 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3123 (single_likely_exit): Update.
3124 * cfgloopmanip.c (scale_loop_profile): Update.
3126 (set_zero_probability): Update.
3127 (lv_adjust_loop_entry_edge): Update.
3128 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
3129 (purge_dead_edges): Update.
3130 (rtl_account_profile_record): Update.
3131 * cgraph.c (cgraph_node::create): Uninitialize count.
3132 (symbol_table::create_edge): Uninitialize count.
3133 (cgraph_update_edges_for_call_stmt_node): Update.
3134 (cgraph_edge::dump_edge_flags): Update.
3135 (cgraph_node::dump): Update.
3136 (cgraph_edge::maybe_hot_p): Update.
3137 * cgraph.h: Include profile-count.h
3138 (create_clone), create_edge, create_indirect_edge): Update.
3139 (cgraph_node): Turn count to profile_count.
3140 (cgraph_edge0: Likewise.
3141 (make_speculative, clone): Update.
3142 (create_edge): Update.
3143 (init_lowered_empty_function): Update.
3144 * cgraphclones.c (cgraph_edge::clone): Update.
3145 (duplicate_thunk_for_node): Update.
3146 (cgraph_node::create_clone): Update.
3147 * cgraphunit.c (cgraph_node::analyze): Update.
3148 (cgraph_node::expand_thunk): Update.
3149 * final.c (dump_basic_block_info): Update.
3150 * gimple-streamer-in.c (input_bb): Update.
3151 * gimple-streamer-out.c (output_bb): Update.
3152 * graphite.c (print_global_statistics): Update.
3153 (print_graphite_scop_statistics): Update.
3154 * hsa-brig.c: Include basic-block.h.
3155 * hsa-dump.c: Include basic-block.h.
3156 * hsa-gen.c (T sum_slice): Update.
3157 (convert_switch_statements):Update.
3158 * hsa-regalloc.c: Include basic-block.h.
3159 * ipa-chkp.c (chkp_produce_thunks): Update.
3160 * ipa-cp.c (struct caller_statistics): Update.
3161 (init_caller_stats): Update.
3162 (gather_caller_stats): Update.
3163 (ipcp_cloning_candidate_p): Update.
3164 (good_cloning_opportunity_p): Update.
3165 (get_info_about_necessary_edges): Update.
3166 (dump_profile_updates): Update.
3167 (update_profiling_info): Update.
3168 (update_specialized_profile): Update.
3169 (perhaps_add_new_callers): Update.
3170 (decide_about_value): Update.
3171 (ipa_cp_c_finalize): Update.
3172 * ipa-devirt.c (struct odr_type_warn_count): Update.
3173 (struct decl_warn_count): Update.
3174 (struct final_warning_record): Update.
3175 (possible_polymorphic_call_targets): Update.
3176 (ipa_devirt): Update.
3177 * ipa-fnsummary.c (redirect_to_unreachable): Update.
3178 * ipa-icf.c (sem_function::merge): Update.
3179 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
3180 * ipa-inline.c (compute_uninlined_call_time): Update.
3181 (compute_inlined_call_time): Update.
3182 (want_inline_small_function_p): Update.
3183 (want_inline_self_recursive_call_p): Update.
3184 (edge_badness): Update.
3185 (lookup_recursive_calls): Update.
3186 (recursive_inlining): Update.
3187 (inline_small_functions): Update.
3188 (dump_overall_stats): Update.
3189 (dump_inline_stats): Update.
3190 * ipa-profile.c (ipa_profile_generate_summary): Update.
3191 (ipa_propagate_frequency): Update.
3192 (ipa_profile): Update.
3193 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
3194 * ipa-utils.c (ipa_merge_profiles): Update.
3195 * loop-doloop.c (doloop_modify): Update.
3196 * loop-unroll.c (report_unroll): Update.
3197 (unroll_loop_runtime_iterations): Update.
3198 * lto-cgraph.c (lto_output_edge): Update.
3199 (lto_output_node): Update.
3200 (input_node): Update.
3201 (input_edge): Update.
3202 (merge_profile_summaries): Update.
3203 * lto-streamer-in.c (input_cfg): Update.
3204 * lto-streamer-out.c (output_cfg): Update.
3205 * mcf.c (create_fixup_graph): Update.
3206 (adjust_cfg_counts): Update.
3207 (sum_edge_counts): Update.
3208 * modulo-sched.c (sms_schedule): Update.
3209 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3210 * predict.c (maybe_hot_count_p): Update.
3211 (probably_never_executed): Update.
3212 (dump_prediction): Update.
3213 (combine_predictions_for_bb): Update.
3214 (propagate_freq): Update.
3215 (handle_missing_profiles): Update.
3216 (counts_to_freqs): Update.
3217 (rebuild_frequencies): Update.
3218 (force_edge_cold): Update.
3219 * predict.h: Include profile-count.h
3220 (maybe_hot_count_p, counts_to_freqs): UPdate.
3221 * print-rtl-function.c: Do not include cfg.h
3222 * print-rtl.c: Include basic-block.h
3223 * profile-count.c: New file.
3224 * profile-count.h: New file.
3225 * profile.c (is_edge_inconsistent): Update.
3226 (correct_negative_edge_counts): Update.
3227 (is_inconsistent): Update.
3228 (set_bb_counts): Update.
3229 (read_profile_edge_counts): Update.
3230 (compute_frequency_overlap): Update.
3231 (compute_branch_probabilities): Update; Initialize and deinitialize
3233 (branch_prob): Update.
3234 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
3235 (edge_gcov_count): New.
3236 (bb_gcov_count): New.
3237 * shrink-wrap.c (try_shrink_wrapping): Update.
3238 * tracer.c (better_p): Update.
3239 * trans-mem.c (expand_transaction): Update.
3240 (ipa_tm_insert_irr_call): Update.
3241 (ipa_tm_insert_gettmclone_call): Update.
3242 * tree-call-cdce.c: Update.
3243 * tree-cfg.c (gimple_duplicate_sese_region): Update.
3244 (gimple_duplicate_sese_tail): Update.
3245 (gimple_account_profile_record): Update.
3246 (execute_fixup_cfg): Update.
3247 * tree-inline.c (copy_bb): Update.
3248 (copy_edges_for_bb): Update.
3249 (initialize_cfun): Update.
3250 (freqs_to_counts): Update.
3251 (copy_cfg_body): Update.
3252 (expand_call_inline): Update.
3253 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3254 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
3255 (try_unroll_loop_completely): Update.
3256 (try_peel_loop): Update.
3257 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3258 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
3259 * tree-ssa-loop-split.c (connect_loops): Update.
3260 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3261 * tree-ssa-reassoc.c (branch_fixup): Update.
3262 * tree-ssa-tail-merge.c (replace_block_by): Update.
3263 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
3264 (compute_path_counts): Update.
3265 (update_profile): Update.
3266 (recompute_probabilities): Update.
3267 (update_joiner_offpath_counts): Update.
3268 (estimated_freqs_path): Update.
3269 (freqs_to_counts_path): Update.
3270 (clear_counts_path): Update.
3271 (ssa_fix_duplicate_block_edges): Update.
3272 (duplicate_thread_path): Update.
3273 * tree-switch-conversion.c (case_bit_test_cmp): Update.
3274 (struct switch_conv_info): Update.
3275 * tree-tailcall.c (decrease_profile): Update.
3276 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3277 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3278 * value-prof.c (check_counter): Update.
3279 (gimple_divmod_fixed_value): Update.
3280 (gimple_mod_pow2): Update.
3281 (gimple_mod_subtract): Update.
3282 (gimple_ic_transform): Update.
3283 (gimple_stringop_fixed_value): Update.
3284 * value-prof.h (gimple_ic): Update.
3286 2017-06-02 Carl Love <cel@us.ibm.com>
3288 * config/rs6000/rs6000-c: Add support for built-in functions
3289 vector double vec_doublee (vector signed int);
3290 vector double vec_doublee (vector unsigned int);
3291 vector double vec_doublee (vector float);
3292 vector double vec_doubleh (vector signed int);
3293 vector double vec_doubleh (vector unsigned int);
3294 vector double vec_doubleh (vector float);
3295 vector double vec_doublel (vector signed int);
3296 vector double vec_doublel (vector unsigned int);
3297 vector double vec_doublel (vector float);
3298 vector double vec_doubleo (vector signed int);
3299 vector double vec_doubleo (vector unsigned int);
3300 vector double vec_doubleo (vector float);.
3301 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
3302 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
3304 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
3305 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
3306 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
3308 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
3309 vec_doublel, vec_doubleh.
3310 * doc/extend.texi: Update the built-in documentation file for the
3311 new built-in functions.
3313 2017-06-02 David Malcolm <dmalcolm@redhat.com>
3316 * ipa-inline-analysis.c (free_growth_caches): Set
3317 edge_removal_hook_holder to NULL after removing it.
3319 2017-06-02 Sudakshina Das <sudi.das@arm.com>
3321 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
3322 comparision with zero.
3324 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
3325 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3326 for early expansion of vec_min and vec_max builtins.
3327 (builtin_function_type): Add min/max unsigned variants to those
3328 identified as having unsigned arguments.
3330 2017-06-02 Olivier Hainque <hainque@adacore.com>
3332 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
3334 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3336 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
3337 Use VALL_F16 iterator rather than VALL.
3339 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3341 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
3342 Emit CBNZ inside loop when doing a strong exchange and comparing
3343 against zero. Generate the CC flags after the loop.
3345 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
3347 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
3348 (dl_section_ref): New.
3349 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
3350 On AIX, append an expression to subtract the size of the
3351 section length to dl_section_ref.
3353 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
3355 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3356 for early expansion of vector absolute builtins.
3358 2017-06-02 Richard Biener <rguenther@suse.de>
3360 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
3361 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
3363 2017-06-02 Richard Biener <rguenther@suse.de>
3365 PR tree-optimization/80948
3366 * tree-tailcall.c (find_tail_calls): Track stmts to move in
3369 2017-06-02 Richard Biener <rguenther@suse.de>
3371 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
3373 * tree-vect-stmts.c (process_use): Do not mark backedge defs
3374 for inductions as relevant.
3376 2017-06-02 Richard Biener <rguenther@suse.de>
3378 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
3379 (vectorizable_induction): ... this. Remove dead code.
3381 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
3383 * builtins. (expand_builtin_alloca): Remove second parameter and
3384 infer its value from the first parameter instead.
3385 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
3387 2017-06-02 Jakub Jelinek <jakub@redhat.com>
3389 PR rtl-optimization/80903
3390 * loop-doloop.c (add_test): Unshare sequence.
3392 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
3394 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
3396 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
3398 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
3400 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
3401 xlogue_layout::get_instance, logue_layout::xlogue_layout,
3402 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
3403 (xlogue_layout::get_stub_rtx): Make static.
3404 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
3405 (xlogue_layout::compute_stub_managed_regs): Rename to...
3406 (xlogue_layout::count_stub_managed_regs): ...this.
3407 (xlogue_layout::is_stub_managed_reg): New function.
3408 (xlogue_layout::m_stub_names): Rename to...
3409 (xlogue_layout::s_stub_names): ...this, make static.
3410 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
3411 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
3412 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
3413 xlogue_layout::s_stub_names): Instantiate statics.
3414 (stub_managed_regs): Remove.
3415 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
3416 (disable_call_ms2sysv_xlogues): Rename to...
3417 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
3418 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
3420 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
3421 change after reload_completed.
3422 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
3424 (ix86_expand_prologue): Likewise.
3425 (ix86_expand_epilogue): Likewise.
3426 (ix86_expand_split_stack_prologue): Likewise.
3427 (ix86_compute_frame_layout): Remove frame parameter ...
3428 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
3429 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
3431 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
3432 (ix86_frame): Move from here ...
3433 * config/i386/i386.h (ix86_frame): ... to here.
3434 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
3435 complete ix86_frame data structure instead. Remove some_ld_name.
3437 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
3439 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
3440 symbols that hold a DECL_VALUE_EXPR.
3442 2017-06-01 Martin Jambor <mjambor@suse.cz>
3444 PR tree-optimization/80898
3445 * tree-sra.c (process_subtree_disqualification): Removed.
3446 (disqualify_candidate): Do not acll
3447 process_subtree_disqualification.
3448 (subtree_mark_written_and_enqueue): New function.
3449 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
3450 RHS has been disqualified and re-queue LHS if necessary. Apart
3451 from that, ignore disqualified RHS.
3453 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3455 * config/s390/s390.c (s390_emit_epilogue): Disable early return
3456 address fetch for z10 or later.
3458 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
3460 * config/arc/arc.md (tst_movb): Add guard when splitting.
3462 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
3464 * config/arc/arc.c (arc_can_eliminate): Test against
3465 arc_frame_pointer_needed.
3467 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
3469 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
3470 to prevent store reordering.
3471 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
3472 (type): Add block type.
3473 (stack_tie): Define special instruction to be used in
3476 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
3478 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
3479 constraint. It is not valid for the pattern.
3480 (noncommutative_binary_comparison): Likewise.
3482 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
3484 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
3487 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
3489 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
3490 be used by the reg-alloc.
3492 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
3494 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
3495 reg-alloc when having mul64 or mul32x16 instructions.
3496 (mulsidi3): Likewise.
3497 (umulsidi3): Likewise.
3498 (mulsi32x16): New pattern.
3499 (mulsi64): Likewise.
3500 (mulsidi64): Likewise.
3501 (umulsidi64): Likewise.
3502 (MUL32x16_REG): Define.
3503 (mul64_600): Use MUL32x16_REG.
3504 (mac64_600): Likewise.
3505 (umul64_600): Likewise.
3506 (umac64_600): Likewise.
3508 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
3510 * config/arc/arc.md (mulsi3_700): Make it commutative.
3512 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
3514 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
3516 (*sign_extendsidi2_insn): Likewise for movstosw.
3518 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
3520 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
3521 the type of the input discriminant value. Convert the
3522 discriminant value of signedness vary.
3524 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
3526 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
3527 Add to -Wall section.
3529 2017-06-01 Richard Biener <rguenther@suse.de>
3532 * fold-const.c (fold_plusminus_mult_expr): If the factored
3533 factor may be zero use a wrapping type for the inner operation.
3534 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
3535 and handle moved defs.
3536 (process_assignment): Properly guard the unary op case. Return a
3537 tri-state indicating that moving the stmt before the call may allow
3538 to continue. Pass through to_move.
3539 (find_tail_calls): Handle moving unrelated defs before
3542 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
3545 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
3546 splitter result in the canonical way.
3548 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
3550 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
3551 also for 32bit target. Update insn attributes.
3552 (zero-extendsidi2 splitter): Allow all registers for operand 1.
3554 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
3556 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
3557 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
3558 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
3559 (_mm_maskz_min_ss): New intrinsics.
3561 2017-05-31 Martin Liska <mliska@suse.cz>
3563 * tree-vect-loop.c (vect_create_epilog_for_reduction):
3564 Change comment style to one we normally use.
3565 (vectorizable_reduction): Likewise.
3566 (vectorizable_induction): Likewise.
3567 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3568 (vectorizable_call): Likewise.
3569 (vectorizable_simd_clone_call): Likewise.
3570 (vectorizable_conversion): Likewise.
3571 (vectorizable_assignment): Likewise.
3572 (vectorizable_shift): Likewise.
3573 (vectorizable_operation): Likewise.
3574 (vectorizable_store): Likewise.
3575 (vectorizable_load): Likewise.
3576 * tree-vectorizer.h: Likewise.
3578 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
3580 * passes.c (emergency_dump_function): New.
3581 * tree-pass.h (emergency_dump_function): Declare.
3582 * plugin.c (plugins_internal_error_function): Remove.
3583 * plugin.h (plugins_internal_error_function): Remove declaration.
3584 * toplev.c (internal_error_function): New static function. Use it...
3585 (general_init): ...here.
3587 2017-05-31 Graham Markall <graham.markall@embecosm.com>
3589 * config/arc/arc.c (arc_print_operand): Handle constant operands.
3590 (arc_rtx_costs): Add costs for new patterns.
3591 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
3592 * config/arc/predicates.md: Add _1_2_3_operand predicate.
3594 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
3596 * tree-ssa-strlen.c (get_next_strinfo): New function.
3597 (get_stridx_plus_constant): Use it.
3598 (zero_length_string): Likewise.
3599 (adjust_related_strinfos): Likewise.
3600 (adjust_last_stmt): Likewise.
3602 2017-05-31 Richard Biener <rguenther@suse.de>
3605 * config/i386/i386.c (ix86_expand_builtin): Remove assert
3606 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
3608 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
3610 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
3611 loop_vinfo argument and use of dependence distance vectors.
3612 Check instead whether the two references differ only in their
3613 initial value and assume that they have the same alignment if the
3614 difference is a multiple of the vector alignment.
3615 (vect_analyze_data_refs_alignment): Update call accordingly.
3617 2017-05-31 Martin Liska <mliska@suse.cz>
3620 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
3622 2017-05-31 Bin Cheng <bin.cheng@arm.com>
3624 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
3625 (create_intersect_range_checks): Move from ...
3626 * tree-data-ref.c (create_intersect_range_checks_index)
3627 (create_intersect_range_checks): ... to here.
3628 (create_runtime_alias_checks): New function factored from ...
3629 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
3630 here. Call above function.
3631 * tree-data-ref.h (create_runtime_alias_checks): New function.
3633 2017-05-31 Bin Cheng <bin.cheng@arm.com>
3635 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
3636 segment length for dr_b and compute it in wide_int.
3638 2017-05-31 Richard Biener <rguenther@suse.de>
3640 PR tree-optimization/80906
3641 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
3642 and pass through iv_map.
3643 (copy_bb_and_scalar_dependences): Adjust.
3644 (translate_pending_phi_nodes): Likewise.
3645 (copy_loop_close_phi_args): Handle code-generating IVs instead
3648 2017-05-30 David Malcolm <dmalcolm@redhat.com>
3650 * diagnostic-color.c (color_dict): Add "type-diff".
3651 (parse_gcc_colors): Update comment.
3652 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
3653 -fdiagnostics-show-template-tree and -fno-elide-type.
3654 (GCC_COLORS): Add type-diff to example.
3656 (-fdiagnostics-show-template-tree): New.
3657 (-fno-elide-type): New.
3658 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
3659 the pp_format_decoder callback. Call any m_format_postprocessor's
3661 (pretty_printer::pretty_printer): Initialize
3662 m_format_postprocessor.
3663 (pretty_printer::~pretty_printer): Delete any
3664 m_format_postprocessor.
3665 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
3666 (class format_postprocessor): New class.
3667 (struct pretty_printer::format_decoder): Document the new parameters.
3668 (struct pretty_printer::m_format_postprocessor): New field.
3669 * tree-diagnostic.c (default_tree_printer): Update for new
3670 bool and const char ** params.
3671 * tree-diagnostic.h (default_tree_printer): Likewise.
3673 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
3675 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
3676 (lwa_operand): Delete rs6000_gen_cell_microcode test.
3677 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
3678 rs6000_gen_cell_microcode code.
3679 (rs6000_final_prescan_insn): Delete.
3680 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
3681 "warn-cell-microcode" entries.
3682 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
3683 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
3684 throughout. Change cc_reg_not_micro_cr0_operand to
3685 cc_reg_not_cr0_operand throughout.
3686 (*extendhi<mode>2_noload): Delete.
3687 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
3688 (mwarn-cell-microcode): Delete.
3689 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
3690 -mgen-cell-microcode and -mwarn-cell-microcode.
3692 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
3695 * config/i386/constraints.md (Yd): New constraint.
3697 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
3698 and (?Yd, r) alternatives. Update insn attributes.
3699 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
3700 and (?*Yd, r) alternatives. Update insn attributes.
3701 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
3703 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
3705 * gimplify.c (gimplify_modify_expr): Don't create a
3706 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
3709 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
3711 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
3713 2017-05-30 Richard Biener <rguenther@suse.de>
3715 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
3716 and reduc_def fields.
3717 (STMT_VINFO_REDUC_TYPE): New define.
3718 (STMT_VINFO_REDUC_DEF): Likewise.
3719 (vect_force_simple_reduction): Adjust prototype.
3720 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
3721 (vect_is_simple_reduction): Remove check_reduction argument.
3722 (vect_force_simple_reduction): Adjust and set
3723 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
3724 (vectorizable_reduction): Do not re-do reduction analysis
3725 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
3726 * tree-parloops.c (gather_scalar_reductions): Adjust.
3728 2017-05-30 Richard Biener <rguenther@suse.de>
3731 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
3734 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
3736 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
3737 Introduce unknown_misalignment parameter and remove vf.
3738 (vect_peeling_hash_get_lowest_cost):
3739 Pass unknown_misalignment parameter.
3740 (vect_enhance_data_refs_alignment):
3741 Fix unsupportable data ref treatment.
3743 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
3745 * tree-vect-data-refs.c (vect_get_data_access_cost):
3746 Workaround for SLP handling.
3747 (vect_enhance_data_refs_alignment):
3748 Compute costs for doing no peeling at all, compare to the best
3749 peeling costs so far and avoid peeling if cheaper.
3751 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
3753 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
3754 Return peeling info and set costs to zero for unlimited cost
3756 (vect_enhance_data_refs_alignment): Also inspect all datarefs
3757 with unknown misalignment. Compute and costs for unknown
3758 misalignment, compare them to the costs for known misalignment
3759 and choose the cheapest for peeling.
3761 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
3763 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
3764 (vect_get_peeling_costs_all_drs): Create function.
3765 (vect_peeling_hash_get_lowest_cost):
3766 Use vect_get_peeling_costs_all_drs.
3767 (vect_peeling_supportable): Create function.
3768 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
3770 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
3772 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
3773 DR_HAS_NEGATIVE_STEP.
3774 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
3775 (vect_enhance_data_refs_alignment): Use.
3776 (vect_duplicate_ssa_name_ptr_info): Use.
3777 * tree-vectorizer.h (dr_misalignment): Use.
3778 (known_alignment_for_access_p): Use.
3780 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3783 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
3785 (has_section_name): New function.
3787 2017-05-30 Martin Liska <mliska@suse.cz>
3790 * auto-profile.c (get_function_decl_from_block): Fix
3793 2017-05-30 Richard Biener <rguenther@suse.de>
3796 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
3798 2017-05-30 Martin Liska <mliska@suse.cz>
3800 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
3801 * dumpfile.h (struct dump_file_info): Remove ctors.
3803 2017-05-30 Martin Liska <mliska@suse.cz>
3805 * predict.def: Fix GNU coding style.
3807 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
3809 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
3810 Mark 'to' argument with ATTRIBUTE_UNUSED.
3812 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
3814 * config/xtensa/xtensa.c (xtensa_emit_call): Use
3815 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
3816 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
3819 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
3821 * doc/install.texi (Options specification): Restore entry of
3822 --enable-sjlj-exceptions.
3824 2017-05-27 Michael Eager <eager@eagercon.com>
3827 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
3829 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
3831 * config/microblaze/microblaze.h
3832 (FIXED_REGISTERS): Update in macro.
3833 (CALL_USED_REGISTERS): Update in macro.
3835 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
3837 * doc/install.texi: Add links to macOS binary distributions.
3839 2017-05-27 Jakub Jelinek <jakub@redhat.com>
3843 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
3845 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
3847 2017-05-26 Martin Liska <mliska@suse.cz>
3849 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
3851 2017-05-26 Martin Liska <mliska@suse.cz>
3853 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
3854 always leading ';; '.
3855 (dump_bb_info): Likewise.
3856 (brief_dump_cfg): Likewise.
3857 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
3858 * dumpfile.c: Remove usage of TDF_VERBOSE.
3859 * dumpfile.h (enum dump_kind): Likewise.
3860 (dump_gimple_bb_header): Do not use TDF_COMMENT.
3861 * print-tree.c (debug_verbose): Remove.
3862 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
3863 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
3864 * tree-diagnostic.c (default_tree_printer): Replace
3865 TDF_DIAGNOSTIC with TDF_SLIM.
3867 2017-05-26 Bin Cheng <bin.cheng@arm.com>
3869 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
3870 in parameter loop, rather than loop_vinfo.
3871 (create_intersect_range_checks): Ditto.
3872 (vect_create_cond_for_alias_checks): Update call to above functions.
3874 2017-05-26 Bin Cheng <bin.cheng@arm.com>
3876 PR tree-optimization/80815
3877 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
3878 for merging runtime alias checks. Handle negative DR_STEPs.
3880 2017-05-26 Bin Cheng <bin.cheng@arm.com>
3882 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
3884 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
3885 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
3886 out code pruning runtime alias checks.
3887 * tree-data-ref.c (prune_runtime_alias_test_list): New function
3888 factored out from above.
3889 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
3891 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
3893 (prune_runtime_alias_test_list): New decalaration.
3895 2017-05-26 Bin Cheng <bin.cheng@arm.com>
3897 * tree-vect-data-refs.c (compare_tree): Rename and move ...
3898 * tree-data-ref.c (data_ref_compare_tree): ... to here.
3899 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
3900 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
3901 (operator==, comp_dr_with_seg_len_pair): Ditto.
3902 (vect_prune_runtime_alias_test_list): Ditto.
3904 2017-05-26 Martin Liska <mliska@suse.cz>
3907 * params.def: Bound partial-inlining-entry-probability param.
3909 2017-05-26 Marek Polacek <polacek@redhat.com>
3912 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
3915 2017-05-26 Richard Biener <rguenther@suse.de>
3917 PR tree-optimization/80842
3918 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
3921 2017-05-26 Richard Biener <rguenther@suse.de>
3923 PR tree-optimization/80844
3924 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
3926 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
3928 * doc/md.texi (Machine Constraints): Update x86 family
3929 machine constraints section to match 'config/i386/constraints.md'.
3931 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
3933 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
3935 2017-05-25 Nathan Sidwell <nathan@acm.org>
3937 * doc/invoke.texi (--enable-languages): Update documentation.
3939 2017-05-25 Martin Liska <mliska@suse.cz>
3941 * dumpfile.c: Add TDF_FOLDING.
3942 * dumpfile.h (enum dump_kind): Likewise.
3943 * genmatch.c (dt_simplify::gen_1): Use it.
3945 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
3947 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
3949 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
3951 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
3952 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
3954 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
3956 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
3957 * match.pd (X == C): Rewrite it here.
3958 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
3959 with_certain_nonzero_bits2): New predicates.
3960 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
3962 2017-05-24 Nathan Sidwell <nathan@acm.org>
3964 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
3967 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
3970 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
3972 * config/powerpcspe: New port. Files are copied from the rs6000
3973 port, with "rs6000" in filenames replaced by "powerpcspe".
3975 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
3977 PR rtl-optimization/80754
3978 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
3980 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
3982 * config/sparc/sparc.md (length): Return the correct value for -mflat
3983 sibcalls to match output_sibcall.
3985 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
3989 * config/rs6000/rs6000.c (struct machine_function): Add new field
3991 (rs6000_get_separate_components): Init that field, use it.
3992 (rs6000_components_for_bb): Use the field.
3994 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
3996 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
3998 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
4001 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
4003 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4006 * config/s390/s390.c (s390_check_qrst_address): Check incoming
4007 address against address_operand predicate.
4008 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
4010 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
4012 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
4013 parameters passed indirectly.
4015 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
4017 * config/i386/i386.md (*movdi_internal): Remove SSE4
4018 alternative 18 (?r, *v). Update insn attributes.
4019 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
4020 Update insn attributes.
4021 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
4022 Update insn attributes.
4023 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
4024 alternative 1 (r, v). Remove isa attribute.
4025 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
4026 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
4027 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
4029 2017-05-23 Tom de Vries <tom@codesourcery.com>
4031 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
4034 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
4036 * cgraphunit.c (symbol_table::process_new_functions): Update.
4037 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
4038 (inline_generate_summary): Rename to ...
4039 (ipa_fn_summary_generate): ... this one.
4040 (inline_read_summary): Rename to ...
4041 (ipa_fn_summary_read): ... this one.
4042 (inline_write_summary): Rename to ...
4043 (ipa_fn_summary_write): ... this one.
4044 (inline_free_summary): Rename to ...
4045 (ipa_free_fn_summary): ... this one.
4046 (pass_data_local_fn_summary, pass_local_fn_summary,
4047 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
4048 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
4049 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
4050 make_pass_ipa_fn_summary): New.
4051 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
4052 inline_write_summary, inline_free_summary): Remove.
4053 (ipa_free_fn_summary) : New.
4054 * ipa-inline.c (ipa_inline): Update.
4055 (pass_ipa_inline): Do not generate summaries.
4056 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
4058 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
4059 and add pass_ipa_fn_summary.
4060 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
4062 (make_pass_inline_parameters): Remove.
4064 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
4066 * omp-low.c (struct omp_context): Remove "default_kind" member.
4069 * omp-offload.c (execute_oacc_device_lower): Remove the
4070 parallelism dimensions function attributes for unparallelized
4071 OpenACC kernels constructs.
4073 2017-05-23 Martin Liska <mliska@suse.cz>
4075 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
4077 (cgraph_edge::make_speculative): Likewise.
4078 (cgraph_edge::resolve_speculation): Likewise.
4079 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
4080 (cgraph_node::dump): Likewise.
4081 * cgraph.h: Likewise.
4082 * cgraphunit.c (analyze_functions): Likewise.
4083 (symbol_table::compile): Likewise.
4084 * ipa-cp.c (print_all_lattices): Likewise.
4085 (determine_versionability): Likewise.
4086 (initialize_node_lattices): Likewise.
4087 (ipcp_verify_propagated_values): Likewise.
4088 (estimate_local_effects): Likewise.
4089 (update_profiling_info): Likewise.
4090 (create_specialized_node): Likewise.
4091 (perhaps_add_new_callers): Likewise.
4092 (decide_about_value): Likewise.
4093 (decide_whether_version_node): Likewise.
4094 (identify_dead_nodes): Likewise.
4095 (ipcp_store_bits_results): Likewise.
4096 * ipa-devirt.c (dump_targets): Likewise.
4097 (ipa_devirt): Likewise.
4098 * ipa-icf.c (sem_item::dump): Likewise.
4099 (sem_function::equals): Likewise.
4100 (sem_variable::equals): Likewise.
4101 (sem_item_optimizer::read_section): Likewise.
4102 (sem_item_optimizer::execute): Likewise.
4103 (congruence_class::dump): Likewise.
4104 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
4105 (dump_inline_summary): Likewise.
4106 (estimate_node_size_and_time): Likewise.
4107 (inline_analyze_function): Likewise.
4108 * ipa-inline-transform.c (inline_call): Likewise.
4109 * ipa-inline.c (report_inline_failed_reason): Likewise.
4110 (want_early_inline_function_p): Likewise.
4111 (edge_badness): Likewise.
4112 (update_edge_key): Likewise.
4113 (inline_small_functions): Likewise.
4114 * ipa-profile.c (ipa_profile): Likewise.
4115 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
4116 (ipa_make_edge_direct_to_target): Likewise.
4117 (remove_described_reference): Likewise.
4118 (ipa_impossible_devirt_target): Likewise.
4119 (propagate_controlled_uses): Likewise.
4120 (ipa_print_node_params): Likewise.
4121 (ipcp_transform_function): Likewise.
4122 * ipa-pure-const.c (pure_const_read_summary): Likewise.
4123 (propagate_pure_const): Likewise.
4124 * ipa-reference.c (generate_summary): Likewise.
4125 (read_write_all_from_decl): Likewise.
4126 (propagate): Likewise.
4127 (ipa_reference_read_optimization_summary): Likewise.
4128 * ipa-utils.c (ipa_merge_profiles): Likewise.
4129 * ipa.c (walk_polymorphic_call_targets): Likewise.
4130 (symbol_table::remove_unreachable_nodes): Likewise.
4131 (ipa_single_use): Likewise.
4132 * passes.c (execute_todo): Likewise.
4133 * predict.c (drop_profile): Likewise.
4134 * symtab.c (symtab_node::get_dump_name): New function.
4135 (symtab_node::dump_name): Likewise.
4136 (symtab_node::dump_asm_name): Likewise.
4137 (symtab_node::dump_references): Likewise.
4138 (symtab_node::dump_referring): Likewise.
4139 (symtab_node::dump_base): Likewise.
4140 (symtab_node::debug_symtab): Likewise.
4141 * tree-sra.c (convert_callers_for_node): Likewise.
4142 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
4143 * value-prof.c (init_node_map): Likewise.
4145 2017-05-23 Martin Liska <mliska@suse.cz>
4147 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
4148 and symtab_node::debug_symtab to symbol_table::debug.
4149 * cgraphunit.c (analyze_functions): Use the renamed function.
4150 (symbol_table::compile): Likewise.
4151 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
4152 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
4153 * passes.c (execute_todo): Likewise.
4154 * symtab.c (symbol_table::dump): New function.
4155 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
4157 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
4159 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
4160 that nonconst implies exec.
4162 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
4164 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
4165 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
4166 (inline_edge_summary_vec): Turn into ...
4167 (ipa_call_summaries): ... this one.
4168 (redirect_to_unreachable, edge_set_predicate,
4169 evaluate_properties_for_edge, inline_summary_alloc,
4170 reset_ipa_call_summary, reset_inline_summary,
4171 inline_summary_t::duplicate): Update.
4172 (inline_edge_duplication_hook): Turn to ...
4173 (ipa_call_summary_t::duplicate): ... this one.
4174 (inline_edge_removal_hook): Turn to ...
4175 (ipa_call_summary_t::remove): ... this one.
4176 (dump_inline_edge_summary): Turn to ...
4177 (dump_ipa_call_summary): ... this one.
4178 (estimate_function_body_sizes): Update.
4179 (inline_update_callee_summaries): Update.
4180 (remap_edge_change_prob): Update.
4181 (remap_edge_summaries): Update.
4182 (inline_merge_summary): Update.
4183 (do_estimate_edge_time): Update.
4184 (inline_generate_summary): Update.
4185 (inline_read_section): Update.
4186 (inline_read_summary): Update.
4187 (inline_free_summary): Update.
4188 * ipa-inline.c (can_inline_edge_p): Update.
4189 (compute_inlined_call_time): Update.
4190 (want_inline_small_function_p): Update.
4191 (edge_badness): Update.
4192 (early_inliner): Update.
4193 * ipa-inline.h (inline_edge_summary): Turn to ...
4194 (ipa_call_summary): ... this one.
4195 (ipa_call_summary_t): New class.
4196 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
4197 (ipa_call_summaries): New.
4198 (inline_edge_summary): Remove.
4199 (estimate_edge_growth): Update.
4200 * ipa-profile.c (ipa_propagate_frequency_1): Update.
4201 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
4202 * ipa-split.c (execute_split_functions): Update.
4203 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
4205 2017-05-23 Tom de Vries <tom@codesourcery.com>
4207 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
4208 attributes): Document rdrand effective target.
4210 2017-05-23 Tom de Vries <tom@codesourcery.com>
4212 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
4213 attributes): Sort alphabetically.
4215 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
4217 * config/avr/genmultilib.awk: Use gsub instead of gensub.
4219 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
4222 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
4223 V2DF/V2DI splat into two separate patterns, one that handles
4224 registers, and the other that only handles memory. Drop support
4225 for splatting from a GPR on ISA 2.07 and then splitting the
4226 splat into direct move and splat.
4227 (vsx_splat_<mode>_reg): Likewise.
4228 (vsx_splat_<mode>_mem): Likewise.
4230 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
4232 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
4234 2017-05-22 Jakub Jelinek <jakub@redhat.com>
4237 * omp-low.c (finish_taskreg_remap): New function.
4238 (finish_taskreg_scan): If unit size of ctx->record_type
4239 is non-constant, unshare the size expression and replace
4240 decls in it with possible outer var refs.
4243 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
4244 GOVD_SHARED rather than GOVD_PRIVATE with it.
4245 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
4246 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
4249 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
4250 as last argument to build_outer_var_ref for pointer bases of array
4253 2017-05-19 Martin Sebor <msebor@redhat.com>
4255 * print-tree.c (print_node): Print DECL_READ_P flag.
4257 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
4259 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
4260 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
4261 * cgraph.c: Likewise.
4262 * cgraphunit.c: Likewise.
4263 * gengtype.c: Likewise.
4264 * ipa-cp.c: Likewise.
4265 * ipa-devirt.c: Likewise.
4266 * ipa-icf.c: Likewise.
4267 * ipa-predicate.c: Likewise.
4268 * ipa-profile.c: Likewise.
4269 * ipa-prop.c: Likewise.
4270 * ipa-split.c: Likewise.
4272 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
4273 edge_predicate_pool, dump_inline_hints,
4274 inline_summary::account_size_time, redirect_to_unreachable,
4275 edge_set_predicate, set_hint_predicate,
4276 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
4277 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
4278 inline_summary_t::remove, remap_hint_predicate_after_duplication,
4279 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
4280 ipa_call_summary_t::remove, initialize_growth_caches,
4281 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
4282 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
4283 mark_modified, unmodified_parm_1, unmodified_parm,
4284 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
4285 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
4286 compute_bb_predicates, will_be_nonconstant_expr_predicate,
4287 will_be_nonconstant_predicate, record_modified_bb_info,
4288 get_minimal_bb, record_modified, param_change_prob,
4289 phi_result_unknown_predicate, predicate_for_phi_result,
4290 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
4291 estimate_function_body_sizes, compute_inline_parameters,
4292 compute_inline_parameters_for_curren, pass_data_inline_parameters,
4293 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
4294 inline_update_callee_summaries, remap_edge_change_prob,
4295 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
4296 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
4297 inline_analyze_function, inline_summary_t::insert,
4298 inline_generate_summary, read_ipa_call_summary, inline_read_section,
4299 inline_read_summary, write_ipa_call_summary, inline_write_summary,
4300 inline_free_summary): Move to ipa-fnsummary.h
4301 (predicate_t): Remove.
4302 * ipa-fnsummary.c: New file.
4303 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
4304 (enum inline_hints_vals, inline_hints, agg_position_info,
4305 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
4306 inline_summaries, ipa_call_summary, ipa_call_summary_t,
4307 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
4308 dump_inline_summary, dump_inline_hints, inline_generate_summary,
4309 inline_read_summary, inline_write_summary, inline_free_summary,
4310 inline_analyze_function, initialize_inline_failed,
4311 inline_merge_summary, inline_update_overall_summary,
4312 compute_inline_parameters): Move to ipa-fnsummary.h
4313 * ipa-fnsummary.h: New file.
4314 * ipa-inline-transform.h: Include ipa-inline.h.
4315 * ipa-inline.c: LIkewise.
4317 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
4319 * ipa-inline.c (edge_badness): Use inlined_time instead of
4320 inline_summaries->get.
4322 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
4324 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
4326 2017-05-22 Nathan Sidwell <nathan@acm.org>
4328 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
4329 (fdump-lang): Document 'raw' option.
4330 * dumpfile.h (TDI_tu): Delete.
4331 * dumpfile.c (dump_files): Remove translation-unit.
4332 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
4334 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
4336 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
4337 command option from $(AWK) call.
4338 * config/avr/genmultilib.awk: Simplify and rewrite so that it
4339 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
4340 [FORMAT]: Remove handling of variable.
4341 * config/avr/t-multilib: Regenerate.
4343 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
4345 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
4347 (dump_inline_summary): Do not print self_time.
4348 (estimate_function_body_sizes): Do not set self_time.
4349 (compute_inline_parameters): Likewise.
4350 (inline_read_section, inline_write_summary): Do not stream self_time.
4351 * ipa-inline.h (inline_summary): Drop self_time.
4353 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
4355 * ipa-inline-analysis.c (account_size_time): Rename to ...
4356 (inline_summary::account_size_time): ... this one.
4357 (reset_ipa_call_summary): Turn to ...
4358 (ipa_call_summary::reset): ... this one.
4359 (reset_inline_summary): Turn to ...
4360 (inline_summary::reset): ... this one.
4361 (inline_summary_t::remove): Update.
4362 (inline_summary_t::duplicate): Update.
4363 (ipa_call_summary_t::remove): Update.
4364 (dump_inline_summary): Update.
4365 (estimate_function_body_sizes): Update.
4366 (compute_inline_parameters): Update.
4367 (estimate_node_size_and_time): Update.
4368 (inline_merge_summary): Update.
4369 (inline_update_overall_summary): Update.
4370 (inline_read_section): Update.
4371 (inline_write_summary): Update.
4372 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
4373 add account_size_time and reset member functions.
4374 (ipa_call_summary): Add reset function.
4375 * ipa-predicate.h (predicate::operator &): Constify.
4377 2017-05-22 Richard Biener <rguenther@suse.de>
4379 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
4381 2017-05-19 Jason Merrill <jason@redhat.com>
4383 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
4385 2017-05-19 Marek Polacek <polacek@redhat.com>
4388 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
4389 TYPE_OVERFLOW_WRAPS checks.
4391 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
4393 * tree-core.h (enum omp_clause_default_kind): Add
4394 "OMP_CLAUSE_DEFAULT_PRESENT".
4395 * tree-pretty-print.c (dump_omp_clause): Handle it.
4396 * gimplify.c (enum gimplify_omp_var_data): Add
4397 "GOVD_MAP_FORCE_PRESENT".
4398 (gimplify_adjust_omp_clauses_1): Map it to
4399 "GOMP_MAP_FORCE_PRESENT".
4400 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
4402 * gimplify.c (oacc_default_clause): Clarify.
4404 2017-05-19 Nathan Sidwell <nathan@acm.org>
4406 LANG_HOOK_REGISTER_DUMPS
4407 * toplev.c (general_init): Call register dump lang hook.
4408 * doc/invoke.texi: Document -fdump-lang option family.
4409 * dumpfile.c (dump_files): Remove class dump here.
4410 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
4411 * dumpfile.h (tree_dump_index): Remove TDI_class.
4412 * langhooks-def.h (lhd_register_dumps): Declare.
4413 (LANG_HOOKS_REGISTER_DUMPS): Define.
4414 (LANG_HOOKS_INITIALIZER): Add it.
4415 * langhooks.c (lhd_register_dumps): Define.
4416 * langhooks.h (struct lang_hooks): Add register_dumps.
4418 2017-05-19 Nathan Sidwell <nathan@acm.org>
4420 * context.h (context::set_passes): New.
4421 * context.c (context::context): Do not create pass manager.
4422 * toplev.c (general_init): Create pass manager here.
4424 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
4426 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
4427 use this splitter if two add or or instructions would also work for
4428 the constant we want to generate.
4430 2017-05-19 Richard Biener <rguenther@suse.de>
4433 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
4434 predicate evaluation.
4436 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
4438 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
4440 * ipa-inline.c (want_inline_small_function_p): Do not cast to
4443 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
4445 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
4446 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
4447 (inline_edge_summary_vec): Turn into ...
4448 (ipa_call_summaries): ... this one.
4449 (redirect_to_unreachable, edge_set_predicate,
4450 evaluate_properties_for_edge, inline_summary_alloc,
4451 reset_ipa_call_summary, reset_inline_summary,
4452 inline_summary_t::duplicate): Update.
4453 (inline_edge_duplication_hook): Turn to ...
4454 (ipa_call_summary_t::duplicate): ... this one.
4455 (inline_edge_removal_hook): Turn to ...
4456 (ipa_call_summary_t::remove): ... this one.
4457 (dump_inline_edge_summary): Turn to ...
4458 (dump_ipa_call_summary): ... this one.
4459 (estimate_function_body_sizes): Update.
4460 (inline_update_callee_summaries): Update.
4461 (remap_edge_change_prob): Update.
4462 (remap_edge_summaries): Update.
4463 (inline_merge_summary): Update.
4464 (do_estimate_edge_time): Update.
4465 (inline_generate_summary): Update.
4466 (inline_read_section): Update.
4467 (inline_read_summary): Update.
4468 (inline_free_summary): Update.
4469 * ipa-inline.c (can_inline_edge_p): Update.
4470 (compute_inlined_call_time): Update.
4471 (want_inline_small_function_p): Update.
4472 (edge_badness): Update.
4473 (early_inliner): Update.
4474 * ipa-inline.h (inline_edge_summary): Turn to ...
4475 (ipa_call_summary): ... this one.
4476 (ipa_call_summary_t): New class.
4477 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
4478 (ipa_call_summaries): New.
4479 (inline_edge_summary): Remove.
4480 (estimate_edge_growth): Update.
4481 * ipa-profile.c (ipa_propagate_frequency_1): Update.
4482 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
4483 * ipa-split.c (execute_split_functions): Update.
4484 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
4486 2017-05-19 Richard Biener <rguenther@suse.de>
4489 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
4491 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
4493 * config/rs6000/rs6000.c (struct machine_function): Add field
4494 fpr_is_wrapped_separately.
4495 (rs6000_get_separate_components): Use 64 components. Handle the
4497 (rs6000_components_for_bb): Handle the FPR components.
4498 (rs6000_emit_prologue_components): Handle the FPR components.
4499 (rs6000_emit_epilogue_components): Handle the FPR components.
4500 (rs6000_set_handled_components): Handle the FPR components.
4501 (rs6000_emit_prologue): Don't output prologue code for those FPRs
4502 that are already separately shrink-wrapped.
4503 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
4504 that are already separately shrink-wrapped.
4506 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
4509 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
4512 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
4513 (define_peephole2 for Altivec d-form load): Add peepholes to catch
4514 cases where the register allocator uses a move and an offsettable
4515 memory operation to/from a FPR register on ISA 2.06/2.07.
4516 (define_peephole2 for Altivec d-form store): Likewise.
4518 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
4521 * config/i386/mmx.md (*mov<mode>_internal): Enable
4522 alternatives 11, 12, 13 and 14 also for 32bit targets.
4523 Remove alternatives 15, 16, 17 and 18.
4524 * config/i386/sse.md (vec_concatv2di): Change
4525 alternative (!x, *y) to (x, ?!*Yn).
4527 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
4529 * dumpfile.h (enum dump_kind): Remove stray comma.
4531 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
4533 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
4534 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
4535 predicate::num_conditions
4536 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
4537 (CHANGED): turn into predicate::changed.
4538 (agg_position_info): Move to ipa-predicate.h
4539 (add_condition, predicate::add_clause, predicate::operator &=,
4540 predicate::or_with, predicate::evaluate, predicate::probability,
4541 dump_condition, dump_clause, predicate::dump,
4542 predicate::remap_after_duplication, predicate::remap_after_inlining,
4543 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
4544 (evaluate_conditions_for_known_args): Update.
4545 (set_cond_stmt_execution_predicate): Update.
4546 * ipa-inline.h: Include ipa-predicate.h
4547 (condition, inline_param_summary, conditions, agg_position_info,
4548 predicate): Move to ipa-predicate.h
4549 * ipa-predicate.c: New file.
4550 * ipa-predicate.h: New file.
4552 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
4554 * final.c (leaf_function_p): Check we are not in a sequence.
4556 2017-05-18 Martin Liska <mliska@suse.cz>
4558 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
4559 * dumpfile.c (dump_register): Use new enum dump_kind.
4560 (get_dump_file_name): Likewise.
4561 (dump_enable_all): Likewise.
4562 (dump_switch_p_1): Likewise.
4563 (enable_rtl_dump_file): Remove usage of TDF_RTL.
4564 * dumpfile.h (enum dump_kind): New enum type.
4565 (struct dump_file_info): Create constructor and
4566 format fields and comments.
4567 * passes.c (pass_manager::register_one_dump_file):
4569 * statistics.c (statistics_early_init): Likewise.
4570 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
4571 TDF_TREE with TDF_SLIM.
4572 (gather_memory_references_ref): Likewise.
4574 2017-05-18 Martin Liska <mliska@suse.cz>
4576 * vec.h (struct vnull): Use it.
4578 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
4580 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
4581 (true_predicate, false_predicate, true_predicate_p,
4582 false_predicate_p): Remove.
4583 (single_cond_predicate, not_inlined_predicate): Turn to member function
4585 (add_condition): Update.
4586 (add_clause): Turn to...
4587 (predicate::add_clause): ... this one; update; allow passing NULL
4589 (and_predicates): Turn to ...
4590 (predicate::operator &=): ... this one.
4591 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
4592 (or_predicates): Turn to ...
4593 (predicate::or_with): ... this one.
4594 (evaluate_predicate): Turn to ...
4595 (predicate::evaluate): ... this one.
4596 (predicate_probability): Turn to ...
4597 (predicate::probability): ... this one.
4598 (dump_condition): Update.
4599 (dump_predicate): Turn to ...
4600 (predicate::dump): ... this one.
4601 (account_size_time): Update.
4602 (edge_set_predicate): Update.
4603 (set_hint_predicate): UPdate.
4604 (evaluate_conditions_for_known_args): Update.
4605 (evaluate_properties_for_edge): Update.
4606 (remap_predicate_after_duplication): Turn to...
4607 (predicate::remap_after_duplication): ... this one.
4608 (remap_hint_predicate_after_duplication): Update.
4609 (inline_summary_t::duplicate): UPdate.
4610 (dump_inline_edge_summary): Update.
4611 (dump_inline_summary): Update.
4612 (set_cond_stmt_execution_predicate): Update.
4613 (set_switch_stmt_execution_predicate): Update.
4614 (compute_bb_predicates): Update.
4615 (will_be_nonconstant_expr_predicate): Update.
4616 (will_be_nonconstant_predicate): Update.
4617 (phi_result_unknown_predicate): Update.
4618 (predicate_for_phi_result): Update.
4619 (array_index_predicate): Update.
4620 (estimate_function_body_sizes): Update.
4621 (estimate_node_size_and_time): Update.
4622 (estimate_ipcp_clone_size_and_time): Update.
4623 (remap_predicate): Rename to ...
4624 (predicate::remap_after_inlining): ... this one.
4625 (remap_hint_predicate): Update.
4626 (inline_merge_summary): Update.
4627 (inline_update_overall_summary): Update.
4628 (estimate_size_after_inlining): Update.
4629 (read_predicate): Rename to ...
4630 (predicate::stream_in): ... this one.
4631 (read_inline_edge_summary): Update.
4632 (write_predicate): Rename to ...
4633 (predicate::stream_out): ... this one.
4634 (write_inline_edge_summary): Update.
4635 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
4636 (clause_t): Turn to uint32_t
4637 (predicate): Turn to class; implement constructor and operators
4639 (size_time_entry): Update.
4640 (inline_summary): Update.
4641 (inline_edge_summary): Update.
4643 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
4645 * fold-const.c (fold_binary_loc): Move transformation...
4646 * match.pd (C - X CMP X): ... here.
4648 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
4650 * config/sparc/sparc.c (sparc_option_override): Set function
4651 alignment for -mcpu=niagara7 to 64 to match the I$ line.
4652 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
4654 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
4656 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
4658 2017-05-18 Marek Polacek <polacek@redhat.com>
4661 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
4662 (pass_ubsan::execute): Call gimple_assign_single_p instead of
4663 gimple_assign_load_p.
4665 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
4668 * real.c (do_compare): Give decimal_do_compare preference over
4669 comparing just the signs.
4671 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
4673 * doc/md.texi (Canonicalization of Instructions): Describe the
4674 canonical form of instructions that inherently set a condition
4677 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
4680 * tree-cfg.c: Move deletion of unreachable case statements to after
4681 the merging of consecutive case labels.
4683 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
4685 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
4686 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
4687 restoring of callee-saved registers.
4689 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
4691 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
4692 * config/visium/visium.c (single_set_and_flags): Likewise.
4693 * config/visium/visium.md (Substitutions): Likewise.
4695 2017-05-17 Martin Liska <mliska@suse.cz>
4697 * cfg.c: Introduce dump_flags_t type and
4698 use it instead of int type.
4700 * cfghooks.c: Likewise.
4701 * cfghooks.h (struct cfg_hooks): Likewise.
4702 * cfgrtl.c: Likewise.
4703 * cfgrtl.h: Likewise.
4704 * cgraph.c (cgraph_node::get_body): Likewise.
4705 * coretypes.h: Likewise.
4706 * domwalk.c: Likewise.
4707 * domwalk.h: Likewise.
4708 * dumpfile.c (struct dump_option_value_info): Likewise.
4709 (dump_enable_all): Likewise.
4710 (dump_switch_p_1): Likewise.
4711 (opt_info_switch_p): Likewise.
4712 * dumpfile.h (enum tree_dump_index): Likewise.
4713 (struct dump_file_info): Likewise.
4714 * genemit.c: Likewise.
4715 * generic-match-head.c: Likewise.
4716 * gengtype.c (open_base_files): Likewise.
4717 * gimple-pretty-print.c: Likewise.
4718 * gimple-pretty-print.h: Likewise.
4719 * graph.c (print_graph_cfg): Likewise.
4720 * graphite-scop-detection.c (dot_all_sese): Likewise.
4721 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
4722 * loop-unroll.c (report_unroll): Likewise.
4723 * passes.c (pass_manager::register_one_dump_file): Likewise.
4724 * print-tree.c: Likewise.
4725 * statistics.c: Likewise.
4726 * tree-cfg.c: Likewise.
4727 * tree-cfg.h: Likewise.
4728 * tree-dfa.c: Likewise.
4729 * tree-dfa.h: Likewise.
4730 * tree-dump.c (dump_function): Likewise.
4731 * tree-dump.h (struct dump_info): Likewise.
4732 * tree-pretty-print.c: Likewise.
4733 * tree-pretty-print.h: Likewise.
4734 * tree-ssa-live.c: Likewise.
4735 * tree-ssa-live.h: Likewise.
4736 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
4737 * tree-vect-loop.c: Likewise.
4738 * tree-vect-slp.c: Likewise.
4740 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4741 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4743 PR tree-optimization/80457
4744 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
4745 of all arguments to a statement as scalar_to_vec operations.
4746 (vectorizable_call): Adjust call to vect_model_simple_cost for
4748 (vectorizable_conversion): Likewise.
4749 (vectorizable_assignment): Likewise.
4750 (vectorizable_shift): Likewise.
4751 (vectorizable_operation): Likewise.
4752 (vectorizable_comparison): Likewise.
4753 (vect_is_simple_cond): Record the def types for operands.
4754 (vectorizable_condition): Likewise, call vect_model_simple_cost.
4755 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
4756 for statement argument count.
4758 2017-05-16 Carl Love <cel@us.ibm.com>
4760 * config/rs6000/rs6000-c: Add support for built-in functions
4761 vector unsigned long long vec_bperm (vector unsigned long long,
4762 vector unsigned char)
4763 vector signed long long vec_mule (vector signed int,
4765 vector unsigned long long vec_mule (vector unsigned int,
4766 vector unsigned int)
4767 vector signed long long vec_mulo (vector signed int,
4769 vector unsigned long long vec_mulo (vector unsigned int,
4770 vector unsigned int)
4771 vector signed char vec_sldw (vector signed char,
4774 vector unsigned char vec_sldw (vector unsigned char,
4775 vector unsigned char,
4777 vector signed short vec_sldw (vector signed short,
4778 vector signed short,
4780 vector unsigned short vec_sldw (vector unsigned short,
4781 vector unsigned short,
4783 vector signed int vec_sldw (vector signed int,
4786 vector unsigned int vec_sldw (vector unsigned int,
4787 vector unsigned int,
4789 vector signed long long vec_sldw (vector signed long long,
4790 vector signed long long,
4792 vector unsigned long long vec_sldw (vector unsigned long long,
4793 vector unsigned long long,
4795 * config/rs6000/rs6000-c: Add support for built-in functions
4796 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
4797 * config/rs6000/altivec.h: Add defintion for vec_sldw.
4798 * doc/extend.texi: Update the built-in documentation for the
4799 new built-in functions.
4801 2017-05-16 Marek Polacek <polacek@redhat.com>
4805 * tree.c (save_expr): Don't fold the expression.
4807 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
4809 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
4810 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
4811 and (?*y,m). Update insn attributes.
4813 2017-05-16 Martin Liska <mliska@suse.cz>
4815 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
4816 flags argument of print_gimple_stmt, print_gimple_expr,
4817 print_generic_stmt and print_generic_expr.
4818 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
4819 * coretypes.h: Likewise.
4820 * except.c (dump_eh_tree): Likewise.
4821 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
4822 * gimple-pretty-print.h: Likewise.
4823 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
4824 (backprop::push_to_worklist): Likewise.
4825 (backprop::pop_from_worklist): Likewise.
4826 (backprop::process_use): Likewise.
4827 (backprop::intersect_uses): Likewise.
4828 (note_replacement): Likewise.
4829 * gimple-ssa-store-merging.c
4830 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
4831 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
4832 (pass_store_merging::execute): Likewise.
4833 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
4834 (ssa_base_cand_dump_callback): Likewise.
4835 (dump_incr_vec): Likewise.
4836 (replace_refs): Likewise.
4837 (replace_mult_candidate): Likewise.
4838 (create_add_on_incoming_edge): Likewise.
4839 (create_phi_basis): Likewise.
4840 (insert_initializers): Likewise.
4841 (all_phi_incrs_profitable): Likewise.
4842 (introduce_cast_before_cand): Likewise.
4843 (replace_one_candidate): Likewise.
4844 * gimplify.c (gimplify_expr): Likewise.
4845 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
4846 (set_rename): Likewise.
4847 (rename_uses): Likewise.
4848 (copy_loop_phi_nodes): Likewise.
4849 (add_close_phis_to_merge_points): Likewise.
4850 (copy_loop_close_phi_args): Likewise.
4851 (copy_cond_phi_args): Likewise.
4852 (graphite_copy_stmts_from_block): Likewise.
4853 (translate_pending_phi_nodes): Likewise.
4854 * graphite-poly.c (print_pdr): Likewise.
4855 (dump_gbb_cases): Likewise.
4856 (dump_gbb_conditions): Likewise.
4857 (print_scop_params): Likewise.
4858 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
4859 (build_cross_bb_scalars_use): Likewise.
4860 (gather_bbs::before_dom_children): Likewise.
4861 * hsa-dump.c (dump_hsa_immed): Likewise.
4862 * ipa-cp.c (print_ipcp_constant_value): Likewise.
4863 (get_replacement_map): Likewise.
4864 * ipa-inline-analysis.c (dump_condition): Likewise.
4865 (estimate_function_body_sizes): Likewise.
4866 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
4867 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
4868 * ipa-prop.c (ipa_dump_param): Likewise.
4869 (ipa_print_node_jump_functions_for_edge): Likewise.
4870 (ipa_modify_call_arguments): Likewise.
4871 (ipa_modify_expr): Likewise.
4872 (ipa_dump_param_adjustments): Likewise.
4873 (ipa_dump_agg_replacement_values): Likewise.
4874 (ipcp_modif_dom_walker::before_dom_children): Likewise.
4875 * ipa-pure-const.c (check_stmt): Likewise.
4876 (pass_nothrow::execute): Likewise.
4877 * ipa-split.c (execute_split_functions): Likewise.
4878 * omp-offload.c (dump_oacc_loop_part): Likewise.
4879 (dump_oacc_loop): Likewise.
4880 * trans-mem.c (tm_log_emit): Likewise.
4881 (tm_memopt_accumulate_memops): Likewise.
4882 (dump_tm_memopt_set): Likewise.
4883 (dump_tm_memopt_transform): Likewise.
4884 * tree-cfg.c (gimple_verify_flow_info): Likewise.
4885 (print_loop): Likewise.
4886 * tree-chkp-opt.c (chkp_print_addr): Likewise.
4887 (chkp_gather_checks_info): Likewise.
4888 (chkp_get_check_result): Likewise.
4889 (chkp_remove_check_if_pass): Likewise.
4890 (chkp_use_outer_bounds_if_possible): Likewise.
4891 (chkp_reduce_bounds_lifetime): Likewise.
4892 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
4893 (chkp_mark_completed_bounds): Likewise.
4894 (chkp_register_incomplete_bounds): Likewise.
4895 (chkp_mark_invalid_bounds): Likewise.
4896 (chkp_maybe_copy_and_register_bounds): Likewise.
4897 (chkp_build_returned_bound): Likewise.
4898 (chkp_get_bound_for_parm): Likewise.
4899 (chkp_build_bndldx): Likewise.
4900 (chkp_get_bounds_by_definition): Likewise.
4901 (chkp_generate_extern_var_bounds): Likewise.
4902 (chkp_get_bounds_for_decl_addr): Likewise.
4903 * tree-chrec.c (chrec_apply): Likewise.
4904 * tree-data-ref.c (dump_data_reference): Likewise.
4905 (dump_subscript): Likewise.
4906 (dump_data_dependence_relation): Likewise.
4907 (analyze_overlapping_iterations): Likewise.
4908 * tree-inline.c (expand_call_inline): Likewise.
4909 (tree_function_versioning): Likewise.
4910 * tree-into-ssa.c (dump_defs_stack): Likewise.
4911 (dump_currdefs): Likewise.
4912 (dump_names_replaced_by): Likewise.
4913 (dump_update_ssa): Likewise.
4914 (update_ssa): Likewise.
4915 * tree-object-size.c (pass_object_sizes::execute): Likewise.
4916 * tree-parloops.c (build_new_reduction): Likewise.
4917 (try_create_reduction_list): Likewise.
4918 (ref_conflicts_with_region): Likewise.
4919 (oacc_entry_exit_ok_1): Likewise.
4920 (oacc_entry_exit_single_gang): Likewise.
4921 * tree-pretty-print.h: Likewise.
4922 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
4923 (get_scalar_evolution): Likewise.
4924 (add_to_evolution): Likewise.
4925 (get_loop_exit_condition): Likewise.
4926 (analyze_evolution_in_loop): Likewise.
4927 (analyze_initial_condition): Likewise.
4928 (analyze_scalar_evolution): Likewise.
4929 (instantiate_scev): Likewise.
4930 (number_of_latch_executions): Likewise.
4931 (gather_chrec_stats): Likewise.
4932 (final_value_replacement_loop): Likewise.
4933 (scev_const_prop): Likewise.
4934 * tree-sra.c (dump_access): Likewise.
4935 (disqualify_candidate): Likewise.
4936 (create_access): Likewise.
4938 (maybe_add_sra_candidate): Likewise.
4939 (create_access_replacement): Likewise.
4940 (analyze_access_subtree): Likewise.
4941 (analyze_all_variable_accesses): Likewise.
4942 (sra_modify_assign): Likewise.
4943 (initialize_constant_pool_replacements): Likewise.
4944 (find_param_candidates): Likewise.
4945 (decide_one_param_reduction): Likewise.
4946 (replace_removed_params_ssa_names): Likewise.
4947 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
4948 * tree-ssa-copy.c (dump_copy_of): Likewise.
4949 (copy_prop_visit_cond_stmt): Likewise.
4950 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
4951 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
4952 (record_equivalences_from_stmt): Likewise.
4953 * tree-ssa-dse.c (compute_trims): Likewise.
4954 (delete_dead_call): Likewise.
4955 (delete_dead_assignment): Likewise.
4956 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
4957 (forward_propagate_into_cond): Likewise.
4958 (pass_forwprop::execute): Likewise.
4959 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
4960 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
4962 (move_computations_worker): Likewise.
4963 (execute_sm): Likewise.
4964 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
4965 (remove_exits_and_undefined_stmts): Likewise.
4966 (remove_redundant_iv_tests): Likewise.
4967 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
4968 (adjust_iv_update_pos): Likewise.
4969 * tree-ssa-math-opts.c (bswap_replace): Likewise.
4970 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
4971 (value_replacement): Likewise.
4972 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
4973 * tree-ssa-pre.c (print_pre_expr): Likewise.
4974 (get_representative_for): Likewise.
4975 (create_expression_by_pieces): Likewise.
4976 (insert_into_preds_of_block): Likewise.
4977 (eliminate_insert): Likewise.
4978 (eliminate_dom_walker::before_dom_children): Likewise.
4979 (eliminate): Likewise.
4980 (remove_dead_inserted_code): Likewise.
4981 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
4982 * tree-ssa-reassoc.c (get_rank): Likewise.
4983 (eliminate_duplicate_pair): Likewise.
4984 (eliminate_plus_minus_pair): Likewise.
4985 (eliminate_not_pairs): Likewise.
4986 (undistribute_ops_list): Likewise.
4987 (eliminate_redundant_comparison): Likewise.
4988 (update_range_test): Likewise.
4989 (optimize_range_tests_var_bound): Likewise.
4990 (optimize_vec_cond_expr): Likewise.
4991 (rewrite_expr_tree): Likewise.
4992 (rewrite_expr_tree_parallel): Likewise.
4993 (linearize_expr): Likewise.
4994 (break_up_subtract): Likewise.
4995 (linearize_expr_tree): Likewise.
4996 (attempt_builtin_powi): Likewise.
4997 (attempt_builtin_copysign): Likewise.
4998 (transform_stmt_to_copy): Likewise.
4999 (transform_stmt_to_multiply): Likewise.
5000 (dump_ops_vector): Likewise.
5001 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
5002 (print_scc): Likewise.
5003 (set_ssa_val_to): Likewise.
5004 (visit_reference_op_store): Likewise.
5005 (visit_use): Likewise.
5006 (sccvn_dom_walker::before_dom_children): Likewise.
5007 (run_scc_vn): Likewise.
5008 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
5010 (expr_hash_elt::print): Likewise.
5011 (const_and_copies::pop_to_marker): Likewise.
5012 (const_and_copies::record_const_or_copy_raw): Likewise.
5013 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
5014 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
5015 (dump_predicates): Likewise.
5016 (find_uninit_use): Likewise.
5017 (warn_uninitialized_phi): Likewise.
5018 (pass_late_warn_uninitialized::execute): Likewise.
5019 * tree-ssa.c (verify_vssa): Likewise.
5020 (verify_ssa): Likewise.
5021 (maybe_optimize_var): Likewise.
5022 * tree-vrp.c (dump_value_range): Likewise.
5023 (dump_all_value_ranges): Likewise.
5024 (dump_asserts_for): Likewise.
5025 (register_edge_assert_for_2): Likewise.
5026 (vrp_visit_cond_stmt): Likewise.
5027 (vrp_visit_switch_stmt): Likewise.
5028 (vrp_visit_stmt): Likewise.
5029 (vrp_visit_phi_node): Likewise.
5030 (simplify_cond_using_ranges_1): Likewise.
5031 (fold_predicate_in): Likewise.
5032 (evrp_dom_walker::before_dom_children): Likewise.
5033 (evrp_dom_walker::push_value_range): Likewise.
5034 (evrp_dom_walker::pop_value_range): Likewise.
5035 (execute_early_vrp): Likewise.
5037 2017-05-16 Richard Biener <rguenther@suse.de>
5039 * dwarf2out.c (loc_list_from_tree_1): Do not create
5040 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
5042 2017-05-16 Richard Biener <rguenther@suse.de>
5044 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
5046 (note_variable_value_in_expr): If we resolved the decl ref
5047 do not push to the stack.
5049 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
5051 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
5052 operations in fast-math mode.
5053 (vaddq_f16): Likewise.
5054 (vmul_f16): Likewise.
5055 (vmulq_f16): Likewise.
5056 (vsub_f16): Likewise.
5057 (vsubq_f16): Likewise.
5058 * config/arm/neon.md (add<mode>3): New.
5060 (fma:<VH:mode>3): New. Also remove outdated comment.
5063 2017-05-16 Martin Liska <mliska@suse.cz>
5067 * ipa-devirt.c (warn_types_mismatch): Fix typo.
5068 (odr_types_equivalent_p): Likewise.
5070 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
5072 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
5074 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
5077 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
5078 non-interunit SSE move alternatives with '?'.
5079 (zero-extendsidi peephole2): New peephole to skip intermediate
5080 general register in SSE zero-extend sequence.
5082 2017-05-15 Jeff Law <law@redhat.com>
5084 * reorg.c (relax_delay_slots): Create a new variable to hold
5085 the temporary target rather than clobbering TARGET_LABEL.
5087 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
5088 missing argument to extract_bit_field call.
5089 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
5091 2017-05-15 Martin Liska <mliska@suse.cz>
5094 * gcc.c (process_command): Do not allow empty argument of -o option.
5096 2017-05-15 Renlin Li <renlin.li@arm.com>
5098 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
5099 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
5100 * config/aarch64/constraints.md (Usf): Add long call check.
5101 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
5102 (call_value): Likewise.
5103 (sibcall): Likewise.
5104 (sibcall_value): Likewise.
5106 (call_value_insn): New.
5107 (sibcall_insn): Update rtx pattern.
5108 (sibcall_value_insn): Likewise.
5109 (call_internal): Remove.
5110 (call_value_internal): Likewise.
5111 (sibcall_internal): Likewise.
5112 (sibcall_value_internal): Likewise.
5113 (call_reg): Likewise.
5114 (call_symbol): Likewise.
5115 (call_value_reg): Likewise.
5116 (call_value_symbol): Likewise.
5118 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
5121 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
5123 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
5125 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
5126 compatible with CCGOCmode and with CCZmode.
5128 2017-05-14 Martin Sebor <msebor@redhat.com>
5131 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
5132 (gimple_fold_builtin_snprintf): Same.
5133 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
5134 (gimple_fold_builtin_snprintf): Same.
5135 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
5137 (is_call_safe): New function.
5138 (try_substitute_return_value): Call it.
5139 (try_simplify_call): New function.
5140 (pass_sprintf_length::handle_gimple_call): Call it.
5142 2017-05-14 Martin Sebor <msebor@redhat.com>
5145 * builtins.c (expand_builtin_stpncpy): Simplify.
5147 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
5149 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
5150 * config/i386/i386.h
5151 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
5152 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
5153 (struct machine_function): Add new members call_ms2sysv,
5154 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
5155 (struct machine_frame_state): New fields sp_realigned and
5156 sp_realigned_offset.
5157 * config/i386/i386.c
5158 (enum xlogue_stub): New enum.
5159 (enum xlogue_stub_sets): New enum.
5160 (class xlogue_layout): New class.
5161 (struct ix86_frame): New fields stack_realign_allocate_offset,
5162 stack_realign_offset and outlined_save_offset. Modify comments to
5163 detail stack layout when using out-of-line stubs.
5164 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
5165 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
5166 -mcall-ms2sysv-xlogues.
5167 (stub_managed_regs): New static variable.
5168 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
5169 registers managed by out-of-line stub.
5170 (disable_call_ms2sysv_xlogues): New function.
5171 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
5172 m->call_ms2sysv when appropriate and compute frame layout for
5174 (sp_valid_at, fp_valid_at): New inline functions.
5175 (choose_basereg): New function.
5176 (choose_baseaddr): Add align parameter, use choose_basereg and modify
5178 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
5179 Use align parameter of choose_baseaddr to generated aligned SSE movs
5181 (pro_epilogue_adjust_stack): Modify to track
5182 machine_frame_state::sp_realigned.
5183 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
5184 (ix86_nsaved_sseregs): Likewise.
5185 (ix86_emit_save_regs): Likewise.
5186 (ix86_emit_save_regs_using_mov): Likewise.
5187 (ix86_emit_save_sse_regs_using_mov): Likewise.
5188 (get_scratch_register_on_entry): Likewise.
5189 (gen_frame_set): New function.
5190 (gen_frame_load): Likewise.
5191 (gen_frame_store): Likewise.
5192 (emit_outlined_ms2sysv_save): Likewise.
5193 (emit_outlined_ms2sysv_restore): Likewise.
5194 (ix86_expand_prologue): Modify stack re-alignment code and call
5195 emit_outlined_ms2sysv_save when appropriate.
5196 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
5197 parameter rtx_insn *insn, which allows the function to be used to only
5199 (ix86_expand_epilogue): Modify validity checks of frame and stack
5200 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
5201 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
5202 * config/i386/predicates.md
5203 (save_multiple): New predicate.
5204 (restore_multiple): Likewise.
5205 * config/i386/sse.md
5206 (save_multiple<mode>): New pattern.
5207 (save_multiple_realign<mode>): Likewise.
5208 (restore_multiple<mode>): Likewise.
5209 (restore_multiple_and_return<mode>): Likewise.
5210 (restore_multiple_leave_return<mode>): Likewise.
5211 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
5213 2017-05-14 Julia Koval <julia.koval@intel.com>
5215 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
5216 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
5217 (__builtin_ia32_xsetbv): New builtins.
5218 * config/i386/i386.c (ix86_expand_special_args_builtin):
5220 (ix86_expand_builtin): Special expand for new intrinsics.
5221 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
5222 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
5223 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
5225 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5227 * cfganal.c (inverted_post_order_compute): Change argument type
5229 * cfganal.h (inverted_post_order_compute): Adjust prototype.
5230 * df-core.c (rest_of_handle_df_initialize): Adjust.
5231 (rest_of_handle_df_finish): Likewise.
5232 (df_analyze_1): Likewise.
5233 (df_analyze): Likewise.
5234 (loop_inverted_post_order_compute): Change argument to be a vec *.
5235 (df_analyze_loop): Adjust.
5236 (df_get_n_blocks): Likewise.
5237 (df_get_postorder): Likewise.
5238 * df.h (struct df_d): Change field to be a vec.
5239 * lcm.c (compute_laterin): Adjust.
5240 (compute_available): Likewise.
5241 * lra-lives.c (lra_create_live_ranges_1): Likewise.
5242 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
5243 * tree-ssa-pre.c (compute_antic): Likewise.
5245 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5247 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
5248 (depth_first_search::depth_first_search): Change structure init
5249 function to this constructor.
5250 (depth_first_search::add_bb): Rename function to this member.
5251 (depth_first_search::execute): Likewise.
5252 (flow_dfs_compute_reverse_finish): Adjust.
5254 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5256 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
5257 (longest_simple_path): Likewise.
5258 * shrink-wrap.c (spread_components): Likewise.
5259 (disqualify_problematic_components): Likewise.
5260 (emit_common_heads_for_components): Likewise.
5261 (emit_common_tails_for_components): Likewise.
5262 (insert_prologue_epilogue_for_components): Likewise.
5264 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5266 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
5269 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5271 * df-core.c (df_set_blocks): Start using auto_bitmap.
5272 (df_compact_blocks): Likewise.
5273 * df-problems.c (df_rd_confluence_n): Likewise.
5274 * df-scan.c (df_insn_rescan_all): Likewise.
5275 (df_process_deferred_rescans): Likewise.
5276 (df_update_entry_block_defs): Likewise.
5277 (df_update_exit_block_uses): Likewise.
5278 (df_entry_block_bitmap_verify): Likewise.
5279 (df_exit_block_bitmap_verify): Likewise.
5280 (df_scan_verify): Likewise.
5281 * lra-constraints.c (lra_constraints): Likewise.
5282 (undo_optional_reloads): Likewise.
5283 (lra_undo_inheritance): Likewise.
5284 * lra-remat.c (calculate_gen_cands): Likewise.
5285 (do_remat): Likewise.
5286 * lra-spills.c (assign_spill_hard_regs): Likewise.
5287 (spill_pseudos): Likewise.
5288 * tree-ssa-pre.c (bitmap_set_and): Likewise.
5289 (bitmap_set_subtract_values): Likewise.
5291 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5293 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
5294 management with auto_bitmap.
5295 (fix_inter_tick): Likewise.
5296 (fix_recovery_deps): Likewise.
5297 * ira.c (add_store_equivs): Likewise.
5298 (find_moveable_pseudos): Likewise.
5299 (split_live_ranges_for_shrink_wrap): Likewise.
5300 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
5301 (rtx_reuse_manager::seen_def_p): Likewise.
5302 (rtx_reuse_manager::set_seen_def): Likewise.
5303 * print-rtl.h (class rtx_reuse_manager): Likewise.
5305 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5307 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
5309 (migrate_btr_def): Likewise.
5310 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
5311 * df-core.c (loop_post_order_compute): Likewise.
5312 (loop_inverted_post_order_compute): Likewise.
5313 * hsa-common.h: Likewise.
5314 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
5315 * init-regs.c (initialize_uninitialized_regs): Likewise.
5316 * ipa-inline.c (resolve_noninline_speculation): Likewise.
5317 (inline_small_functions): Likewise.
5318 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
5319 * ira.c (combine_and_move_insns): Likewise.
5320 (build_insn_chain): Likewise.
5321 * loop-invariant.c (find_invariants): Likewise.
5322 * lower-subreg.c (propagate_pseudo_copies): Likewise.
5323 * predict.c (tree_predict_by_opcode): Likewise.
5324 (predict_paths_leading_to): Likewise.
5325 (predict_paths_leading_to_edge): Likewise.
5326 (estimate_loops_at_level): Likewise.
5327 (estimate_loops): Likewise.
5328 * shrink-wrap.c (try_shrink_wrapping): Likewise.
5329 (spread_components): Likewise.
5330 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
5331 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
5332 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
5333 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
5334 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
5335 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
5336 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
5337 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
5338 (mark_threaded_blocks): Likewise.
5339 (thread_through_all_blocks): Likewise.
5340 * tree-ssa.c (verify_ssa): Likewise.
5341 (execute_update_addresses_taken): Likewise.
5342 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
5344 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5346 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
5348 (post_order_compute): Likewise.
5349 (inverted_post_order_compute): Likewise.
5350 (pre_and_rev_post_order_compute_fn): Likewise.
5352 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5354 * genrecog.c (int_set::int_set): Explicitly construct our
5355 auto_vec base class.
5356 * vec.h (auto_vec::auto_vec): New constructor.
5358 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5360 * bitmap.h (class auto_bitmap): New constructor taking
5361 bitmap_obstack * argument.
5363 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5365 * bitmap.h (class auto_bitmap): Change type of m_bits to
5366 bitmap_head, and adjust ctor / dtor and member operators.
5368 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
5370 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
5371 when returned register mode doesn't match original mode.
5373 2017-05-12 Jeff Law <law@redhat.com>
5374 Jakub Jelinek <jakub@redhat.com>
5376 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
5377 we look for cc setter after the compare-elim changes.
5378 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
5379 within the vector to match what compare-elim now expects.
5380 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
5381 (xorsi3_flags, one_cmplsi2_flags): Likewise.
5383 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
5384 after the compare-elim changes.
5385 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
5386 the vector to match what compare-elim now expects.
5387 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
5388 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
5389 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
5390 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
5391 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
5393 * config/visium/visium.c (single_set_and_flags): Fix where
5394 we look for cc setter after the compare-elim changes.
5395 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
5396 with the vector to match what compare-elim now expects.
5397 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
5398 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
5399 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
5400 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
5401 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
5402 (neg<mode>2_insn_set_overflow): Likewise.
5404 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
5407 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
5408 maybe_expand_insn call, set ops[0].target. If still set after call,
5409 set alt_rtl. Add extra arg to recursive calls.
5410 (extract_bit_field): Add alt_rtl argument. Pass to
5412 * expmed.h (extract_bit_field): Fix prototype.
5413 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
5414 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
5415 to extract_bit_field_calls.
5416 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
5417 Pass alt_rtl to extract_bit_field calls.
5418 * calls.c (store_unaligned_arguments_into_psuedos)
5419 load_register_parameters): Pass extra NULL to extract_bit_field calls.
5420 * optabs.c (maybe_legitimize_operand): Clear op->target when call
5422 * optabs.h (struct expand_operand): Add target bitfield.
5424 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
5426 * compare-elim.c (try_eliminate_compare): Canonicalize
5427 operation with embedded compare to
5428 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
5429 (set (reg) (operation)].
5431 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
5433 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
5436 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
5437 cost of adding a carry flag for ADC instruction.
5438 [case MINUS]: Ignore the cost of subtracting a carry flag
5439 for SBB instruction.
5441 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
5443 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
5445 * config/rs6000/bmiintrin.h: New file.
5446 * config/rs6000/bmi2intrin.h: New file.
5447 * config/rs6000/x86intrin.h: New file.
5449 2017-05-12 Jeff Law <law@redhat.com>
5451 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
5454 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
5457 * tree-cfg.c: Remove cfg edges of unreachable case statements.
5459 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
5461 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
5462 early expansion of vector divide builtins.
5463 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
5464 builtins identified as having unsigned arguments.
5466 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
5468 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
5469 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
5470 expansion of vector logical operations (and, andc, or, xor,
5473 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
5475 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
5476 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
5478 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
5480 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
5481 early GIMPLE expansion of vector multiplies.
5483 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
5485 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
5486 TARGET_HAVE_MOVT conditional.
5487 (movt splitter): Likewise.
5489 2017-05-12 Richard Biener <rguenther@suse.de>
5491 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
5492 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
5493 Fold all stmts not inplace.
5495 2017-05-12 Richard Biener <rguenther@suse.de>
5497 PR tree-optimization/80713
5498 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
5499 inserted_exprs bit for not removed stmts.
5501 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
5504 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
5505 parallelized" attribute for parallelized OpenACC kernels.
5506 * omp-offload.c (execute_oacc_device_lower): Use it.
5508 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
5509 Set "oacc kernels" attribute.
5510 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
5511 parameter. Adjust all users.
5512 (oacc_fn_attrib_kernels_p): Remove function.
5513 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
5514 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
5515 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
5516 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
5517 assert "oacc kernels" attribute is set.
5519 2017-05-11 Carl Love <cel@us.ibm.com>
5521 * config/rs6000/rs6000-c: Add support for built-in functions
5522 vector unsigned char vec_popcnt (vector signed char)
5523 vector unsigned char vec_popcnt (vector unsigned char)
5524 vector unsigned short vec_popcnt (vector signed short)
5525 vector unsigned short vec_popcnt (vector unsigned short)
5526 vector unsigned int vec_popcnt (vector signed int)
5527 vector unsigned int vec_popcnt (vector unsigned int)
5528 vector unsigned long long vec_popcnt (vector signed long long)
5529 vector unsigned long long vec_popcnt (vector unsigned long long)
5530 vector signed long long vec_slo (vector signed long long,
5532 vector signed long long vec_slo (vector signed long long,
5533 vector unsigned char)
5534 vector unsigned long long vec_slo (vector unsigned long long,
5536 vector unsigned long long vec_slo (vector unsigned long long,
5537 vector unsigned char)
5538 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
5539 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
5540 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
5541 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
5542 * doc/extend.texi: Update the built-in documentation file for the
5543 new built-in functions.
5545 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
5547 * attribs.h (sorted_attr_string): Move machine independent
5548 functions for target clone support from the i386 port to common
5549 code. Rename ix86_function_versions to common_function_versions.
5550 Rename make_name to make_unique_name.
5551 (common_function_versions): Likewise.
5552 (make_unique_name): Likewise.
5553 (make_dispatcher_decl): Likewise.
5554 (is_function_default_version): Likewise.
5555 * attribs.c (attr_strcmp): Likewise.
5556 (sorted_attr_string): Likewise.
5557 (common_function_versions): Likewise.
5558 (make_unique_name): Likewise.
5559 (make_dispatcher_decl): Likewise.
5560 (is_function_default_version): Likewise.
5561 * config/i386/i386.c (attr_strcmp): Likewise.
5562 (sorted_attr_string): Likewise.
5563 (ix86_function_versions): Likewise.
5564 (make_name): Likewise.
5565 (make_dispatcher_decl): Likewise.
5566 (is_function_default_version): Likewise.
5567 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
5569 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5572 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
5573 Account for direct move costs for vec_construct of integer
5576 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
5579 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
5580 (UNSPEC_STX_ATOMIC): Ditto.
5581 (loaddi_via_sse): New insn.
5582 (storedi_via_sse): Ditto.
5583 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
5584 Update corresponding peephole2 patterns.
5585 (atomic_storedi_fpu): Ditto.
5587 2017-05-11 Julia Koval <julia.koval@intel.com>
5589 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
5590 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
5592 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
5593 (__builtin_ia32_rsqrt14ss_mask): New builtins.
5594 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
5596 2017-05-11 Nathan Sidwell <nathan@acm.org>
5598 * graphite-poly.c: Include dumpfile.h.
5600 * dumpfle.h (dump_function): Declare here ...
5601 * tree-dump.h (dump_function): ... not here.
5602 * dumpfile.c: #include tree-cfg.h.
5603 (dump_function): Move here from ...
5604 * tree-dump.c (dump_function): ... here.
5605 * gimplify.c: #include splay-tree.h, not tree-dump.h.
5606 * graphite-poly.c: Don't include tree-dump.h.
5607 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
5608 * print-tree.c: Likewise.
5609 * stor-layout.c: Likewise.
5610 * tree-nested.c: Likewise.
5612 * dumpfile.c (dump_start): Use TDF_FLAGS.
5613 (dump_enable_all): Fix TDF_KIND check thinko.
5615 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
5617 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5618 array entries to represent two legal parameterizations of the
5619 overloaded __builtin_cmpb function, as represented by the
5620 P6_OV_BUILTIN_CMPB constant.
5621 (altivec_resolve_overloaded_builtin): Add special case handling
5622 for the __builtin_cmpb function, as represented by the
5623 P6_OV_BUILTIN_CMPB constant.
5624 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
5625 (BU_P6_64BIT_2): New macro.
5626 (BU_P6_OVERLOAD_2): New macro
5627 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
5628 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
5629 (CMPB): Add overload support to represent both 32-bit and 64-bit
5630 compare-bytes function.
5631 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
5632 support for TARGET_CMPB.
5633 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
5634 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
5635 documentation of the __builtin_cmpb overloaded built-in function.
5637 2017-05-11 Richard Biener <rguenther@suse.de>
5639 PR tree-optimization/80705
5640 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
5641 bases are not vectorizable.
5643 2017-05-11 Bin Cheng <bin.cheng@arm.com>
5645 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
5646 when counting register pressure.
5648 2017-05-11 Bin Cheng <bin.cheng@arm.com>
5650 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
5651 (iv_ca_more_deps): Renamed to ...
5652 (iv_ca_compare_deps): ... this.
5653 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
5655 2017-05-11 Bin Cheng <bin.cheng@arm.com>
5657 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
5659 (determine_group_iv_costs): ... here.
5660 (find_inv_vars_cb): Record inv var if it's not recorded before.
5662 2017-05-11 Bin Cheng <bin.cheng@arm.com>
5664 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
5665 (get_shiftadd_cost): Ditto.
5667 2017-05-11 Bin Cheng <bin.cheng@arm.com>
5669 * tree-ssa-address.c: Include header file.
5670 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
5672 (add_to_parts): Refactor.
5673 (addr_to_parts): New parameter. Update use of move_hint_to_base.
5674 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
5677 2017-05-11 Bin Cheng <bin.cheng@arm.com>
5679 PR tree-optimization/53090
5680 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
5682 (extract_cond_operands): Detect condition with IV on both sides
5683 and return COMP_IV_EXPR_2.
5684 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
5685 (rewrite_use_compare): Simplify by removing call to function
5686 extract_cond_operands.
5688 2017-05-11 Bin Cheng <bin.cheng@arm.com>
5690 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
5691 (extract_cond_operands): Detect condition comparing against non-
5692 invariant bound and return appropriate enum value.
5693 (find_interesting_uses_cond): Update use of extract_cond_operands.
5694 Handle its return value accordingly.
5695 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
5697 2017-05-11 Bin Cheng <bin.cheng@arm.com>
5699 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
5700 nonlinear iv_use computation in loop invariant sensitive way.
5702 2017-05-11 Bin Cheng <bin.cheng@arm.com>
5704 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
5705 (find_iv_candidates): Call relate_compare_use_with_all_cands.
5707 2017-05-11 Bin Cheng <bin.cheng@arm.com>
5709 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
5710 (dump_cand): Support iv_cand.inv_exprs.
5711 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
5713 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
5716 2017-05-11 Bin Cheng <bin.cheng@arm.com>
5718 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
5720 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
5721 as local function. Include necessary header files.
5722 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
5724 2017-05-11 Bin Cheng <bin.cheng@arm.com>
5726 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
5728 2017-05-11 Bin Cheng <bin.cheng@arm.com>
5730 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
5731 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
5732 RSHIFT_EXPR and BIT_NOT_EXPR.
5734 2017-05-11 Bin Cheng <bin.cheng@arm.com>
5736 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
5737 (adjust_setup_cost): New parameter supporting round up adjustment.
5738 (struct address_cost_data): Delete.
5739 (force_expr_to_var_cost): Don't bound cost with spill_cost.
5740 (split_address_cost, ptr_difference_cost): Delete.
5741 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
5742 (struct ainc_cost_data): New struct.
5743 (get_address_cost_ainc): New function.
5744 (get_address_cost, get_computation_cost): Reimplement.
5745 (determine_group_iv_cost_address): Record inv_expr for all uses of
5747 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
5748 (iv_ca_has_deps): Reimplemented to ...
5749 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
5751 (iv_ca_extend): Call iv_ca_more_deps.
5753 2017-05-11 Bin Cheng <bin.cheng@arm.com>
5755 * tree-ssa-address.c (struct mem_address): Move to header file.
5756 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
5757 * tree-ssa-address.h (struct mem_address): Move from C file.
5758 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
5760 2017-05-11 Bin Cheng <bin.cheng@arm.com>
5762 * tree-affine.h (aff_combination_type): New interface.
5763 (aff_combination_zero_p): Remove static.
5764 (aff_combination_const_p): New interface.
5765 (aff_combination_singleton_var_p): New interfaces.
5767 2017-05-11 Richard Biener <rguenther@suse.de>
5769 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
5770 Skip unreachable blocks and destinations.
5771 (eliminate): Move stmt removal and fixup ...
5772 (fini_eliminate): ... here. Skip inserted exprs.
5773 (pass_pre::execute): Move fini_pre after fini_eliminate.
5774 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
5775 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
5776 PRE to get rid of dead code that has invalid SSA form and
5777 split critical edges again.
5779 2017-05-11 Bin Cheng <bin.cheng@arm.com>
5781 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
5783 2017-05-11 Richard Biener <rguenther@suse.de>
5785 * passes.c (execute_function_todo): Verify loops if they are
5786 said to be up-to-date.
5787 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
5788 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
5790 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
5793 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
5794 handle calling assemble_external ourself.
5797 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
5798 modes with zero size. Enhance comment.
5800 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5802 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
5803 built-ins for vec_xl and vec_xst with short and char pointer
5806 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
5808 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
5809 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
5810 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
5811 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
5812 (_mm_maskz_min_round_ss): New intrinsics.
5813 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
5814 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
5815 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
5816 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
5817 (__builtin_ia32_minss_mask_round): New builtins.
5818 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
5819 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
5820 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
5822 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
5823 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
5825 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
5828 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
5830 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
5831 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
5832 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
5833 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
5834 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
5835 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
5836 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
5837 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
5838 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
5839 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
5840 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
5841 (__builtin_ia32_mulss_mask_round): New builtins.
5842 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
5843 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
5844 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
5846 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
5847 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
5849 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
5852 2017-05-10 Julia Koval <julia.koval@intel.com>
5854 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
5855 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
5856 (_mm256_setr_m128i): New intrinsics.
5858 2017-05-10 Julia Koval <julia.koval@intel.com>
5860 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
5861 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
5862 (_mm_maskz_rcp14_ss): New intrinsics.
5863 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
5864 (__builtin_ia32_rcp14ss_mask): New builtins.
5865 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
5867 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
5869 PR tree-optimization/51513
5870 * tree-cfg.c (gimple_seq_unreachable_p): New function.
5871 (assert_unreachable_fallthru_edge_p): Use it.
5872 (group_case_labels_stmt): Likewise.
5873 * tree-cfg.h: Prototype it.
5874 * stmt.c: Include cfghooks.h and tree-cfg.h.
5875 (emit_case_dispatch_table) <gap_label>: New local variable.
5876 Use it to fill dispatch table gaps.
5877 Test for default_label before updating probabilities.
5878 (expand_case) <default_label>: Remove unneeded initialization.
5879 Test for unreachable default case statement and remove its edge.
5880 Set default_label accordingly.
5881 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
5883 2017-05-10 Carl Love <cel@us.ibm.com>
5885 * config/rs6000/rs6000-c: Add support for built-in functions
5886 vector signed char vec_neg (vector signed char)
5887 vector signed short int vec_neg (vector short int)
5888 vector signed int vec_neg (vector signed int)
5889 vector signed long long vec_neg (vector signed long long)
5890 vector float vec_neg (vector float)
5891 vector double vec_neg (vector double)
5892 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
5894 * config/rs6000/altivec.h: Add define for vec_neg
5895 * doc/extend.texi: Update the built-in documentation for the
5896 new built-in functions.
5898 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5900 PR tree-optimization/77644
5901 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
5903 2017-05-10 Nathan Sidwell <nathan@acm.org>
5905 * dumpfile.h (TDI_lang_all): New.
5906 (TDF_KIND): New. Renumber others
5907 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
5909 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
5911 (get_dump_file_name): Adjust suffix generation.
5912 (dump_enable_all): Use TDF_KIND.
5913 * doc/invoke.texi (-fdump-lang-all): Document.
5915 * dumpfile.h: Tabify.
5917 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
5920 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
5921 Move member access before delete.
5923 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
5925 * tree-inline.c (expand_call_inline): Split block at stmt
5928 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5931 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
5932 are now unused after splitting mov{sf,sd}_hardfloat.
5933 (f32_lr2): Likewise.
5935 (f32_lm2): Likewise.
5937 (f32_li2): Likewise.
5940 (f32_sr2): Likewise.
5942 (f32_sm2): Likewise.
5944 (f32_si2): Likewise.
5947 (f32_vsx): Likewise.
5949 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
5950 For movsf, order stores so the VSX stores occur before the GPR
5951 store which encourages the register allocator to use a traditional
5952 FPR instead of a GPR. For movsd, order the stores so that the GPR
5953 store comes before the VSX stores to allow the power6 to work.
5954 This is due to the power6 not having a 32-bit integer store
5955 instruction from a FPR.
5956 (movsf_hardfloat): Likewise.
5957 (movsd_hardfloat): Likewise.
5959 2017-05-09 Martin Sebor <msebor@redhat.com>
5961 PR translation/80280
5962 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
5965 PR translation/80280
5966 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
5967 data member added in r247778.
5968 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
5970 2017-05-09 Nathan Sidwell <nathan@acm.org>
5972 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
5974 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
5977 2017-05-09 Marek Polacek <polacek@redhat.com>
5979 * doc/invoke.texi: Fix typo.
5981 2017-05-09 Richard Biener <rguenther@suse.de>
5983 * tree-vrp.c (vrp_val_is_max): Adjust comment.
5984 (vrp_val_is_min): Likewise.
5985 (set_value_range_to_value): Likewise.
5986 (set_value_range_to_nonnegative): Likewise.
5987 (gimple_assign_nonzero_p): Likewise.
5988 (gimple_stmt_nonzero_p): Likewise.
5989 (vrp_int_const_binop): Likewise. Remove unreachable case.
5990 (adjust_range_with_scev): Adjust comments.
5991 (compare_range_with_value): Likewise.
5992 (extract_range_from_phi_node): Likewise.
5993 (test_for_singularity): Likewise.
5995 2017-05-09 Richard Biener <rguenther@suse.de>
5997 * tree-vrp.c (get_single_symbol): Add assert that we don't
5998 get overflowed constants as invariant part.
5999 (compare_values_warnv): Add comment before the TREE_NO_WARNING
6000 checks. Use wi::cmp instead of recursing for integer constants.
6001 (compare_values): Just ignore whether we assumed undefined
6002 overflow instead of failing the compare.
6003 (extract_range_for_var_from_comparison_expr): Add comment before the
6004 TREE_NO_WARNING sets.
6005 (test_for_singularity): Likewise.
6006 (extract_range_from_comparison): Do not disable optimization
6007 when we assumed undefined overflow.
6008 (extract_range_basic): Remove init of unused var.
6010 2017-05-09 Richard Biener <rguenther@suse.de>
6012 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
6013 (extract_range_from_multiplicative_op_1): Adjust.
6014 (extract_range_from_binary_expr_1): Use int_const_binop.
6016 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
6019 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
6020 rs6000_store_data_bypass_p in seven define_bypass directives and
6021 in several comments.
6022 * config/rs6000/rs6000-protos.h: Add prototype for
6023 rs6000_store_data_bypass_p function.
6024 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
6025 function implements slightly different (rs6000-specific) semantics
6026 than store_data_bypass_p, returning false rather than aborting
6027 with assertion error when arguments do not satisfy the
6028 requirements of store data bypass.
6029 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
6030 rs6000_store_data_bypass_p.
6032 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
6034 * config/xtensa/xtensa-protos.h
6035 (xtensa_initial_elimination_offset): New declaration.
6036 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
6037 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
6038 macro definition, add case for FRAME_POINTER_REGNUM when
6039 FRAME_GROWS_DOWNWARD.
6040 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
6041 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
6042 xtensa_initial_elimination_offset.
6044 2017-05-08 Nathan Sidwell <nathan@acm.org>
6046 * doc/invoke.texi: Alphabetize -fdump options.
6048 2017-05-08 Martin Sebor <msebor@redhat.com>
6050 PR translation/80280
6051 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
6053 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
6055 * target.def (compute_frame_layout): New optional target hook.
6056 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
6057 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
6058 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
6060 * reload1.c (verify_initial_elim_offsets): Likewise.
6061 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
6062 (use_simple_return_p): Call arm_compute_frame_layout if needed.
6063 (arm_get_frame_offsets): Split up into this ...
6064 (arm_compute_frame_layout): ... and this function.
6066 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
6068 * config/aarch64/constraints.md (Usa): New constraint.
6069 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
6071 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
6073 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
6074 with_multilib_list after it has been checked.
6076 2017-05-08 Richard Biener <rguenther@suse.de>
6078 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
6079 (bitmap_set_subtract_values): Likewise.
6081 2017-05-08 Richard Biener <rguenther@suse.de>
6083 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
6084 (gimple_assign_nonzero): ... this and remove strict_overflow_p
6086 (gimple_stmt_nonzero_warnv_p): Rename to ...
6087 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
6089 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
6090 (extract_range_basic): Adjust, do not disable propagation on
6091 strict overflow sensitive simplification.
6092 (vrp_visit_cond_stmt): Likewise.
6094 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
6096 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
6097 body size unconditionally.
6099 2017-05-07 Jeff Law <law@redhat.com>
6102 2017-05-06 Jeff Law <law@redhat.com>
6103 PR tree-optimization/78496
6104 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
6107 PR tree-optimization/78496
6108 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
6109 (simplify_stmt_using_ranges): Call it.
6110 (vrp_dom_walker::before_dom_children): Extract equivalences
6111 from an ASSERT_EXPR with an equality comparison against a
6114 2017-05-06 Jeff Law <law@redhat.com>
6116 PR tree-optimization/78496
6117 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
6120 PR tree-optimization/78496
6121 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
6122 (simplify_stmt_using_ranges): Call it.
6123 (vrp_dom_walker::before_dom_children): Extract equivalences
6124 from an ASSERT_EXPR with an equality comparison against a
6127 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
6129 * lra-constraints.c (lra_copy_reg_equiv): New function.
6130 (split_reg): Use it to copy equivalence information from the
6131 original register to the spill register.
6133 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
6135 PR rtl-optimization/75964
6136 * simplify-rtx.c (simplify_const_relational_operation): Remove
6137 invalid handling of comparisons of integer ABS.
6139 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
6141 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
6143 (init_regs): Remove declaration.
6144 (function_arg_advance_32): Initialize error_p as boolean variable.
6146 2017-05-05 Nathan Sidwell <nathan@acm.org>
6148 * store-motion.c (remove_reachable_equiv_notes): Reformat long
6149 lines. Use for (;;).
6151 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6153 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
6154 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
6155 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
6156 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
6157 VF=2 that require versioning.
6159 2017-05-05 David Malcolm <dmalcolm@redhat.com>
6161 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
6164 2017-05-05 David Malcolm <dmalcolm@redhat.com>
6166 * diagnostic.h (diagnostic_override_option_index): Convert from
6167 macro to inline function.
6169 2017-05-05 David Malcolm <dmalcolm@redhat.com>
6171 * diagnostic.c (last_module_changed_p): New function.
6172 (set_last_module): New function.
6173 (diagnostic_report_current_module): Convert macro usage to
6174 the above functions.
6175 * diagnostic.h (diagnostic_context::last_module): Strengthen
6176 from const line_map * to const line_map_ordinary *.
6177 (diagnostic_last_module_changed): Delete macro.
6178 (diagnostic_set_last_module): Delete macro.
6180 2017-05-05 David Malcolm <dmalcolm@redhat.com>
6182 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
6183 with diagnostic_report_diagnostic.
6184 (diagnostic_n_impl_richloc): Likewise.
6185 * diagnostic.h (report_diagnostic): Delete macro.
6186 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
6187 with diagnostic_report_diagnostic.
6188 * substring-locations.c (format_warning_va): Likewise.
6190 2017-05-05 David Malcolm <dmalcolm@redhat.com>
6192 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
6193 save/restor of format_spec. Move option-printing code to...
6194 (print_option_information): ...this new function, and
6195 reimplement by simply printing to the pretty_printer,
6196 rather than appending to the format string.
6198 2017-05-05 David Malcolm <dmalcolm@redhat.com>
6200 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
6201 handling logic into...
6202 (update_effective_level_from_pragmas): ...this new function.
6204 2017-05-04 Andrew Waterman <andrew@sifive.com>
6206 * config/riscv/riscv.opt (mstrict-align): New option.
6207 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
6208 (SLOW_UNALIGNED_ACCESS): Define.
6209 (riscv_slow_unaligned_access): Declare.
6210 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
6212 (riscv_slow_unaligned_access): New variable.
6213 (rocket_tune_info): Set slow_unaligned_access to true.
6214 (optimize_size_tune_info): Set slow_unaligned_access to false.
6215 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
6216 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
6217 (riscv_option_override): Set riscv_slow_unaligned_access.
6218 * doc/invoke.texi: Add -mstrict-align to RISC-V.
6220 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
6222 * config/riscv/riscv.md: Unify indentation.
6224 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6229 * config/rs6000/rs6000.md (u code attribute): Add FIX and
6231 (extendsi<mode>2): Add support for doing sign extension via
6232 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
6233 don't have ISA 3.0 instructions.
6234 (extendsi<mode>2 splitter): Likewise.
6235 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
6236 generate the normal insns since SImode can now go in vector
6237 registers. Disallow the special UNSPECs needed for previous
6238 machines to hide SImode being used. Add new insns
6239 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
6240 (fix_trunc<mode>si2_stfiwx): Likewise.
6241 (fix_trunc<mode>si2_internal): Likewise.
6242 (fixuns_trunc<mode>si2): Likewise.
6243 (fixuns_trunc<mode>si2_stfiwx): Likewise.
6244 (fctiw<u>z_<mode>_smallint): Likewise.
6245 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
6246 of floating point to 32-bit integer from doing a direct move to
6247 the GPR registers to do a store.
6248 (fctiwz_<mode>): Break long line.
6250 2017-05-05 Bin Cheng <bin.cheng@arm.com>
6252 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
6253 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
6254 (addr_list, addr_offset_valid_p): New.
6255 (split_address_groups): Check offset validity with above function.
6256 (gt-tree-ssa-loop-ivopts.h): Include header file.
6258 2017-05-05 Nathan Sidwell <nathan@acm.org>
6260 * config.gcc (arm*-*-*): Add missing 'fi'.
6262 2017-05-05 Steve Ellcey <sellcey@cavium.com>
6264 * doc/invoke.texi (-fopt-info): Explicitly say order of options
6265 included in -fopt-info does not matter.
6266 * doc/optinfo.texi (-fopt-info): Fix description of default
6267 behavour. Explicitly say order of options included in -fopt-info
6270 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
6272 * config.gcc: Allow combinations of aprofile and rmprofile values for
6273 --with-multilib-list.
6274 * config/arm/t-multilib: New file.
6275 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
6276 variables. Remove setting of ISA and floating-point ABI in
6277 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
6278 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
6279 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
6280 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
6282 * config/arm/t-rmprofile: Likewise except for the matches changes.
6283 * doc/install.texi (--with-multilib-list): Document the combination of
6284 aprofile and rmprofile values and warn about pitfalls in doing that.
6286 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
6288 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
6289 (movdi_aarch64): Likewise.
6291 2017-05-05 Jakub Jelinek <jakub@redhat.com>
6293 PR tree-optimization/80632
6294 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
6296 (build_arrays): Initialize it for virtual phis.
6297 (fix_phi_nodes): Use it for virtual phis.
6299 PR tree-optimization/80558
6300 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
6301 [x, y] op z into [x op, y op z] for op & or | if conditions
6304 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
6305 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
6308 * config/arm/arm.md (use_literal_pool): Remove.
6309 (64-bit immediate split): No longer takes cost into consideration
6310 if arm_disable_literal_pool is enabled.
6311 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
6312 used when arm_disable_literal_pool is enabled.
6313 (arm_max_const_double_inline_cost): Remove use of
6314 arm_disable_literal_pool.
6315 (push_minipool_fix): Add assert.
6316 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
6317 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
6318 (no_literal_pool_sf_immediate): New.
6320 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6322 PR tree-optimization/80613
6323 * tree-ssa-dce.c (propagate_necessity): Remove cases for
6324 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
6326 2017-05-05 Richard Biener <rguenther@suse.de>
6328 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
6330 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
6332 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
6333 of this flag from insn conditions due to removal from r247495.
6335 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
6337 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
6339 (arm_early_store_addr_dep_ptr): Likewise.
6340 * config/arm/aarch-common-protos.h
6341 (arm_early_load_addr_dep_ptr): Add prototype.
6342 (arm_early_store_addr_dep_ptr): Likewise.
6343 * config/arm/cortex-a53.md: Add new bypasses.
6345 2017-05-05 Jakub Jelinek <jakub@redhat.com>
6347 * tree.c (next_type_uid): Change type to unsigned.
6348 (type_hash_canon): Decrement back next_type_uid if
6349 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
6350 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
6353 2017-05-04 Martin Sebor <msebor@redhat.com>
6355 * builtins.c: Fix a trivial typo in a comment.
6358 * builtins.c (check_sizes): Adjust to handle reading past the end.
6359 Avoid printing excessive upper bound of ranges. Use %E to print
6360 tree nodes instead of converting them to %wu.
6361 (expand_builtin_memchr): New function.
6362 (compute_dest_size): Rename...
6363 (compute_objsize): ...to this.
6364 (expand_builtin_memcpy): Adjust.
6365 (expand_builtin_mempcpy): Adjust.
6366 (expand_builtin_strcat): Adjust.
6367 (expand_builtin_strcpy): Adjust.
6368 (check_strncat_sizes): Adjust.
6369 (expand_builtin_strncat): Adjust.
6370 (expand_builtin_strncpy): Adjust and simplify.
6371 (expand_builtin_memset): Adjust.
6372 (expand_builtin_bzero): Adjust.
6373 (expand_builtin_memcmp): Adjust.
6374 (expand_builtin): Handle memcmp.
6375 (maybe_emit_chk_warning): Check strncat just once.
6377 2017-05-04 Martin Sebor <msebor@redhat.com>
6379 PR preprocessor/79214
6382 * builtins.c (check_sizes): Add inlining context and issue
6383 warnings even when -Wno-system-headers is set.
6384 (check_strncat_sizes): Same.
6385 (expand_builtin_strncat): Same.
6386 (expand_builtin_memmove): New function.
6387 (expand_builtin_stpncpy): Same.
6388 (expand_builtin): Handle memmove and stpncpy.
6390 2017-05-04 Bin Cheng <bin.cheng@arm.com>
6392 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
6393 which is not used any more.
6395 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
6397 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
6399 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
6401 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
6402 (cortexa53_tunings): Likewise.
6403 (cortexa57_tunings): Likewise.
6404 (cortexa72_tunings): Likewise.
6405 (cortexa73_tunings): Likewise.
6407 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
6409 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
6410 Set loop alignment to 8.
6412 2017-05-04 Martin Sebor <msebor@redhat.com>
6414 PR translation/80280
6415 * builtins.c (expand_builtin_object_size): Add missing quoting to
6416 %D and like directives.
6417 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
6418 (hsa_type_for_tree_type): Same.
6419 (verify_function_arguments): Same.
6420 * symtab.c (symbol_table::change_decl_assembler_name): Same.
6421 * varasm.c (get_section): Same.
6424 2017-05-04 Martin Sebor <msebor@redhat.com>
6426 PR translation/80280
6427 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
6429 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
6431 * config/aarch64/aarch64.c (generic_addrcost_table):
6432 Change HI/TI mode setting.
6434 2017-05-04 Martin Jambor <mjambor@suse.cz>
6436 PR tree-optimization/80622
6437 * tree-sra.c (comes_initialized_p): New function.
6438 (build_accesses_from_assign): Only set write lazily when
6439 comes_initialized_p is false.
6440 (analyze_access_subtree): Use comes_initialized_p.
6441 (propagate_subaccesses_across_link): Assert !comes_initialized_p
6442 instead of testing for PARM_DECL.
6444 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6446 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
6447 constraint on operand 0 to allow more general addressing modes.
6448 Adjust output template.
6449 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
6451 * config/aarch64/aarch64-protos.h
6452 (aarch64_address_valid_for_prefetch_p): Declare prototype.
6453 * config/aarch64/constraints.md (Dp): New address constraint.
6454 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
6457 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
6459 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
6460 update use of estimate_ipcp_clone_size_and_time.
6461 (estimate_local_effects): Update use of
6462 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
6463 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
6464 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
6465 Return nonspecialized time.
6467 2017-05-04 Richard Biener <rguenther@suse.de>
6469 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
6470 for the last VUSE which def dominates the PHI. Directly call
6472 (get_continuation_for_phi_1): Remove.
6474 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
6476 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
6477 to explain the use of truncating division. Cap the number of
6478 iterations to the maximum given by nb_iterations_upper_bound,
6481 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
6483 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
6484 * configure: Regenerate.
6485 * config.in: Regenerate.
6486 * config/i386/driver-mingw32.c: new file.
6487 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
6488 * config.host: Link driver-mingw32.o on MinGW host.
6489 * doc/install.texi: Document new --enable-mingw-wildcard configure
6492 2017-05-04 Marek Polacek <polacek@redhat.com>
6494 PR tree-optimization/80612
6495 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
6497 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
6498 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
6500 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
6501 (movt splitter): Likewise.
6502 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
6503 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
6504 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
6505 block for Thumb-1 with MOVT.
6506 (thumb2_legitimate_address_p): Move code block ...
6507 (can_avoid_literal_pool_for_label_p): ... into this new function.
6508 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
6510 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
6511 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
6512 "M-profile targets with the MOVT instruction".
6514 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
6516 * gcc/config/arm/arm-builtins.c (arm_init_builtins): Rename
6517 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
6518 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
6520 2017-05-04 Martin Liska <mliska@suse.cz>
6522 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
6525 2017-05-04 Richard Biener <rguenther@suse.de>
6527 * tree.c (array_at_struct_end_p): Handle arrays at struct
6528 end with flexarrays more conservatively. Refactor and treat
6529 arrays of arrays or aggregates more strict. Fix
6530 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
6531 * tree.c (array_at_struct_end_p): Adjust prototype.
6532 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
6533 * gimple-fold.c (get_range_strlen): Likewise.
6534 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
6536 2017-05-04 Richard Biener <rguenther@suse.de>
6538 PR tree-optimization/31130
6539 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
6541 (supports_overflow_infinity): Likewise.
6542 (is_negative_overflow_infinity): Likewise.
6543 (is_positive_overflow_infinity): Likewise.
6544 (is_overflow_infinity): Likewise.
6545 (stmt_overflow_infinity): Likewise.
6546 (overflow_infinity_range_p): Likewise.
6547 (usable_range_p): Remove as always returning true.
6548 (make_overflow_infinity): Remove.
6549 (negative_overflow_infinity): Likewise.
6550 (positive_overflow_infinity): Likewise.
6551 (avoid_overflow_infinity): Likewise.
6552 (set_value_range): Adjust accordingly.
6553 (set_value_range_to_nonnegative): Likewise, remove now unused
6554 overflow_infinity arg.
6555 (vrp_operand_equal_p): Adjust.
6556 (update_value_range): Likewise.
6557 (range_int_cst_singleton_p): Likewise.
6558 (operand_less_p): Likewise.
6559 (compare_values_warnv): Likewise.
6560 (extract_range_for_var_from_comparison_expr): Likewise.
6561 (vrp_int_const_binop): Likewise.
6562 (zero_nonzero_bits_from_vr): Likewise.
6563 (extract_range_from_multiplicative_op_1): Likewise.
6564 (extract_range_from_binary_expr_1): Likewise.
6565 (extract_range_from_unary_expr): Likewise.
6566 (extract_range_from_comparison): Likewise.
6567 (extract_range_basic): Likewise.
6568 (adjust_range_with_scev): Likewise.
6569 (compare_ranges): Likewise.
6570 (compare_range_with_value): Likewise.
6571 (dump_value_range): Likewise.
6572 (test_for_singularity): Likewise, remove strict_overflow_p parameter
6574 (simplify_cond_using_ranges): Adjust.
6576 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
6578 * brig-builtins.def: Added a builtin for class_f64.
6579 * builtin-types.def: Added a builtin type needed by class_f64.
6581 2017-05-03 Jason Merrill <jason@redhat.com>
6583 * timevar.def: Add TV_CONSTEXPR.
6585 2017-05-03 David Malcolm <dmalcolm@redhat.com>
6587 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
6589 2017-05-03 Martin Jambor <mjambor@suse.cz>
6591 * ipa-prop.c (ipa_update_after_lto_read): Removed.
6592 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
6593 * ipa-cp.c (ipcp_propagate_stage): Do not call
6594 ipa_update_after_lto_read.
6595 * ipa-inline.c (ipa_inline): Likewise.
6597 2017-05-03 Martin Jambor <mjambor@suse.cz>
6599 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
6600 tag. Added a default constructor and a destructor.
6601 (ipa_edge_args_sum_t): New class;
6602 (ipa_edge_args_sum): Declare.
6603 (ipa_edge_args_vector): Remove declaration.
6604 (IPA_EDGE_REF): Use ipa_edge_args_sum.
6605 (ipa_free_edge_args_substructures): Remove declaration.
6606 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
6607 (ipa_edge_args_info_available_for_edge_p): Likewise.
6608 * ipa-prop.c (ipa_edge_args_vector): Removed.
6609 (edge_removal_hook_holder): Likewise.
6610 (edge_duplication_hook_holder): Likewise.
6611 (ipa_edge_args_sum): New variable.
6612 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
6613 ipa_edge_args_vector.
6614 (ipa_free_edge_args_substructures): Likewise.
6615 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
6616 ipa_edge_args_vector.
6617 (ipa_edge_removal_hook): Turned into method
6618 ipa_edge_args_sum_t::remove.
6619 (ipa_edge_duplication_hook): Turned into method
6620 ipa_edge_args_sum_t::duplicate.
6621 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
6622 registering edge hooks.
6623 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
6624 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
6625 ipa_edge_args_sum instead of ipa_edge_args_vector.
6626 * ipa-profile.c (ipa_profile): Likewise.
6628 2017-05-03 Martin Jambor <mjambor@suse.cz>
6630 * symbol-summary.h (function_summary): New method exists.
6631 (function_summary::symtab_removal): Deallocate through release.
6632 (call_summary): New class.
6633 (gt_ggc_mx): New overload.
6634 (gt_pch_nx): Likewise.
6635 (gt_pch_nx): Likewise.
6637 2017-05-03 Jeff Law <law@redhat.com>
6639 PR tree-optimization/78496
6640 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
6641 from simplify_cond_using_ranges. Split off code to walk
6642 backwards through casts into ...
6643 (simplify_cond_using_ranges_2): New function.
6644 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
6645 (execute_vrp): After identifying jump threads, call
6646 simplify_cond_using_ranges_2.
6648 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
6651 * ipa-inline.h (inline_summary): Add ctor.
6652 (create_ggc): Do not use ggc_cleared_alloc.
6654 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
6655 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6657 * gcc.c (handle_braces): Support escaping in switch matching
6659 * doc/invoke.texi (Spec Files): Document it.
6660 Remove superfluous @code markup in items.
6662 2017-05-03 David Malcolm <dmalcolm@redhat.com>
6664 * diagnostic-show-locus.c (struct column_range): New struct.
6665 (get_affected_columns): New function.
6666 (get_printed_columns): New function.
6667 (struct correction): New struct.
6668 (correction::ensure_capacity): New function.
6669 (correction::ensure_terminated): New function.
6670 (struct line_corrections): New struct.
6671 (line_corrections::~line_corrections): New dtor.
6672 (line_corrections::add_hint): New function.
6673 (layout::print_trailing_fixits): Reimplement in terms of the new
6675 (selftest::test_overlapped_fixit_printing): New function.
6676 (selftest::diagnostic_show_locus_c_tests): Call it.
6678 2017-05-03 Nathan Sidwell <nathan@acm.org>
6680 Canonicalize canonical type hashing
6681 * tree.h (type_hash_canon_hash): Declare.
6682 * tree.c (type_hash_list, attribute_hash_list): Move into
6683 type_hash_canon_hash.
6684 (build_type_attribute_qual_variant): Break out hash code calc into
6685 type_hash_canon_hash.
6686 (type_hash_canon_hash): New. Generic type hash computation.
6687 (build_range_type_1, build_array_type_1, build_function_type,
6688 build_method_type_directly, build_offset_type, build_complex_type,
6689 make_vector_type): Call it.
6691 2017-05-03 Richard Biener <rguenther@suse.de>
6693 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
6694 When all DRs have unknown misaligned do not always peel
6695 when there is a store but apply the same costing model as if
6696 there were only loads.
6698 2017-05-03 Richard Biener <rguenther@suse.de>
6701 PR tree-optimization/80492
6702 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
6703 compare_base_decls returning dont-know properly.
6705 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
6707 * config/arm/iterators.md (CCSI): New mode iterator.
6708 (arch): New mode attribute.
6709 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
6710 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
6711 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
6712 code iterator for success result mode.
6713 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
6714 the corresponding new insn generators.
6716 2017-05-03 Bin Cheng <bin.cheng@arm.com>
6719 2017-05-02 Bin Cheng <bin.cheng@arm.com>
6720 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
6722 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
6724 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
6725 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
6726 (DDR_A): Wrap DDR argument in brackets.
6727 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
6728 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
6729 (DDR_REVERSED_P): Likewise.
6731 2017-05-03 Jakub Jelinek <jakub@redhat.com>
6733 PR tree-optimization/79472
6734 * tree-switch-conversion.c (struct switch_conv_info): Add
6735 contiguous_range and default_case_nonstandard fields.
6736 (collect_switch_conv_info): Compute contiguous_range and
6737 default_case_nonstandard fields, don't clear final_bb if
6738 contiguous_range and only the default case doesn't have the required
6740 (check_all_empty_except_final): Set default_case_nonstandard instead
6741 of failing if contiguous_range and the default case doesn't have empty
6743 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
6744 and only the default case doesn't have the required constants. Skip
6746 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
6747 if default_case_nonstandard.
6748 (build_constructors): Build constant 1 just once. Assert that default
6749 values aren't inserted in between cases if contiguous_range. Skip
6751 (build_arrays): Skip virtual phis.
6752 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
6753 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
6754 Handle virtual phis.
6755 (gen_inbound_check): Handle default_case_nonstandard case.
6756 (process_switch): Adjust check_final_bb caller. Call
6757 gather_default_values with the first non-default case instead of
6758 default case if default_case_nonstandard.
6760 2017-05-02 Nathan Sidwell <nathan@acm.org>
6762 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
6763 check. Fix formatting.
6765 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
6767 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
6768 errors when comparing specialized and unspecialized times.
6770 2017-05-02 David Malcolm <dmalcolm@redhat.com>
6772 * diagnostic-show-locus.c
6773 (layout::should_print_annotation_line_p): Make private.
6774 (layout::print_annotation_line): Make private.
6775 (layout::annotation_line_showed_range_p): Make private.
6776 (layout::show_ruler): Make private.
6777 (layout::print_source_line): Make private. Pass in line and
6778 line_width, rather than calling location_get_source_line. Drop
6780 (layout::print_leading_fixits): New method.
6781 (layout::print_any_fixits): Rename to...
6782 (layout::print_trailing_fixits): ...this, and make private.
6783 Don't print newline fixits.
6784 (diagnostic_show_locus): Move logic for printing one row into...
6785 (layout::print_line): ...this new function. Move the
6786 location_get_source_line call and error-handling from
6787 print_source_line to here. Call print_leading_fixits, and rename
6788 print_any_fixits to print_trailing_fixits.
6789 (selftest::test_fixit_insert_containing_newline): Update now that
6790 newlines are partially supported.
6791 (selftest::test_fixit_insert_containing_newline_2): New test.
6792 (selftest::test_fixit_replace_containing_newline): Update comments.
6793 (selftest::diagnostic_show_locus_c_tests): Call the new test.
6794 * edit-context.c (class added_line): New class.
6795 (class edited_line): Describe newline handling in comment.
6796 (edited_line::actually_edited_p): New method.
6797 (edited_line::print_content): Delete redundant decl.
6798 (edited_line::m_predecessors): New field.
6799 (edited_file::print_content): Call edited_line::print_content.
6800 (edited_file::print_diff): Update to support newlines.
6801 (edited_file::print_diff_hunk): Likewise.
6802 (edited_file::print_run_of_changed_lines): New function.
6803 (edited_file::print_diff_line): Convert to...
6804 (print_diff_line): ...this.
6805 (edited_file::get_effective_line_count): New function.
6806 (edited_line::edited_line): Initialize new field m_predecessors.
6807 (edited_line::~edited_line): Clean up m_predecessors.
6808 (edited_line::apply_fixit): Handle newlines.
6809 (edited_line::get_effective_line_count): New function.
6810 (edited_line::print_content): New function.
6811 (edited_line::print_diff_lines): New function.
6812 (selftest::test_applying_fixits_insert_containing_newline): New
6814 (selftest::test_applying_fixits_replace_containing_newline): New
6816 (selftest::insert_line): New function.
6817 (selftest::test_applying_fixits_multiple_lines): Add example of
6819 (selftest::edit_context_c_tests): Call the new tests.
6821 2017-05-02 Bin Cheng <bin.cheng@arm.com>
6823 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
6824 parameter cand. Update dump information.
6825 (get_computation_cost): Update uses.
6827 2017-05-02 Bin Cheng <bin.cheng@arm.com>
6829 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
6830 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
6831 (get_computation_at, rewrite_use_address): Update use of
6832 get_computation_aff.
6834 2017-05-02 Bin Cheng <bin.cheng@arm.com>
6836 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
6837 (get_computation): Delete.
6838 (get_computation_cost): Implement like get_computation_cost_at.
6839 Use get_computation_at.
6840 (get_computation_cost_at): Delete.
6841 (rewrite_use_nonlinear_expr): Use get_computation_at.
6842 (rewrite_use_compare, remove_unused_ivs): Ditto.
6844 2017-05-02 Bin Cheng <bin.cheng@arm.com>
6846 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
6848 2017-05-02 Bin Cheng <bin.cheng@arm.com>
6850 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
6851 (ivopts_global_cost_for_size): Rename parameter and update uses.
6852 (iv_ca_recount_cost): Update uses.
6853 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
6854 candidates seperately in n_invs and n_cands.
6855 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
6857 2017-05-02 Bin Cheng <bin.cheng@arm.com>
6859 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
6860 (find_inv_vars_cb): New.
6861 (find_depends): Renamed to ...
6862 (find_inv_vars): ... this.
6863 (add_candidate_1, force_var_cost): Call find_inv_vars.
6864 (split_address_cost, determine_group_iv_cost_cond): Ditto.
6866 2017-05-02 Bin Cheng <bin.cheng@arm.com>
6868 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
6869 inv_vars. Add inv_exprs.
6870 (struct iv_cand): Rename depends_on to inv_vars.
6871 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
6872 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
6873 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
6874 (dump_cand): Dump inv_vars.
6875 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
6876 (record_invariant, find_depends, add_candidate_1): Ditto.
6877 (set_group_iv_cost, force_var_cost): Ditto.
6878 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
6879 (get_computation_cost_at, get_computation_cost): Ditto.
6880 (determine_group_iv_cost_generic): Ditto.
6881 (determine_group_iv_cost_address): Ditto.
6882 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
6883 (determine_group_iv_costs): Ditto.
6884 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
6885 (iv_ca_set_remove_invariants): Renamed to ...
6886 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
6887 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
6888 (iv_ca_set_add_invariants): Renamed to ...
6889 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
6890 (iv_ca_set_cp): Use iv_ca_set_add_invs.
6891 (iv_ca_has_deps): Support inv_vars and inv_exprs.
6892 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
6893 (create_new_ivs): Remove useless dump.
6895 2017-05-02 Bin Cheng <bin.cheng@arm.com>
6897 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
6899 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
6900 (iv_ca_set_no_cp, create_new_iv): Ditto.
6902 2017-05-02 Bin Cheng <bin.cheng@arm.com>
6904 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
6906 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
6908 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
6909 function tree_check2.
6911 2017-05-02 Martin Liska <mliska@suse.cz>
6913 * doc/gcov.texi: Add missing preposition.
6914 * gcov.c (function_info::function_info): Properly fill up
6915 all member variables.
6917 2017-05-02 Tamar Christina <tamar.christina@arm.com>
6919 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
6921 2017-05-02 Tamar Christina <tamar.christina@arm.com>
6923 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
6925 2017-05-02 Martin Liska <mliska@suse.cz>
6928 * lto-streamer-in.c (lto_read_tree_1): Remove
6930 * lto-streamer.c (struct tree_hash_entry): Likewise.
6931 (struct tree_entry_hasher): Likewise.
6932 (tree_entry_hasher::hash): Likewise.
6933 (tree_entry_hasher::equal): Likewise.
6934 (lto_streamer_init): Likewise.
6935 (lto_orig_address_map): Likewise.
6936 (lto_orig_address_get): Likewise.
6937 (lto_orig_address_remove): Likewise.
6938 * lto-streamer.h: Likewise.
6939 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
6940 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
6942 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
6944 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
6945 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
6946 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
6947 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
6948 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
6949 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
6950 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
6951 (mm_maskz_sub_ss): New intrinsics.
6952 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
6953 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
6954 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
6955 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
6956 (__builtin_ia32_subss_mask_round): New builtins.
6957 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
6958 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
6959 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
6961 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
6962 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
6964 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
6967 2017-05-02 Martin Jambor <mjambor@suse.cz>
6969 PR tree-optimization/78687
6970 * tree-sra.c (access): New field parent.
6971 (process_subtree_disqualification): New function.
6972 (disqualify_candidate): Call it.
6973 (build_accesses_from_assign): Reset write flag if creating an
6975 (build_access_subtree): Fill in parent field and also prpagate
6976 down grp_write flag.
6977 (create_artificial_child_access): New parameter set_grp_write, set
6978 grp_write to its value.
6979 (propagate_subaccesses_across_link): Also propagate grp_write flag
6981 (propagate_all_subaccesses): Push the closest parent back to work
6982 queue if add_access_to_work_queue returned true.
6984 2017-05-02 Richard Biener <rguenther@suse.de>
6986 * common.opt (fstrict-overflow): Alias negative to fwrapv.
6987 * doc/invoke.texi (fstrict-overflow): Remove all traces of
6988 -fstrict-overflow documentation.
6989 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
6990 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
6991 flag_strict_overflow.
6992 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
6993 * lto-opts.c (lto_write_options): Do not stream it.
6994 * lto-wrapper.c (merge_and_complain): Do not handle it.
6995 * opts.c (default_options_table): Do not set -fstrict-overflow.
6996 (finish_options): Likewise do not clear it when sanitizing.
6997 * simplify-rtx.c (simplify_const_relational_operation): Do not
6998 test flag_strict_overflow.
7000 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
7002 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
7003 using enabled attribute.
7004 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
7005 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
7006 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
7007 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
7008 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
7009 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
7010 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
7011 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
7012 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
7013 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
7015 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
7017 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
7019 2017-05-02 Richard Biener <rguenther@suse.de>
7021 PR tree-optimization/80591
7023 2017-04-10 Richard Biener <rguenther@suse.de>
7025 * tree-ssa-structalias.c (find_func_aliases): Properly handle
7028 2017-05-02 Richard Biener <rguenther@suse.de>
7030 PR tree-optimization/80549
7031 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
7032 (cleanup_tree_cfg_noloop): Create forwarders to known loop
7033 headers if they do not have a preheader.
7035 2017-05-02 Martin Liska <mliska@suse.cz>
7038 * common.opt: Fix typo.
7039 * doc/invoke.texi: Likewise.
7041 2017-05-01 Jan Beulich <jbeulich@suse.com>
7043 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
7044 swapping, add (x,x,m,x,n) alternative.
7046 2017-05-01 Nathan Sidwell <nathan@acm.org>
7048 * calls.c (combine_pending_stack_adjustment_and_call): Remove
7049 unnecessary unadjusted_alignment check.
7051 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
7054 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
7056 * gimplify.c (gimplify_cilk_detach): New function.
7057 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
7058 * tree-core.h: Document EXPR_CILK_SPAWN.
7059 * tree.h (EXPR_CILK_SPAWN): Define.
7061 2017-05-01 David Malcolm <dmalcolm@redhat.com>
7063 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
7064 to use new fixit_hint representation, using the "replace" logic.
7065 (get_line_span_for_fixit_hint): Likewise.
7066 (layout::print_any_fixits): Likewise.
7067 (selftest::test_one_liner_many_fixits): Rename to...
7068 (selftest::test_one_liner_many_fixits_1): ...this, and update
7069 comment and expected output to reflect that the multiple fix-it
7070 hints are now consolidated into one insertion.
7071 (selftest::test_one_liner_many_fixits_2): New test.
7072 (selftest::test_diagnostic_show_locus_one_liner): Update for
7074 (selftest::test_fixit_consolidation): Update for fix-it API
7076 * diagnostic.c (print_parseable_fixits): Likewise.
7077 * edit-context.c (edited_line::m_line_events): Convert from
7078 auto_vec <line_event *> to auto_vec <line_event>.
7079 (class line_event): Convert from abstract base class to a concrete
7080 class, taking over the role of replace_event.
7081 (class insert_event): Delete.
7082 (class replace_event): Rename to class line_event. Convert to
7084 (edit_context::add_fixits): Reimplement.
7085 (edit_context::apply_insert): Delete.
7086 (edit_context::apply_replace): Rename to...
7087 (edit_context::apply_fixit): ...this. Convert to half-open range.
7088 (edited_file::apply_insert): Delete.
7089 (edited_file::apply_replace): Rename to...
7090 (edited_file::apply_fixit): ...this.
7091 (edited_line::~edited_line): Drop deletion of events.
7092 (edited_line::apply_insert): Delete.
7093 (edited_line::apply_replace): Rename to...
7094 (edited_line::apply_fixit): ...this. Convert to half-open range.
7095 Update for change to type of m_line_events.
7096 * edit-context.h (edit_context::apply_insert): Delete.
7097 (edit_context::apply_replace): Rename to...
7098 (edit_context::apply_fixit): ...this.
7100 2017-05-01 Martin Sebor <msebor@redhat.com>
7102 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
7105 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
7108 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
7109 __get_cpuid_max returns 0.
7110 (__get_cpuid_count): Ditto.
7112 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
7114 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
7115 replacement expression is another instance of one of its arguments.
7117 2017-05-01 Jakub Jelinek <jakub@redhat.com>
7120 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
7121 check for stack push/pop autoinc.
7122 * config/i386/i386.c (ix86_agi_dependent): Return false
7123 if the only reason why modified_in_p returned true is that
7124 addr is SP based and set_insn is a push or pop.
7126 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
7128 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
7131 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
7134 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
7135 (account_size_time): Use two predicates - exec_pred and
7137 (evaluate_conditions_for_known_args): Compute both clause and
7139 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
7140 (inline_summary_t::duplicate): Update.
7141 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
7143 (compute_inline_parameters): Likewise.
7144 (estimate_edge_size_and_time): Update caluclation of time.
7145 (estimate_node_size_and_time): Compute both time and nonspecialized
7147 (estimate_ipcp_clone_size_and_time): Update.
7148 (inline_merge_summary): Update.
7149 (do_estimate_edge_time): Update.
7150 (do_estimate_edge_size): Update.
7151 (do_estimate_edge_hints): Update.
7152 (inline_read_section, inline_write_summary): Stream both new predicates.
7153 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
7155 (compute_inlined_call_time): Cleanup.
7156 (big_speedup_p): Update.
7157 (edge_badness): Update.
7158 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
7159 (size_time_entry): Replace predicate by exec_predicate and
7161 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
7162 (estimate_edge_time): Return also nonspec_time.
7163 (reset_edge_growth_cache): Update.
7165 2017-04-29 Jakub Jelinek <jakub@redhat.com>
7167 PR rtl-optimization/80491
7168 * ifcvt.c (noce_process_if_block): When looking for x setter
7169 with missing else_bb, don't check only the insn right before
7170 cond_earliest, but look for the last insn that x is modified in
7173 PR rtl-optimization/80491
7174 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
7176 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
7178 PR tree-optimization/80487
7179 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
7181 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7183 PR tree-optimization/79697
7184 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
7185 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
7186 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
7188 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
7189 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
7191 2017-04-28 Martin Sebor <msebor@redhat.com>
7193 PR tree-optimization/80523
7194 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
7195 (init_target_to_host_charmap, target_to_host, target_strtol10): New
7197 (maybe_warn, format_directive, parse_directive): Use new functions.
7198 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
7200 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
7202 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
7204 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
7206 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
7207 target_header_dir): Set correctly.
7208 * configure: Regenerated.
7209 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
7210 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
7211 instead of SYSTEM_HEADER_DIR.
7213 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
7215 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
7216 (estimate_local_effects): Likewise.
7217 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
7218 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
7219 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
7220 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
7221 do_estimate_edge_time, estimate_edge_time): Likewise.
7222 * ipa-inline-analysis.c (estimate_node_size_and_time,
7223 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
7224 (estimate_time_after_inlining): Remove.
7226 2017-04-28 Martin Liska <mliska@suse.cz>
7228 * doc/gcov.texi: Enhance documentation of gcov.
7230 2017-04-28 Martin Liska <mliska@suse.cz>
7232 * doc/gcov.texi: Sort options in alphabetic order.
7233 * doc/gcov-dump.texi: Likewise.
7234 * doc/gcov-tool.texi: Likewise.
7235 * gcov.c (print_usage): Likewise.
7236 * gcov-dump.c (print_usage): Likewise.
7237 * gcov-tool.c (print_merge_usage_message): Likewise.
7238 (print_rewrite_usage_message): Likewise.
7239 (print_overlap_usage_message): Likewise.
7241 2017-04-28 Martin Liska <mliska@suse.cz>
7243 PR gcov-profile/53915
7244 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
7246 2017-04-28 Martin Liska <mliska@suse.cz>
7248 PR gcov-profile/79891
7249 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
7250 is marked by compiler as living on a line.
7251 (get_cycles_count): Remove usage of the union.
7252 (output_intermediate_file): Likewise.
7253 (find_source): Fix GNU coding style.
7254 (accumulate_line_counts): Remove old non-all block mode.
7255 (output_lines): Remove usage of the union.
7256 * profile.c (output_location): Include all BBs, even if
7257 belonging to a same line (and file) as a previous BB.
7259 2017-04-28 Martin Liska <mliska@suse.cz>
7261 * gcov.c (process_args): Handle new argument 'w'.
7262 (read_graph_file): Assign ID to BBs.
7263 (output_branch_count): Display BB # if verbose flag is set.
7264 (output_lines): Likewise for arcs.
7265 (print_usage): Add '--verbose' option help.
7266 * doc/gcov.texi: Document --verbose (-w) option.
7268 2017-04-28 Martin Liska <mliska@suse.cz>
7270 * gcov.c (struct block_location_info): New struct.
7271 (process_file): Fill up the new structure.
7272 (read_graph_file): Replace usage of encoding by the newly added
7274 (add_line_counts): Likewise.
7275 (accumulate_line_counts): Remove usage of the union.
7276 (function_info::function_info): New function.
7277 (function_info::~function_info): Likewise.
7278 (process_file): Call delete instead of release_function.
7279 (release_function): Release the function.
7280 (release_structures): Call delete instead of release_function.
7281 (solve_flow_graph): Replace usage of num_blocks.
7282 (find_exception_blocks): Likewise.
7283 (output_lines): Fix GNU coding style.
7285 2017-04-28 Martin Liska <mliska@suse.cz>
7288 * coverage.c (coverage_remove_note_file): New function.
7289 * coverage.h: Declare the function.
7290 * toplev.c (finalize): Clean if an error has been seen.
7292 2017-04-28 Martin Liska <mliska@suse.cz>
7294 PR gcov-profile/80031
7295 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
7296 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
7297 * gcov.c (read_graph_file): Read just number of blocks.
7298 * profile.c (branch_prob): Do not stream 0 flags per a basic
7301 2017-04-28 Martin Liska <mliska@suse.cz>
7303 * gcov-dump.c (tag_*): Add new argument to declarations.
7304 (dump_gcov_file): Likewise.
7305 (tag_blocks): Add and use new argument depth.
7306 (tag_arcs): Likewise.
7307 (tag_lines): Likewise.
7308 (tag_counters): Likewise.
7309 (tag_summary): Likewise.
7310 (dump_working_sets): Use depth to do a proper indentation.
7312 2017-04-28 Jakub Jelinek <jakub@redhat.com>
7315 * cgraph.h (symtab_node::debug_symtab): No longer inline.
7316 * symtab.c (symtab_node::debug_symtab): Move definition here.
7318 2017-04-28 Richard Biener <rguenther@suse.de>
7320 * lto-streamer.h (LTO_major_version): Bump to 7.
7322 2017-04-28 Richard Biener <rguenther@suse.de>
7324 * tree-vrp.c (assert_info): New struct.
7325 (add_assert_info): New helper.
7326 (register_edge_assert_for_2): Refactor to add asserts to a vector
7328 (register_edge_assert_for_1): Likewise.
7329 (register_edge_assert_for): Likewise.
7330 (finish_register_edge_assert_for): New helper actually registering
7331 asserts where live on edge.
7332 (find_conditional_asserts): Adjust.
7333 (find_switch_asserts): Likewise.
7334 (evrp_dom_walker::try_find_new_range): Generalize.
7335 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
7337 2017-04-27 Marek Polacek <polacek@redhat.com>
7340 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
7341 arg10 and arg11 to itype.
7343 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
7345 * doc/extend.texi (Object Size Checking): Improve grammar.
7347 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
7350 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
7351 that the logic for permitting reciprocal estimates matches that
7354 2017-04-27 Jakub Jelinek <jakub@redhat.com>
7357 * tree.c (type_cache_hasher::equal): Only compare
7358 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
7359 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
7360 non-aggregate element types.
7361 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
7362 about the flag on ARRAY_TYPEs in the comment, formatting fix.
7364 2017-04-27 Richard Biener <rguenther@suse.de>
7367 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
7368 stripping ARRAY_REFs from MEM_EXPR make sure we're not
7369 keeping a reference to a trailing array.
7371 2017-04-27 Richard Biener <rguenther@suse.de>
7374 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
7375 being in loop-closed SSA form conservatively.
7376 (chrec_fold_multiply_poly_poly): Likewise.
7378 2017-04-27 Tamar Christina <tamar.christina@arm.com>
7381 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
7382 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
7384 2017-04-27 Jakub Jelinek <jakub@redhat.com>
7387 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
7388 (aarch64_function_arg_alignment): Return unsigned int again, but still
7389 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
7390 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
7391 Don't emit -Wpsabi note.
7392 (aarch64_function_arg_boundary): Likewise.
7393 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
7396 2017-04-26 Nathan Sidwell <nathan@acm.org>
7398 * tree.h (crc32_unsigned_n): Declare.
7399 (crc32_unsigned, crc32_unsigned): Make inline.
7400 * tree.c (crc32_unsigned_bits): Replace with ...
7401 (crc32_unsigned_n): ... this.
7402 (crc32_unsigned, crc32_byte): Remove.
7403 (crc32_string): Remove unnecessary braces.
7405 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
7407 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
7408 * ipa-inline-analysis.c (MAX_TIME): Remove.
7409 (account_size_time): Use sreal for time.
7410 (dump_inline_summary): Update.
7411 (estimate_function_body_sizes): Update.
7412 (estimate_edge_size_and_time): Update.
7413 (estimate_calls_size_and_time): Update.
7414 (estimate_node_size_and_time): Update.
7415 (inline_merge_summary): Update.
7416 (inline_update_overall_summary): Update.
7417 (estimate_time_after_inlining): Update.
7418 (inline_read_section): Update.
7419 (inline_write_summary): Update.
7420 * ipa-inline.c (compute_uninlined_call_time): Update.
7421 (compute_inlined_call_time): Update.
7422 (recursive_inlining): Update.
7423 (inline_small_functions): Update.
7424 (dump_overall_stats): Update.
7425 * ipa-inline.h: Include sreal.h.
7426 (size_time_entry): Turn time to sreal.
7427 (inline_summary): Turn self_time nad time to sreal.
7429 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
7431 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
7433 (sreal::stream_out, sreal::stream_in): New.
7434 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
7436 2017-04-25 Jakub Jelinek <jakub@redhat.com>
7438 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
7441 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
7444 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
7446 (dimode_scalar_chain::compute_convert_gain): Ditto.
7447 (dimode_scalar_chain::make_vector_copies): Ditto.
7448 (dimode_scalar_chain::convert_reg): Ditto.
7449 (dimode_scalar_chain::convert_insn): Ditto.
7450 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
7451 (VI248_AVX512BW_1): New mode iterator.
7452 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
7453 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
7456 2017-04-25 Martin Sebor <msebor@redhat.com>
7458 PR tree-optimization/80497
7459 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
7460 constants are representable in HOST_WIDE_INT.
7461 (parse_directive): Ditto.
7463 2017-04-25 Martin Sebor <msebor@redhat.com>
7466 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
7467 (new_zero_array): Adjust signature.
7468 (dom_info::dom_init): Used unsigned rather that size_t.
7469 (dom_info::dom_info): Same.
7471 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7472 Jakub Jelinek <jakub@redhat.com>
7475 * config/arm/arm.c: Include gimple.h.
7476 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
7477 returns negative, increment ncrn only if it returned positive.
7478 (arm_needs_doubleword_align): Return int instead of bool,
7479 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
7480 members, but if there is any such non-FIELD_DECL
7481 > PARM_BOUNDARY aligned decl, return -1 instead of false.
7482 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
7483 returns negative, increment nregs only if it returned positive.
7484 (arm_setup_incoming_varargs): Likewise.
7485 (arm_function_arg_boundary): Emit -Wpsabi note if
7486 arm_needs_doubleword_align returns negative, return
7487 DOUBLEWORD_ALIGNMENT only if it returned positive.
7489 2017-04-25 Marek Polacek <polacek@redhat.com>
7492 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
7493 first argument to type.
7495 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
7498 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
7499 type checks to test for compatibility instead of equality.
7501 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7502 Jakub Jelinek <jakub@redhat.com>
7505 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
7507 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
7508 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
7509 the alignment computation, but return their maximum in warn_alignment.
7510 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
7511 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
7513 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
7514 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
7517 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
7519 * config/arc/simdext.md (dmpyh): Fix typo.
7521 2017-04-25 Richard Biener <rguenther@suse.de>
7523 PR tree-optimization/80492
7524 * alias.c (compare_base_decls): Handle registers with asm
7525 specification conservatively.
7526 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
7527 compare_base_decls returning dont-know properly.
7529 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
7531 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
7532 (legitimate_offset_address_p): New function.
7533 (arc_legitimate_address_p): Use above function.
7535 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
7537 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
7539 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
7541 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
7542 ACCH registers whenever they are available.
7544 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
7546 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
7547 double regs fix when not used.
7549 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
7551 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
7553 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
7554 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
7556 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
7558 * config/arc/arc.c (arc_output_addsi): Check for h-register class
7559 when emitting short ADD instructions.
7561 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
7563 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
7565 (cmpsi_cc_c_insn): Likewise.
7566 (cbranchsi4_scratch): Compute proper instruction length using
7567 compact_hreg_operand.
7568 * config/arc/predicates.md (compact_hreg_operand): New predicate.
7570 2017-04-25 Richard Biener <rguenther@suse.de>
7573 * passes.c (pass_manager::pass_manager): Initialize
7576 2017-04-25 Richard Biener <rguenther@suse.de>
7578 PR tree-optimization/79201
7579 * tree-ssa-sink.c (statement_sink_location): Handle calls.
7581 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7584 * config/s390/vector.md: Split MEM->GPR vector moves for
7585 non-s_operand addresses.
7587 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7590 * config/s390/predicates.md (reload_const_wide_int_operand): New
7592 * config/s390/s390.md ("movti"): Remove d/P alternative.
7593 ("movti_bigconst"): New pattern definition.
7595 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
7598 * s390-protos.h (s390_expand_cs_hqi): Removed.
7599 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
7600 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
7601 modes as well as CCZ1mode and CCZmode.
7602 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
7603 signature of s390_emit_compare_and_swap.
7604 (s390_expand_cs_hqi): Likewise, make static.
7605 (s390_expand_cs_tdsi): Generate an explicit compare before trying
7606 compare-and-swap, in some cases.
7607 (s390_expand_cs): Wrapper function.
7608 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
7610 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
7611 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
7612 patterns for small and large integers. Forbid symref memory operands.
7613 Move expander to s390.c. Require cc register.
7614 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
7615 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
7616 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
7617 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
7618 symref memory operands. Remove CC mode and call s390_match_ccmode
7620 ("atomic_exchange<mode>"): Allow and implement all integer modes.
7622 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
7624 * config/s390/s390.md (define_peephole2): New peephole to help
7625 combining the load-and-test pattern with volatile memory.
7627 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
7629 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
7630 with CCZmode for TARGET_Z196.
7632 2017-04-25 Jakub Jelinek <jakub@redhat.com>
7634 PR rtl-optimization/80501
7635 * combine.c (make_compound_operation_int): Set subreg_code to SET
7636 even for AND with mask of the sign bit of mode.
7638 PR rtl-optimization/80500
7639 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
7640 sum's initial value.
7642 2017-04-25 Julian Brown <julian@codesourcery.com>
7643 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7645 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
7647 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
7649 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
7651 2017-04-25 Julian Brown <julian@codesourcery.com>
7652 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7654 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
7655 (thunderx2t99_sha): New Reservation.
7657 2017-04-25 Julian Brown <julian@codesourcery.com>
7658 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7660 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
7661 type for 1-element load.
7663 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
7665 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
7667 2017-04-24 Martin Jambor <mjambor@suse.cz>
7669 PR tree-optimization/80293
7670 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
7671 char arrays not totally scalarizable if it is false.
7672 (analyze_all_variable_accesses): Pass correct value in the new
7673 parameter. Add a statistics counter.
7675 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
7678 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
7680 2017-04-24 Richard Biener <rguenther@suse.de>
7682 PR tree-optimization/80494
7683 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
7684 out for complex types.
7686 2017-04-24 Richard Biener <rguenther@suse.de>
7688 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
7689 * tree-ssa-sccvn.c (print_scc): Print SCC size.
7690 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
7691 (DFS): Adjust and never fail.
7692 (sccvn_dom_walker::fail): Remove.
7693 (sccvn_dom_walker::before_dom_children): Adjust.
7694 (run_scc_vn): Likewise and never fail.
7695 * tree-ssa-pre.c (pass_pre::execute): Adjust.
7696 (pass_fre::execute): Likewise.
7698 2017-04-24 Richard Biener <rguenther@suse.de>
7700 PR tree-optimization/79725
7701 * tree-ssa-sink.c (statement_sink_location): Return whether
7702 failure reason was zero uses. Move that check later.
7703 (sink_code_in_bb): Deal with zero uses by removing the stmt
7706 2017-04-24 Richard Biener <rguenther@suse.de>
7709 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
7710 pointer-based references.
7712 2017-04-24 Richard Biener <rguenther@suse.de>
7715 * pass_manager.h (pass_manager::operator new): Remove.
7716 (pass_manager::operator delete): Likewise.
7717 * passes.c (pass_manager::operator new): Remove.
7718 (pass_manager::operator delete): Likewise.
7719 (pass_manager::pass_manager): Zero individual pass members.
7721 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
7724 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
7725 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
7726 Check "XEXP (src, 1)" operand here.
7727 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
7728 Check "XEXP (src, 1)" operand here.
7729 (dimode_scalar_chain::make_vector_copies): Detect count register
7730 of a shift instruction. Zero extend count register from QImode
7731 to DImode to satisfy vector shift pattern count operand predicate.
7732 Substitute vector shift count operand with a DImode copy.
7733 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
7736 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
7738 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
7739 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
7740 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
7741 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
7742 (UNSPEC_NOREX_MEM): Remove definition.
7744 2017-04-21 Richard Biener <rguenther@suse.de>
7746 PR tree-optimization/79547
7747 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
7748 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
7749 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
7750 without any constraints.
7752 2017-04-21 Richard Biener <rguenther@suse.de>
7754 PR tree-optimization/78847
7755 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
7757 2017-04-21 Richard Biener <rguenther@suse.de>
7759 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
7760 (build_distinct_type_copy): Likewise.
7761 (build_variant_type_copy): Likewise.
7762 * tree.c (build_qualified_type): Pass down mem-stat info.
7763 (build_distinct_type_copy): Likewise.
7764 (build_variant_type_copy): Likewise.
7766 2017-04-21 Richard Biener <rguenther@suse.de>
7768 PR tree-optimization/80237
7769 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
7771 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
7772 for a simplified result.
7774 2016-04-21 Richard Biener <rguenther@suse.de>
7776 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
7777 sth as strict as a simple_iv but a chrec without symbols and an
7778 operand defined in the loop we are peeling (and not some subloop).
7779 (propagate_constants_for_unrolling): Propagate all constants.
7781 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
7784 * config/i386/i386.c (print_reg): Remove assert for disalowed
7785 regno values, call output_operand_lossage instead.
7787 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
7790 * config/i386/constraints.md (Yc): New register constraint.
7791 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
7792 Use Yc constraint for alternative 2 of operand 0. Remove
7793 preferred_for_speed attribute.
7795 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
7797 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
7798 lastprivate clauses in SIMT case.
7800 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
7802 * doc/invoke.texi (-Wextra-semi): Document new warning option.
7804 2017-04-20 Richard Biener <rguenther@suse.de>
7806 PR tree-optimization/57796
7807 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
7809 (vect_model_load_cost): Cost gathers as N scalar loads.
7811 2017-04-20 Richard Biener <rguenther@suse.de>
7813 * ggc-page.c (ggc_allocated_p): Rename to ...
7814 (safe_lookup_page_table_entry): ... this and return the lookup
7816 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
7818 2017-04-20 Richard Biener <rguenther@suse.de>
7820 PR tree-optimization/80453
7821 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
7822 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
7823 from the conditions.
7824 (vn_phi_eq): Pass them down.
7825 (vn_phi_lookup): Record them.
7826 (vn_phi_insert): Likewise.
7828 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
7830 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
7831 uninitialized variable warning to avoid buffer overrun.
7833 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
7836 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
7837 is suppressed for '{ 0 }' in C.
7839 2017-04-20 Jakub Jelinek <jakub@redhat.com>
7841 * BASE-VER: Set to 8.0.0.
7843 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
7845 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
7846 priority .init_array and .fini_array section with SECTION_NOTYPE
7849 2017-04-20 Jakub Jelinek <jakub@redhat.com>
7852 * tree.h (build_array_type): Add typeless_storage default argument.
7853 * tree.c (type_cache_hasher::equal): Also compare
7854 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
7855 (build_array_type): Add typeless_storage argument, set
7856 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
7858 (build_nonshared_array_type): Adjust build_array_type_1 caller.
7859 (build_array_type): Likewise. Add typeless_storage argument.
7861 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
7862 Jakub Jelinek <jakub@redhat.com>
7864 PR tree-optimization/80426
7865 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
7866 operation on symbolic operands, also compute the overflow for the
7867 invariant part when the operation degenerates into a negation.
7869 2017-04-19 Jakub Jelinek <jakub@redhat.com>
7872 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
7873 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
7876 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
7878 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
7881 * config/avr/avr.c (tree.h): Include it.
7882 (cgraph.h): Include it.
7883 (avr_encode_section_info): Don't warn for uninitialized progmem
7884 variable if it's just an alias.
7886 2017-04-19 Richard Biener <rguenther@suse.de>
7889 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
7890 when needed by AutoPGO.
7892 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
7895 * doc/lto.texi: Remove an extra 'that'.
7897 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
7899 PR rtl-optimization/80429
7900 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
7901 are only used in debug insns.
7903 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
7904 Vladimir Makarov <vmakarov@redhat.com>
7906 * config/sparc/predicates.md (input_operand): Add comment. Return
7907 true for any memory operand when LRA is in progress.
7908 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
7910 2017-04-18 Jeff Law <law@redhat.com>
7913 * mips.md ({return,simple_return}_internal): Do not overwrite
7916 2017-04-18 Jakub Jelinek <jakub@redhat.com>
7918 PR tree-optimization/80443
7919 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
7920 instead of adding 1, subtract -1 and similarly instead of subtracting
7923 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
7925 PR rtl-optimization/80357
7926 * haifa-sched.c (tmp_bitmap): New variable.
7927 (model_recompute): Handle duplicate use records.
7928 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
7929 (free_global_sched_pressure_data): Free it.
7931 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
7934 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
7935 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
7936 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
7937 instead of SYSTEM_HEADER_DIR.
7939 2017-04-18 Jeff Law <law@redhat.com>
7942 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
7943 predecessors after walking up the insn chain.
7945 2017-04-18 Jakub Jelinek <jakub@redhat.com>
7948 * dwarf2out.c (modified_type_die): Try harder not to emit internal
7949 sizetype type into debug info.
7951 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
7954 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
7955 unneeded test for TARGET_UPPER_REGS_SF.
7956 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
7958 2017-04-18 Jakub Jelinek <jakub@redhat.com>
7961 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
7962 instead of gsi_after_labels.
7964 2017-04-18 Jeff Law <law@redhat.com>
7966 * regcprop.c (maybe_mode_change): Avoid creating copies of the
7970 2017-04-13 Jeff Law <law@redhat.com>
7971 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
7972 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
7974 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
7977 * config/avr/avr.c (intl.h): Include it.
7978 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
7980 2017-04-18 Martin Liska <mliska@suse.cz>
7982 PR gcov-profile/78783
7983 * gcov-tool.c (gcov_output_files): Validate that destination
7984 file is either removed by the tool or by a user.
7986 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
7987 Guy Benyei <guybe@mellanox.com>
7989 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
7990 block, and do not negate it, the stored id is already negative.
7992 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
7994 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
7996 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7999 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
8000 masks of options that should be turned off if the VSX vector
8001 options are turned off.
8002 (OTHER_P8_VECTOR_MASKS): Likewise.
8003 (OTHER_VSX_VECTOR_MASKS): Likewise.
8004 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
8005 rs6000_disable_incompatible_switches to validate no type switches
8007 (rs6000_incompatible_switch): New function to disallow turning on
8008 other vector options if -mno-vsx, -mno-power8-vector, or
8009 -mno-power9-vector are specified.
8011 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
8013 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
8015 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
8017 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
8018 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
8019 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
8020 (ARG_POINTER_CFA_OFFSET): Likewise.
8022 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
8024 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
8025 conditions to take advantage of various optimizations.
8027 2017-04-13 Jeff Law <law@redhat.com>
8029 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
8030 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
8031 (zero_extendsidi2_dext): Likewise.
8033 2017-04-13 Jakub Jelinek <jakub@redhat.com>
8036 * fold-const.c (fold_ternary_loc): Revert
8037 use op0 instead of fold_convert_loc (loc, type, arg0) part of
8040 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
8042 PR rtl-optimization/80343
8043 * lra-remat.c (update_scratch_ops): Assign original hard reg to
8046 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
8049 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
8050 to ubsan_encode_value.
8052 2017-04-13 Jeff Law <law@redhat.com>
8054 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
8055 appearing in DEBUG_INSNs.
8057 2017-04-13 Martin Liska <mliska@suse.cz>
8059 PR gcov-profile/80413
8060 * gcov-io.c (gcov_write_string): Copy to buffer just when
8061 allocated size is greater than zero.
8063 2017-04-13 Jakub Jelinek <jakub@redhat.com>
8066 * dwarf2out.c (decls_for_scope): Ignore declarations of
8067 current_function_decl in BLOCK_NONLOCALIZED_VARS.
8069 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
8072 * ipa-visibility.c (non_local_p): Fix typos.
8073 (localize_node): When localizing symbol in same comdat group,
8074 dissolve the group only when we know external symbols are going
8076 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
8078 2017-04-12 Jakub Jelinek <jakub@redhat.com>
8080 PR tree-optimization/79390
8081 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
8082 order does not result in usable sequence, retry with reversed operand
8088 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
8089 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
8090 op0 instead of fold_convert_loc (loc, type, arg0).
8092 2017-04-12 Jeff Law <law@redhat.com>
8094 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
8095 has a delay slot in the generated code.
8097 * config/cris/cris.md (cris_preferred_reload_class): Return
8098 GENNONACR_REGS rather than GENERAL_REGS.
8100 2017-04-12 Jakub Jelinek <jakub@redhat.com>
8103 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
8104 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
8105 signedness of the result type.
8107 2017-04-12 Richard Biener <rguenther@suse.de>
8108 Jeff Law <law@redhat.com>
8110 PR tree-optimization/80359
8111 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
8112 trim stores to TARGET_MEM_REFs.
8114 2017-04-12 Richard Biener <rguenther@suse.de>
8116 PR tree-optimization/79390
8117 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
8118 threading case even more.
8120 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
8123 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
8124 for quad_address_p for TImode, instead of just not indexed_address.
8126 2017-04-12 Richard Biener <rguenther@suse.de>
8127 Bernd Edlinger <bernd.edlinger@hotmail.de>
8130 * alias.c (component_uses_parent_alias_set_from): Handle
8131 TYPE_TYPELESS_STORAGE.
8132 (get_alias_set): Likewise.
8133 * tree-core.h (tree_type_common): Add typeless_storage flag.
8134 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
8135 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
8136 for types containing members with TYPE_TYPELESS_STORAGE.
8137 (place_field): Likewise.
8138 (layout_type): Likewise for ARRAY_TYPE.
8139 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
8140 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
8141 TYPE_TYPELESS_STORAGE.
8142 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
8144 2017-04-12 Jakub Jelinek <jakub@redhat.com>
8147 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
8148 first argument to type.
8150 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8154 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
8155 CONST0_RTX (mode) rather than const0_rtx where appropriate.
8156 (rs6000_expand_binop_builtin): Likewise.
8157 (rs6000_expand_ternop_builtin): Likewise; also add missing
8158 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
8159 vshasigma built-ins.
8160 * doc/extend.texi: Document that vec_xxpermdi's third argument
8163 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
8165 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
8166 Use shift_const cost parameter when calculating gain of STV shifts.
8168 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
8170 PR rtl-optimization/70478
8171 * lra-constraints.c (process_alt_operands): Check memory for
8172 disfavoring memory insn operand.
8174 2017-04-11 Jakub Jelinek <jakub@redhat.com>
8177 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
8178 left shift in unsigned HOST_WIDE_INT type.
8180 PR rtl-optimization/80385
8181 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
8182 (not (neg X)) into (plus X -1) for complex or non-integral modes.
8185 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
8186 if they have any depend clauses.
8188 2017-04-11 Martin Liska <mliska@suse.cz>
8191 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
8192 * ipa-split.c (split_function): Create a local comdat symbol
8193 if caller is in a comdat group.
8195 2017-04-11 Martin Liska <mliska@suse.cz>
8198 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
8201 2017-04-11 Martin Sebor <msebor@redhat.com>
8204 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
8205 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
8207 (directive::set_width, directive::set_precision, format_character):
8209 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
8212 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
8215 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
8216 conflict, set target->arch_name instead of target->cpu_name.
8218 2017-04-11 Richard Biener <rguenther@suse.de>
8220 PR tree-optimization/80374
8221 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
8222 build_zero_cst, remove fold_convertible_p check again.
8224 2017-04-11 Martin Liska <mliska@suse.cz>
8227 * ubsan.c (instrument_object_size): Do not instrument register
8230 2017-04-11 Jakub Jelinek <jakub@redhat.com>
8233 * config/i386/i386-builtin-types.def
8234 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
8235 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
8236 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
8237 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
8238 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
8239 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
8240 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
8241 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
8242 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
8243 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
8244 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
8245 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
8246 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
8247 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
8248 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
8249 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
8250 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
8251 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
8252 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
8253 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
8254 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
8255 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
8256 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
8257 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
8258 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
8259 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
8260 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
8261 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
8262 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
8263 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
8264 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
8265 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
8266 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
8267 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
8268 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
8269 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
8270 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
8271 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
8272 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
8273 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
8274 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
8275 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
8276 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
8277 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
8278 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
8279 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
8281 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
8282 flag to second_arg_count, handle 4 argument function type _COUNT
8283 aliases, handle second_arg_count on second argument rather than last.
8285 2017-04-10 Jeff Law <law@redhat.com>
8287 PR tree-optimization/80374
8288 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
8289 record anything if we can not convert integer_zero_node to the
8292 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
8295 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8296 Enhance special handling given to the TARGET_P9_MINMAX option in
8297 relation to certain other options.
8299 2017-04-10 Bin Cheng <bin.cheng@arm.com>
8301 PR tree-optimization/80153
8302 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
8303 remove POINTER_PLUS_EXPR's base part directly, rather than through
8306 2017-04-10 Richard Biener <rguenther@suse.de>
8307 Bin Cheng <bin.cheng@arm.com>
8309 PR tree-optimization/80153
8310 * tree-affine.c (aff_combination_to_tree): Get base pointer from
8311 the first element of pointer type aff_tree. Build result expr in
8313 (add_elt_to_tree): Convert to type unconditionally. Remove other
8315 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
8316 (rewrite_use_nonlinear_expr): Check invariant using iv information.
8318 2017-04-10 Richard Biener <rguenther@suse.de>
8320 * tree-ssa-structalias.c (find_func_aliases): Properly handle
8323 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
8325 PR rtl-optimization/70478
8326 * lra-constraints.c (curr_small_class_check): New.
8327 (update_and_check_small_class_inputs): New.
8328 (process_alt_operands): Update curr_small_class_check. Disfavor
8329 alternative insn memory operands. Check available regs for small
8332 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
8335 * config/mips/mips.opt (-mvirt): Update description.
8336 * doc/invoke.texi (-mvirt): Likewise.
8338 2017-04-10 Richard Biener <rguenther@suse.de>
8341 * fold-const.c (fold_binary_loc): Look at unstripped ops when
8342 looking for NEGATE_EXPR in -A / -B to A / B folding.
8344 2017-04-10 Martin Liska <mliska@suse.cz>
8346 PR gcov-profile/80224
8347 * gcov.c (print_usage): Fix usage string.
8348 (get_gcov_intermediate_filename): Remove.
8349 (output_gcov_file): Use both for normal and intermediate format.
8350 (generate_results): Do not initialize special file for
8351 intermediate format.
8353 2017-04-10 Richard Biener <rguenther@suse.de>
8355 PR tree-optimization/80304
8356 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
8359 2017-04-10 Nathan Sidwell <nathan@acm.org>
8362 * config/rs6000/rs6000.c (rs6000_vector_type): New.
8363 (rs6000_init_builtins): Use it.
8365 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8367 * config/arm/arm.md (<mrc>): Add mode to SET source.
8370 2017-04-10 Richard Biener <rguenther@suse.de>
8373 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
8375 2017-04-10 Jakub Jelinek <jakub@redhat.com>
8378 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
8379 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
8380 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
8381 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
8382 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
8383 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
8384 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
8385 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
8386 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
8387 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
8388 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
8389 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
8390 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
8391 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
8392 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
8393 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
8394 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
8395 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
8396 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
8397 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
8398 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
8399 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
8400 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
8402 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
8404 PR rtl-optimization/70478
8405 * lra-constraints.c: Reverse the last patch.
8407 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
8409 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
8410 Add comment for WCHAR_T.
8412 2017-04-08 Martin Liska <mliska@suse.cz>
8415 2017-04-07 Martin Liska <mliska@suse.cz>
8418 * ipa-split.c (split_function): Add function part to a same comdat
8421 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8424 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
8426 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
8428 * rs6000/rs6000.c (vec_load_pendulum): Rename...
8429 (vec_pairing): ...to this.
8430 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
8431 (rs6000_sched_init): Adjust for name change.
8432 (struct rs6000_sched_context): Likewise.
8433 (rs6000_init_sched_context): Likewise.
8434 (rs6000_set_sched_context): Likewise.
8436 2017-04-07 Jakub Jelinek <jakub@redhat.com>
8442 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
8444 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
8445 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
8446 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
8448 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
8450 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
8452 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
8454 PR rtl-optimization/70703
8455 * ira-color.c (update_conflict_hard_regno_costs): Use
8456 int64_t instead of HOST_WIDE_INT.
8458 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
8460 PR rtl-optimization/70478
8461 * lra-constraints.c (process_alt_operands): Disfavor alternative
8462 insn memory operands.
8464 2017-04-07 Jeff Law <law@redhat.com>
8466 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
8467 CALL and NOTE_INSN_CALL_ARG_LOCATION.
8469 2017-04-07 Martin Liska <mliska@suse.cz>
8472 * config/aarch64/aarch64.c (aarch64_process_target_attr):
8473 Show error message instead of an ICE.
8475 2017-04-07 Martin Liska <mliska@suse.cz>
8478 * ipa-split.c (split_function): Add function part to a same comdat
8481 2017-04-07 Richard Biener <rguenther@suse.de>
8484 * tree.c (get_unwidened): Also handle ! for_type case for
8486 * convert.c (do_narrow): Split out from ...
8487 (convert_to_integer_1): ... here. Do not pass final truncation
8488 type to get_unwidened for TRUNC_DIV_EXPR.
8490 2017-04-07 Richard Biener <rguenther@suse.de>
8492 * tree-affine.c (wide_int_ext_for_comb): Take type rather
8494 (aff_combination_const): Adjust.
8495 (aff_combination_scale): Likewise.
8496 (aff_combination_add_elt): Likewise.
8497 (aff_combination_add_cst): Likewise.
8498 (aff_combination_convert): Likewise.
8499 (add_elt_to_tree): Likewise. Remove unused argument.
8500 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
8502 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
8504 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
8506 * config/arm/arm.c (arm_default_short_enums): Use
8507 ARM_DEFAULT_SHORT_ENUMS.
8508 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
8510 2017-04-06 Jakub Jelinek <jakub@redhat.com>
8513 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
8514 members with redundant out-of-class redeclaration.
8516 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
8519 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
8520 * config/i386/i386.md (*zero_extendsidi2):
8521 Add (?*x,*x) and (?*v,*v) alternatives.
8523 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
8526 * config/i386/i386.c (ix86_expand_builtin)
8527 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
8528 mode from insn data. Convert operands to insn operand mode.
8529 Copy operands that don't satisfy insn predicate to a register.
8531 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
8533 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
8536 2017-04-06 Richard Biener <rguenther@suse.de>
8538 PR tree-optimization/80334
8539 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
8540 preserve alignment of accesses.
8542 2017-04-06 Richard Biener <rguenther@suse.de>
8544 PR tree-optimization/80262
8545 * tree-sra.c (build_ref_for_offset): Preserve address-space
8547 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
8548 Drop useless address-space information on MEM_REF offsets.
8550 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
8552 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
8554 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
8556 PR rtl-optimization/70703
8557 * ira-color.c (update_conflict_hard_regno_costs): Use
8558 HOST_WIDE_INT instead of long.
8560 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
8563 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
8564 not defined for x86_64 target. Add -mmmx target option when __SSE2__
8566 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
8567 for x86_64 target. Handle -m3dnowa option.
8569 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
8571 PR rtl-optimization/70703
8572 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
8573 (update_conflict_hard_regno_costs): Use long instead of unsigned
8574 arithmetic for cost calculation.
8576 2017-04-05 Jakub Jelinek <jakub@redhat.com>
8577 Bernd Edlinger <bernd.edlinger@hotmail.de>
8580 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
8583 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
8586 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
8587 ptr_mode with Pmode throughout.
8588 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
8589 into probe_stack_range and use DImode.
8591 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
8594 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
8595 call_eh_return is true.
8597 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8599 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
8600 Initialize last_match_fntype_index.
8602 2017-04-05 Jakub Jelinek <jakub@redhat.com>
8605 * tree-nvr.c: Include internal-fn.h.
8606 (pass_return_slot::execute): Ignore internal calls without
8609 2017-04-04 Jakub Jelinek <jakub@redhat.com>
8610 Richard Biener <rguenther@suse.de>
8613 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
8614 captures used multiple times, except for the last use.
8615 * generic-match-head.c: Include gimplify.h.
8617 2017-04-04 Jakub Jelinek <jakub@redhat.com>
8619 PR tree-optimization/79390
8620 * target.h (struct noce_if_info): Declare.
8621 * targhooks.h (default_noce_conversion_profitable_p): Declare.
8622 * target.def (noce_conversion_profitable_p): New target hook.
8623 * ifcvt.h (struct noce_if_info): New type, moved from ...
8624 * ifcvt.c (struct noce_if_info): ... here.
8625 (noce_conversion_profitable_p): Renamed to ...
8626 (default_noce_conversion_profitable_p): ... this. No longer
8628 (noce_try_store_flag_constants, noce_try_addcc,
8629 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
8630 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
8631 instead of noce_conversion_profitable_p.
8632 * config/i386/i386.c: Include ifcvt.h.
8633 (ix86_option_override_internal): Don't override
8634 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
8635 (ix86_noce_conversion_profitable_p): New function.
8636 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
8637 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
8638 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
8639 * doc/tm.texi: Regenerated.
8641 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8643 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
8646 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
8649 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
8650 instructions for small multiply cores.
8652 2017-04-04 Jeff Law <law@redhat.com>
8654 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
8656 (mips_expand_vec_perm_const): Initialize elements in orig_perm
8657 that are not set by the loop over the elements.
8659 2017-04-04 Jakub Jelinek <jakub@redhat.com>
8662 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
8663 int mode, convert_modes it to mode as unsigned, otherwise use
8664 lowpart_subreg to mode rather than SImode.
8665 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
8666 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
8667 Use DImode instead of SImode for the shift count operand.
8668 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
8671 2017-04-04 Richard Biener <rguenther@suse.de>
8674 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
8675 arithmetic done for the negate or the plus. Simplify.
8676 (A - (-B) -> A + B): Likewise.
8677 * fold-const.c (split_tree): Make sure to not negate pointers.
8679 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
8681 PR rtl-optimization/60818
8682 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
8683 a compare of comparisons with the thing compared if this results
8684 in a different machine mode.
8686 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
8688 * alias.c (base_alias_check): Fix typo in comment.
8689 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
8690 * cgraphunit.c (symbol_table::compile): Likewise.
8691 * collect2.c (maybe_run_lto_and_relink): Likewise.
8692 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
8693 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
8694 * config/avr/avr.c (avr_map_op_t): Likewise.
8695 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
8696 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
8697 * config/epiphany/epiphany.md (movcc): Likewise.
8698 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
8699 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
8701 * config/mips/mips.c (mips_save_restore_reg): Likewise.
8702 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
8703 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
8704 * config/sh/sh.c (sh_rtx_costs): Likewise.
8705 * fold-const.c (fold_truth_andor): Likewise.
8706 * genautomata.c (collapse_flag): Likewise.
8707 * gengtype.h (struct type::u::s): Likewise.
8708 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
8709 * input.c (FORMAT_AMOUNT): Likewise.
8710 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
8711 (known_aggs_to_agg_replacement_list): Likewise.
8712 * ipa-inline-analysis.c: Likewise.
8713 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
8714 * ipa-polymorphic-call.c
8715 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
8716 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
8717 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
8719 * modulo-sched.c (apply_reg_moves): Likewise.
8720 * omp-expand.c (build_omp_regions_1): Likewise.
8721 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
8722 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
8723 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
8724 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
8725 * value-prof.c: Likewise.
8726 * var-tracking.c (val_reset): Likewise.
8728 2017-04-03 Richard Biener <rguenther@suse.de>
8730 PR tree-optimization/80275
8731 * fold-const.c (split_address_to_core_and_offset): Handle
8734 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
8736 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
8737 descriptors is at least equal to that of functions.
8739 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
8741 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
8743 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
8746 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
8747 (mov<IMOD4:mode>): New expander.
8748 (*mov<IMOD4:mode>_internal): New insn and split pattern.
8750 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
8752 PR rtl-optimization/79405
8753 * fwprop.c (propagations_left): New variable.
8754 (forward_propagate_into): Decrement it.
8755 (fwprop_init): Initialize it.
8756 (fw_prop): If the variable has reached zero, stop propagating.
8757 (fwprop_addr): Ditto.
8759 2017-03-31 Jakub Jelinek <jakub@redhat.com>
8762 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
8763 a FUNCTION_DECL, pass it as decl instead of origin to
8766 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
8768 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
8771 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
8774 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
8775 TARGET_VSX_SMALL_INTEGER.
8777 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8779 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8780 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
8782 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
8784 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
8785 extraction from odd-numbered MSA register.
8787 2017-03-31 Jakub Jelinek <jakub@redhat.com>
8790 * expmed.c (store_bit_field_1): Don't attempt to create
8791 a word subreg out of hard registers wider than word if they
8792 have HARD_REGNO_NREGS of 1 for their mode.
8795 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
8796 conversions to integer types wider than word and pointer.
8799 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
8800 (rtx_equal_for_cselib_p): Pass 0 to it.
8801 * cselib.c (cselib_hasher::equal): Likewise.
8802 (rtx_equal_for_cselib_1): Add depth argument. If depth
8803 is 128, don't look up VALUE locs and punt. Increment
8804 depth in recursive calls when walking VALUE locs.
8806 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
8808 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
8809 (make_gcov_file_name): Use the canonical path name for generating
8811 (read_line): Fix handling of files with ascii null bytes.
8813 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
8815 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
8816 to initialise a vector register instead
8817 of using a const_int.
8819 2017-03-30 Jakub Jelinek <jakub@redhat.com>
8821 PR translation/80189
8822 * gimplify.c (omp_default_clause): Use %qs instead of %s in
8823 diagnostic messages.
8825 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
8828 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
8829 (dfp_diex_<mode>): Update mode of operand 1.
8830 * doc/extend.texi (dxex, dxexq): Document change to return type.
8831 (diex, diexq): Document change to argument type.
8833 2017-03-30 Martin Jambor <mjambor@suse.cz>
8836 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
8837 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
8838 it reflects the signature changes performed at the callee side.
8839 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
8840 to cgraph_build_function_type_skip_args.
8841 (build_function_decl_skip_args): Adjust call to the above function.
8843 2017-03-30 Jakub Jelinek <jakub@redhat.com>
8846 * config/i386/sse.md
8847 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
8848 register as dest whenever it is a MEM not rtx_equal_p to the
8849 corresponding dup operand, and when forcing into reg move the
8850 reg into the memory afterwards.
8851 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
8852 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
8853 for the force_reg mode.
8854 (avx512vl_vextractf128<mode>): Use register as dest either
8855 always when a MEM, or when it is a MEM not rtx_equal_p to the
8856 corresponding dup operand, or even not when it is a CONST_VECTOR
8857 depending on the mode and lo vs. hi.
8858 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
8860 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
8861 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
8862 Likewise. Require that operands[2] is even.
8863 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
8864 Remove extraneous parens. Require that operands[2] is a multiple
8866 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
8867 operands[0] is a MEM if <mask_applied>, the predicates/constraints
8868 disallow memory then.
8870 2017-03-30 Richard Biener <rguenther@suse.de>
8872 PR tree-optimization/77498
8873 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
8874 to non-constants over backedges.
8876 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
8878 PR rtl-optimization/80233
8879 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
8880 as last_combined_insn. Do not test for BARRIER_P separately.
8882 2017-03-29 Andreas Schwab <schwab@suse.de>
8885 * calls.c (prepare_call_address): Convert funexp to Pmode before
8886 copying to temp reg.
8888 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8890 PR tree-optimization/80158
8891 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
8892 Handle possible future case of more than one alternate
8894 (replace_rhs_if_not_dup): Likewise.
8895 (replace_one_candidate): Likewise.
8897 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
8899 PR rtl-optimization/80193
8900 * ira.c (ira): Do not check allocation for LRA.
8902 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
8904 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
8905 (nvptx_output_simt_exit): Declare.
8906 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
8907 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
8908 (init_softstack_frame): Move initialization of crtl->is_leaf to...
8909 (nvptx_declare_function_name): ...here. Emit declaration of local
8910 memory space buffer for omp_simt_enter insn.
8911 (nvptx_output_unisimt_switch): New.
8912 (nvptx_output_softstack_switch): New.
8913 (nvptx_output_simt_enter): New.
8914 (nvptx_output_simt_exit): New.
8915 * config/nvptx/nvptx.h (struct machine_function): New fields
8916 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
8917 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
8918 (UNSPECV_SIMT_EXIT): Ditto.
8919 (omp_simt_enter_insn): New insn.
8920 (omp_simt_enter): New expansion.
8921 (omp_simt_exit): New insn.
8922 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
8924 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
8925 (expand_GOMP_SIMT_ENTER_ALLOC): New.
8926 (expand_GOMP_SIMT_EXIT): New.
8927 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
8928 (GOMP_SIMT_ENTER_ALLOC): Ditto.
8929 (GOMP_SIMT_EXIT): Ditto.
8930 * target-insns.def (omp_simt_enter): New insn.
8931 (omp_simt_exit): Ditto.
8932 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
8934 (lower_rec_simd_input_clauses): Implement SIMT privatization.
8935 (lower_rec_input_clauses): Likewise.
8936 (lower_lastprivate_clauses): Handle SIMT privatization.
8938 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
8939 (ompdevlow_adjust_simt_enter): New.
8940 (find_simtpriv_var_op): New.
8941 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
8942 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
8944 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
8945 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
8946 (copy_decl_for_dup_finish): Ditto.
8948 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
8950 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
8953 * config/i386/i386.c (ix86_option_override_internal): Always
8954 allow -mpreferred-stack-boundary=3 for 64-bit targets.
8956 2017-03-28 Bin Cheng <bin.cheng@arm.com>
8958 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
8960 2017-03-28 Bin Cheng <bin.cheng@arm.com>
8962 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
8963 mark new edge's irreducible flag accordign to it.
8964 (vect_do_peeling): Check loop preheader edge's irreducible flag
8965 and pass it to function slpeel_add_loop_guard.
8967 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
8969 PR tree-optimization/80218
8970 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
8971 Update block frequencies and counts.
8973 2017-03-28 Richard Biener <rguenther@suse.de>
8975 PR tree-optimization/78644
8976 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
8977 of a simplification result we may not use it at all.
8979 2017-03-28 Richard Biener <rguenther@suse.de>
8982 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
8983 without arguments, generate default definition of a SSA name.
8985 2017-03-28 Richard Biener <rguenther@suse.de>
8988 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
8989 TYPE_REF_CAN_ALIAS_ALL references.
8990 * fold-const.c (fold_indirect_ref_1): Likewise.
8992 2017-03-28 Martin Liska <mliska@suse.cz>
8995 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
8996 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
8998 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
8999 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
9001 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
9002 (EXTRA_SPECS): Define.
9003 (SUBTARGET_EXTRA_SPECS): Likewise.
9004 (SUBTARGET_CPP_SPEC): Likewise.
9005 * config/arc/elf.h (EXTRA_SPECS): Renamed to
9006 SUBTARGET_EXTRA_SPECS.
9007 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
9009 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
9011 * config/arc/simdext.md (vst64_insn): Update pattern.
9012 (vld32wh_insn): Likewise.
9013 (vld32wl_insn): Likewise.
9014 (vld64_insn): Likewise.
9015 (vld32_insn): Likewise.
9017 2017-03-28 Marek Polacek <polacek@redhat.com>
9020 * fold-const.c (fold_comparison): Use protected_set_expr_location
9021 instead of SET_EXPR_LOCATION.
9023 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
9025 * tree.c (add_expr): Avoid name lookup warning.
9027 2017-03-27 Jeff Law <law@redhat.com>
9029 PR tree-optimization/80216
9030 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
9031 function name. Limit recursion depth.
9032 (record_temporary_equivalences): Corresponding changes.
9034 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
9036 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
9039 2017-03-27 Jakub Jelinek <jakub@redhat.com>
9042 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
9044 * cfgcleanup.c (reg_note_cfa_p): New array.
9045 (insns_have_identical_cfa_notes): New function.
9046 (old_insns_match_p): Don't cross-jump in between /f
9047 and non-/f instructions. If both i1 and i2 are frame related,
9048 verify all CFA notes, their order and content.
9050 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9053 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
9054 HImode and SImode with zero extend to DImode to one insn.
9055 (bswap<mode>2_extenddi): Likewise.
9056 (bswapsi2_extenddi): Likewise.
9057 (bswaphi2_extendsi): Likewise.
9058 (bswaphi2): Combine bswap HImode and SImode into one insn.
9059 Separate memory insns from swapping register.
9060 (bswapsi2): Likewise.
9061 (bswap<mode>2): Likewise.
9062 (bswaphi2_internal): Delete, no longer used.
9063 (bswapsi2_internal): Likewise.
9064 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
9065 store, and gpr<-gpr swap insns.
9066 (bswap<mode>2_store): Likewise.
9067 (bswaphi2_reg): Register only splitter, combine with the splitter.
9068 (bswaphi2 splitter): Likewise.
9069 (bswapsi2_reg): Likewise.
9070 (bswapsi2 splitter): Likewise.
9071 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
9072 the insns into load, store, and register/register insns.
9073 (bswapdi2_ldbrx): Likewise.
9074 (bswapdi2_load): Likewise.
9075 (bswapdi2_store): Likewise.
9076 (bswapdi2_reg): Likewise.
9078 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
9080 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
9081 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
9083 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
9086 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
9088 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
9089 special handling for target option conflicts between dform
9090 options (-mpower9-dform, -mpower9-dform-vector,
9091 -mpower9-dform-scalar) and -mno-direct-move.
9093 2017-03-27 Richard Biener <rguenther@suse.de>
9095 PR tree-optimization/80181
9096 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
9098 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
9100 * config/arc/predicates.md (move_double_src_operand): Replace the
9101 call to move_double_src_operand with a call to address_operand.
9103 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
9105 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
9106 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
9107 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
9109 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
9111 * config/arc/predicates.md (long_immediate_loadstore_operand):
9112 Consider scaled addresses cases.
9114 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
9116 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
9117 restored when in interrupt.
9118 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
9119 doesn't have delay slot.
9121 2017-03-27 Richard Biener <rguenther@suse.de>
9124 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
9125 inlined thunk clones.
9127 2017-03-27 Jakub Jelinek <jakub@redhat.com>
9130 * asan.c (instrument_derefs): Copy over last operand from
9131 original COMPONENT_REF to the new COMPONENT_REF with
9132 DECL_BIT_FIELD_REPRESENTATIVE.
9133 * ubsan.c (instrument_object_size): Likewise.
9135 2017-03-27 Richard Biener <rguenther@suse.de>
9137 PR tree-optimization/80170
9138 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
9139 sure DR/SCEV didnt fold in constants we do not see when looking
9140 at the reference base alignment.
9142 2017-03-27 Richard Biener <rguenther@suse.de>
9145 * gimple-fold.c (fold_ctor_reference): Properly guard against
9146 NULL return value from canonicalize_constructor_val.
9148 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
9151 * config/i386/i386.c (ix86_expand_builtin)
9152 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
9153 flags reg setting and flags reg using instructions.
9154 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
9155 clobbering instructions to zero extend op2.
9157 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
9159 * doc/install.texi (Configuration) <--with-aix-soname>:
9160 Update link to AIX ld.
9162 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
9164 PR rtl-optimization/80160
9165 PR rtl-optimization/80159
9166 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
9167 reg_alternate_class into account.
9169 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
9172 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
9173 to consider in curr_insn_transform.
9175 2017-03-24 Jakub Jelinek <jakub@redhat.com>
9177 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
9178 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
9179 and emit_mode_inner.
9181 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9183 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
9184 argument to the overloaded builtin variants. Use the new flag to
9185 deprecate certain builtin variants.
9186 * config/s390/s390-builtin-types.def: Add new builtin types.
9187 * config/s390/s390-builtins.h: Support new flags field for
9188 overloaded builtins.
9189 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
9190 (s390_macro_to_expand): Enable vector float data type.
9191 (s390_cpu_cpp_builtins_internal): Indicate support of the new
9192 builtins by incrementing the __VEC__ version number.
9193 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
9195 (s390_resolve_overloaded_builtin): Emit error messages depending
9196 on the builtin flags.
9197 * config/s390/s390.c (s390_expand_builtin): Support additional
9198 flags argument. Change error message to match the messages
9199 emitted in s390-c.c.
9200 * config/s390/s390.md: New UNSPEC_* constants.
9201 (op_type): Add new instruction types.
9202 * config/s390/vecintrin.h: Add new builtins and test data class
9204 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
9205 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
9206 (VEC_INEXACT, VEC_NOINEXACT): New constants.
9207 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
9208 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
9209 ("vec_mergel<mode>"): V_HW -> VEC_HW.
9211 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
9212 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
9213 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
9214 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
9216 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
9217 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
9218 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
9219 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
9221 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
9222 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
9223 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
9224 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
9225 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
9226 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
9227 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
9229 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
9230 ("vec_scatter_element<V_HW_4:mode>_DI")
9231 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
9232 ("vec_fpint<mode>", "vflls")
9233 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
9234 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
9235 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
9236 ("*vec_cmphe<mode>_cc"): ... these.
9238 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
9239 mode constant instead of magic value.
9241 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9243 * config/s390/s390.c (s390_expand_vec_compare): Support other
9244 vector floating point modes than just V2DF.
9245 (s390_expand_vcond): Likewise.
9246 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
9247 (s390_cannot_change_mode_class): Prevent mode changes between TF
9248 and V1TF in vector registers.
9249 * config/s390/s390.md (DF, SF): New mode attributes.
9250 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
9251 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
9252 SFmode support for VRs.
9253 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
9255 (VFT, VF_HW): New mode iterators.
9256 (vw, sdx): New mode attributes.
9257 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
9258 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
9259 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
9260 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
9261 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
9262 also the new vector floating point modes. Renaming to ...
9264 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
9265 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
9266 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
9267 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
9268 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
9269 ("vec_unordered<mode>"): ... these.
9271 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
9272 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
9273 ("*vec_extendv2df"): New insn definitions.
9275 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9277 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
9278 ("mulditi3_2", "*muldi3_sign"): New patterns.
9279 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
9280 rename the pattern definition.
9282 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9284 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
9286 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
9288 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9290 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
9291 instruction if possible.
9292 * config/s390/vector.md (vec_halfnumelts): New mode
9294 ("*vec_vllezlf<mode>"): New pattern.
9296 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9298 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
9299 ("popcountv4si2", "popcountv2di2"): Rename to ...
9300 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
9301 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
9303 ("popcount<mode>2_vxe"): New pattern.
9305 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9307 * common/config/s390/s390-common.c (processor_flags_table): Add
9309 * config.gcc: Add arch12.
9310 * config/s390/driver-native.c (s390_host_detect_local_cpu):
9311 Default to arch12 for unknown CPU model numbers.
9312 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
9313 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
9314 PROCESSOR_max sanity check.
9315 * config/s390/s390-opts.h (enum processor_type): Add
9317 * config/s390/s390.c (processor_table): Add arch12.
9318 (s390_expand_builtin): Add check for B_VXE flag.
9319 (s390_issue_rate): Add PROCESSOR_ARCH12.
9320 (s390_get_sched_attrmask): Likewise.
9321 (s390_get_unit_mask): Likewise.
9322 (s390_sched_score): Enable z13 scheduling for arch12.
9323 (s390_sched_reorder): Likewise.
9324 (s390_sched_variable_issue): Likewise.
9325 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
9327 (s390_tune_attr): Use z13 scheduling also for arch12.
9328 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
9329 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
9330 (TARGET_VXE_P): New macros.
9331 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
9332 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
9333 * config/s390/s390.opt: Add arch12 as processor_type.
9335 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9337 * config/s390/s390.md
9338 ("fixuns_truncdddi2", "fixuns_trunctddi2")
9339 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
9340 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
9342 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
9343 Rename expanders to ...
9345 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
9346 ("fixuns_truncdddi2_emu"): ... these.
9348 ("fixuns_trunc<mode>si2_emu"): New expander.
9350 ("*fixuns_truncdfdi2_z13"): Rename to ...
9351 ("*fixuns_truncdfdi2_vx"): ... this.
9353 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9355 * config/s390/2964.md: Remove the single element vector compare
9356 instructions which are no longer used.
9357 * config/s390/s390.c (s390_select_ccmode): Remove handling of
9359 (s390_canonicalize_comparison): Remove handling of DFmode
9361 (s390_expand_vec_compare_scalar): Remove function.
9362 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
9363 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
9365 ("*cmp<mode>_ccs"): Add wfcdb instruction.
9367 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9369 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
9371 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
9372 will anyway by matched by mov<mode>_64dfp.
9374 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9376 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
9377 vlef/vstef. Add missing operand to vleif.
9379 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9381 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
9382 pair for all vector types with 64 bit elements.
9383 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
9384 * config/s390/vector.md (V_HW_64): ... here.
9385 (V_128_NOSINGLE): New mode iterator.
9386 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
9387 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
9388 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
9389 ("*vec_load_pairv2di"): Change to ...
9390 ("*vec_load_pair<mode>"): ... this one.
9392 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9394 * config/s390/constraints.md: Add comments.
9395 (jKK): Reject element sizes > 8 bytes.
9396 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
9398 * config/s390/s390.md: Add the s_operand checks formerly in
9399 s390_split_ok_p to various splitters where they are still
9401 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
9402 for 128 bit vectors. Plus two splitters.
9404 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9406 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
9409 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9412 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
9413 error if the boundary argument is not constant.
9415 2017-03-24 Jakub Jelinek <jakub@redhat.com>
9417 PR rtl-optimization/80112
9418 * loop-doloop.c (doloop_condition_get): Don't check condition
9419 if cmp isn't SET with IF_THEN_ELSE src.
9421 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9423 PR tree-optimization/80158
9424 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
9425 replacing a candidate statement, also replace it for the
9426 candidate's alternate interpretation.
9427 (replace_rhs_if_not_dup): Likewise.
9428 (replace_one_candidate): Likewise.
9430 2017-03-24 Richard Biener <rguenther@suse.de>
9432 PR tree-optimization/80167
9433 * graphite-isl-ast-to-gimple.c
9434 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
9436 (translate_isl_ast_to_gimple::get_rename): Likewise.
9438 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
9440 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
9441 handling of certain combinations of target options, including the
9442 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
9443 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
9445 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9448 * config/arm/arm.md (*load_multiple): Add reload_completed to
9451 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9452 Richard Biener <rguenth@suse.de>
9454 PR tree-optimization/79908
9455 PR tree-optimization/80136
9456 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
9457 been cast away, gimplify_and_add suffices.
9459 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
9461 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
9463 2017-03-23 Richard Biener <rguenther@suse.de>
9465 PR tree-optimization/80032
9466 * gimplify.c (gimple_push_cleanup): Forced unconditional
9467 cleanups still have to go to the conditional_cleanups
9470 2017-03-22 Jakub Jelinek <jakub@redhat.com>
9472 PR tree-optimization/80072
9473 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
9475 (next_operand_entry_id): Change type to unsigned int.
9476 (sort_by_operand_rank): Make sure to return the right return value
9477 even if unsigned fields are bigger than INT_MAX.
9478 (struct oecount): Change cnt and id type to unsigned int.
9479 (oecount_hasher::equal): Formatting fix.
9480 (oecount_cmp): Make sure to return the right return value
9481 even if unsigned fields are bigger than INT_MAX.
9482 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
9485 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
9486 TREE_READONLY on result if writing it more than once.
9489 * doc/invoke.texi (-fsanitize=thread): Document that with
9490 -fnon-call-exceptions atomics are not able to throw
9494 * tsan.c: Include tree-eh.h.
9495 (instrument_builtin_call): Call maybe_clean_eh_stmt or
9496 maybe_clean_or_replace_eh_stmt where needed.
9497 (instrument_memory_accesses): Add cfg_changed argument.
9498 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
9499 if it returned true.
9500 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
9502 PR rtl-optimization/63191
9503 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
9504 wrapper function, moved the whole old content into ...
9505 (ix86_delegitimize_address_1): ... this. New inline function.
9506 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
9507 true as last argument instead of ix86_delegitimize_address.
9509 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
9511 * config/aarch64/aarch64.c (generic_branch_cost): Copy
9512 cortexa57_branch_cost.
9514 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
9516 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
9518 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9521 * doc/md.texi (Constraints): Document wA constraint.
9522 * config/rs6000/constraints.md (wA): New.
9523 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
9524 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
9525 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
9526 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
9528 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
9531 * gimplify.c (is_oacc_declared): New function.
9532 (oacc_default_clause): Use it to set default flags for acc declared
9533 variables inside parallel regions.
9534 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
9536 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
9537 declare attribute to any decl as necessary.
9539 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
9542 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
9543 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
9544 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
9545 (arm_arch_lpae): This.
9546 * config/arm/arm.c (arm_arch7ve): Rename into ...
9547 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
9548 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
9551 2017-03-22 Martin Liska <mliska@suse.cz>
9554 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
9555 error message instead of an ICE.
9557 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9559 * doc/extend.texi (6.11 Additional Floating Types): Revise.
9561 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
9563 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
9565 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
9568 2017-03-21 Martin Sebor <msebor@redhat.com>
9570 * doc/extend.texi: Use "cannot" instead of "can't."
9571 * doc/hostconfig.texi: Same.
9572 * doc/install.texi: Same.
9573 * doc/invoke.texi: Same.
9574 * doc/loop.texi: Same.
9575 * doc/md.texi: Same.
9576 * doc/objc.texi: Same.
9577 * doc/rtl.texi: Same.
9578 * doc/tm.texi: Same.
9579 * doc/tm.texi.in: Same.
9580 * doc/trouble.texi: Same.
9582 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
9585 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
9586 (collect_checksum_attributes): Set it.
9587 (die_checksum_ordered): Use it.
9589 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9591 PR tree-optimization/79908
9592 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
9593 change: For a VA_ARG whose LHS has been cast away, use
9594 force_gimple_operand to construct the side effects.
9596 2017-03-21 David Malcolm <dmalcolm@redhat.com>
9598 PR translation/80001
9599 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
9600 more amenable to translation.
9601 (oacc_loop_auto_partitions): Likewise.
9603 2017-03-21 Marek Polacek <polacek@redhat.com>
9604 Martin Sebor <msebor@redhat.com>
9606 PR tree-optimization/80109
9607 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
9610 2017-03-21 Jakub Jelinek <jakub@redhat.com>
9611 Segher Boessenkool <segher@kernel.crashing.org>
9614 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
9615 check reg_used_between_p between insn and one of succ or succ2
9616 depending on if succ is artificial insn not inserted into insn
9619 2017-03-21 Martin Liska <mliska@suse.cz>
9621 PR gcov-profile/80081
9622 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
9623 * doc/gcc.texi: Include gcov-dump stuff.
9624 * doc/gcov-dump.texi: New file.
9626 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
9628 PR rtl-optimization/79150
9629 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
9630 conditional jump, if the jump is the last insn of the loop.
9632 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9633 Richard Biener <rguenth@suse.de>
9635 PR tree-optimization/79908
9636 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
9637 been cast away, use force_gimple_operand to construct the side
9640 2017-03-21 Martin Liska <mliska@suse.cz>
9643 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
9646 2017-03-21 Brad Spengler <spender@grsecurity.net>
9649 * plugin.c (htab_hash_plugin): New function.
9650 (add_new_plugin): Use it and adjust.
9651 (parse_plugin_arg_opt): Adjust.
9652 (init_one_plugin): Likewise.
9654 2017-03-21 Richard Biener <rguenther@suse.de>
9656 PR tree-optimization/80032
9657 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
9658 if set force the cleanup to happen unconditionally.
9659 (gimplify_target_expr): Push inserted clobbers with force_uncond
9660 to avoid them being removed by control-dependent DCE.
9662 2017-03-21 Richard Biener <rguenther@suse.de>
9664 PR tree-optimization/80122
9665 * tree-inline.c (copy_bb): Do not expans va-arg packs or
9666 va_arg_pack_len when the inlined call stmt requires pack
9668 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
9670 2017-03-21 Jakub Jelinek <jakub@redhat.com>
9673 * tsan.c (instrument_builtin_call): If the memory model argument
9674 is not a constant, assume it is valid.
9677 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
9680 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
9682 PR rtl-optimization/79910
9683 * combine.c (can_combine_p): Do not allow combining an I0 or I1
9684 if its dest is used by an insn before I2 (other than the combined
9685 insns themselves, which are properly handled already).
9687 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
9690 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
9692 * combine.c (record_used_regs): New static function.
9693 (try_combine): Handle situations where there is an additional
9694 instruction between I2 and I3 which needs to have a LOG_LINK
9698 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
9700 * combine.c (try_combine): Delete redundant i1 test. Call
9701 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
9703 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9706 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
9709 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9711 PR tree-optimization/80054
9712 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
9713 the optimization if a PHI or any of its arguments is not dominated
9714 by the candidate's basis. Use gphi* rather than gimple* as
9716 (replace_profitable_candidates): Clean up a gimple* variable that
9717 should be a gphi* variable.
9719 2017-03-20 Martin Sebor <msebor@redhat.com>
9722 * doc/extend.texi (attribute constructor): Document present limitation.
9724 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
9727 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
9728 __POWER9_VECTOR__ #ifdef control, change template definition to
9729 use Power9-specific built-in function.
9730 (vec_any_eq): Likewise.
9731 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
9732 to control outcomes from this test.
9733 (vector_ae_<mode>p): For VEC_F modes, likewise.
9735 2017-03-20 Ian Lance Taylor <iant@google.com>
9737 * config/i386/i386.c (ix86_function_regparm): Save an extra
9738 register for -fsplit-stack with DECL_STATIC_CHAIN.
9740 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
9743 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
9744 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
9746 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
9748 * config/riscv/riscv.c (riscv_print_operand): Use "fence
9750 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
9753 2017-03-20 Marek Polacek <polacek@redhat.com>
9756 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
9758 2017-03-20 Richard Biener <rguenther@suse.de>
9760 PR tree-optimization/80113
9761 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
9762 allocate extra SSA name for PHI def.
9763 (add_close_phis_to_outer_loops): Likewise.
9764 (add_close_phis_to_merge_points): Likewise.
9765 (copy_loop_close_phi_args): Likewise.
9766 (copy_cond_phi_nodes): Likewise.
9768 2017-03-20 Martin Liska <mliska@suse.cz>
9771 * tree-chkp.c (chkp_build_returned_bound): Do not build
9772 returned bounds for a LHS that's not a BOUNDED_P type.
9774 2017-03-20 Martin Liska <mliska@suse.cz>
9778 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
9779 COMPLEX_CST and VECTOR_CST.
9781 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9784 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
9785 target operand. A new splitter adds the clobber statement in case
9786 the target operand is dead anyway.
9788 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
9790 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
9791 to age-old versions of binutils and glibc.
9793 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
9795 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
9797 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
9799 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
9801 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
9803 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
9804 requirement for binutils 2.13.
9806 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
9808 * combine.c (try_combine): Delete redundant i1 test. Call
9809 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
9811 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
9813 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
9814 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
9816 <riscv64-*-elf>: Re-arrange section
9817 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
9818 <riscv32-*-linux>: Likewise.
9819 <riscv64-*-elf>: Likewise
9820 <riscv64-*-linux>: Likewise.
9822 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
9825 * aarch64.opt(verbose-cost-dump): Fix typo.
9827 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
9830 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
9831 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
9833 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
9835 * reload.c (find_reloads): When reloading a nonoffsettable address,
9836 use RELOAD_OTHER for it and its address reloads.
9838 PR rtl-optimization/79910
9839 * combine.c (record_used_regs): New static function.
9840 (try_combine): Handle situations where there is an additional
9841 instruction between I2 and I3 which needs to have a LOG_LINK
9844 2017-03-17 Jeff Law <law@redhat.com>
9846 PR tree-optimization/71437
9847 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
9848 conditional in the hash table first.
9849 (vrp_dom_walker::before_dom_children): Extract condition from
9850 ASSERT_EXPR. Record condition, its inverion and any implied
9853 2017-03-17 Marek Polacek <polacek@redhat.com>
9854 Markus Trippelsdorf <markus@trippelsdorf.de>
9856 PR tree-optimization/80079
9857 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
9860 2017-03-17 Richard Biener <rguenther@suse.de>
9863 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
9864 Properly verify the LHS before the RHS possibly claims to be
9866 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
9869 2017-03-17 Martin Jambor <mjambor@suse.cz>
9871 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
9872 (List of -O2 options): Likewise.
9873 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
9876 2017-03-17 Tom de Vries <tom@codesourcery.com>
9878 * gcov-dump.c (print_usage): Print bug_report_url.
9880 2017-03-17 Richard Biener <rguenther@suse.de>
9883 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
9884 (parser::peek): Likewise.
9886 2017-03-17 Richard Biener <rguenther@suse.de>
9888 PR tree-optimization/80048
9889 * sese.c (free_sese_info): Properly release rename_map and
9890 copied_bb_map elements.
9892 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
9894 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
9895 Add linked-list forward and backlinks. Insert on
9896 construction, remove on destruction.
9897 (class pass_store_merging): Add m_stores_head field.
9898 (pass_store_merging::terminate_and_process_all_chains):
9899 Iterate over m_stores_head list.
9900 (pass_store_merging::terminate_all_aliasing_chains):
9902 (pass_store_merging::execute): Check for debug stmts first.
9903 Push new chains onto the m_stores_head stack.
9905 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
9908 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
9909 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
9910 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
9912 2017-03-16 Jeff Law <law@redhat.com>
9914 PR tree-optimization/71437
9915 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
9916 member function. Implementation moved into after_dom_children
9917 member function and into the threader's thread_outgoing_edges
9919 (dom_opt_dom_walker::after_dom_children): Simplify by moving
9920 some code into new thread_outgoing_edges.
9921 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
9922 definition. Simplify marker handling (do it here). Assume we always
9923 have the available expression and the const/copies tables.
9924 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
9926 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
9927 * tree-vrp.c (equiv_stack): No longer file scoped.
9928 (vrp_dom_walker): New class.
9929 (vrp_dom_walker::before_dom_children): New member function.
9930 (vrp_dom_walker::after_dom_children): Likewise.
9931 (identify_jump_threads): Setup domwalker. Use it rather than
9932 walking edges in a random order by hand. Simplify setup/finalization.
9933 (finalize_jump_threads): Remove.
9934 (vrp_finalize): Do not call identify_jump_threads here.
9935 (execute_vrp): Do it here instead and call thread_through_all_blocks
9938 PR tree-optimization/71437
9939 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
9941 (simplify_stmt_for_jump_threading): Add basic_block argument. All
9943 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
9944 (dom_opt_dom_walker::thread_across_edge): Remove
9945 handle_dominating_asserts argument. All callers changed.
9946 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
9947 changes. Remove calls to lhs_of_dominating_assert. Other
9948 uses of handle_dominating_asserts turn into unconditional code
9949 (simplify_control_stmt_condition_1): Likewise.
9950 (simplify_control_stmt_condition): Likewise.
9951 (thread_through_normal_block, thread_across_edge): Likewise.
9952 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
9953 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
9954 object if it is not an SSA_NAME.
9955 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
9956 before calling into the VRP specific simplifiers.
9957 (identify_jump_threads): Remove handle_dominating_asserts
9960 2017-03-16 Jakub Jelinek <jakub@redhat.com>
9963 * tree-diagnostic.c (default_tree_printer): No longer static.
9964 * tree-diagnostic.h (default_tree_printer): New prototype.
9966 2017-03-16 Tamar Christina <tamar.christina@arm.com>
9968 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
9969 Change ins into fmov.
9971 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9973 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
9974 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
9975 Use h_con constraint for operand 1.
9976 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
9977 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
9979 2017-03-15 Jeff Law <law@redhat.com>
9981 PR tree-optimization/71437
9982 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
9983 (record_temporary_equivalences): Use it.
9985 PR tree-optimization/71437
9986 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
9987 tree-ssa-scopedtables.
9988 (lookup_avail_expr, build_and_record_new_cond): Likewise.
9989 (record_conditions, record_cond, vuse_eq): Likewise.
9990 (record_edge_info): Adjust to API tweak of record_conditions.
9991 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
9992 (record_temporary_equivalences, optimize_stmt): Likewise.
9993 (eliminate_redundant_computations): Likewise.
9994 (record_equivalences_from_stmt): Likewise.
9995 * tree-ssa-scopedtables.c: Include options.h and params.h.
9996 (vuse_eq): New function, moved from tree-ssa-dom.c
9997 (build_and_record_new_cond): Likewise.
9998 (record_conditions): Likewise. Accept vector of conditions rather
9999 than edge_equivalence structure for first argument.
10000 for the first argument.
10001 (avail_exprs_stack::lookup_avail_expr): New member function, moved
10002 from tree-ssa-dom.c.
10003 (avail_exprs_stack::record_cond): Likewise.
10004 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
10005 from tree-ssa-dom.c.
10006 (avail_exprs_stack): Add new member functions lookup_avail_expr
10008 (record_conditions): Declare.
10010 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
10013 * lra-constraints.c (process_alt_operands): Increase reject for
10014 reloading an input/output operand.
10016 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
10019 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
10020 insns to convert from signed/unsigned char/short to IEEE 128-bit
10022 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
10024 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
10027 * config/i386/i386.c (ix86_vector_duplicate_value): Create
10028 subreg of inner mode for values already in registers.
10030 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
10032 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
10033 iteration reg is used after the loop.
10035 2017-03-14 Martin Sebor <msebor@redhat.com>
10037 PR tree-optimization/79800
10038 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
10039 precision in negative-positive range.
10040 (format_floating): Call non-const overload with adjusted precision.
10042 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
10045 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
10048 2017-03-14 Martin Sebor <msebor@redhat.com>
10050 PR middle-end/80020
10051 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
10052 * builtins.def (aligned_alloc): Use it.
10055 * Makefile.in (GTFILES): Add calls.c.
10056 * calls.c: Include "gt-calls.h".
10058 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
10060 PR rtl-optimization/79728
10061 * regs.h (struct target_regs): New field
10062 x_contains_allocatable_regs_of_mode.
10063 (contains_allocatable_regs_of_mode): New macro.
10064 * reginfo.c (init_reg_sets_1): Initialize it, and change
10065 contains_reg_of_mode so it includes global regs as well.
10066 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
10067 rather than contains_regs_of_mode.
10069 2017-03-14 Martin Liska <mliska@suse.cz>
10071 * doc/invoke.texi: Document options that can't be combined with
10072 -fcheck-pointer-bounds.
10074 2017-03-14 Martin Liska <mliska@suse.cz>
10076 PR middle-end/79831
10077 * doc/invoke.texi (-Wchkp): Document the option.
10079 2017-03-14 Martin Liska <mliska@suse.cz>
10081 * Makefile.in: Install gcov-dump.
10083 2017-03-14 Martin Liska <mliska@suse.cz>
10085 * multiple_target.c (expand_target_clones): Bail out for
10086 an invalid attribute.
10088 2017-03-14 Richard Biener <rguenther@suse.de>
10090 * alias.c (struct alias_set_entry): Pack properly.
10091 * cfgloop.h (struct loop): Likewise.
10092 * cse.c (struct set): Likewise.
10093 * ipa-utils.c (struct searchc_env): Likewise.
10094 * loop-invariant.c (struct invariant): Likewise.
10095 * lra-remat.c (struct cand): Likewise.
10096 * recog.c (struct change_t): Likewise.
10097 * rtl.h (struct address_info): Likewise.
10098 * symbol-summary.h (function_summary): Likewise.
10099 * tree-loop-distribution.c (struct partition): Likewise.
10100 * tree-object-size.c (struct object_size_info): Likewise.
10101 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
10102 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
10103 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
10104 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
10105 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
10106 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
10107 (struct _stmt_vec_info): Likewise.
10109 2017-03-14 Martin Liska <mliska@suse.cz>
10112 * multiple_target.c (create_dispatcher_calls): Check that
10113 a target can create a function dispatcher.
10115 2017-03-14 Martin Liska <mliska@suse.cz>
10118 * multiple_target.c (expand_target_clones): Drop local.local
10119 flag for default implementation.
10121 2017-03-14 Richard Biener <rguenther@suse.de>
10123 PR tree-optimization/80030
10124 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
10126 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
10128 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
10129 gcc_fallthrough() instead of __attribute__((fallthrough));
10131 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
10133 * doc/gcc.texi: Remove "up" link to (DIR).
10134 * doc/gccint.texi: Ditto.
10136 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
10138 * doc/install.texi (Specific) <avr>: Remove reference to
10141 2017-03-13 Jeff Law <law@redhat.com>
10143 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
10144 attribute rather than comments.
10146 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
10147 match_scratch operand is highest.
10149 2017-03-13 Martin Liska <mliska@suse.cz>
10151 PR middle-end/78339
10152 * ipa-pure-const.c (warn_function_noreturn): If the declarations
10153 is a CHKP clone, use original declaration.
10155 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
10157 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
10158 (arc_conditional_register_usage): Use a different allocation order
10159 when optimizing for size.
10160 * common/config/arc/arc-common.c (arc_option_optimization_table):
10161 Section anchors default on when optimizing for size.
10163 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
10165 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
10167 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
10169 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
10170 * config/arc/arc.md (cpu_facility): Add cd variant.
10171 (*movqi_insn): Add code density variant.
10172 (*movhi_insn): Likewise.
10173 (*movqi_insn): Likewise.
10174 (*addsi3_mixed): Likewise.
10175 (subsi3_insn): Likewise.
10177 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
10179 * config/arc/arc.md (movsi_cond_exec): Update constraint.
10181 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
10183 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
10184 expressions with MINUS and UNARY ops.
10186 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10189 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
10191 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
10192 between vec_select and vector argument.
10193 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
10194 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
10195 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
10196 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
10197 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
10198 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
10200 2017-03-13 Richard Biener <rguenther@suse.de>
10203 * params.def (vect-max-peeling-for-alignment): Fix typo.
10205 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
10207 * doc/install.texi (Specific) <mips-*-*>: Remove description of
10208 issue that only occurred with binutils below 2.18.
10210 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
10212 * doc/install.texi (Specific) <cris-axis-elf>: No longer
10213 refer to binutils 2.11/2.12 minimum.
10215 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
10217 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
10218 ftp.kernel.org and simplify binutils requirement.
10220 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
10222 * doc/invoke.texi (Warning Options): Fix spelling of link-time
10224 (Optimize Options): Ditto. Also remove redundancy.
10226 2017-03-10 David Malcolm <dmalcolm@redhat.com>
10228 PR translation/79848
10229 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
10231 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
10232 to G_ to avoid double translation.
10234 2017-03-10 David Malcolm <dmalcolm@redhat.com>
10236 PR translation/79923
10237 * auto-profile.c (get_combined_location): Convert leading
10238 character of diagnostics to lower case and remove trailing period.
10239 (read_profile): Likewise for various diagnostics.
10240 * config/arm/arm.c (arm_option_override): Remove trailing period
10241 from various diagnostics.
10242 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
10243 (msp430_expand_delay_cycles): Likewise.
10245 2017-03-10 David Malcolm <dmalcolm@redhat.com>
10248 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
10249 full command-line argument, rather than just "str".
10250 (aarch64_validate_march): Likewise.
10251 (aarch64_validate_mtune): Likewise.
10253 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
10255 PR rtl-optimization/78911
10256 * lra-assigns.c (must_not_spill_p): New function.
10257 (spill_for): Use it.
10259 2017-03-10 Jakub Jelinek <jakub@redhat.com>
10261 PR tree-optimization/79981
10262 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
10263 ATOMIC_COMPARE_EXCHANGE ifn result.
10264 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
10265 IFN_ATOMIC_COMPARE_EXCHANGE.
10267 2017-03-10 David Malcolm <dmalcolm@redhat.com>
10270 * opts.c (parse_sanitizer_options): Add missing question mark to
10271 "did you mean" message.
10273 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10275 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
10277 (VMULEUH_UNS): Likewise.
10278 (VMULOUB_UNS): Likewise.
10279 (VMULOUH_UNS): Likewise.
10280 * config/rs6000/rs6000.c (builtin_function_type): Remove
10281 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
10283 2017-03-10 David Malcolm <dmalcolm@redhat.com>
10286 * read-rtl-function.c (function_reader::read_rtx_operand): Update
10287 x with result of extra_parsing_for_operand_code_0.
10288 (function_reader::extra_parsing_for_operand_code_0): Convert
10289 return type from void to rtx, returning x. When reading
10290 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
10291 larger size containing struct block_symbol.
10293 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
10295 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
10296 -mfloat128-hardware without -m64.
10298 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
10301 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
10302 entries to the case statement that marks unsigned arguments to
10303 overloaded functions.
10305 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
10307 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
10308 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
10310 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
10313 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
10314 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
10316 2017-03-10 Martin Liska <mliska@suse.cz>
10320 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
10321 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
10324 2017-03-10 Olivier Hainque <hainque@adacore.com>
10326 * tree-switch-conversion (array_value_type): Start by resetting
10327 candidate type to it's main variant.
10329 2017-03-10 Jakub Jelinek <jakub@redhat.com>
10331 PR rtl-optimization/79909
10332 * combine.c (try_combine): Use simplify_replace_rtx on individual
10333 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
10334 of the whole CALL_INSN_FUNCTION_USAGE.
10336 PR tree-optimization/79972
10337 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
10338 get_range_info on SSA_NAMEs. Formatting fixes.
10340 2017-03-10 Richard Biener <rguenther@suse.de>
10341 Jakub Jelinek <jakub@redhat.com>
10343 PR tree-optimization/77975
10344 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
10345 edge to be constant.
10346 (get_val_for): For constant x return it. Formatting fix.
10347 (loop_niter_by_eval): Avoid pointless looping if the next iteration
10348 would use the same bases as the current one.
10350 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10352 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
10353 instead of vec_select for V1TImode.
10354 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
10356 (VSX_LE_128): Add V1TI to this mode iterator.
10357 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
10358 (*vsx_le_perm_store_<mode>): Likewise.
10359 (pre-reload splitter for VSX stores): Likewise.
10360 (post-reload splitter for VSX stores): Likewise.
10361 (*vsx_xxpermdi2_le_<mode>): Likewise.
10362 (*vsx_lxvd2x2_le_<mode>): Likewise.
10363 (*vsx_stxvd2x2_le_<mode>): Likewise.
10365 2017-03-09 Michael Eager <eager@eagercon.com>
10367 Correct failures with --enable-checking=yes,rtl.
10369 * config/microblaze/microblaze.c (microblaze_expand_shift):
10370 Replace GET_CODE test with CONST_INT_P and INTVAL test with
10371 test for const0_rtx.
10372 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
10373 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
10375 2017-03-09 Richard Biener <rguenther@suse.de>
10377 PR tree-optimization/79977
10378 * graphite-scop-detection.c (scop_detection::merge_sese):
10379 Handle the case of extra exits to blocks dominating the entry.
10381 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
10383 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
10386 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
10388 PR rtl-optimization/79949
10389 * lra-constraints.c (process_alt_operands): Check memory when
10390 trying to predict a cycle. Print about the overall increase.
10392 2017-03-09 Richard Biener <rguenther@suse.de>
10394 PR middle-end/79971
10395 * gimple-expr.c (useless_type_conversion_p): Preserve
10396 TYPE_SATURATING for fixed-point types.
10398 2017-03-09 Richard Biener <rguenther@suse.de>
10401 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
10402 alignment of BLKmode params.
10404 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10407 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
10408 (VALL_NO_V2Q): Likewise.
10410 * config/aarch64/aarch64-simd.md
10411 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
10413 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
10414 VALL_NO_V2Q mode iterator.
10415 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
10417 2017-03-09 Martin Liska <mliska@suse.cz>
10419 PR tree-optimization/79631
10420 * tree-chkp-opt.c (chkp_is_constant_addr): Call
10421 tree_int_cst_sign_bit just for INTEGER constants.
10423 2017-03-09 Martin Liska <mliska@suse.cz>
10427 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
10430 2017-03-09 Marek Polacek <polacek@redhat.com>
10433 * tree.c (inchash::add_expr): Handle TREE_VEC.
10435 2017-03-09 Martin Liska <mliska@suse.cz>
10438 (chkp_narrow_size_and_offset): New function.
10439 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
10440 (void chkp_parse_bit_field_ref): New function.
10441 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
10442 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
10444 2017-03-09 Martin Liska <mliska@suse.cz>
10447 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
10448 (chkp_find_bounds_1): Remove gcc_unreachable.
10450 2017-03-09 Jakub Jelinek <jakub@redhat.com>
10453 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
10454 BUILT_IN_SYNC*, determine the access type from the size suffix and
10455 always build a MEM_REF with that type. Handle forgotten
10456 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
10459 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
10460 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
10461 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
10462 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
10463 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
10464 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
10465 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
10466 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
10467 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
10468 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
10469 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
10470 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
10471 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
10472 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
10473 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
10474 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
10475 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
10476 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
10477 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
10478 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
10479 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
10480 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
10481 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
10482 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
10483 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
10484 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
10485 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
10486 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
10487 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
10488 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
10489 definitions outside of __OPTIMIZE__ guarded section.
10492 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
10493 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
10494 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
10495 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
10498 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10500 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
10501 ("vfenez<mode>"): Add missing constraints.
10503 2017-03-08 Martin Sebor <msebor@redhat.com>
10506 * config/nds32/nds32.c (nds32_option_override):
10507 Fix misspelled diagnostic.
10509 2017-03-08 Jakub Jelinek <jakub@redhat.com>
10512 * gimplify.c (gimplify_omp_for): Replace index var in outer
10513 taskloop statement with an artificial variable and add
10514 OMP_CLAUSE_PRIVATE clause for it.
10516 2017-03-08 Richard Biener <rguenther@suse.de>
10518 PR tree-optimization/79955
10519 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
10520 for accesses that are completely outside of the variable.
10522 2017-03-08 Andrew Haley <aph@redhat.com>
10524 PR tree-optimization/79943
10525 * tree-ssa-loop-split.c (compute_new_first_bound): When
10526 calculating the new upper bound, (END-BEG) should be added, not
10529 2017-03-08 Jakub Jelinek <jakub@redhat.com>
10531 * config/avr/avr.md (setmemhi): Make sure match_dup
10532 operand number comes before match_scratch.
10534 2017-03-08 Richard Biener <rguenther@suse.de>
10536 PR tree-optimization/79920
10537 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
10538 with ncopies == 1 to ...
10539 (vect_transform_slp_perm_load): ... here. Properly compute
10540 all element loads by iterating VF times over the group. Do
10541 not handle ncopies (computed in a broken way) in
10542 vect_create_mask_and_perm.
10544 2017-03-08 Jakub Jelinek <jakub@redhat.com>
10547 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
10548 is a uniform vector, use uniform_vector_p return value instead of
10549 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
10551 2017-03-07 Marek Polacek <polacek@redhat.com>
10553 PR middle-end/79809
10554 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
10555 (alloca_call_type): Likewise.
10557 2017-03-07 Martin Liska <mliska@suse.cz>
10559 * gcov.c (process_args): Put comment to correct location.
10561 2017-03-07 Martin Liska <mliska@suse.cz>
10563 PR middle-end/68270
10564 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
10565 Use array_at_struct_end_p instead of DECL_CHAIN (field).
10566 (chkp_narrow_bounds_for_field): Likewise.
10567 (chkp_parse_array_and_component_ref): Pass one more argument to
10570 2017-03-07 Richard Biener <rguenther@suse.de>
10572 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
10575 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
10577 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
10578 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
10580 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10583 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
10584 to end of description.
10585 (PARAM_MAX_STORES_TO_MERGE): Likewise.
10587 2017-03-07 Jakub Jelinek <jakub@redhat.com>
10589 PR rtl-optimization/79901
10590 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
10592 (*avx512f_<code><mode>3<mask_name>): ... this.
10593 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
10594 iterator instead of VI8_AVX2_AVX512BW.
10596 PR rtl-optimization/79901
10597 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
10598 min/max expander, expand it using expand_vec_cond_expr.
10601 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
10604 2017-03-06 Jakub Jelinek <jakub@redhat.com>
10607 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
10608 to void * for PCH reasons.
10609 * dwarf2out.c (output_loc_operands, output_die): Cast
10610 v.val_vec.array to unsigned char *.
10612 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
10615 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
10618 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
10620 PR rtl-optimization/79571
10621 * lra-constraints.c (process_alt_operands): Calculate static
10622 reject and subtract it from overall when only addresses will be
10625 2017-03-06 Julia Koval <julia.koval@intel.com>
10628 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
10629 incoming stack boundary to 128 for 64-bit targets.
10631 2017-03-06 Richard Biener <rguenther@suse.de>
10633 PR tree-optimization/79894
10634 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
10635 to NULL after folding it.
10637 2017-03-06 Richard Biener <rguenther@suse.de>
10639 PR tree-optimization/79824
10640 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
10641 check disabling peeling for gaps.
10643 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
10645 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
10646 attributes): Document gettimeofday.
10648 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
10650 * config/s390/s390.c (s390_option_override_internal): Set
10651 PARAM_MIN_VECT_LOOP_BOUND
10653 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
10655 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
10656 * config/s390/s390.md: Likewise.
10658 2017-03-06 Jakub Jelinek <jakub@redhat.com>
10661 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
10662 (<avx2_avx512>_perm<mode>): Rename to ...
10663 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
10665 (<avx512>_perm<mode>_mask): Rename to ...
10666 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
10668 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
10669 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
10670 instead of VI8F_256_512.
10671 (avx512f_perm<mode>): New define_expand.
10672 (avx512f_perm<mode>_mask): Likewise.
10673 (avx512f_perm<mode>_1<mask_name>): New define_insn.
10674 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
10676 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
10678 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
10679 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
10681 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
10683 2017-03-06 Martin Liska <mliska@suse.cz>
10686 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
10687 when having a SSA NAME w/o VAR_DECL assigned to it.
10689 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
10691 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
10692 msa_dpsub_<su>_d): Fix MODE for vec_select.
10694 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
10696 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
10698 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
10700 2017-03-06 Richard Biener <rguenther@suse.de>
10702 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
10703 * plugin.c (register_plugin_info): Likewise.
10704 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
10706 2017-03-05 Jakub Jelinek <jakub@redhat.com>
10708 * config/i386/sse.md (sse_storehps, sse_storelps,
10709 avx_<castmode><avxsizesuffix>_<castmode>,
10710 avx512f_<castmode><avxsizesuffix>_<castmode>,
10711 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
10712 in condition that at least one operand is not a MEM.
10714 2017-03-03 Jakub Jelinek <jakub@redhat.com>
10716 PR middle-end/79805
10717 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
10718 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
10720 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
10721 gimple_call_nothrow_p flag based on whether original builtin can throw.
10722 If it can, emit following stmts on the fallthrough edge.
10723 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
10724 don't create new bb if inserting just debug stmts on the edge, try to
10725 insert them on the fallthru bb or just reset debug stmts.
10727 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
10730 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
10731 restore recog_data (including the operand rtxes inside it) around
10732 the call to get_insn_template.
10734 2017-03-03 Martin Sebor <msebor@redhat.com>
10736 PR tree-optimization/79699
10737 * context.c (context::~context): Free MPFR caches to avoid
10738 a memory leak on program exit.
10740 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10742 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
10743 Use wide_int::ulow () instead of .elt (0).
10745 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
10747 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
10748 (*pushxf): Limit oF constraint to 32bit targets and add oC
10749 constraint for 64bit targets.
10750 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
10751 (*pushdf): Change rmF constraint to rmC.
10753 2017-03-03 Martin Liska <mliska@suse.cz>
10755 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
10756 Remove unused variable.
10758 2017-03-03 Jakub Jelinek <jakub@redhat.com>
10761 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
10762 is a memory operand, increase num_memory.
10763 (ix86_expand_args_builtin): Likewise.
10765 2017-03-03 Jan Hubicka <jh@suse.cz>
10768 * ipa-devirt.c (maybe_record_node): Properly handle
10769 __cxa_pure_virtual visibility.
10771 2017-03-03 Martin Liska <mliska@suse.cz>
10773 PR tree-optimization/79803
10774 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
10776 (pass_loop_prefetch::execute): Disabled optimization if an
10777 assumption about L1 cache size is not met.
10779 2017-03-03 Martin Liska <mliska@suse.cz>
10781 PR rtl-optimization/79574
10782 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
10783 (hash_scan_set): Likewise.
10784 (dump_hash_table): Likewise.
10785 (hoist_code): Likewise.
10787 2017-03-03 Richard Biener <rguenther@suse.de>
10789 * fixed-value.c (fixed_from_string): Restore use of elt (1)
10790 in place of uhigh ().
10791 (fixed_convert_from_real): Likewise.
10793 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
10796 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
10798 2017-03-03 Richard Biener <rguenther@suse.de>
10800 PR middle-end/79818
10801 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
10802 TYPE_OVERFLOW_UNDEFINED check.
10804 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10806 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
10808 (vector_ae_<mode>_p): Likewise.
10809 (vector_nez_<mode>_p): Likewise.
10810 (vector_ne_v2di_p): Likewise.
10811 (vector_ae_v2di_p): Likewise.
10812 (vector_ne_<mode>_p): Likewise.
10813 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
10815 (vsx_tsqrt<mode>2_fe): Likewise.
10817 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
10820 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
10822 2017-03-02 Jakub Jelinek <jakub@redhat.com>
10824 PR rtl-optimization/79780
10825 * cprop.c (one_cprop_pass): When second and further conditional trap
10826 in a single basic block is turned into an unconditional trap, turn it
10827 into a deleted note to avoid RTL verification failures.
10829 2017-03-02 Richard Biener <rguenther@suse.de>
10831 * fold-const.c (const_binop): Use ulow () instead of elt (0).
10833 2017-03-02 Richard Biener <rguenther@suse.de>
10835 PR tree-optimization/79345
10837 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
10838 param and abort the walk, returning -1 if it is hit.
10839 (walk_aliased_vdefs): Take a limit param and pass it on.
10840 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
10841 defaulting to 0 and return a signed int.
10842 * tree-ssa-uninit.c (struct check_defs_data): New struct.
10843 (check_defs): New helper.
10844 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
10845 about uninitialized memory.
10846 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
10847 bogus uninitialized warning.
10848 (fixed_convert_from_real): Likewise.
10850 2017-03-02 Bin Cheng <bin.cheng@arm.com>
10852 PR tree-optimization/66768
10853 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
10854 iv_use if base object can't be determined.
10856 2017-03-02 Jakub Jelinek <jakub@redhat.com>
10858 PR tree-optimization/79345
10859 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
10860 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
10861 (get_pattern_stats): Initialize it.
10862 * genemit.c (gen_expand): Verify match_scratch numbers come after
10863 match_operand/match_dup numbers.
10864 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
10865 match_scratch numbers.
10866 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
10868 * config/s390/s390.md (trunctdsd2): Likewise.
10870 2017-03-02 Richard Biener <rguenther@suse.de>
10872 * wide-int.h (wide_int_storage::operator=): Implement in terms
10875 2017-03-02 Richard Biener <rguenther@suse.de>
10877 PR tree-optimization/79777
10878 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
10879 the to insert expression to sth existing.
10881 2017-03-01 Martin Sebor <msebor@redhat.com>
10883 PR middle-end/79692
10884 * gimple-ssa-sprintf.c
10885 (directive::known_width_and_precision): New function.
10886 (format_integer): Use it.
10887 (get_mpfr_format_length): Consider the full range of precision
10888 when computing %g output with the # flag. Set the likely byte
10889 count to 3 rather than 1 when precision is indeterminate.
10890 (format_floating): Correct the lower bound of precision.
10892 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10894 * doc/invoke.texi: Document default code model for 64-bit Linux.
10896 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10899 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
10900 udiv rather than div since input pattern is unsigned.
10902 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
10904 * config/i386/i386.c (print_reg): Warn for values of
10905 unsupported size in integer register.
10907 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
10910 * config/rs6000/predicates.md (current_file_function_operand): Do
10911 not allow self calls to be local if the function is replaceable.
10913 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
10916 * config/rs6000/altivec.h (vec_ctz and others): Change the
10917 preprocessor macro that controls conditional compilation from
10918 _ARCH_PWR9 to __POWER9_VECTOR__.
10919 (vec_all_ne): Change parameterization of __altivec_scalar_pred
10920 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
10921 control (instead of _ARCH_PWR9 control) so that template
10922 definition uses power9-specific function.
10923 (vec_any_eq): Likewise.
10924 (vec_all_ne): Change macro definition to use a power9-specific
10925 expansion under #ifdef __POWER9_VECTOR__ control (instead of
10926 _ARCH_PWR9 control).
10927 (vec_any_eq) Likewise.
10928 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
10929 expansion for CMPNEF to remove support for xvcmpnesp instruction.
10930 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
10931 support for xvcmpnedp instruction.
10932 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
10933 macro expansion so that Power9 implementation of vec_all_ne does
10934 not use the AltiVec predicate framework.
10935 (VCMPNEH_P): Likewise.
10936 (VCMPNEW_P): Likewise.
10937 (VCMPNED_P): Likewise.
10938 (VCMPNEFP_P): Likewise.
10939 (VCMPNEDP_P): Likewise.
10940 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
10941 implementation of vec_any_eq to not use AltiVec predicate
10943 (VCMPAEH_P): Likewise.
10944 (VCMPAEW_P): Likewise.
10945 (VCMPAED_P): Likewise.
10946 (VCMPAEFP_P): Likewise.
10947 (VCMPAEDP_P): Likewise.
10948 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
10949 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
10950 not use the AltiVec predicate framework.
10951 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
10952 of vec_any_eq to not use AltiVec predicate framework.
10953 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
10954 support for predefined __POWER9_VECTOR__ macro to indicate that
10955 Power9 instruction selection is enabled.
10956 (altivec_overloaded_builtins): Remove extraneous
10957 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
10958 function argument types RS6000_BTI_bool_V16QI and
10959 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
10960 entry for overloaded function argument types RS6000_BTI_bool_V4SI
10961 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
10962 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
10963 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
10964 Power9 for implementations of vec_cmpne. Change the signature for
10965 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
10966 (representing vec_all_ne) to remove the previously described first
10967 argument of type RS6000_BTI_INTSI, as this was an artifact of
10968 reliance on the AltiVec predicate framework, which is no longer
10969 used in the implementation of these functions. Add
10970 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
10971 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
10972 since, unlike the AltiVec predicate framework implementation, we
10973 do not share function descriptors between vec_alle and vec_anyeq.
10974 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
10975 set of modes that receive special treatment even when
10976 TARGET_P9_VECTOR is true. The special treatment emits code that
10977 does not depend on Power9 instructions.
10978 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
10979 define_expand to not rely on AltiVec predicate framework.
10980 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
10982 (vector_ne_v2di_p): Change this define_expand to not rely on
10983 AltiVec predicate framework.
10984 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
10986 (vector_ne_<mode>_p): Change this define_expand to not rely on
10987 AltiVec predicate framework.
10988 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
10990 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
10991 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
10992 define_insn pattern.
10993 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
10994 define_insn pattern because the xvcmpne<VSs>. instruction is not
10996 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
10997 instruction is not supported.
10999 2017-03-01 Jakub Jelinek <jakub@redhat.com>
11001 * config/nvptx/nvptx.c: Include intl.h.
11003 2017-03-01 Martin Jambor <mjambor@suse.cz>
11006 * ipa-prop.h (ipa_bits): Removed field known.
11007 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
11008 to pointers. Adjusted their comments to warn about their sharing.
11009 (ipcp_transformation_summary): Change bits to a vector of pointers.
11010 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
11011 (ipa_get_ipa_bits_for_value): Declare.
11012 * tree-vrp.h (value_range): Mark as GTY((for_user)).
11013 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
11014 (ipa_bits_hash_table): Likewise.
11015 (ipa_vr_ggc_hash_traits): Likewise.
11016 (ipa_vr_hash_table): Likewise.
11017 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
11018 being pointers and vr_known being removed.
11019 (ipa_set_jf_unknown): Likewise.
11020 (ipa_get_ipa_bits_for_value): New function.
11021 (ipa_set_jfunc_bits): Likewise.
11022 (ipa_get_value_range): New overloaded functions.
11023 (ipa_set_jfunc_vr): Likewise.
11024 (ipa_compute_jump_functions_for_edge): Use the above functions to
11025 construct bits and vr parts of jump functions.
11026 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
11027 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
11029 (ipcp_grow_transformations_if_necessary): Also allocate
11030 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
11032 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
11033 them. Fix too long lines.
11034 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
11035 vr_known being removed.
11036 (ipa_read_jump_function): Use new setter functions to construct bits
11037 and vr parts of jump functions or set them to NULL.
11038 (write_ipcp_transformation_info): Adjust for bits being pointers.
11039 (read_ipcp_transformation_info): Likewise.
11040 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
11042 Include gt-ipa-prop.h.
11043 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
11045 (ipcp_store_bits_results): Likewise.
11046 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
11047 Do not write to existing jump functions but use a temporary instead.
11049 2017-03-01 Jakub Jelinek <jakub@redhat.com>
11052 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
11053 attempt to use its first operand as BIT_FIELD_REF base.
11055 2017-03-01 Richard Biener <rguenther@suse.de>
11057 PR middle-end/79721
11058 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
11059 interpolating formula in wrapping arithmetic.
11060 (chrec_apply): Convert chrec_evaluate return value to wanted type.
11062 2017-03-01 Jakub Jelinek <jakub@redhat.com>
11064 PR tree-optimization/79734
11065 * tree-vect-generic.c (expand_vector_condition): Optimize
11066 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
11067 Handle VEC_COND_EXPR where comparison has different inner width from
11068 type's inner width.
11070 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
11072 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
11073 markup, and similar issues. Remove @opindex entries for things
11074 that aren't options. Add missing -mmpy-option entries.
11076 2017-02-28 Jakub Jelinek <jakub@redhat.com>
11078 PR tree-optimization/79737
11079 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
11080 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
11081 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
11082 instead of byte_size. Formatting fix.
11083 (shift_bytes_in_array_right): Formatting fix.
11085 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
11088 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
11089 condition on optimize for the leaf function test.
11091 2017-02-28 Martin Liska <mliska@suse.cz>
11094 * read-rtl-function.c (function_reader::handle_unknown_directive):
11095 Bail out when one uses -flto.
11097 2017-02-28 Martin Liska <mliska@suse.cz>
11099 * common.opt: Replace space with tabular for options of <number>
11101 * config/i386/i386.opt: Show <number> value for
11102 -mlarge-data-threshold.
11103 * opts.c (print_filtered_help): Do not display number in hexadecimal
11106 2017-02-28 Martin Liska <mliska@suse.cz>
11108 * common.opt: Fix --help=option -Q for options which are of
11111 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
11113 * config/i386/i386.c (print_reg): Error out for values
11114 of 8-bit size in invalid integer register.
11116 2017-02-28 Martin Sebor <msebor@redhat.com>
11118 PR tree-optimization/79691
11119 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
11121 2017-02-28 Jakub Jelinek <jakub@redhat.com>
11124 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
11125 gcc_unreachable with output_operand_lossage.
11127 2017-02-28 Richard Biener <rguenther@suse.de>
11129 PR tree-optimization/79740
11130 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
11132 (visit_nary_op): Insert the nary into the hashtable if we
11133 pattern-matched sth.
11134 * tree-ssa-pre.c (eliminate_insert): Robustify.
11136 2017-02-28 Richard Biener <rguenther@suse.de>
11138 PR middle-end/79731
11139 * fold-const.c (decode_field_reference): Reject out-of-bound
11142 2017-02-28 Jakub Jelinek <jakub@redhat.com>
11144 * config/i386/i386.c: Include intl.h.
11145 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
11146 instead of just cond ? "..." : "...".
11147 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
11148 * coverage.c (read_counts_file): Likewise.
11149 * omp-offload.c: Include intl.h.
11150 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
11151 of just cond ? "..." : "...".
11152 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
11153 of just cond ? "..." : "...".
11155 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
11158 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
11160 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
11161 'tune for' CPU name.
11162 * config/arm/arm-cpu-data.h: Regenerated.
11164 2017-02-28 Richard Biener <rguenther@suse.de>
11166 PR tree-optimization/79732
11167 * tree-inline.c (expand_call_inline): Do not shadow var.
11169 2017-02-28 Richard Biener <rguenther@suse.de>
11171 PR tree-optimization/79723
11172 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
11173 address-space properly.
11175 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
11177 * doc/optinfo.texi (Optimization groups): Fix option used for
11179 * doc/invoke.texi (-fopt-info): Document "omp".
11180 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
11181 (OPTGROUP_ALL): Add OPTGROUP_OMP.
11182 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
11183 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
11184 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
11186 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
11188 * dumpfile.c (optgroup_options): Instead of "openmp", associate
11189 OPTGROUP_OMP with "omp".
11191 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
11194 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
11195 for arithmetic shift of unsigned V2DI.
11197 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
11199 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
11200 arc/linux.h headers.
11201 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
11202 (LINK_SPEC): Likewise.
11203 (ARC_TLS_EXTRA_START_SPEC): Likewise.
11204 (EXTRA_SPECS): Likewise.
11205 (STARTFILE_SPEC): Likewise.
11206 (ENDFILE_SPEC): Likewise.
11207 (LIB_SPEC): Likewise.
11208 (TARGET_SDATA_DEFAULT): Likewise.
11209 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
11210 (MULTILIB_DEFAULTS): Likewise.
11211 (DWARF2_UNWIND_INFO): Likewise.
11212 * config/arc/big.h: New file.
11213 * config/arc/elf.h: Likewise.
11214 * config/arc/linux.h: Likewise.
11215 * config/arc/t-uClibc: Remove.
11217 2017-02-27 Bin Cheng <bin.cheng@arm.com>
11219 PR tree-optimization/77536
11220 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
11221 (tree_transform_and_unroll_loop): Use above function to compute the
11222 estimated niter of unrolled loop and use it when scaling profile.
11223 Also use count info rather than frequency if it's non-zero.
11224 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
11225 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
11226 (vect_transform_loop): Call above function.
11228 2017-02-27 Richard Biener <rguenther@suse.de>
11230 PR tree-optimization/45397
11231 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
11232 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
11233 (visit_nary_op): Add pattern matching for CSEing sign-changed
11234 or truncated operations with wider ones.
11236 2017-02-27 Richard Biener <rguenther@suse.de>
11238 PR tree-optimization/79690
11239 * tree-vect-stmts.c (vectorizable_store): Use vector type
11240 built from the DR with address-space.
11242 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
11244 * doc/invoke.texi (Optimize Options): Refine the description
11245 of asan-use-after-return.
11247 2017-02-25 Alan Modra <amodra@gmail.com>
11249 PR rtl-optimization/79584
11250 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
11251 base, not ad->base_term, the reg within base. Remove assertion
11252 that ad->base == ad->base_term. Replace gen_int_mode using
11253 bogus mode with const0_rtx.
11255 2017-02-25 Jakub Jelinek <jakub@redhat.com>
11257 PR middle-end/79396
11258 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
11259 FMA_EXPR like tcc_binary or tcc_unary.
11261 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
11264 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
11266 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
11267 (output_loc_operands): Handle DW_OP_call_ref and
11268 DW_OP_GNU_variable_value.
11269 (struct variable_value_struct): New type.
11270 (struct variable_value_hasher): Likewise.
11271 (variable_value_hash): New variable.
11272 (string_types): Remove.
11273 (copy_loc_descr): New function.
11274 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
11275 (prepend_loc_descr_to_each): New function.
11276 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
11277 instead of add_loc_descr_to_each if the first argument is single
11278 location list and the second has multiple.
11279 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
11280 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
11281 when looking for variable value which doesn't have other location info.
11282 (loc_list_from_tree): Formatting fix.
11283 (gen_array_type_die): Simplify DW_AT_string_length handling.
11284 (adjust_string_types): Remove.
11285 (gen_subprogram_die): Don't call adjust_string_types nor test/set
11286 string_types. Call resolve_variable_values.
11287 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
11288 (resolve_addr_in_expr): Likewise. Add A argument.
11289 (copy_deref_exprloc): Remove deref argument. Adjust for the
11290 original expression being DW_OP_GNU_variable_value with optionally
11291 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
11292 optionally after it.
11293 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
11294 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
11295 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
11296 (variable_value_hasher::hash, variable_value_hasher::equal): New
11298 (resolve_variable_value_in_expr, resolve_variable_value,
11299 resolve_variable_values, note_variable_value_in_expr,
11300 note_variable_value): New functions.
11301 (dwarf2out_early_finish): Call note_variable_value on all toplevel
11304 2017-02-24 Jakub Jelinek <jakub@redhat.com>
11307 * opts.h (handle_generated_option): Add GENERATED_P argument.
11308 * opts-common.c (handle_option): Adjust function comment.
11309 (handle_generated_option): Add GENERATED_P argument, pass it to
11311 (control_warning_option): Pass false to handle_generated_option
11313 * opts.c (maybe_default_option): Pass true to handle_generated_option
11315 * optc-gen.awk: Likewise.
11317 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
11319 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
11320 a REG, look at the REG it is a SUBREG of.
11321 (splitter for cmpeqsi_t): Ditto.
11323 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
11325 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
11326 the special USEs with the pattern of the insn, not the insn itself.
11328 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
11331 * doc/invoke.texi: Document -mload-store-pairs.
11333 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
11334 Sandra Loosemore <sandra@codesourcery.com>
11336 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
11337 argument isn't a CONST_INT.
11338 (nios2_alternate_compare_const): Assert op is a CONST_INT.
11339 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
11340 (nios2_validate_compare): Bypass alternate compare logic if *op2
11341 is not a CONST_INT.
11342 (ldstwm_operation_p): Return false if first_base is not a REG or
11343 if first_offset is not a CONST_INT.
11345 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
11347 * config/cris/cris.md: Use correct operand in a define_peephole2.
11349 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
11351 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
11353 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
11355 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
11356 this_insn if it is an INSN or JUMP_INSN.
11357 (force_offsettable): Look at base, not at addr.
11358 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
11359 on things that aren't necessarily CONST_INTs.
11361 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
11363 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
11364 -mfpmath=sse is the default also for x86-32 targets with SSE2
11365 instruction set when @option{-ffast-math} is enabled
11367 2017-02-24 Jeff Law <law@redhat.com>
11369 PR rtl-optimizatoin/79286
11370 * ira.c (update_equiv_regs): Drop may_trap_p exception to
11373 2017-02-24 Richard Biener <rguenther@suse.de>
11375 PR tree-optimization/79389
11376 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
11379 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
11381 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
11382 function comment to reflect reality.
11383 (loop_exits_before_overflow): Fix typo in function description.
11385 2017-02-24 Richard Biener <rguenther@suse.de>
11387 PR tree-optimization/79389
11388 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
11389 properly that a threading opportunity exists. Detect conditional
11390 copy/constant propagation opportunities.
11392 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
11394 * config/visium/visium.md (type): Add trap.
11395 (b): New mode attribute.
11396 (*btst): Rename into...
11397 (*btst<mode>): ...this and adjust.
11398 (*cbranchsi4_btst_insn): Rename into...
11399 (*cbranch<mode>4_btst_insn): ...this and adjust.
11400 (trap): New define_insn.
11402 2017-02-23 Jakub Jelinek <jakub@redhat.com>
11404 PR tree-optimization/79389
11405 * ifcvt.c (struct noce_if_info): Add rev_cond field.
11406 (noce_reversed_cond_code): New function.
11407 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
11408 reversed_comparison_code. Formatting fix.
11409 (noce_try_store_flag): Test rev_cond != NULL in addition to
11410 reversed_comparison_code.
11411 (noce_try_store_flag_constants): Likewise.
11412 (noce_try_store_flag_mask): Likewise.
11413 (noce_try_addcc): Use rev_cond if non-NULL instead of
11414 reversed_comparison_code.
11415 (noce_try_cmove_arith): Likewise. Formatting fixes.
11416 (noce_try_minmax, noce_try_abs): Clear rev_cond.
11417 (noce_find_if_block): Initialize rev_cond.
11418 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
11419 instead of false as last argument never attempt to reverse it
11422 2017-02-23 Bin Cheng <bin.cheng@arm.com>
11424 PR tree-optimization/79663
11425 * tree-predcom.c (combine_chains): Process refs in reverse order
11426 only for ZERO length chains, and add explaining comment.
11428 2017-02-23 Jeff Law <law@redhat.com>
11430 PR tree-optimization/79578
11431 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
11432 in call to operand_equal_p.
11434 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
11437 * config/i386/cpuid.h: Fix another undefined behavior.
11439 2017-02-23 Richard Biener <rguenther@suse.de>
11441 PR tree-optimization/79683
11442 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
11443 vector types for data-refs.
11445 2017-02-23 Martin Liska <mliska@suse.cz>
11447 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
11449 2017-02-23 Jakub Jelinek <jakub@redhat.com>
11451 PR middle-end/79665
11452 * internal-fn.c (get_range_pos_neg): Moved to ...
11453 * tree.c (get_range_pos_neg): ... here. No longer static.
11454 * tree.h (get_range_pos_neg): New prototype.
11455 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
11456 are known to be in between 0 and signed maximum inclusive, try to
11457 expand both unsigned and signed divmod and use the cheaper one from
11460 2017-02-22 Jeff Law <law@redhat.com>
11462 PR tree-optimization/79578
11463 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
11464 to compare base operands.
11466 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
11469 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
11470 gpc_reg_operand instead of fpr_reg_operand.
11472 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
11474 * config/mips/mips.c (mips_return_in_memory): Force FP
11475 vector types to be returned in memory for o32 ABI.
11477 2017-02-22 Jakub Jelinek <jakub@redhat.com>
11479 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
11480 instead of DW_TAG_member for static data member declarations and don't
11481 set no_linkage_name for static inline data members.
11482 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
11485 2017-02-22 Martin Liska <mliska@suse.cz>
11487 * doc/invoke.texi: Replace inequality signs with square brackets
11490 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11492 PR tree-optimization/68644
11493 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
11495 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
11498 * lra-constraints.c (simplify_operand_subreg): Handle
11499 WORD_REGISTER_OPERATIONS targets.
11501 2017-02-22 Jakub Jelinek <jakub@redhat.com>
11504 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
11505 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
11506 elimination by swapping fld*.
11508 2017-02-22 Richard Biener <rguenther@suse.de>
11510 PR tree-optimization/79673
11511 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
11512 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
11513 irrelevant address-space qualifiers and avoiding a
11514 ADDR_SPACE_CONVERT_EXPR from fold_convert.
11516 2017-02-22 Richard Biener <rguenther@suse.de>
11518 PR tree-optimization/79666
11519 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
11520 to not symbolically negate if that may introduce undefined
11523 2017-02-22 Martin Liska <mliska@suse.cz>
11526 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
11527 * data-streamer-out.c (streamer_write_gcov_count_stream):
11529 * value-prof.c (stream_out_histogram_value): Make assert more
11530 precise based on type of counter.
11532 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
11535 * config/i386/i386.md (standard_x87sse_constant_load splitter):
11536 Use nonimmediate_operand instead of memory_operand for operand 1.
11537 (float-extend standard_x87sse_constant_load splitter): Ditto.
11539 2017-02-21 Jeff Law <law@redhat.com>
11541 PR tree-optimization/79621
11542 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
11543 blocks with edges to themselves.
11545 2017-02-21 Jakub Jelinek <jakub@redhat.com>
11548 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
11549 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
11550 Use gimple_call_builtin_p.
11553 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
11554 on temporarily removed DEBUG_INSNs.
11556 PR tree-optimization/79649
11557 * tree-loop-distribution.c (classify_partition): Give up on
11558 non-generic address space loads/stores.
11560 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
11562 * doc/loop.texi (Loop manipulation): Remove nonexistent
11563 tree_ssa_loop_version from the documentation.
11564 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
11566 2017-02-21 Jakub Jelinek <jakub@redhat.com>
11569 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
11570 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
11571 * config/rs6000/rs6000.c: Include except.h.
11572 (rs6000_expand_split_stack_prologue): Call
11573 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
11575 2017-02-21 Martin Jambor <mjambor@suse.cz>
11578 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
11579 have been analyzed.
11581 2017-02-21 Martin Jambor <mjambor@suse.cz>
11583 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
11584 for backward compatibility only.
11585 * doc/invoke.texi (Option Summary): Remove all references to
11586 -fipa-cp-alignment.
11588 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
11592 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
11594 * lra-constraints.c (curr_insn_transform): Handle
11595 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
11597 2017-02-21 Martin Liska <mliska@suse.cz>
11599 * config/i386/i386.opt: Replace -masm-dialect with -masm.
11601 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
11603 PR translation/79638
11604 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
11606 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
11609 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
11610 (arm_function_ok_for_sibcall): Return false for an indirect call by
11611 descriptor if all the argument registers are used.
11612 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
11613 alignment of the function.
11615 2017-02-21 Jakub Jelinek <jakub@redhat.com>
11617 PR tree-optimization/61441
11618 * simplify-rtx.c (simplify_const_unary_operation): For
11619 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
11620 the sNaN unmodified.
11622 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
11624 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
11625 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
11626 instead of SYSTEM_HEADER_DIR.
11628 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
11629 Martin Liška <mliska@suse.cz>
11631 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
11632 Fix typos and grammar, use active voice, and clarify.
11634 2017-02-20 Marek Polacek <polacek@redhat.com>
11636 PR middle-end/79537
11637 * gimplify.c (gimplify_expr): Handle unused *&&L;.
11640 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
11642 2017-02-20 Jakub Jelinek <jakub@redhat.com>
11645 * config/i386/i386.c (ix86_expand_builtin): Handle
11646 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
11647 ix86_builtins_isa[fcode].isa as a requirement of those
11648 flags and any other flag in the bitmask.
11649 (ix86_init_mmx_sse_builtins): Use 0 instead of
11650 ~OPTION_MASK_ISA_64BIT as mask.
11651 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
11652 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
11653 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
11654 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
11656 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
11659 * lra-constraints.c (split_reg): Check requested split mode
11660 is supported by the register.
11662 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
11664 * lra-constraints.c (simplify_operand_subreg): Remove early
11667 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
11670 * lra-constraints.c (curr_insn_transform): Tighten condition
11671 for converting SUBREG reloads from OP_OUT to OP_INOUT.
11673 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
11676 * lra-constraints.c (curr_insn_transform): Handle
11677 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
11679 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
11682 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
11684 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
11686 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
11689 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
11692 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11694 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
11697 2017-02-18 Jakub Jelinek <jakub@redhat.com>
11699 * final.c (last_columnnum, override_columnnum): New variables.
11700 (final_start_function): Set last_columnnum, pass it to begin_prologue
11701 hook and pass 0 to dwarf2out_begin_prologue.
11702 (final_scan_insn): Update override_columnnum. Pass last_columnnum
11703 to source_line debug hook.
11704 (notice_source_line): Compute last_columnnum and for debug_column_info
11705 return true on column changes.
11706 * debug.h (struct gcc_debug_hooks): Add column argument to
11707 source_line and begin_prologue hooks.
11708 (debug_nothing_int_charstar_int_bool): Remove prototype.
11709 (debug_nothing_int_int_charstar,
11710 debug_nothing_int_int_charstar_int_bool): New prototypes.
11711 (dwarf2out_begin_prologue): Add column argument.
11712 * debug.c (do_nothing_debug_hooks): Adjust source_line and
11713 begin_prologue hooks.
11714 (debug_nothing_int_charstar_int_bool): Remove.
11715 (debug_nothing_int_int_charstar,
11716 debug_nothing_int_int_charstar_int_bool): New functions.
11717 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
11718 through to dwarf2out_source_line.
11719 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
11720 (dwarf2out_source_line): Add column argument, emit it if requested.
11721 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
11723 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
11724 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
11725 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
11726 through to dwarf2out_begin_prologue.
11727 (vmsdbgout_source_line): Add column argument, pass it through to
11728 dwarf2out_source_line.
11729 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
11730 dbxout_source_line caller.
11731 (dbxout_source_line): Add column argument.
11733 * common.opt (gno-column-info, gcolumn-info): New options.
11734 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
11735 (check_die): Also test for multiple DW_AT_decl_column attributes.
11736 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
11737 DW_AT_decl_column if requested.
11738 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
11740 (gen_variable_die): Likewise.
11741 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
11742 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
11745 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
11746 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
11747 (ix86_handle_option): Handle OPT_m3dnowa.
11748 * doc/invoke.texi (-m3dnowa): Document.
11749 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
11750 -m3dnowa instead of -m3dnow -march=athlon.
11753 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
11754 instead of gcc_assert for K, r and R code checks. Formatting fixes.
11756 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11759 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
11760 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
11761 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
11762 generator for vsx_xxpermdi_<mode>_be.
11763 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
11764 force big-endian semantics.
11765 (vsx_xxpermdi_<mode>_be): New define_expand with same
11766 implementation as previous version of vsx_xxpermdi_<mode>.
11768 2017-02-17 Jakub Jelinek <jakub@redhat.com>
11770 PR tree-optimization/79327
11771 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
11772 variable, its initialization and use.
11774 2017-02-17 Julia Koval <julia.koval@intel.com>
11776 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
11777 (OPTION_MASK_ISA_PKU_UNSET): New.
11778 (ix86_handle_option): Handle -mrdpid.
11779 * config/i386/cpuid.h (bit_RDPID): New.
11780 * config/i386/driver-i386.c (host_detect_local_cpu):
11781 Detect RDPID feature.
11782 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
11783 * config/i386/i386-c.c (ix86_target_macros_internal):
11785 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
11786 (ix86_valid_target_attribute_inner_p): Add "rdpid".
11787 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
11788 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
11789 * config/i386/i386.md (define_insn "rdpid"): New.
11790 * config/i386/i386.opt Add -mrdpid.
11791 * config/i386/immintrin.h (_rdpid_u32): New.
11793 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
11795 PR rtl-optimization/79541
11796 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
11797 instead of transforming it into USE.
11799 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
11801 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
11802 If HONOR_SNANS (SFmode) force the input to a register.
11803 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
11804 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
11805 an frsp or similar insn.
11807 2017-02-17 Martin Liska <mliska@suse.cz>
11809 PR rtl-optimization/79577
11810 * params.def (selsched-max-sched-times): Increase minimum to 1.
11812 2017-02-17 Martin Liska <mliska@suse.cz>
11814 PR rtl-optimization/79574
11815 * gcse.c (want_to_gcse_p): Prevent integer overflow.
11817 2017-02-17 Martin Liska <mliska@suse.cz>
11819 PR tree-optimization/79529
11820 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
11821 ssa_defined_default_def_p to handle cases which are implicitly
11823 * tree-ssa.c (ssa_defined_default_def_p): New function.
11824 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
11825 which are implicitly defined.
11826 * tree-ssa.h (ssa_defined_default_def_p): Declare.
11828 2017-02-17 Richard Biener <rguenther@suse.de>
11830 PR middle-end/79576
11831 * params.def (max-ssa-name-query-depth): Limit to 10.
11833 2017-02-17 Richard Biener <rguenther@suse.de>
11835 PR tree-optimization/79552
11836 * tree-ssa-structalias.c (visit_loadstore): Properly verify
11839 2017-02-17 Richard Biener <rguenther@suse.de>
11842 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
11844 2017-02-17 Marek Polacek <polacek@redhat.com>
11846 PR middle-end/79536
11847 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
11848 (fold_negate_expr): New wrapper.
11850 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
11852 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
11853 Correct terminology and de-emphasize pre-standard behavior.
11855 2017-02-16 Alan Modra <amodra@gmail.com>
11857 PR rtl-optimization/79286
11858 * ira.c (def_dominates_uses): New function.
11859 (update_equiv_regs): Don't create an equivalence for insns that
11860 may trap where the register def does not dominate the use.
11862 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
11864 PR rtl-optimization/78127
11865 * lra.c (lra): Call lra_eliminate before finish the loop after
11868 2017-02-16 Richard Biener <rguenther@suse.de>
11870 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
11872 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
11873 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
11874 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
11875 (isl_val_int_from_wi): New function.
11876 (extract_affine_gmp): Rename to ...
11877 (extract_affine_wi): ... this, take a widest_int.
11878 (extract_affine_int): Just wrap extract_affine_wi.
11879 (add_param_constraints): Use isl_val_int_from_wi.
11880 (add_loop_constraints): Likewise, and extract_affine_wi.
11882 2017-02-15 Jeff Law <law@redhat.com>
11884 PR middle-end/79521
11885 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
11886 ira_init_register_move_cost_if_necessary.
11888 2017-02-15 Martin Sebor <msebor@redhat.com>
11890 PR middle-end/32003
11891 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
11892 removed in a prior commit.
11894 2017-02-15 Bin Cheng <bin.cheng@arm.com>
11896 PR tree-optimization/79347
11897 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
11898 counters during peeling.
11900 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
11902 * Makefile.in (site.exp): Remove "set ISLVER".
11904 2017-02-15 Jakub Jelinek <jakub@redhat.com>
11907 * real.c (real_from_integer): Call real_convert even for decimal.
11909 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
11912 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
11914 2017-02-14 Andrew Pinski <apinski@cavium.com>
11916 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
11917 cores and change the partno/implementer to be correct.
11918 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
11919 the 'B" as the implementer.
11920 * config/aarch64/aarch64-tune.md: Regenerate.
11922 2017-02-14 Carl Love <cel@us.ibm.com>
11924 * config/rs6000/rs6000.c: Add case statement entry to make the
11925 xvcvuxdsp built-in argument unsigned.
11926 * config/rs6000/vsx.md: Fix the source and return operand types so they
11927 match the instruction definitions from the ISA document. Fix typo
11928 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
11931 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
11934 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
11935 member early_clobber_alts.
11936 * lra-lives.c (reg_early_clobber_p): New.
11937 (process_bb_lives): Use it.
11938 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
11939 (debug_operand_data): Initialize early_clobber_alts.
11940 (setup_operand_alternative): Set up early_clobber_alts.
11941 (collect_non_operand_hard_regs): Ditto. Pass early clobber
11942 alternatives to new_insn_reg.
11943 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
11945 (lra_update_insn_regno_info): Pass the new arg.
11947 2017-02-14 Jakub Jelinek <jakub@redhat.com>
11949 PR middle-end/79505
11950 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
11951 (new_oacc_loop_raw): Don't clear already cleared fields.
11954 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
11955 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
11956 _mm512_prefetch_i64gather_ps): New inline functions and macros.
11958 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
11961 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
11963 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
11966 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
11967 the extra instruction to the right place to store 128-bit constant
11970 2017-02-14 Martin Sebor <msebor@redhat.com>
11972 PR middle-end/79448
11973 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
11974 warning for strings of unknown length.
11976 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
11978 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
11980 2017-02-14 Jeff Law <law@redhat.com>
11983 * ira-costs.c (scan_one_insn): Initialize register move costs
11984 for pseudos seen in USE/CLOBBER insns.
11986 PR tree-optimization/79095
11987 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
11988 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
11989 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
11990 if the operands are known to be not equal, then the resulting range
11992 (intersect_ranges): If the new range is ~[0,0] and the old range is
11993 wide, then prefer ~[0,0].
11994 * tree-vrp.c (overflow_comparison_p_1): New function.
11995 (overflow_comparison_p): New function.
11996 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
11997 if NAME is used in an overflow test.
11998 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
11999 overflow check that can be expressed as an equality test, then adjust
12000 ops to be that equality test.
12002 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12004 * config/s390/s390-builtin-types.def: Remove flags argument.
12005 * config/s390/s390.c (s390_init_builtins): Likewise.
12007 2017-02-14 Martin Liska <mliska@suse.cz>
12009 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
12010 vector. Fix trailing white spaces.
12012 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
12014 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
12017 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12019 PR rtl-optimization/68664
12020 * config/arm/arm.c (arm_sched_can_speculate_insn):
12021 New function. Declare prototype.
12022 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
12024 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12026 PR rtl-optimization/68664
12027 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
12029 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
12031 2017-02-14 Amit Pawar <amit.pawar@amd.com>
12033 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
12034 max skip bytes for function, loop and jump.
12036 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12038 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
12039 ABS_EXPR for gimple dump.
12041 2017-02-14 Jakub Jelinek <jakub@redhat.com>
12044 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
12046 PR tree-optimization/79408
12047 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
12048 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
12049 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
12050 also if rhs1 is INTEGER_CST.
12052 2017-02-14 Richard Biener <rguenther@suse.de>
12054 PR middle-end/79432
12055 * tree-into-ssa.c (insert_phi_nodes): When the function can
12056 have abnormal edges rewrite SSA names with broken use-def
12057 dominance out of SSA and register them for PHI insertion.
12059 2017-02-13 Martin Sebor <msebor@redhat.com>
12061 PR middle-end/79496
12062 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
12063 clearing info.nowrite flag when snprintf size argument is a range.
12065 2017-02-13 Jakub Jelinek <jakub@redhat.com>
12067 * cprop.c (cprop_jump): Add missing space in string literal.
12068 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
12069 (get_constraint_for_component_ref): Likewise.
12070 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
12071 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
12072 * lra-constraints.c (process_alt_operands): Likewise.
12073 * ipa-inline.c (inline_small_functions): Likewise.
12074 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
12075 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
12076 * trans-mem.c (diagnose_tm_1_op): Likewise.
12077 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
12078 (grid_parallel_clauses_gridifiable): Likewise.
12080 * config/nvptx/mkoffload.c (process): Add space in between
12083 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
12084 "MOD4_SSE_REGS" and "ALL_REGS".
12086 * spellcheck.c (test_data): Add , in between "foo" and "food".
12088 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12091 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
12092 boundary crossing check and subsequent code generation agree.
12094 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12096 * config/aarch64/aarch64.c (has_memory_op): Delete.
12097 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
12100 2017-02-13 Jakub Jelinek <jakub@redhat.com>
12102 PR rtl-optimization/79388
12103 PR rtl-optimization/79450
12104 * combine.c (distribute_notes): When removing TEM_INSN for which
12105 corresponding dest has last value recorded, invalidate that last
12108 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12110 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
12111 of explicit '@'. Add missing assembly comment marker on branch costs
12114 2017-02-13 Nathan Sidwell <nathan@acm.org>
12116 * gengtype-lex.l (<in_struct>): Add '/'.
12118 2017-02-13 Martin Liska <mliska@suse.cz>
12121 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
12123 2017-02-13 Richard Biener <rguenther@suse.de>
12125 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
12127 * configure: Re-generate.
12128 * config.in: Likewise.
12129 * graphite-dependences.c: Simplify as if
12130 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
12131 * graphite-isl-ast-to-gimple.c: Likewise.
12132 * graphite-optimize-isl.c: Likewise.
12133 * graphite-poly.c: Likewise.
12134 * graphite-sese-to-poly.c: Likewise.
12135 * graphite.h: Likewise.
12136 * toplev.c: Include isl/version.h and use isl_version () for
12137 printing the ISL version.
12138 * doc/install.texi: Update ISL requirement.
12140 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
12142 * doc/standards.texi (Standards): Update reference to
12145 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
12147 * doc/extend.texi (Named Address Spaces): sourceware.org now
12149 * doc/install.texi (Binaries): Ditto.
12152 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
12154 * doc/cpp.texi: Replace "stringify"/"stringification" with C
12155 standard terminology "stringize"/"stringizing" throughout.
12156 * doc/cppinternals.texi: Likewise.
12158 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
12160 * doc/extend.texi: Fix some spelling mistakes and typos.
12161 * doc/invoke.texi: Likewise.
12163 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
12166 * params.def (inline-min-speedup) Change from 10 to 8.
12168 2017-02-11 Jakub Jelinek <jakub@redhat.com>
12170 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
12173 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
12176 * ipa-inline-analysis.c (get_minimal_bb): New function.
12177 (record_modified): Use it.
12178 (remap_edge_change_prob): Handle also ancestor functions.
12180 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
12182 * doc/contrib.texi (Contributors): Remove broken link into
12183 the Mauve CVS repository.
12185 2017-02-11 Jakub Jelinek <jakub@redhat.com>
12187 PR middle-end/79454
12188 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
12189 result computation whenever lhs doesn't have vector mode, not
12190 just when it has BLKmode.
12192 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
12194 * doc/makefile.texi (profiledbootstrap): Refer to the
12195 installation instructions only in textual form.
12197 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12200 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
12202 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
12204 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
12205 (Specific): Update mingw-w64 reference.
12207 (Specific): Remove broken link to Renesas RX processor.
12209 2017-02-10 Richard Biener <rguenther@suse.de>
12211 * toplev.c (process_options): Do not mention obsolete graphite
12212 options when printing sorry message about missing graphite support.
12213 Mention -floop-nest-optimize.
12215 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
12217 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
12218 (vtst_p16): Likewise.
12219 (vtstq_p8): Likewise.
12220 (vtstq_p16): Likewise.
12222 (vtstq_p64): Likewise.
12223 * config/arm/arm_neon.h (vgetq_lane_p64): New.
12224 (vset_lane_p64): New.
12225 (vsetq_lane_p64): New.
12227 2017-02-10 Jakub Jelinek <jakub@redhat.com>
12229 PR tree-optimization/79411
12230 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
12231 stmt operands are SSA_NAMEs used in abnormal phis.
12232 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
12235 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
12238 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
12241 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
12243 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
12245 2017-02-09 Jakub Jelinek <jakub@redhat.com>
12247 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
12251 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
12252 not arbitrary TREE_CONSTANT.
12255 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
12256 "omp declare target link" attribute unless is_global_var.
12257 * omp-offload.c (find_link_var_op): Likewise.
12259 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
12260 Chung-Lin Tang <cltang@codesourcery.com>
12262 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
12264 (gimplify_adjust_omp_clauses): Don't delete TILE.
12265 (gimplify_omp_for): Deal with TILE.
12266 * internal-fn.c (expand_GOACC_TILE): New function.
12267 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
12269 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
12270 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
12272 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
12273 avoid DIV for outermost collapse var.
12274 (expand_oacc_for): Insert tile element loop as needed. Adjust.
12275 Remove out of date comments, fix whitespace.
12276 * omp-general.c (omp_extract_for_data): Deal with tiling.
12277 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
12278 adjust OLF_DIM_BASE value.
12279 (struct omp_for_data): Add tiling field.
12280 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
12281 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
12282 for auto loops. Remove default auto determining, moved to
12283 oacc_loop_fixed_partitions.
12284 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
12285 stmts, add e_mask field.
12286 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
12287 (oacc_thread_numbers): Use oacc_dim_call.
12288 (oacc_xform_tile): New.
12289 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
12290 (finish_oacc_loop): Adjust for ifns vector.
12291 (oacc_loop_discover_walk): Append loop abstraction sites to list,
12292 add case for GOACC_TILE fns.
12293 (oacc_loop_xform_loop): Delete.
12294 (oacc_loop_process): Iterate over call list directly, and add
12295 handling for GOACC_TILE fns.
12296 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
12298 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
12299 vector partitioning to outer loops. Assign 2 partitions to loops
12300 when available. Add TILE handling.
12301 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
12302 (execite_oacc_device_lower): Process GOACC_TILE fns,
12303 ignore unknown specs.
12304 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
12305 * tree.c (omp_clause_num_ops): Adjust TILE ops.
12306 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
12308 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
12310 * configure.ac (ACX_BUGURL): Update.
12311 * configure: Regenerate.
12313 2017-02-09 Richard Biener <rguenther@suse.de>
12315 PR tree-optimization/69823
12316 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
12317 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
12319 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
12321 * config/arc/arc-c.def: Add __NPS400__ definition.
12322 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
12323 (TARGET_NPS400): Define.
12325 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
12327 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
12329 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
12330 pointer, arch_info.
12331 (arc_cpu_types): Fill the arch_info field with a pointer into the
12332 arc_arch_types table.
12333 (arc_selected_cpu): Declare.
12334 * config/arc/arc.c (arc_selected_cpu): Make global.
12335 (arc_selected_arch): Delete.
12336 (arc_base_cpu): Delete.
12337 (arc_override_options): Remove references to deleted variables,
12338 update access to arch information.
12339 (ARC_OPT): Update access to arch information.
12340 (ARC_OPTX): Likewise.
12341 * config/arc/arc.h (arc_base_cpu): Remove declaration.
12342 (TARGET_ARC600): Update access to arch information.
12343 (TARGET_ARC601): Likewise.
12344 (TARGET_ARC700): Likewise.
12345 (TARGET_EM): Likewise.
12346 (TARGET_HS): Likewise.
12347 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
12350 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
12353 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
12354 condition/operands for integer GE/LE/GEU/LEU operations.
12356 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
12358 PR translation/79397
12359 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
12362 2017-02-08 Martin Jambor <mjambor@suse.cz>
12365 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
12366 whether allocation happened.
12367 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
12368 nothing was allocated.
12370 2017-02-08 Jakub Jelinek <jakub@redhat.com>
12372 PR tree-optimization/79408
12373 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
12374 constant, but SSA_NAME with a known integer range, use the minimum
12375 of that range instead of op1 to determine if modulo can be replaced
12376 with its first operand.
12378 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12380 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
12382 2017-02-08 Richard Biener <rguenther@suse.de>
12384 PR tree-optimization/71824
12385 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
12386 Check all loops contained in the merged region.
12388 2017-02-07 Andrew Pinski <apinski@cavium.com>
12390 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
12392 2017-02-07 Andrew Pinski <apinski@cavium.com>
12394 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
12395 (thunderxt88): Likewise.
12396 (thunderxt81): Disable LSE and change v8.1 to v8.
12397 (thunderxt83): Likewise.
12399 2017-02-07 Jakub Jelinek <jakub@redhat.com>
12400 Richard Biener <rguenther@suse.de>
12402 PR middle-end/79399
12403 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
12404 type from int to size_t.
12405 * ira-costs.c (struct_costs_size): Change type from int to size_t.
12407 2017-02-07 Jakub Jelinek <jakub@redhat.com>
12409 PR rtl-optimization/79386
12410 * cprop.c (bypass_conditional_jumps): Initialize
12411 bypass_last_basic_block already before splitting bbs after
12412 unconditional traps...
12413 (bypass_conditional_jumps): ... rather than here.
12416 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
12417 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
12418 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
12419 fix -masm=intel patterns.
12421 2017-02-07 Richard Biener <rguenther@suse.de>
12423 PR tree-optimization/79256
12424 PR middle-end/79278
12425 * builtins.c (get_object_alignment_2): Use min_align_of_type
12426 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
12427 and ADJUST_FIELD_ALIGN.
12429 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
12431 * doc/tm.texi: Regenerate.
12432 * stor-layout.c (layout_decl): Adjust.
12433 (update_alignment_for_field): Likewise.
12434 (place_field): Likewise.
12435 (min_align_of_type): Likewise.
12436 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
12437 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
12438 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
12439 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
12440 * config/frv/frv.c (frv_adjust_field_align): Likewise.
12441 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
12442 * config/i386/i386.c (x86_field_alignment): Likewise.
12443 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
12444 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
12445 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
12446 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
12447 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
12448 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
12452 2017-01-30 Richard Biener <rguenther@suse.de>
12454 PR tree-optimization/79256
12455 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
12456 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
12459 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
12461 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
12462 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
12463 builtins to SImode and emit a zero-extend, if necessary.
12465 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
12467 * docs/invoke.texi (RISC-V Options): Alphabetize.
12469 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
12471 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
12474 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
12476 * config/riscv/riscv.c: New file.
12477 * gcc/common/config/riscv/riscv-common.c: Likewise.
12478 * config.gcc: Likewise.
12479 * config/riscv/constraints.md: Likewise.
12480 * config/riscv/elf.h: Likewise.
12481 * config/riscv/generic.md: Likewise.
12482 * config/riscv/linux.h: Likewise.
12483 * config/riscv/multilib-generator: Likewise.
12484 * config/riscv/peephole.md: Likewise.
12485 * config/riscv/pic.md: Likewise.
12486 * config/riscv/predicates.md: Likewise.
12487 * config/riscv/riscv-builtins.c: Likewise.
12488 * config/riscv/riscv-c.c: Likewise.
12489 * config/riscv/riscv-ftypes.def: Likewise.
12490 * config/riscv/riscv-modes.def: Likewise.
12491 * config/riscv/riscv-opts.h: Likewise.
12492 * config/riscv/riscv-protos.h: Likewise.
12493 * config/riscv/riscv.h: Likewise.
12494 * config/riscv/riscv.md: Likewise.
12495 * config/riscv/riscv.opt: Likewise.
12496 * config/riscv/sync.md: Likewise.
12497 * config/riscv/t-elf-multilib: Likewise.
12498 * config/riscv/t-linux: Likewise.
12499 * config/riscv/t-linux-multilib: Likewise.
12500 * config/riscv/t-riscv: Likewise.
12501 * configure.ac: Likewise.
12502 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
12503 Waterman as RISC-V maintainers.
12504 * doc/install.texi: Add RISC-V entries.
12505 * doc/invoke.texi: Add RISC-V options section.
12506 * doc/md.texi: Add RISC-V constraints section.
12507 * configure: Regenerated.
12509 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
12512 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
12513 false values to be constant vectors with all 0 or all 1 bits set.
12514 (vcondu<mode><mode>): Likewise.
12515 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
12517 (fpmask_comparison_operator): Update comment.
12518 (vecint_comparison_operator): New predicate.
12519 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
12520 vector conditionals when the true and false values are constant
12521 vectors with all 0 bits or all 1 bits set.
12523 2017-02-06 Martin Sebor <msebor@redhat.com>
12525 PR tree-optimization/79376
12526 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
12528 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
12530 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
12531 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
12532 to simplify split condition.
12534 2017-02-06 Jakub Jelinek <jakub@redhat.com>
12536 * omp-expand.c (oxpand_omp_atomic_fetch_op,
12537 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
12540 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
12542 PR rtl-optimization/68664
12543 * target.def (can_speculate_insn): New hook.
12544 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
12545 * doc/tm.texi: Regenerate.
12546 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
12547 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
12548 (rs6000_sched_can_speculate_insn): New function.
12550 2017-02-06 Jakub Jelinek <jakub@redhat.com>
12552 PR tree-optimization/79284
12553 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
12554 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
12555 vectorizable_mask_load_store, vectorizable_operation,
12556 vect_is_simple_cond, get_same_sized_vectype): Use it instead
12557 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
12558 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
12559 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
12560 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
12561 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
12562 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
12563 is_gimple_assign (stmt). Replace another such test with
12564 is_gimple_assign (stmt).
12566 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
12569 * config/avr/avr.c (rtl-iter.h): Include it.
12570 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
12571 (avr_legitimate_combined_insn): ...and implementation.
12573 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
12575 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
12576 * config/s390/s390.c (s390_const_operand_ok)
12577 (s390_canonicalize_comparison, s390_extract_part)
12578 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
12579 (s390_contiguous_bitmask_p, s390_rtx_costs)
12580 (legitimize_pic_address): Likewise.
12581 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
12582 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
12583 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
12584 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
12585 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
12587 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
12589 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
12590 REGNO($0) == REGNO($1).
12592 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12594 * config/s390/linux.h(SIZE_TYPE): Add comment.
12596 2017-02-06 Julian Brown <julian@codesourcery.com>
12597 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12598 Virendra Pathak <virendra.pathak@broadcom.com>
12600 * config/aarch64/aarch64-cores.def: Change the scheduler
12602 * config/aarch64/aarch64.md: Include thunderx2t99.md.
12603 * config/aarch64/thunderx2t99.md: New file.
12605 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
12607 * doc/standards.texi (Go Language): Update link to language
12610 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
12612 * tree-eh.c (lower_resx): Sanitize profile.
12613 (cleanup_empty_eh_move_lp): Likewise.
12615 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
12618 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
12620 * cfgloopmanip.h (loop_version): Update prototype.
12621 * modulo-sched.c (sms_schedule): Update call of loop_version.
12622 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
12623 * tree-parloops.c (gen_parallel_loop): Likewise.
12624 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
12625 * tree-ssa-loop-split.c (split_loop): Likewise.
12626 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
12627 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
12629 2017-02-05 Martin Liska <mliska@suse.cz>
12632 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
12634 (print_operand_address): Initialize a struct to zero.
12636 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
12638 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
12639 garbage collector only in textual form.
12641 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
12643 * doc/extend.texi (x86 specific memory model extensions for
12644 transactional memory): Simplify a phrase.
12646 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
12649 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
12650 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
12651 (atomic_storedi_1): Likewise.
12653 2017-02-04 Jakub Jelinek <jakub@redhat.com>
12655 PR tree-optimization/79338
12656 * tree-parloops.c (gather_scalar_reductions): Don't call
12657 vect_analyze_loop_form for loop->inner before destroying loop's
12660 2017-02-03 Martin Sebor <msebor@redhat.com>
12662 PR tree-optimization/79327
12663 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
12664 when precision has resulted in leading zeros.
12665 (format_integer): Adjust the likely counter to assume an unknown
12666 argument that may be zero is non-zero.
12668 2017-02-03 Jason Merrill <jason@redhat.com>
12671 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
12672 avoid copying non-taken branch.
12674 2017-02-03 Jakub Jelinek <jakub@redhat.com>
12676 PR tree-optimization/79340
12677 * tree-vect-loop.c (vectorizable_reduction): Release
12678 vec_defs elements after safe_splicing them into other vectors.
12681 PR tree-optimization/79327
12682 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
12683 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
12685 (format_integer): Use wide_int_to_tree instead of build_int_cst
12686 + to_?hwi. If argmin is NULL, just set argmin and argmax to
12687 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
12688 of shortest and longest sequence.
12690 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
12692 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
12693 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
12695 2017-02-03 Walter Lee <walt@tilera.com>
12698 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
12699 after initial stackframe link reg save.
12700 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
12702 2017-02-03 Jakub Jelinek <jakub@redhat.com>
12705 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
12706 wu for stxssp alternative.
12708 2017-02-03 Martin Sebor <msebor@redhat.com>
12710 PR tree-optimization/79352
12711 * gimple-fold.c (get_range_strlen): Add argument.
12712 (get_range_strlen): Change return type to bool.
12713 (get_maxval_strlen): Pass in a dummy argument.
12714 * gimple-fold.h (get_range_strlen): Change return type to bool.
12715 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
12716 * tree.h (array_at_struct_end_p): Add argument.
12717 * tree.c (array_at_struct_end_p): Handle it.
12719 2017-02-03 Martin Liska <mliska@suse.cz>
12722 * multiple_target.c (create_dispatcher_calls): Redirect edge
12723 from a caller of a dispatcher.
12724 (expand_target_clones): Make the clones local.
12725 (ipa_target_clone): Do both target clones and resolvers.
12726 (ipa_dispatcher_calls): Remove the pass.
12727 (pass_dispatcher_calls::gate): Likewise.
12728 (make_pass_dispatcher_calls): Likewise.
12729 * passes.def (pass_target_clone): Put as very first IPA early
12732 2017-02-03 Martin Liska <mliska@suse.cz>
12734 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
12735 in case of a function with ifunc attribute.
12737 2017-02-03 Martin Liska <mliska@suse.cz>
12739 * cgraph.c (cgraph_node::dump): Dump function version info.
12740 * symtab.c (symtab_node::dump_base): Add missing new line.
12742 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
12744 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
12745 (ifcombine_ifandif): Use it.
12747 2017-02-03 Martin Liska <mliska@suse.cz>
12749 * doc/invoke.texi: Document default value for
12750 use-after-scope-direct-emission-threshold.
12752 2017-02-03 Martin Liska <mliska@suse.cz>
12754 PR tree-optimization/79339
12755 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
12756 (format_floating): Likewise.
12758 2017-02-03 Martin Liska <mliska@suse.cz>
12761 * ipa-prop.c (ipa_node_params_t::insert): Remove current
12763 (ipa_node_params_t::remove): Likewise.
12764 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
12765 initialization from removed ipa_node_params_t::insert.
12766 (ipa_node_params::~ipa_node_params): Move from removed
12767 ipa_node_params_t::release.
12768 * symbol-summary.h (symbol_summary::m_released): New member.
12769 Do not release a summary twice. Do not allow to call finalizer
12770 for types of a summary that live in GGC memory.
12772 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12774 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
12777 2017-02-02 Martin Sebor <msebor@redhat.com>
12779 PR middle-end/79275
12780 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
12781 (format_string): Tighten up the range of output for non-constant
12782 strings and correct the expected range for wide non-constant strings.
12784 2017-02-02 Martin Sebor <msebor@redhat.com>
12786 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
12788 PR middle-end/32003
12789 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
12791 (-fdump-tree-@var): Add to index and document how to come up
12792 with pass-specific option and dump file names.
12793 (-fdump-passes): Clarify where to look for output.
12795 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
12797 PR middle-end/77445
12798 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
12799 statistics of the analyzed path; allow threading for speed when
12800 any of BBs along the path are optimized for speed.
12802 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
12804 PR middle-end/78468
12805 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
12806 settings of the virtual registers.
12809 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
12811 * explow.c (get_dynamic_stack_size): Take known alignment of stack
12812 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
12815 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12817 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
12818 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
12820 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12822 * config/s390/s390.md: Add missing comments with the expanded
12824 * config/s390/vector.md: Likewise.
12825 * config/s390/vx-builtins.md: Likewise.
12827 2017-02-02 Jakub Jelinek <jakub@redhat.com>
12830 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
12831 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
12832 conditions on a single line.
12834 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12836 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
12837 __S390_VX__ to __VX__.
12839 2017-02-01 Andrew Pinski <apinski@cavium.com>
12841 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
12842 stmt_info to record_stmt_cost.
12843 (vect_get_known_peeling_cost): Pass stmt_info if known to
12845 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
12846 cpu_vector_cost field into
12847 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
12848 field into vec_int_stmt_cost and vec_fp_stmt_cost.
12849 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
12850 splitting of scalar_stmt_cost and vec_stmt_cost.
12851 (thunderx_vector_cost): Likewise.
12852 (cortexa57_vector_cost): LIkewise.
12853 (exynosm1_vector_cost): Likewise.
12854 (xgene1_vector_cost): Likewise.
12855 (thunderx2t99_vector_cost): Improve after the splitting of the two
12857 (aarch64_builtin_vectorization_cost): Update for the splitting of
12858 scalar_stmt_cost and vec_stmt_cost.
12860 2017-02-01 Torvald Riegel <triegel@redhat.com>
12861 Richard Henderson <rth@redhat.com>
12863 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
12864 conditional on existance of a fast atomic load.
12865 * optabs-query.c (can_atomic_load_p): New function.
12866 * optabs-query.h (can_atomic_load_p): Declare it.
12867 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
12868 no fast atomic load is available for the particular size of access.
12869 (expand_atomic_compare_and_swap): Likewise.
12870 (expand_atomic_load): Likewise.
12871 (expand_atomic_store): Likewise.
12872 (expand_atomic_fetch_op): Likewise.
12873 * testsuite/lib/target-supports.exp
12874 (check_effective_target_sync_int_128): Remove x86 because it provides
12875 no fast atomic load.
12876 (check_effective_target_sync_int_128_runtime): Likewise.
12878 2017-02-01 Richard Biener <rguenther@suse.de>
12880 * graphite.c: Include tree-vectorizer.h for find_loop_location.
12881 (graphite_transform_loops): Provide opt-info for optimized nests.
12882 * tree-parloop.c (parallelize_loops): Provide opt-info for
12883 parallelized loops.
12885 2017-02-01 Richard Biener <rguenther@suse.de>
12887 PR middle-end/79315
12888 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
12889 was not set before.
12891 2017-02-01 Richard Biener <rguenther@suse.de>
12893 PR tree-optimization/71824
12894 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
12895 Verify the loops are valid in the merged SESE region.
12896 (scop_detection::can_represent_loop_1): Check analyzing the
12897 evolution of the number of iterations in the region succeeds.
12899 2017-01-31 Ian Lance Taylor <iant@golang.org>
12901 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
12902 REG_ARGS_SIZE note to 32-bit push insns and call insn.
12904 2017-01-31 David Malcolm <dmalcolm@redhat.com>
12906 PR preprocessor/79210
12907 * input.c (get_substring_ranges_for_loc): Replace line_width
12908 assertion with error-handling.
12910 2017-01-31 Richard Biener <rguenther@suse.de>
12912 PR tree-optimization/77318
12913 * graphite-sese-to-poly.c (extract_affine): Fix assert.
12914 (create_pw_aff_from_tree): Take loop parameter.
12915 (add_condition_to_pbb): Pass loop of the condition to
12916 create_pw_aff_from_tree.
12918 2017-01-31 Jakub Jelinek <jakub@redhat.com>
12920 * config/s390/s390.c (s390_asan_shadow_offset): New function.
12921 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
12923 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
12927 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
12929 (convert_int_to_float128): Likewise.
12930 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
12931 (convert_int_to_float128): Likewise.
12932 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
12933 (UNSPEC_IEEE128_CONVERT): Likewise.
12934 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
12935 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
12936 Use local variables for IBM extended format.
12937 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
12938 (fix_trunc<mode>si2_fprs): Likewise.
12939 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
12940 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
12941 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
12942 to know that we can now have integers of all sizes in vector
12944 (fix<uns>_<mode>di2_hw): Likewise.
12945 (float<uns>_<mode>si2_hw): Likewise.
12946 (fix_<mode>si2_hw): Likewise.
12947 (fixuns_<mode>si2_hw): Likewise.
12948 (float<uns>_<mode>di2_hw): Likewise.
12949 (float_<mode>di2_hw): Likewise.
12950 (float_<mode>si2_hw): Likewise.
12951 (floatuns_<mode>di2_hw): Likewise.
12952 (floatuns_<mode>si2_hw): Likewise.
12953 (xscvqp<su>wz_<mode>): Delete, no longer used.
12954 (xscvqp<su>dz_<mode>): Likewise.
12955 (xscv<su>dqp_<mode>): Likewise.
12956 (ieee128_mfvsrd_64bit): Likewise.
12957 (ieee128_mfvsrd_32bit): Likewise.
12958 (ieee128_mfvsrwz): Likewise.
12959 (ieee128_mtvsrw): Likewise.
12960 (ieee128_mtvsrd_64bit): Likewise.
12961 (ieee128_mtvsrd_32bit): Likewise.
12963 2017-01-31 Martin Liska <mliska@suse.cz>
12966 * ipa-prop.c (ipa_free_all_node_params): Call release method
12967 instead of ~sumbol_summary to not to trigger double times
12970 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
12972 PR tree-optimization/71691
12973 * bitmap.h (class auto_bitmap): New.
12974 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
12975 is_maybe_undefined instead of ssa_undefined_value_p.
12977 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12979 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
12980 __S390_ARCH_LEVEL__ to __ARCH__.
12982 2017-01-31 Jakub Jelinek <jakub@redhat.com>
12984 PR tree-optimization/79267
12985 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
12986 if should_remove_lhs_p is true.
12988 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
12991 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
12992 (add_alignment_attribute): New.
12993 (base_type_die): Add alignment attribute.
12994 (subrange_type_die): Likewise.
12995 (modified_type_die): Likewise.
12996 (gen_array_type_die): Likewise.
12997 (gen_descr_array_type_die: Likewise.
12998 (gen_enumeration_type_die): Likewise.
12999 (gen_subprogram_die): Likewise.
13000 (gen_variable_die): Likewise.
13001 (gen_field_die): Likewise.
13002 (gen_ptr_to_mbr_type_die): Likewise.
13003 (gen_struct_or_union_type_die): Likewise.
13004 (gen_subroutine_type_die): Likewise.
13005 (gen_typedef_die): Likewise.
13006 (base_type_cmp): Compare alignment attribute.
13008 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13011 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
13012 (setb_unsigned) New pattern for setb with CCUNS.
13013 * config/rs6000/rs6000.c (expand_block_compare): Use a different
13014 subfc./subfe sequence to avoid overflow problems. Generate a
13015 shorter sequence with cmpld/setb for power9.
13016 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
13017 for generating subfc. instruction.
13018 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
13019 now uses this instruction.
13021 2017-01-30 Ian Lance Taylor <iant@google.com>
13024 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
13025 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
13027 2017-01-30 Martin Sebor <msebor@redhat.com>
13029 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
13030 Move constant to the right of a relational operator.
13031 (get_mpfr_format_length, format_character, format_string): Ditto.
13032 (should_warn_p, maybe_warn): Same.
13034 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
13036 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
13039 * asan.c (get_translation_unit_decl): Remove function.
13040 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
13042 2017-01-30 Martin Liska <mliska@suse.cz>
13044 PR gcov-profile/79259
13045 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
13046 -fprofile-generate.
13048 2017-01-30 Martin Liska <mliska@suse.cz>
13051 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
13052 Initialize variables with NULL value.
13054 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
13057 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
13059 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
13061 2017-01-30 Richard Biener <rguenther@suse.de>
13063 PR tree-optimization/79276
13064 * tree-vrp.c (process_assert_insertions): Properly adjust common
13065 when removing a duplicate.
13067 * gcc.dg/torture/pr79276.c: New testcase.
13069 2017-01-30 Richard Biener <rguenther@suse.de>
13071 PR tree-optimization/79256
13072 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
13073 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
13075 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
13077 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
13080 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
13081 ("*r<noxa>sbg_<mode>_sll_bitmask")
13082 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
13083 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
13084 Use contiguous_bitmask_nowrap_operand.
13086 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13089 * config/rs6000/altivec.h (vec_xl): Revise #define.
13090 (vec_xst): Likewise.
13092 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
13094 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
13096 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
13098 PR rtl-optimization/79194
13099 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
13100 traps before call to bypass_conditional_jumps.
13102 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
13104 PR tree-optimization/71374
13105 * lra-constraints.c (check_conflict_input_operands): New.
13106 (match_reload): Use it.
13108 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
13111 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
13112 account to calculate conflict_set.
13114 2017-01-27 Bin Cheng <bin.cheng@arm.com>
13116 PR rtl-optimization/78559
13117 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
13118 other_insn in combine.
13120 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
13122 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
13123 uint16_type_node for BT_UINT16.
13125 2017-01-27 David Malcolm <dmalcolm@redhat.com>
13127 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
13128 "RTL Tests" to menu.
13129 (GIMPLE Tests): New node.
13130 (RTL Tests): New node.
13132 2017-01-27 Richard Biener <rguenther@suse.de>
13134 PR tree-optimization/79245
13135 * tree-loop-distribution.c (distribute_loop): Apply cost
13136 modeling also to detected patterns.
13138 2017-01-27 Richard Biener <rguenther@suse.de>
13140 PR tree-optimization/71433
13141 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
13142 (compare_assert_loc): New function.
13143 (process_assert_insertions): Sort and optimize assert locations
13144 to remove duplicates and push down identical assertions on
13145 edges to their destination block.
13147 2017-01-27 Richard Biener <rguenther@suse.de>
13149 PR tree-optimization/79244
13150 * tree-vrp.c (remove_range_assertions): Forcefully propagate
13151 out SSA names even if abnormal.
13153 2017-01-27 Jakub Jelinek <jakub@redhat.com>
13155 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
13156 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
13157 instead of MPFR_RNDN.
13159 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
13162 * arm.c (arm_option_override): Don't call build_target_option_node
13163 until after doing all option overrides.
13164 (arm_valid_target_attribute_tree): Likewise.
13166 2017-01-27 Martin Liska <mliska@suse.cz>
13168 * doc/invoke.texi (-fprofile-arcs): Document profiling support
13169 for {cd}tors and C++ {cd}tors.
13171 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
13173 * config/s390/s390.md ("*setmem_long_and")
13174 ("*setmem_long_and_31z"): Use zero_extend instead of and.
13176 2017-01-26 Martin Sebor <msebor@redhat.com>
13178 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
13181 2017-01-26 Martin Sebor <msebor@redhat.com>
13183 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
13184 HAVE_DFmode before using XFmode or DFmode.
13185 (parse_directive): Avoid using the z length modifier to avoid
13186 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
13188 PR middle-end/78703
13189 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
13190 to accept adjustment as an array.
13191 (get_int_range): New function.
13192 (struct directive): Make width and prec arrays.
13193 (directive::set_width, directive::set_precision): Call get_int_range.
13194 (format_integer, format_floating): Handle width and precision ranges.
13195 (format_string, parse_directive): Same.
13197 2017-01-26 Jakub Jelinek <jakub@redhat.com>
13200 * dwarf2out.c (generate_skeleton_bottom_up): For children with
13201 comdat_type_p set, just clone them, but keep the children in the
13205 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
13206 which have direct callers with -fvar-tracking-assignments enabled
13208 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
13209 inside of type units.
13211 2017-01-26 Martin Sebor <msebor@redhat.com>
13213 PR middle-end/78703
13214 * gimple-ssa-sprintf.c (struct result_range): Add likely and
13216 (struct format_result): Replace number_chars, number_chars_min,
13217 and number_chars_max with a single member of struct result_range.
13219 (format_result::operator+=): Adjust.
13220 (struct fmtresult): Remove bounded. Handle likely and unlikely
13222 (fmtresult::adjust_for_width_or_precision): New function.
13223 (fmtresult:type_max_digits): New function.
13224 (bytes_remaining): Handle likely and unlikely counters.
13225 (min_bytes_remaining): Remove.
13226 (format_percent): Simplify.
13227 (format_integer, format_floating): Set likely and unlikely counters.
13228 (get_string_length, format_character, format_string): Same.
13229 (format_plain, should_warn_p): New function.
13230 (maybe_warn): Call should_warn_p. Update diagnostic messages
13231 and handle those for all directives, including plain strings.
13232 (format_directive): Handle likely and unlikely counters.
13233 Remove unnecessary quoting from diagnostics. Add an informational
13235 (add_bytes): Remove.
13236 (pass_sprintf_length::compute_format_length): Simplify.
13237 (try_substitute_return_value): Handle likely and unlikely counters.
13239 2017-01-26 Carl Love <cel@us.ibm.com>
13241 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
13242 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
13244 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
13247 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
13248 endianess for subregs into account.
13249 * lra-constraints.c (lra_constraints): Do risky transformations
13250 always on the first iteration.
13251 * lra-lives.c (check_pseudos_live_through_calls): Add arg
13252 last_call_used_reg_set.
13253 (process_bb_lives): Define and use last_call_used_reg_set.
13254 * lra.c (lra): Always continue after lra_constraints on the first
13257 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
13259 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
13261 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
13263 2017-01-26 Jakub Jelinek <jakub@redhat.com>
13265 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
13266 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
13267 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
13268 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
13269 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
13270 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
13271 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
13272 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
13273 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
13275 2017-01-26 Marek Polacek <polacek@redhat.com>
13278 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
13279 for the third operand.
13281 2017-01-26 Jakub Jelinek <jakub@redhat.com>
13283 PR middle-end/79236
13284 * omp-low.c (struct omp_context): Add simt_stmt field.
13285 (scan_omp_for): Return omp_context *.
13286 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
13287 context to the _simt_ SIMD stmt.
13288 (lower_omp_for): For combined SIMD with sibling _simt_
13289 SIMD, make sure to use the same decls in _looptemp_
13290 clauses as in the sibling.
13292 2017-01-26 David Sherwood <david.sherwood@arm.com>
13294 PR middle-end/79212
13295 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
13298 2017-01-26 Jakub Jelinek <jakub@redhat.com>
13301 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
13302 emit fld b; fld a; if possible.
13304 * brig-builtins.def: Update copyright years.
13305 * config/arm/arm_acle_builtins.def: Update copyright years.
13307 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13310 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
13311 constraint instead of o for the stxsd instruction.
13313 2017-01-25 Carl Love <cel@us.ibm.com>
13315 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
13316 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
13318 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
13320 * doc/invoke.texi (C++ Dialect Options): Fix typo.
13322 2017-01-25 Richard Biener <rguenther@suse.de>
13324 PR tree-optimization/69264
13325 * target.def (vector_alignment_reachable): Improve documentation.
13326 * doc/tm.texi: Regenerate.
13327 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
13329 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
13330 earlier changes with respect to TYPE_USER_ALIGN.
13331 (vector_alignment_reachable_p): Likewise. Improve dumping.
13333 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13336 * config/arm/arm.md (xordi3): Force constant operand into a register
13339 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13341 * doc/invoke.texi (-fstore-merging): Correct default optimization
13342 levels at which it is enabled.
13343 (-O): Move -fstore-merging from list to...
13346 2017-01-25 Richard Biener <rguenther@suse.de>
13349 * omp-expand.c: Include debug.h.
13350 (expand_omp_taskreg): Make sure to generate early debug before
13351 outlining anything from a function.
13352 (expand_omp_target): Likewise.
13353 (grid_expand_target_grid_body): Likewise.
13355 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
13358 * asan.c (get_translation_unit_decl): New function.
13359 (asan_add_global): Extract modules file name from globals
13360 TRANSLATION_UNIT_DECL name.
13362 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
13365 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
13366 for long calls with APCS frame and VFP.
13368 2017-01-24 David Malcolm <dmalcolm@redhat.com>
13370 * cfg.c (original_copy_tables_initialized_p): New function.
13371 * cfg.h (original_copy_tables_initialized_p): New decl.
13372 * cfgrtl.c (relink_block_chain): Guard the call to
13373 free_original_copy_tables with a call to
13374 original_copy_tables_initialized_p.
13375 * cgraph.h (symtab_node::native_rtl_p): New decl.
13376 * cgraphunit.c (symtab_node::native_rtl_p): New function.
13377 (symtab_node::needed_p): Don't assert for early assembly output
13378 for __RTL functions.
13379 (cgraph_node::finalize_function): Set "force_output" for __RTL
13381 (cgraph_node::analyze): Bail out early for __RTL functions.
13382 (analyze_functions): Update assertion to support __RTL functions.
13383 (cgraph_node::expand): Bail out early for __RTL functions.
13384 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
13386 * function.h (struct function): Update comment for field
13388 * gimple-expr.c: Include "tree-pass.h".
13389 (gimple_has_body_p): Return false for __RTL functions.
13390 * Makefile.in (OBJS): Add run-rtl-passes.o.
13391 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
13393 (gcc::pass_manager::get_clean_slate): New accessor.
13394 * passes.c: Include "insn-addr.h".
13395 (should_skip_pass_p): Add logging. Update logic for running
13396 "expand" to be compatible with both __GIMPLE and __RTL. Guard
13397 property-provider override so it is only done for gimple passes.
13399 (skip_pass): New function.
13400 (execute_one_pass): Call skip_pass when skipping passes.
13401 * read-md.c (md_reader::read_char): Support filtering
13402 the input to a subset of line numbers.
13403 (md_reader::md_reader): Initialize fields
13404 m_first_line and m_last_line.
13405 (md_reader::read_file_fragment): New function.
13406 * read-md.h (md_reader::read_file_fragment): New decl.
13407 (md_reader::m_first_line): New field.
13408 (md_reader::m_last_line): New field.
13409 * read-rtl-function.c (function_reader::create_function): Only
13410 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
13411 curr_properties. Set DECL_INITIAL to a dummy block.
13412 (read_rtl_function_body_from_file_range): New function.
13413 * read-rtl-function.h (read_rtl_function_body_from_file_range):
13415 * run-rtl-passes.c: New file.
13416 * run-rtl-passes.h: New file.
13418 2017-01-24 Jeff Law <law@redhat.com>
13420 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
13423 2017-01-24 Bin Cheng <bin.cheng@arm.com>
13425 PR tree-optimization/79159
13426 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
13427 (record_nonwrapping_iv): Improve boundary using above function if no
13428 value range information.
13430 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
13431 Martin Jambor <mjambor@suse.cz>
13433 * brig-builtins.def: New file.
13434 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
13435 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
13436 (DEF_HSAIL_SAT_BUILTIN): Likewise.
13437 (DEF_HSAIL_INTR_BUILTIN): Likewise.
13438 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
13439 * builtin-types.def (BT_INT8): New.
13440 (BT_INT16): Likewise.
13441 (BT_UINT8): Likewise.
13442 (BT_UINT16): Likewise.
13443 (BT_FN_ULONG): Likewise.
13444 (BT_FN_UINT_INT): Likewise.
13445 (BT_FN_UINT_ULONG): Likewise.
13446 (BT_FN_UINT_LONG): Likewise.
13447 (BT_FN_UINT_PTR): Likewise.
13448 (BT_FN_ULONG_PTR): Likewise.
13449 (BT_FN_INT8_FLOAT): Likewise.
13450 (BT_FN_INT16_FLOAT): Likewise.
13451 (BT_FN_UINT32_FLOAT): Likewise.
13452 (BT_FN_UINT16_FLOAT): Likewise.
13453 (BT_FN_UINT8_FLOAT): Likewise.
13454 (BT_FN_UINT64_FLOAT): Likewise.
13455 (BT_FN_UINT16_UINT32): Likewise.
13456 (BT_FN_UINT32_UINT16): Likewise.
13457 (BT_FN_UINT16_UINT16_UINT16): Likewise.
13458 (BT_FN_INT_PTR_INT): Likewise.
13459 (BT_FN_UINT_PTR_UINT): Likewise.
13460 (BT_FN_LONG_PTR_LONG): Likewise.
13461 (BT_FN_ULONG_PTR_ULONG): Likewise.
13462 (BT_FN_VOID_UINT64_UINT64): Likewise.
13463 (BT_FN_UINT8_UINT8_UINT8): Likewise.
13464 (BT_FN_INT8_INT8_INT8): Likewise.
13465 (BT_FN_INT16_INT16_INT16): Likewise.
13466 (BT_FN_INT_INT_INT): Likewise.
13467 (BT_FN_UINT_FLOAT_UINT): Likewise.
13468 (BT_FN_FLOAT_UINT_UINT): Likewise.
13469 (BT_FN_ULONG_UINT_UINT): Likewise.
13470 (BT_FN_ULONG_UINT_PTR): Likewise.
13471 (BT_FN_ULONG_ULONG_ULONG): Likewise.
13472 (BT_FN_UINT_UINT_UINT): Likewise.
13473 (BT_FN_VOID_UINT_PTR): Likewise.
13474 (BT_FN_UINT_UINT_PTR: Likewise.
13475 (BT_FN_UINT32_UINT64_PTR): Likewise.
13476 (BT_FN_INT_INT_UINT_UINT): Likewise.
13477 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
13478 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
13479 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
13480 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
13481 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
13482 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
13483 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
13484 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
13485 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
13486 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
13487 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
13488 * doc/frontends.texi: List BRIG FE.
13489 * doc/install.texi (Testing): Add BRIG tesring requirements.
13490 * doc/invoke.texi (Overall Options): Mention BRIG.
13491 * doc/standards.texi (Standards): Doucment BRIG HSA version.
13493 2017-01-24 Richard Biener <rguenther@suse.de>
13495 PR translation/79208
13496 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
13498 2017-01-24 Martin Jambor <mjambor@suse.cz>
13501 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
13502 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
13503 and known_contexts.
13505 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
13507 PR middle-end/79123
13508 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
13509 casts from signed to unsigned really don't have a range.
13511 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
13513 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
13514 GMP_RNDx for compatiblity.
13516 2017-01-24 Martin Liska <mliska@suse.cz>
13519 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
13520 that would prevent us to call alloca with -1 as argument.
13522 2017-01-24 Jakub Jelinek <jakub@redhat.com>
13524 * dwarf2out.c (output_compilation_unit_header, output_file_names):
13525 Avoid -Wformat-security warning.
13527 2017-01-23 Andrew Pinski <apinski@cavium.com>
13529 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
13532 2017-01-23 Martin Sebor <msebor@redhat.com>
13534 PR middle-end/78703
13535 * gimple-ssa-sprintf.c (warn_level): New global.
13536 (format_integer): Use it here and throughout the rest of the file.
13537 Use the same switch to compute sign as base.
13538 (maybe_warn): New function.
13539 (format_directive): Factor out warnings into maybe_warn.
13540 Add debugging output. Use warn_level.
13541 (add_bytes): Use warn_level.
13542 (pass_sprintf_length::compute_format_length): Add debugging output.
13543 (try_substitute_return_value): Same.
13544 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
13546 PR middle-end/78703
13547 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
13548 (struct fmtresult, format_integer, format_floating): Adjust.
13549 (fmtresult::fmtresult): Set max correctly in two argument ctor.
13550 (get_string_length, format_string,format_directive): Same.
13551 (pass_sprintf_length::compute_format_length): Same.
13552 (try_substitute_return_value): Simplify slightly.
13554 PR middle-end/78703
13555 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
13556 (fmtresult::operator+=): Outlined.
13557 (struct fmtresult): Add ctors.
13558 (struct conversion_spec): Rename...
13559 (struct directive): ...to this. Add and remove data members.
13560 (directive::set_width, directive::set_precision): New functions.
13561 (format_percent): Use fmtresult ctor.
13562 (get_width_and_precision): Remove.
13563 (format_integer): Make naming changes. Avoid computing width and
13565 (format_floating): Same. Adjust indentation.
13566 (format_character, format_none): New functions.
13567 (format_string): Moved character handling to format_character.
13568 (format_directive): Remove arguments, change return type.
13569 (parse_directive): New function.
13570 (pass_sprintf_length::compute_format_length): Move directive
13571 parsing to parse_directive.
13573 2017-01-23 Jakub Jelinek <jakub@redhat.com>
13575 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
13576 (assign_assembler_name_if_needed): ... this.
13577 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
13578 (assign_assembler_name_if_needed): ... this.
13579 (free_lang_data_in_cgraph): Adjust callers.
13580 * cgraphunit.c (cgraph_node::analyze): Likewise.
13581 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
13584 2017-01-23 Richard Biener <rguenther@suse.de>
13586 PR tree-optimization/79088
13587 PR tree-optimization/79188
13588 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
13589 resetting loop bounds after last path deletion. Reset loop
13590 bounds of the target loop, make code match the comments.
13591 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
13592 Make sure loops need no fixups.
13594 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
13596 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
13597 exponent support with double type for first argument.
13598 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
13599 type returned by __builtin_vec_extract_sig,
13600 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
13601 functions from "vector int" to "vector unsigned int" or from
13602 "vector long long int" to "vector unsigned long long int".
13603 Changed type returned by __builtin_vec_extract_exp,
13604 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
13605 functions from "vector int" to "vector unsigned int" or from
13606 "vector long long int" to "vector unsigned long long int".
13607 Changed return type of __builtin_vec_test_data_class,
13608 __builtin_vec_test_data_class_sp, and
13609 __builtin_vec_test_data_class_dp from "vector int" to
13610 "vector bool int" or from "vector long long int" to "vector bool
13611 long long int" and changed second argument type from "unsigned
13612 int" to "int". Added new overloaded function forms "vector float
13613 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
13614 "vector float __builtin_vec_insert_exp_sp (vector float, vector
13615 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
13616 double, vector unsigned long long int)" and "vector double
13617 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
13618 long int)". Changed return type of
13619 __builtin_scalar_test_data_class and
13620 __builtin_scalar_test_data_class_sp and
13621 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
13622 int" and changed second argument from "unsigned int" to "int".
13623 Changed type returned by __builtin_scalar_test_neg,
13624 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
13625 from "int" to "bool int". Added new overloaded function form
13626 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
13627 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
13628 exponent double-precision with floating point first argument.
13629 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
13630 documentation of scalar_test_data_class, scalar_test_neg,
13631 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
13632 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
13633 vec_test_data_class built-in functions to reflect refinements in
13634 their type signatures.
13636 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
13638 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
13640 (aarch64_elf_asm_destructor): Likewise.
13642 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
13644 PR rtl-optimization/78634
13645 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
13646 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
13647 * ifcvt.c (noce_try_cmove): Add missing cost check.
13649 PR rtl-optimization/71724
13650 * combine.c (if_then_else_cond): Look for situations where it is
13651 beneficial to undo the work of one of the recursive calls.
13653 2017-01-23 Bin Cheng <bin.cheng@arm.com>
13655 PR tree-optimization/70754
13656 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
13657 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
13658 combined stmt before it if not NULL.
13659 (combine_chains): Process refs reversely and compute dominance point
13662 2017-01-23 Martin Liska <mliska@suse.cz>
13664 PR tree-optimization/79196
13665 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
13666 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
13668 (strlen_optimize_stmt): Call the renamed function.
13670 2017-01-23 Michael Matz <matz@suse.de>
13672 PR tree-optimization/78384
13673 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
13675 2017-01-23 Richard Biener <rguenther@suse.de>
13677 PR tree-optimization/79186
13678 * tree-vrp.c (register_new_assert_for): Make sure we've seen
13679 both incoming edges before moving an assert.
13681 2017-01-23 Martin Jambor <mjambor@suse.cz>
13683 * ipa-prop.c (load_from_param_1): Removed.
13684 (load_from_unmodified_param): Bits from load_from_param_1 put back
13686 (load_from_param): Removed.
13687 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
13688 with stmt. Reverted back to use of load_from_unmodified_param.
13690 2017-01-23 Martin Jambor <mjambor@suse.cz>
13693 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
13694 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
13695 field a pointer to garbage collected vector, mark lattices and
13696 ipcp_orig_node with GTY((skip)).
13697 (ipa_get_param_count): Adjust to descriptors being a pointer.
13698 (ipa_get_param): Likewise.
13699 (ipa_get_type): Likewise.
13700 (ipa_get_param_move_cost): Likewise.
13701 (ipa_set_param_used): Likewise.
13702 (ipa_get_controlled_uses): Likewise.
13703 (ipa_set_controlled_uses): Likewise.
13704 (ipa_is_param_used): Likewise.
13705 (ipa_node_params_t): Move into garbage collector. New methods insert
13707 (ipa_node_params_sum): Annotate wth GTY(()).
13708 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
13710 (ipa_load_from_parm_agg): Adjust declaration.
13711 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
13712 * ipa-profile.c (ipa_profile): Likewise.
13713 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
13714 (ipa_populate_param_decls): Make descriptors parameter garbage
13716 (ipa_dump_param): Adjust to descriptors being a pointer.
13717 (ipa_alloc_node_params): Likewise.
13718 (ipa_initialize_node_params): Likewise.
13719 (load_from_param_1): Make descriptors parameter garbage collected.
13720 (load_from_unmodified_param): Likewise.
13721 (load_from_param): Likewise.
13722 (ipa_load_from_parm_agg): Likewise.
13723 (ipa_node_params::~ipa_node_params): Removed.
13724 (ipa_free_all_node_params): Remove call to delete operator.
13725 (ipa_node_params_t::insert): New.
13726 (ipa_node_params_t::remove): Likewise.
13727 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
13728 copy known_csts and known_contexts vectors.
13729 (ipa_read_node_info): Adjust to descriptors being a pointer.
13730 (ipcp_modif_dom_walker): Make m_descriptors field garbage
13732 (ipcp_transform_function): Make descriptors variable garbage
13735 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
13737 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
13738 * config/i386/avx512dqintrin.h: Ditto.
13739 * config/i386/avx512fintrin.h: Ditto.
13740 * gcc/config/i386/i386.c: Handle new builtins.
13741 * config/i386/i386-builtin.def: Add new builtins.
13742 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
13743 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
13745 2017-01-23 Jakub Jelinek <jakub@redhat.com>
13746 Martin Liska <mliska@suse.cz>
13748 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
13749 * asan.c (asan_expand_poison_ifn): Support stores and use
13750 appropriate ASAN report function.
13751 * internal-fn.c (expand_ASAN_POISON_USE): New function.
13752 * internal-fn.def (ASAN_POISON_USE): Declare.
13753 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
13754 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
13755 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
13756 ASAN_POISON calls w/o LHS.
13757 * tree-ssa.c (execute_update_addresses_taken): Create clobber
13758 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
13759 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
13760 * gimplify.c (asan_poison_variables): Add attribute
13761 use_after_scope_memory to variables that really needs to live
13763 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
13764 having the attribute.
13766 2017-01-23 Martin Liska <mliska@suse.cz>
13768 * asan.c (create_asan_shadow_var): New function.
13769 (asan_expand_poison_ifn): Likewise.
13770 * asan.h (asan_expand_poison_ifn): New declaration.
13771 * internal-fn.c (expand_ASAN_POISON): Likewise.
13772 * internal-fn.def (ASAN_POISON): New builtin.
13773 * sanopt.c (pass_sanopt::execute): Expand
13774 asan_expand_poison_ifn.
13775 * tree-inline.c (copy_decl_for_dup_finish): Make function
13777 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
13778 * tree-ssa.c (is_asan_mark_p): New function.
13779 (execute_update_addresses_taken): Rewrite local variables
13780 (identified just by use-after-scope as addressable) into SSA.
13782 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
13784 * doc/install.texi (Specific): opensource.apple.com uses https
13785 now. Remove trailing slash.
13787 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
13789 * README.Portability: Remove note on an Irix compatibility issue.
13791 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
13793 * gcov.c (INCLUDE_ALGORITHM): Define.
13794 (INCLUDE_VECTOR): Define.
13795 No longer include <vector> and <algorithm> directly.
13797 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
13799 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
13801 * doc/invoke.texi (Code Gen Options): Ditto.
13803 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
13806 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
13808 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
13810 rtl-optimization/79125
13811 * cprop.c (local_cprop_pass): Handle cases where we make an
13812 unconditional trap.
13814 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
13818 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
13819 read from, for big endian.
13821 2017-01-20 Jiong Wang <jiong.wang@arm.com>
13823 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
13824 register pauth builtins for LP64 only.
13826 2017-01-20 Marek Polacek <polacek@redhat.com>
13829 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
13832 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
13834 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
13836 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
13837 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
13838 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
13840 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13843 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
13844 in big-endian mode when they are not a single duplicated value.
13846 2017-01-20 Richard Biener <rguenther@suse.de>
13848 * BASE-VER: Bump to 7.0.1.
13850 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
13852 * omp-low.c (omplow_simd_context): New struct. Use it...
13853 (lower_rec_simd_input_clauses): ...here and...
13854 (lower_rec_input_clauses): ...here to hold common data. Adjust all
13855 references to idx, lane, max_vf, is_simt.
13857 2017-01-20 Graham Markall <graham.markall@embecosm.com>
13859 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
13862 2017-01-20 Martin Jambor <mjambor@suse.cz>
13864 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
13865 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
13867 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
13868 (GTFILES): Rename hsa.c to hsa-common.c.
13869 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
13870 * hsa-dump.c: Likewise.
13871 * hsa-gen.c: Likewise.
13872 * hsa-regalloc.c: Likewise.
13873 * ipa-hsa.c: Likewise.
13874 * omp-expand.c: Likewise.
13875 * omp-low.c: Likewise.
13876 * toplev.c: Likewise.
13878 2017-01-20 Marek Polacek <polacek@redhat.com>
13881 * doc/invoke.texi: Document -Wduplicated-branches.
13882 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
13883 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
13884 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
13885 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
13886 return 0 only when not OEP_LEXICOGRAPHIC.
13887 (fold_build_cleanup_point_expr): Use the expression
13888 location when building CLEANUP_POINT_EXPR.
13889 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
13890 * tree.c (add_expr): Handle error_mark_node.
13892 2017-01-20 Martin Liska <mliska@suse.cz>
13895 * tree-profile.c (init_ic_make_global_vars): Do not call
13897 (gimple_init_gcov_profiler): Likewise.
13899 2017-01-20 Martin Liska <mliska@suse.cz>
13902 * cgraph.h (maybe_create_reference): Remove argument and
13904 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
13906 * ipa-cp.c (create_specialized_node): Likewise.
13907 * symtab.c (symtab_node::maybe_create_reference): Handle
13908 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
13910 2017-01-20 Martin Liska <mliska@suse.cz>
13912 * read-rtl-function.c (function_reader::create_function): Use
13913 build_decl instread of build_decl_stat.
13915 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
13917 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
13918 * config/i386/avx512dqintrin.h: Ditto.
13919 * config/i386/avx512fintrin.h: Ditto.
13920 * config/i386/i386-builtin-types.def: Add new types.
13921 * gcc/config/i386/i386.c: Handle new types.
13922 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
13923 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
13924 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
13925 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
13926 (__builtin_ia32_kshiftridi): New.
13927 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
13929 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
13933 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
13934 define to rs6000_init_stack_protect_guard.
13935 (rs6000_init_stack_protect_guard): New function.
13937 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
13938 Yunqiang Su <yunqiang.su@imgtec.com>
13940 * config.gcc (supported_defaults): Add madd4.
13941 (with_madd4): Add validation.
13942 (all_defaults): Add madd4.
13943 * config/mips/mips.opt (mmadd4): New option.
13944 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
13946 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
13948 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
13949 (ISA_HAS_FUSED_MADD4): Likewise.
13950 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
13951 * gcc/doc/install.texi (--with-madd4): Document the new option.
13953 2017-01-19 Jiong Wang <jiong.wang@arm.com>
13955 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
13956 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
13957 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
13958 (aarch64_init_pauth_hint_builtins): New.
13959 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
13960 (aarch64_expand_builtin): Expand new builtins.
13962 2017-01-19 Jiong Wang <jiong.wang@arm.com>
13964 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
13965 * combine-stack-adj.c (no_unhandled_cfa): Handle
13966 REG_CFA_TOGGLE_RA_MANGLE.
13967 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
13968 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
13969 info for return address signing.
13970 (aarch64_expand_epilogue): Likewise.
13972 2017-01-19 Jiong Wang <jiong.wang@arm.com>
13974 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
13975 * config/aarch64/aarch64-protos.h
13976 (aarch64_return_address_signing_enabled): New declaration.
13977 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
13979 (aarch64_expand_prologue): Sign return address before it's pushed onto
13981 (aarch64_expand_epilogue): Authenticate return address fetched from
13983 (aarch64_override_options): Sanity check for ILP32 and ISA level.
13984 (aarch64_attributes): New function attributes for "sign-return-address".
13985 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
13986 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
13987 ("*do_return"): Generate combined instructions according to key index.
13988 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
13989 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
13991 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
13992 * config/aarch64/aarch64.opt (msign-return-address=): New.
13993 * doc/extend.texi (AArch64 Function Attributes): Documents
13994 "sign-return-address=".
13995 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
13997 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
13999 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
14000 overall option summary.
14002 2017-01-19 Jiong Wang <jiong.wang@arm.com>
14004 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
14005 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
14006 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
14007 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
14009 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
14011 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
14012 -mpower9-minmax by default for -mcpu=power9.
14013 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
14014 128-bit floating point.
14016 2017-01-20 Alan Modra <amodra@gmail.com>
14018 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
14019 optimizing for size.
14021 2017-01-20 Alan Modra <amodra@gmail.com>
14024 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
14025 for strcmp and strncmp from corresponding builtin decl.
14027 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
14029 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
14030 instead of i386/rtems-64.h.
14031 * config/i386/rtems-64.h: Remove.
14033 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
14037 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
14039 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
14041 2017-01-19 Tamar Christina <tamar.christina@arm.com>
14043 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
14044 Change int to HOST_WIDE_INT.
14045 * config/aarch64/aarch64-protos.h
14046 (aarch64_simd_gen_const_vector_dup): Likewise.
14047 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
14049 2017-01-19 David Malcolm <dmalcolm@redhat.com>
14051 * langhooks-def.h (lhd_type_for_size): New decl.
14052 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
14053 * langhooks.c (lhd_type_for_size): New function, taken from
14056 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
14058 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
14059 define_bypass for CR latency.
14060 (power9-cracked-alu): Update bypass latency and remove power9-branch.
14061 (power9-alu2): Add define_bypass for CR latency.
14063 (power9-mul): Update insn latency.
14064 (power9-mul-compare): Update insn latency, bypass latency and remove
14067 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14069 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
14071 * config/aarch64/aarch64.md
14072 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
14073 aarch64_nopcrelative_literal_loads.
14074 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
14076 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
14078 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
14079 TARGET_LOONGSON_3A.
14080 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
14082 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
14085 * config.gcc (supported_defaults): Add lxc1-sxc1.
14086 (with_lxc1_sxc1): Add validation.
14087 (all_defaults): Add lxc1-sxc1.
14088 * config/mips/mips.opt (mlxc1-sxc1): New option.
14089 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
14091 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
14092 __mips_no_lxc1_sxc1.
14093 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
14094 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
14095 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
14097 2017-01-19 Richard Biener <rguenther@suse.de>
14099 PR tree-optimization/72488
14100 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
14101 sure to restore SSA info.
14102 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
14104 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
14106 PR rtl-optimization/79121
14107 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
14108 of the inner type when shifting an extended value.
14110 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
14113 * symtab.c (symtab_node::equal_address_to): Fix comparing of
14114 interposable aliases.
14116 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
14119 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
14120 Use the evmergelohi instruction.
14121 (mov_si<mode>_e500_subreg4_2_le): Likewise.
14122 (mov_sitf_e500_subreg8_2_be): Likewise.
14123 (mov_sitf_e500_subreg12_2_le): Likewise.
14124 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
14125 (mov_si<mode>_e500_subreg4_2_be): Likewise.
14126 (mov_sitf_e500_subreg8_2_le): Likewise.
14127 (mov_sitf_e500_subreg12_2_be): Likewise.
14129 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14131 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
14132 attribute from vecsimple to vecperm.
14133 (altivec_vbpermq2): Likewise.
14135 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14138 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
14140 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14141 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
14142 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
14143 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
14144 case where N arg is SIZE_MAX.
14145 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
14146 (cmpstrsi): Add pattern.
14148 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
14150 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14151 __builtin_vec_revb builtins.
14152 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
14153 built-in functions to support generation of the ISA 3.0 XXBR<x>
14154 vector byte reverse instructions.
14155 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
14156 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
14157 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
14158 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
14159 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
14160 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
14161 (P9V_BUILTIN_VEC_REVB): Likewise.
14162 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
14163 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
14164 (p9_xxbrq_v16qi): Likewise.
14165 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
14166 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
14167 (p9_xxbrh_v8hi): Likewise.
14168 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
14169 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
14170 vec_revb built-in functions.
14172 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
14174 PR rtl-optimization/78952
14175 * config/i386/i386.md (any_extract): New code iterator.
14176 (*insvqi_2): Use any_extract for source operand.
14177 (*insvqi_3): Use any_shiftrt for source operand.
14179 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
14181 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
14183 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
14185 2017-01-18 Matthias Klose <doko@ubuntu.com>
14187 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
14189 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14191 * config/rs6000/altivec.h (vec_bperm): Change #define.
14192 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
14193 (altivec_vbpermq2): New define_insn.
14194 (altivec_vbpermd): Likewise.
14195 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
14196 function interface.
14197 (VBPERMD): Likewise.
14198 (VBPERM): New polymorphic function interface.
14199 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
14200 Add entries for P9V_BUILTIN_VEC_VBPERM.
14201 * doc/extend.texi: Add interfaces for vec_bperm.
14203 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14205 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
14206 first letter of error messages.
14207 (s390_resolve_overloaded_builtin): Likewise.
14208 * config/s390/s390.c (s390_expand_builtin): Likewise.
14209 (s390_invalid_arg_for_unprototyped_fn): Likewise.
14210 (s390_valid_target_attribute_inner_p): Likewise.
14211 * config/s390/s390.md ("tabort"): Likewise.
14213 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
14215 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
14216 (ISA_AVOID_DIV_HILO): New macro.
14217 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
14218 (ISA_HAS_DDIV): Likewise.
14220 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
14222 * doc/invoke.texi (fabi-version): Correct number of occurrences.
14224 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
14226 * doc/invoke.texi (fabi-version): Spelling fix.
14228 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
14231 * doc/invoke.texi (fabi-version): Mention mangling fix for
14234 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
14237 * doc/invoke.texi (fabi-version): Document discriminator mangling.
14239 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
14242 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
14243 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
14245 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
14247 (stack_protect_test): Ditto.
14248 * config/rs6000/rs6000.opt (mstack-protector-guard=,
14249 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
14251 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
14252 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
14253 -mstack-protector-guard-offset=.
14254 (RS/6000 and PowerPC Options): Ditto.
14256 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
14258 * config/i386/i386.h (MASK_CLASS_P): New define.
14259 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
14260 there are no registers from different register sets also when
14261 mask registers are used. Update function comment.
14262 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
14263 to (*k/*r) and (*k/*km) alternatives.
14265 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
14267 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
14268 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
14269 (EH_RETURN_HANDLER_RTX): New define.
14270 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
14271 Force frame pointer in EH return functions.
14272 (aarch64_expand_epilogue): Add barrier for eh_return.
14273 (aarch64_final_eh_return_addr): Remove.
14274 (aarch64_eh_return_handler_rtx): New function.
14275 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
14277 (aarch64_eh_return_handler_rtx): New prototype.
14279 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14281 * config/rs6000/altivec.h (vec_rlmi): New #define.
14282 (vec_vrlnm): Likewise.
14283 (vec_rlnm): Likewise.
14284 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
14285 (UNSPEC_VRLNM): Likewise.
14286 (VIlong): New mode iterator.
14287 (altivec_vrl<VI_char>mi): New define_insn.
14288 (altivec_vrl<VI_char>nm): Likewise.
14289 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
14291 (VRLDNM): Likewise.
14292 (RLNM): New polymorphic function entry.
14293 (VRLWMI): New monomorphic function entry.
14294 (VRLDMI): Likewise.
14295 (RLMI): New polymorphic function entry.
14296 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
14297 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
14298 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
14301 2017-01-17 Jakub Jelinek <jakub@redhat.com>
14304 * dwarf2out.c (field_byte_offset): Restore the
14305 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
14306 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
14307 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
14310 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
14313 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
14315 2017-01-17 Jakub Jelinek <jakub@redhat.com>
14318 * dwarf2out.c (add_data_member_location_attribute): For constant
14319 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
14320 instead of DW_AT_data_member_location, DW_AT_bit_offset and
14321 DW_AT_byte_size attributes.
14323 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
14325 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
14326 after forcing to constant memory when the code model is medium.
14328 2017-01-17 Julia Koval <julia.koval@intel.com>
14331 * config/i386/avx512fintrin.h
14332 (_mm512_i32gather_ps): Change __addr type to void const*.
14333 (_mm512_mask_i32gather_ps): Ditto.
14334 (_mm512_i32gather_pd): Ditto.
14335 (_mm512_mask_i32gather_pd): Ditto.
14336 (_mm512_i64gather_ps): Ditto.
14337 (_mm512_mask_i64gather_ps): Ditto.
14338 (_mm512_i64gather_pd): Ditto.
14339 (_mm512_mask_i64gather_pd): Ditto.
14340 (_mm512_i32gather_epi32): Ditto.
14341 (_mm512_mask_i32gather_epi32): Ditto.
14342 (_mm512_i32gather_epi64): Ditto.
14343 (_mm512_mask_i32gather_epi64): Ditto.
14344 (_mm512_i64gather_epi32): Ditto.
14345 (_mm512_mask_i64gather_epi32): Ditto.
14346 (_mm512_i64gather_epi64): Ditto.
14347 (_mm512_mask_i64gather_epi64): Ditto.
14348 (_mm512_i32scatter_ps): Change __addr type to void*.
14349 (_mm512_mask_i32scatter_ps): Ditto.
14350 (_mm512_i32scatter_pd): Ditto.
14351 (_mm512_mask_i32scatter_pd): Ditto.
14352 (_mm512_i64scatter_ps): Ditto.
14353 (_mm512_mask_i64scatter_ps): Ditto.
14354 (_mm512_i64scatter_pd): Ditto.
14355 (_mm512_mask_i64scatter_pd): Ditto.
14356 (_mm512_i32scatter_epi32): Ditto.
14357 (_mm512_mask_i32scatter_epi32): Ditto.
14358 (_mm512_i32scatter_epi64): Ditto.
14359 (_mm512_mask_i32scatter_epi64): Ditto.
14360 (_mm512_i64scatter_epi32): Ditto.
14361 (_mm512_mask_i64scatter_epi32): Ditto.
14362 (_mm512_i64scatter_epi64): Ditto.
14363 (_mm512_mask_i64scatter_epi64): Ditto.
14364 * config/i386/avx512pfintrin.h
14365 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
14366 (_mm512_mask_prefetch_i32gather_ps): Ditto.
14367 (_mm512_mask_prefetch_i64gather_pd): Ditto.
14368 (_mm512_mask_prefetch_i64gather_ps): Ditto.
14369 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
14370 (_mm512_prefetch_i32scatter_ps): Ditto.
14371 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
14372 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
14373 (_mm512_prefetch_i64scatter_pd): Ditto.
14374 (_mm512_prefetch_i64scatter_ps): Ditto.
14375 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
14376 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
14377 * config/i386/avx512vlintrin.h
14378 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
14379 (_mm_mmask_i32gather_ps): Ditto.
14380 (_mm256_mmask_i32gather_pd): Ditto.
14381 (_mm_mmask_i32gather_pd): Ditto.
14382 (_mm256_mmask_i64gather_ps): Ditto.
14383 (_mm_mmask_i64gather_ps): Ditto.
14384 (_mm256_mmask_i64gather_pd): Ditto.
14385 (_mm_mmask_i64gather_pd): Ditto.
14386 (_mm256_mmask_i32gather_epi32): Ditto.
14387 (_mm_mmask_i32gather_epi32): Ditto.
14388 (_mm256_mmask_i32gather_epi64): Ditto.
14389 (_mm_mmask_i32gather_epi64): Ditto.
14390 (_mm256_mmask_i64gather_epi32): Ditto.
14391 (_mm_mmask_i64gather_epi32): Ditto.
14392 (_mm256_mmask_i64gather_epi64): Ditto.
14393 (_mm_mmask_i64gather_epi64): Ditto.
14394 (_mm256_i32scatter_ps): Change __addr type to void*.
14395 (_mm256_mask_i32scatter_ps): Ditto.
14396 (_mm_i32scatter_ps): Ditto.
14397 (_mm_mask_i32scatter_ps): Ditto.
14398 (_mm256_i32scatter_pd): Ditto.
14399 (_mm256_mask_i32scatter_pd): Ditto.
14400 (_mm_i32scatter_pd): Ditto.
14401 (_mm_mask_i32scatter_pd): Ditto.
14402 (_mm256_i64scatter_ps): Ditto.
14403 (_mm256_mask_i64scatter_ps): Ditto.
14404 (_mm_i64scatter_ps): Ditto.
14405 (_mm_mask_i64scatter_ps): Ditto.
14406 (_mm256_i64scatter_pd): Ditto.
14407 (_mm256_mask_i64scatter_pd): Ditto.
14408 (_mm_i64scatter_pd): Ditto.
14409 (_mm_mask_i64scatter_pd): Ditto.
14410 (_mm256_i32scatter_epi32): Ditto.
14411 (_mm256_mask_i32scatter_epi32): Ditto.
14412 (_mm_i32scatter_epi32): Ditto.
14413 (_mm_mask_i32scatter_epi32): Ditto.
14414 (_mm256_i32scatter_epi64): Ditto.
14415 (_mm256_mask_i32scatter_epi64): Ditto.
14416 (_mm_i32scatter_epi64): Ditto.
14417 (_mm_mask_i32scatter_epi64): Ditto.
14418 (_mm256_i64scatter_epi32): Ditto.
14419 (_mm256_mask_i64scatter_epi32): Ditto.
14420 (_mm_i64scatter_epi32): Ditto.
14421 (_mm_mask_i64scatter_epi32): Ditto.
14422 (_mm256_i64scatter_epi64): Ditto.
14423 (_mm256_mask_i64scatter_epi64): Ditto.
14424 (_mm_i64scatter_epi64): Ditto.
14425 (_mm_mask_i64scatter_epi64): Ditto.
14426 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
14427 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
14428 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
14429 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
14430 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
14431 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
14432 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
14433 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
14434 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
14435 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
14436 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
14437 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
14438 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
14439 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
14440 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
14441 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
14442 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
14443 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
14444 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
14445 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
14446 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
14447 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
14448 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
14449 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
14450 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
14451 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
14452 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
14453 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
14454 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
14455 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
14456 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
14457 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
14458 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
14459 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
14460 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
14461 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
14462 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
14463 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
14464 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
14465 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
14466 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
14467 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
14468 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
14469 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
14470 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
14471 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
14472 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
14473 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
14474 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
14475 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
14476 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
14477 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
14478 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
14479 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
14480 definitions accordingly.
14482 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
14483 Kuan-Lin Chen <kuanlinchentw@gmail.com>
14486 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
14489 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
14492 * ira-conflicts.c (ira_build_conflicts): Update total conflict
14493 hard regs for inner regno.
14495 2017-01-17 Martin Liska <mliska@suse.cz>
14498 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
14499 assumption and add comment.
14501 2017-01-17 Nathan Sidwell <nathan@acm.org>
14503 * ipa-visibility.c (localize_node): New function, broken out of ...
14504 (function_and_variable_visibility): ... here. Call it.
14506 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
14508 PR middle-end/77445
14509 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
14510 correctly set frequency of oudgoing edge.
14511 (duplicate_thread_path): Fix profile updating.
14513 2017-01-17 Jakub Jelinek <jakub@redhat.com>
14516 * configure.ac: Add GCC_BASE_VER.
14517 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
14518 version from BASE-VER file.
14519 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
14520 (gcc.o): Depend on $(BASEVER).
14521 * common.opt (dumpfullversion): New option.
14522 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
14523 * doc/invoke.texi: Document -dumpfullversion.
14524 * doc/install.texi: Document --with-gcc-major-version-only.
14525 * configure: Regenerated.
14527 2017-01-17 Richard Biener <rguenther@suse.de>
14529 PR tree-optimization/71433
14530 * tree-vrp.c (register_new_assert_for): Merge same asserts
14531 on all incoming edges.
14532 (process_assert_insertions_for): Handle insertions at the
14535 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
14537 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
14538 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
14540 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
14543 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
14546 2017-01-17 Alan Modra <amodra@gmail.com>
14549 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
14550 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
14551 symbolic stack limit when pic.
14553 2017-01-16 Martin Sebor <msebor@redhat.com>
14555 PR tree-optimization/78608
14556 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
14558 2017-01-16 Jeff Law <law@redhat.com>
14561 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
14562 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
14563 for several include directories that may be relative to sysroot.
14564 * config/i386/x-mingw32 (gplus_includedir): Define.
14565 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
14566 (native_system_includedir): Likewise.
14567 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
14568 override if TARGET_SYSTEM_ROOT is defined.
14569 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
14571 PR tree-optimization/79090
14572 PR tree-optimization/33562
14573 PR tree-optimization/61912
14574 PR tree-optimization/77485
14575 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
14576 and computed trims into the dump file.
14578 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
14580 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
14582 2017-01-16 Jakub Jelinek <jakub@redhat.com>
14585 * gimplify.c (gimplify_init_constructor): If want_value and
14586 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
14590 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
14591 sequence. Formatting fixes.
14592 (doloop_optimize): Formatting fixes.
14595 * gcc.c (debug_level_greater_than_spec_func): New function.
14596 (static_spec_functions): Add debug-level-gt spec function.
14597 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
14599 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
14600 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
14601 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
14602 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
14603 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
14604 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
14606 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
14608 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
14609 QImode fixups to general and mask registers only.
14611 2017-01-16 Carl Love <cel@us.ibm.com>
14613 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
14614 for built-in functions
14615 vector signed char vec_nabs (vector signed char)
14616 vector signed short vec_nabs (vector signed short)
14617 vector signed int vec_nabs (vector signed int)
14618 vector signed long long vec_nabs (vector signed long long)
14619 vector float vec_nabs (vector float)
14620 vector double vec_nabs (vector double)
14621 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
14623 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
14624 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
14625 * doc/extend.texi: Update the documentation file for the new built-in
14628 2017-01-16 Martin Sebor <msebor@redhat.com>
14630 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
14633 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14635 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
14636 UNSPEC_VSX__XXSPLTD to require special splat handling.
14638 2017-01-16 David Malcolm <dmalcolm@redhat.com>
14641 * system.h: Poison strndup.
14643 2017-01-16 Alan Modra <amodra@gmail.com>
14646 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
14649 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
14651 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
14653 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
14655 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
14656 call recog here. Assert that INSN_CODE (insn) is non-negative.
14658 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
14661 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
14663 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
14664 in the currently scheduled RTL fragment.
14666 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
14668 PR rtl-optimization/78751
14669 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
14672 2017-01-14 Jeff Law <law@redhat.com>
14674 PR tree-optimization/79090
14675 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
14676 variable length stores.
14677 (compute_trims): Delete dead assignment to *trim_tail.
14678 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
14681 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
14683 PR rtl-optimization/78626
14684 PR rtl-optimization/78727
14685 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
14686 of a block, and split such blocks after everything else is finished.
14688 2017-01-14 Alan Modra <amodra@gmail.com>
14691 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
14692 target legitimate_combined_insn.
14693 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
14694 (rs6000_legitimate_combined_insn): New function.
14695 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
14697 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
14698 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
14699 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
14701 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
14703 * doc/frontends.texi (G++ and GCC): Remove references to Java.
14705 2017-01-13 Jeff Law <law@redhat.com>
14707 PR tree-optimization/33562
14708 PR tree-optimization/61912
14709 PR tree-optimization/77485
14710 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
14712 (delete_dead_assignment): Likewise.
14713 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
14714 statement to delete_dead_call and delete_dead_assignment.
14716 2017-01-13 David Malcolm <dmalcolm@redhat.com>
14719 * substring-locations.c (format_warning_va): Strengthen case 1 so
14720 that both endpoints of the substring must be within the format
14721 range for just the substring to be printed.
14723 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
14725 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
14726 * config/i386/i386.c (ix86_target_string): Add missing options
14727 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
14728 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
14729 flags_other and ix86_target_other to flags2_other. Display unknown
14731 (ix86_valid_target_attribute_inner_p): Add missing options and
14732 reorder options by implied ISAs, as in ix86_target_string.
14734 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
14736 * hash-table.h (hash_table::too_empty_p): New function.
14737 (hash_table::expand): Use it.
14738 (hash_table::traverse): Likewise.
14739 (hash_table::empty_slot): Use sizeof (value_type) instead of
14740 sizeof (PTR) to convert bytes to elements. Shrink the table
14741 if the current size is excessive for the current number of
14744 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
14746 * ira-costs.c (record_reg_classes): Break from the inner loop
14747 early once alt_fail is known to be true. Update outer loop
14748 handling accordingly.
14750 2017-01-13 Jeff Law <law@redhat.com>
14752 * tree-ssa-dse.c (decrement_count): New function.
14753 (increment_start_addr, maybe_trim_memstar_call): Likewise.
14754 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
14755 when we know the partially dead statement is a mem* function.
14757 PR tree-optimization/61912
14758 PR tree-optimization/77485
14759 * tree-ssa-dse.c: Include expr.h.
14760 (maybe_trim_constructor_store): New function.
14761 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
14763 PR tree-optimization/33562
14764 PR tree-optimization/61912
14765 PR tree-optimization/77485
14766 * doc/invoke.texi: Document new dse-max-object-size param.
14767 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
14768 * tree-ssa-dse.c: Include params.h.
14769 (dse_store_status): New enum.
14770 (initialize_ao_ref_for_dse): New, partially extracted from
14772 (valid_ao_ref_for_dse, normalize_ref): New.
14773 (setup_live_bytes_from_ref, compute_trims): Likewise.
14774 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
14775 (maybe_trim_partially_dead_store): Likewise.
14776 (maybe_trim_complex_store): Likewise.
14777 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
14778 Track what bytes live from the original store. Return tri-state
14779 for dead, partially dead or live.
14780 (dse_dom_walker): Add constructor, destructor and new private members.
14781 (delete_dead_call, delete_dead_assignment): New extracted from
14783 (dse_optimize_stmt): Make a member of dse_dom_walker.
14784 Use initialize_ao_ref_for_dse.
14786 PR tree-optimization/33562
14787 PR tree-optimization/61912
14788 PR tree-optimization/77485
14789 * sbitmap.h (bitmap_count_bits): Prototype.
14790 (bitmap_clear_range, bitmap_set_range): Likewise.
14791 * sbitmap.c (bitmap_clear_range): New function.
14792 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
14794 2017-01-13 Martin Liska <mliska@suse.cz>
14797 * function.c (set_cfun): Add new argument force.
14798 * function.h (set_cfun): Likewise.
14799 * ipa-inline-transform.c (inline_call): Use the function when
14800 strict alising from is dropped for function we inline to.
14802 2017-01-13 Richard Biener <rguenther@suse.de>
14804 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
14805 for dumping GIMPLE INTEGER_CSTs.
14807 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14809 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
14810 to 201112L since C++17.
14812 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
14815 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
14816 if -fsanitize=kernel-address is present.
14818 2017-01-13 Richard Biener <rguenther@suse.de>
14820 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
14821 as _Literal ( type ) number in case usual suffixes do not
14822 preserve all information.
14824 2017-01-13 Richard Biener <rguenther@suse.de>
14826 PR tree-optimization/77283
14827 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
14828 and ssa-iterators.h.
14829 (is_feasible_trace): Implement a cost model based on joiner
14832 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
14835 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
14836 char or short to __float128/_Float128 directly.
14838 2017-01-12 Martin Sebor <msebor@redhat.com>
14840 to -Wformat-overflow.
14841 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
14842 (min_bytes_remaining): Same.
14843 (get_string_length): Same.
14844 (format_string): Same.
14845 (format_directive): Same.
14847 (pass_sprintf_length::handle_gimple_call): Same.
14849 2017-01-12 Jakub Jelinek <jakub@redhat.com>
14851 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
14852 info.nowrite calls with no lhs that can't throw. Return bool
14853 whether gsi_remove has been called or not.
14854 (pass_sprintf_length::handle_gimple_call): Return bool whether
14855 try_substitute_return_value called gsi_remove. Formatting fix.
14856 (pass_sprintf_length::execute): Don't use gsi_remove if
14857 handle_gimple_call returned true.
14860 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
14861 be removed due to side-effects, don't remove following barrier nor
14862 turn the successor edge into fallthru edge.
14864 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14867 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
14868 element-reversing loads and stores as not swappable.
14870 2017-01-12 Nathan Sidwell <nathan@acm.org>
14871 Nicolai Stange <nicstange@gmail.com>
14873 * combine.c (try_combine): Don't ignore result of overlap checking
14874 loop. Combine overlap & asm check into single loop.
14876 2017-01-12 Richard Biener <rguenther@suse.de>
14878 * tree-pretty-print.c (dump_generic_node): Provide -gimple
14879 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
14881 2017-01-12 Richard Biener <rguenther@suse.de>
14883 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
14884 and TS_TARGET_OPTION directly derive from TS_BASE.
14885 * tree-core.h (tree_optimization_option): Derive from tree_base.
14886 (tree_target_option): Likewise.
14888 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
14890 * config/i386/i386.c (memory_address_length): Increase len
14891 only when rip_relative_addr_p returns false.
14893 2017-01-11 Julia Koval <julia.koval@intel.com>
14895 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
14896 (OPTION_MASK_ISA_SGX_SET): New.
14897 (ix86_handle_option): Handle OPT_msgx.
14898 * config.gcc: Added sgxintrin.h.
14899 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
14900 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
14901 * config/i386/i386.c (ix86_target_string): Add -msgx.
14903 (ix86_option_override_internal): Handle new options.
14904 (ix86_valid_target_attribute_inner_p): Add sgx.
14905 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
14906 * config/i386/i386.opt: Add msgx.
14907 * config/i386/sgxintrin.h: New file.
14908 * config/i386/x86intrin.h: Add sgxintrin.h.
14910 2017-01-11 Jakub Jelinek <jakub@redhat.com>
14913 * fold-const.c (maybe_nonzero_address): Return 1 for function
14915 (tree_single_nonzero_warnv_p): Don't handle function local objects
14919 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
14922 2017-01-11 David Malcolm <dmalcolm@redhat.com>
14925 * opts.c: Include "spellcheck.h"
14926 (struct string_fragment): New struct.
14927 (struct edit_distance_traits<const string_fragment &>): New
14929 (get_closest_sanitizer_option): New function.
14930 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
14932 2017-01-11 Jakub Jelinek <jakub@redhat.com>
14934 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
14936 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
14937 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
14938 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
14939 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
14940 for initial die_offset if dwarf_split_debug_info.
14941 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
14942 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
14944 (output_skeleton_debug_sections): Formatting fix. Use
14945 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
14946 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
14948 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
14950 * config/arm/cortex-a53.md: Add bypasses for
14951 cortex_a53_r2f_cvt.
14952 (cortex_a53_r2f): Only use for transfers.
14953 (cortex_a53_f2r): Likewise.
14954 (cortex_a53_r2f_cvt): Add reservation for conversions.
14955 (cortex_a53_f2r_cvt): Likewise.
14957 2017-01-11 Tamar Christina <tamar.christina@arm.com>
14959 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
14960 to all inlined functions, change static to extern.
14962 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
14965 * config/arm/arm.c (legitimize_pic_address): Handle reference to
14967 (arm_assemble_integer): Likewise.
14969 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
14971 * config.gcc: Use new awk script to check CPU, FPU and architecture
14972 parameters for --with-... options.
14973 * config/arm/parsecpu.awk: New file
14974 * config/arm/arm-cpus.in: New file.
14975 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
14977 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
14979 * config/arm/t-arm: Update dependency rules.
14980 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
14981 of processing .def files.
14982 * config/arm/genopt.sh: Deleted.
14983 * config/arm/gentune.sh: Deleted.
14984 * config/arm/arm-cores.def: Deleted.
14985 * config/arm/arm-arches.def: Deleted.
14986 * config/arm/arm-fpus.def: Deleted.
14987 * config/arm/arm-tune.md: Regenerated.
14988 * config/arm/arm-tables.opt: Regenerated.
14989 * config/arm/arm-cpu.h: New generated file.
14990 * config/arm/arm-cpu-data.h: New generated file.
14991 * config/arm/arm-cpu-cdata.h: New generated file.
14993 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
14996 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
14998 (input_varpool_node): Unpack dynamically_initialized bit.
15000 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
15002 PR rtl-optimization/79032
15003 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
15004 the alignment of the adjusted memory reference against that of MODE,
15005 instead of the alignment of the original memory reference.
15007 2017-01-11 Martin Jambor <mjambor@suse.cz>
15009 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
15011 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
15012 decorated functions.
15014 2017-01-11 Richard Biener <rguenther@suse.de>
15016 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
15017 set range/nonnull info for PHI results. Do not set it on
15018 stmts marked for removal.
15020 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
15022 * expr.c (store_field): In the bitfield case, fetch the return value
15023 from the registers before applying a single big-endian adjustment.
15024 Always do a final load for a BLKmode value not larger than a word.
15026 2017-01-10 David Malcolm <dmalcolm@redhat.com>
15029 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
15030 that we correctly handle column numbers greater than
15031 LINE_MAP_MAX_COLUMN_NUMBER.
15033 2017-01-10 Martin Sebor <msebor@redhat.com>
15035 PR middle-end/78245
15036 * gimple-ssa-sprintf.c (get_destination_size): Call
15037 {init,fini}object_sizes.
15038 * tree-object-size.c (addr_object_size): Adjust.
15039 (pass_through_call): Adjust.
15040 (pass_object_sizes::execute): Adjust.
15041 * tree-object-size.h (fini_object_sizes): Declare.
15043 2017-01-10 Martin Sebor <msebor@redhat.com>
15045 PR tree-optimization/78775
15046 * builtins.c (get_size_range): Move...
15047 * calls.c: ...to here.
15048 (alloc_max_size): Accept zero argument.
15049 (operand_signed_p): Remove.
15050 (maybe_warn_alloc_args_overflow): Call get_size_range.
15051 * calls.h (get_size_range): Declare.
15053 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
15055 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
15056 from TI's devices.csv file as of September 2016.
15057 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
15059 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
15061 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
15062 * doc/invoke.texi: Likewise.
15063 * doc/md.texi: Likewise.
15064 * doc/objc.texi: Likewise.
15066 2017-01-10 Joshua Conner <joshconner@google.com>
15068 * config/arm/fuchsia-elf.h: New file.
15069 * config/fuchsia.h: New file.
15070 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
15071 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
15073 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
15075 2016-01-10 Richard Biener <rguenther@suse.de>
15077 PR tree-optimization/79034
15078 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
15079 Propagate out degenerate PHIs in the joiner.
15081 2017-01-10 Martin Liska <mliska@suse.cz>
15083 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
15084 (sort_congruence_classes_by_decl_uid): Likewise.
15085 (sort_congruence_class_groups_by_decl_uid): Likewise.
15086 (sem_item_optimizer::merge_classes): Sort class, groups in these
15087 classes and members in the groups by DECL_UID of declarations.
15088 This would make merge operations stable.
15090 2017-01-10 Martin Liska <mliska@suse.cz>
15092 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
15093 usage of m_classes_vec.
15094 (sem_item_optimizer::~sem_item_optimizer): Likewise.
15095 (sem_item_optimizer::get_group_by_hash): Likewise.
15096 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
15097 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
15098 (sem_item_optimizer::verify_classes): Likewise.
15099 (sem_item_optimizer::process_cong_reduction): Likewise.
15100 (sem_item_optimizer::dump_cong_classes): Likewise.
15101 (sem_item_optimizer::merge_classes): Likewise.
15102 * ipa-icf.h (congruence_class_hash): Rename from
15103 congruence_class_group_hash. Remove declaration of m_classes_vec.
15105 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
15107 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
15108 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
15109 * config.gcc: Add avx512vpopcntdqintrin.h.
15110 * config/i386/avx512vpopcntdqintrin.h: New.
15111 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
15112 * config/i386/i386-builtin-types.def: Add new types.
15113 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
15114 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
15115 __builtin_ia32_vpopcountq_v8di_mask): New.
15116 * config/i386/i386-c.c (ix86_target_macros_internal): Define
15117 __AVX512VPOPCNTDQ__.
15118 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
15119 (PTA_AVX512VPOPCNTDQ): Define.
15120 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
15121 TARGET_AVX512VPOPCNTDQ_P): Define.
15122 * config/i386/i386.opt: Add mavx512vpopcntdq.
15123 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
15124 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
15126 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
15128 PR middle-end/77484
15129 * predict.def (PRED_CALL): Set to 67.
15131 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
15133 * expr.c (store_field): In the bitfield case, if the value comes from
15134 a function call and is of an aggregate type returned in registers, do
15135 not modify the field mode; extract the value in all cases if the mode
15136 is BLKmode and the size is not larger than a word.
15138 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
15141 * config/i386/cpuid.h: Fix undefined behavior.
15143 2017-01-04 Jeff Law <law@redhat.com>
15145 PR tree-optimization/79007
15146 PR tree-optimization/67955
15147 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
15148 conservative for pt.null when flag_non_call_exceptions is on.
15150 2017-01-09 Jakub Jelinek <jakub@redhat.com>
15152 PR translation/79019
15153 PR translation/79020
15154 * params.def (PARAM_INLINE_MIN_SPEEDUP,
15155 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
15156 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
15158 * config/avr/avr.opt (maccumulate-args): Likewise.
15159 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
15160 * common.opt (freport-bug): Likewise.
15161 * cif-code.def (CIF_FINAL_ERROR): Likewise.
15162 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
15163 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
15164 translatable string.
15165 * config/i386/i386.c (function_value_32): Likewise.
15166 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
15167 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
15169 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
15170 * common/config/msp430/msp430-common.c (msp430_handle_option):
15172 * symtab.c (symtab_node::verify_base): Likewise.
15173 * opts.c (set_debug_level): Likewise.
15174 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
15175 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
15176 missing whitespace to translatable strings.
15177 * config/avr/avr.md (bswapsi2): Fix typo in comment.
15178 * config/sh/superh.h: Likewise.
15179 * config/i386/xopintrin.h: Likewise.
15180 * config/i386/znver1.md: Likewise.
15181 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
15182 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
15183 * double-int.h (struct double_int): Likewise.
15184 * double-int.c (div_and_round_double): Likewise.
15185 * wide-int.cc: Likewise.
15186 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
15187 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
15188 * cfgcleanup.c (crossjumps_occured): Renamed to ...
15189 (crossjumps_occurred): ... this.
15190 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
15193 PR tree-optimization/78899
15194 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
15195 returning bool return struct loop *, NULL for failure and the new
15197 (versionable_outer_loop_p): Don't version outer loop if it has
15198 dont_vectorized bit set.
15199 (tree_if_conversion): When versioning outer loop, ensure
15200 tree_if_conversion is performed also on the inner loop of the
15201 non-vectorizable outer loop copy.
15202 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
15203 LOOP_VECTORIZED in inner loop of the scalar outer loop and
15204 prevent vectorization of it.
15205 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
15206 the outer loop vectorization of the non-scalar version is attempted
15207 before vectorization of the inner loop in scalar version. If
15208 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
15209 vectorization of its inner loop.
15210 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
15211 has 2 inner loops, rename also on edges from bb whose single pred
15212 is outer_loop->header. Fix typo in function comment.
15214 2017-01-09 Martin Sebor <msebor@redhat.com>
15217 * asan.c (asan_emit_stack_protection): Increase local buffer size
15218 to avoid snprintf truncation warning.
15220 2017-01-09 Andrew Pinski <apinski@cavium.com>
15222 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
15223 to reference thunderx2t99 for the tuning structure
15224 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
15226 (thunderx2t99_extra_costs): This.
15227 * config/aarch64/aarch64-tune.md: Regenerate.
15228 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
15229 (vulcan_addrcost_table): This.
15230 (vulcan_regmove_cost): Rename to ...
15231 (thunderx2t99_regmove_cost): This.
15232 (vulcan_vector_cost): Rename to ...
15233 (thunderx2t99_vector_cost): this.
15234 (vulcan_branch_cost): Rename to ...
15235 (thunderx2t99_branch_cost): This.
15236 (vulcan_tunings): Rename to ...
15237 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
15238 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
15240 2017-01-09 Martin Jambor <mjambor@suse.cz>
15244 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
15245 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
15246 (propagate_vr_accross_jump_function): Use the above function for all
15247 value range computations for pass-through jump functions and type
15248 converasion from explicit value range values.
15249 (ipcp_propagate_stage): Do not attempt to deduce types of formal
15250 parameters from TYPE_ARG_TYPES.
15251 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
15252 (ipa_write_node_info): Stream type of the actual argument.
15253 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
15255 2017-01-09 Martin Liska <mliska@suse.cz>
15258 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
15259 (lookup_compiler): Do not show error message with have_E.
15261 2017-01-09 Jakub Jelinek <jakub@redhat.com>
15263 PR tree-optimization/78938
15264 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
15265 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
15266 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
15267 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
15270 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15272 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
15275 2017-01-09 Richard Biener <rguenther@suse.de>
15277 PR tree-optimization/78997
15278 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
15279 name condition properly.
15281 2017-01-09 Richard Biener <rguenther@suse.de>
15284 * dwarf2out.c (is_cxx): New overload with context.
15285 (is_naming_typedef_decl): Use it.
15287 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
15289 * invoke.texi (Option Summary): Correct spacing in option lists
15290 and add line breaks to fix over-long lines.
15292 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
15294 PR middle-end/17660
15296 * extend.texi (Common Variable Attributes): Add xref to GCC
15297 Internals manual to explain mode attribute keywords.
15299 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
15302 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
15303 and Preprocessor Options.
15304 (Options for Linking): Document -pthread here....
15305 (RS/6000 and PowerPC Options): ...not here.
15306 (Solaris 2 Options): ...or here.
15307 * doc/cppopts.texi: Document -pthread.
15309 2017-01-08 Martin Sebor <msebor@redhat.com>
15311 PR middle-end/77708
15312 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
15313 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
15314 New member functions.
15315 (format_directive): Used them.
15317 (pass_sprintf_length::handle_gimple_call): Same.
15318 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
15319 to avoid truncation for any argument.
15320 (extract_affine_mul): Same.
15321 * tree.c (get_file_function_name): Same.
15323 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
15325 PR middle-end/77484
15326 * predict.def (PRED_INDIR_CALL): Set to 86.
15328 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
15330 PR preprocessor/54124
15331 * doc/cppopts.texi: Reformat -d subtable to list the full name
15332 of the options. Add cross-reference to the docs for the general
15333 compiler -d options.
15334 * doc/invoke.texi (Developer Options): Add cross-reference to the
15335 preprocessor-specific -d option documentation.
15337 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
15339 PR preprocessor/13498
15340 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
15341 redudant material, and reflect new command-line options.
15342 (System Headers): Likewise.
15344 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
15346 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
15347 -isystem, and -idirafter. Copy-edit.
15348 * doc/cppopts.texi: Copy-edit. Remove contradiction about
15349 default for -ftrack-macro-expansion. Delete obsolete and
15350 badly-formatted implementation details about -fdebug-cpp output.
15351 * doc/cppwarnopts.texi: Copy-edit.
15353 2017-01-07 David Malcolm <dmalcolm@redhat.com>
15356 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
15357 that the transition from a max line width >= 1<<10 to narrower
15358 lines works correctly.
15360 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
15362 * doc/options.texi (PerFunction): New.
15363 * opt-functions.awk (switch_flags): Map both Optimization and
15364 PerFunction to CL_OPTIMIZATION.
15365 * opth-gen.awk: Test for PerFunction flag along with
15367 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
15368 it only when the latter is present. Skip those that don't in
15369 the hash function generator.
15370 * common.opt (fvar-tracking): Mark as PerFunction instead of
15372 (fvar-tracking-assignments): Likewise.
15373 (fvar-tracking-assignments-toggle): Likewise.
15374 (fvar-tracking-uninit): Likewise.
15376 2017-01-07 Jakub Jelinek <jakub@redhat.com>
15378 PR translation/79018
15379 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
15382 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
15385 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
15386 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
15387 TARGET_LONG_JUMP_TABLE_OFFSETS.
15388 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
15389 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
15390 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
15391 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
15392 * config/m68k/m68k.md (tablejump expander): Likewise.
15393 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
15394 TARGET_LONG_JUMP_TABLE_OFFSETS.
15395 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
15396 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
15398 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
15399 David Holsgrove <david.holsgrove@xilinx.com>
15401 * common/config/microblaze/microblaze-common.c
15402 (TARGET_EXCEPT_UNWIND_INFO): Remove.
15403 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
15405 * config/microblaze/microblaze.c (microblaze_must_save_register)
15406 (microblaze_expand_epilogue, microblaze_return_addr): Handle
15408 (microblaze_eh_return): New function.
15409 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
15410 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
15411 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
15412 * config/microblaze/microblaze.md (eh_return): New pattern.
15414 2017-01-06 Jakub Jelinek <jakub@redhat.com>
15416 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
15417 GCC_DIAGNOSTIC_STRINGIFY): Define.
15419 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
15421 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
15423 * config/arm/arm.md (<mcrr>): New.
15425 * config/arm/arm.c (arm_arch5te): New.
15426 (arm_option_override): Set arm_arch5te.
15427 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
15429 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
15430 (arm_mcrr_qualifiers): ... this. New.
15431 (MRRC_QUALIFIERS): Define to...
15432 (arm_mrrc_qualifiers): ... this. New.
15433 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
15435 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
15436 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
15437 (MRRCI, mrrc, MRRC): New.
15438 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
15439 VUNSPEC_MRRC2): New.
15441 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
15443 * config/arm/arm.md (<mcr>): New.
15445 * config/arm/arm.c (arm_coproc_builtin_available): Add
15446 support for mcr, mrc, mcr2 and mrc2.
15447 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
15448 (arm_mcr_qualifiers): ... this. New.
15449 (MRC_QUALIFIERS): Define to ...
15450 (arm_mrc_qualifiers): ... this. New.
15451 (MCR_QUALIFIERS): Define to ...
15452 (arm_mcr_qualifiers): ... this. New.
15453 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
15455 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
15456 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
15457 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
15458 VUNSPEC_MRC2): New.
15460 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
15462 * config/arm/arm.md (*ldc): New.
15466 * config/arm/arm.c (arm_coproc_builtin_available): Add
15467 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
15468 (arm_coproc_ldc_stc_legitimate_address): New.
15469 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
15470 'qualifier_const_pointer'.
15471 (LDC_QUALIFIERS): Define to...
15472 (arm_ldc_qualifiers): ... this. New.
15473 (STC_QUALIFIERS): Define to...
15474 (arm_stc_qualifiers): ... this. New.
15475 * config/arm/arm-protos.h
15476 (arm_coproc_ldc_stc_legitimate_address): New.
15477 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
15478 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
15479 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
15480 stc2, stcl, stc2l): New.
15481 * config/arm/constraints.md (Uz): New.
15482 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
15483 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
15484 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
15485 VUNSPEC_STC2L): New.
15487 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
15489 * config/arm/arm.md (<cdp>): New.
15490 * config/arm/arm.c (neon_const_bounds): Rename this ...
15491 (arm_const_bounds): ... this.
15492 (arm_coproc_builtin_available): New.
15493 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
15494 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
15495 (CDP_QUALIFIERS): Define to...
15496 (arm_cdp_qualifiers): ... this. New.
15498 (arm_expand_builtin_args): Add case for 6 arguments.
15499 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
15500 (arm_const_bounds): ... this.
15501 (arm_coproc_builtin_available): New.
15502 * config/arm/arm_acle.h (__arm_cdp): New.
15504 * config/arm/arm_acle_builtins.def (cdp): New.
15506 * config/arm/iterators.md (CDPI,CDP,cdp): New.
15507 * config/arm/neon.md: Rename all 'neon_const_bounds' to
15508 'arm_const_bounds'.
15509 * config/arm/types.md (coproc): New.
15510 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
15511 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
15512 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
15513 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
15515 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
15517 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
15518 (UBINOP_QUALIFIERS): New.
15520 (acle_builtin_data): New. Change comment.
15521 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
15522 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
15523 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
15524 arm_acle_builtins.def.
15525 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
15526 (arm_init_acle_builtins): New.
15527 (CRC32_BUILTIN): Remove.
15528 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
15529 crc32cb, crc32ch and crc32cw.
15530 (arm_init_crc32_builtins): Remove.
15531 (arm_init_builtins): Use arm_init_acle_builtins rather
15532 than arm_init_crc32_builtins.
15533 (arm_expand_acle_builtin): New.
15534 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
15535 * config/arm/arm_acle_builtins.def: New.
15537 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
15539 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
15540 (arm_builtin_datum): ... this.
15541 (arm_init_neon_builtin): Rename to ...
15542 (arm_init_builtin): ... this. Add a new parameters PREFIX
15543 and USE_SIG_IN_NAME.
15544 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
15545 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
15546 'arm_builtin_datum'.
15547 (arm_init_vfp_builtins): Likewise.
15548 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
15549 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
15550 (arm_expand_neon_args): Rename to ...
15551 (arm_expand_builtin_args): ... this. Rename builtin_arg
15552 enum values and differentiate between ARG_BUILTIN_MEMORY
15553 and ARG_BUILTIN_NEON_MEMORY.
15554 (arm_expand_neon_builtin_1): Rename to ...
15555 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
15556 values, arm_expand_builtin_args and add bool parameter NEON.
15557 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
15558 (arm_expand_vfp_builtin): Likewise.
15559 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
15561 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
15563 PR middle-end/77484
15564 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
15565 * predict.c (tree_estimate_probability_bb): Reverse direction of
15566 polymorphic call predictor.
15568 2017-01-06 David Malcolm <dmalcolm@redhat.com>
15570 * passes.c (execute_one_pass): Split out pass-skipping logic into...
15571 (determine_pass_name_match): ...this new function and...
15572 (should_skip_pass_p): ...this new function.
15574 2017-01-06 Nathan Sidwell <nathan@acm.org>
15576 * ipa-visibility.c (function_and_variable_visibility): Reformat
15577 comments and long lines. Remove extrneous if.
15578 * symtab.c (symtab_node::make_decl_local): Fix code format.
15579 (symtab_node::set_section_for_node): Fix comment typo.
15581 2017-01-06 Martin Liska <mliska@suse.cz>
15584 * lra-constraints.c: Rename invariant to lra_invariant.
15585 * predict.c (set_even_probabilities): Initialize e to NULL.
15587 2017-01-05 Martin Sebor <msebor@redhat.com>
15589 PR tree-optimization/78910
15590 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
15591 (format_integer): Correct off-by-one error in the handling
15592 of precision with negative numbers in signed conversions..
15594 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
15596 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
15598 2017-01-05 Jakub Jelinek <jakub@redhat.com>
15600 PR tree-optimization/71016
15601 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
15602 factor_out_conditional_conversion. Formatting fix.
15603 (factor_out_conditional_conversion): Add cond_stmt argument.
15604 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
15605 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
15608 2017-01-05 David Malcolm <dmalcolm@redhat.com>
15610 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
15611 read-rtl-function.o, and selftest-rtl.o.
15612 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
15613 (selftest::aarch64_test_loading_full_dump): New function.
15614 (selftest::aarch64_run_selftests): New function.
15615 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
15616 selftest::aarch64_run_selftests.
15617 * config/i386/i386.c
15618 (selftest::ix86_test_loading_dump_fragment_1): New function.
15619 (selftest::ix86_test_loading_call_insn): New function.
15620 (selftest::ix86_test_loading_full_dump): New function.
15621 (selftest::ix86_test_loading_unspec): New function.
15622 (selftest::ix86_run_selftests): Call the new functions.
15623 * emit-rtl.c (maybe_set_max_label_num): New function.
15624 * emit-rtl.h (maybe_set_max_label_num): New decl.
15625 * function.c (instantiate_decls): Guard call to
15626 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
15627 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
15629 * gensupport.c (gen_reader::gen_reader): Pass "false"
15630 for new "compact" param of rtx_reader.
15631 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
15632 rather than an empty string for NULL strings.
15633 * read-md.c: Potentially include config.h rather than bconfig.h.
15634 Wrap include of errors.h with #ifdef GENERATOR_FILE.
15635 (have_error): New global, copied from errors.c.
15636 (md_reader::read_name): Rename to...
15637 (md_reader::read_name_1): ...this, adding "out_loc" param,
15638 and converting "missing name or number" to returning false, rather
15640 (md_reader::read_name): Reimplement in terms of read_name_1.
15641 (md_reader::read_name_or_nil): New function.
15642 (md_reader::read_string): Handle "(nil)" by returning NULL.
15643 (md_reader::md_reader): Add new param "compact".
15644 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
15645 (md_reader::read_file): New method.
15646 * read-md.h (md_reader::md_reader): Add new param "compact".
15647 (md_reader::read_file): New method.
15648 (md_reader::is_compact): New accessor.
15649 (md_reader::read_name): Convert return type from void to file_location.
15650 (md_reader::read_name_or_nil): New decl.
15651 (md_reader::read_name_1): New decl.
15652 (md_reader::m_compact): New field.
15653 (noop_reader::noop_reader): Pass "false" for new "compact" param
15655 (rtx_reader::rtx_reader): Add new "compact" param.
15656 (rtx_reader::read_rtx_operand): Make virtual and convert return
15657 type from void to rtx.
15658 (rtx_reader::read_until): New decl.
15659 (rtx_reader::handle_any_trailing_information): New virtual function.
15660 (rtx_reader::postprocess): New virtual function.
15661 (rtx_reader::finalize_string): New virtual function.
15662 (rtx_reader::m_in_call_function_usage): New field.
15663 (rtx_reader::m_reuse_rtx_by_id): New field.
15664 * read-rtl-function.c: New file.
15665 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
15666 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
15667 (selftest::verify_three_block_rtl_cfg): New decl.
15668 * read-rtl-function.h: New file.
15669 * read-rtl.c: Potentially include config.h rather than bconfig.h.
15670 For host, include function.h, memmodel.h, and emit-rtl.h.
15671 (one_time_initialization): New function.
15672 (struct compact_insn_name): New struct.
15673 (compact_insn_names): New array.
15674 (find_code): Handle insn codes in compact dumps.
15675 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
15676 (bind_subst_iter_and_attr): Likewise.
15677 (add_condition_to_string): Likewise.
15678 (add_condition_to_rtx): Likewise.
15679 (apply_attribute_uses): Likewise.
15680 (add_current_iterators): Likewise.
15681 (apply_iterators): Likewise.
15682 (initialize_iterators): Guard usage of apply_subst_iterator with
15683 #ifdef GENERATOR_FILE.
15684 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
15685 (md_reader::read_mapping): Likewise.
15686 (add_define_attr_for_define_subst): Likewise.
15687 (add_define_subst_attr): Likewise.
15688 (read_subst_mapping): Likewise.
15689 (check_code_iterator): Likewise.
15690 (rtx_reader::read_rtx): Likewise. Move one-time initialization
15692 (one_time_initialization): New function.
15693 (rtx_reader::read_until): New method.
15694 (read_flags): New function.
15695 (parse_reg_note_name): New function.
15696 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
15697 Handle reuse_rtx ids.
15698 Wrap iterator lookup within #ifdef GENERATOR_FILE.
15699 Add parsing support for RTL dumps, mirroring the special-cases in
15700 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
15701 values, and calling handle_any_trailing_information.
15702 (rtx_reader::read_rtx_operand): Convert return type from void
15703 to rtx, returning return_rtx. Handle case 'e'. Call
15704 finalize_string on XSTR and XTMPL fields.
15705 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
15706 "(nil)" values were omitted. Call the postprocess vfunc on the
15708 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
15709 class ctor. Initialize m_in_call_function_usage. Call
15710 one_time_initialization.
15711 * rtl-tests.c (selftest::test_uncond_jump): Call
15712 set_new_first_and_last_insn.
15713 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
15714 * selftest-rtl.c: New file.
15715 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
15716 (selftest::get_insn_by_uid): New decl.
15717 * selftest-run-tests.c (selftest::run_tests): Call
15718 read_rtl_function_c_tests.
15719 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
15720 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
15723 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
15725 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
15726 operands in a special way. Assert that pos+len <= mode precision.
15728 2017-01-05 Jakub Jelinek <jakub@redhat.com>
15730 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
15731 3 argument Alias with unlimited for the negative form.
15732 (fno-vect-cost-model): Removed.
15734 2017-01-05 Martin Liska <mliska@suse.cz>
15736 * hsa-gen.c (gen_hsa_divmod): New function.
15737 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
15739 2017-01-05 Martin Liska <mliska@suse.cz>
15742 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
15745 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15747 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
15748 small constant length operands.
15750 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15752 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
15753 between loop iterations.
15755 2017-01-05 Martin Liska <mliska@suse.cz>
15758 * gimplify.c (gimplify_decl_expr): Compare to
15759 asan_poisoned_variables instread of checking flags.
15760 (gimplify_target_expr): Likewise.
15761 (gimplify_expr): Likewise.
15762 (gimplify_function_tree): Conditionally initialize
15763 asan_poisoned_variables.
15765 2017-01-04 Jeff Law <law@redhat.com>
15767 PR tree-optimizatin/78812
15768 * rtl.h (contains_mem_rtx_p): Prototype.
15769 * ifcvt.c (containts_mem_rtx_p): Move from here to...
15770 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
15771 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
15772 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
15773 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
15775 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
15777 * input.c (assert_char_at_range): Default-initialize actual_range.
15779 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
15781 * df-scan.c (df_ref_create_structure): Make regno unsigned,
15782 to match the caller.
15784 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
15786 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
15787 insns after final jump in test to emit dummy move.
15789 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
15791 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
15792 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
15794 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
15796 * multiple_target.c (create_dispatcher_calls): Init e_next.
15797 * tree-ssa-loop-split.c (split_loop): Init border.
15798 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
15801 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
15806 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
15807 (altivec_register_operand): Do not return true if the operand
15808 contains a SUBREG mixing SImode and SFmode.
15809 (vsx_register_operand): Likewise.
15810 (vsx_reg_sfsubreg_ok): New predicate.
15811 (vfloat_operand): Do not return true if the operand contains a
15812 SUBREG mixing SImode and SFmode.
15813 (vint_operand): Likewise.
15814 (vlogical_operand): Likewise.
15815 (gpc_reg_operand): Likewise.
15816 (int_reg_operand): Likewise.
15817 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
15818 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
15819 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
15821 (rs6000_emit_move_si_sf_subreg): New helper function.
15822 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
15823 fixup SUBREGs involving SImode and SFmode.
15824 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
15825 numbers for the new peephole2 optimization.
15826 (peephole2 for SFmode unions): New peephole2 to optimize cases in
15827 the GLIBC math library that do AND/IOR/XOR operations on single
15828 precision floating point.
15829 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
15830 target macros to say whether we need to avoid SUBREGs mixing
15832 (TARGET_ALLOW_SF_SUBREG): Likewise.
15833 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
15834 (UNSPEC_SI_FROM_SF): Likewise.
15835 (iorxor): Change spacing.
15836 (and_ior_xor): New iterator for AND, IOR, and XOR.
15837 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
15838 (movdi_from_sf_zero_ext): Likewise.
15839 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
15840 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
15841 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
15842 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
15843 (fms<mode>4): Likewise.
15844 (fnma<mode>4): Likewise.
15845 (fnms<mode>4): Likewise.
15846 (nfma<mode>4): Likewise.
15847 (nfms<mode>4): Likewise.
15849 2017-01-04 Marek Polacek <polacek@redhat.com>
15852 * doc/invoke.texi: Document -Wpointer-compare.
15854 2017-01-04 Jakub Jelinek <jakub@redhat.com>
15856 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
15859 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
15860 descriptions for -gdwarf-5 and emit them as uleb128 instead of
15863 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
15866 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
15867 documentation of the powerpc_popcntb_ok attribute.
15868 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
15869 code to issue warning messages if a requested CPU configuration is
15870 not supported by the binary (assembler and loader) toolchain.
15871 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
15872 made to define a built-in function that has been disabled.
15873 (paired_init_builtins): Add assertion to prevent ICE if attempt is
15874 made to define a built-in function that has been disabled.
15875 (altivec_init_builtins): Add comment explaining why definition
15876 of the DST built-in functions is not preceded by an assertion
15877 check. Add assertions to prevent ICE if attempts are made to
15878 define an altivec predicate or an abs* built-in function that has
15880 (htm_init_builtins): Add comment explaining why definition of the
15881 htm built-in functions is not preceded by an assertion check.
15883 2017-01-04 Jeff Law <law@redhat.com>
15885 PR tree-optimizatin/67955
15886 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
15887 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
15888 the points-to solution does not include pt_null. Use DECL_PT_UID
15891 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
15893 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
15894 Use gen_int_mode instead of gen_lopwart for const_int operands.
15896 2017-01-04 Jakub Jelinek <jakub@redhat.com>
15898 PR tree-optimization/71563
15899 * match.pd: Simplify X << Y into X if Y is known to be 0 or
15900 out of range value - has low bits known to be zero.
15902 2017-01-04 Alan Modra <amodra@gmail.com>
15904 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
15905 * configure: Regenerate.
15906 * config.in: Regenerate.
15908 2017-01-04 Jakub Jelinek <jakub@redhat.com>
15911 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
15912 a substring of the message, but strcmp with the whole message. Ifdef
15913 ENABLE_NLS, translate the message first using dgettext.
15915 2017-01-03 Jeff Law <law@redhat.com>
15917 PR tree-optimizatin/78856
15918 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
15919 (mark_threaded_blocks): Remove code to truncate thread paths that
15920 cross multiple loop headers. Instead invalidate the cached loop
15921 iteration information and handle case of a thread path walking
15922 into an irreducible region.
15924 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
15927 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
15928 assertions. Add support for doing the signbit if the IEEE 128-bit
15929 floating point value is in a GPR.
15930 * config/rs6000/rs6000.md (Fsignbit): Delete.
15931 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
15932 Update the length attribute if the value is in a GPR.
15933 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
15934 the sign or zero extension instruction, since the value is always 0/1.
15935 (signbit<mode>2_dm2): Delete using <Fsignbit>.
15938 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
15939 extracting SImode to a GPR register so that we can generate a
15940 store, limit the vector to be in a traditional Altivec register
15941 for the vextuwrx instruction.
15943 2017-01-03 Ian Lance Taylor <iant@google.com>
15945 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
15947 2017-01-03 Martin Sebor <msebor@redhat.com>
15949 PR tree-optimization/78696
15950 * gimple-ssa-sprintf.c (format_floating): Correct handling of
15951 precision. Use MPFR for %f for greater fidelity. Correct handling
15953 (pass_sprintf_length::compute_format_length): Set width and precision
15954 specified by asrerisk to void_node for vararg functions.
15955 (try_substitute_return_value): Adjust dump output.
15957 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
15959 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
15961 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
15963 * doc/invoke.texi (SPARC options): Document -mlra as the default.
15964 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
15965 -mlra/-mno-lra was passed to the compiler.
15967 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
15969 PR rtl-optimization/65618
15970 * emit-rtl.c (try_split): Move initialization of "before" and
15971 "after" to just before the call to emit_insn_after_setloc.
15973 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
15975 * doc/md.texi (Standard Names): Remove reference to Java frontend.
15977 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
15979 * dwarf2out.c (gen_enumeration_type_die): When
15980 -gno-strict-dwarf, add a DW_AT_encoding attribute.
15982 2017-01-03 Jakub Jelinek <jakub@redhat.com>
15984 PR tree-optimization/78965
15985 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
15986 Change first argument from const call_info & to call_info &. For %n
15987 set info.nowrite to false.
15989 PR middle-end/78901
15990 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
15991 possibly throwing calls.
15993 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
15994 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
15995 and fns handling, rather than in a separate case SSA_NAME.
15997 2017-01-02 Jeff Law <law@redhat.com>
15999 * config/darwin-driver.c (darwin_driver_init): Const-correctness
16000 fixes for first_period and second_period variables.
16002 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
16005 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
16006 (*insvqi_1): New insn pattern.
16007 (*insvqi_1_mem_rex64): Ditto.
16008 (*insvqi_2): Ditto.
16009 (*insvqi_3): Rename from *insvqi.
16011 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
16013 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
16015 * doc/cfg.texi (Edges): Remove reference to Java.
16016 (Maintaining the CFG): Ditto.
16018 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
16020 PR middle-end/77674
16021 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
16022 transparent aliases.
16024 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
16026 PR middle-end/77484
16027 * predict.def (PRED_CALL): Update hitrate.
16028 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
16029 * predict.c (tree_estimate_probability_bb): Split CALL predictor
16030 into direct/indirect/polymorphic variants.
16032 2017-01-01 Jakub Jelinek <jakub@redhat.com>
16034 Update copyright years.
16036 * gcc.c (process_command): Update copyright notice dates.
16037 * gcov-dump.c (print_version): Ditto.
16038 * gcov.c (print_version): Ditto.
16039 * gcov-tool.c (print_version): Ditto.
16040 * gengtype.c (create_file): Ditto.
16041 * doc/cpp.texi: Bump @copying's copyright year.
16042 * doc/cppinternals.texi: Ditto.
16043 * doc/gcc.texi: Ditto.
16044 * doc/gccint.texi: Ditto.
16045 * doc/gcov.texi: Ditto.
16046 * doc/install.texi: Ditto.
16047 * doc/invoke.texi: Ditto.
16049 Copyright (C) 2017 Free Software Foundation, Inc.
16051 Copying and distribution of this file, with or without modification,
16052 are permitted in any medium without royalty provided the copyright
16053 notice and this notice are preserved.