1 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
4 padding_bits_to_clear_ptr.
5 (cmse_nonsecure_entry_clear_before_return): Likewise.
7 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
9 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
10 auto_sbitap instead of integer bitfield to control register needing
13 2017-11-22 Jakub Jelinek <jakub@redhat.com>
15 PR tree-optimization/83044
16 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
17 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
18 that need the upper bound. Subtract offset from
19 get_addr_base_and_unit_offset only if positive and subtract it
20 before division by eltsize rather than after it.
23 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
24 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
25 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
29 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
30 expand_binop with *mul_widen_optab, make sure at least one of the
31 operands doesn't have VOIDmode.
34 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
36 PR rtl-optimization/82044
37 PR tree-optimization/82042
38 * dse.c (record_store): Check for overflow.
39 (check_mem_read_rtx): Properly check for overflow if width == -1, call
40 add_wild_read instead of clear_rhs_from_active_local_stores on
41 overflow and log it into dump_file.
43 2017-11-22 Richard Biener <rguenther@suse.de>
45 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
46 fake edges to exit when looking for a place to insert.
47 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
49 (insert_into_preds_of_block): Commit edge insertion immediately,
50 assert that doesn't require new BBs.
51 (fini_pre): Release expressions.
52 (pass_pre::execute): Shuffle things around a bit, if the fn
53 is too large do not compute AVAIL either as this is really the
56 2017-11-22 Richard Biener <rguenther@suse.de>
58 PR tree-optimization/83089
59 * tree-if-conv.c (pass_if_conversion::execute): If anything
60 changed reset SCEV and free the number of iteration estimates.
62 2017-11-21 Martin Sebor <msebor@redhat.com>
64 PR tree-optimization/82945
65 * calls.h (warn_nonstring_bound): Remove unused function.
67 2017-11-21 Martin Sebor <msebor@redhat.com>
69 PR tree-optimization/82945
70 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
71 * calls.h (maybe_warn_nonstring_arg): Declare new function.
72 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
74 (initialize_argument_information): Call maybe_warn_nonstring_arg.
75 * calls.h (get_attr_nonstring_decl): Declare new function.
76 * doc/extend.texi (attribute nonstring): Update.
77 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
78 get_attr_nonstring_decl and handle it.
79 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
80 detection of nul-termination.
81 (strlen_to_stridx): Change to a pointer.
82 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
83 (pass_strlen::execute): Same.
85 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
87 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
88 for the new option -mprefer-vector-width=[none|128|256|512].
89 * config/i386/i386.c (ix86_target_string): remove old style options
90 -mprefer-avx256 and make -mprefer-avx128 as alias.
91 (ix86_option_override_internal): Apply defaults for the
92 -mprefer-vector-width=[128|256] option.
93 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
94 Implement macros to work with -mprefer-vector-width=.
95 * config/i386/i386.opt: Implemented option
96 -mprefer-vector-width=[none|128|256|512].
97 * doc/invoke.texi: Documentation for
98 -mprefer-vector-width=[none|128|256|512].
100 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
102 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
104 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
106 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
108 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
110 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
112 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
113 * expr.c (expand_expr_real_2): Likewise.
114 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
115 fold_binary_loc): Likewise.
116 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
117 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
118 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
119 MINUS_EXPR transformations.
120 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
121 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
122 * tree-inline.c (estimate_operator_cost): Likewise.
123 * tree-pretty-print.c (dump_generic_node, op_code_prio,
124 op_symbol_code): Likewise.
125 * tree-vect-stmts.c (vectorizable_operation): Likewise.
126 * vr-values.c (extract_range_from_binary_expr): Likewise.
127 * varasm.c (initializer_constant_valid_p_1): Likewise.
128 * tree.def: New tree code POINTER_DIFF_EXPR.
130 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
132 * config/i386/i386.md (*bswap<mode>2_movbe): Add
133 integer suffix to movbe mnemonic.
134 (*bswaphi2_movbe): Ditto.
135 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
137 2017-11-21 Jakub Jelinek <jakub@redhat.com>
140 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
141 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
142 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
144 2017-11-21 Martin Liska <mliska@suse.cz>
146 * tree-inline.c (expand_call_inline): Remove not needed
149 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
150 Jakub Jelinek <jakub@redhat.com>
153 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
154 Remove static keyword from f variable.
156 2017-11-21 Jakub Jelinek <jakub@redhat.com>
158 PR tree-optimization/83086
159 * gimple-ssa-store-merging.c
160 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
161 rather than n.base_addr.
163 2017-11-21 Martin Liska <mliska@suse.cz>
165 PR rtl-optimization/82044
166 PR tree-optimization/82042
167 * dse.c (check_mem_read_rtx): Check for overflow.
169 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
171 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
174 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
176 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
177 and its corresponding call arg location note.
179 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
180 Andrew Burgess <andrew.burgess@embecosm.com>
182 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
184 (arc_return_slot_offset): Likewise.
185 (arc_eh_return_address_location): New declaration.
186 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
187 (MUST_SAVE_REGISTER): Add exception handler case.
188 (MUST_SAVE_RETURN_ADDR): Likewise.
189 (arc_frame_pointer_required): Likewise.
190 (arc_frame_pointer_needed): New function.
191 (arc_compute_frame_size): Changed.
192 (arc_expand_prologue): Likewise.
193 (arc_expand_epilogue): Likewise.
194 (arc_initial_elimination_offset): Likewise.
195 (arc_return_slot_offset): Delete.
196 (arc_eh_return_address_location): New function.
197 (arc_builtin_setjmp_frame_value): Likewise.
198 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
199 (EH_RETURN_STACKADJ_RTX): Define.
200 (EH_RETURN_HANDLER_RTX): Likewise.
201 * config/arc/arc.md (eh_return): Delete.
203 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
205 * print-tree.h (debug_vec_tree): Remove prototype.
206 * gdbinit.in (pvt): Remove macro.
208 2017-11-21 Jakub Jelinek <jakub@redhat.com>
210 PR tree-optimization/83047
211 * gimple-ssa-store-merging.c
212 (imm_store_chain_info::output_merged_store): If the loads with the
213 same vuse are in different basic blocks, for load_gsi pick a load
214 location that is dominated by the other loads.
217 * config/i386/i386.c (ix86_memmodel_check): Start
218 -Winvalid-memory-model diagnostics with lowercase letter.
221 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
222 set in_cold_section_p to the partition containing loc_list->first.
223 When seeing loc_list->last_before_switch node, update secname and
224 perform range_across_switch second partition handling only after that.
227 * run-rtl-passes.c: Include debug.h.
228 (run_rtl_passes): Call debug_hooks->assembly_start.
229 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
233 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
234 OPTAB_DIRECT in calls to expand_simple_binop.
236 2017-11-20 David Malcolm <dmalcolm@redhat.com>
239 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
241 2017-11-20 David Malcolm <dmalcolm@redhat.com>
244 * spellcheck.h (best_match::blithely_get_best_candidate): New
247 2017-11-20 Jakub Jelinek <jakub@redhat.com>
249 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
250 with lower case letter.
252 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
254 * config/i386/i386.md (bswaphi2): New expander.
255 (*bswaphi2_movbe): New insn pattern.
256 (bswaphi -> rorhi pepehole2): New peephole pattern.
258 2017-11-20 Jeff Law <law@redhat.com>
260 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
261 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
262 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
263 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
264 gimple-ssa-evrp-analyze.h.
266 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
268 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
270 2017-11-20 Bin Cheng <bin.cheng@arm.com>
272 * tree-predcom.c: Add general comment on Store-Store chains.
273 (split_data_refs_to_components): Postpone clearing eliminate_store_p
275 (get_chain_last_ref_at): Rename into...
276 (get_chain_last_write_at): ...this.
277 (get_chain_last_write_before_load): New function.
278 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
279 CT_STORE_STORE when write reference is added.
280 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
281 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
282 (initialize_root_vars_store_elim_1): Ditto.
283 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
284 definition is created.
285 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
286 chain by replacing it with dominant stored value.
288 2017-11-20 Bin Cheng <bin.cheng@arm.com>
290 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
292 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
294 * vr-values.c (extract_range_from_binary_expr): Use a full range
297 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
299 * config/arm/arm.md (R4_REGNUM): Define constant.
300 (nonsecure_call_internal): Remove r4 clobber.
301 (nonsecure_call_value_internal): Likewise.
302 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
303 clobber and resequence match_operands.
304 (nonsecure_call_value_reg_thumb1_v5): Likewise.
305 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
306 (nonsecure_call_value_reg_thumb2): Likewise.
308 2017-11-20 Jakub Jelinek <jakub@redhat.com>
310 PR tree-optimization/78821
311 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
312 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
313 to the address of the base rather than the base itself.
314 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
315 (find_bswap_or_nop_finalize): New function.
316 (find_bswap_or_nop): Use it.
317 (bswap_replace): Return a tree rather than bool, change first
318 argument from gimple * to gimple_stmt_iterator, allow inserting
319 into an empty sequence, allow ins_stmt to be NULL - then emit
320 all stmts into gsi. Fix up MEM_REF address gimplification.
321 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
322 (struct store_immediate_info): Add N and INS_STMT non-static
324 (store_immediate_info::store_immediate_info): Initialize them
325 from newly added ctor args.
326 (merged_store_group::apply_stores): Formatting fixes. Sort by
328 (stmts_may_clobber_ref_p): For stores call also
329 refs_anti_dependent_p.
330 (gather_bswap_load_refs): New function.
331 (imm_store_chain_info::try_coalesce_bswap): New method.
332 (imm_store_chain_info::coalesce_immediate_stores): Use it.
333 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
334 (imm_store_chain_info::output_merged_store): Fail if number of
335 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
336 and NOP_EXPR rhs_code.
337 (pass_store_merging::process_store): Compute n and ins_stmt, if
338 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
339 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
341 (pass_store_merging::execute): Calculate dominators.
343 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
344 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
345 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
346 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
347 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
348 class pass_optimize_bswap, bswap_replace,
349 pass_optimize_bswap::execute): Moved to ...
350 * gimple-ssa-store-merging.c: ... this file.
351 Include optabs-tree.h.
352 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
353 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
354 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
355 anonymous namespace, remove static keywords.
356 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
357 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
359 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
362 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
364 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
366 * vec.h (debug_helper): New function.
367 (DEFINE_DEBUG_VEC): New macro.
368 * hash-set.h (debug_helper): New function.
369 (DEFINE_DEBUG_HASH_SET): New macro.
370 * cfg.c (debug_slim (edge)): New function.
371 Call DEFINE_DEBUG_VEC for edges.
372 Call DEFINE_DEBUG_HASH_SET for edges.
373 * cfghooks.c (debug_slim (basic_block)): New function.
374 Call DEFINE_DEBUG_VEC for basic blocks.
375 Call DEFINE_DEBUG_HASH_SET for basic blocks.
376 * print-tree.c (debug_slim): New function to handle trees.
377 Call DEFINE_DEBUG_VEC for trees.
378 Call DEFINE_DEBUG_HASH_SET for trees.
379 (debug (vec<tree, va_gc>) &): Remove.
380 (debug (<vec<tree, va_gc>) *): Remove.
381 * print-rtl.c (debug_slim): New function to handle const_rtx.
382 Call DEFINE_DEBUG_VEC for rtx_def.
383 Call DEFINE_DEBUG_VEC for rtx_insn.
384 Call DEFINE_DEBUG_HASH_SET for rtx_def.
385 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
386 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
387 (debug (vec<rtx_insn *> *ptr): Remove.
388 (debug_insn_vector): Remove.
389 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
391 2017-11-20 Tom de Vries <tom@codesourcery.com>
393 PR rtl-optimization/82020
394 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
395 IF_THEN_ELSE condition.
397 2017-11-19 Jeff Law <law@redhat.com>
399 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
400 of degenerates resulting from ignoring an edge.
402 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
405 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
407 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
410 * profile-count.c (profile_count::to_sreal_scale): Fix return value
411 for uninitialied counts.
413 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
416 * tree-inline.c (estimate_num_insns): Set to 1 at least.
418 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
421 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
423 2017-11-19 Tom de Vries <tom@codesourcery.com>
425 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
428 2017-11-19 Tom de Vries <tom@codesourcery.com>
430 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
433 2017-11-19 Tom de Vries <tom@codesourcery.com>
435 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
438 2017-11-19 Tom de Vries <tom@codesourcery.com>
440 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
442 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
444 2017-11-19 Tom de Vries <tom@codesourcery.com>
446 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
448 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
449 semicolon after MCORE_EXPORT_NAME call.
451 2017-11-19 Tom de Vries <tom@codesourcery.com>
454 * vmsdbgout.c (vmsdbgout_early_finish): New function.
455 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
457 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
459 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
460 (cgraph_edge::clone): Cleanup updating of profile.
461 * ipa-cp.c (update_profiling_info): Likewise.
462 * ipa-inline-transform.c (inline_transform): Likewise.
463 * ipa-inline.c (inline_small_functions): Add missing space to dump.
464 * ipa-split.c (execute_split_functions): Do not split when function
466 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
467 * profile-count.c (profile_count::dump): Add global0.
468 (profile_count::to_cgraph_frequency): Do not ICE when entry is
470 (profile_count::to_sreal_scale): Likewise.
471 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
472 (profile_count::combine_with_ipa_count): New function.
473 * profile-count.h (profile_guessed_global0adjusted): New.
474 (profile_count::adjusted_zero): New.
475 (profile_count::global0adjusted): New.
476 (profile_count::combine_with_ipa_count): New.
477 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
478 correct profile of return block of split functions.
479 (copy_cfg_body): Remove unused profile_count.
480 (copy_body): Likewise.
481 (expand_call_inline): Update.
482 (tree_function_versioning): Update.
484 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
486 * hash-set.h (hash_set::empty): New.
487 * tree-ssa-threadbackward.h: Delete.
488 * tree-ssa-threadbackward.c (class thread_jumps): New.
489 Move max_threaded_paths into class.
490 (fsm_find_thread_path): Remove arguments that are now in class.
491 (profitable_jump_thread_path): Rename to...
492 (thread_jumps::profitable_jump_thread_path): ...this.
493 (convert_and_register_jump_thread_path): Rename to...
494 (thread_jumps::convert_and_register_current_path): ...this.
495 (check_subpath_and_update_thread_path): Rename to...
496 (thread_jumps::check_subpath_and_update_thread_path): ...this.
497 (register_jump_thread_path_if_profitable): Rename to...
498 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
499 (handle_phi): Rename to...
500 (thread_jumps::handle_phi): ...this.
501 (handle_assignment): Rename to...
502 (thread_jumps::handle_assignment): ...this.
503 (fsm_find_control_statement_thread_paths): Rename to...
504 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
505 (find_jump_threads_backwards): Rename to...
506 (thread_jumps::find_jump_threads_backwards): ...this.
507 Initialize path local data.
508 (pass_thread_jumps::execute): Call find_jump_threads_backwards
509 from within thread_jumps class.
510 (pass_early_thread_jumps::execute): Same.
512 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
514 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
516 2017-11-17 Steve Ellcey <sellcey@cavium.com>
518 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
519 to canonical location.
521 2017-11-17 Steve Ellcey <sellcey@cavium.com>
524 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
526 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
528 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
530 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
531 rcrt1.o%s/grcrt1.o%s for -static-pie.
533 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
535 * i386.c (ix86_multiplication_cost, ix86_division_cost,
536 ix86_shift_rotate_cost): Break out from ...
537 (ix86_rtx_costs): ... here.
538 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
541 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
543 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
544 when entry block was promoted unlikely.
545 (estimate_bb_frequencies): Increase frequency scale.
546 * profile-count.h (profile_count): Export precision info.
548 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
550 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
551 disturbin profile of entry block.
553 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
555 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
558 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
560 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
561 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
563 2017-11-17 Jeff Law <law@redhat.com>
565 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
566 from evrp_dom_walker class. Various methods moved into new class.
567 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
568 (evrp_range_analyzer::enter): New method.
569 (evrp_range_analyzer::leave): New method.
570 (evrp_dom_walker): Remove delegators no longer needed by this class.
571 Replace vr_values data member with evrp_range_analyzer
573 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
574 method extracted from evrp_dom_walker::before_dom_children.
575 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
576 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
578 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
579 Add private copy constructor and move assignment operators.
580 Privatize methods and class data where trivially possible.
581 (evrp_dom_walker::cleanup): New function, extracted from
582 execute_early_vrp. Simplify access to class data.
584 * vr-values.h (get_output_for_vrp): Prototype.
585 * vr-values.c (get_output_for_vrp): New function extracted from
586 vrp_visit_assignment_or_call and extract_range_from_stmt.
587 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
589 2017-11-17 Luis Machado <luis.machado@linaro.org>
591 * config/aarch64/aarch64.c
592 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
593 (qdf24xx_tunings) <autoprefetcher_model>: Set to
594 tune_params::AUTOPREFETCHER_WEAK.
596 2017-11-17 Tamar Christina <tamar.christina@arm.com>
599 * config/arm/arm.c (arm_valid_target_attribute_rec):
600 Parse "arch=" and "+<ext>".
601 (arm_valid_target_attribute_tree): Re-init global options.
602 (arm_option_override): Make non-static.
603 (arm_options_perform_arch_sanity_checks): Make errors fatal.
604 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
605 (__ARM_FEATURE_CRC32): Support undef.
606 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
607 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
609 2017-11-17 David Malcolm <dmalcolm@redhat.com>
611 * gdbinit.in (break-on-diagnostic): New command.
613 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
615 * config/i386/i386.c (ix86_expand_epilogue): Change simple
616 return to indirect jump for EH return if control-flow
617 protection is enabled. Change explicit 'false' argument in
618 pro_epilogue_adjust_stack with a value of
620 * config/i386/i386.md (simple_return_indirect_internal):
621 Remove SImode restriction to support 64-bit.
623 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
625 * combine.c (added_notes_insn): New.
626 (try_combine): Handle added_notes_insn like added_links_insn.
627 Rewrite return value code.
628 (distribute_notes): Set added_notes_insn to the earliest insn we added
631 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
633 PR rtl-optimization/82621
634 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
635 dest of one of those SETs is unused.
637 2017-11-17 Richard Biener <rguenther@suse.de>
640 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
641 * tree-pretty-print.c (dump_generic_node): Handle
642 annot_expr_parallel_kind.
643 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
644 * gimplify.c (gimple_boolify): Likewise.
646 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
648 * config.gcc (extra_headers): Add cet.h for x86 targets.
649 * config/i386/cet.h: New file.
650 * doc/install.texi: Add --enable-cet/--disable-cet.
652 2017-11-17 Richard Biener <rguenther@suse.de>
654 PR tree-optimization/83017
655 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
656 (gen_parallel_loop): Properly count iterations.
657 (parallelize_loops): Handle loop->can_be_parallel independent
658 of flag_loop_parallelize_all. Make static profitability test match
660 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
661 * invoke.texi (parloops-min-per-thread): Document.
663 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
665 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
666 upstreaming review comments.
668 2017-11-17 Tamar Christina <tamar.christina@arm.com>
670 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
671 with fast unaligned access.
672 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
674 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
676 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
677 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
679 2017-11-17 Richard Biener <rguenther@suse.de>
681 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
682 folding of references.
684 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
687 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
688 of replacing call to strncmp with corresponding call to strcmp when
691 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
693 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
694 option prefer-avx256 for skylake-avx512 configuration.
695 * config/i386/i386.c (ix86_option_override_internal): Ditto.
696 (get_builtin_code_for_version): Ditto.
698 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
699 Monk Chiang <sh.chiang04@gmail.com>
701 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
702 (FIXED_REGISTERS): Reserve more register numbers.
703 (CALL_USED_REGISTERS): Likewise.
704 (REG_ALLOC_ORDER): Likewise.
705 (REG_CLASS_CONTENTS): Likewise.
706 (REGISTER_NAMES): Likewise.
708 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
709 Kito Cheng <kito.cheng@gmail.com>
711 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
713 * config/nds32/iterators.md: Add vector mode iterators and attributes.
715 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
717 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
718 parameter list for vec_splats.
720 2017-11-16 Joseph Myers <joseph@codesourcery.com>
722 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
724 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
726 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
729 * doc/install.texi: Document incompatibility of Perl >=5.6.26
730 with the required version of automake 1.11.6.
732 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
734 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
736 (DU_C2_3_power9): Correct reservation combinations.
737 (FP_DIV_power9, VEC_DIV_power9): New.
738 (power9-alu): Split out rotate/shift...
739 (power9-rot): ...to here, correct dispatch resource.
740 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
742 (power9-fp): Correct latency.
743 (power9-sdiv): Add div/sqrt resource.
744 (power9-ddiv): Correct latency, add div/sqrt resource.
745 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
746 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
748 (power9-qpdiv, power9-qpmul): Adjust resource usage.
750 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
752 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
753 switch statement mapping KF built-ins to TF built-ins if we don't
754 have the proper ISA 3.0 assembler support.
756 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
758 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
759 (lower_emutls_function_body): Do not compute it.
761 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
763 * ipa-split.c (split_bb_info): Turn time to sreal.
764 (split_point): Likewise.
765 (dump_split_point): Likewise.
766 (fine_split_points): Likewise.
767 (execute_split_functions): Only zero split_bbs; turn time to sreals.
769 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
771 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
774 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
776 * predict.c (combine_predictions_for_bb): Preserve zero predicted
778 (expensive_function_p): Remove useless assert.
779 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
781 2017-11-16 Martin Sebor <msebor@redhat.com>
783 PR tree-optimization/82588
784 PR tree-optimization/82583
785 * tree-vrp.c (check_array_ref): Handle flexible array members,
786 string literals, and inner indices.
787 (search_for_addr_array): Add detail to diagnostics.
789 2017-11-16 Nathan Sidwell <nathan@acm.org>
793 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
794 (SET_DECL_ASSEMBLER_NAME): Forward to
795 overwrite_decl_assembler_name.
796 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
797 (overwrite_decl_assembler_name): Declare.
798 * tree.c (overwrite_decl_assembler_name): New.
799 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
800 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
801 (LANG_HOOKS_INITIALIZER): Add it.
802 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
803 * langhooks.c (lhd_set_decl_assembler_name): Use
804 SET_DECL_ASSEMBLER_NAME.
805 (lhd_overwrite_decl_assembler_name): Default implementation.
807 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
808 Jackson Woodruff <jackson.woodruff@arm.com>
810 PR tree-optimization/71026
811 * match.pd: Canonicalize constant multiplies in division.
813 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
815 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
816 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
817 Remove OPT_fomit_frame_pointer entry.
818 * common/config/arc/arc-common.c: Likewise.
819 * common/config/arm/arm-common.c: Likewise.
820 * common/config/avr/avr-common.c: Likewise.
821 * common/config/c6x/c6x-common.c: Likewise.
822 * common/config/cr16/cr16-common.c: Likewise.
823 * common/config/cris/cris-common.c: Likewise.
824 * common/config/epiphany/epiphany-common.c: Likewise.
825 * common/config/fr30/fr30-common.c: Likewise.
826 * common/config/frv/frv-common.c: Likewise.
827 * common/config/ia64/ia64-common.c: Likewise.
828 * common/config/iq2000/iq2000-common.c: Likewise.
829 * common/config/lm32/lm32-common.c: Likewise.
830 * common/config/m32r/m32r-common.c: Likewise.
831 * common/config/mcore/mcore-common.c: Likewise.
832 * common/config/microblaze/microblaze-common.c: Likewise.
833 * common/config/mips/mips-common.c: Likewise.
834 * common/config/mmix/mmix-common.c: Likewise.
835 * common/config/mn10300/mn10300-common.c: Likewise.
836 * common/config/nios2/nios2-common.c: Likewise.
837 * common/config/pa/pa-common.c: Likewise.
838 * common/config/pdp11/pdp11-common.c: Likewise.
839 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
840 * common/config/riscv/riscv-common.c: Likewise.
841 * common/config/rs6000/rs6000-common.c: Likewise.
842 * common/config/rx/rx-common.c: Likewise.
843 * common/config/s390/s390-common.c: Likewise.
844 * common/config/sh/sh-common.c: Likewise.
845 * common/config/sparc/sparc-common.c: Likewise.
846 * common/config/tilegx/tilegx-common.c: Likewise.
847 * common/config/tilepro/tilepro-common.c: Likewise.
848 * common/config/v850/v850-common.c: Likewise.
849 * common/config/visium/visium-common.c: Likewise.
850 * common/config/xstormy16/xstormy16-common.c: Likewise.
851 * common/config/xtensa/xtensa-common.c: Likewise.
852 * invoke.texi (-fomit-frame-pointer): Update documentation.
854 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
856 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
858 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
860 * cfg.c (scale_bbs_frequencies_int,
861 cale_bbs_frequencies_gcov_type): Remove.
862 * cfg.h (scale_bbs_frequencies_int,
863 cale_bbs_frequencies_gcov_type): Remove.
865 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
867 * tree-ssa-loop-manip.c
868 (scale_dominated_blocks_in_loop): Update to profile counts.
869 (tree_transform_and_unroll_loop): Likewise.
871 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
873 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
874 scale_bbs_frequencies_int.
876 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
878 * final.c (compute_alignments): Use counts rather than frequencies.
880 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
882 * cfgloopanal.c: Include sreal.h
883 (average_num_loop_insns): Use counts and sreal for accounting.
885 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
887 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
890 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
891 Kito Cheng <kito.cheng@gmail.com>
893 * config/nds32/constraints.md: Provide more constraints.
894 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
895 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
896 support constraints usage.
898 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
900 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
902 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
903 Kito Cheng <kito.cheng@gmail.com>
905 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
906 * config/nds32/nds32.opt: Refine the layout.
907 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
908 TARGET_EXT_STRING): Support new options.
909 * config/nds32/nds32.h: Likewise.
910 * config/nds32/nds32.md: Likewise.
911 * config/nds32/nds32-predicates.c: Likewise.
912 * config/nds32/constraints.md: Likewise.
913 * common/config/nds32/nds32-common.c: Likewise.
915 2017-11-16 Julia Koval <julia.koval@intel.com>
918 * config/i386/gfniintrin.h: Add sse check.
919 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
921 2017-11-16 Julia Koval <julia.koval@intel.com>
923 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
924 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
925 (ix86_handle_option): Handle -mavx512vbmi2.
926 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
927 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
928 * config/i386/i386-c.c (__AVX512VBMI2__): New.
929 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
930 (ix86_valid_target_attribute_inner_p): Ditto.
931 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
932 * config/i386/i386.opt (mavx512vbmi2): New option.
933 * doc/invoke.texi: Add new option.
935 [2017-11-16 Julia Koval <julia.koval@intel.com>
937 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
938 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
939 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
940 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
941 _mm512_gf2p8mul_epi8): New intrinsics.
942 * config/i386/i386-builtin-types.def
943 (V64QI_FTYPE_V64QI_V64QI): New type.
944 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
945 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
946 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
947 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
948 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
949 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
951 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
953 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
954 explicitly as a stream of bytes.
956 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
958 * config/rs6000/altivec.h (vec_xst_be): New #define.
959 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
960 and externalize from *altivec_vperm_<mode>_internal.
961 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
963 (XL_BE_V8HI): Likewise.
964 (XL_BE_V4SI): Likewise.
965 (XL_BE_V4SI): Likewise.
966 (XL_BE_V2DI): Likewise.
967 (XL_BE_V4SF): Likewise.
968 (XL_BE_V2DF): Likewise.
969 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
970 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
971 all array entries with these keys: VSX_BUILTIN_VEC_XL,
972 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
973 VSX_BUILTIN_VEC_XST_BE.
974 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
975 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
977 (altivec_init_builtins): Replace conditional calls to def_builtin
978 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
979 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
980 with unconditional calls. Remove calls to def_builtin for
981 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
982 __builtin_vec_xst_be.
983 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
984 to define_expand, and add alternate RTL generation for P8.
985 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
987 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
988 add alternate RTL generation for P8.
989 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
990 vsx_ld_elemrev_v16qi.
991 (vsx_st_elemrev_v8hi): Convert define_insn
992 to define_expand, and add alternate RTL generation for P8.
993 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
995 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
996 add alternate RTL generation for P8.
997 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
998 vsx_st_elemrev_v16qi.
1000 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
1003 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
1004 TARGET_AVX512ER check.
1005 (ix86_option_override_internal): Set MASK_VZEROUPPER if
1006 neither -mzeroupper nor -mno-zeroupper is used and
1007 TARGET_EMIT_VZEROUPPER is set.
1008 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
1009 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
1011 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
1013 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1014 folding of vector compares.
1015 (fold_build_vec_cmp): New helper function.
1016 (fold_compare_helper): New helper function.
1017 (builtin_function_type): Add compare builtins to the list of functions
1018 having unsigned arguments. Cosmetic updates to comment indentation.
1019 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
1020 the not+eq combination.
1022 2017-11-15 Bin Cheng <bin.cheng@arm.com>
1024 PR tree-optimization/82726
1025 PR tree-optimization/70754
1026 * tree-predcom.c (order_drefs_by_pos): New function.
1027 (combine_chains): Move code setting has_max_use_after to...
1028 (try_combine_chains): ...here. New parameter. Sort combined chains
1029 according to position information.
1030 (tree_predictive_commoning_loop): Update call to above function.
1031 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
1033 2017-11-15 Bin Cheng <bin.cheng@arm.com>
1035 PR tree-optimization/82726
1037 2017-01-23 Bin Cheng <bin.cheng@arm.com>
1039 PR tree-optimization/70754
1040 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
1041 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
1042 combined stmt before it if not NULL.
1043 (combine_chains): Process refs reversely and compute dominance point
1047 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1049 PR tree-optimization/79663
1050 * tree-predcom.c (combine_chains): Process refs in reverse order
1051 only for ZERO length chains, and add explaining comment.
1053 2017-11-15 Tamar Christina <tamar.christina@arm.com>
1055 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
1056 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
1057 * doc/invoke.texi (ARM Options): Add armv8.3-a.
1059 2017-11-15 Tamar Christina <tamar.christina@arm.com>
1061 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
1063 2017-11-15 Martin Liska <mliska@suse.cz>
1065 * tree-cfg.c (pass_warn_function_return::execute):
1066 Compare warn_return_type for greater than zero.
1068 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
1072 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
1073 to return true on Xeon and not on Xeon Phi.
1074 (ix86_check_avx256_register): Changed to ...
1075 (ix86_check_avx_upper_register): ... this. Add extra check for
1076 VALID_AVX512F_REG_OR_XI_MODE.
1077 (ix86_avx_u128_mode_needed): Changed
1078 ix86_check_avx256_register to ix86_check_avx_upper_register.
1079 (ix86_check_avx256_stores): Changed to ...
1080 (ix86_check_avx_upper_stores): ... this. Changed
1081 ix86_check_avx256_register to ix86_check_avx_upper_register.
1082 (ix86_avx_u128_mode_after): Changed
1083 avx_reg256_found to avx_upper_reg_found. Changed
1084 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
1085 (ix86_avx_u128_mode_entry): Changed
1086 ix86_check_avx256_register to ix86_check_avx_upper_register.
1087 (ix86_avx_u128_mode_exit): Ditto.
1088 * config/i386/i386.h: (host_detect_local_cpu): New define.
1090 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1092 * config/arm/xgene1.md (xgene1): Split into automatons
1093 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
1094 (xgene1_f_load): Adjust reservations and/or types.
1095 (xgene1_f_store): Likewise.
1096 (xgene1_load_pair): Likewise.
1097 (xgene1_store_pair): Likewise.
1098 (xgene1_fp_load1): Likewise.
1099 (xgene1_load1): Likewise.
1100 (xgene1_store1): Likewise.
1101 (xgene1_move): Likewise.
1102 (xgene1_alu): Likewise.
1103 (xgene1_simd): Likewise.
1104 (xgene1_bfm): Likewise.
1105 (xgene1_neon_load1): Likewise.
1106 (xgene1_neon_store1): Likewise.
1107 (xgene1_neon_logic): Likewise.
1108 (xgene1_neon_st1): Likewise.
1109 (xgene1_neon_ld1r): Likewise.
1110 (xgene1_alu_cond): Added.
1111 (xgene1_shift_reg): Likwise.
1112 (xgene1_bfx): Likewise.
1113 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
1115 2017-11-15 Jakub Jelinek <jakub@redhat.com>
1118 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
1120 (can_widen_mult_without_libcall): New function.
1121 (expand_mul_overflow): If only checking unsigned mul overflow,
1122 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
1123 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
1124 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
1125 (expand_DIVMOD): Formatting fix.
1126 * expmed.h (expand_mult): Add NO_LIBCALL argument.
1127 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
1128 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
1130 PR tree-optimization/82977
1131 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
1132 constructed temporary to strlen_to_stridx.put.
1134 2017-11-15 Martin Liska <mliska@suse.cz>
1136 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
1137 * configure: Regenerate.
1139 2017-11-15 Martin Liska <mliska@suse.cz>
1142 * config/sh/sh-mem.cc: Use proper probability for
1145 2017-11-14 Jeff Law <law@redhat.com>
1147 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
1148 the red zone for stack_clash_protection_final_dynamic_probe targets
1149 when the total dynamic stack size is zero bytes.
1151 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
1152 blocks is post order.
1154 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
1156 * dumpfile.h (TDF_COMPARE_DEBUG): New.
1157 * final.c (rest_of_clean_state): Set it for the
1158 -fcompare-debug dump.
1159 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
1160 class when TDF_COMPARE_DEBUG is set.
1162 * dwarf2out.c (gen_producer_string): Discard
1165 2017-11-15 Joseph Myers <joseph@codesourcery.com>
1168 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
1169 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
1170 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
1171 (__TGMATH_REAL_2_3): Remove macros.
1172 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
1173 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
1175 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
1177 (remquo): Define using __TGMATH_REAL_3.
1179 2017-11-14 Jeff Law <law@redhat.com>
1181 * vr-values.c: New file with contents extracted from tree-vrp.c.
1182 * Makefile.in (OBJS): Add vr-values.o
1183 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
1184 (set_value_range, set_and_canonicalize_value_range): Likewise.
1185 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
1186 (value_range_constant_singleton, symbolic_range_p): Likewise.
1187 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
1188 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
1189 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
1190 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
1191 (find_case_label_range, find_case_label_index): Likewise.
1192 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
1193 (range_int_cst_singleton_p, value_inside_range): Likewise.
1194 (get_single_symbol): Likewise.
1195 (switch_update): Move structure definition here.
1196 (to_remove_edges, to_update_switch_stmts): Provide externs.
1197 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
1198 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
1199 (vrp_val_is_min, set_value_range): Likewise.
1200 (set_and_canonicalize_value_range, copy_value_range): Likewise.
1201 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
1202 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
1203 (range_is_nonnull, range_int_cst_p): Likewwise.
1204 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
1205 (get_single_symbol, operand_less_p): Likewise
1206 (compare_values_warnv, compare_values): Likewise.
1207 (value_inside_range, value_range_constant_singleton): Likewise.
1208 (zero_nonzero_bitgs_from_vr): Likewise.
1209 (extract_range_from_binary_expr_1): Likewise.
1210 (overflow_comparison_p): Likewise.
1211 (to_remove_edges, to_update_switch_stmts): Likewise.
1212 (find_case_label-index, find_case_label_range): Likewise.
1213 (switch_update, set_value_range_to_nonnegative): Remove.
1214 (set_value_range_to_truthvalue): Likewise.
1215 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
1216 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
1217 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
1218 (find_case_label_ranges, test_for_singularity): Likewise.
1219 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
1220 (x_vr_values): Move to its remaining use site.
1222 2017-11-10 Jeff Law <law@redhat.com>
1224 * vr-values.h (VR_INITIALIZER): Move #define here.
1225 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
1226 * Makefile.in (OBJS): Add tree-evrp.o
1227 * tree-vrp.h (assert_info): Move structure definition here.
1228 (set_value_range_to_varying): Prototype.
1229 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
1230 (infer_value_range, register_edge_assert_for): Likewise.
1231 (stmt_interesting_for_vrp): Likewise.
1232 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
1233 (set_value_range_to_varying): No longer static.
1234 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
1235 (infer_value_range, register_edge_assert_for): Likewise.
1237 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
1239 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
1240 generate the XXBRD instruction.
1242 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
1243 (rs6000_builtin_type_compatible): Treat _Float128 and long double
1244 as being compatible if -mabi=ieeelongdouble.
1245 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
1246 to setup float128 built-ins with hardware support.
1247 (BU_FLOAT128_HW_2): Likewise.
1248 (BU_FLOAT128_HW_3): Likewise.
1249 (BU_FLOAT128_HW_VSX_1): Likewise.
1250 (BU_FLOAT128_HW_VSX_2): Likewise.
1251 (scalar_extract_expq): Change float128 built-in functions to
1252 accommodate having both KFmode and TFmode functions. Use the
1253 KFmode variant as the default.
1254 (scalar_extract_sigq): Likewise.
1255 (scalar_test_neg_qp): Likewise.
1256 (scalar_insert_exp_q): Likewise.
1257 (scalar_insert_exp_qp): Likewise.
1258 (scalar_test_data_class_qp): Likewise.
1259 (sqrtf128_round_to_odd): Delete processing the round to odd
1260 built-in functions as special built-in functions, and define them
1261 as float128 built-ins. Use the KFmode variant as the default.
1262 (truncf128_round_to_odd): Likewise.
1263 (addf128_round_to_odd): Likewise.
1264 (subf128_round_to_odd): Likewise.
1265 (mulf128_round_to_odd): Likewise.
1266 (divf128_round_to_odd): Likewise.
1267 (fmaf128_round_to_odd): Likewise.
1268 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1269 support for KFmode and TFmode xststdcqp calls.
1270 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
1271 point, switch the built-in handlers for the get/set float128
1272 exponent, get float128 mantissa, float128 test built-ins, and the
1273 float128 round to odd built-in functions. Eliminate creating the
1274 float128 round to odd built-in functions as special built-ins.
1275 (rs6000_init_builtins): Eliminate special creation of the float128
1276 round to odd built-in functions.
1277 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
1278 function insns to support both TFmode and KFmode variants.
1279 (xsxsigqp_<mode>): Likewise.
1280 (xsiexpqpf_<mode>): Likewise.
1281 (xsiexpqp_<mode>): Likewise.
1282 (xststdcqp_<mode>): Likewise.
1283 (xststdcnegqp_<mode>): Likewise.
1284 (xststdcqp_<mode>): Likewise.
1286 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1288 * tree-ssa-threadupdate.c (compute_path_counts): Remove
1289 unused path_in_freq_ptr parameter.
1290 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
1292 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1294 * ipa-inline.c (edge_badness): Dump sreal frequency.
1295 (compute_inlined_call_time): Match natural implementaiton ...
1296 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
1297 forgotten division by CGRAPH_FREQ_BASE.
1299 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1301 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
1302 Solaris 11. Update comment.
1303 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
1305 * config/sol2.h (STARTFILE_SPEC): Likewise.
1306 * configure: Regenerate.
1308 2017-11-14 Carl Love <cel@us.ibm.com>
1310 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
1311 le_ and be_ prefixes to swap* variables. Remove
1312 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
1315 2017-11-14 Jason Merrill <jason@redhat.com>
1317 Support GTY((cache)) on hash_map.
1318 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
1319 (ggc_cache_remove): Override it instead of ggc_mx.
1320 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
1321 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
1322 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
1323 (simple_cache_map_traits): Override maybe_mx.
1324 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
1325 (hash_map): Friend gt_cleare_cache.
1326 (gt_cleare_cache): New.
1327 * tree.h (tree_cache_traits): New hash_map traits class.
1328 (tree_cache_map): New typedef.
1330 2017-11-14 Richard Biener <rguenther@suse.de>
1332 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
1333 paramter and handling.
1334 (cleanup_control_flow_bb): Likewise.
1335 (cleanup_control_flow_pre): New helper performing a DFS walk
1336 to call cleanup_control_flow_bb in PRE order.
1337 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
1338 via cleanup_control_flow_pre.
1340 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
1342 * config/aarch64/aarch64-simd.md
1343 (aarch64_simd_bsl<mode>_internal): Remove DImode.
1344 (*aarch64_simd_bsl<mode>_alt): Likewise.
1345 (aarch64_simd_bsldi_internal): New.
1346 (aarch64_simd_bsldi_alt): Likewise.
1348 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1350 * tracer.c (better_p): Do not compare frequencies.
1351 * reg-stack.c (better_edge): Likewise.
1352 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
1355 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1357 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
1358 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
1359 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
1360 * lto-streamer-in.c (input_function): Use update_max_bb_count.
1361 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
1362 * predict.c (maybe_hot_frequency_p): Inline to ...
1363 (maybe_hot_count_p): ... here; rewrite to counts.
1364 (counts_to_freqs): Rename to ...
1365 (update_max_bb_count): ... this one.
1366 (expensive_function_p): Use counts.
1367 (estimate_bb_frequencies): Update.
1368 (rebuild_frequencies): Update.
1369 * predict.h (counts_to_freqs): Rename to ...
1370 (update_max_bb_count): ... this one.
1371 * profile.c (compute_branch_probabilities): Add debug info
1372 * tree-inline.c (expand_call_inline): Update debug info.
1373 (optimize_inline_calls): Use update_max_bb_count..
1374 (tree_function_versioning): Use update_max_bb_count..
1375 * value-prof.c (gimple_value_profile_transformations):
1376 Do not use update_max_bb_count.
1378 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1380 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
1381 always use frequencies.
1383 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1385 * bb-reorder.c: Remove frequencies from comments.
1386 (better_edge_p): Use profile counts.
1387 (find_traces): Dump profile counts.
1388 (rotate_loop): Use profile counts.
1389 (find_traces_1_round): Likewise.
1390 (connect_better_edge_p): Use counts instead of probabilities for
1392 (copy_bb_p): Drop early check for non-0 frequency.
1393 (sanitize_hot_paths): Update comments.
1395 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1397 * ipa-split.c (struct split_point): Add count.
1398 (consider_split): Do not compute incoming frequency; compute incoming
1399 count and store it to split_point.
1400 (split_function): Set count of the call to split part correctly.
1402 2017-11-13 Carl Love <cel@us.ibm.com>
1404 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
1406 2017-11-13 Tom Tromey <tom@tromey.com>
1408 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
1410 2017-11-13 Carl Love <cel@us.ibm.com>
1412 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1413 Add support for builtins:
1414 unsigned int vec_first_{,miss}_match_{,or_eos}index,
1415 vector {un,}signed {char,int,short},
1416 vector {un,}signed {char,int,short}) arguments.
1417 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
1418 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
1419 Add BU_P9V_AV_2 expansions for the builtins.
1420 * config/rs6000/altivec.h (vec_first_match_index,
1421 vec_first_mismatch_index, vec_first_match_or_eos_index,
1422 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
1423 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
1424 new extern declaration.
1425 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
1426 * config/rs6000/vsx.md (first_match_index_<mode>,
1427 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
1428 first_mismatch_or_eos_index_<mode>): Add define expand.
1429 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
1430 * doc/extend.texi: Update the built-in documenation file for the new
1433 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
1435 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
1436 into the min/max operations for _Float<N> and _Float<N>X types.
1438 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1441 * dwarf2out.c (do_eh_frame): New static variable.
1442 (dwarf2out_begin_prologue): Set it.
1443 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
1445 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1447 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
1449 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
1451 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1454 * debug.h (dwarf2out_do_eh_frame): Declare.
1455 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
1456 (dwarf2out_do_frame): Use it.
1457 (dwarf2out_do_cfi_asm): Likewise.
1458 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
1459 (dwarf2out_assembly_start): Likewise.
1460 (dwarf2out_begin_prologue): Fix comment.
1461 * toplev.c (compile_file): Always call dwarf2out_frame_finish
1462 if the target needs either debug or unwind DWARF2 info.
1463 * lto-opts.c (lto_write_options): Do not save -fexceptions,
1464 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
1465 -ftrapping-math, -ftrapv and -fwrapv.
1467 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1469 * cgraph.c (cgraph_edge::sreal_frequency): New function.
1470 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
1471 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
1472 (estimate_edge_size_and_time): Likewise.
1473 (ipa_merge_fn_summary_after_inlining): Likewise.
1474 * ipa-inline.c (cgraph_freq_base_rec): Remove.
1475 (compute_uninlined_call_time): Use sreal_frequency.
1476 (compute_inlined_call_time): Likewise.
1477 (ipa_inline): Do not initialize cgraph_freq_base_rec.
1478 * profile-count.c: Include sreal.h.
1479 (profile_count::to_sreal_scale): New.
1480 * profile-count.h: Forward declare sreal.
1481 (profile_count::to_sreal_scale): Declare.
1483 2017-11-13 Nathan Sidwell <nathan@acm.org>
1485 * diagnostic.c (maybe_line_and_column): New.
1486 (diagnostic_get_location_text): Use it.
1487 (diagnostic_report_current_module): Likewise.
1488 (test_diagnostic_get_location_text): Add tests.
1490 2017-11-13 Luis Machado <luis.machado@linaro.org>
1492 * doc/md.texi (Specifying processor pipeline description): Fix
1493 incorrect latency for the div instruction example.
1495 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1497 PR tree-optimization/78821
1498 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
1499 that bit_not_p is the same.
1500 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1501 (split_group): Count precisely bit_not_p bits in each statement.
1502 (invert_op): New function.
1503 (imm_store_chain_info::output_merged_store): Use invert_op to
1504 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
1505 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
1507 2017-11-13 Martin Liska <mliska@suse.cz>
1509 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
1510 (struct source_info): Likewise.
1511 (add_branch_counts): Likewise.
1512 (add_line_counts): Likewise.
1513 (function_summary): Likewise.
1514 (output_intermediate_line): Likewise.
1515 (generate_results): Likewise.
1517 2017-11-13 Martin Liska <mliska@suse.cz>
1519 * gcov.c (struct block_info): Remove typedef for block_t.
1520 (struct line_info): Likewise.
1521 (line_info::has_block): Likewise.
1522 (EXIT_BLOCK): Likewise.
1523 (unblock): Likewise.
1524 (circuit): Likewise.
1525 (get_cycles_count): Likewise.
1526 (process_file): Likewise.
1527 (read_graph_file): Likewise.
1528 (solve_flow_graph): Likewise.
1529 (find_exception_blocks): Likewise.
1530 (add_line_counts): Likewise.
1531 (accumulate_line_info): Likewise.
1532 (output_line_details): Likewise.
1534 2017-11-13 Martin Liska <mliska@suse.cz>
1536 * gcov.c (struct arc_info): Remove typedef for arc_t.
1537 (struct line_info): Likewise.
1538 (add_branch_counts): Likewise.
1539 (output_branch_count): Likewise.
1540 (function_info::~function_info): Likewise.
1541 (circuit): Likewise.
1542 (output_intermediate_line): Likewise.
1543 (read_graph_file): Likewise.
1544 (solve_flow_graph): Likewise.
1545 (find_exception_blocks): Likewise.
1546 (add_line_counts): Likewise.
1547 (accumulate_line_info): Likewise.
1548 (output_line_details): Likewise.
1549 (output_function_details): Likewise.
1551 2017-11-13 Martin Liska <mliska@suse.cz>
1553 * gcov.c (struct function_info): Remove typedef for function_t.
1554 (struct source_info): Likewise.
1555 (source_info::get_functions_at_location): Likewise.
1556 (solve_flow_graph): Likewise.
1557 (find_exception_blocks): Likewise.
1558 (add_line_counts): Likewise.
1559 (output_intermediate_file): Likewise.
1560 (process_file): Likewise.
1561 (generate_results): Likewise.
1562 (release_structures): Likewise.
1563 (read_graph_file): Likewise.
1564 (read_count_file): Likewise.
1565 (accumulate_line_counts): Likewise.
1566 (output_lines): Likewise.
1568 2017-11-13 Martin Liska <mliska@suse.cz>
1570 * gcov.c (function_info::function_info): Remove num_counts
1571 and add vector<gcov_type>.
1572 (function_info::~function_info): Use the vector.
1573 (process_file): Likewise.
1574 (read_graph_file): Likewise.
1575 (read_count_file): Likewise.
1576 (solve_flow_graph): Likewise.
1578 2017-11-13 Martin Liska <mliska@suse.cz>
1580 * gcov.c (function_info::is_artificial): New function.
1581 (process_file): Erase all artificial early.
1582 (generate_results): Skip as all artificial are already
1585 2017-11-13 Martin Liska <mliska@suse.cz>
1587 * gcov.c (read_graph_file): Store to global vector of functions.
1588 (read_count_file): Iterate the vector.
1589 (process_file): Likewise.
1590 (generate_results): Likewise.
1591 (release_structures): Likewise.
1593 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1595 PR tree-optimization/82954
1596 * gimple-ssa-store-merging.c
1597 (imm_store_chain_info::coalesce_immediate_stores): If
1598 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
1600 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
1602 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
1603 Upddate call to ENDIAN_LANE_N.
1604 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
1605 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1606 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
1607 and use aarch64_endian_lane_rtx.
1608 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
1610 2017-11-12 Tom de Vries <tom@codesourcery.com>
1612 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
1614 2017-11-12 Tom de Vries <tom@codesourcery.com>
1616 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
1618 2017-11-12 Tom de Vries <tom@codesourcery.com>
1620 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
1622 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1623 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1624 * defaults.h (ASM_OUTPUT_LABELREF): Same.
1626 2017-11-11 Martin Sebor <msebor@redhat.com>
1629 * doc/extend.texi (attribute nonstring): Remove spurious argument.
1632 * prefic.c (translate_name): Replace strncpy with memcpy to
1633 avoid -Wstringop-truncation.
1635 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1637 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
1639 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1641 * predict.c (maybe_hot_frequency_p): Do not use cfun.
1643 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1645 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
1648 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1650 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1651 enable generating XXBRH if the value is in a vector register.
1652 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1653 value is in a vector register.
1654 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1655 register to register bswap64's instead of doing the GPR sequence
1656 used on previous machines.
1657 (bswapdi2_xxbrd): New insn.
1658 (bswapdi2_reg): Disallow on ISA 3.0.
1659 (register to register bswap64 splitter): Do not split the insn on
1660 ISA 3.0 systems that use XXBRD.
1662 2017-11-10 Martin Sebor <msebor@redhat.com>
1665 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1667 (find_subframework_file): Same.
1669 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1671 * auto-profile.c (afdo_indirect_call): Drop frequency.
1672 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1673 (cgraph_node::create_edge): Drop frequency argument.
1674 (cgraph_node::create_indirect_edge): Drop frequency argument.
1675 (cgraph_edge::make_speculative): Drop frequency arguments.
1676 (cgraph_edge::resolve_speculation): Do not update frequencies
1677 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1678 (cgraph_node::dump): Check consistency in IPA mode.
1679 (cgraph_edge::maybe_hot_p): Use IPA counter.
1680 (cgraph_edge::verify_count_and_frequency): Rename to ...
1681 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1682 (cgraph_node::verify_node): Update.
1683 * cgraph.h (struct cgraph_edge): Drop frequency.
1684 (cgraph_edge::frequency): New function.
1685 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1687 (cgraph_edge::rebuild_edges): Likewise.
1688 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1689 (duplicate_thunk_for_node): Do not pass frequency.
1690 (cgraph_node::create_clone): Scale only counts.
1691 (cgraph_node::create_virtual_clone): Do not pass frequency.
1692 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1693 (cgraph_node::create_version_clone): Do not pass frequency.
1694 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1695 (cgraph_node::expand_thunk): Do not pass frequency.
1696 (cgraph_node::create_wrapper): Do not pass frequency.
1697 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1699 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1700 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1701 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1702 (gather_caller_stats): Use frequency function.
1703 (ipcp_cloning_candidate_p): Use frequency function.
1704 (ipcp_propagate_stage): Use frequency function.
1705 (get_info_about_necessary_edges): Use frequency function.
1706 (update_profiling_info): Update only IPA profile.
1707 (update_specialized_profile): Use frequency functoin.
1708 (perhaps_add_new_callers): Update only IPA profile.
1709 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1710 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1711 (dump_ipa_call_summary): Use frequency function.
1712 (estimate_edge_size_and_time): Use frequency function.
1713 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1714 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1715 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1716 (update_noncloned_counts): ... ths one; scale counts only.
1717 (clone_inlined_nodes): Do not scale frequency.
1718 (inline_call): Do not pass frequency.
1719 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1720 (compute_inlined_call_time): Use IPA profile.
1721 (want_inline_small_function_p): Use IPA profile.
1722 (want_inline_self_recursive_call_p): Use IPA profile.
1723 (edge_badness): Use IPA profile.
1724 (lookup_recursive_calls): Use IPA profile.
1725 (recursive_inlining): Do not pass frequency.
1726 (resolve_noninline_speculation): Do not update frequency.
1727 (inline_small_functions): Collect max of IPA profile.
1728 (dump_overall_stats): Dump IPA porfile.
1729 (dump_inline_stats): Dump IPA porfile.
1730 (ipa_inline): Collect IPA stats.
1731 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1732 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1733 (ipa_propagate_frequency): Use frequency function.
1734 (ipa_profile): Cleanup.
1735 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1736 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1737 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1738 (input_node): Do not stream frequency.
1739 (input_edge): Do not stream frequency.
1740 (merge_profile_summaries): Scale only IPA profiles.
1741 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1742 * predict.c (drop_profile): Do not recompute frequency.
1743 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1744 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1745 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1746 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1747 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1748 * tree-inline.c (copy_bb): Do not scale frequency.
1749 (expand_call_inline): Do not scale frequency.
1750 (tree_function_versioning): Do not scale frequency.
1751 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1753 2017-11-10 Julia Koval <julia.koval@intel.com>
1755 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1756 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1757 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1758 (_mm256_maskz_gf2p8affine_epi64_epi8)
1759 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1760 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1761 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1762 (__builtin_ia32_vgf2p8affineqb_v32qi)
1763 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1764 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1766 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1770 (arm_option_override): Refactor.
1771 (arm_option_reconfigure_globals): New.
1772 (arm_options_perform_arch_sanity_checks): New.
1773 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1775 (arm_options_perform_arch_sanity_checks): Likewise
1777 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1779 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1780 (power9-qpmul): New.
1781 * rs6000/rs6000.md ("type" attr): Add qmul.
1782 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1783 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1784 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1786 2017-11-10 Martin Sebor <msebor@redhat.com>
1789 * builtins.c (compute_objsize): Handle arrays that
1790 compute_builtin_object_size likes to fail for. Make extern.
1791 * builtins.h (compute_objsize): Declare.
1792 (check_strncpy_sizes): New function.
1793 (expand_builtin_strncpy): Call check_strncpy_sizes.
1794 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1795 -Wstringop-truncation.
1796 (gimple_fold_builtin_strncat): Same.
1797 * gimple.c (gimple_build_call_from_tree): Set call location.
1798 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1799 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1800 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1801 (handle_builtin_strlen): Use strlen_to_stridx.
1802 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1804 Use strlen_to_stridx.
1805 (pass_strlen::execute): Release strlen_to_stridx.
1806 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1807 (-Wstringop-truncation): Document new option.
1809 2017-11-10 Martin Liska <mliska@suse.cz>
1811 PR gcov-profile/82702
1812 * gcov.c (main): Handle intermediate files in a different
1814 (get_gcov_intermediate_filename): New function.
1815 (output_gcov_file): Remove support of intermediate files.
1816 (generate_results): Allocate intermediate file.
1817 (release_structures): Clean-up properly fn_end.
1818 (output_intermediate_file): Start iterating with line 1.
1820 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1822 PR tree-optimization/82929
1823 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1824 ops_swapped_p non-static data member.
1825 (store_immediate_info::store_immediate_info): Clear it.
1826 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1827 ops set ops_swapped_p.
1828 (count_multiple_uses): Handle ops_swapped_p.
1830 2017-11-10 Martin Liska <mliska@suse.cz>
1832 * coverage.c (coverage_init): Stream information about
1833 support of has_unexecuted_blocks.
1834 * doc/gcov.texi: Document that.
1835 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1836 * gcov.c (read_graph_file): Likewise.
1837 (output_line_beginning): Fix a small issue with
1840 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1842 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1843 reference of trivial component.
1845 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1848 * gimple-ssa-store-merging.c
1849 (pass_store_merging::terminate_all_aliasing_chains): For
1850 gimple_store_p stmts also call refs_output_dependent_p.
1852 PR rtl-optimization/82913
1853 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1856 2017-11-09 Jeff Law <law@redhat.com>
1858 * vr-values.h: New file with vr_values class.
1859 * tree-vrp.c: Include vr-values.h
1860 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1861 data objects into the vr_values class.
1862 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1863 (get_value_range): Make it a member function within vr_values class.
1864 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1865 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1866 (op_with_constant_singleton_value_range): Likewise.
1867 (extract_range_for_var_from_comparison_expr): Likewise.
1868 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1869 (extract_range_from_binary_expr): Likewise.
1870 (extract_range_from_unary_expr): Likewise.
1871 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1872 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1873 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1874 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1875 (compare_name_with_value, compare_names): Likewise.
1876 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1877 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1878 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1879 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1880 (extract_range_from_phi_node): Likewise.
1881 (simplify_truth_ops_using_ranges): Likewise.
1882 (simplify_div_or_mod_using_ranges): Likewise.
1883 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1884 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1885 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1886 (simplify_float_conversion_using_ranges): Likewise.
1887 (simplify_internal_call_using_ranges): Likewise.
1888 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1889 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1890 poitner via x_vr_values for calls into gimple folder.
1891 (vrp_initialize_lattice): Make this the vr_values ctor.
1892 (vrp_free_lattice): Make this the vr_values dtor.
1893 (set_vr_value): New function.
1894 (class vrp_prop): Add vr_values data member. Add various member
1895 functions as well as member functions that delegate to vr_values.
1896 (check_array_ref): Make a member function within vrp_prop class.
1897 (search_for_addr_array, vrp_initialize): Likewise.
1898 (vrp_finalize): Likewise. Revamp to avoid direct access to
1899 vr_value, values_propagated, etc.
1900 (check_array_bounds): Extract vrp_prop class instance pointer from
1901 walk info structure. Use it to call member functions.
1902 (check_all_array_refs): Make a member function within vrp_prop class.
1903 Smuggle class instance pointer via walk info structure.
1904 (x_vr_values): New local static.
1905 (vrp_valueize): Use x_vr_values to get class instance.
1906 (vr_valueize_1): Likewise.
1907 (class vrp_folder): Add vr_values data member. Add various member
1908 functions as well as member functions that delegate to vr_values.
1909 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1910 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1911 class instance from vr_values. Use it to call member functions.
1912 (vrp_dom_walker): Add vr_values data member.
1913 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1914 instance via x_vr_values.
1915 (identify_jump_threads): Accept vr_values as argument. Store
1916 it into the walker structure.
1917 (evrp_dom_walker): Add vr_values class data member. Add various
1919 (evrp_dom_walker::try_find_new_range): Use vr_values data
1920 member to access the memory allocator.
1921 (evrp_dom_walker::before_dom_children): Store vr_values class
1922 instance into the vrp_folder class.
1923 (evrp_dom_walker::push_value_range): Rework to avoid direct
1924 access to num_vr_values and vr_value.
1925 (evrp_dom_walker::pop_value_range): Likewise.
1926 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1927 Use vr_values to get to dump_all_value_ranges member function.
1928 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1929 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1930 Pass vr_values class instance down to identify_jump_threads.
1931 Remove call to vrp_free_lattice.
1932 (debug_all_value_ranges): Remove.
1934 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1935 (vrp_folder): Likewise.
1937 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1938 Get it from the existing bitmap instead.
1939 (vrp_intersect_ranges_1): Likewise.
1941 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1943 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1945 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1946 set bit_not_p to it.
1947 (imm_store_chain_info::coalesce_immediate_stores): Break group
1948 if bit_not_p is different.
1949 (count_multiple_uses, split_group,
1950 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1951 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1952 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1953 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1954 to store_immediate_info ctor.
1956 2017-11-09 Jim Wilson <jimw@sifive.com>
1958 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1959 (scan_prog_file): Likewise.
1961 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1963 * bb-reorder.c (max_entry_frequency): Remove.
1964 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1965 connect_traces, push_to_next_round_p): Remove prototypes.
1966 (find_traces_1_round): Use counts only.
1967 (push_to_next_round_p): Likewise.
1968 (find_traces): Likewise.
1969 (rotate_loop): Likewise.
1970 (find_traces_1_round): Likewise.
1971 (connect_traces): Likewise.
1972 (edge_order): Likewise.
1974 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1976 * config/arm/arm.c (output_return_instruction): Add comments to
1977 indicate requirement for cmse_nonsecure_entry return to account
1978 for the size of clearing instruction output here.
1979 (thumb_exit): Likewise.
1980 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1981 return in hardfloat mode.
1983 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1985 * config/rs6000/rs6000.c (machine_function): Add a bool,
1986 "toc_is_wrapped_separately".
1987 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1988 if it wasn't explicitly set or unset, we are optimizing for speed, and
1989 doing separate shrink-wrapping.
1990 (rs6000_get_separate_components): Enable the TOC component if
1991 saving the TOC register in the prologue.
1992 (rs6000_components_for_bb): Handle the TOC component.
1993 (rs6000_emit_prologue_components): Store the TOC register where needed.
1994 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1995 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1997 2017-11-09 Martin Jambor <mjambor@suse.cz>
1999 * ipa-param-manipulation.c: New file.
2000 * ipa-param-manipulation.h: Likewise.
2001 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
2002 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
2003 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
2004 (ipa_parm_adjustment): Likewise.
2005 (ipa_parm_adjustment_vec): Likewise.
2006 (ipa_get_vector_of_formal_parms): Moved declaration to
2007 ipa-param-manipulation.h.
2008 (ipa_get_vector_of_formal_parm_types): Likewise.
2009 (ipa_modify_formal_parameters): Likewise.
2010 (ipa_modify_call_arguments): Likewise.
2011 (ipa_combine_adjustments): Likewise.
2012 (ipa_dump_param_adjustments): Likewise.
2013 (ipa_modify_expr): Likewise.
2014 (ipa_get_adjustment_candidate): Likewise.
2015 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
2016 ipa-param-manipulation.c.
2017 (ipa_get_vector_of_formal_parm_types): Likewise.
2018 (ipa_modify_formal_parameters): Likewise.
2019 (ipa_modify_call_arguments): Likewise.
2020 (ipa_modify_expr): Likewise.
2021 (get_ssa_base_param): Likewise.
2022 (ipa_get_adjustment_candidate): Likewise.
2023 (index_in_adjustments_multiple_times_p): Likewise.
2024 (ipa_combine_adjustments): Likewise.
2025 (ipa_dump_param_adjustments): Likewise.
2026 * tree-sra.c: Also include ipa-param-manipulation.h
2027 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
2030 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2031 Alan Hayward <alan.hayward@arm.com>
2032 David Sherwood <david.sherwood@arm.com>
2034 * doc/sourcebuild.texi (vect_masked_store): Document.
2036 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2037 Alan Hayward <alan.hayward@arm.com>
2038 David Sherwood <david.sherwood@arm.com>
2040 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
2042 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2043 Alan Hayward <alan.hayward@arm.com>
2044 David Sherwood <david.sherwood@arm.com>
2046 * doc/sourcebuild.texi (vect_variable_length): Document.
2048 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2049 Alan Hayward <alan.hayward@arm.com>
2050 David Sherwood <david.sherwood@arm.com>
2052 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
2054 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2055 Alan Hayward <alan.hayward@arm.com>
2056 David Sherwood <david.sherwood@arm.com>
2058 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
2060 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2061 Alan Hayward <alan.hayward@arm.com>
2062 David Sherwood <david.sherwood@arm.com>
2064 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
2065 previously undocumented selectors.
2066 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
2068 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2070 * doc/rtl.texi (const_vector): Say that elements can be
2071 const_wide_ints too.
2072 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
2073 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
2074 (gen_vec_duplicate): Use it instead of CONSTANT_P.
2075 * optabs.c (expand_vector_broadcast): Likewise.
2077 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2078 Alan Hayward <alan.hayward@arm.com>
2079 David Sherwood <david.sherwood@arm.com>
2081 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
2082 scaled index even if the unscaled address was invalid.
2083 Don't increase the complexity of using a scale in that case.
2085 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2086 Alan Hayward <alan.hayward@arm.com>
2087 David Sherwood <david.sherwood@arm.com>
2089 * doc/rtl.texi: Rewrite the subreg rules so that they partition
2090 the inner register into REGMODE_NATURAL_SIZE bytes rather than
2091 UNITS_PER_WORD bytes.
2092 * emit-rtl.c (validate_subreg): Divide subregs into blocks
2093 based on REGMODE_NATURAL_SIZE of the inner mode.
2094 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
2095 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
2096 * expmed.c (lowpart_bit_field_p): Divide the value up into
2097 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
2098 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
2099 whether something is likely to occupy more than one register.
2101 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
2104 * ipa-inline-transform.c (update_noncloned_frequencies): Use
2105 profile_count::adjust_for_ipa_scaling.
2106 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
2107 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
2109 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
2111 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2113 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
2114 (split_group): Add total_orig and total_new arguments, estimate the
2115 number of statements related to the store group without store merging
2116 and with store merging.
2117 (imm_store_chain_info::output_merged_store): Adjust split_group
2118 callers, punt if estimated number of statements with store merging
2119 is not smaller than estimated number of statements without it.
2121 (handled_load): Remove has_single_use checks.
2122 (pass_store_merging::process_store): Likewise.
2124 2017-11-09 Richard Biener <rguenther@suse.de>
2126 PR tree-optimization/82902
2127 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
2129 2017-11-09 Martin Liska <mliska@suse.cz>
2132 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
2135 2017-11-09 Martin Liska <mliska@suse.cz>
2137 PR tree-optimization/82669
2138 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
2140 2017-11-09 Martin Liska <mliska@suse.cz>
2142 PR gcov-profile/48463
2143 * coverage.c (coverage_begin_function): Output also end locus
2144 of a function and information whether the function is
2146 * gcov-dump.c (tag_function): Parse and print the information.
2147 * gcov.c (INCLUDE_MAP): Add include.
2148 (INCLUDE_SET): Likewise.
2149 (struct line_info): Move earlier in the source file because
2150 of vector<line_info> in function_info structure.
2151 (line_info::line_info): Likewise.
2152 (line_info::has_block): Likewise.
2153 (struct source_info): Add new member index.
2154 (source_info::get_functions_at_location): New function.
2155 (function_info::group_line_p): New function.
2156 (output_intermediate_line): New function.
2157 (output_intermediate_file): Use the mentioned function.
2158 (struct function_start): New.
2159 (struct function_start_pair_hash): Likewise.
2160 (process_file): Add code that identifies group functions.
2161 Assign lines either to global or function scope.
2162 (generate_results): Skip artificial functions.
2163 (find_source): Assign index for each source file.
2164 (read_graph_file): Read new flag artificial and end_line.
2165 (add_line_counts): Assign it either to global of function scope.
2166 (accumulate_line_counts): Isolate core of the function to
2167 accumulate_line_info and call it for both function and global
2169 (accumulate_line_info): New function.
2170 (output_line_beginning): Fix GNU coding style.
2171 (print_source_line): New function.
2172 (output_line_details): Likewise.
2173 (output_function_details): Likewise.
2174 (output_lines): Iterate both source (global) scope and function
2176 (struct function_line_start_cmp): New class.
2177 * doc/gcov.texi: Reflect changes in documentation.
2179 2017-11-09 Jakub Jelinek <jakub@redhat.com>
2182 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
2183 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
2184 and similarly for not instead of neg.
2186 2017-11-08 Andi Kleen <ak@linux.intel.com>
2188 * config/i386/i386.opt: Add -mforce-indirect-call.
2189 * config/i386/predicates.md: Check for flag_force_indirect_call.
2190 * doc/invoke.texi: Document -mforce-indirect-call
2192 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
2194 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
2196 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
2197 config/riscv/riscv.c (predict.h): New include.
2198 (riscv_slow_unaligned_access_p): No longer static.
2199 (riscv_block_move_straight): Add require.
2200 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
2202 2017-11-08 Jakub Jelinek <jakub@redhat.com>
2205 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
2206 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
2207 nonimmediate_operand predicate for operand 1 instead of
2210 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2212 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
2214 * config/aarch64/constraints.md (Uml): New constraint.
2215 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
2218 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2220 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
2221 of two vec_duplicates into a vec_concat.
2223 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2225 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
2226 Simplify vec_merge of vec_duplicate and vec_concat.
2227 * config/aarch64/constraints.md (Utq): New constraint.
2228 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
2231 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2233 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
2234 Simplify vec_merge of vec_duplicate and const_vector.
2235 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
2237 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
2238 mode iterator. Update predicate on operand 1 to
2239 handle non-const_vec constants. Delete constraints.
2240 (*aarch64_combinez_be<mode>): Likewise for operand 2.
2242 2017-11-08 Jakub Jelinek <jakub@redhat.com>
2244 PR tree-optimization/78821
2245 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
2247 (store_operand_info::store_operand_info): Initialize it to false.
2248 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
2249 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
2250 store in the group, and if chain_info is non-NULL, to ignore altogether
2252 (compatible_load_p): Fail if bit_not_p does not match.
2253 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
2254 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
2255 (pass_store_merging::process_store): Adjust
2256 terminate_all_aliasing_chains calls to pass NULL in all current spots,
2257 call terminate_all_aliasing_chains newly when adding a store into
2258 a chain with non-NULL chain_info.
2260 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
2262 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
2264 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
2266 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2268 (aarch64_layout_frame): Initialise emit_frame_chain.
2269 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
2270 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
2272 2017-11-08 Martin Liska <mliska@suse.cz>
2274 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
2275 of gimple_call_internal_p.
2277 2017-11-07 Tom de Vries <tom@codesourcery.com>
2279 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
2281 2017-11-07 Tom de Vries <tom@codesourcery.com>
2283 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
2286 2017-11-08 Martin Liska <mliska@suse.cz>
2289 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
2291 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2293 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
2295 2017-11-07 Jakub Jelinek <jakub@redhat.com>
2298 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
2299 (*cmp<mode>_ccz_1): New insn with $k alternative.
2302 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
2303 RTX_COMM_COMPARE as commutative as well.
2304 (ix86_binary_operator_ok): Formatting fix.
2305 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
2306 *<code><mode>3<mask_name><round_saeonly_name>,
2307 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
2308 *<s>mul<mode>3_highpart<mask_name>,
2309 *vec_widen_umult_even_v16si<mask_name>,
2310 *vec_widen_umult_even_v8si<mask_name>,
2311 *vec_widen_umult_even_v4si<mask_name>,
2312 *vec_widen_smult_even_v16si<mask_name>,
2313 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
2314 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
2315 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
2316 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
2317 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
2318 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
2319 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
2320 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
2321 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
2322 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
2323 ix86_binary_operator_ok. Formatting fixes.
2324 (*<plusminus_insn><mode>3<mask_name><round_name>,
2325 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
2328 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
2330 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
2331 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
2332 eq and ne if TARGET_ISEL.
2333 (cmp): New code_iterator.
2334 (UNS, UNSU_, UNSIK): New code_attrs.
2335 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
2336 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
2338 ("eq<mode>3"): ... this.
2339 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
2341 ("ne<mode>3"): ... this.
2343 2017-11-07 Julia Koval <julia.koval@intel.com>
2346 * common/config/i386/i386-common.c
2347 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
2348 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
2349 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
2350 * config/i386/i386.opt: Ditto.
2351 * config/i386/i386.c (ix86_target_string): Ditto.
2352 (ix86_option_override_internal): Ditto.
2353 (ix86_init_mpx_builtins): Move MPX to args2.
2354 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
2355 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2356 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2357 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2358 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2359 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2360 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
2362 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
2365 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
2366 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
2367 (zero-extendsidi peephole2): Remove peephole.
2369 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2372 * stor-layout.c: Include attribs.h.
2373 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
2374 explicit lookup of "aligned" attribute.
2376 2017-11-07 Andrew Waterman <andrew@sifive.com>
2378 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
2379 (riscv_expand_block_move): Likewise.
2380 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
2382 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
2383 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
2384 * config/riscv/riscv.c (riscv_block_move_straight): New function.
2385 (riscv_adjust_block_mem): Likewise.
2386 (riscv_block_move_loop): Likewise.
2387 (riscv_expand_block_move): Likewise.
2388 gcc/config/riscv/riscv.md (movmemsi): New pattern.
2390 2017-11-07 Michael Clark <michaeljclark@mac.com>
2392 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
2393 (MUSL_DYNAMIC_LINKER): Likewise.
2395 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
2397 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
2398 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
2401 2017-11-07 Richard Biener <rguenther@suse.de>
2403 * match.pd: Fix build.
2405 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
2406 Jackson Woodruff <jackson.woodruff@arm.com>
2408 PR tree-optimization/71026
2409 * match.pd: Canonicalize negate in division.
2411 2017-11-07 Sudakshina Das <sudi.das@arm.com>
2414 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
2416 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2418 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
2419 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
2421 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2423 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
2424 non-scalar integral types.
2425 * match.pd (negate_expr_p): Handle MINUS_EXPR.
2426 (-(A-B), -(~A)): New transformations.
2428 2017-11-07 Tom de Vries <tom@codesourcery.com>
2430 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
2431 semicolon after "do {} while (0)".
2432 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2433 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2434 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2435 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2436 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2438 2017-11-07 Tom de Vries <tom@codesourcery.com>
2440 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
2441 after "do {} while (0)".
2442 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2443 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2444 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2445 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2446 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2448 2017-11-07 Tom de Vries <tom@codesourcery.com>
2451 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
2453 (arm_rtx_costs_internal): Add missing semicolon after
2454 HANDLE_NARROW_SHIFT_ARITH call.
2456 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2458 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2459 disable isel if it was not set explicitly.
2461 2017-11-06 James Bowman <james.bowman@ftdichip.com>
2463 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
2464 (add_type_attribute) likewise.
2466 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
2468 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
2469 of struct ix86_frame.
2470 (ix86_initial_elimination_offset): Likewise.
2471 (ix86_expand_split_stack_prologue): Likewise.
2473 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
2475 * tree-vrp.h (enum value_range_type): Update stale comment.
2477 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2478 Alan Hayward <alan.hayward@arm.com>
2479 David Sherwood <david.sherwood@arm.com>
2481 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
2482 (aarch64_expand_vec_perm_const): Take the number of units too.
2483 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
2484 (aarch64_expand_vec_perm_const): Likewise.
2485 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
2486 (vec_perm<mode>): Update accordingly.
2488 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2489 Alan Hayward <alan.hayward@arm.com>
2490 David Sherwood <david.sherwood@arm.com>
2492 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
2493 Take the number of units too.
2494 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
2495 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
2496 but check for a vector mode before rather than after the call.
2497 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
2498 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
2499 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
2500 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
2501 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
2502 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
2503 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
2504 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
2505 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
2506 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
2507 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
2508 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
2509 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
2510 (aarch64_sqdmull2_n<mode>): Update accordingly.
2512 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2513 Alan Hayward <alan.hayward@arm.com>
2514 David Sherwood <david.sherwood@arm.com>
2516 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
2517 the number of units too.
2518 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
2519 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
2520 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
2521 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
2522 (vec_store_lanesxi<mode>): Update accordingly.
2524 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2525 Alan Hayward <alan.hayward@arm.com>
2526 David Sherwood <david.sherwood@arm.com>
2528 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
2529 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
2530 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
2531 of units rather than the mode.
2532 * config/aarch64/iterators.md (nunits): New mode attribute.
2533 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
2534 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
2535 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
2536 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
2537 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
2538 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
2539 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
2540 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
2541 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
2542 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
2543 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2544 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
2545 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
2546 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
2547 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
2548 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
2549 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
2550 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
2551 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
2552 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
2553 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
2554 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
2555 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
2556 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2557 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
2558 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
2559 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2560 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
2561 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
2562 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
2563 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
2564 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
2565 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
2566 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
2567 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
2568 (aarch64_simd_vec_setv2di): Likewise.
2570 2017-11-06 Carl Love <cel@us.ibm.com>
2572 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
2574 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
2575 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
2577 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
2579 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
2580 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
2581 Add power 8 macro expansions.
2582 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
2583 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
2584 power 8 instructions. (VSX_XXBR): Add iterator.
2586 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
2588 * config/arm/arm.md (predicable_short_it): Change default to "no",
2589 improve documentation, remove uses that are identical to the default.
2590 (enabled_for_depr_it): Rename to enabled_for_short_it.
2591 * gcc/config/arm/arm-fixed.md (predicable_short_it):
2592 Remove default uses.
2593 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
2594 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
2595 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
2596 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
2598 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2601 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
2602 float128 helper macros, which are no longer used after deleting
2603 the old 'q' built-in functions, and moving the round to odd
2604 built-in functions to being special built-in functions.
2605 (BU_FLOAT128_2): Likewise.
2606 (BU_FLOAT128_1_HW): Likewise.
2607 (BU_FLOAT128_2_HW): Likewise.
2608 (BU_FLOAT128_3_HW): Likewise.
2609 (FABSQ): Delete old 'q' built-in functions.
2610 (COPYSIGNQ): Likewise.
2611 (SQRTF128_ODD): Move round to odd built-in functions to be
2612 special built-in functions, so that we can handle
2613 -mabi=ieeelongdouble.
2614 (TRUNCF128_ODD): Likewise.
2615 (ADDF128_ODD): Likewise.
2616 (SUBF128_ODD): Likewise.
2617 (MULF128_ODD): Likewise.
2618 (DIVF128_ODD): Likewise.
2619 (FMAF128_ODD): Likewise.
2620 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
2621 built-in names to 'f128'.
2622 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
2623 old 'q' built-in functions, as the machine independent code for
2624 'f128' built-in functions handles this.
2625 (rs6000_expand_builtin): Add expansion for float128 round to odd
2626 functions, keying off on -mabi=ieeelongdouble of whether to use
2627 the KFmode or TFmode variant.
2628 (rs6000_init_builtins): Initialize the _Float128 round to odd
2630 * doc/extend.texi (PowerPC Built-in Functions): Document the old
2631 _Float128 'q' built-in functions are now mapped into the new
2632 'f128' built-in functions.
2634 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
2636 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
2637 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
2639 2017-11-06 David Malcolm <dmalcolm@redhat.com>
2642 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
2643 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
2644 * toplev.c: Include "ipa-fnsummary.h".
2645 (toplev::finalize): Call ipa_fnsummary_c_finalize.
2647 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2649 PR tree-optimization/82838
2650 * gimple-ssa-store-merging.c
2651 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2652 on a separate gimple_seq which is then appended to seq.
2654 2017-11-06 Jeff Law <law@redhat.com>
2657 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2658 (get_probe_interval): New functions.
2659 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2660 (ix86_adjust_stack_and_probe): Likewise.
2661 (output_adjust_stack_and_probe): Likewise.
2662 (ix86_emit_probe_stack_range): Likewise.
2663 (ix86_expand_prologue): Likewise.
2665 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2667 PR tree-optimization/82816
2668 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2669 if the modes of the two types are the same.
2670 (convert_plusminus_to_widen): Likewise.
2672 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2674 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2676 (usadv16qi): New define_expand.
2677 (usadv8hi): New define_expand.
2679 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2682 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2684 (inline_transform): Likewise.
2685 * predict.c (counts_to_freqs): Remove useless conditional.
2686 * profile-count.h (profile_count::apply_scale): Move sanity check.
2687 * tree-inline.c (copy_bb): Always scale.
2688 (copy_cfg_body): Likewise.
2690 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2693 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2695 (*cmp_ite0): Add enabled_for_depr_it attribute.
2696 (*cmp_ite1): Likewise.
2698 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2700 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2703 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2705 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2706 return the value by pointer.
2707 (extract_range_from_multiplicative_op_1): Update accordingly.
2708 Return as soon as an operation fails.
2710 2017-11-05 Tom de Vries <tom@codesourcery.com>
2713 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2714 (DEF_SANITIZER_BUILTIN): ... here.
2715 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2716 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2718 2017-11-05 Tom de Vries <tom@codesourcery.com>
2721 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2723 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2724 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2725 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2727 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2728 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2729 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2731 2017-11-05 Tom de Vries <tom@codesourcery.com>
2734 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2737 2017-11-04 Michael Clark <michaeljclark@mac.com>
2739 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2740 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2742 (*addsi3_extended): Likewise.
2743 (*addsi3_extended2): Likewise.
2744 (<optab>si3): Likewise.
2745 (<optab>di3): Likewise.
2746 (<optab><mode>3): Likewise.
2747 (<*optabe>si3_internal): Likewise.
2748 (zero_extendqi<SUPERQI:mode>2): Likewise.
2749 (*add<mode>hi3): Likewise.
2750 (*xor<mode>hi3): Likewise.
2751 (<optab>di3): Likewise.
2752 (*<optab>si3_extend): Likewise.
2753 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2754 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2755 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2757 2017-11-04 Andrew Waterman <andrew@sifive.com>
2759 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2760 TARGET_STRICT_ALIGN based upon -mtune argument.
2762 2017-11-04 Andrew Waterman <andrew@sifive.com>
2764 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2766 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2768 config/i386/i386.c (choose_basereg): Use optional scratch
2769 register and add assertion.
2770 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2771 needed, and don't allocate stack.
2772 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2773 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2774 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2776 2017-11-03 Jeff Law <law@redhat.com>
2778 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2779 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2780 to probe at the start of a noreturn function.
2782 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2784 PR tree-optimization/78821
2785 * gimple-ssa-store-merging.c: Update the file comment.
2786 (MAX_STORE_ALIAS_CHECKS): Define.
2787 (struct store_operand_info): New type.
2788 (store_operand_info::store_operand_info): New constructor.
2789 (struct store_immediate_info): Add rhs_code and ops data members.
2790 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2791 and op1r arguments to the ctor, initialize corresponding data members.
2792 (struct merged_store_group): Add load_align_base and load_align
2794 (merged_store_group::merged_store_group): Initialize them.
2795 (merged_store_group::do_merge): Update them.
2796 (merged_store_group::apply_stores): Pick the constant for
2797 encode_tree_to_bitpos from one of the two operands, or skip
2798 encode_tree_to_bitpos if neither operand is a constant.
2799 (class pass_store_merging): Add process_store method decl. Remove
2800 bool argument from terminate_all_aliasing_chains method decl.
2801 (pass_store_merging::terminate_all_aliasing_chains): Remove
2802 var_offset_p argument and corresponding handling.
2803 (stmts_may_clobber_ref_p): New function.
2804 (compatible_load_p): New function.
2805 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2806 if there is overlap and rhs_code is not INTEGER_CST. For
2807 non-overlapping stores terminate group if rhs is not mergeable.
2808 (get_alias_type_for_stmts): Change first argument from
2809 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2810 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2811 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2813 (get_location_for_stmts): Change first argument from
2814 auto_vec<gimple *> & to vec<gimple *> &.
2815 (struct split_store): Remove orig_stmts data member, add orig_stores.
2816 (split_store::split_store): Create orig_stores rather than orig_stmts.
2817 (find_constituent_stmts): Renamed to ...
2818 (find_constituent_stores): ... this. Change second argument from
2819 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2820 to info structures rather than the statements.
2821 (split_group): Rename ALLOW_UNALIGNED argument to
2822 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2823 it. Adjust find_constituent_stores caller.
2824 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2825 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2826 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2827 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2828 (mem_valid_for_store_merging): New function.
2829 (handled_load): New function.
2830 (pass_store_merging::process_store): New method.
2831 (pass_store_merging::execute): Use process_store method. Adjust
2832 terminate_all_aliasing_chains caller.
2834 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2836 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2837 Return true for more constants, symbols and label references.
2838 (aarch64_valid_floating_const): Remove unused function.
2840 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2843 * config/aarch64/aarch64.c (aarch64_layout_frame):
2844 Undo forcing of LR at bottom of frame.
2846 2017-11-03 Jeff Law <law@redhat.com>
2849 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2850 for int_registers_saved.
2852 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2853 extracted from tree-ssa-dom.c.
2854 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2855 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2856 (record_equivalences_from_incoming_edge): Add additional argument
2857 to single_pred_edge_ignoring_loop_edges call.
2858 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2859 (uncprop_dom_walker::before_dom_children): Add additional argument
2860 to single_pred_edge_ignoring_loop_edges call.
2861 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2862 single_pred_edge_ignoring_loop_edges rather than open coding.
2863 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2865 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2867 * match.pd (-(-A)): Rewrite.
2869 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2871 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2872 (rs6000_emit_int_cmove): New declaration.
2873 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2874 (rs6000_emit_sISEL): Delete.
2875 (rs6000_emit_int_cmove): Make non-static.
2876 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2877 instead of rs6000_emit_sISEL.
2879 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2881 * asan.c (create_cond_insert_point): Maintain profile.
2882 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2884 * basic-block.h (struct basic_block_def): Remove frequency.
2885 (EDGE_FREQUENCY): Use to_frequency
2886 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2888 (find_traces): Update to use to_frequency.
2889 (find_traces_1_round): Likewise; use only IPA counts.
2890 (bb_to_key): Likewise.
2891 (connect_traces): Use IPA counts only.
2892 (copy_bb_p): Update to use to_frequency.
2893 (fix_up_crossing_landing_pad): Likewise.
2894 (sanitize_hot_paths): Likewise.
2895 * bt-load.c (basic_block_freq): Likewise.
2896 * cfg.c (init_flow): Set count_max to uninitialized.
2897 (check_bb_profile): Remove frequencies; check counts.
2898 (dump_bb_info): Do not dump frequencies.
2899 (update_bb_profile_for_threading): Update counts only.
2900 (scale_bbs_frequencies_int): Likewise.
2901 (MAX_SAFE_MULTIPLIER): Remove.
2902 (scale_bbs_frequencies_gcov_type): Update counts only.
2903 (scale_bbs_frequencies_profile_count): Update counts only.
2904 (scale_bbs_frequencies): Update counts only.
2905 * cfg.h (struct control_flow_graph): Add count-max.
2906 (update_bb_profile_for_threading): Update prototype.
2907 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2908 (find_many_sub_basic_blocks): Likewise.
2909 * cfgcleanup.c (try_forward_edges): Likewise.
2910 (try_crossjump_to_edge): Likewise.
2911 * cfgexpand.c (expand_gimple_cond): Likewise.
2912 (expand_gimple_tailcall): Likewise.
2913 (construct_init_block): Likewise.
2914 (construct_exit_block): Likewise.
2915 * cfghooks.c (verify_flow_info): Check consistency of counts.
2916 (dump_bb_for_graph): Do not dump frequencies.
2917 (split_block_1): Do not update frequencies.
2918 (split_edge): Do not update frequencies.
2919 (make_forwarder_block): Do not update frequencies.
2920 (duplicate_block): Do not update frequencies.
2921 (account_profile_record): Do not update frequencies.
2922 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2923 for global heuristics.
2924 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2925 (expected_loop_iterations_unbounded): Use counts only.
2926 * cfgloopmanip.c (scale_loop_profile): Simplify.
2927 (create_empty_loop_on_edge): Simplify
2929 (duplicate_loop_to_header_edge): Simplify
2930 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2931 (update_br_prob_note): Take care of removing note when profile
2933 (relink_block_chain): Do not dump frequency.
2934 (rtl_account_profile_record): Use to_frequency.
2935 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2936 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2937 (cgraph_update_edges_for_call_stmt_node): Likewise.
2938 (cgraph_edge::verify_count_and_frequency): Update.
2939 (cgraph_node::verify_node): Temporarily disable frequency verification.
2940 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2941 to_cgraph_frequency.
2942 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2943 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2945 (cgraph_node::expand_thunk): Update profile.
2946 * except.c (dw2_build_landing_pads): Do not update frequency.
2947 * final.c (compute_alignments): Use to_frequency.
2948 (dump_basic_block_info): Do not dump frequency.
2949 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2950 (dump_gimple_bb_header): Do not dump frequency.
2951 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2953 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2954 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2955 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2956 (init_before_recovery): Do not update frequency.
2957 (sched_create_recovery_edges): Do not update frequency.
2958 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2959 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2960 (ipa_cp_c_finalize): Set max_count to uninitialized.
2961 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2962 (param_change_prob): Use counts.
2963 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2965 * ipa-split.c (consider_split): Use to_frequency.
2966 (split_function): Use to_frequency.
2967 * ira-build.c (loop_compare_func): Likewise.
2968 (mark_loops_for_removal): Likewise.
2969 (mark_all_loops_for_removal): Likewise.
2970 * loop-doloop.c (doloop_modify): Do not update frequency.
2971 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2973 * lto-streamer-in.c (input_function): Update count_max.
2974 * omp-expand.c (expand_omp_taskreg): Update count_max.
2975 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2976 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2977 (maybe_hot_count_p): Use ipa counts only.
2978 (maybe_hot_bb_p): Simplify.
2979 (maybe_hot_edge_p): Simplify.
2980 (probably_never_executed): Do not take frequency argument.
2981 (probably_never_executed_bb_p): Do not pass frequency.
2982 (probably_never_executed_edge_p): Likewise.
2983 (combine_predictions_for_bb): Check that profile is nonzero.
2984 (propagate_freq): Do not set frequency.
2985 (drop_profile): Simplify.
2986 (counts_to_freqs): Simplify.
2987 (expensive_function_p): Use to_frequency.
2988 (propagate_unlikely_bbs_forward): Simplify.
2989 (determine_unlikely_bbs): Simplify.
2990 (estimate_bb_frequencies): Add hack to silence graphite issues.
2991 (compute_function_frequency): Use ipa counts.
2992 (pass_profile::execute): Update.
2993 (rebuild_frequencies): Use counts only.
2994 (force_edge_cold): Use counts only.
2995 * profile-count.c (profile_count::dump): Dump new count types.
2996 (profile_count::differs_from_p): Check compatiblity.
2997 (profile_count::to_frequency): New function.
2998 (profile_count::to_cgraph_frequency): New function.
2999 * profile-count.h (struct function): Declare.
3000 (enum profile_quality): Add profile_guessed_local and
3001 profile_guessed_global0.
3002 (class profile_proability): Decrease number of bits to 29;
3003 update from_reg_br_prob_note and to_reg_br_prob_note.
3004 (class profile_count: Update comment; decrease number of bits
3005 to 61. Check compatibility.
3006 (profile_count::compatible_p): New private member function.
3007 (profile_count::ipa_p): New member function.
3008 (profile_count::operator<): Handle global zero correctly.
3009 (profile_count::operator>): Handle global zero correctly.
3010 (profile_count::operator<=): Handle global zero correctly.
3011 (profile_count::operator>=): Handle global zero correctly.
3012 (profile_count::nonzero_p): New member function.
3013 (profile_count::force_nonzero): New member function.
3014 (profile_count::max): New member function.
3015 (profile_count::apply_scale): Handle IPA scalling.
3016 (profile_count::guessed_local): New member function.
3017 (profile_count::global0): New member function.
3018 (profile_count::ipa): New member function.
3019 (profile_count::to_frequency): Declare.
3020 (profile_count::to_cgraph_frequency): Declare.
3021 * profile.c (OVERLAP_BASE): Delete.
3022 (compute_frequency_overlap): Delete.
3023 (compute_branch_probabilities): Do not use compute_frequency_overlap.
3024 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
3025 * sched-ebb.c (rank): Use counts only.
3026 * shrink-wrap.c (handle_simple_exit): Use counts only.
3027 (try_shrink_wrapping): Use counts only.
3028 (place_prologue_for_one_component): Use counts only.
3029 * tracer.c (find_best_predecessor): Use to_frequency.
3030 (find_trace): Use to_frequency.
3031 (tail_duplicate): Use to_frequency.
3032 * trans-mem.c (expand_transaction): Do not update frequency.
3033 * tree-call-cdce.c: Do not update frequency.
3034 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
3035 (gimple_merge_blocks): Likewise.
3036 (gimple_split_edge): Likewise.
3037 (gimple_duplicate_sese_region): Likewise.
3038 (gimple_duplicate_sese_tail): Likewise.
3039 (move_sese_region_to_fn): Likewise.
3040 (gimple_account_profile_record): Likewise.
3041 (insert_cond_bb): Likewise.
3042 * tree-complex.c (expand_complex_div_wide): Likewise.
3043 * tree-eh.c (lower_resx): Update profile.
3044 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
3046 (initialize_cfun): Do not initialize frequencies
3047 (freqs_to_counts): Delete.
3048 (copy_cfg_body): Ignore count parameter.
3049 (copy_body): Update.
3050 (expand_call_inline): Update count_max.
3051 (optimize_inline_calls): Update count_max.
3052 (tree_function_versioning): Update count_max.
3053 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
3054 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
3056 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
3057 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
3058 (try_peel_loop): Likewise.
3059 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
3061 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
3062 (tree_transform_and_unroll_loop): Do not use frequencies
3063 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
3064 Use reliable prediction only.
3065 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
3066 * tree-ssa-sink.c (select_best_block): Use to_frequency.
3067 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
3068 probability scaling.
3069 * tree-ssa-threadupdate.c (create_block_for_threading): Do
3070 not update frequency
3071 (any_remaining_duplicated_blocks): Likewise.
3072 (update_profile): Likewise.
3073 (estimated_freqs_path): Delete.
3074 (freqs_to_counts_path): Delete.
3075 (clear_counts_path): Delete.
3076 (ssa_fix_duplicate_block_edges): Likewise.
3077 (duplicate_thread_path): Likewise.
3078 * tree-switch-conversion.c (gen_inbound_check): Use counts.
3079 * tree-tailcall.c (decrease_profile): Do not update frequency.
3080 (eliminate_tail_call): Likewise.
3081 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
3082 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
3083 (optimize_mask_stores): Likewise.
3084 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
3085 * ubsan.c (ubsan_expand_null_ifn): Update profile.
3086 (ubsan_expand_ptr_ifn): Update profile.
3087 * value-prof.c (gimple_ic): Simplify.
3088 * value-prof.h (gimple_ic): Update prototype.
3089 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
3090 * ipa-inline.c (compute_uninlined_call_time): Be sure that
3092 (want_inline_self_recursive_call_p): Likewise.
3093 (resolve_noninline_speculation): Only cummulate defined counts.
3094 (inline_small_functions): Use nonzero_p.
3095 (ipa_inline): Do not access freed node.
3097 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3099 * config/aarch64/aarch64.c (aarch64_override_options_internal):
3100 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
3102 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
3104 * config/riscv/riscv.c (riscv_legitimize_move): Handle
3105 non-legitimate address.
3107 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3109 * config/rs6000/rs6000.md (*lt0_disi): Delete.
3110 (*lt0_<mode>di, *lt0_<mode>si): New.
3112 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
3114 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
3115 TARGET_PAIRED_FLOAT.
3117 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
3118 Jim Wilson <jim.wilson@linaro.org>
3120 * config/aarch64/aarch64-cores.def (saphira): New CPU.
3121 * config/aarch64/aarch64-tune.md: Regenerated.
3122 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
3123 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
3125 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
3127 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
3128 (arc_expand_prologue): Restore blink for millicode.
3129 * config/arc/linux.h (LINK_EH_SPEC): Defined.
3131 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
3134 * config/i386/i386.c (ix86_vector_duplicate_value): Use
3135 gen_vec_duplicate after forcing the scalar into a register.
3137 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
3139 * combine (try_combine): Print the insns input to try_combine to the
3142 2017-11-02 Steve Ellcey <sellcey@cavium.com>
3145 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
3146 Remove second argument from aarch64_process_target_attr call.
3147 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
3149 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
3151 (aarch64_handle_attr_arch): Remove second argument.
3152 (aarch64_handle_attr_cpu): Ditto.
3153 (aarch64_handle_attr_tune): Ditto.
3154 (aarch64_handle_attr_isa_flags): Ditto.
3155 (aarch64_process_one_target_attr): Ditto.
3156 (aarch64_process_target_attr): Ditto.
3157 (aarch64_option_valid_attribute_p): Remove second argument.
3158 on aarch64_process_target_attr call.
3160 2017-11-02 David Malcolm <dmalcolm@redhat.com>
3162 * diagnostic.c: Include "selftest-diagnostic.h".
3163 (selftest::assert_location_text): New function.
3164 (selftest::test_diagnostic_get_location_text): New function.
3165 (selftest::diagnostic_c_tests): Call it.
3167 2017-11-02 David Malcolm <dmalcolm@redhat.com>
3169 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
3170 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
3171 (class selftest::test_diagnostic_context): Move to...
3172 * selftest-diagnostic.c: New file.
3173 * selftest-diagnostic.h: New file.
3175 2017-11-02 James Bowman <james.bowman@ftdichip.com>
3177 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
3178 offset range for FT32B.
3179 * config/ft32/ft32.h: option "mcompress" enables relaxation.
3180 * config/ft32/ft32.md: Add TARGET_NOPM.
3181 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
3182 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
3184 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
3186 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
3188 2017-11-02 Jeff Law <law@redhat.com>
3190 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
3191 virtual keyword on FINAL OVERRIDE members.
3193 * tree-ssa-propagate.h (ssa_propagation_engine): Group
3194 virtuals together. Add virtual destructor.
3195 (substitute_and_fold_engine): Similarly.
3197 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
3199 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
3201 2017-11-02 Richard Biener <rguenther@suse.de>
3203 PR tree-optimization/82795
3204 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
3206 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3208 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
3210 (gcc_GAS_CHECK_FEATURE): Remove.
3211 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
3212 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
3213 * configure: Regenerate.
3215 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
3217 * config/arc/arc.c (hwloop_optimize): Account for empty
3220 2017-11-02 Richard Biener <rguenther@suse.de>
3223 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
3224 Truncate ARRAY_REF index and element size.
3226 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
3228 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
3230 2017-11-01 Jeff Law <law@redhat.com>
3232 * tree-ssa-ccp.c (ccp_folder): New class derived from
3233 substitute_and_fold_engine.
3234 (ccp_folder::get_value): New member function.
3235 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
3236 (ccp_fold_stmt): Remove prototype.
3237 (ccp_finalize): Call substitute_and_fold from the ccp_class.
3238 * tree-ssa-copy.c (copy_folder): New class derived from
3239 substitute_and_fold_engine.
3240 (copy_folder::get_value): Renamed from get_value.
3241 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
3242 * tree-vrp.c (vrp_folder): New class derived from
3243 substitute_and_fold_engine.
3244 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
3245 (vrp_folder::get_value): New member function.
3246 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
3247 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
3248 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
3249 provide a class interface to folder/substitute routines.
3250 (ssa_prop_fold_stmt_fn): Remove typedef.
3251 (ssa_prop_get_value_fn): Likewise.
3252 (subsitute_and_fold): Remove prototype.
3253 (replace_uses_in): Likewise.
3254 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
3255 Renamed from replace_uses_in. Call the virtual member function
3256 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
3257 (substitute_and_fold_dom_walker): Remove initialization of
3258 data member entries for calbacks. Add substitute_and_fold_engine
3259 member and initialize it.
3260 (substitute_and_fold_dom_walker::before_dom_children0: Use the
3261 member functions for get_value, replace_phi_args_in c
3262 replace_uses_in, and fold_stmt calls.
3263 (substitute_and_fold_engine::substitute_and_fold): Renamed from
3264 substitute_and_fold. Remove assert. Update ctor call.
3266 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
3267 (ssa_prop_visit_phi_fn): Likewise.
3268 (class ssa_propagation_engine): New class to provide an interface
3270 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
3272 (ssa_prop_visit_phi): Likewise.
3273 (ssa_propagation_engine::simulate_stmt): Moved into class.
3274 Call visit_phi/visit_stmt from the class rather than via
3275 file scoped static variables.
3276 (ssa_propagation_engine::simulate_block): Moved into class.
3277 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
3278 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
3279 set file scoped statics for the visit_stmt/visit_phi callbacks.
3280 * tree-complex.c (complex_propagate): New class derived from
3281 ssa_propagation_engine.
3282 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
3283 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
3284 (tree_lower_complex): Call ssa_propagate via the complex_propagate
3286 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
3287 ssa_propagation_engine.
3288 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
3289 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
3290 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
3291 * tree-ssa-copy.c (copy_prop): New class derived from
3292 ssa_propagation_engine.
3293 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
3294 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
3295 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
3296 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
3297 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
3298 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
3299 (execute_vrp): Call ssa_propagate from the vrp_prop class.
3301 2017-11-01 Jakub Jelinek <jakub@redhat.com>
3303 PR rtl-optimization/82778
3304 PR rtl-optimization/82597
3305 * compare-elim.c (struct comparison): Add in_a_setter field.
3306 (find_comparison_dom_walker::before_dom_children): Remove killed
3307 bitmap and df_simulate_find_defs call, instead walk the defs.
3308 Compute last_setter and initialize in_a_setter. Merge definitions
3309 with first initialization for a few variables.
3310 (try_validate_parallel): Use insn_invalid_p instead of
3311 recog_memoized. Return insn rather than just the pattern.
3312 (try_merge_compare): Fix up comment. Don't uselessly test if
3313 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
3315 (execute_compare_elim_after_reload): Remove df_chain_add_problem
3318 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3319 Alan Hayward <alan.hayward@arm.com>
3320 David Sherwood <david.sherwood@arm.com>
3322 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
3323 aarch64_hard_regno_nregs to get the number of registers
3326 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3327 Alan Hayward <alan.hayward@arm.com>
3328 David Sherwood <david.sherwood@arm.com>
3330 * config/aarch64/constraints.md (Upl): Rename to...
3332 * config/aarch64/aarch64.md
3333 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
3336 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3337 Alan Hayward <alan.hayward@arm.com>
3338 David Sherwood <david.sherwood@arm.com>
3340 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
3341 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
3344 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3345 Alan Hayward <alan.hayward@arm.com>
3346 David Sherwood <david.sherwood@arm.com>
3348 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
3349 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
3350 (aarch64_evpc_dup): Generate rtl direcly, rather than using
3352 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
3353 of a single element.
3354 * config/aarch64/iterators.md: Add a comment above the permute
3355 unspecs to say that they are generated directly by
3356 aarch64_expand_vec_perm_const.
3357 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
3359 2017-11-01 Nathan Sidwell <nathan@acm.org>
3361 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
3363 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
3365 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
3366 and medany code models, and describe what they do.
3368 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3370 Revert accidental duplicate:
3372 * combine.c (can_change_dest_mode): Reject changes in
3373 REGMODE_NATURAL_SIZE.
3375 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
3377 PR rtl-optimization/64682
3378 PR rtl-optimization/69567
3379 PR rtl-optimization/69737
3380 PR rtl-optimization/82683
3381 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
3382 register mentioned in the note, drop the note, unless it came from I3,
3383 in which case it should go to I3 again.
3385 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3387 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
3388 and return false if not.
3389 (clear_bytes_written_by, live_bytes_read): Update accordingly.
3391 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3393 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
3394 range is known to be empty.
3396 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3397 Alan Hayward <alan.hayward@arm.com>
3398 David Sherwood <david.sherwood@arm.com>
3400 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
3401 and CONST_VECTOR_NUNITS instead of computing the number of units from
3402 the byte sizes of the vector and element.
3403 (simplify_binary_operation_1): Likewise.
3404 (simplify_const_binary_operation): Likewise.
3405 (simplify_ternary_operation): Likewise.
3407 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3408 Alan Hayward <alan.hayward@arm.com>
3409 David Sherwood <david.sherwood@arm.com>
3411 * var-tracking.c (INT_MEM_OFFSET): Replace with...
3412 (int_mem_offset): ...this new function.
3413 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
3414 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
3415 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
3418 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3419 Alan Hayward <alan.hayward@arm.com>
3420 David Sherwood <david.sherwood@arm.com>
3422 * lower-subreg.c (interesting_mode_p): New function.
3423 (compute_costs, find_decomposable_subregs, decompose_register)
3424 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
3425 (resolve_clobber, dump_choices): Use it.
3427 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3428 Alan Hayward <alan.hayward@arm.com>
3429 David Sherwood <david.sherwood@arm.com>
3431 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
3433 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3434 Alan Hayward <alan.hayward@arm.com>
3435 David Sherwood <david.sherwood@arm.com>
3437 * alias.c (find_base_value, find_base_term): Only process integer
3438 truncations. Check the precision rather than the size.
3440 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3441 Alan Hayward <alan.hayward@arm.com>
3442 David Sherwood <david.sherwood@arm.com>
3444 * machmode.h (is_narrower_int_mode): New function
3445 * optabs.c (expand_float, expand_fix): Use it.
3446 * dwarf2out.c (rotate_loc_descriptor): Likewise.
3448 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3449 Alan Hayward <alan.hayward@arm.com>
3450 David Sherwood <david.sherwood@arm.com>
3452 * rtl.h (narrower_subreg_mode): New function.
3453 * ira-color.c (update_costs_from_allocno): Use it.
3455 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3456 Alan Hayward <alan.hayward@arm.com>
3457 David Sherwood <david.sherwood@arm.com>
3459 * optabs-query.h (convert_optab_p): New function, split out from...
3460 (convert_optab_handler): ...here.
3461 (widening_optab_handler): Delete.
3462 (find_widening_optab_handler): Remove permit_non_widening parameter.
3463 (find_widening_optab_handler_and_mode): Likewise. Provide an
3464 override that operates on mode class wrappers.
3465 * optabs-query.c (widening_optab_handler): Delete.
3466 (find_widening_optab_handler_and_mode): Remove permit_non_widening
3467 parameter. Assert that the two modes are the same class and that
3468 the "from" mode is narrower than the "to" mode. Use
3469 convert_optab_handler instead of widening_optab_handler.
3470 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
3471 instead of widening_optab_handler.
3472 * expr.c (expand_expr_real_2): Update calls to
3473 find_widening_optab_handler.
3474 * optabs.c (expand_widen_pattern_expr): Likewise.
3475 (expand_binop_directly): Take the insn_code as a parameter.
3476 (expand_binop): Only call find_widening_optab_handler for
3477 conversion optabs; use optab_handler otherwise. Update calls
3478 to find_widening_optab_handler and expand_binop_directly.
3479 Use convert_optab_handler instead of widening_optab_handler.
3480 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
3481 find_widening_optab_handler and use scalar_mode rather than
3483 (convert_plusminus_to_widen): Likewise.
3485 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3486 Alan Hayward <alan.hayward@arm.com>
3487 David Sherwood <david.sherwood@arm.com>
3489 * machmode.h (fixed_size_mode): New class.
3490 * rtl.h (get_pool_mode): Return fixed_size_mode.
3491 * gengtype.c (main): Add fixed_size_mode.
3492 * target.def (get_raw_result_mode): Return a fixed_size_mode.
3493 (get_raw_arg_mode): Likewise.
3494 * doc/tm.texi: Regenerate.
3495 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
3496 * targhooks.c (default_get_reg_raw_mode): Likewise.
3497 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
3498 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
3499 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
3500 (msp430_get_raw_result_mode): Likewise.
3501 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
3502 * dbxout.c (dbxout_parms): Require fixed-size modes.
3503 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
3504 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3505 * omp-low.c (lower_oacc_reductions): Likewise.
3506 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
3507 (simplify_subreg): Update accordingly.
3508 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
3509 (force_const_mem): Update accordingly. Return NULL_RTX for modes
3510 that aren't fixed-size.
3511 (get_pool_mode): Return a fixed_size_mode.
3512 (output_constant_pool_2): Take a fixed_size_mode.
3514 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3515 Alan Hayward <alan.hayward@arm.com>
3516 David Sherwood <david.sherwood@arm.com>
3518 * doc/rtl.texi (vec_series): Document.
3519 (const): Say that the operand can be a vec_series.
3520 * rtl.def (VEC_SERIES): New rtx code.
3521 * rtl.h (const_vec_series_p_1): Declare.
3522 (const_vec_series_p): New function.
3523 * emit-rtl.h (gen_const_vec_series): Declare.
3524 (gen_vec_series): Likewise.
3525 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
3526 (gen_vec_series): Likewise.
3527 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
3528 * simplify-rtx.c (simplify_unary_operation): Handle negations
3530 (simplify_binary_operation_series): New function.
3531 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
3532 (test_vector_ops_series): New function.
3533 (test_vector_ops): Call it.
3534 * config/powerpcspe/altivec.md (altivec_lvsl): Use
3535 gen_const_vec_series.
3536 (altivec_lvsr): Likewise.
3537 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
3539 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3540 Alan Hayward <alan.hayward@arm.com>
3541 David Sherwood <david.sherwood@arm.com>
3543 * doc/rtl.texi (const): Update description of address constants.
3544 Say that vector constants are allowed too.
3545 * common.md (E, F): Use CONSTANT_P instead of checking for
3547 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
3548 checking for CONST_VECTOR.
3549 * expmed.c (make_tree): Use build_vector_from_val for a CONST
3551 * expr.c (expand_expr_real_2): Check for vector modes instead
3552 of checking for CONST_VECTOR.
3553 * rtl.h (const_vec_p): New function.
3554 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
3555 (unwrap_const_vec_duplicate): Handle them here too.
3557 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3558 David Malcolm <dmalcolm@redhat.com>
3559 Alan Hayward <alan.hayward@arm.com>
3560 David Sherwood <david.sherwood@arm.com>
3562 * rtl.h (vec_duplicate_p): New function.
3563 * selftest-rtl.c (assert_rtx_eq_at): New function.
3564 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
3565 (assert_rtx_eq_at): Declare.
3566 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
3567 * selftest-run-tests.c (selftest::run_tests): Call it.
3568 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
3569 (simplify_unary_operation_1): Recursively handle vector duplicates.
3570 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
3572 (simplify_subreg): Handle subregs of vector duplicates.
3573 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
3574 (selftest::simplify_rtx_c_tests): New functions.
3576 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3577 Alan Hayward <alan.hayward@arm.com>
3578 David Sherwood <david.sherwood@arm.com>
3580 * emit-rtl.h (gen_const_vec_duplicate): Declare.
3581 (gen_vec_duplicate): Likewise.
3582 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
3584 (gen_const_vector): ...here.
3585 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
3586 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
3587 whose elements are all equal.
3588 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
3589 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
3590 (simplify_relational_operation): Likewise.
3591 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3593 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
3594 (aarch64_expand_vector_init): Likewise.
3595 * config/arm/arm.c (neon_vdup_constant): Likewise.
3596 (neon_expand_vector_init): Likewise.
3597 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
3598 (arm_block_set_unaligned_vect): Likewise.
3599 (arm_block_set_aligned_vect): Likewise.
3600 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
3601 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
3602 (expand_vec_perm_even_odd_pack): Likewise.
3603 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
3604 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
3605 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
3606 gen_const_vec_duplicate.
3607 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
3608 * config/mips/mips.c (mips_gen_const_int_vector): Use
3609 gen_const_vec_duplicate.
3610 (mips_expand_vector_init): Use CONST0_RTX.
3611 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
3612 (define_split): Use gen_const_vec_duplicate.
3613 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
3614 (define_split): Use gen_const_vec_duplicate.
3615 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
3616 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
3617 * config/spu/spu.c (spu_const): Likewise.
3619 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3620 Alan Hayward <alan.hayward@arm.com>
3621 David Sherwood <david.sherwood@arm.com>
3623 * combine.c (can_change_dest_mode): Reject changes in
3624 REGMODE_NATURAL_SIZE.
3626 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3628 * configure.ac (--enable-libssp): New.
3629 (gcc_cv_libc_provides_ssp): Check for explicit setting before
3630 trying to determine target-specific default. Adjust indentation.
3631 * configure: Regenerated.
3632 * doc/install.texi (Configuration): Expand --disable-libssp
3635 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
3637 config/i386/i386.c (ix86_expand_epilogue): Correct stack
3640 2017-10-31 Martin Jambor <mjambor@suse.cz>
3643 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
3645 2017-10-31 David Malcolm <dmalcolm@redhat.com>
3647 * auto-profile.c (autofdo_source_profile::read): Use
3648 UNKNOWN_LOCATION rather than 0.
3649 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3650 (warning_at): ...this overload.
3651 (warning_at_rich_loc_n): Rename to...
3652 (warning_n): ...this overload.
3653 (error_at_rich_loc): Rename to...
3654 (error_at): ...this overload.
3655 (pedwarn_at_rich_loc): Rename to...
3656 (pedwarn): ...this overload.
3657 (permerror_at_rich_loc): Rename to...
3658 (permerror): ...this overload.
3659 (inform_at_rich_loc): Rename to...
3660 (inform): ...this overload.
3661 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3662 (diagnostic_n_impl_richloc): Rename to...
3663 (diagnostic_n_impl): ...this rich_location *-based decl.
3664 (inform_at_rich_loc): Rename to...
3665 (inform): ...this, and add an assertion.
3666 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3667 (warning_at_rich_loc): Rename to...
3668 (warning_at): ...this, and add an assertion.
3669 (warning_at_rich_loc_n): Rename to...
3670 (warning_n): ...this, and add an assertion.
3671 (warning_n): Update location_t-based implementation for removal of
3672 location_t-based diagnostic_n_impl.
3673 (pedwarn_at_rich_loc): Rename to...
3674 (pedwarn): ...this, and add an assertion.
3675 (permerror_at_rich_loc): Rename to...
3676 (permerror): ...this, and add an assertion.
3677 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3678 (error_at_rich_loc): Rename to...
3679 (error_at): ...this, and add an assertion.
3680 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3681 (driver::do_spec_on_infiles): Likewise.
3682 * substring-locations.c (format_warning_va): Update for renaming
3683 of inform_at_rich_loc.
3685 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3687 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3688 _Float<N>X built-in functions so that the variant without the
3689 "__builtin_" prefix is only enabled for the GNU C and Objective C
3690 languages when they are in non-strict ANSI/ISO mode.
3691 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3692 * target.def (floatn_builtin_p): Add a target hook to control
3693 whether _Float<N> and _Float<N>X built-in functions without the
3694 "__builtin_" prefix are enabled, and return true for C and
3695 Objective C in the default hook. Include langhooks.h in
3697 * targhooks.h (default_floatn_builtin_p): Likewise.
3698 * targhooks.c (default_floatn_builtin_p): Likewise.
3699 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3700 floatn_builtin_p target hook.
3701 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3703 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3704 Eric Botcazou <ebotcazou@adacore.com>
3706 PR rtl-optimization/81803
3707 * lra-constraints.c (curr_insn_transform): Also reload the whole
3708 register for a strict subreg no wider than a word if this is for
3709 a WORD_REGISTER_OPERATIONS target.
3711 2017-10-31 Jason Merrill <jason@redhat.com>
3713 * gdbinit.in: Skip over inlines from timevar.h.
3715 2017-10-31 Martin Liska <mliska@suse.cz>
3717 * doc/gcov.texi: Document new option.
3718 * gcov.c (print_usage): Likewise print it.
3719 (process_args): Support the argument.
3720 (format_count): New function.
3721 (format_gcov): Use the function.
3723 2017-10-31 Martin Liska <mliska@suse.cz>
3725 * gcov.c (struct name_map): do not use typedef.
3726 Define operator== and operator<.
3727 (name_search): Remove.
3728 (name_sort): Remove.
3729 (main): Do not allocate names.
3730 (process_file): Add vertical space.
3731 (generate_results): Use std::find.
3732 (release_structures): Do not release memory.
3733 (find_source): Use std::find.
3735 2017-10-31 Martin Liska <mliska@suse.cz>
3737 * gcov.c (struct line_info): Remove it's typedef.
3738 (line_info::line_info): Add proper ctor.
3739 (line_info::has_block): Do not use a typedef.
3740 (struct source_info): Do not use typedef.
3741 (circuit): Likewise.
3742 (get_cycles_count): Likewise.
3743 (output_intermediate_file): Iterate via vector iterator.
3744 (add_line_counts): Use std::vector methods.
3745 (accumulate_line_counts): Likewise.
3746 (output_lines): Likewise.
3748 2017-10-31 Martin Liska <mliska@suse.cz>
3750 * gcov.c (struct source_info): Remove typedef.
3751 (source_info::source_info): Add proper ctor.
3752 (accumulate_line_counts): Use struct, not it's typedef.
3753 (output_gcov_file): Likewise.
3754 (output_lines): Likewise.
3755 (main): Do not allocate an array.
3756 (output_intermediate_file): Use size of vector container.
3757 (process_file): Resize the vector.
3758 (generate_results): Do not preallocate, use newly added vector
3760 (release_structures): Do not release sources.
3761 (find_source): Use vector methods.
3762 (add_line_counts): Do not use typedef.
3764 2017-10-31 Martin Liska <mliska@suse.cz>
3766 * doc/gcov.texi: Document that.
3767 * gcov.c (add_line_counts): Mark lines with a non-executed
3769 (output_line_beginning): Handle such lines.
3770 (output_lines): Pass new argument.
3771 (output_intermediate_file): Print it in intermediate format.
3773 2017-10-31 Martin Liska <mliska@suse.cz>
3775 * color-macros.h: New file.
3776 * diagnostic-color.c: Factor out color related to macros to
3778 * doc/gcov.texi: Document -k option.
3779 * gcov.c (INCLUDE_STRING): Include string.h.
3780 (print_usage): Add -k option.
3781 (process_args): Parse it.
3782 (pad_count_string): New function.
3783 (output_line_beginning): Likewise.
3784 (DEFAULT_LINE_START): New macro.
3785 (output_lines): Support color output.
3787 2017-10-31 Martin Liska <mliska@suse.cz>
3789 PR gcov-profile/82633
3790 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3791 their interaction with GCOV infrastructure.
3792 * configure.ac: Add -fkeep-{inline,static}-functions to
3794 * configure: Regenerate.
3796 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3799 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3801 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3804 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3805 into a register if it does not fit into an immediate offset field.
3807 2017-10-31 Olivier Hainque <hainque@adacore.com>
3809 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3811 2017-10-31 Julia Koval <julia.koval@intel.com>
3813 * config.gcc: Add gfniintrin.h.
3814 * config/i386/gfniintrin.h: New.
3815 * config/i386/i386-builtin-types.def
3816 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3817 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3818 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3819 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3820 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3821 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3822 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3823 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3824 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3825 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3826 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3827 * config/i386/immintrin.h: Include gfniintrin.h.
3828 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3830 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3832 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3834 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3836 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3837 (arm_ashldi3_1bit): Remove pattern.
3838 (ashrdi3): Remove shift by 1 expansion.
3839 (arm_ashrdi3_1bit): Remove pattern.
3840 (lshrdi3): Remove shift by 1 expansion.
3841 (arm_lshrdi3_1bit): Remove pattern.
3842 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3843 cost of ashldi3 by 1.
3844 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3845 (<shift>di3_neon): Likewise.
3847 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3849 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3850 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3851 and (*aarch64_simd_mov<VQ:mode>).
3852 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3853 pattern alternative.
3854 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3855 attributes to match pattern alternative.
3857 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3859 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3860 * config/rs6000/emmintrin.h: New file.
3861 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3863 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3865 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3866 * (movdi_vfp_cortexa8): Remove pattern.
3868 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3870 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3872 (Specific, alpha*-dec-osf5.1): Remove.
3873 (Specific, mips-sgi-irix5): Remove.
3874 (Specific, mips-sgi-irix6): Remove.
3876 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3879 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3880 arguments to clear_bit_region_be.
3882 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3884 * gimplify.c: Include memmodel.h.
3886 2017-10-30 Martin Jambor <mjambor@suse.cz>
3888 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3889 condition whether loop should be executed at all.
3891 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3893 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3894 gimple folding of vec_madd() intrinsics.
3895 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3896 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3897 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3899 2017-10-30 Richard Biener <rguenther@suse.de>
3901 PR tree-optimization/82762
3903 2017-10-23 Richard Biener <rguenther@suse.de>
3905 PR tree-optimization/82129
3907 2017-08-01 Richard Biener <rguenther@suse.de>
3909 PR tree-optimization/81181
3910 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3911 (compute_antic): ... end of iteration here.
3913 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3915 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3916 -std=iso9899:2017 and -std=gnu17.
3917 * doc/standards.texi (C Language): Document C17 support.
3918 * doc/cpp.texi (Overview): Mention -std=c17.
3919 (Standard Predefined Macros): Document C11 and C17 values of
3920 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3921 * doc/extend.texi (Inline): Do not list individual options for
3922 standards newer than C99.
3923 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3925 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3928 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3930 * asan.c (asan_finish_file): Align asan globals array by shadow
3933 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3936 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3937 (struct store_immediate_info): Add bitregion_start and bitregion_end
3939 (store_immediate_info::store_immediate_info): Add brs and bre
3940 arguments and initialize bitregion_{start,end} from those.
3941 (struct merged_store_group): Add bitregion_start, bitregion_end,
3942 align_base and mask fields. Drop unnecessary struct keyword from
3943 struct store_immediate_info. Add do_merge method.
3944 (clear_bit_region_be): Use memset instead of loop storing zeros.
3945 (merged_store_group::do_merge): New method.
3946 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3947 stores as long as the surrounding bitregions have no gaps.
3948 (merged_store_group::merge_overlapping): Use do_merge.
3949 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3950 is byte aligned, rather than requiring that start and width are
3951 byte aligned. Drop unnecessary struct keyword from
3952 struct store_immediate_info. Allocate and populate also mask array.
3953 Make start of the arrays relative to bitregion_start rather than
3954 start and size them according to bitregion_{end,start} difference.
3955 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3956 struct store_immediate_info.
3957 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3958 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3959 struct keyword from struct store_immediate_info.
3960 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3961 between stores as long as the surrounding bitregions have no gaps.
3963 (struct split_store): Add orig non-static data member.
3964 (split_store::split_store): Initialize orig to false.
3965 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3966 if there is exactly a single original stmt. Change stmts argument
3967 to pointer from reference, if NULL, don't push anything to it. Add
3968 first argument, use it to optimize skipping over orig stmts that
3969 are known to be before bitpos already. Simplify.
3970 (split_group): Return unsigned int count how many stores are or
3971 would be needed rather than a bool. Add allow_unaligned argument.
3972 Change split_stores argument from reference to pointer, if NULL,
3973 only do a dry run computing how many stores would be produced.
3974 Rewritten algorithm to use both alignment and misalign if
3975 !allow_unaligned and handle bitfield stores with gaps.
3976 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3977 from bitregion_start instead of start. Compute allow_unaligned
3978 here, if true, do 2 split_group dry runs to compute which one
3979 produces fewer stores and prefer aligned if equal. Punt if
3980 new count is bigger or equal than original before emitting any
3981 statements, rather than during that. Remove no longer needed
3982 new_ssa_names tracking. Replace num_stmts with
3983 split_stores.length (). Use 32-bit stack allocated entries
3984 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3985 if possible. Handle bitfields with gaps.
3986 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3987 Compute bitregion_{start,end} for the stores and construct
3988 store_immediate_info with that. Formatting fixes.
3990 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3993 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3994 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3996 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3998 * gimplify.c: Include tm_p.h.
4000 * common.opt (gcoff): Re-add as ignored option.
4001 (gcoff1, gcoff2, gcoff3): Likewise.
4003 * Makefile.in (OBJS): Delete sdbout.o.
4004 (GTFILES): Delete $(srcdir)/sdbout.c.
4005 * debug.h: Delete sdb_debug_hooks.
4006 * final.c: Delete sdbout.h include.
4007 (final_scan_insn): Delete SDB_DEBUG check.
4008 (rest_of_clean_state): Likewise.
4009 * output.h: Delete sdb_begin_function_line.
4012 * toplev.c: Delete sdbout.h include.
4013 (process_options): Delete SDB_DEBUG check.
4014 * tree-core.h (tree_type_common): Delete pointer field of
4016 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
4017 TYPE_SYMTAB_POINTER.
4018 * tree.h (TYPE_SYMTAB_POINTER): Delete.
4019 (TYPE_SYMTAB_IS_POINTER): Delete.
4020 (TYPE_SYMTAB_IS_DIE): Renumber.
4021 * xcoffout.c: Refer to former sdbout.c file.
4022 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
4024 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
4025 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
4026 (Debugging Options): Delete -gcoff.
4027 (-gstabs): Delete SDB reference.
4029 (-gcoff@var{level}): Delete.
4030 * doc/passes.texi (Debugging information output): Delete SDB and
4031 sdbout.c references.
4032 * doc/tm.texi: Regenerate.
4033 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
4034 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
4036 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
4037 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
4038 (SDB_DEBUGGING_INFO): Delete.
4039 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
4040 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
4041 * target.def (output_source_filename): Delete COFF reference.
4043 * common.opt (gcoff): Delete.
4044 (gxcoff+): Update Negative chain.
4045 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
4047 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
4048 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
4049 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
4051 (expand_function_start): Change sdb reference to past tense.
4052 (expand_function_end): Change sdb reference to past tense.
4053 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
4054 * opts.c (debug_type_names): Delete coff entry.
4055 (common_handle_option): Delete OPT_gcoff case.
4056 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
4058 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
4059 * config/cris/cris.h: Delete SDB reference in comment.
4060 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
4061 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
4062 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
4063 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
4065 (ix86_expand_prologue): Likewise.
4066 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
4067 * config/ia64/ia64.h: Likewise.
4068 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
4069 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
4071 * config/mmix/mmix.h: Likewise.
4072 * config/nds32/nds32.c: Likewise.
4073 * config/stormy/storym16.h: Likewise.
4074 * config/visium/visium.h: Likewise.
4075 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
4077 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
4079 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
4080 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
4081 FRAME_POINTER_REGNUM point at high end of local var area.
4083 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
4085 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
4086 Move comment around. Do not reset best_edge for a copiable
4087 destination if the copy would cause a partition change.
4088 (better_edge_p): Remove redundant check.
4090 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
4092 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
4094 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4096 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
4097 for math functions that have _Float<N> and _Float<N>X variants.
4098 (mathfn_built_in_2): Add support for math functions that have
4099 _Float<N> and _Float<N>X variants.
4100 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
4101 (expand_builtin_mathfn_ternary): Add support for fma with
4102 _Float<N> and _Float<N>X variants.
4103 (expand_builtin): Likewise.
4104 (fold_builtin_3): Likewise.
4105 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
4106 create math function _Float<N> and _Float<N>X variants as external
4108 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
4109 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
4110 the __builtin_ prefix and if not strict ansi, without the prefix.
4111 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
4112 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
4113 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
4114 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
4115 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
4116 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
4117 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
4118 function signatures for fma _Float<N> and _Float<N>X variants.
4119 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
4120 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
4121 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
4122 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
4123 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
4124 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
4125 * gencfn-macros.c (print_case_cfn): Add support for math functions
4126 that have _Float<N> and _Float<N>X variants.
4127 (print_define_operator_list): Likewise.
4128 (fltfn_suffixes): Likewise.
4130 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
4131 for math functions that have _Float<N> and _Float<N>X variants.
4132 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
4133 and _Float<N>X variants.
4134 (COPYSIGN): Likewise.
4137 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
4138 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
4140 (integer_valued_read_call_p): Likewise.
4141 * fold-const-call.c (fold_const_call_ss): Likewise.
4142 (fold_const_call_sss): Add support for copysign, fmin, and fmax
4143 _Float<N> and _Float<N>X variants.
4144 (fold_const_call_ssss): Add support for fma _Float<N> and
4145 _Float<N>X variants.
4146 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
4147 support for copysign and fma _Float<N> and _Float<N>X variants.
4148 (backprop::process_builtin_call_use): Likewise.
4149 * tree-call-cdce.c (can_test_argument_range); Add support for
4150 sqrt _Float<N> and _Float<N>X variants.
4151 (edom_only_function): Likewise.
4152 (get_no_error_domain): Likewise.
4153 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
4154 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
4155 copysign _Float<N> and _Float<N>X variants.
4156 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
4157 handled by machine independent code.
4158 (FMAF128): Likewise.
4159 * doc/cpp.texi (Common Predefined Macros): Document defining
4160 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
4161 fma _Float<N> and _Float<N>X variants.
4163 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
4166 * config/i386/i386-modes.def (CCFPU): Remove definition.
4167 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
4168 (ix86_cc_modes_compatible): Ditto.
4169 (ix86_expand_carry_flag_compare): Ditto.
4170 (ix86_expand_int_movcc): Ditto.
4171 (ix86_expand_int_addcc): Ditto.
4172 (ix86_reverse_condition): Ditto.
4173 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
4174 Return true/false for unordered/ordered fp comparisons.
4175 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
4176 (ix86_prepare_fp_compare_args): Update for rename.
4177 (ix86_expand_fp_compare): Update for rename. Generate unordered
4178 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
4179 (ix86_expand_sse_compare_and_jump): Ditto.
4180 * config/i386/predicates.md (fcmov_comparison_operator):
4181 Remove CCFPU mode handling.
4182 (ix86_comparison_operator): Ditto.
4183 (ix86_carry_flag_operator): Ditto.
4184 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
4185 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
4186 (*cmpu<mode>_cc_i387): Ditto.
4187 (FPCMP): Remove mode iterator.
4188 (unord): Remove mode attribute.
4189 (unord_subst): New define_subst transformation
4190 (unord): New define_subst attribute.
4192 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
4193 (*cmpi<unord>xf_i387): Ditto.
4194 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
4195 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
4196 using unord_subst transformation.
4197 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
4198 (round_saeonly): Also handle CCFP mode.
4199 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
4200 Remove UNSPEC_SAHF unspec handling.
4202 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
4204 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
4206 2017-10-27 Jeff Law <law@redhat.com>
4208 * gimple-ssa-sprintf.c: Include domwalk.h.
4209 (class sprintf_dom_walker): New class, derived from dom_walker.
4210 (sprintf_dom_walker::before_dom_children): New function.
4211 (struct call_info): Moved into sprintf_dom_walker class
4212 (compute_formath_length, handle_gimple_call): Likewise.
4213 (sprintf_length::execute): Call the dominator walker rather
4214 than walking the statements.
4216 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
4217 gimple statement locations.
4218 (check_array_bounds): Corresponding changes. Get the statement's
4219 location directly from wi->stmt.
4221 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
4224 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
4226 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
4228 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
4229 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
4231 2017-10-27 Jakub Jelinek <jakub@redhat.com>
4234 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
4235 * config/i386/i386.c (maybe_get_pool_constant): Removed.
4236 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
4237 maybe_get_pool_constant.
4238 * config/i386/predicates.md (zero_extended_scalar_load_operand):
4241 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4243 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
4244 2.26 caveat. Update gas and gld versions.
4245 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
4248 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4250 * cgraph.h (set_malloc_flag): Declare.
4251 * cgraph.c (set_malloc_flag_1): New function.
4252 (set_malloc_flag): Likewise.
4253 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
4254 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
4256 (read_ipa_call_summary): Add support for reading is_return_callee.
4257 (write_ipa_call_summary): Stream is_return_callee.
4258 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
4259 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
4260 ipa-prop.h, ipa-fnsummary.h.
4261 (pure_const_names): Change to static.
4262 (malloc_state_e): Define.
4263 (malloc_state_names): Define.
4264 (funct_state_d): Add field malloc_state.
4265 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
4266 (check_retval_uses): New function.
4267 (malloc_candidate_p): Likewise.
4268 (analyze_function): Add support for malloc attribute.
4269 (pure_const_write_summary): Stream malloc_state.
4270 (pure_const_read_summary): Add support for reading malloc_state.
4271 (dump_malloc_lattice): New function.
4272 (propagate_malloc): New function.
4273 (warn_function_malloc): New function.
4274 (ipa_pure_const::execute): Call propagate_malloc and
4275 ipa_free_fn_summary.
4276 (pass_local_pure_const::execute): Add support for malloc attribute.
4277 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
4278 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
4280 2017-10-27 Martin Liska <mliska@suse.cz>
4282 PR gcov-profile/82457
4283 * doc/invoke.texi: Document that one needs a non-strict ISO mode
4284 for fork-like functions to be properly instrumented.
4286 2017-10-27 Richard Biener <rguenther@suse.de>
4289 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
4290 info when we redirected EH.
4292 2017-10-26 Michael Collison <michael.collison@arm.com>
4294 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
4296 (<optab>_trunchf<GPI:mode>2: New pattern.
4297 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
4298 * config/aarch64/iterators.md (wv): New mode attribute.
4299 (vf, VF): New mode attributes.
4300 (vgp, VGP): New mode attributes.
4301 (s): Update attribute with SImode and DImode prefixes.
4303 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
4305 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
4306 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
4307 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
4308 (nios2_option_overide): Initialize it. Don't allow R0-relative
4309 addressing with PIC.
4310 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
4311 (nios2_symbolic_constant_p): Likewise.
4312 (nios2_legitimate_address_p): Likewise.
4313 (nios2_r0rel_section_name_p): New.
4314 (nios2_symbol_ref_in_r0rel_data_p): New.
4315 (nios2_emit_move_sequence): Handle r0rel_constant_p.
4316 (r0rel_constant_p): New.
4317 (nios2_print_operand_address): Handle r0rel_constant_p.
4318 (nios2_cdx_narrow_form_p): Likewise.
4319 * config/nios2/nios2.opt (mr0rel-sec=): New option.
4320 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
4321 (Nios II Options): Document -mr0rel-sec.
4323 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
4325 * config/nios2/nios2.c: Include xregex.h.
4326 (nios2_gprel_sec_regex): New.
4327 (nios2_option_overide): Initialize it. Don't allow GP-relative
4328 addressing with PIC.
4329 (nios2_small_section_name_p): Check for regex match.
4330 * config/nios2/nios2.opt (mgprel-sec=): New option.
4331 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
4332 (Nios II Options): Document -mgprel-sec.
4334 2017-10-26 Jim Wilson <wilson@tuliptree.org>
4336 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
4338 2017-10-26 Tom de Vries <tom@codesourcery.com>
4340 PR tree-optimization/82707
4341 * gimple.c (gimple_copy): Fix unsharing of
4342 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
4344 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
4346 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
4347 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
4348 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
4349 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
4350 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
4351 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
4352 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
4353 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
4354 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
4355 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
4356 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
4357 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
4358 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
4359 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
4360 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
4361 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
4362 _mm512_mask_cmpunord_ps_mask): New intrinsics.
4364 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4366 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
4368 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
4369 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
4370 warning to rs6000.c. Remove the Undocumented flag, since it has
4372 (-mabi=ibmlongdouble): Likewise.
4373 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
4374 already set, set the default format for long double.
4375 (rs6000_debug_reg_global): Print whether long double is IBM or
4377 (rs6000_option_override_internal): Rework setting long double
4378 format. Only warn if the user is changing the long double default
4379 and they did not use -Wno-psabi.
4380 * doc/invoke.texi (PowerPC options): Update the documentation for
4381 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
4383 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4384 Alan Hayward <alan.hayward@arm.com>
4385 David Sherwood <david.sherwood@arm.com>
4387 * rtl.h (wider_subreg_mode): New function.
4388 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
4389 rather than an unsigned int *.
4390 * ira-color.c (regno_max_ref_width): Replace with...
4391 (regno_max_ref_mode): ...this new variable.
4392 (coalesced_pseudo_reg_slot_compare): Update accordingly.
4393 Use wider_subreg_mode.
4394 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
4395 rather than an unsigned int *.
4396 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
4397 (process_alt_operands): Likewise.
4398 (invariant_p): Likewise.
4399 * lra-spills.c (assign_mem_slot): Likewise.
4400 (add_pseudo_to_slot): Likewise.
4401 * lra.c (collect_non_operand_hard_regs): Likewise.
4402 (add_regs_to_insn_regno_info): Likewise.
4403 * reload1.c (regno_max_ref_width): Replace with...
4404 (regno_max_ref_mode): ...this new variable.
4405 (reload): Update accordingly. Update call to
4406 ira_sort_regnos_for_alter_reg.
4407 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
4408 (init_eliminable_invariants): Update to use regno_max_ref_mode.
4409 (scan_paradoxical_subregs): Likewise.
4411 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4413 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
4414 (aarch64_frame): Add emit_frame_chain boolean.
4415 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4416 Move eh_return case to aarch64_layout_frame.
4417 (aarch64_layout_frame): Initialize emit_frame_chain.
4418 (aarch64_expand_prologue): Use emit_frame_chain.
4420 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4422 * config/aarch64/aarch64.c (aarch64_layout_frame):
4423 Ensure LR is always stored at the bottom of the callee-saves.
4424 Remove rarely used frame layout which saves callee-saves at top of
4425 frame, so the store of LR can be used as a valid probe in all cases.
4427 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4429 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
4430 Improve unaligned TImode/TFmode base/offset split.
4432 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4433 Alan Hayward <alan.hayward@arm.com>
4434 David Sherwood <david.sherwood@arm.com>
4436 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
4437 * combine.c (find_single_use_1): Likewise.
4438 (expand_field_assignment): Likewise.
4439 (move_deaths): Likewise.
4440 * lra-constraints.c (simplify_operand_subreg): Likewise.
4441 (curr_insn_transform): Likewise.
4442 * lra.c (collect_non_operand_hard_regs): Likewise.
4443 (add_regs_to_insn_regno_info): Likewise.
4444 * rtlanal.c (reg_referenced_p): Likewise.
4445 (covers_regno_no_parallel_p): Likewise.
4447 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4449 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
4450 Don't print any bits outside the precision of the value.
4451 * wide-int.cc (test_printing): Add some new tests.
4453 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4455 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
4456 supports -xbrace_comment option.
4457 * configure: Regenerate.
4458 * config.in: Regenerate.
4459 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
4460 (ASM_CPU_SPEC): Use it.
4462 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4464 * target.def (static_rtx_alignment): New hook.
4465 * targhooks.h (default_static_rtx_alignment): Declare.
4466 * targhooks.c (default_static_rtx_alignment): New function.
4467 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
4468 * doc/tm.texi: Regenerate.
4469 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
4470 instead of targetm.constant_alignment. Remove call to
4472 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4473 (cris_preferred_mininum_alignment): New function, split out from...
4474 (cris_constant_alignment): ...here.
4475 (cris_static_rtx_alignment): New function.
4476 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
4478 (ix86_constant_alignment): ...here.
4479 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4480 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4481 (mmix_static_rtx_alignment): New function.
4482 * config/spu/spu.c (spu_static_rtx_alignment): New function.
4483 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4485 2017-10-26 Tamar Christina <tamar.christina@arm.com>
4488 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
4489 Add flag_trapping_math and flag_fp_int_builtin_inexact.
4491 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
4493 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
4494 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
4495 mark as a sign-extending load.
4496 (local_pic_load_u): Define.
4498 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
4501 * fold-const.c (operand_equal_for_comparison_p): Also return true
4502 if ARG0 is a simple variant of ARG1 with narrower precision.
4503 (fold_ternary_loc): Always pass unstripped operands to the predicate.
4505 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
4507 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
4509 * i386.h (processor_costs): Add gather_static, gather_per_elt,
4510 scatter_static, scatter_per_elt.
4511 * x86-tune-costs.h: Add new cost entries.
4513 2017-10-25 Richard Biener <rguenther@suse.de>
4515 * tree-ssa-sccvn.h (vn_eliminate): Declare.
4516 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
4517 class pass_fre): Move to ...
4518 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
4519 class pass_fre): ... here and adjust for statistics.
4521 2017-10-25 Jakub Jelinek <jakub@redhat.com>
4524 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
4525 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
4526 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
4527 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
4530 2017-10-25 Richard Biener <rguenther@suse.de>
4532 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
4533 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
4534 eliminate_push_avail, eliminate_insert): Move inside...
4535 (class eliminate_dom_walker): ... this class in preparation
4537 (fini_eliminate): Remove by merging with ...
4538 (eliminate): ... this function. Adjust for class changes.
4539 (pass_pre::execute): Remove fini_eliminate call.
4540 (pass_fre::execute): Likewise.
4542 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4545 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
4546 (VPERMI2, VPERMI2I): New mode iterators.
4547 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
4548 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
4550 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
4551 mode iterator. Remove 3 old define_insn patterns.
4552 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
4553 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
4554 VPERMI2 mode iterator, remove the other two expanders.
4555 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
4556 to use VPERMI2 mode iterator, add another alternative for vpermi2*
4557 instructions, remove the other two patterns.
4558 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
4559 mode iterator, remove the other two patterns.
4560 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
4561 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
4562 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
4563 and adjust argument order accordingly.
4564 (ix86_expand_vec_perm): Adjust caller.
4565 (expand_vec_perm_1): Likewise.
4566 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
4567 (expand_vec_perm_vpermt2_vpshub2): ... this.
4568 (ix86_expand_vec_perm_const_1): Adjust caller.
4569 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
4572 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
4573 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
4574 (vec_shl_<mode>): Remove unused expander.
4575 (avx512bw_<shift_insn><mode>3): New define_insn.
4576 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
4577 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
4579 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
4582 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
4585 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4587 PR rtl-optimization/82396
4588 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
4589 (autopref_multipass_init): Simplify initialization.
4590 (autopref_rank_data): Simplify sort order.
4591 * gcc/sched-int.h (autopref_multipass_data_): Remove
4592 multi_mem_insn_p, min_offset and max_offset.
4594 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4597 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4598 Check special value of flag_omit_frame_pointer.
4599 (aarch64_can_eliminate): Likewise.
4600 (aarch64_override_options_after_change_1): Simplify handling of
4601 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
4603 2017-10-24 Richard Biener <rguenther@suse.de>
4605 PR tree-optimization/82697
4606 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
4607 zero for conditional load and unconditional store.
4609 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4611 * doc/install.texi: Document bootstrap-cet.
4613 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4616 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
4617 ENDBR instruction at function entrance if function is only
4620 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4623 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
4624 patterns to better describe from which operation the CF is computed.
4625 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
4626 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
4627 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
4628 is 0, use _0 suffixed expanders instead of emitting a comparison
4631 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
4633 * config/i386/i386.md(*movsf_internal, *movdf_internal):
4634 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
4636 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
4639 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
4640 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
4641 * loop-iv.c (iv_get_reaching_def): Likewise.
4642 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
4643 variable is promoted and the partition contains undefined values.
4645 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4647 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4649 (nios2_address_cost): Define.
4650 (nios2_legitimize_address): Recognize (exp + constant) directly.
4651 (TARGET_ADDRESS_COST): Define.
4653 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4655 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4656 (nios2_symbolic_memory_operand_p): Declare.
4657 (nios2_split_large_constant): Declare.
4658 (nios2_split_symbolic_memory_operand): Declare.
4659 * config/nios2/nios2.c: Adjust includes.
4660 (nios2_symbolic_constant_allowed): New.
4661 (nios2_symbolic_constant_p): New.
4662 (nios2_plus_symbolic_constant_p): New.
4663 (nios2_valid_addr_expr_p): Recognize addresses involving
4665 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4666 (nios2_symbolic_memory_operand_p): New.
4667 (nios2_large_constant_p): New.
4668 (nios2_split_large_constant): New.
4669 (nios2_split_plus_large_constant): New.
4670 (nios2_split_symbolic_memory_operand): New.
4671 (nios2_legitimize_address): Code refactoring. Handle addresses
4672 involving symbolic constants.
4673 (nios2_emit_move_sequence): Likewise.
4674 (nios2_print_operand): Improve error output.
4675 (nios2_print_operand_address): Handle LO_SUM.
4676 (nios2_cdx_narrow_form_p): Likewise.
4677 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4678 operands involving symbolic constants.
4679 (movhi_internal, movsi_internal): Likewise.
4680 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4681 (extendhisi2, extendqi<mode>2): Likewise.
4683 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4685 * tree-pass.h (PROP_rtl_split_insns): Define.
4686 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4688 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4690 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4692 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4695 * target.def (const_not_ok_for_debug_p): Default to
4696 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4697 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4698 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4699 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4700 which targetm.const_not_ok_for_debug_p returned true.
4701 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4703 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4705 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4706 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4707 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4709 (ix86_const_not_ok_for_debug_p): New function.
4710 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4711 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4713 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4716 * system.h: Conditionally include "unique-ptr.h" if
4717 INCLUDE_UNIQUE_PTR is defined.
4718 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4719 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4721 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4723 * config/rl78/rl78.md: New define_expand "subdi3".
4725 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4728 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4729 DF_REF_INSN if DF_REF_INSN_INFO is false.
4731 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4733 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4734 xmm_move instead of sse_move.
4735 (sse_store_index): New function.
4736 (ix86_register_move_cost): Be more sensible about mismatch stall;
4737 model AVX moves correctly; make difference between sse->integer and
4739 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4740 moves; make difference between SSE and AVX.
4741 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4742 and zmm_move. Increase size of sse load and store tables;
4743 add unaligned load and store tables; add ssemmx_to_integer.
4744 * x86-tune-costs.h: Update all entries according to real
4745 move latencies from Agner Fog's manual and chip documentation.
4747 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4750 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4751 * config/i386/constraints.md (Wf): New constraint.
4752 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4753 (cmp<dwi>_doubleword): Removed.
4754 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4755 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4756 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4757 expand with cmp<dwi>_doubleword. For LTU and GEU use
4758 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4760 * common.opt (gcolumn-info): Enable by default.
4761 * doc/invoke.texi (gcolumn-info): Document new default.
4763 2017-10-23 Richard Biener <rguenther@suse.de>
4765 PR tree-optimization/82672
4766 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4767 Fold the stmt if we propagated into it.
4769 2017-10-23 Richard Biener <rguenther@suse.de>
4771 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4772 (bitmap_remove_expr_from_set): ... this. All callers call this
4773 for non-constant values.
4774 (bitmap_set_subtract): Rename to...
4775 (bitmap_set_subtract_expressions): ... this. Adjust and
4777 (bitmap_set_contains_value): Remove superfluous check.
4778 (bitmap_set_replace_value): Inline into single caller ...
4779 (bitmap_value_replace_in_set): ... here and simplify.
4780 (dependent_clean): Merge into ...
4781 (clean): ... this using an overload. Adjust.
4782 (prune_clobbered_mems): Adjust.
4783 (compute_antic_aux): Likewise.
4784 (compute_partial_antic_aux): Likewise.
4786 2017-10-23 Richard Biener <rguenther@suse.de>
4788 PR tree-optimization/82129
4790 2017-08-01 Richard Biener <rguenther@suse.de>
4792 PR tree-optimization/81181
4793 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4794 (compute_antic): ... end of iteration here.
4796 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4798 * target.def (starting_frame_offset): New hook.
4799 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4800 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4801 * doc/tm.texi.in: Regenerate.
4802 * hooks.h (hook_hwi_void_0): Declare.
4803 * hooks.c (hook_hwi_void_0): New function.
4804 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4805 STARTING_FRAME_OFFSET.
4806 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4807 * reload1.c (reload): Likewise.
4808 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4809 instead of STARTING_FRAME_OFFSET.
4810 * function.c (try_fit_stack_local): Likewise.
4811 (assign_stack_local_1): Likewise
4812 (instantiate_virtual_regs): Likewise.
4813 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4814 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4815 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4816 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4817 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4818 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4819 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4820 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4821 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4822 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4823 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4824 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4825 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4826 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4827 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4828 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4829 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4830 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4831 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4832 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4833 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4834 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4835 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4836 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4837 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4838 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4839 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4840 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4841 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4842 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4843 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4844 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4845 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4846 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4847 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4848 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4849 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4850 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4851 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4852 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4853 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4854 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4855 return a HOST_WIDE_INT.
4856 (avr_builtin_setjmp_frame_value): Use it instead of
4857 STARTING_FRAME_OFFSET.
4858 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4859 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4860 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4862 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4863 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4864 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4865 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4866 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4867 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4868 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4869 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4870 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4871 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4872 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4873 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4875 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4876 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4877 * config/mips/mips.c (mips_compute_frame_info): Refer to
4878 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4879 (mips_starting_frame_offset): New function.
4880 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4881 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4882 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4883 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4884 and return a HOST_WIDE_INT.
4885 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4886 (mmix_initial_elimination_offset): Refer to
4887 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4888 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4889 * config/pa/pa.c (pa_starting_frame_offset): New function.
4890 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4891 (pa_expand_prologue): Likewise.
4892 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4893 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4894 !FRAME_GROWS_DOWNWARD handling to...
4895 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4896 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4897 !FRAME_GROWS_DOWNWARD handling to...
4898 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4899 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4900 !FRAME_GROWS_DOWNWARD handling to...
4901 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4902 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4904 (rs6000_starting_frame_offset): New function.
4905 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4906 !FRAME_GROWS_DOWNWARD handling to...
4907 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4908 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4909 !FRAME_GROWS_DOWNWARD handling to...
4910 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4911 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4912 !FRAME_GROWS_DOWNWARD handling to...
4913 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4914 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4915 (rs6000_starting_frame_offset): New function.
4916 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4917 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4918 * config/vax/vax.c (vax_starting_frame_offset): New function.
4919 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4920 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4921 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4922 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4923 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4924 * system.h (STARTING_FRAME_OFFSET): Poison.
4926 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4928 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4929 SCALAR_TYPE_MODE instead of TYPE_MODE.
4931 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4932 Alan Hayward <alan.hayward@arm.com>
4933 David Sherwood <david.sherwood@arm.com>
4935 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4937 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4938 Alan Hayward <alan.hayward@arm.com>
4939 David Sherwood <david.sherwood@arm.com>
4941 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4943 2017-10-23 Richard Biener <rguenther@suse.de>
4945 PR tree-optimization/82129
4946 * tree-ssa-pre.c (bitmap_set_and): Remove.
4947 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4948 canonicalizing expressions in the set to those with lowest
4949 ID rather than taking that from the first edge.
4951 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4953 * combine.c (rtx_equal_for_field_assignment_p): Use
4954 byte_lowpart_offset.
4956 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4957 Alan Hayward <alan.hayward@arm.com>
4958 David Sherwood <david.sherwood@arm.com>
4960 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4961 to a SUBREG_PROMOTED_VAR.
4963 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4964 Alan Hayward <alan.hayward@arm.com>
4965 David Sherwood <david.sherwood@arm.com>
4967 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4968 (expand_debug_source_expr): Likewise.
4969 * combine.c (combine_simplify_rtx): Likewise.
4970 * cse.c (fold_rtx): Likewise.
4971 * optabs.c (expand_float): Likewise.
4972 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4973 (simplify_binary_operation_1): Likewise.
4975 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4976 Alan Hayward <alan.hayward@arm.com>
4977 David Sherwood <david.sherwood@arm.com>
4979 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4980 (record_promoted_value): Likewise.
4981 * expr.c (expand_expr_real_2): Likewise.
4982 * ree.c (update_reg_equal_equiv_notes): Likewise.
4983 (combine_set_extension): Likewise.
4984 * rtlanal.c (low_bitmask_len): Likewise.
4985 * simplify-rtx.c (neg_const_int): Likewise.
4986 (simplify_binary_operation_1): Likewise.
4988 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4989 Alan Hayward <alan.hayward@arm.com>
4990 David Sherwood <david.sherwood@arm.com>
4992 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4993 * regcprop.c (maybe_mode_change): Likewise.
4994 * reload1.c (alter_reg): Likewise.
4996 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4998 * inchash.h (inchash::hash::add_wide_int): New function.
4999 * lto-streamer-out.c (hash_tree): Use it.
5001 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
5003 * inchash.h (inchash::hash::add_wide_int): Rename to...
5004 (inchash::hash::add_hwi): ...this.
5005 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
5006 (polymorphic_call_target_hasher::hash): Likewise.
5007 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
5008 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
5009 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
5010 * lto-streamer-out.c (hash_tree): Likewise.
5011 * optc-save-gen.awk: Likewise.
5012 * tree.c (add_expr): Likewise.
5014 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
5017 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
5018 for ordered inequality comparisons even with TARGET_IEEE_FP.
5020 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
5023 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
5024 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5025 Expand with cmp<dwi>_doubleword.
5027 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5029 * extend.texi: Add x86 specific to 'nocf_check' attribute.
5030 List CET intrinsics.
5031 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
5032 specific to -fcf-protection option.
5034 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5036 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
5037 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
5038 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
5039 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
5040 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
5041 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
5042 (extra_objs): Add cet.o for Linux/x86 targets.
5043 (tmake_file): Add i386/t-cet for Linux/x86 targets.
5044 * config/i386/cet.c: New file.
5045 * config/i386/cetintrin.h: Likewise.
5046 * config/i386/t-cet: Likewise.
5047 * config/i386/cpuid.h (bit_SHSTK): New.
5048 (bit_IBT): Likewise.
5049 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
5050 pass IBT and SHSTK bits.
5051 * config/i386/i386-builtin-types.def
5052 (VOID_FTYPE_UNSIGNED_PVOID): New.
5053 (VOID_FTYPE_UINT64_PVOID): Likewise.
5054 * config/i386/i386-builtin.def: Add CET intrinsics.
5055 * config/i386/i386-c.c (ix86_target_macros_internal): Add
5056 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
5057 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
5058 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
5060 * config/i386/i386.c (rest_of_insert_endbranch): New.
5061 (pass_data_insert_endbranch): Likewise.
5062 (pass_insert_endbranch): Likewise.
5063 (make_pass_insert_endbranch): Likewise.
5064 (ix86_notrack_prefixed_insn_p): Likewise.
5065 (ix86_target_string): Add -mibt, -mshstk flags.
5066 (ix86_option_override_internal): Add flag_cf_protection
5068 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
5069 (ix86_print_operand): Add 'notrack' prefix output.
5070 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
5071 (ix86_expand_builtin): Expand CET intrinsics.
5072 (x86_output_mi_thunk): Add 'endbranch' instruction.
5073 * config/i386/i386.h (TARGET_IBT): New.
5074 (TARGET_IBT_P): Likewise.
5075 (TARGET_SHSTK): Likewise.
5076 (TARGET_SHSTK_P): Likewise.
5077 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
5078 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
5079 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
5080 (builtin_setjmp_setup): New pattern.
5081 (builtin_longjmp): Likewise.
5082 (rdssp<mode>): Likewise.
5083 (incssp<mode>): Likewise.
5084 (saveprevssp): Likewise.
5085 (rstorssp): Likewise.
5086 (wrss<mode>): Likewise.
5087 (wruss<mode>): Likewise.
5088 (setssbsy): Likewise.
5089 (clrssbsy): Likewise.
5090 (nop_endbr): Likewise.
5091 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
5093 * config/i386/immintrin.h: Include <cetintrin.h>.
5094 * config/i386/linux-common.h
5095 (file_end_indicate_exec_stack_and_cet): New prototype.
5096 (TARGET_ASM_FILE_END): New.
5098 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
5100 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
5101 latencies instead of having separate table; make difference between
5102 integer and float costs.
5103 * i386.h (processor_costs): Remove scalar_stmt_cost,
5104 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
5105 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
5107 * x86-tune-costs.h: Remove entries which has been removed in
5108 procesor_costs from all tables; make cond_taken_branch_cost
5109 and cond_not_taken_branch_cost COST_N_INSNS based.
5111 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
5113 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
5115 2017-10-20 Jakub Jelinek <jakub@redhat.com>
5117 * config/i386/i386.md (isa): Remove fma_avx512f.
5118 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
5119 <avx512>_fmadd_<mode>_mask3<round_name>,
5120 <avx512>_fmsub_<mode>_mask<round_name>,
5121 <avx512>_fmsub_<mode>_mask3<round_name>,
5122 <avx512>_fnmadd_<mode>_mask<round_name>,
5123 <avx512>_fnmadd_<mode>_mask3<round_name>,
5124 <avx512>_fnmsub_<mode>_mask<round_name>,
5125 <avx512>_fnmsub_<mode>_mask3<round_name>,
5126 <avx512>_fmaddsub_<mode>_mask<round_name>,
5127 <avx512>_fmaddsub_<mode>_mask3<round_name>,
5128 <avx512>_fmsubadd_<mode>_mask<round_name>,
5129 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
5130 (*vec_widen_umult_even_v16si<mask_name>,
5131 *vec_widen_smult_even_v16si<mask_name>): Likewise.
5132 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
5134 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5136 * extend.texi: Add 'nocf_check' documentation.
5137 * gimple.texi: Add second parameter to
5138 gimple_build_call_from_tree.
5139 * invoke.texi: Add -fcf-protection documentation.
5140 * rtl.texi: Add REG_CALL_NOTRACK documenation.
5142 2017-10-20 Richard Biener <rguenther@suse.de>
5144 PR tree-optimization/82473
5145 * tree-vect-loop.c (vectorizable_reduction): Properly get at
5146 the largest input type.
5148 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5150 * c-attribs.c (handle_nocf_check_attribute): New function.
5151 (c_common_attribute_table): Add 'nocf_check' handling.
5152 * gimple-parser.c: Add second argument NULL to
5153 gimple_build_call_from_tree.
5154 * attrib.c (comp_type_attributes): Check nocf_check attribute.
5155 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
5157 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
5158 * common.opt: Add fcf-protection flag.
5159 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
5160 * flag-types.h: Add enum cf_protection_level.
5161 * gimple.c (gimple_build_call_from_tree): Add second parameter.
5162 Add 'nocf_check' attribute propagation to gimple call.
5163 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
5164 (gimple_build_call_from_tree): Update prototype.
5165 (gimple_call_nocf_check_p): New function.
5166 (gimple_call_set_nocf_check): Likewise.
5167 * gimplify.c: Add second argument to gimple_build_call_from_tree.
5168 * ipa-icf.c: Add nocf_check attribute in statement hash.
5169 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
5170 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
5171 * toplev.c (process_options): Add flag_cf_protection handling.
5173 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5175 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
5177 2017-10-20 Richard Biener <rguenther@suse.de>
5179 PR tree-optimization/82603
5180 * tree-if-conv.c (predicate_mem_writes): Make sure to only
5181 remove false predicated stores.
5183 2017-10-20 Richard Biener <rguenther@suse.de>
5185 * graphite-isl-ast-to-gimple.c
5186 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
5187 Remove return value and simplify, dump copied stmt after lhs
5189 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
5190 Reduce dump verbosity.
5191 (gsi_insert_earliest): Likewise.
5192 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
5193 * graphite.c (print_global_statistics): Adjust dumping.
5194 (print_graphite_scop_statistics): Likewise.
5195 (print_graphite_statistics): Do not dump loops here.
5196 (graphite_transform_loops): But here.
5198 2017-10-20 Nicolas Roche <roche@adacore.com>
5200 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
5201 * configure: Regenerate.
5203 2017-10-20 Jakub Jelinek <jakub@redhat.com>
5206 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
5207 functions when optimizing replace GIMPLE_RETURN stmts with
5208 calls to __builtin_unreachable ().
5211 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
5212 for -fsanitize=thread link of executables.
5213 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
5214 link of executables.
5217 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
5219 (<shift_insn><mode>3<mask_name>): Change the last of the 3
5220 define_insns for logical vector shifts to use VI248_AVX512BW
5221 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
5222 condition, useless isa and prefix attributes. Change the first
5223 2 of these define_insns to ...
5224 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
5225 define_insn for avx512vl.
5226 (<shift_insn><mode>3): ... and this, new define_insn without
5227 masking for non-avx512vl.
5230 * config/i386/sse.md (*andnot<mode>3,
5231 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
5232 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
5233 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
5234 not applied use empty suffix even for TARGET_AVX512VL.
5235 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
5236 is applied, supply evex,evex or evex,evex,evex instead of just
5239 2017-10-20 Julia Koval <julia.koval@intel.com>
5241 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
5242 (OPTION_MASK_ISA_GFNI_UNSET): New.
5243 (ix86_handle_option): Handle OPT_mgfni.
5244 * config/i386/cpuid.h (bit_GFNI): New.
5245 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
5246 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
5247 * config/i386/i386.c (ix86_target_string): Add -mgfni.
5248 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
5249 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
5250 * config/i386/i386.opt: Add mgfni.
5252 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
5254 * config/msp430/msp430.c (msp430_option_override): Disable
5255 -fdelete-null-pointer-checks.
5256 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
5258 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5260 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
5261 of x87 and SSE instructions.
5263 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5265 * asan.c (create_cond_insert_point): Do not update edge count.
5266 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
5267 (afdo_propagate_circuit): Likewise.
5268 (afdo_calculate_branch_prob): Likewise.
5269 (afdo_annotate_cfg): Likewise.
5270 * basic-block.h (struct edge_def): Remove count.
5271 (edge_def::count): New accessor.
5272 * bb-reorder.c (rotate_loop): Update.
5273 (find_traces_1_round): Update.
5274 (connect_traces): Update.
5275 (sanitize_hot_paths): Update.
5276 * cfg.c (unchecked_make_edge): Update.
5277 (make_single_succ_edge): Update.
5278 (check_bb_profile): Update.
5279 (dump_edge_info): Update.
5280 (update_bb_profile_for_threading): Update.
5281 (scale_bbs_frequencies_int): Update.
5282 (scale_bbs_frequencies_gcov_type): Update.
5283 (scale_bbs_frequencies_profile_count): Update.
5284 (scale_bbs_frequencies): Update.
5285 * cfganal.c (connect_infinite_loops_to_exit): Update.
5286 * cfgbuild.c (compute_outgoing_frequencies): Update.
5287 (find_many_sub_basic_blocks): Update.
5288 * cfgcleanup.c (try_forward_edges): Update.
5289 (try_crossjump_to_edge): Update
5290 * cfgexpand.c (expand_gimple_cond): Update
5291 (expand_gimple_tailcall): Update
5292 (construct_exit_block): Update
5293 * cfghooks.c (verify_flow_info): Update
5294 (redirect_edge_succ_nodup): Update
5295 (split_edge): Update
5296 (make_forwarder_block): Update
5297 (duplicate_block): Update
5298 (account_profile_record): Update
5299 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
5300 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
5301 * cfgloopmanip.c (scale_loop_profile): Update.
5303 (lv_adjust_loop_entry_edge): Update.
5304 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
5305 (force_nonfallthru_and_redirect): Update.
5306 (purge_dead_edges): Update.
5307 (rtl_flow_call_edges_add): Update.
5308 * cgraphunit.c (init_lowered_empty_function): Update.
5309 (cgraph_node::expand_thunk): Update.
5310 * gimple-pretty-print.c (dump_probability): Update.
5311 (dump_edge_probability): Update.
5312 * gimple-ssa-isolate-paths.c (isolate_path): Update.
5313 * haifa-sched.c (sched_create_recovery_edges): Update.
5314 * hsa-gen.c (convert_switch_statements): Update.
5315 * ifcvt.c (dead_or_predicable): Update.
5316 * ipa-inline-transform.c (inline_transform): Update.
5317 * ipa-split.c (split_function): Update.
5318 * ipa-utils.c (ipa_merge_profiles): Update.
5319 * loop-doloop.c (add_test): Update.
5320 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
5321 * lto-streamer-in.c (input_cfg): Update.
5322 (input_function): Update.
5323 * lto-streamer-out.c (output_cfg): Update.
5324 * modulo-sched.c (sms_schedule): Update.
5325 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
5326 * predict.c (maybe_hot_edge_p): Update.
5327 (unlikely_executed_edge_p): Update.
5328 (probably_never_executed_edge_p): Update.
5329 (dump_prediction): Update.
5330 (drop_profile): Update.
5331 (propagate_unlikely_bbs_forward): Update.
5332 (determine_unlikely_bbs): Update.
5333 (force_edge_cold): Update.
5334 * profile.c (compute_branch_probabilities): Update.
5335 * reg-stack.c (better_edge): Update.
5336 * shrink-wrap.c (handle_simple_exit): Update.
5337 * tracer.c (better_p): Update.
5338 * trans-mem.c (expand_transaction): Update.
5339 (split_bb_make_tm_edge): Update.
5340 * tree-call-cdce.c: Update.
5341 * tree-cfg.c (gimple_find_sub_bbs): Update.
5342 (gimple_split_edge): Update.
5343 (gimple_duplicate_sese_region): Update.
5344 (gimple_duplicate_sese_tail): Update.
5345 (gimple_flow_call_edges_add): Update.
5346 (insert_cond_bb): Update.
5347 (execute_fixup_cfg): Update.
5348 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
5349 * tree-complex.c (expand_complex_div_wide): Update.
5350 * tree-eh.c (lower_resx): Update.
5351 (unsplit_eh): Update.
5352 (cleanup_empty_eh_move_lp): Update.
5353 * tree-inline.c (copy_edges_for_bb): Update.
5354 (freqs_to_counts): Update.
5355 (copy_cfg_body): Update.
5356 * tree-ssa-dce.c (remove_dead_stmt): Update.
5357 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
5358 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
5359 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
5360 (unloop_loops): Update.
5361 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
5362 * tree-ssa-loop-split.c (connect_loops): Update.
5363 (split_loop): Update.
5364 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
5365 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
5366 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
5367 * tree-ssa-reassoc.c (branch_fixup): Update.
5368 * tree-ssa-tail-merge.c (replace_block_by): Update.
5369 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
5370 (compute_path_counts): Update.
5371 (update_profile): Update.
5372 (recompute_probabilities): Update.
5373 (update_joiner_offpath_counts): Update.
5374 (estimated_freqs_path): Update.
5375 (freqs_to_counts_path): Update.
5376 (clear_counts_path): Update.
5377 (ssa_fix_duplicate_block_edges): Update.
5378 (duplicate_thread_path): Update.
5379 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
5380 (case_bit_test_cmp): Update.
5381 (collect_switch_conv_info): Update.
5382 (gen_inbound_check): Update.
5383 (do_jump_if_equal): Update.
5384 (emit_cmp_and_jump_insns): Update.
5385 * tree-tailcall.c (decrease_profile): Update.
5386 (eliminate_tail_call): Update.
5387 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
5388 (vect_do_peeling): Update.
5389 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
5390 * ubsan.c (ubsan_expand_null_ifn): Update.
5391 (ubsan_expand_ptr_ifn): Update.
5392 * value-prof.c (gimple_divmod_fixed_value): Update.
5393 (gimple_mod_pow2): Update.
5394 (gimple_mod_subtract): Update.
5395 (gimple_ic): Update.
5396 (gimple_stringop_fixed_value): Update.
5398 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5401 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
5403 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
5405 PR rtl-optimization/82395
5406 * ira-color.c (allocno_priority_compare_func): Fix comparison step
5407 based on non_spilled_static_chain_regno_p.
5409 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5411 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
5412 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
5413 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
5415 2017-10-19 Martin Sebor <msebor@redhat.com>
5417 PR tree-optimization/82596
5418 * tree.c (array_at_struct_end_p): Handle STRING_CST.
5420 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5422 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
5423 (get_mem_refs_of_builtin_call): Likewise.
5424 * builtins.c (expand_builtin_apply): Adjust call to
5425 allocate_dynamic_stack_space.
5426 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
5427 the third argument to allocate_dynamic_stack_space, otherwise -1.
5428 (expand_builtin): Deal with all alloca variants.
5429 (is_inexpensive_builtin): Likewise.
5430 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
5431 * calls.c (special_function_p): Deal with all alloca variants.
5432 (initialize_argument_information): Adjust call to
5433 allocate_dynamic_stack_space.
5434 (expand_call): Likewise.
5435 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
5436 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
5437 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
5438 use it for the stack usage computation.
5439 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
5440 * function.c (gimplify_parameters): Call build_alloca_call_expr.
5441 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
5442 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
5443 (in_loop_p): Remove first argument and useless check.
5444 (pass_walloca::execute): Remove useless test and adjust call to above.
5445 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
5446 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
5447 (gimplify_call_expr): Deal with all alloca variants.
5448 * hsa-gen.c (gen_hsa_alloca): Likewise.
5449 (gen_hsa_insns_for_call): Likewise.
5450 * ipa-pure-const.c (special_builtin_state): Likewise.
5451 * tree-chkp.c (chkp_build_returned_bound): Likewise.
5452 * tree-object-size.c (alloc_object_size): Likewise.
5453 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5454 (call_may_clobber_ref_p_1): Likewise.
5455 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
5456 (ccp_fold_stmt): Likewise.
5457 (optimize_stack_restore): Likewise.
5458 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5459 (mark_all_reaching_defs_necessary_1): Likewise.
5460 (propagate_necessity): Likewise.
5461 (eliminate_unnecessary_stmts): Likewise.
5462 * tree.c (build_common_builtin_nodes): Build
5463 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
5464 (build_alloca_call_expr): New function.
5465 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
5466 (CASE_BUILT_IN_ALLOCA): Likewise.
5467 (build_alloca_call_expr): Declare.
5468 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
5470 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5473 * dwarf2out.c (new_die_raw): New static inline function.
5474 (new_die): Use it to create the DIE.
5475 (add_AT_external_die_ref): Likewise.
5476 (clone_die): Likewise.
5477 (clone_as_declaration): Likewise.
5478 (dwarf2out_vms_debug_main_pointer): Likewise.
5479 (base_type_die): Likewise. Remove early return for corner cases.
5480 Do not call add_pubtype on the DIE here.
5481 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
5482 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
5483 typedefs for base types with DW_AT_endianity. Make sure a DIE with
5484 native order exists for base types, attach the DIE manually and call
5485 add_pubtype on it. Do not equate a reverse order DIE to the type.
5487 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
5489 * config/arm/arm.c (align_ok_ldrd_strd): New function.
5490 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
5492 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
5494 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5496 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
5497 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
5498 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
5499 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
5500 * opts.c (sanitizer_opts): Add builtin.
5501 * ubsan.c (instrument_builtin): New function.
5502 (pass_ubsan::execute): Call it.
5503 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
5504 * doc/invoke.texi: Document -fsanitize=builtin.
5506 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
5507 builtins, store max (log2 (align), 0) into uchar field instead of
5508 align into uptr field.
5509 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
5510 store uchar 0 field instead of uptr 0 field.
5511 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
5512 instead of passing one address of struct with 2 locations pass
5513 two addresses of structs with 1 location each.
5514 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
5515 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
5516 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
5517 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
5518 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
5519 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
5520 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
5521 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
5523 2017-10-19 Martin Liska <mliska@suse.cz>
5526 * file-find.c (remove_prefix): Remove.
5527 * file-find.h (remove_prefix): Likewise.
5528 * gcc-ar.c: Remove smartness of lookup.
5530 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
5532 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
5533 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
5534 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
5536 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5539 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
5540 (setcc + and to xor + setcc): New peephole2.
5542 2017-10-19 Tom de Vries <tom@codesourcery.com>
5544 * doc/sourcebuild.texi (Test Directives, Variants of
5545 dg-require-support): Add dg-require-stack-size.
5547 2017-10-19 Martin Liska <mliska@suse.cz>
5550 * gimplify.c (gimplify_decl_expr): Do not instrument variables
5551 that have a large alignment.
5552 (gimplify_target_expr): Likewise.
5554 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
5556 PR rtl-optimization/82602
5557 * ira.c (rtx_moveable_p): Return false for volatile asm.
5559 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
5562 * config/i386/i386-modes.def (CCGZ): New CC mode.
5563 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
5564 * config/i386/predicates.md (ix86_comparison_operator):
5566 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5567 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
5568 with double-word subtraction.
5569 (put_condition_code): Handle CCGZmode.
5571 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
5573 * wide-int.cc (debug (const wide_int &)): New.
5574 (debug (const wide_int *)): New.
5575 (debug (const widest_int &)): New.
5576 (debug (const widest_int *)): New.
5578 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
5581 * lra-constraints.c (curr_insn_transform): Use non-input operand
5582 instead of output one for matched reload.
5584 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5586 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
5587 (tree-ssa-loop-ivopts.h): New header file.
5588 (struct builtin_info): New fields.
5589 (classify_builtin_1): Compute and record base and offset parts for
5590 memset builtin partition by calling strip_offset.
5591 (offset_cmp, fuse_memset_builtins): New functions.
5592 (finalize_partitions): Fuse adjacent memset partitions by calling
5594 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
5595 Expose the interface.
5596 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
5598 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5600 PR tree-optimization/82574
5601 * tree-loop-distribution.c (find_single_drs): New parameter. Check
5602 that data reference must be executed exactly once per iteration
5603 against the outermost loop in nest.
5604 (classify_partition): Update call to above function.
5606 2017-10-18 Richard Biener <rguenther@suse.de>
5608 PR tree-optimization/82591
5609 * graphite.c (graphite_transform_loops): Move code gen message
5611 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5612 Here. Handle scop_to_isl_ast failing.
5613 (scop_to_isl_ast): Limit the number of ISL operations.
5615 2017-10-18 Richard Biener <rguenther@suse.de>
5617 * graphite-isl-ast-to-gimple.c
5618 (translate_isl_ast_to_gimple::set_rename): Simplify.
5619 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
5620 (graphite_copy_stmts_from_block): ... here.
5621 (copy_bb_and_scalar_dependences): Simplify.
5622 (add_parameters_to_ivs_params): Canonicalize.
5623 (generate_entry_out_of_ssa_copies): Simplify.
5624 * graphite-sese-to-poly.c (extract_affine_name): Simplify
5625 by passing in ISL dimension.
5626 (parameter_index_in_region_1): Rename to ...
5627 (parameter_index_in_region): ... this.
5628 (extract_affine): Adjust assert, pass down parameter index.
5629 (add_param_constraints): Use range-info when available.
5630 (build_scop_context): Adjust.
5631 * sese.c (new_sese_info): Adjust.
5632 (free_sese_info): Likewise.
5633 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
5634 Remove unused typedefs.
5635 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
5637 2017-10-18 Martin Liska <mliska@suse.cz>
5639 * combine.c (simplify_compare_const): Add gcc_fallthrough.
5641 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5643 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
5644 (s390_sched_init): Do not reset s390_sched_state if we entered the
5645 current basic block via a fallthru edge and all others are unlikely.
5647 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5649 * config/s390/s390.c (NUM_SIDES): New variable.
5650 (LONGRUNNING_THRESHOLD): New variable.
5651 (LATENCY_FACTOR): New variable.
5652 (s390_sched_score): Decrease score for long-running instructions on
5654 (s390_sched_variable_issue): Perform bookkeeping for long-running
5657 2017-10-18 Richard Biener <rguenther@suse.de>
5659 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5660 Simplify with removal of the parameter rename map.
5661 (set_rename): Likewise.
5662 (should_copy_to_new_region): Likewise.
5663 (graphite_copy_stmts_from_block): Likewise.
5664 (copy_bb_and_scalar_dependences): Remove initialization of
5665 unused copied_bb_map.
5667 (copy_internal_parameters): Likewise.
5668 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5669 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5670 Use INTEGRAL_TYPE_P.
5671 (parameter_index_in_region_1): Rename to ...
5672 (assign_parameter_index_in_region): ... this. Assert we have
5673 a parameter we handle.
5674 (scan_tree_for_params): Adjust.
5675 * sese.h (parameter_rename_map_t): Remove.
5676 (struct sese_info_t): Remove unused parameter_rename_map and
5677 copied_bb_map members.
5678 * sese.c (new_sese_info): Adjust.
5679 (free_sese_info): Likewise.
5681 2017-10-18 Martin Liska <mliska@suse.cz>
5684 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5685 on an abnormal edge.
5687 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5689 * doc/invoke.texi (ffunction-sections and fdata-sections):
5692 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5694 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5695 the use statement can throw internally.
5697 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5699 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5700 any RTX present on the RHS of a SET.
5701 * compare-elim.c (try_eliminate_compare): Restore comment.
5703 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5705 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5706 may be also called on tcc_type nodes.
5707 * langhooks.c (lhd_tree_size): Likewise.
5709 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5711 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5712 format_warning_at_substring.
5713 (maybe_warn): Convert source_range * param to a location_t. Pass
5714 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5715 (format_directive): Remove code to extract source_ranges and
5716 source_range * in favor of just a location_t.
5717 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5719 * substring-locations.c (format_warning_va): Convert
5720 source_range * param to a location_t.
5721 (format_warning_at_substring): Likewise.
5722 * substring-locations.h (format_warning_va): Likewise.
5723 (format_warning_at_substring): Likewise.
5725 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5727 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5729 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5730 and scatter/gather ops.
5732 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5733 vec_gather_load and vec_scatter_store.
5734 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5735 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5736 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5737 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5738 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5739 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5741 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5743 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5744 Detect FTST instruction and handle its register pops. Only pop
5745 second operand if can_pop_second_op is true.
5746 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5747 set can_pop_second_op to false in the compare_for_stack_reg call.
5749 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5750 output_fp_compare for stack register operands.
5751 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5752 instructions here. Do not emit stack register pops here. Assert
5753 that FCOMPP pops next to top stack register. Rewrite function.
5755 2017-10-17 Nathan Sidwell <nathan@acm.org>
5758 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5759 use DECL_ASSEMBLER_NAME_RAW.
5762 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5765 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5766 Wilco Dijkstra <wilco.dijkstra@arm.com>
5768 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5769 converstion to Pmode from the buf_addr.
5771 2017-10-17 Richard Biener <rguenther@suse.de>
5773 * graphite-dependences.c (scop_get_reads_and_writes): Change
5774 output parameters to references.
5776 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5778 PR 71026/tree-optimization
5779 * fold-const.c (distribute_real_division): Removed.
5780 (fold_binary_loc): Remove calls to distribute_real_divison.
5782 2017-10-17 Richard Biener <rguenther@suse.de>
5784 * graphite-scop-detection.c
5785 (scop_detection::stmt_has_simple_data_refs_p): Always use
5786 the full nest as region.
5787 (try_generate_gimple_bb): Likewise.
5788 * sese.c (scalar_evolution_in_region): Simplify now that
5789 SCEV can handle instantiation in regions.
5790 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5791 in the non-loop part of a function if requested.
5793 2017-10-17 Richard Biener <rguenther@suse.de>
5795 PR tree-optimization/82563
5796 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5798 (graphite_regenerate_ast_isl): Call it.
5799 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5801 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5803 PR tree-optimization/82549
5804 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5805 Formatting fixes. Instead of calling make_bit_field_ref with negative
5808 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5810 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5811 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5812 _mm_maskz_reduce_ss): New.
5813 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5814 __builtin_ia32_reducess_mask): Ditto..
5815 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5816 * config/i386/sse.md (reduces<mode>): Renamed to ...
5817 (reduces<mode><mask_scalar_name>): ... this.
5818 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5820 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5821 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5823 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5825 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5826 * selftest-run-tests.c (selftest::run_tests): Call
5827 selftest::unique_ptr_tests_cc_tests.
5828 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5829 * unique-ptr-tests.cc: New file.
5831 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5834 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5836 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5837 (make_hard_regno_born, make_hard_regno_dead): Update
5838 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5840 2017-10-16 Jeff Law <law@redhat.com>
5842 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5844 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5846 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5848 2017-10-16 Olivier Hainque <hainque@adacore.com>
5850 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5851 with_cpu if we were configured for an e500v2 target cpu name.
5853 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5855 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5856 * doc/invoke.texi: Document +nodsp as a valid extension for
5859 2017-10-16 Martin Liska <mliska@suse.cz>
5861 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5862 (test_set_range): Likewise.
5863 (test_range_functions): Rename to ...
5864 (test_bit_in_range): ... this.
5865 (sbitmap_c_tests): Add new test.
5867 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5869 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5871 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5872 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5874 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5876 * config/aarch64/aarch64-builtins.c
5877 (aarch64_types_quadopu_lane_qualifiers): New.
5878 (TYPES_QUADOPU_LANE): New.
5879 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5880 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5881 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5882 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5883 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5884 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5885 (Vdottype, DOTPROD): New.
5886 (sur): Add SDOT and UDOT.
5888 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5890 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5891 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5892 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5894 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5895 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5896 Enable TARGET_DOTPROD.
5897 (cortex-a75.cortex-a55): Likewise.
5898 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5900 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5902 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5903 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5905 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5907 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5908 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5909 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5910 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5911 * config/arm/types.md (neon_dot, neon_dot_q): New.
5912 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5914 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5916 * config/arm/arm.h (TARGET_DOTPROD): New.
5917 * config/arm/arm.c (arm_arch_dotprod): New.
5918 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5919 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5920 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5921 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5922 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5923 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5924 * doc/invoke.texi (armv8.2-a): Document dotprod
5926 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5928 * i386.c (ix86_vec_cost): New function.
5929 (ix86_rtx_costs): Handle vector operations better.
5930 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5931 * x86-tune-costs.h: Add new costs to all tables.
5933 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5935 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5937 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5938 divsd, sqrtss and sqrtsd
5939 * x86-tune-costs.h: Add new entries to all costs.
5940 (znver1_cost): Fix to match real instruction latencies.
5942 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5943 Michael Collison <michael.collison@arm.com>
5945 * compare-elim.c: Include emit-rtl.h.
5946 (can_merge_compare_into_arith): New function.
5947 (try_validate_parallel): Likewise.
5948 (try_merge_compare): Likewise.
5949 (try_eliminate_compare): Call the above when no previous clobber
5951 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5954 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5958 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5959 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5963 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5964 to be any operand_equal_p operands. For & (B - 1) require
5965 B to be power of 2. Recognize
5966 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5968 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5971 * optabs.c (expand_memory_blockage): Fix call of
5972 targetm.have_memory_blockage.
5974 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5977 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5978 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5979 objects to extra_objs instead of overwriting it.
5981 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5983 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5984 Use any_fp_register_operand as operand[3] predicate. Simplify
5985 equality test for operands[2] and operands[4] memory location.
5986 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5987 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5988 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5989 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5990 any_fp_register_operand as operand[1] predicate. Simplify
5991 equality test for operands[0] and operands[3] memory location.
5992 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5993 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5994 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5996 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5998 * target-insns.def: Add memory_blockage.
5999 * optabs.c (expand_memory_blockage): New function.
6000 (expand_asm_memory_barrier): Rename ...
6001 (expand_asm_memory_blockage): ... to this.
6002 (expand_mem_thread_fence): Call expand_memory_blockage
6003 instead of expand_asm_memory_barrier.
6004 (expand_mem_singnal_fence): Ditto.
6005 (expand_atomic_load): Ditto.
6006 (expand_atomic_store): Ditto.
6007 * doc/md.texi (Standard Pattern Names For Generation):
6008 Document memory_blockage instruction pattern.
6010 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
6012 * config/rl78/rl78.c (rl78_emit_libcall): New function.
6013 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
6014 * config/rl78/rl78.md: New define_expand "adddi3".
6016 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
6018 * cfghooks.c (verify_flow_info): Disable check that all probabilities
6021 2017-10-13 Jeff Law <law@redhat.com>
6023 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
6025 2017-10-13 Jakub Jelinek <jakub@redhat.com>
6028 * internal-fn.c (expand_mul_overflow): If both operands have
6029 the same highpart of -1 or 0 and the topmost bit of lowpart
6030 is different, overflow is if res <= 0 rather than res < 0.
6032 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
6034 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
6035 TARGET_P9_VECTOR code for unaligned_load case.
6037 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
6039 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
6041 2017-10-13 Nathan Sidwell <nathan@acm.org>
6043 * tree-core.h (tree_contains_struct): Make bool.
6044 * tree.c (tree_contains_struct): Likewise.
6045 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
6046 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
6047 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
6048 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
6049 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
6051 2017-10-13 Richard Biener <rguenther@suse.de>
6053 * graphite-isl-ast-to-gimple.c
6054 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
6055 parameters and dominance check.
6056 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
6057 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
6058 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
6059 Do not update SSA form here or do intermediate IL verification.
6060 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
6061 (graphite_initialize): Remove check on the number of loops in
6062 the function and inline into graphite_transform_loops.
6063 (graphite_finalize): Inline into graphite_transform_loops.
6064 (graphite_transform_loops): Perform SSA update and IL verification
6066 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
6068 2017-10-13 Richard Biener <rguenther@suse.de>
6070 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
6071 graphite_expression_type_precision): Avoid global constructor
6073 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
6074 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
6075 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
6076 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
6077 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
6078 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
6080 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
6083 * config/i386/i386.h (ix86_red_zone_size): New.
6084 * config/i386/i386.md (push peephole2s): Replace
6085 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
6087 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6088 Alan Hayward <alan.hayward@arm.com>
6089 David Sherwood <david.sherwood@arm.com>
6091 * combine.c (can_change_dest_mode): Reject changes in
6092 REGMODE_NATURAL_SIZE.
6094 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6095 Alan Hayward <alan.hayward@arm.com>
6096 David Sherwood <david.sherwood@arm.com>
6098 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
6099 (expand_debug_source_expr): Likewise.
6100 * combine.c (combine_simplify_rtx): Likewise.
6101 * cse.c (fold_rtx): Likewise.
6102 * fwprop.c (canonicalize_address): Likewise.
6103 * targhooks.c (default_shift_truncation_mask): Likewise.
6105 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6106 Alan Hayward <alan.hayward@arm.com>
6107 David Sherwood <david.sherwood@arm.com>
6109 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
6110 (widened_mode): Likewise.
6111 (expand_unop): Likewise.
6112 * ree.c (transform_ifelse): Likewise.
6113 (merge_def_and_ext): Likewise.
6114 (combine_reaching_defs): Likewise.
6115 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6117 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6118 Alan Hayward <alan.hayward@arm.com>
6119 David Sherwood <david.sherwood@arm.com>
6121 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
6122 * combine.c (gen_lowpart_for_combine): Likewise.
6123 * dwarf2out.c (rtl_for_decl_location): Likewise.
6124 * final.c (alter_subreg): Likewise.
6125 * rtlhooks.c (gen_lowpart_general): Likewise.
6126 (gen_lowpart_if_possible): Likewise.
6128 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
6129 Alan Hayward <alan.hayward@arm.com>
6130 David Sherwood <david.sherwood@arm.com>
6132 * calls.c (expand_call): Use subreg_lowpart_offset.
6133 * cse.c (cse_insn): Likewise.
6134 * regcprop.c (copy_value): Likewise.
6135 (copyprop_hardreg_forward_1): Likewise.
6137 2017-10-13 Jakub Jelinek <jakub@redhat.com>
6140 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
6141 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
6142 =Q constraints to +Q and into insn condition add check
6143 that operands[0] and operands[1] are equal.
6144 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
6145 =Q constraints to +Q and into insn condition add check
6146 that operands[0] is equal to either operands[1] or operands[2].
6149 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
6150 instead of handling MINUS_EXPR twice (once for each argument),
6151 canonicalize operand order and handle just once, use rtype where
6152 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
6155 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
6156 any values of __C while still being pattern recognizable as a simple
6159 2017-10-13 Richard Biener <rguenther@suse.de>
6161 PR tree-optimization/82451
6163 2017-10-02 Richard Biener <rguenther@suse.de>
6165 PR tree-optimization/82355
6166 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6167 a mapping for the enclosing loop but avoid generating one for
6169 (copy_bb_and_scalar_dependences): Remove premature codegen
6170 error on PHIs in blocks duplicated into multiple places.
6171 * graphite-scop-detection.c
6172 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6173 in the region use it as loop and nest to analyze the DR in.
6174 (try_generate_gimple_bb): Likewise.
6175 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6176 (add_loop_constraints): For blocks in a loop not in the region
6177 create a dimension with a single iteration.
6178 * sese.h (gbb_loop_at_index): Remove assert.
6180 * cfgloop.c (loop_preheader_edge): For the loop tree root
6181 return the single successor of the entry block.
6182 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6183 Reset the SCEV hashtable and niters.
6184 * graphite-scop-detection.c
6185 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
6186 assert that we only have POLYNOMIAL_CHREC that vary in loops
6187 contained in the region.
6188 (scop_detection::graphite_can_represent_expr): Adjust.
6189 (scop_detection::stmt_has_simple_data_refs_p): For loops
6190 not in the region set loop to NULL. The nest is now the
6191 entry edge to the region.
6192 (try_generate_gimple_bb): Likewise.
6193 * sese.c (scalar_evolution_in_region): Adjust for
6194 instantiate_scev change.
6195 * tree-data-ref.h (graphite_find_data_references_in_stmt):
6196 Make nest parameter the edge into the region.
6197 (create_data_ref): Likewise.
6198 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
6199 entry edge into a region and adjust instantiate_scev calls.
6200 (create_data_ref): Likewise.
6201 (graphite_find_data_references_in_stmt): Likewise.
6202 (find_data_references_in_stmt): Pass the loop preheader edge
6203 from the nest argument.
6204 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
6205 parameter the edge into the region.
6206 (instantiate_parameters): Use the loop preheader edge as entry.
6207 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
6209 (get_instantiated_value_entry): Make instantiate_below parameter
6210 the edge into the region.
6211 (instantiate_scev_name): Likewise. Adjust dominance checks,
6212 when we cannot use loop-based instantiation instantiate by
6213 walking use-def chains.
6214 (instantiate_scev_poly): Adjust.
6215 (instantiate_scev_binary): Likewise.
6216 (instantiate_scev_convert): Likewise.
6217 (instantiate_scev_not): Likewise.
6218 (instantiate_array_ref): Remove.
6219 (instantiate_scev_3): Likewise.
6220 (instantiate_scev_2): Likewise.
6221 (instantiate_scev_1): Likewise.
6222 (instantiate_scev_r): Do not blindly handle N-operand trees.
6223 Do not instantiate array-refs. Handle all constants and invariants.
6224 (instantiate_scev): Make instantiate_below parameter
6225 the edge into the region.
6226 (resolve_mixers): Use the loop preheader edge for the region
6227 parameter to instantiate_scev_r.
6228 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
6230 2017-10-13 Richard Biener <rguenther@suse.de>
6232 PR tree-optimization/82525
6233 * graphite-isl-ast-to-gimple.c
6234 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
6236 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
6237 Fail code generation when we cannot represent the isl integer.
6238 (binary_op_to_tree): Elide modulo operations that are no-ops
6239 in the type we code generate. Remove now superfluous code
6242 2017-10-13 Richard Biener <rguenther@suse.de>
6244 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
6245 (scop_detection::harmful_loop_in_region): Remove premature
6246 IV type restriction.
6247 (scop_detection::graphite_can_represent_scev): We can handle
6248 pointer IVs just fine.
6250 2017-10-13 Alan Modra <amodra@gmail.com>
6252 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
6253 "Clobbers and Scratch Registers". Add paragraph on
6254 alternative to clobbers for scratch registers and OpenBLAS
6257 2017-10-13 Alan Modra <amodra@gmail.com>
6259 * doc/extend.texi (Clobbers): Correct vax example. Delete old
6260 example of a memory input for a string of known length. Move
6261 commentary out of table. Add a number of new examples
6262 covering array memory inputs.
6264 2017-10-12 Martin Liska <mliska@suse.cz>
6266 PR tree-optimization/82493
6267 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
6268 (test_range_functions): New function.
6269 (sbitmap_c_tests): Likewise.
6270 * selftest-run-tests.c (selftest::run_tests): Run new tests.
6271 * selftest.h (sbitmap_c_tests): New function.
6273 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
6275 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6277 * config/rs6000/amo.h: Fix spacing issue.
6279 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6282 * config/i386/i386.md (*ashl<mode>3_mask_1,
6283 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
6284 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
6287 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
6289 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
6290 (profile_probability): Set max_probability
6291 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
6293 * profile-count.c (profile_probability::differs_from_p): Do not
6294 rely on max_probaiblity == 10000
6296 2017-10-12 Jeff Law <law@redhat.com>
6298 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
6301 2017-10-12 Martin Sebor <msebor@redhat.com>
6305 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
6306 (handle_alias_pairs): Call it.
6307 * common.opt (-Wattribute-alias): New option.
6308 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
6309 * doc/invoke.texi (-Wattribute-alias): Document.
6311 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
6314 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6316 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6318 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6319 (make_hard_regno_born, make_hard_regno_dead): Update
6320 bb_killed_pseudos and bb_gen_pseudos.
6322 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
6324 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
6326 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
6328 * config/alpha/alpha.c (alpha_split_conditional_move):
6329 Use std::swap instead of manually swapping.
6330 (alpha_stdarg_optimize_hook): Ditto.
6331 (alpha_canonicalize_comparison): Ditto.
6333 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6335 * tree-loop-distribution.c (struct builtin_info): New struct.
6336 (struct partition): Refactor fields into struct builtin_info.
6337 (partition_free): Free struct builtin_info.
6338 (build_size_arg_loc, build_addr_arg_loc): Delete.
6339 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
6340 information from struct builtin_info.
6341 (find_single_drs): New function refactored from classify_partition.
6342 Also moved builtin validity checks to this function.
6343 (compute_access_range, alloc_builtin): New functions.
6344 (classify_builtin_st, classify_builtin_ldst): New functions.
6345 (classify_partition): Refactor code into functions find_single_drs,
6346 classify_builtin_st and classify_builtin_ldst.
6347 (distribute_loop): Don't do runtime alias check when distributing
6349 (find_seed_stmts_for_distribution): New function.
6350 (pass_loop_distribution::execute): Refactor code finding seed
6351 stmts into above function. Support distribution for the innermost
6352 two-level loop nest. Adjust dump information.
6354 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6356 * tree-loop-distribution.c: Adjust the general comment.
6357 (NUM_PARTITION_THRESHOLD): New macro.
6358 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
6359 (classify_partition): Skip builtin pattern of loop nest's inner loop.
6360 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
6361 in call to build_partition_graph.
6362 (finalize_partitions): New parameter. Make loop distribution more
6363 conservative by fusing more partitions.
6364 (distribute_loop): Don't do runtime alias check in case of loop nest
6366 (find_seed_stmts_for_distribution): New function.
6367 (prepare_perfect_loop_nest): New function.
6368 (pass_loop_distribution::execute): Refactor code finding seed stmts
6369 and loop nest into above functions. Support loop nest distribution.
6370 Adjust dump information accordingly.
6372 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6374 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
6375 and set PTYPE_SEQUENTIAL for merged partition.
6377 2017-10-12 Richard Biener <rguenther@suse.de>
6379 PR tree-optimization/69728
6381 2017-09-19 Richard Biener <rguenther@suse.de>
6383 PR tree-optimization/69728
6384 * graphite-sese-to-poly.c (schedule_error): New global.
6385 (add_loop_schedule): Handle empty domain by failing the
6387 (build_original_schedule): Handle schedule_error.
6389 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
6390 domain by returning an unchanged schedule.
6392 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6394 * genrecog.c (validate_pattern): For VEC_SELECT verify that
6395 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
6397 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
6399 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
6402 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6405 * expr.c (store_field): Don't optimize away bitsize == 0 store
6406 from CALL_EXPR with addressable return type.
6408 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6410 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
6411 * config/rs6000/rs6000.md (sel): Delete mode attribute.
6412 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
6413 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
6414 TARGET_ISEL instead of TARGET_ISEL<sel>.
6416 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
6418 * config/rs6000/rs6000.c
6419 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
6421 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6423 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
6425 (reg_or_cint_operand): Fix comment.
6426 (reg_or_zero_operand): New predicate.
6427 * config/rs6000/rs6000-protos.h (output_isel): Delete.
6428 * config/rs6000/rs6000.c (output_isel): Delete.
6429 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
6430 instead of reg_or_cint_operand. Output instruction directly (not via
6432 (isel_unsigned_<mode>): Ditto.
6433 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
6434 gpc_reg_operand. Add an instruction alternative for this. Output
6435 instruction directly.
6436 (*isel_reversed_unsigned_<mode>): Ditto.
6438 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6440 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
6441 (TARGET_CANONICALIZE_COMPARISON): Define.
6443 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
6446 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6447 Check whether the dest is REG before adding REG_EQUIV note.
6449 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6452 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6454 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6455 (make_hard_regno_born, make_hard_regno_dead): Update
6456 bb_killed_pseudos and bb_gen_pseudos.
6458 2017-10-11 Nathan Sidwell <nathan@acm.org>
6460 * incpath.h (enum incpath_kind): Name enum, prefix values.
6461 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
6462 * incpath.c (heads, tails): Use INC_MAX.
6463 (add_env_var_paths, add_standard_paths): Use incpath_kind.
6464 (merge_include_chains, split_quote_chain,
6465 register_include_chains): Update incpath_kind names.
6466 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
6467 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
6469 (add_framework_path, darwin_register_objc_includes): Likewise.
6470 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
6472 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6474 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
6475 Do not use float_operator operator predicate.
6476 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6477 * config/i386/predicates.md (float_operator): Remove predicate.
6479 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6481 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
6482 (*jccxf_i387): Ditto.
6483 (*jcc<mode>_i387): Ditto.
6484 (*jccu<mode>_i387): Ditto.
6485 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
6486 (*jcc_*_i387 splitters): Remove.
6487 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
6488 * config/i386/i386.c (ix86_split_fp_branch): Remove.
6489 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
6492 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6494 * profile-count.h (slow_safe_scale_64bit): New function.
6495 (safe_scale_64bit): New inline.
6496 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
6497 * profile-count.c: Include wide-int.h
6498 (slow_safe_scale_64bit): New.
6500 2017-10-11 Nathan Sidwell <nathan@acm.org>
6502 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
6503 HAS_DECL_ASSEMBLER_NAME_P.
6504 * gimple-expr.c (gimple_decl_printable_name: Check
6505 HAS_DECL_ASSEMBLER_NAME_P too.
6506 * ipa-utils.h (type_in_anonymous_namespace_p): Check
6507 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
6508 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
6509 * passes.c (rest_of_decl_compilation): Check
6510 HAS_DECL_ASSEMBLER_NAME_P too.
6511 * recog.c (verify_changes): Likewise.
6512 * tree-pretty-print.c (dump_decl_name): Likewise.
6513 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
6515 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
6516 (SET_DECL_ASSEMBLER_NAME): Use it.
6517 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
6518 (COPY_DECL_ASSEMBLER_NAME): Likewise.
6519 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
6521 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6523 * config.gcc (i386, x86_64): Add extra objects.
6524 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
6525 (ix86_min_insn_size): Declare.
6526 (ix86_issue_rate): Declare.
6527 (ix86_adjust_cost): Declare.
6528 (ia32_multipass_dfa_lookahead): Declare.
6529 (ix86_macro_fusion_p): Declare.
6530 (ix86_macro_fusion_pair_p): Declare.
6531 (ix86_bd_has_dispatch): Declare.
6532 (ix86_bd_do_dispatch): Declare.
6533 (ix86_core2i7_init_hooks): Declare.
6534 (ix86_atom_sched_reorder): Declare.
6535 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
6536 (COSTS_N_BYTES): Move to x86-tune-costs.h.
6537 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
6538 (rip_relative_addr_p): Rename to ...
6539 (ix86_rip_relative_addr_p): ... this one; export.
6540 (memory_address_length): Update.
6541 (ix86_issue_rate): Move to x86-tune-sched.c.
6542 (ix86_flags_dependent): Move to x86-tune-sched.c.
6543 (ix86_agi_dependent): Move to x86-tune-sched.c.
6544 (exact_dependency_1): Move to x86-tune-sched.c.
6545 (exact_store_load_dependency): Move to x86-tune-sched.c.
6546 (ix86_adjust_cost): Move to x86-tune-sched.c.
6547 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
6548 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
6549 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
6550 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
6551 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
6552 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
6553 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
6554 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
6555 (min_insn_size): Rename to ...
6556 (ix86_min_insn_size): ... this one; export.
6557 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
6558 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
6559 (core2i7_first_cycle_multipass_backtrack): Move to
6560 x86-tune-sched-core.c.
6561 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
6562 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
6563 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
6564 (ix86_avoid_jump_mispredicts): Update.
6565 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
6566 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
6567 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
6568 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
6569 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
6570 (MAX_INSN): Move to ix86-tune-sched-bd.c.
6571 (MAX_IMM): Move to ix86-tune-sched-bd.c.
6572 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
6573 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
6574 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
6575 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
6576 (MAX_STORE): Move to ix86-tune-sched-bd.c.
6577 (BIG): Move to ix86-tune-sched-bd.c.
6578 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
6579 (enum insn_path): Move to ix86-tune-sched-bd.c.
6580 (get_mem_group): Move to ix86-tune-sched-bd.c.
6581 (is_cmp): Move to ix86-tune-sched-bd.c.
6582 (dispatch_violation): Move to ix86-tune-sched-bd.c.
6583 (is_branch): Move to ix86-tune-sched-bd.c.
6584 (is_prefetch): Move to ix86-tune-sched-bd.c.
6585 (init_window): Move to ix86-tune-sched-bd.c.
6586 (allocate_window): Move to ix86-tune-sched-bd.c.
6587 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
6588 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
6589 (process_end_window): Move to ix86-tune-sched-bd.c.
6590 (allocate_next_window): Move to ix86-tune-sched-bd.c.
6591 (find_constant): Move to ix86-tune-sched-bd.c.
6592 (get_num_immediates): Move to ix86-tune-sched-bd.c.
6593 (has_immediate): Move to ix86-tune-sched-bd.c.
6594 (get_insn_path): Move to ix86-tune-sched-bd.c.
6595 (get_insn_group): Move to ix86-tune-sched-bd.c.
6596 (count_num_restricted): Move to ix86-tune-sched-bd.c.
6597 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
6598 (add_insn_window): Move to ix86-tune-sched-bd.c.
6599 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
6600 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
6601 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
6602 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
6603 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
6604 (do_dispatch): Move to ix86-tune-sched-bd.c.
6605 (has_dispatch): Move to ix86-tune-sched-bd.c.
6606 * config/i386/t-i386: Add new object files.
6607 * config/i386/x86-tune-costs.h: New file.
6608 * config/i386/x86-tune-sched-atom.c: New file.
6609 * config/i386/x86-tune-sched-bd.c: New file.
6610 * config/i386/x86-tune-sched-core.c: New file.
6611 * config/i386/x86-tune-sched.c: New file.
6613 2017-10-11 Liu Hao <lh_mouse@126.com>
6615 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
6616 the generic version below instead.
6617 (should_colorize): Recognize Windows consoles as terminals
6619 * pretty-print.c [__MINGW32__] (write_all): New function.
6620 [__MINGW32__] (find_esc_head): Likewise.
6621 [__MINGW32__] (find_esc_terminator): Likewise.
6622 [__MINGW32__] (eat_esc_sequence): Likewise.
6623 [__MINGW32__] (mingw_ansi_fputs): New function that handles
6625 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
6628 2017-10-11 Richard Biener <rguenther@suse.de>
6630 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
6631 Properly call analyze_scalar_evolution with the loop of the stmt.
6633 2017-10-11 Richard Biener <rguenther@suse.de>
6635 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
6636 * tree-core.h (tree_base): Add chrec_var union member.
6637 * tree.h (CHREC_VAR): Remove.
6638 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
6639 * tree-chrec.h (build_polynomial_chrec): Adjust.
6640 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
6641 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
6643 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
6645 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
6646 * match.pd: ... here.
6647 ((T) X == (T) Y): Relax condition.
6649 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6651 PR tree-optimization/82472
6652 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6654 (break_alias_scc_partitions): Update postorder number.
6656 2017-10-11 Martin Liska <mliska@suse.cz>
6659 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6661 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6663 2017-10-11 Martin Liska <mliska@suse.cz>
6665 * print-rtl.c (print_insn): Move declaration of idbuf
6666 to same scope as name.
6668 2017-10-11 Martin Liska <mliska@suse.cz>
6673 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6676 2017-10-11 Martin Liska <mliska@suse.cz>
6679 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6682 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6684 * config.gcc (default_gnu_indirect_function): Default to yes for
6685 arm*-*-linux* with glibc.
6687 2017-10-11 Richard Biener <rguenther@suse.de>
6689 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6690 default-defs and types we do not want to analyze.
6691 (interpret_loop_phi): Replace unreachable code with an assert.
6692 (compute_scalar_evolution_in_loop): Remove and inline ...
6693 (analyze_scalar_evolution_1): ... here, replacing condition with
6694 what makes the intent clearer. Remove handling of cases
6695 get_scalar_evolution now handles.
6697 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6699 PR rtl-optimization/81434
6700 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6701 comment for main loop. In sched_group_found if, also add checks for
6702 pass and min_cost_group.
6704 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6706 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6707 (rs6000_insn_cost): New function.
6708 * config/rs6000/rs6000.md (cost): New attribute.
6710 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6711 H.J. Lu <hongjiu.lu@intel.com>
6715 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6716 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6717 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6718 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6719 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6720 OPTION_MASK_ISA_MMX and some other bit set require both
6721 mmx and the ISAs without the mmx bit.
6722 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6723 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6724 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6725 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6726 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6727 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6728 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6729 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6730 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6731 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6732 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6733 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6734 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6735 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6736 Add OPTION_MASK_ISA_MMX.
6738 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6740 * config.gcc (armv7*-*-freebsd*): New target.
6741 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6743 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6745 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6746 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6749 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6751 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6752 a template parameter.
6753 (WIDE_INT_REF_FOR): Update accordingly.
6754 * tree.h (wi::int_traits <const_tree>): Delete.
6755 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6756 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6757 (wi::tree_to_wide_ref): New typedef.
6758 (wi::to_wide): New function.
6759 * calls.c (get_size_range): Use wi::to_wide when operating on
6761 * cgraph.c (cgraph_node::create_thunk): Likewise.
6762 * config/i386/i386.c (ix86_data_alignment): Likewise.
6763 (ix86_local_alignment): Likewise.
6764 * dbxout.c (stabstr_O): Likewise.
6765 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6766 * expr.c (const_vector_from_tree): Likewise.
6767 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6768 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6769 (fold_negate_expr_1, int_const_binop_1, const_binop)
6770 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6771 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6772 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6773 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6774 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6775 (fold_not_const, round_up_loc): Likewise.
6776 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6777 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6778 (alloca_call_type): Likewise.
6779 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6780 * godump.c (go_output_typedef): Likewise.
6781 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6782 * internal-fn.c (get_min_precision): Likewise.
6783 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6784 * ipa-polymorphic-call.c
6785 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6786 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6787 (ipa_modify_call_arguments): Likewise.
6788 * match.pd: Likewise.
6789 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6790 * print-tree.c (print_node_brief, print_node): Likewise.
6791 * stmt.c (expand_case): Likewise.
6792 * stor-layout.c (layout_type): Likewise.
6793 * tree-affine.c (tree_to_aff_combination): Likewise.
6794 * tree-cfg.c (group_case_labels_stmt): Likewise.
6795 * tree-data-ref.c (dr_analyze_indices): Likewise.
6796 (prune_runtime_alias_test_list): Likewise.
6797 * tree-dump.c (dequeue_and_dump): Likewise.
6798 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6799 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6800 * tree-pretty-print.c (dump_generic_node): Likewise.
6801 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6802 (simple_iv_with_niters): Likewise.
6803 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6804 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6805 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6806 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6807 (refine_value_range_using_guard, number_of_iterations_ne_max)
6808 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6809 (get_cst_init_from_scev, record_nonwrapping_iv)
6810 (scev_var_range_cant_overflow): Likewise.
6811 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6812 * tree-ssa-pre.c (compute_avail): Likewise.
6813 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6814 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6815 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6816 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6817 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6818 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6819 (dump_case_nodes, try_switch_expansion): Likewise.
6820 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6821 (vect_do_peeling): Likewise.
6822 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6823 * tree-vect-stmts.c (vectorizable_load): Likewise.
6824 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6825 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6826 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6827 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6828 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6829 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6830 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6831 (evrp_dom_walker::before_dom_children): Likewise.
6832 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6833 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6834 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6835 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6836 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6837 (get_range_pos_neg): Likewise.
6838 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6839 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6840 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6841 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6842 * config/avr/avr.c (avr_fold_builtin): Likewise.
6843 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6844 * config/msp430/msp430.c (msp430_attr): Likewise.
6845 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6846 * config/powerpcspe/powerpcspe-c.c
6847 (altivec_resolve_overloaded_builtin): Likewise.
6848 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6849 (rs6000_expand_ternop_builtin): Likewise.
6850 * config/rs6000/rs6000-c.c
6851 (altivec_resolve_overloaded_builtin): Likewise.
6852 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6853 (rs6000_expand_ternop_builtin): Likewise.
6854 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6856 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6858 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6859 when copying loop nest with only one inner loop.
6861 2017-10-10 Richard Biener <rguenther@suse.de>
6863 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6864 blocks if SCEV is active.
6865 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6867 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6868 (scev_initialize): Assert we are not yet initialized.
6870 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6872 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6873 inner loop's exit stmt by making it always exit the loop, otherwise
6874 we would generate an infinite empty loop.
6876 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6878 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6879 renaming variables in new preheader if it's deleted.
6881 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6883 * tree-loop-distribution.c (struct partition): Remove unused field
6884 loops of the structure.
6885 (partition_alloc, partition_free): Ditto.
6886 (build_rdg_partition_for_vertex): Ditto.
6888 2017-10-09 Jeff Law <law@redhat.com>
6890 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6891 return type to match prototype and documentation.
6893 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6895 * config/rs6000/rs6000.c (processor_costs): Move to ...
6896 * config/rs6000/rs6000.h: ... here.
6897 (rs6000_cost): Declare.
6899 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6901 * except.c (setjmp_fn): New global variable.
6902 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6903 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6904 if DONT_USE_BUILTIN_SETJMP is defined.
6906 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6908 * target.def (insn_cost): New hook.
6909 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6910 * doc/tm.texi: Regenerate.
6911 * rtlanal.c (insn_cost): Use the new hook.
6913 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6915 * combine.c (combine_validate_cost): Compute the new insn_cost,
6916 not just pattern_cost.
6917 (try_combine): Adjust comment.
6919 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6921 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6923 * combine.c (uid_insn_cost): Adjust comment.
6924 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6926 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6927 * dse.c (find_shift_sequence): Ditto.
6928 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6929 (bb_valid_for_noce_process_p): Use pattern_cost.
6930 * rtl.h (insn_rtx_cost): Delete.
6931 (pattern_cost): New prototype.
6932 (insn_cost): New prototype.
6933 * rtlanal.c (insn_rtx_cost): Rename to...
6934 (pattern_cost): ... this.
6937 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6939 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6940 (*jcc<mode>_0_r_i387): Ditto.
6941 (*jccxf_r_i387): Ditto.
6942 (*jcc<mode>_r_i387): Ditto.
6943 (*jccu<mode>_r_i387): Ditto.
6944 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6945 (*jcc): Rename from *jcc_1.
6947 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6949 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6950 deferred rescans after the lvx/stvx recombination pre-pass.
6952 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6954 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6955 memory operation instruction support.
6956 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6957 (rs6000-ibm-aix[789]*): Likewise.
6958 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6959 Document new functions.
6961 2017-10-09 Richard Biener <rguenther@suse.de>
6963 PR tree-optimization/82397
6964 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6965 equality only for semantically equal trees.
6967 2017-10-09 Richard Biener <rguenther@suse.de>
6969 PR tree-optimization/82449
6970 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6971 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6972 allow constant addresses.
6973 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6976 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6978 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6981 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6984 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6987 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6990 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6992 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6995 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6996 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6998 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
7000 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
7001 (WI_BINARY_PREDICATE_RESULT): Likewise.
7002 (wi::binary_traits::operator_result): New type.
7003 (wi::binary_traits::predicate_result): Likewise.
7004 (generic_wide_int::operator~, unary generic_wide_int::operator-)
7005 (generic_wide_int::operator==, generic_wide_int::operator!=)
7006 (generic_wide_int::operator&, generic_wide_int::and_not)
7007 (generic_wide_int::operator|, generic_wide_int::or_not)
7008 (generic_wide_int::operator^, generic_wide_int::operator+
7009 (binary generic_wide_int::operator-, generic_wide_int::operator*):
7011 (operator~, unary operator-, operator==, operator!=, operator&)
7012 (operator|, operator^, operator+, binary operator-, operator*): New
7014 * expr.c (get_inner_reference): Use wi::bit_and_not.
7015 * fold-const.c (fold_binary_loc): Likewise.
7016 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
7017 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
7018 (bit_value_binop): Likewise.
7019 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
7020 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
7021 (extract_range_from_binary_expr_1): Likewise.
7022 (masked_increment): Likewise.
7023 (simplify_bit_ops_using_ranges): Likewise.
7025 2017-10-09 Martin Jambor <mjambor@suse.cz>
7028 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
7029 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
7030 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
7032 (hsa_fixup_mov_insn_type): New function.
7033 (hsa_op_with_type::get_in_type): Use it.
7034 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
7035 immediates in an assert.
7036 (hsa_op_with_type::extend_int_to_32bit): New method.
7037 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
7038 types. Convert to dest type if necessary.
7039 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
7040 (reg_for_gimple_ssa): Pass false as min32int to
7041 hsa_type_for_scalar_tree_type.
7042 (gen_hsa_addr): Fixup type when creating addresable temporary.
7043 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
7044 (gen_hsa_unary_operation): Extend operands and convert to dest type if
7045 necessary. Call hsa_fixup_mov_insn_type.
7046 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
7047 extend operands and convert to dest type if necessary.
7048 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
7049 to dest type if necessary.
7050 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
7051 if conversion nt necessary and size matches.
7052 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
7053 to dest type if necessary.
7054 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
7055 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
7057 (gen_hsa_clrsb): Likewise.
7058 (gen_hsa_ffs): Likewise.
7059 (gen_hsa_divmod): Extend operands and convert to dest type if
7061 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
7063 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
7065 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
7066 Remove empty default arguments. Use a brace block as output
7068 (conditional return): Ditto.
7070 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
7071 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
7073 (group_ending_nop): Ditto.
7074 (doloop_end): Ditto.
7075 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
7076 (splitters for those): Ditto.
7078 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
7080 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
7081 a conditional jump (and the compare for it) so that pc_rtx is the
7083 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
7084 for the deleted and renamed ctr<mode>_internal[234] patterns.
7085 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
7086 Delete second conditional return pattern.
7087 (ctr<mode>_internal2): Delete this second bdnz pattern.
7088 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
7089 (ctr<mode>_internal4): Delete this second bdz pattern.
7091 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
7093 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
7094 (always_initialized_rtx_for_ssa_name_p): New predicate.
7095 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
7096 (finish_out_of_ssa): Free new field of SA.
7097 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
7098 * tree-ssa-coalesce.c: Include tree-ssa.h.
7099 (get_parm_default_def_partitions): Remove extern keyword.
7100 (get_undefined_value_partitions): New function.
7101 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
7102 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
7104 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
7106 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
7108 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
7110 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
7112 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
7113 for targets that preffer 128bit.
7115 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
7117 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
7119 2017-10-08 Olivier Hainque <hainque@adacore.com>
7121 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
7122 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
7124 (thumb_set_return_address): Likewise.
7126 2017-10-08 Olivier Hainque <hainque@adacore.com>
7128 * common/config/arm/arm-common.c (arm_except_unwind_info):
7129 Handle DWARF2_UNWIND_INFO.
7131 2017-10-07 Michael Collison <michael.collison@arm.com>
7133 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
7136 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
7138 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
7139 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
7140 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
7141 defined, force the creation of a new block for a dispatch label.
7143 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
7145 * invoke.texi (Wsuggest-attribute=cold): Document.
7146 * common.opt (Wsuggest-attribute=cold): New
7147 * ipa-pure-const.c (warn_function_cold): New function.
7148 * predict.c (compute_function_frequency): Use it.
7149 * predict.h (warn_function_cold): Declare.
7151 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
7153 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
7156 2017-10-06 Martin Liska <mliska@suse.cz>
7158 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
7159 keyword for member functions.
7160 (struct sanopt_tree_couple): New struct.
7161 (struct sanopt_tree_couple_hash): New function.
7162 (struct sanopt_ctx): Add new hash_map.
7163 (has_dominating_ubsan_ptr_check): New function.
7164 (record_ubsan_ptr_check_stmt): Likewise.
7165 (maybe_optimize_ubsan_ptr_ifn): Likewise.
7166 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
7167 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
7169 2017-10-06 Sudakshina Das <sudi.das@arm.com>
7172 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
7173 aarch64_simd_valid_immediate on CONST_VECTORs.
7174 (aarch64_reg_or_bic_imm): Likewise.
7176 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
7178 PR rtl-optimization/82396
7179 * haifa-sched.c (ready_sort_real): Disable qsort checking.
7181 2017-10-06 Sebastian Pop <sebpop@gmail.com>
7183 * graphite-dependences.c (scop_get_reads): Move code to...
7184 (scop_get_must_writes): Move code to...
7185 (scop_get_may_writes): Move code to...
7186 (scop_get_reads_and_writes): ... here.
7187 (scop_get_dependences): Call scop_get_reads_and_writes.
7189 2017-10-06 Jakub Jelinek <jakub@redhat.com>
7191 PR tree-optimization/82434
7192 * fold-const.h (can_native_encode_type_p,
7193 can_native_encode_string_p): Remove.
7194 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
7195 don't encode anything, just return what would be otherwise returned.
7196 (native_encode_fixed, native_encode_complex, native_encode_vector):
7198 (native_encode_string): Likewise. Inline by hand
7199 can_native_encode_string_p.
7200 (can_native_encode_type_p): Remove.
7201 (can_native_encode_string_p): Remove.
7202 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
7203 STRING_CSTs using can_native_encode_string_p, test all
7204 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
7205 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
7206 argument from native_encode_expr.
7207 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
7208 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
7211 2017-10-06 Richard Biener <rguenther@suse.de>
7213 PR tree-optimization/82397
7214 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
7215 operand_equal_p but rely on data_ref_compare_tree for detecting
7217 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
7218 to match up with dr_group_sort_cmp.
7220 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7223 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
7225 * config/s390/s390-builtin-types.def: Regenerate.
7227 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7230 * config/s390/s390-builtin-types.def: Regenerate.
7231 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
7232 Change flag from B_VXE to B_VX.
7233 (s390_vec_min_dbl): Remove B_VXE flag.
7235 2017-10-06 Richard Biener <rguenther@suse.de>
7237 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
7238 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
7239 translate_isl_ast_to_gimple::is_valid_rename,
7240 translate_isl_ast_to_gimple::get_rename,
7241 translate_isl_ast_to_gimple::get_def_bb_for_const,
7242 translate_isl_ast_to_gimple::get_new_name,
7243 translate_isl_ast_to_gimple::collect_all_ssa_names,
7244 translate_isl_ast_to_gimple::copy_loop_phi_args,
7245 translate_isl_ast_to_gimple::collect_all_ssa_names,
7246 translate_isl_ast_to_gimple::copy_loop_phi_args,
7247 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
7248 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
7249 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
7250 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
7251 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
7252 translate_isl_ast_to_gimple::copy_cond_phi_args,
7253 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
7254 translate_isl_ast_to_gimple::edge_for_new_close_phis,
7255 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
7256 translate_isl_ast_to_gimple::rename_uses,
7257 translate_isl_ast_to_gimple::rename_all_uses): Remove.
7258 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
7259 (set_rename_for_each_def): Likewise.
7260 (graphite_copy_stmts_from_block): Handle debug stmt resetting
7261 here. Handle rewriting SCEV analyzable uses here.
7262 (copy_bb_and_scalar_dependences): Generate code for PHI
7264 (graphite_regenerate_ast_isl): Adjust.
7265 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
7266 (add_write, add_read): New functions.
7267 (build_cross_bb_scalars_def): Use it and simplify.
7268 (build_cross_bb_scalars_use): Likewise.
7269 (graphite_find_cross_bb_scalar_vars): Inline into...
7270 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
7271 simulating out-of-SSA. Compute liveout and add dependencies.
7272 (build_scops): Force an empty entry block.
7273 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
7275 (sese_build_liveouts): Declare.
7276 (sese_trivially_empty_bb_p): Likewise.
7277 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
7278 compute liveout and debug_liveout.
7279 (sese_bad_liveouts_use): Remove.
7280 (sese_reset_debug_liveouts_bb): Likewise.
7281 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
7282 (sese_build_liveouts): Build liveout and debug_liveout and store
7284 (new_sese_info): Adjust.
7285 (free_sese_info): Likewise.
7286 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
7287 do not build liveout here.
7288 (move_sese_in_condition): Adjust region entry.
7289 (scev_analyzable_p): Match up with chrec_apply requirements.
7290 (sese_trivially_empty_bb_p): New.
7291 * tree-into-ssa.c (get_reaching_def): Properly support generating
7292 default-defs for incremental rewrite of anonymous names.
7294 2017-10-06 Richard Biener <rguenther@suse.de>
7296 * graphite-sese-to-poly.c (extract_affine): For casts increasing
7297 precision do not perform modulo reduction.
7299 2017-10-06 Richard Biener <rguenther@suse.de>
7301 PR tree-optimization/82436
7302 * tree-vect-slp.c (vect_supported_load_permutation_p): More
7303 conservatively choose the vectorization factor when checking
7304 whether we can perform the required load permutation.
7305 (vect_transform_slp_perm_load): Assert when we may not fail.
7307 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
7309 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
7310 message for incompatible -msdata=* and -mcall-* options.
7312 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7314 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
7315 rate for post-reload scheduling.
7317 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7319 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
7321 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7323 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
7324 to improve monte carlo in scimark.
7326 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7328 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
7329 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
7330 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
7331 pentium4_cost, nocona_cost): Set reassociation width to 1.
7332 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
7333 width to 2 for fp operations and 1 otherwise.
7334 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
7336 (atom_cost): Set reassociation width to 2.
7337 (slm_cost, generic_cost): Set fp reassociation width
7338 to 2 and 1 otherwise.
7339 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
7340 (core_cost): Set fp reassociation width to 4 and vector to 2.
7341 (ix86_reassociation_width): Rewrite using cost table; special case
7342 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
7343 and TARGET_AVX128_OPTIMAL.
7344 * config/i386/i386.h (processor_costs): Add
7345 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
7346 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
7347 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
7348 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
7349 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
7350 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
7352 2017-10-05 Nathan Sidwell <nathan@acm.org>
7354 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
7356 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7358 * config/arm/arm.c (arm_test_fpu_data): New.
7359 (arm_run_selftests): Call arm_test_fpu_data.
7361 2017-10-04 Nathan Sidwell <nathan@acm.org>
7363 * toplev.c (toplev::main): Remove excess parens on pretty_printer
7365 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
7367 2017-10-04 Sudakshina Das <sudi.das@arm.com>
7369 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
7370 check type for aarch64_simd_valid_immediate.
7371 (aarch64_output_simd_mov_immediate): Update prototype.
7372 (aarch64_simd_valid_immediate): Update prototype.
7373 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
7374 support for ORR-immediate.
7375 (and<mode>3): modified pattern to add support for BIC-immediate.
7376 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
7377 now checks for valid immediate for BIC and ORR based on new enum
7379 (aarch64_output_simd_mov_immediate): Function now used to output
7380 BIC/ORR imm as well based on new enum argument.
7381 * config/aarch64/constraints.md (Do): New vector immediate constraint.
7383 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
7384 (aarch64_reg_or_bic_imm): Likewise.
7386 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7388 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
7389 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
7391 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7395 PR rtl-optimization/82396
7396 * haifa-sched.c (autopref_multipass_init): Simplify
7398 (autopref_rank_data): Simplify sort order.
7399 * sched-int.h (autopref_multipass_data_): Remove
7400 multi_mem_insn_p, min_offset and max_offset.
7402 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7404 * doc/sourcebuild.texi: Document vect_peeling_profitable.
7406 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7408 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
7411 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7413 * doc/sourcebuild.texi: Document vect_long_mult.
7415 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
7417 PR tree-optimization/82413
7418 * fold-const.c (build_range_check): Use widest_int when comparing
7419 the maximum ETYPE value with HIGH.
7421 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7423 PR rtl-optimization/82396
7424 * haifa-sched.c (autopref_multipass_init): Simplify
7426 (autopref_rank_data): Simplify sort order.
7427 * sched-int.h (autopref_multipass_data_): Remove
7428 multi_mem_insn_p, min_offset and max_offset.
7430 2017-10-04 Jakub Jelinek <jakub@redhat.com>
7432 PR tree-optimization/82381
7433 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
7434 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
7437 PR tree-optimization/82374
7438 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
7439 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
7440 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
7441 current_function_decl to the new decl.
7443 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
7445 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
7446 helper macro for IEEE float128 hardware built-in functions.
7447 (SQRTF128_ODD): Add built-in functions with the round-to-odd
7449 (TRUNCF128_ODD): Likewise.
7450 (ADDF128_ODD): Likewise.
7451 (SUBF128_ODD): Likewise.
7452 (MULF128_ODD): Likewise.
7453 (DIVF128_ODD): Likewise.
7454 (FMAF128_ODD): Likewise.
7455 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
7456 UNSPEC_TRUNC_ROUND_TO_ODD.
7457 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
7458 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
7459 floating point round to odd instructions.
7460 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
7461 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
7462 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
7463 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
7464 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
7465 (trunc<mode>sf2_hw): Change the truncate with round to odd
7466 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
7467 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
7468 to odd hardware instructions.
7469 (sub<mode>3_odd): Likewise.
7470 (mul<mode>3_odd): Likewise.
7471 (div<mode>3_odd): Likewise.
7472 (sqrt<mode>2_odd): Likewise.
7473 (fma<mode>4_odd): Likewise.
7474 (fms<mode>4_odd): Likewise.
7475 (nfma<mode>4_odd): Likewise.
7476 (nfms<mode>4_odd): Likewise.
7477 (trunc<mode>df2_odd): Change the truncate with round to odd
7478 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
7480 * doc/extend.texi (PowerPC built-in functions): Update documentation
7481 for existing IEEE float128-bit built-in functions. Add built-in
7482 functions that generate the IEEE 128-bit floating point round to
7485 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
7487 PR rtl-optimization/77729
7488 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
7489 to (X&(C1&~C2))|C2 transformations.
7491 2017-10-03 Martin Jambor <mjambor@suse.cz>
7493 PR tree-optimization/82363
7494 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
7495 mismatch, mark lacc written regardless of racc.
7497 2017-10-03 Jakub Jelinek <jakub@redhat.com>
7499 PR tree-optimization/82381
7500 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
7501 stmt_to_insert nor wheather SSA_NAMEs are default defs.
7502 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
7503 fallthrough into reassoc_stmt_dominates_stmt_p.
7506 * combine.c (combine_instructions): Don't combine in unreachable
7509 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7512 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
7513 function to not use the have_cpu variable. Do not set cpu_index,
7514 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
7516 (rs6000_valid_attribute_p): Remove duplicate initializations of
7517 old_optimize and func_optimize.
7518 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
7519 (rs6000_activate_target_options): Make global.
7520 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
7523 2017-10-02 Jakub Jelinek <jakub@redhat.com>
7525 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
7526 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
7527 Set *poffset to 0 and *psize and *pmax_size to -1 if
7528 *poffset + *psize overflows in HOST_WIDE_INT.
7530 PR tree-optimization/82387
7531 PR tree-optimization/82388
7532 PR tree-optimization/82389
7533 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
7534 instead of live_bytes non-NULL.
7536 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
7539 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
7542 2017-10-02 Richard Biener <rguenther@suse.de>
7544 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
7545 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
7546 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
7548 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7550 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
7551 requested precision matches the type's.
7552 * calls.c (alloc_max_size): Calculate the new candidate size as
7553 a widest_int and use wi::to_widest when comparing it with the
7554 current candidate size.
7555 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
7556 zero rather than integer_zero_node.
7557 * match.pd: Check for a no-op conversion before using wi::add
7558 rather than after. Use tree_to_uhwi when summing small shift
7559 counts into an unsigned int.
7561 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7562 Alan Hayward <alan.hayward@arm.com>
7563 David Sherwood <david.sherwood@arm.com>
7566 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
7567 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
7568 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
7569 POINTER_AND_FP_REGS.
7571 2017-10-02 Richard Biener <rguenther@suse.de>
7573 PR tree-optimization/82355
7574 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7575 a mapping for the enclosing loop but avoid generating one for
7577 (copy_bb_and_scalar_dependences): Remove premature codegen
7578 error on PHIs in blocks duplicated into multiple places.
7579 * graphite-scop-detection.c
7580 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7581 in the region use it as loop and nest to analyze the DR in.
7582 (try_generate_gimple_bb): Likewise.
7583 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7584 (add_loop_constraints): For blocks in a loop not in the region
7585 create a dimension with a single iteration.
7586 * sese.h (gbb_loop_at_index): Remove assert.
7588 2017-10-01 Kevin Buettner <kevinb@redhat.com>
7590 * omp-expand.c (adjust_context_scope): New function.
7591 (expand_parallel_call): Call adjust_context_scope.
7593 2017-10-01 Jeff Law <law@redhat.com>
7595 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
7596 dom_opt_dom_walker class with direct access to private members.
7597 Add comments. Call test_for_singularity.
7598 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
7599 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
7600 m_dummy_cond anymore.
7601 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
7603 (pass_dominator:execute): Build the dummy_cond here and pass it
7604 to the dom_opt_dom_walker ctor.
7605 (test_for_singularity): New function.
7607 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
7608 Maya Rashish <coypu@sdf.org>
7610 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
7611 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
7612 (alpha*-*-netbsd*) Use nbsd_tm_file.
7613 (arm*-*-netbsdelf*) Likewise.
7614 (i[34567]86-*-netbsdelf*) Likewise.
7615 (x86_64-*-netbsd*) Likewise.
7616 (mips*-*-netbsd*) Likewise.
7617 (powerpc-*-netbsd*) Likewise.
7618 (sh*-*-netbsd*) Likewise.
7619 (sparc-*-netbsdelf*) Likewise.
7620 (sparc64-*-netbsd*) Likewise.
7621 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
7623 (vax-*-netbsdelf*) Likewise.
7624 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
7625 (UINT_FAST8_TYPE) Likewise.
7626 (INT_FAST16_TYPE) Check CHAR_FAST16.
7627 (UINT_FAST16_TYPE) Likewise.
7629 2017-09-30 Jakub Jelinek <jakub@redhat.com>
7632 * config/i386/i386.md
7633 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
7634 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
7635 *divmodsi4_zext_2): New define_insn_and_split.
7636 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
7637 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
7638 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
7639 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
7640 New define_insn_and_split.
7641 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
7642 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
7643 operands[1] having DImode when mode is SImode.
7645 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
7646 always SImode for DIV and MOD in REG_EQUAL notes.
7648 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7651 * match.pd: Fix handling of NaNs in pattern.
7653 2017-09-29 Jeff Law <law@redhat.com>
7655 * sbitmap.c (bitmap_bit_in_range_p): New function.
7656 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7657 * tree-ssa-dse.c (live_bytes_read): New function.
7658 (dse_classify_store): Ignore reads of dead bytes.
7660 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7661 typos and whitespace errors.
7662 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7663 * config/s390/s390.c (s390_emit_prologue): Likewise.
7665 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7668 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7669 with a symbol for LRA.
7671 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7673 PR rtl-optimization/82338
7674 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7676 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7678 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7679 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7680 (qsort_chk): Declare.
7681 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7682 (qsort_chk): New function.
7684 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7686 PR tree-optimization/82337
7687 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7688 phi definition if the PHI result appears in an abnormal PHI.
7689 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7690 basis appears in an abnormal PHI.
7692 2017-09-29 Richard Biener <rguenther@suse.de>
7694 * graphite-isl-ast-to-gimple.c
7695 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7696 (binary_op_to_tree): Use it.
7697 (get_rename_from_scev): Likewise.
7698 (copy_loop_phi_nodes): Likewise.
7699 (copy_bb_and_scalar_dependences): Likewise.
7700 (translate_pending_phi_nodes): Likewise.
7702 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7705 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7706 for movabsq $(i32 << shift), r64.
7708 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7710 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7711 index when encoding %esp as %rsp to avoid 0x67 prefix.
7713 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7715 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7716 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7718 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7720 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7721 Extensions with more than 16 double VFP registers.
7722 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7723 to_clear_mask and all code related to it. Replace the remaining
7724 entry by a sbitmap and adapt code accordingly.
7726 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7728 * brig-builtins.def: Change pure attributes to const.
7730 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7732 * config.gcc (default_gnu_indirect_function): Default to yes for
7733 sparc*-*-linux* with glibc.
7735 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7737 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7738 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7739 when creating .init_array and .fini_array sections with priority
7742 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7745 * config/aarch64/aarch64.c
7746 (aarch64_builtin_support_vector_misalignment): Always return false
7747 when misalignment is unknown.
7749 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7751 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7752 this function to return false if the definition used by the swap
7753 instruction is artificial, or if the memory address from which the
7754 constant value is loaded is not represented by a base address held
7755 in a register or if the base address register is a frame or stack
7756 pointer. Additionally, return false if the base address of the
7757 loaded constant is a SYMBOL_REF but is not considered to be a
7759 (replace_swapped_load_constant): New function.
7760 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7761 loaded constant vector with a load of a swapped constant vector.
7763 2017-09-27 Carl Love <cel@us.ibm.com>
7765 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7766 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7767 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7770 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7772 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7773 first, always call autopref_rank_data otherwise.
7775 2017-09-27 Richard Biener <rguenther@suse.de>
7777 * graphite-scop-detection.c (find_scop_parameters): Move
7778 loop bound handling ...
7779 (gather_bbs::before_dom_children): ... here, avoiding the need
7780 to build scop_info->loop_nest.
7781 (record_loop_in_sese): Remove.
7782 * sese.h (sese_info_t::loop_nest): Remove.
7783 * sese.c (new_sese_info): Do not allocate loop_nest.
7784 (free_sese_info): Do not free loop_nest.
7786 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7789 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7790 lhs from calls if the lhs has addressable type.
7792 2017-09-27 Richard Biener <rguenther@suse.de>
7794 * graphite.h (scop::max_alias_set): New member.
7795 * graphite-scop-detection.c: Remove references to non-existing
7796 --param in comments.
7797 (build_alias_sets): Record the maximum alias set used for drs.
7798 (build_scops): Support zero as unlimited for
7799 --param graphite-max-arrays-per-scop.
7800 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7802 (build_poly_sr_1): ... here. Compute alias set based on the
7803 maximum alias set used for drs rather than
7804 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7806 2017-09-27 Richard Biener <rguenther@suse.de>
7808 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7809 --param loop-block-tile-size=0 to disable tiling.
7811 2017-09-27 Richard Biener <rguenther@suse.de>
7813 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7814 (graphite-max-nb-scop-params): Document special value zero.
7815 * domwalk.h (dom_walker::STOP): New symbolical constant.
7816 (dom_walker::dom_walker): Add optional parameter for bb to
7818 (dom_walker::~dom_walker): Declare.
7819 (dom_walker::before_dom_children): Document STOP return value.
7820 (dom_walker::m_user_bb_to_rpo): New member.
7821 (dom_walker::m_bb_to_rpo): Likewise.
7822 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7823 mapping here if not provided by the user.
7824 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7825 provided by the user.
7826 (dom_walker::STOP): Define.
7827 (dom_walker::walk): Do not compute bb to RPO mapping here.
7828 Support STOP return value from before_dom_children to stop
7830 * graphite-optimize-isl.c (optimize_isl): If the schedule
7831 is the same still generate code if -fgraphite-identity
7832 or -floop-parallelize-all are given.
7833 * graphite-scop-detection.c: Include cfganal.h.
7834 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7836 (gather_bbs::before_dom_children): Return STOP for BBs
7838 (build_scops): Compute bb to RPO mapping and pass it to
7839 the domwalk. Treat --param graphite-max-nb-scop-params=0
7840 as not limiting the number of params.
7841 * graphite.c (graphite_initialize): Remove limit on the
7842 number of basic-blocks in a function.
7843 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7844 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7845 default value of 10.
7847 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7849 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7850 Adjust code to eliminate needing to do the shift right 32-bits
7851 operation after XSCVDPSPN.
7853 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7855 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7856 ((X / Y) != 0 -> X >= Y): Likewise.
7858 2017-09-26 Carl Love <cel@us.ibm.com>
7860 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7861 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7862 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7863 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7864 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7865 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7866 definitions and overloading.
7867 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7868 statement for P9V_BUILTIN_XST_LEN_R.
7869 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7870 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7871 define_expand and define_insn for the instructions and builtins.
7872 * doc/extend.texi: Update the built-in documentation file for the new
7874 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7875 define_insn for the instructions
7877 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7880 * gcc/config/netbsd-protos.h: New file.
7881 * gcc/config/netbsd.c: New file.
7882 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7883 * gcc/config/t-netbsd: New file.
7884 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7885 (tmake_file) Add t-netbsd.
7886 (extra_objs) Add netbsd.o.
7888 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7892 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7894 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7896 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7897 sign extension from a vector register to a GPR by doing a 32-bit
7898 direct move and then an EXTSW.
7899 (extendsi<mode>2 splitter): Likewise.
7900 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7901 right or vector extract after doing XSCVDPSPN. Use
7902 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7904 (movdi_from_sf_zero_ext): Likewise.
7905 (reload_gpr_from_vsxsf): Likewise.
7906 (p8_mfvsrd_4_disf): Delete, no longer used.
7907 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7908 then needing to move the SFmode to a GPR to use the XSCVDPSP
7909 instruction instead of FRSP and XSCVDPSPN.
7910 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7911 it is adjacent to the other XSCVSPDP insns.
7912 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7913 SFmode to be in traditional Altivec registers.
7914 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7915 (vsx_xscvspdpn): Likewise.
7916 (vsx_xscvspdpn_scalar): Likewise.
7918 2017-09-26 Martin Jambor <mjambor@suse.cz>
7920 * tree-sra.c (compare_access_positions): Put integral types first,
7921 stabilize sorting of integral types, remove conditions putting
7922 non-full-precision integers last.
7923 (sort_and_splice_var_accesses): Disable scalarization if a
7924 non-integert would be represented by a non-full-precision integer.
7926 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7928 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7929 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7930 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7931 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7932 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7933 conditionals inside the function instead of around it. Call
7934 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7935 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7937 2017-09-26 Richard Biener <rguenther@suse.de>
7939 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7941 (scop_detection::build_scop_breadth): ... this. Removed.
7942 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7943 (scop_detection::harmful_stmt_in_bb): Likewise.
7944 (scop_detection::graphite_can_represent_stmt): Likewise.
7945 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7946 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7947 (scop_detection::can_represent_loop_1): ... this. Removed.
7948 (scop_detection::harmful_loop_in_region): Simplify after inlining
7949 the above and remove more quadraticness.
7950 (build_scops): Adjust.
7951 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7954 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7957 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7958 REGNO (base) == SP_REG if base is a REG.
7961 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7962 if it is different SSA_NAME.
7963 (optimize_range_tests_cmp_bitwise): New function.
7964 (optimize_range_tests): Call it.
7966 2017-09-26 Richard Biener <rguenther@suse.de>
7968 PR tree-optimization/82321
7969 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7970 for the def being inside the loop.
7972 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7974 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7976 * config/s390/s390-builtins.def: Fix constraint on op4.
7978 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7980 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7981 independent expanders.
7982 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7983 ("vec_ordered", "vec_unordered"): New expanders.
7985 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7987 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7990 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7992 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7993 vec_unpacks_lo_v16qi.
7994 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7996 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7998 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7999 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
8000 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
8002 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8004 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
8006 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
8008 ("*vec_slb<mode>"): New insn pattern.
8009 ("vec_shr_<mode>"): New expander.
8010 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
8011 and force the shift count operand to V16QImode.
8012 ("vec_srb<mode>"): Set shift count mode to V16QI.
8014 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8016 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
8017 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
8018 ("vec_widen_smult_hi_<mode>"): New expander definitions.
8020 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
8023 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
8025 2017-09-26 Richard Biener <rguenther@suse.de>
8027 PR tree-optimization/82320
8028 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
8031 2017-09-25 Jeff Law <law@redhat.com>
8033 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
8034 prototype for new argument.
8035 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
8036 mostly extracted from rs6000_emit_allocate_stack.
8037 (rs6000_emit_probe_stack_range_stack_clash): New function.
8038 (rs6000_emit_allocate_stack): Call
8039 rs6000_emit_probe_stack_range_stack_clash as needed.
8040 (rs6000_emit_probe_stack_range): Add additional argument
8041 to call to gen_probe_stack_range{si,di}.
8042 (output_probe_stack_range): New.
8043 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
8044 (output_probe_stack_range_stack_clash): New.
8045 (rs6000_emit_prologue): Emit notes into dump file as requested.
8046 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
8047 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
8048 Add additional operand and pass it to output_probe_stack_range.
8050 2017-09-25 Bin Cheng <bin.cheng@arm.com>
8052 PR tree-optimization/82163
8053 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
8054 (checking_verify_loop_closed_ssa): New parameter.
8055 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
8056 (check_loop_closed_ssa_stmt): Delete.
8057 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
8058 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
8059 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
8062 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
8064 * brig-builtins.def: Treat HSAIL barrier builtins as
8065 setjmp/longjump style functions.
8067 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
8069 * target.def (constant_alignment): New hook.
8070 * defaults.h (CONSTANT_ALIGNMENT): Delete.
8071 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
8072 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
8073 * doc/tm.texi: Regenerate.
8074 * targhooks.h (default_constant_alignment): Declare.
8075 (constant_alignment_word_strings): Likewise.
8076 * targhooks.c (default_constant_alignment): New function.
8077 (constant_alignment_word_strings): Likewise.
8078 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
8079 instead of CONSTANT_ALIGNMENT.
8080 * varasm.c (align_variable, get_variable_align, build_constant_desc)
8081 (force_const_mem): Likewise.
8082 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
8083 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
8084 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
8085 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8086 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
8088 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
8089 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8090 constant_alignment_word_strings.
8091 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
8092 (CONSTANT_ALIGNMENT): Likewise.
8093 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8094 (arm_constant_alignment): New function.
8095 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
8096 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8097 constant_alignment_word_strings.
8098 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
8099 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8100 constant_alignment_word_strings.
8101 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
8102 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8103 (cris_constant_alignment): New function.
8104 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
8105 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8106 (epiphany_constant_alignment): New function.
8107 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
8108 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8109 constant_alignment_word_strings.
8110 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
8111 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8112 constant_alignment_word_strings.
8113 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
8114 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8115 constant_alignment_word_strings.
8116 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
8117 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
8118 * config/i386/i386.c (ix86_constant_alignment): Make static.
8119 Use the same interface as the target hook.
8120 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8121 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
8122 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8123 constant_alignment_word_strings.
8124 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
8125 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
8126 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8127 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
8128 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8129 constant_alignment_word_strings.
8130 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
8131 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8132 constant_alignment_word_strings.
8133 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
8134 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8135 constant_alignment_word_strings.
8136 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
8137 * config/microblaze/microblaze.c (microblaze_constant_alignment):
8139 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8140 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
8141 * config/mips/mips.c (mips_constant_alignment): New function.
8142 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8143 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
8144 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
8145 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8146 (mmix_constant_alignment): Make static. Use the same interface
8148 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
8149 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8150 constant_alignment_word_strings.
8151 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
8152 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8153 constant_alignment_word_strings.
8154 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
8155 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8156 constant_alignment_word_strings.
8157 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
8158 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8159 (rs6000_constant_alignment): New function.
8160 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
8161 * config/riscv/riscv.c (riscv_constant_alignment): New function.
8162 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8163 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
8164 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8165 (rs6000_constant_alignment): New function.
8166 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
8167 * config/s390/s390.c (s390_constant_alignment): New function.
8168 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8169 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
8170 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8171 constant_alignment_word_strings.
8172 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
8173 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8174 (sparc_constant_alignment): New function.
8175 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
8176 * config/spu/spu.c (spu_constant_alignment): New function.
8177 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8178 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
8179 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8180 constant_alignment_word_strings.
8181 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
8182 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8183 constant_alignment_word_strings.
8184 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
8185 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
8186 constant_alignment_word_strings.
8187 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
8188 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8189 (visium_constant_alignment): New function.
8190 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
8191 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
8192 (xtensa_constant_alignment): New function.
8193 * system.h (CONSTANT_ALIGNMENT): Poison.
8195 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
8197 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8198 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
8199 (rs6000_builtin_valid_without_lhs): New helper function.
8200 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8201 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
8203 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
8205 * target.h (vec_perm_indices): Use unsigned short rather than
8207 (auto_vec_perm_indices): Likewise.
8208 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
8209 Use unsigned int rather than unsigned char.
8210 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
8212 2017-09-25 Richard Biener <rguenther@suse.de>
8214 * cfgloop.h (sort_sibling_loops): Declare.
8215 * cfgloop.c (sort_sibling_loops_cmp): New helper.
8216 (sort_sibling_loops): New function sorting the sibling loop list
8218 * graphite.c (graphite_transform_loops): Sort sibling loops.
8220 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
8222 * target.def (vec_perm_const_ok): Change sel parameter to
8224 * optabs-query.c (can_vec_perm_p): Update accordingly.
8225 * doc/tm.texi: Regenerate.
8226 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
8227 auto_vec_perm_indices and remove separate nelt field.
8228 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
8229 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
8230 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
8231 (aarch64_expand_vec_perm_const): Update accordingly.
8232 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
8233 to vec_perm_indices.
8234 * config/arm/arm.c (expand_vec_perm_d): Change perm to
8235 auto_vec_perm_indices and remove separate nelt field.
8236 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
8237 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
8238 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
8240 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
8241 to vec_perm_indices.
8242 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
8243 sel to vec_perm_indices.
8244 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
8245 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
8246 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
8248 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
8251 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
8254 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
8255 on the FUNCTION_DECL function context if it has a DIE that is a
8258 2017-09-25 Richard Biener <rguenther@suse.de>
8260 PR tree-optimization/82285
8261 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
8264 2017-09-25 Tom de Vries <tom@codesourcery.com>
8268 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
8271 2017-09-25 Richard Biener <rguenther@suse.de>
8273 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
8274 ISL errors other than isl_error_quota happen. Dump if the
8275 schedule is the same.
8276 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
8277 errors instead of aborting inside ISL.
8279 2017-09-25 Iain Sandoe <iain@codesourcery.com>
8282 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
8283 of libgcc_eh for m64.
8284 * config/i386/darwin64.h: Likewise.
8286 2017-09-25 Richard Biener <rguenther@suse.de>
8289 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
8290 attribute for incomplete types nor twice for complete ones.
8292 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
8295 * config/i386/i386.c (ix86_print_operand_address_as): Encode
8296 %esp as %rsp to avoid 0x67 prefix if there is no index or base
8299 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
8302 * config/i386/i386.opt (mprefer-avx256): Use
8303 ix86_target_flags variable.
8304 * config/i386/i386.c (ix86_target_string): Move
8305 -mprefer-avx256 to flag2_opts.
8307 2017-09-22 Jakub Jelinek <jakub@redhat.com>
8310 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
8311 and x != -1 | y != -1 into (x & y) != -1.
8313 2017-09-22 Steve Ellcey <sellcey@cavium.com>
8315 * config.gcc: Add new case statement to set
8316 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
8317 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
8318 s390x-*-linux* case statements. Added aarch64 to the list of
8319 supported architectures.
8321 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8323 PR tree-optimization/82289
8324 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
8325 STMT_VINFO_RELEVANT_P.
8327 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8328 Alan Hayward <alan.hayward@arm.com>
8329 David Sherwood <david.sherwood@arm.com>
8331 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
8332 for VR_RANGE only; don't allow VR_ANTI_RANGE.
8333 (extract_range_from_binary_expr_1): Don't call
8334 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
8336 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8337 Alan Hayward <alan.hayward@arm.com>
8338 David Sherwood <david.sherwood@arm.com>
8340 * target.def (preferred_vector_alignment): New hook.
8341 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
8343 * doc/tm.texi: Regenerate.
8344 * targhooks.h (default_preferred_vector_alignment): Declare.
8345 * targhooks.c (default_preferred_vector_alignment): New function.
8346 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
8348 (DR_TARGET_ALIGNMENT): New macro.
8349 (aligned_access_p): Update commentary.
8350 (vect_known_alignment_in_bytes): New function.
8351 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
8353 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
8354 Calculate the misalignment based on the target alignment rather than
8356 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
8357 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
8358 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
8359 the target alignment, rather than masking the element misalignment
8360 with the number of elements in a vector. Also use the target
8361 alignment when calculating the maximum number of peels.
8362 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
8363 instead of TYPE_ALIGN_UNIT.
8364 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
8365 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
8366 (vect_create_addr_base_for_vector_ref): Update call accordingly.
8367 (vect_create_data_ref_ptr): Likewise.
8368 (vect_setup_realignment): Realign by ANDing with
8369 -DR_TARGET_MISALIGNMENT.
8370 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
8371 the number of peels based on DR_TARGET_ALIGNMENT.
8372 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
8373 with the guaranteed alignment boundary when deciding whether
8375 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
8376 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
8377 (ensure_base_align): Remove stmt_info parameter. Get the
8378 target base alignment from DR_TARGET_ALIGNMENT.
8379 (vectorizable_store): Update call accordingly. Interpret
8380 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
8382 (vectorizable_load): Likewise.
8384 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8385 Alan Hayward <alan.hayward@arm.com>
8386 David Sherwood <david.sherwood@arm.com>
8388 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
8389 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
8390 (vect_enhance_data_refs_alignment): Likewise.
8392 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8394 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8395 error. Only quit immediately if parsing is complete.
8396 (BEGIN): Initialize fatal_err and parse_done.
8397 (begin fpu, end fpu): Check number of arguments.
8398 (begin arch, end arch): Likewise.
8399 (begin cpu, end cpu): Likewise.
8400 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8401 (optalias): Likewise.
8403 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8405 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8406 * config/arm/arm-isa.h: Delete. Move definitions to ...
8407 * arm-cpus.in: ... here. Use new feature and fgroup values.
8408 * config/arm/arm.c (arm_option_override): Use lower case for feature
8410 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8411 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8412 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8414 (print_isa_bits_for): New function.
8415 (gen_isa): New function.
8416 (gen_comm_data): Use print_isa_bits_for.
8417 (define feature): New keyword.
8418 (define fgroup): New keyword.
8419 * config/arm/t-arm (TM_H): Remove.
8420 (GTM_H): Add arm-isa.h.
8421 (arm-isa.h): Add rule to generate file.
8422 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8423 case for feature bit names.
8425 2017-09-22 Richard Biener <rguenther@suse.de>
8427 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
8429 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
8430 print of no dependency loops ...
8431 * graphite.c (graphite_transform_loops): ... here.
8432 (canonicalize_loop_closed_ssa_form): Work from inner to outer
8434 (same_close_phi_node, remove_duplicate_close_phi,
8435 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
8436 (canonicalize_loop_closed_ssa): ... here and simplify.
8437 * graphite-optimize-isl.c: Include tree-vectorizer.h.
8438 (optimize_isl): Use dump_printf_loc to tell when we stopped
8439 optimizing because of an ISL timeout.
8441 2017-09-22 Richard Biener <rguenther@suse.de>
8443 PR tree-optimization/82291
8444 * tree-if-conv.c (predicate_mem_writes): Make sure to
8445 remove writes in blocks predicated with false.
8447 2017-09-22 Richard Biener <rguenther@suse.de>
8449 * sese.c: Include cfganal.h.
8450 (if_region_set_false_region): Remove.
8451 (create_if_region_on_edge): Likewise.
8452 (move_sese_in_condition): Re-implement without destroying
8455 2017-09-22 Richard Biener <rguenther@suse.de>
8457 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
8458 Verify both BBs contain loop PHI nodes before dispatching to
8460 (graphite_regenerate_ast_isl): Do not recompute dominators,
8461 do not verify three times. Restructure for clarity.
8462 * graphite-scop-detection.c (same_close_phi_node,
8463 remove_duplicate_close_phi, make_close_phi_nodes_unique,
8464 defined_in_loop_p, canonicalize_loop_closed_ssa,
8465 canonicalize_loop_closed_ssa_form): Simplify, remove excess
8466 checking and SSA rewrite, move to ...
8467 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
8468 (graphite_initialize): Do not pass in ctx, do not reset the
8469 SCEV cache, compute only dominators.
8470 (graphite_transform_loops): Allocate ISL ctx after
8471 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
8472 Maintain post-dominators only around build_scops.
8473 * sese.c (if_region_set_false_region): Make static. Free
8474 and recompute dominators.
8475 (move_sese_in_condition): Assert we don't get called with
8476 post-dominators computed.
8477 * sese.h (if_region_set_false_region): Remove.
8479 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
8481 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
8482 mode attribute for TARGET_AVX512VL.
8484 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
8486 * config/i386/i386.opt (mprefer-avx256): New option.
8487 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
8489 (ix86_preferred_simd_mode): Return 256-bit AVX modes
8490 for TARGET_PREFER_AVX256.
8491 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
8493 2017-09-21 Jeff Law <law@redhat.com>
8495 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
8496 Fix dump output if the only stack space is for pushed registers.
8498 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8500 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
8503 2017-09-21 Martin Sebor <msebor@redhat.com>
8506 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
8507 code (in C++) or code that triggers warnings.
8509 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
8511 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
8513 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
8515 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
8516 * sched-rgn.c: Ditto.
8517 * sel-sched-ir.c: Ditto.
8519 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
8521 * toplev.h (set_random_seed): Adjust return type.
8522 * toplev.c (init_local_tick): Move eager initialization of random_seed
8523 to get_random_seed. Adjust comment.
8524 (init_random_seed): Inline to get_random_seed, delete.
8525 (get_random_seed): Initialize random_seed lazily.
8526 (set_random_seed): Do not return previous value.
8527 (print_switch_value): Do not call get_random_seed.
8529 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
8531 * cgraph.c (delete_function_version): New, broken out from...
8532 (cgraph_node::delete_function_version): ...here. Rename to
8533 cgraph_node::delete_function_version_by_decl. Update all uses.
8534 (cgraph_node::remove): Call delete_function_version.
8536 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8539 * tree-inline.c (expand_call_inline): Emit clobber stmts for
8540 VAR_DECLs to which addressable non-volatile parameters are mapped
8541 and for id->retvar after the return value assignment. Clear
8542 id->retval and id->retbnd after inlining.
8544 2017-09-21 Richard Biener <rguenther@suse.de>
8546 PR tree-optimization/82276
8547 PR tree-optimization/82244
8548 * tree-vrp.c (build_assert_expr_for): Set
8549 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
8551 (remove_range_assertions): Revert earlier change.
8553 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
8556 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
8558 2017-09-21 Richard Biener <rguenther@suse.de>
8560 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8561 Restore valid IL after code generation errors.
8562 * graphite.c (graphite_transform_loops): Diagnose code
8563 generation issues as MSG_MISSED_OPTIMIZATION and continue
8564 with processing SCOPs.
8566 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8567 Alan Hayward <alan.hayward@arm.com>
8568 David Sherwood <david.sherwood@arm.com>
8570 * calls.c (compute_argument_addresses): Use simplify_gen_binary
8571 rather than choosing between plus_constant and gen_rtx_<CODE>.
8572 * expr.c (emit_push_insn): Likewise.
8573 (expand_expr_real_2): Likewise.
8575 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8576 Alan Hayward <alan.hayward@arm.com>
8577 David Sherwood <david.sherwood@arm.com>
8579 * loop-unroll.c (split_iv): Call copy_rtx on the step.
8581 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8582 Alan Hayward <alan.hayward@arm.com>
8583 David Sherwood <david.sherwood@arm.com>
8585 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
8586 calling tree_to_uhwi.
8588 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8589 Alan Hayward <alan.hayward@arm.com>
8590 David Sherwood <david.sherwood@arm.com>
8592 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
8593 INTEGER_CST rather than a negative test for ADDR_EXPR.
8595 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8596 Alan Hayward <alan.hayward@arm.com>
8597 David Sherwood <david.sherwood@arm.com>
8599 * tree-vrp.c (extract_range_from_binary_expr_1): Check
8600 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
8602 2017-09-21 Richard Biener <rguenther@suse.de>
8604 PR tree-optimization/71351
8605 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
8606 graphite_create_new_loop_guard): Remove, fold remaining parts
8608 (translate_isl_ast_node_for): ... here and simplify.
8610 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8613 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
8614 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
8615 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
8616 alternative always use QI mode, for -Os imov (=R,R) alternative
8617 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
8620 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8621 Jeff Law <law@redhat.com>
8623 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
8624 (allocate_stack_space): New function, partially extracted from
8626 (s390_emit_prologue): Track offset to most recent stack probe.
8627 Code to allocate space moved into allocate_stack_space.
8628 Dump actions when no stack is allocated.
8629 (s390_prologue_plus_offset): New function.
8630 (s390_emit_stack_probe): Likewise.
8632 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
8634 * common.opt (Wa, Wl, Wp, g, gz=): Add
8636 (gno-column-info): Remove.
8637 (gcolumn-info): Drop RejectNegative.
8639 (gno-record-gcc-switches): Remove.
8640 (grecord-gcc-switches): Drop RejectNegative.
8641 (gno-split-dwarf): Remove.
8642 (gsplit-dwarf): Drop RejectNegative.
8643 (gno-strict-dwarf): Remove.
8644 (gstrict-dwarf): Drop RejectNegative.
8645 * config/darwin.opt (gfull, gused): Add RejectNegative.
8646 * dwarf2out.c (gen_producer_string): Drop
8647 gno-record-gcc-switches handler.
8648 * optc-gen.awk: Add g to prefixes with negative forms.
8649 * opts-common.c (remapping_prefix_p): New.
8650 (find_opt): Check it.
8651 (generate_canonical_option): Test g prefix.
8652 (option_map): Add -gno- mapping.
8653 (add_misspelling_candidates): Check remapping_prefix_p.
8655 2017-09-20 Jeff Law <law@redhat.com>
8657 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8658 thinko in stack clash protection support.
8660 * explow.c (compute_stack_clash_protection_loop_data): Use
8661 CONST_INT_P instead of explicit test. Verify object is a
8662 CONST_INT_P before looking at INTVAL.
8663 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8664 instead of explicit test.
8666 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8669 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8670 address instead of to r1 and r11.
8672 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8674 * config.gcc: Support "knm".
8675 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8676 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8678 * config/i386/i386.c (m_KNM): Define.
8679 (processor_target_table): Add "knm".
8681 (ix86_option_override_internal): Add "knm".
8682 (ix86_issue_rate): Add PROCESSOR_KNM.
8683 (ix86_adjust_cost): Ditto.
8684 (ia32_multipass_dfa_lookahead): Ditto.
8685 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8686 (fold_builtin_cpu): Add M_INTEL_KNM.
8687 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8688 (processor_type): Add PROCESSOR_KNM.
8689 * config/i386/x86-tune.def: Add m_KNM.
8690 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8692 2017-09-20 Richard Biener <rguenther@suse.de>
8694 PR tree-optimization/80213
8695 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8696 are allowed in empty BBs as well.
8697 (canonicalize_loop_closed_ssa): Also look for other complex
8699 (scop_detection::get_sese): Include the loop-closed PHI block
8701 (scop_detection::merge_sese): Remove code adding extra blocks.
8702 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8703 (build_scops): Assert the final returned scop is invalid.
8705 2017-09-20 Richard Biener <rguenther@suse.de>
8707 PR tree-optimization/82264
8708 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8710 (vn_phi_lookup): Likewise.
8711 (vn_phi_insert): Likewise.
8713 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8715 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8716 that fits into uhwi or shwi, add DW_AT_const_value regardless
8717 of early_dwarf without going through RTL, using add_AT_unsigned
8720 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8721 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8722 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8723 (DEBUG_MACRO_SECTION): Likewise.
8724 (DEBUG_LTO_MACRO_SECTION): Likewise.
8725 (DEBUG_STR_DWO_SECTION): Likewise.
8726 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8727 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8728 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8729 (DEBUG_LTO_LINE_STR_SECTION): Define.
8730 (init_sections_and_labels): Initialize debug_line_str_section
8731 variable. Initialize debug_loc_section for -gdwarf-5 to
8732 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8734 2017-09-20 Richard Biener <rguenther@suse.de>
8736 * graphite-sese-to-poly.c (extract_affine): Properly handle
8737 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8739 2017-09-20 Richard Biener <rguenther@suse.de>
8741 PR tree-optimization/81373
8742 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8743 Force SESE live-out defs to be handled even if they are
8746 2017-09-19 Jeff Law <law@redhat.com>
8748 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8749 nothing for stack adjustments with REG_STACK_CHECK.
8750 * sched-deps.c (parse_add_or_inc): Reject insns with
8751 REG_STACK_CHECK from dependency breaking.
8752 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8753 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8754 * reg-notes.def (STACK_CHECK): New note.
8756 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8757 (ix86_expand_prologue): Dump stack clash info as needed.
8758 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8760 * function.c (dump_stack_clash_frame_info): New function.
8761 * function.h (dump_stack_clash_frame_info): Prototype.
8762 (enum stack_clash_probes): New enum.
8764 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8765 flag_stack_clash_protection.
8766 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8767 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8768 (arm_frame_pointer_required): Likewise.
8769 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8770 (ia64_expand_prologue): Likewise.
8771 * config/mips/mips.c (mips_expand_prologue): Likewise.
8772 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8773 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8774 (sparc_flat_expand_prologue): Likewise.
8775 * config/spu/spu.c (spu_expand_prologue): Likewise.
8777 * explow.c: Include "params.h".
8778 (anti_adjust_stack_and_probe_stack_clash): New function.
8779 (get_stack_check_protect): Likewise.
8780 (compute_stack_clash_protection_loop_data): Likewise.
8781 (emit_stack_clash_protection_loop_start): Likewise.
8782 (emit_stack_clash_protection_loop_end): Likewise.
8783 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8784 Use anti_adjust_stack_and_probe_stack_clash.
8785 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8786 (emit_stack_clash_protection_loop_start): Likewise.
8787 (emit_stack_clash_protection_loop_end): Likewise.
8788 * rtl.h (get_stack_check_protect): Prototype.
8789 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8790 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8791 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8793 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8795 * doc/tm.texi: Rebuilt.
8796 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8797 get_stack_check_protect.
8798 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8799 * config/arm/arm.c (arm_expand_prologue): Likewise.
8800 (arm_frame_pointer_required): Likewise.
8801 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8802 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8803 * config/mips/mips.c (mips_expand_prologue): Likewise.
8804 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8805 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8806 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8807 (sparc_flat_expand_prologue): Likewise.
8809 * common.opt (-fstack-clash-protection): New option.
8810 * flag-types.h (enum stack_check_type): Note difference between
8811 -fstack-check= and -fstack-clash-protection.
8812 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8813 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8814 * toplev.c (process_options): Issue warnings/errors for cases
8815 not handled with -fstack-clash-protection.
8816 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8817 (-fstack-check): Note additional problem with -fstack-check=generic.
8818 Note that -fstack-check is primarily for Ada and refer users
8819 to -fstack-clash-protection for stack-clash-protection.
8820 Document new params for stack clash protection.
8822 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8824 * config/i386/i386.c (ix86_split_long_move): Do not handle
8825 address used for LEA in a special way.
8827 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8829 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8831 2017-09-19 Martin Sebor <msebor@redhat.com>
8834 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8835 of incompatible types.
8837 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8839 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8840 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8841 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8842 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8844 2017-09-19 Richard Biener <rguenther@suse.de>
8846 PR tree-optimization/82244
8847 * tree-vrp.c (remove_range_assertions): Do not propagate
8848 a constant to abnormals but replace the assert with a copy.
8850 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8852 PR rtl-optimization/57878
8853 PR rtl-optimization/68988
8854 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8855 avoidance test involving non_reload_pseudos. Move frequency test
8856 below the general fragmentation avoidance test.
8858 2017-09-19 Richard Biener <rguenther@suse.de>
8860 PR tree-optimization/69728
8861 * graphite-sese-to-poly.c (schedule_error): New global.
8862 (add_loop_schedule): Handle empty domain by failing the
8864 (build_original_schedule): Handle schedule_error.
8866 2017-09-19 Richard Biener <rguenther@suse.de>
8868 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8869 Do not iterate to sibling loops but only to siblings of inner
8872 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8875 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8876 that the registers are different.
8878 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8880 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8881 to processor_model and "amdfam17h" to arch_names_table.
8882 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8884 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8887 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8889 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8890 Alan Hayward <alan.hayward@arm.com>
8891 David Sherwood <david.sherwood@arm.com>
8893 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8895 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8897 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8898 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8899 vect_schedule_slp_instance.
8900 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8901 Update call to vect_slp_analyze_node_operations. Simplify return
8903 (vect_slp_analyze_bb_1): Update call accordingly.
8904 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8905 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8906 (vect_schedule_slp): Update call accordingly.
8908 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8909 Alan Hayward <alan.hayward@arm.com>
8910 David Sherwood <david.sherwood@arm.com>
8912 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8913 with types that aren't in fact scalar.
8915 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8917 * tree-vect-slp.c (vect_record_max_nunits): New function,
8919 (vect_build_slp_tree_1): ...here.
8920 (vect_build_slp_tree_2): Call it for phis too.
8922 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8924 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8925 to vect_get_vec_def_for_operand when getting the mask operand.
8927 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8928 Alan Hayward <alan.hayward@arm.com>
8929 David Sherwood <david.sherwood@arm.com>
8931 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8934 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8935 Alan Hayward <alan.hayward@arm.com>
8936 David Sherwood <david.sherwood@arm.com>
8938 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8939 calculation for vector booleans.
8941 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8942 Alan Hayward <alan.hayward@arm.com>
8943 David Sherwood <david.sherwood@arm.com>
8945 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8947 (vect_transform_stmt): ...here.
8948 (vect_analyze_stmt): Use it instead of calling
8949 vectorizable_live_operation directly.
8951 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8953 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8954 non-SIMT targets in acc vector loops.
8956 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8958 * configure.ac: Add arc and check if assembler supports gdwarf2.
8959 * configure: Regenerate.
8961 2017-09-18 Richard Biener <rguenther@suse.de>
8963 PR tree-optimization/82220
8964 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8965 epilogue niters from the min_profitable_iters compute.
8967 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8970 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8971 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8972 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8974 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8977 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8978 switching to a new text section.
8980 2017-09-18 Richard Biener <rguenther@suse.de>
8982 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8984 (build_alias_set): Reject aliases with no access function.
8986 2017-09-18 Richard Biener <rguenther@suse.de>
8988 PR tree-optimization/79622
8989 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8991 (build_cross_bb_scalars_use): Likewise.
8993 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8995 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8997 2017-09-18 Alan Modra <amodra@gmail.com>
9000 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
9001 stack_pointer_rtx for count 0. Update comments. Break up
9002 large rtl expression.
9004 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
9006 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
9007 Increase to 20 bytes.
9008 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
9009 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
9010 or avx version of the stub.
9012 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
9015 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
9016 compute the minimum stack alignment. Also update preferred stack
9017 boundary for leaf functions.
9019 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
9021 PR tree-optimization/82228
9022 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
9025 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
9027 * common/config/nds32/nds32-common.c
9028 (nds32_option_optimization_table): Refine formatting.
9029 (nds32_option_optimization_table): Use -fsched-pressure and
9030 -fomit-frame-pointer for specific optimization level.
9032 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
9034 * config/nds32/nds32.c: Refine formatting and comments.
9035 * config/nds32/nds32.h: Likewise.
9036 * config/nds32/nds32.md: Likewise.
9037 * config/nds32/nds32-cost.c: Likewise.
9038 * config/nds32/nds32-isr.c: Likewise.
9039 * config/nds32/nds32-md-auxiliary.c: Likewise.
9040 * config/nds32/nds32-multiple.md: Likewise.
9041 * config/nds32/nds32-predicates.c: Likewise.
9043 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
9044 Jakub Jelinek <jakub@redhat.com>
9046 Add support for -std=c++2a.
9047 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
9049 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
9051 2017-09-15 Steve Ellcey <sellcey@cavium.com>
9054 * doc/extend.texi (Common Function Attributes): Add
9055 references to ARM, AArch64, and S/390 specific attributes.
9056 (Function Specific Option Pragmas): Add AArch64 and S/390
9057 to list of back ends that support the target pragma.
9059 2017-09-15 Nathan Sidwell <nathan@acm.org>
9061 * doc/standards.texi: Fix C++17 description. Update URLs for
9064 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
9066 * common.opt (Wcast-align=strict): New warning option.
9067 * doc/invoke.texi: Document -Wcast-align=strict.
9069 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
9071 * cgraph.h (cgraph_thunk_info): Add comments.
9072 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
9073 assert for VIRTUAL_* arguments stricter.
9075 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
9077 PR tree-optimization/71026
9078 * match.pd: Move RDIV patterns from fold-const.c
9079 * fold-const.c (distribute_real_division): Removed.
9080 (fold_binary_loc): Remove calls to distribute_real_divison.
9082 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9084 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
9085 c++1z and gnu++1z as deprecated. Change other references to
9086 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
9087 Change -Wc++1z-compat to -Wc++17-compat.
9088 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
9089 * dwarf2out.c (highest_c_language): Handle C++17.
9090 (gen_compile_unit_die): Likewise.
9092 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9094 PR rtl-optimization/82192
9095 * combine.c (make_extraction): Don't look through non-paradoxical
9096 SUBREGs or TRUNCATE if pos + len is or might be bigger than
9099 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9100 Alan Hayward <alan.hayward@arm.com>
9101 David Sherwood <david.sherwood@arm.com>
9103 * target.def (function_arg_offset): New hook.
9104 * targhooks.h (default_function_arg_offset): Declare.
9105 * targhooks.c (default_function_arg_offset): New function.
9106 * function.c (locate_and_pad_parm): Use
9107 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
9108 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
9109 (TARGET_FUNCTION_ARG_OFFSET): ...this.
9110 * doc/tm.texi: Regenerate.
9111 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
9112 * config/spu/spu.c (spu_function_arg_offset): New function.
9113 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
9114 * system.h (FUNCTION_ARG_OFFSET): Poison.
9116 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9117 Alan Hayard <alan.hayward@arm.com>
9118 David Sherwood <david.sherwood@arm.com>
9120 * target.def (truly_noop_truncation): New hook.
9121 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
9122 than TRULY_NOOP_TRUNCATION.
9123 * hooks.h (hook_bool_uint_uint_true): Declare.
9124 * hooks.c (hook_bool_uint_uint_true): New function.
9125 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
9126 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
9127 * doc/tm.texi: Regenerate.
9128 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
9129 rather than TRULY_NOOP_TRUNCATION in comments.
9130 (simplify_comparison): Likewise.
9131 (record_truncated_value): Likewise.
9132 * expmed.c (extract_bit_field_1): Likewise.
9133 (extract_split_bit_field): Likewise.
9134 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
9135 instead of TRULY_NOOP_TRUNCATION.
9136 * function.c (assign_parm_setup_block): Likewise.
9137 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
9138 * rtlhooks.c: Include target.h.
9139 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
9140 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
9141 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
9142 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
9143 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
9144 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
9145 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
9146 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
9147 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
9148 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
9149 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
9150 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
9151 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
9152 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
9153 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
9154 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
9155 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
9156 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
9157 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
9158 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
9159 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
9160 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
9161 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
9162 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
9163 * config/mips/mips.c (mips_truly_noop_truncation): New function.
9164 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
9165 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
9166 rather than TRULY_NOOP_TRUNCATION in comments.
9167 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
9168 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
9169 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
9170 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
9171 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
9172 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
9173 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
9174 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
9175 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
9176 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
9177 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
9178 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
9179 rather than TRULY_NOOP_TRUNCATION in comments.
9180 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
9181 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
9182 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
9183 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
9184 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
9185 TRULY_NOOP_TRUNCATION condition.
9186 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
9187 (TRULY_NOOP_TRUNCATION): Delete.
9188 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
9189 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
9190 * config/spu/spu.c (spu_truly_noop_truncation): New function.
9191 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
9192 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
9193 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
9194 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
9195 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
9196 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
9197 rather than TRULY_NOOP_TRUNCATION in comments.
9198 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
9199 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
9200 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
9201 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
9202 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
9203 * system.h (TRULY_NOOP_TRUNCATION): Poison.
9205 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
9208 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
9209 (*cmp_ior): Likewise.
9210 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
9211 (*ior_scc_scc_cmp): Likewise.
9212 (*and_scc_scc): Likewise.
9213 (*and_scc_scc_cmp): Likewise.
9215 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
9216 Alan Hayard <alan.hayward@arm.com>
9217 David Sherwood <david.sherwood@arm.com>
9219 * target.def (can_change_mode_class): New hook.
9220 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
9221 (hard_regno_nregs): Likewise.
9222 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
9223 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
9224 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
9225 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
9226 (LOAD_EXTEND_OP): Update accordingly.
9227 * doc/tm.texi: Regenerate.
9228 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
9229 CANNOT_CHANGE_MODE_CLASS.
9230 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
9231 (REG_CAN_CHANGE_MODE_P): ...this new macro.
9232 * combine.c (simplify_set): Update accordingly.
9233 * emit-rtl.c (validate_subreg): Likewise.
9234 * recog.c (general_operand): Likewise.
9235 * regcprop.c (mode_change_ok): Likewise.
9236 * reload1.c (choose_reload_regs): Likewise.
9237 (inherit_piecemeal_p): Likewise.
9238 * rtlanal.c (simplify_subreg_regno): Likewise.
9239 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
9240 instead of CANNOT_CHANGE_MODE_CLASS.
9241 (reload_cse_simplify_operands): Likewise.
9242 * reload.c (push_reload): Use targetm.can_change_mode_class
9243 instead of CANNOT_CHANGE_MODE_CLASS.
9244 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
9245 REG_CANNOT_CHANGE_MODE_P.
9246 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9247 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
9248 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9249 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9250 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9251 (arm_can_change_mode_class): New function.
9252 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
9253 than CANNOT_CHANGE_MODE_CLASS in comments.
9254 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9255 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
9256 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
9257 (ix86_can_change_mode_class): ...this new function, inverting the
9258 sense of the return value.
9259 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9260 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9261 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9262 (ia64_can_change_mode_class): New function.
9263 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9264 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
9265 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
9266 (m32c_can_change_mode_class): ...this new function, inverting the
9267 sense of the return value.
9268 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9269 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9270 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
9271 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
9272 (mips_can_change_mode_class): ...this new function, inverting the
9273 sense of the return value.
9274 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9275 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9276 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9277 (msp430_can_change_mode_class): New function.
9278 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9279 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
9280 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9281 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9282 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9283 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
9284 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9285 (pa_cannot_change_mode_class): Replace with...
9286 (pa_can_change_mode_class): ...this new function, inverting the
9287 sense of the return value.
9288 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
9289 than CANNOT_CHANGE_MODE_CLASS in comments.
9290 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9291 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
9292 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9293 (pdp11_cannot_change_mode_class): Replace with...
9294 (pdp11_can_change_mode_class): ...this new function, inverting the
9295 sense of the return value.
9296 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9297 * config/powerpcspe/powerpcspe-protos.h
9298 (rs6000_cannot_change_mode_class_ptr): Delete.
9299 * config/powerpcspe/powerpcspe.c
9300 (rs6000_cannot_change_mode_class_ptr): Delete.
9301 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9302 (rs6000_option_override_internal): Assign to
9303 targetm.can_change_mode_class instead of
9304 rs6000_cannot_change_mode_class_ptr.
9305 (rs6000_cannot_change_mode_class): Replace with...
9306 (rs6000_can_change_mode_class): ...this new function, inverting the
9307 sense of the return value.
9308 (rs6000_debug_cannot_change_mode_class): Replace with...
9309 (rs6000_debug_can_change_mode_class): ...this new function.
9310 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9311 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
9312 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9313 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9314 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
9316 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
9317 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9318 (rs6000_option_override_internal): Assign to
9319 targetm.can_change_mode_class instead of
9320 rs6000_cannot_change_mode_class_ptr.
9321 (rs6000_cannot_change_mode_class): Replace with...
9322 (rs6000_can_change_mode_class): ...this new function, inverting the
9323 sense of the return value.
9324 (rs6000_debug_cannot_change_mode_class): Replace with...
9325 (rs6000_debug_can_change_mode_class): ...this new function.
9326 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9327 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
9328 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
9329 (s390_can_change_mode_class): ...this new function, inverting the
9330 sense of the return value.
9331 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9332 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9333 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
9334 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9335 (sh_cannot_change_mode_class): Replace with...
9336 (sh_can_change_mode_class): ...this new function, inverting the
9337 sense of the return value.
9338 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9339 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9340 (sparc_can_change_mode_class): New function.
9341 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9342 * config/spu/spu.c (spu_can_change_mode_class): New function.
9343 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9344 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9345 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9346 (visium_can_change_mode_class): New function.
9347 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
9349 2017-09-15 Richard Biener <rguenther@suse.de>
9351 PR tree-optimization/82217
9352 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
9353 but not undefined case.
9355 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9358 * postreload.c (reload_cse_simplify_operands): Skip
9359 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
9361 2017-09-15 Richard Biener <rguenther@suse.de>
9363 PR tree-optimization/68823
9364 * graphite-scop-detection.c (build_alias_set): If we have a
9365 possible dependence check whether we can handle them by just
9366 looking at the DRs DR_ACCESS_FNs.
9367 (build_scops): If build_alias_set fails, fail the SCOP.
9369 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
9371 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
9372 to support float128 built-in functions that require the ISA 3.0
9374 (BU_FLOAT128_3_HW): Likewise.
9375 (SQRTF128): Add support for the IEEE 128-bit square root and fma
9377 (FMAF128): Likewise.
9379 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9380 support for built-in functions that need the ISA 3.0 IEEE 128-bit
9381 floating point instructions.
9382 (rs6000_invalid_builtin): Likewise.
9383 (rs6000_builtin_mask_names): Likewise.
9384 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
9385 (RS6000_BTM_FLOAT128_HW): Likewise.
9386 (RS6000_BTM_COMMON): Likewise.
9387 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
9389 * doc/extend.texi (RS/6000 built-in functions): Document the
9390 IEEE 128-bit floating point square root and fused multiply-add
9393 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
9395 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
9396 reg (r2) isn't in the set of registers defined in the prologue.
9398 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9399 Alan Hayward <alan.hayward@arm.com>
9400 David Sherwood <david.sherwood@arm.com>
9402 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
9403 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
9404 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
9405 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
9406 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
9408 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
9409 max_vectorization_factor.
9410 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
9412 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9413 Alan Hayward <alan.hayward@arm.com>
9414 David Sherwood <david.sherwood@arm.com>
9416 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
9417 (vect_worthwhile_without_simd_p): Declare.
9418 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
9419 (vectorizable_reduction): Use it.
9420 * tree-vect-stmts.c (vectorizable_shift): Likewise.
9421 (vectorizable_operation): Likewise.
9423 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9424 Alan Hayward <alan.hayward@arm.com>
9425 David Sherwood <david.sherwood@arm.com>
9427 * tree-vectorizer.h (vect_get_num_copies): New function.
9428 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
9429 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9430 (vectorizable_induction): Likewise.
9431 (vectorizable_live_operation): Likewise.
9432 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9433 (vectorizable_bswap): Likewise.
9434 (vectorizable_call): Likewise.
9435 (vectorizable_conversion): Likewise.
9436 (vectorizable_assignment): Likewise.
9437 (vectorizable_shift): Likewise.
9438 (vectorizable_operation): Likewise.
9439 (vectorizable_store): Likewise.
9440 (vectorizable_load): Likewise.
9441 (vectorizable_condition): Likewise.
9442 (vectorizable_comparison): Likewise.
9443 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
9445 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9446 Alan Hayward <alan.hayward@arm.com>
9447 David Sherwood <david.sherwood@arm.com>
9449 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
9450 of vect_init_vector.
9452 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9453 Alan Hayward <alan.hayward@arm.com>
9454 David Sherwood <david.sherwood@arm.com>
9456 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
9457 an inline wrapper that provides a location.
9458 (gimple_build_vector): Likewise.
9459 * gimple-fold.c (gimple_build_vector_from_val): New function.
9460 (gimple_build_vector): Likewise.
9461 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
9462 functions to build the initial value. Always return a gimple value.
9463 (get_initial_defs_for_reduction): Likewise. Only compute
9465 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
9466 vect_init_vector on the results from get_initial_def(s)_for_reduction.
9467 (vectorizable_induction): Use gimple_build_vector rather than
9470 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9471 Alan Hayward <alan.hayward@arm.com>
9472 David Sherwood <david.sherwood@arm.com>
9474 * target.h (vec_perm_indices): New typedef.
9475 (auto_vec_perm_indices): Likewise.
9476 * optabs-query.h: Include target.h
9477 (can_vec_perm_p): Take a vec_perm_indices *.
9478 * optabs-query.c (can_vec_perm_p): Likewise.
9479 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
9480 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9481 * tree-vect-generic.c (lower_vec_perm): Likewise.
9482 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
9483 (vect_grouped_load_supported): Likewise.
9484 (vect_shift_permute_load_chain): Likewise.
9485 (vect_permute_store_chain): Use auto_vec_perm_indices.
9486 (vect_permute_load_chain): Likewise.
9487 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
9488 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
9489 Update uses of can_vec_perm_p.
9490 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
9491 mode with a number of elements. Take a vec_perm_indices *.
9492 (vect_create_epilog_for_reduction): Update accordingly.
9493 Use auto_vec_perm_indices.
9494 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
9495 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
9496 (vect_transform_slp_perm_load): Likewise.
9497 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
9498 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
9499 (vect_gen_perm_mask_checked): Likewise.
9500 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
9501 (vect_gen_perm_mask_checked): Likewise.
9502 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
9503 (vectorizable_store): Likewise.
9504 (vectorizable_load): Likewise.
9505 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
9506 (vectorizable_bswap): Likewise.
9508 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9509 Alan Hayward <alan.hayward@arm.com>
9510 David Sherwood <david.sherwood@arm.com>
9512 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
9513 * tree.c (build_vector): Likewise.
9514 (build_vector_from_ctor): Update accordingly.
9515 (build_vector_from_val): Likewise.
9516 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
9517 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9518 * tree-vect-generic.c (add_rshift): Likewise.
9519 (expand_vector_divmod): Likewise.
9520 (optimize_vector_constructor): Likewise.
9521 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
9522 (vect_transform_slp_perm_load): Likewise.
9523 (vect_schedule_slp_instance): Likewise.
9524 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
9525 (vectorizable_call): Likewise.
9526 (vect_gen_perm_mask_any): Likewise. Add elements in order.
9527 * expmed.c (make_tree): Likewise.
9528 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
9529 a vector passed to build_vector.
9530 (fold_convert_const): Likewise.
9531 (exact_inverse): Likewise.
9532 (fold_ternary_loc): Likewise.
9533 (fold_relational_const): Likewise.
9534 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
9535 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
9536 (const_unop): Likewise. Store the reduction accumulator in a
9537 variable rather than an array.
9538 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
9539 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
9540 the new vector, rather than constructing it after the input arrays.
9541 (native_interpret_vector): Use auto_vec<tree> when building
9542 a vector passed to build_vector. Add elements in order.
9543 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
9544 auto_vec<tree> when building a vector passed to build_vector.
9545 (vect_create_epilog_for_reduction): Likewise.
9546 (vectorizable_induction): Likewise.
9547 (get_initial_def_for_reduction): Likewise. Fix indentation of
9549 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
9551 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
9552 passed to build_vector.
9554 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9555 Alan Hayward <alan.hayward@arm.com>
9556 David Sherwood <david.sherwood@arm.com>
9558 * tree-core.h (tree_base::u): Add an "nelts" field.
9559 (tree_vector): Use VECTOR_CST_NELTS as the length.
9560 * tree.c (tree_size): Likewise.
9561 (make_vector): Initialize VECTOR_CST_NELTS.
9562 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
9563 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
9564 TYPE_VECTOR_SUBPARTS.
9565 * expr.c (const_vector_mask_from_tree): Consistently use "units"
9566 as the number of units, setting it from VECTOR_CST_NELTS.
9567 (const_vector_from_tree): Likewise.
9568 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
9569 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
9570 (fold_negate_expr_1): Likewise.
9571 (fold_convert_const): Likewise.
9572 (const_binop): Likewise. Differentiate the number of output and
9574 (const_unop): Likewise.
9575 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
9576 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
9577 in cases that did the opposite.
9579 2017-09-14 Richard Biener <rguenther@suse.de>
9581 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
9584 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
9586 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
9588 2017-09-14 Jakub Jelinek <jakub@redhat.com>
9591 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
9592 if and where to split a bb, except for splitting before debug insn
9593 sequences followed by non-label real insn. Delete debug insns
9594 in between basic blocks.
9596 * combine.c (make_compound_operation_int): Formatting fixes.
9598 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
9599 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9600 * config/netbsd.h (LINK_EH_SPEC): Likewise.
9601 * config/sol2.h (LINK_EH_SPEC): Likewise.
9602 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9603 * config/s390/linux.h (LINK_SPEC): Likewise.
9604 * config/freebsd.h (LINK_EH_SPEC): Likewise.
9605 * config/openbsd.h (LINK_EH_SPEC): Likewise.
9606 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9607 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
9608 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
9609 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9610 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
9611 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
9613 2017-09-13 Jakub Jelinek <jakub@redhat.com>
9615 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
9617 (ENDFILE_LINUX_SPEC): Likewise.
9618 (LINK_EH_SPEC): Likewise.
9619 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
9620 (LINK_OS_LINUX_SPEC32): Likewise.
9621 (LINK_OS_LINUX_SPEC64): Likewise.
9622 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
9623 (LINK_OS_LINUX_SPEC): Likewise.
9625 2017-09-13 Martin Liska <mliska@suse.cz>
9628 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
9629 CASE_HIGH is NULL_TREE.
9631 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9632 Alan Hayward <alan.hayward@arm.com>
9633 David Sherwood <david.sherwood@arm.com>
9635 * target.def (secondary_memory_needed): New hook.
9636 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
9637 instead of SECONDARY_MEMORY_NEEDED.
9638 (secondary_memory_needed_mode): Likewise.
9639 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
9640 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
9641 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
9642 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
9643 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
9644 * doc/tm.texi: Regenerate.
9645 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
9646 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
9647 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9648 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9649 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9650 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9651 mode argument first and change the reg_class arguments to reg_class_t.
9652 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9653 Make static. Update the call to inline_secondary_memory_needed.
9654 (ix86_register_move_cost): Update the call to
9655 inline_secondary_memory_needed.
9656 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9657 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9659 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9660 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9662 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9663 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9664 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9665 and match hook interface. Add comment from mips.h.
9666 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9667 * config/mmix/mmix.md (truncdfsf2): Refer to
9668 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9670 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9671 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9672 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9673 (pa_secondary_memory_needed): New function.
9674 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9675 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9676 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9677 (pdp11_secondary_memory_needed): Make static and match hook interface.
9678 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9679 * config/powerpcspe/powerpcspe-protos.h
9680 (rs6000_secondary_memory_needed_ptr): Delete.
9681 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9683 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9684 (rs6000_option_override_internal): Assign to
9685 targetm.secondary_memory_needed rather than
9686 rs6000_secondary_memory_needed_ptr.
9687 (rs6000_secondary_memory_needed): Match hook interface.
9688 (rs6000_debug_secondary_memory_needed): Likewise.
9689 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9690 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9691 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9692 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9693 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9694 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9696 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9697 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9698 (rs6000_option_override_internal): Assign to
9699 targetm.secondary_memory_needed rather than
9700 rs6000_secondary_memory_needed_ptr.
9701 (rs6000_secondary_memory_needed): Match hook interface.
9702 (rs6000_debug_secondary_memory_needed): Likewise.
9703 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9704 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9705 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9706 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9707 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9708 (sparc_secondary_memory_needed): New function.
9709 * lra-constraints.c (check_and_process_move): Refer to
9710 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9712 (curr_insn_transform): Likewise.
9713 (process_alt_operands): Use targetm.secondary_memory_needed
9714 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9715 (check_secondary_memory_needed_p): Likewise.
9716 (choose_split_class): Likewise.
9717 * reload.c: Unconditionally include code that was previously
9718 conditional on SECONDARY_MEMORY_NEEDED.
9719 (push_secondary_reload): Use targetm.secondary_memory_needed
9720 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9721 (push_reload): Likewise.
9722 * reload1.c: Unconditionally include code that was previously
9723 conditional on SECONDARY_MEMORY_NEEDED.
9724 (choose_reload_regs): Use targetm.secondary_memory_needed
9725 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9726 (gen_reload): Likewise.
9727 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9729 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9730 Alan Hayward <alan.hayward@arm.com>
9731 David Sherwood <david.sherwood@arm.com>
9733 * target.def (secondary_memory_needed_mode): New hook:
9734 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9735 * targhooks.h (default_secondary_memory_needed_mode): New function.
9736 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9737 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9738 * doc/tm.texi: Regenerate.
9739 * lra-constraints.c (check_and_process_move): Use
9740 targetm.secondary_memory_needed_mode instead of
9741 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9742 (curr_insn_transform): Likewise.
9743 * reload.c (get_secondary_mem): Likewise.
9744 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9745 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9747 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9748 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9749 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9750 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9751 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9753 * config/powerpcspe/powerpcspe-protos.h
9754 (rs6000_secondary_memory_needed_mode): Delete.
9755 * config/powerpcspe/powerpcspe.c
9756 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9757 (rs6000_secondary_memory_needed_mode): Make static.
9758 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9759 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9761 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9763 (rs6000_secondary_memory_needed_mode): Make static.
9764 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9765 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9766 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9767 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9768 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9770 (sparc_secondary_memory_needed_mode): New function.
9771 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9773 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9775 * config/aarch64/constraints.md (Umq): New constraint.
9776 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9778 (mov<mode>): Update condition.
9780 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9782 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9783 when bitposition is the same.
9785 2017-09-13 Richard Biener <rguenther@suse.de>
9787 * dwarf2out.c (output_die_symbol): Remove.
9788 (output_die): Do not output a DIEs symbol.
9790 2017-09-13 Richard Biener <rguenther@suse.de>
9793 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9794 default-def to avoid breaking iterator update with the weird
9795 interaction with cgraph_update_edges_for_call_stmt_node.
9797 2017-09-13 Richard Biener <rguenther@suse.de>
9799 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9800 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9801 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9802 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9804 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9806 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9807 Disable pc relative literal load irrespective of
9808 TARGET_FIX_ERR_A53_84341 for default.
9810 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9812 * config/sparc/sparc.c (output_return): Output the source location of
9813 the insn in the delay slot, if any.
9814 (output_sibcall): Likewise.
9816 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9819 * common.opt (-static-pie): New alias.
9820 (shared): Negate static-pie.
9821 (-no-pie): Update help text.
9823 (static-pie): New option.
9824 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9825 -static-pie support.
9826 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9827 (LINK_EH_SPEC): Likewise.
9828 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9829 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9830 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9831 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9832 (init_gcc_specs): Likewise.
9833 (init_spec): Likewise.
9834 (display_help): Update help message for -pie.
9835 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9838 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9840 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9841 (movdi_aarch64): Likewise.
9842 (movti_aarch64): Likewise.
9844 2017-09-12 Simon Wright <simon@pushface.org>
9847 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9848 calculation of the minor version, always output as 0.
9850 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9853 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9854 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9855 on it early, rather than manual conversion late. For
9856 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9857 instead of performing manual conversion.
9859 2017-09-12 Carl Love <cel@us.ibm.com>
9861 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9862 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9864 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9865 VMULOSW): Add definitions.
9866 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9867 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9868 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9869 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9870 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9872 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9874 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9876 (movti_aarch64): Likewise.
9877 (movdf_aarch64): Likewise.
9878 (movtf_aarch64): Likewise.
9879 (load_pairdi): Likewise.
9880 (store_pairdi): Likewise.
9881 (load_pairdf): Likewise.
9882 (store_pairdf): Likewise.
9883 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9884 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9885 (ldr_got_small_<mode>): Likewise.
9886 (ldr_got_small_28k_<mode>): Likewise.
9887 (ldr_got_tiny): Likewise.
9888 * config/aarch64/iterators.md (ldst_sz): New.
9889 (ldpstp_sz): Likewise.
9890 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9892 (thunderx_load): Split load_8 to load_16.
9893 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9895 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9896 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9897 (xgene1_store_pair): Split store_8 to store_16.
9898 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9899 (falkor_st_0_st_sd): Split store_8 to store_16.
9901 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9903 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9904 and store1/2/3/4 to store_4/8/12/16.
9905 * config/aarch64/aarch64.md: Update for rename.
9906 * config/arm/arm.md: Likewise.: Likewise.
9907 * config/arm/arm.c: Likewise.
9908 * config/arm/thumb1.md: Likewise.
9909 * config/arm/thumb2.md: Likewise.
9910 * config/arm/vfp.md: Likewise.
9911 * config/arm/arm-generic.md: Likewise.
9912 * config/arm/arm1020e.md: Likewise.
9913 * config/arm/arm1026ejs.md: Likewise.
9914 * config/arm/arm1136jfs.md: Likewise.
9915 * config/arm/arm926ejs.md: Likewise.
9916 * config/arm/cortex-a15.md: Likewise.
9917 * config/arm/cortex-a17.md: Likewise.
9918 * config/arm/cortex-a5.md: Likewise.
9919 * config/arm/cortex-a53.md: Likewise.
9920 * config/arm/cortex-a57.md: Likewise.
9921 * config/arm/cortex-a7.md: Likewise.
9922 * config/arm/cortex-a8.md: Likewise.
9923 * config/arm/cortex-a9.md: Likewise.
9924 * config/arm/cortex-m4.md: Likewise.
9925 * config/arm/cortex-m7.md: Likewise.
9926 * config/arm/cortex-r4.md: Likewise.
9927 * config/arm/exynos-m1.md: Likewise.
9928 * config/arm/fa526.md: Likewise.
9929 * config/arm/fa606te.md: Likewise.
9930 * config/arm/fa626te.md: Likewise.
9931 * config/arm/fa726te.md: Likewise.
9932 * config/arm/fmp626.md: Likewise.
9933 * config/arm/iwmmxt.md: Likewise.
9934 * config/arm/ldmstm.md: Likewise.
9935 * config/arm/marvell-pj4.md: Likewise.
9936 * config/arm/xgene1.md: Likewise.
9937 * config/aarch64/thunderx.md: Likewise.
9938 * config/aarch64/thunderx2t99.md: Likewise.
9939 * config/aarch64/falkor.md: Likewise.
9941 2017-09-12 Martin Liska <mliska@suse.cz>
9943 * attribs.c (private_lookup_attribute): New function.
9944 * attribs.h (private_lookup_attribute): Declared here.
9945 (lookup_attribute): Called from this place.
9947 2017-09-12 Richard Biener <rguenther@suse.de>
9949 PR tree-optimization/82157
9950 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9951 stmts with side-effects.
9953 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9954 Alan Hayward <alan.hayward@arm.com>
9955 David Sherwood <david.sherwood@arm.com>
9957 * target.def (hard_regno_nregs): New hook.
9958 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9959 * targhooks.h (default_hard_regno_nregs): Declare.
9960 * targhooks.c (default_hard_regno_nregs): New function.
9961 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9962 (TARGET_HARD_REGNO_NREGS): ...this hook.
9963 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9964 (CLASS_MAX_NREGS): Likewise.
9965 * doc/tm.texi: Regenerate.
9966 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9967 instead of HARD_REGNO_NREGS.
9968 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9969 HARD_REGNO_NREGS in the comment.
9970 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9971 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9972 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9973 Return an unsigned int.
9974 (TARGET_HARD_REGNO_NREGS): Redefine.
9975 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9976 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9977 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9978 (arc_hard_regno_nregs): New function.
9979 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9980 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9981 (arm_hard_regno_nregs): New function.
9982 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9983 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9984 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9985 (TARGET_HARD_REGNO_NREGS): Redefine.
9986 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9987 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9988 (HARD_REGNO_NREGS): Delete.
9989 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9990 (cr16_hard_regno_nregs): New function.
9991 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9992 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9993 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9994 (cris_hard_regno_nregs): New function.
9995 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9996 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9997 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9998 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9999 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
10000 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
10001 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
10002 (frv_hard_regno_nregs): Make static. Take and return an
10004 (frv_class_max_nregs): Remove outdated copy of documentation.
10005 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
10006 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
10007 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
10008 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
10009 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
10010 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
10011 (TARGET_HARD_REGNO_NREGS): Redefine.
10012 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
10013 (CLASS_MAX_NREGS): Update comment.
10014 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
10015 (ia64_hard_regno_nregs): New function.
10016 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
10017 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
10018 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
10019 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
10020 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
10022 (m32c_hard_regno_nregs): Likewise. Make static.
10023 (TARGET_HARD_REGNO_NREGS): Redefine.
10024 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
10025 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
10026 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
10027 (m68k_hard_regno_nregs): New function.
10028 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
10029 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
10030 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
10031 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
10032 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
10033 Take and return an unsigned int.
10034 (TARGET_HARD_REGNO_NREGS): Redefine.
10035 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
10036 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
10037 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
10038 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
10039 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
10040 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
10041 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
10042 (msp430_hard_regno_nregs): Make static. Take and return an
10044 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
10045 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
10046 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
10047 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
10048 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
10049 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
10050 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
10051 (TARGET_HARD_REGNO_NREGS): Redefine.
10052 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
10053 (PA_HARD_REGNO_NREGS): ...this.
10054 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
10055 (PA_HARD_REGNO_NREGS): ...this.
10056 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
10057 (pa_hard_regno_nregs): New function.
10058 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
10059 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
10060 (pdp11_hard_regno_nregs): New function.
10061 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
10062 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
10063 (rs6000_hard_regno_nregs_hook): New function.
10064 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
10065 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
10066 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
10067 Take and return an unsigned int. Move earlier in file.
10068 (TARGET_HARD_REGNO_NREGS): Redefine.
10069 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
10070 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
10071 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
10072 (rl78_hard_regno_nregs): Make static. Take and return an
10074 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
10075 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
10076 (rs6000_hard_regno_nregs_hook): New function.
10077 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
10078 * config/rx/rx.c (rx_hard_regno_nregs): New function.
10079 (TARGET_HARD_REGNO_NREGS): Redefine.
10080 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
10081 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
10082 instead of HARD_REGNO_NREGS.
10083 (s390_hard_regno_nregs): New function.
10084 (s390_hard_regno_mode_ok): Add comment from s390.h.
10085 (TARGET_HARD_REGNO_NREGS): Redefine.
10086 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
10087 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
10088 (sh_hard_regno_nregs): New function.
10089 (sh_pass_in_reg_p): Use it.
10090 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
10091 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
10092 (sparc_hard_regno_nregs): New function.
10093 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
10094 * config/spu/spu.c (spu_hard_regno_nregs): New function.
10095 (spu_function_arg_advance): Use it, supplying a valid register number.
10096 (TARGET_HARD_REGNO_NREGS): Redefine.
10097 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
10098 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
10099 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
10100 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
10101 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
10102 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
10103 (CLASS_MAX_NREGS): Remove copy of old documentation.
10104 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
10105 (visium_hard_regno_nregs): New function.
10106 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
10107 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
10108 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
10109 xtensa_hard_regno_nregs): New function.
10110 * system.h (HARD_REGNO_NREGS): Poison.
10112 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10114 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
10115 hard_regno_nregs instead of HARD_REGNO_NREGS.
10116 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
10117 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
10118 (c6x_expand_epilogue): Likewise.
10119 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
10120 (frv_read_iacc_argument): Likewise.
10121 * config/sh/sh.c: Include regs.h.
10122 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
10123 (regs_used): Likewise.
10124 (output_stack_adjust): Likewise.
10125 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
10126 * expmed.c: Include regs.h.
10127 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
10128 * ree.c: Include regs.h.
10129 (combine_reaching_defs): Use hard_regno_nregs instead of
10131 (add_removable_extension): Likewise.
10133 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10135 * regs.h (hard_regno_nregs): Turn into a function.
10136 (end_hard_regno): Update accordingly.
10137 * caller-save.c (setup_save_areas): Likewise.
10138 (save_call_clobbered_regs): Likewise.
10139 (replace_reg_with_saved_mem): Likewise.
10140 (insert_restore): Likewise.
10141 (insert_save): Likewise.
10142 * combine.c (can_change_dest_mode): Likewise.
10143 (move_deaths): Likewise.
10144 (distribute_notes): Likewise.
10145 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
10146 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
10147 (rs6000_split_multireg_move): Likewise.
10148 (rs6000_register_move_cost): Likewise.
10149 (rs6000_memory_move_cost): Likewise.
10150 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
10151 (rs6000_split_multireg_move): Likewise.
10152 (rs6000_register_move_cost): Likewise.
10153 (rs6000_memory_move_cost): Likewise.
10154 * cselib.c (cselib_reset_table): Likewise.
10155 (cselib_lookup_1): Likewise.
10156 * emit-rtl.c (set_mode_and_regno): Likewise.
10157 * function.c (aggregate_value_p): Likewise.
10158 * ira-color.c (setup_profitable_hard_regs): Likewise.
10159 (check_hard_reg_p): Likewise.
10160 (calculate_saved_nregs): Likewise.
10161 (assign_hard_reg): Likewise.
10162 (improve_allocation): Likewise.
10163 (calculate_spill_cost): Likewise.
10164 * ira-emit.c (modify_move_list): Likewise.
10165 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
10166 (ira_hard_reg_in_set_p): Likewise.
10167 * ira.c (setup_reg_mode_hard_regset): Likewise.
10168 (clarify_prohibited_class_mode_regs): Likewise.
10169 (check_allocation): Likewise.
10170 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10171 (lra_setup_reg_renumber): Likewise.
10172 (setup_try_hard_regno_pseudos): Likewise.
10173 (spill_for): Likewise.
10174 (assign_hard_regno): Likewise.
10175 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
10176 * lra-constraints.c (in_class_p): Likewise.
10177 (lra_constraint_offset): Likewise.
10178 (simplify_operand_subreg): Likewise.
10179 (lra_constraints): Likewise.
10180 (split_reg): Likewise.
10181 (split_if_necessary): Likewise.
10182 (invariant_p): Likewise.
10183 (inherit_in_ebb): Likewise.
10184 * lra-lives.c (process_bb_lives): Likewise.
10185 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
10186 (get_hard_regs): Likewise.
10187 (do_remat): Likewise.
10188 * lra-spills.c (assign_spill_hard_regs): Likewise.
10189 * mode-switching.c (create_pre_exit): Likewise.
10190 * postreload.c (reload_combine_recognize_pattern): Likewise.
10191 * recog.c (peep2_find_free_register): Likewise.
10192 * regcprop.c (kill_value_regno): Likewise.
10193 (set_value_regno): Likewise.
10194 (copy_value): Likewise.
10195 (maybe_mode_change): Likewise.
10196 (find_oldest_value_reg): Likewise.
10197 (copyprop_hardreg_forward_1): Likewise.
10198 * regrename.c (check_new_reg_p): Likewise.
10199 (regrename_do_replace): Likewise.
10200 * reload.c (push_reload): Likewise.
10201 (combine_reloads): Likewise.
10202 (find_dummy_reload): Likewise.
10203 (operands_match_p): Likewise.
10204 (find_reloads): Likewise.
10205 (find_equiv_reg): Likewise.
10206 (reload_adjust_reg_for_mode): Likewise.
10207 * reload1.c (count_pseudo): Likewise.
10208 (count_spilled_pseudo): Likewise.
10209 (find_reg): Likewise.
10210 (clear_reload_reg_in_use): Likewise.
10211 (free_for_value_p): Likewise.
10212 (allocate_reload_reg): Likewise.
10213 (choose_reload_regs): Likewise.
10214 (reload_adjust_reg_for_temp): Likewise.
10215 (emit_reload_insns): Likewise.
10216 (delete_output_reload): Likewise.
10217 * rtlanal.c (subreg_get_info): Likewise.
10218 * sched-deps.c (sched_analyze_reg): Likewise.
10219 * sel-sched.c (init_regs_for_mode): Likewise.
10220 (mark_unavailable_hard_regs): Likewise.
10221 (choose_best_reg_1): Likewise.
10222 (verify_target_availability): Likewise.
10223 * valtrack.c (dead_debug_insert_temp): Likewise.
10224 * var-tracking.c (track_loc_p): Likewise.
10225 (emit_note_insn_var_location): Likewise.
10226 * varasm.c (make_decl_rtl): Likewise.
10227 * reginfo.c (choose_hard_reg_mode): Likewise.
10228 (init_reg_modes_target): Refer directly to
10229 this_target_regs->x_hard_regno_nregs.
10231 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10233 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
10234 instead of hard_regno_nregs.
10236 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10238 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
10239 end_hard_regno instead of hard_regno_nregs.
10240 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
10241 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
10242 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
10243 * ira-color.c (improve_allocation): Likewise.
10244 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10245 * lra-lives.c (mark_regno_live): Likewise.
10246 (mark_regno_dead): Likewise.
10247 * lra-remat.c (operand_to_remat): Likewise.
10248 * lra.c (collect_non_operand_hard_regs): Likewise.
10249 * postreload.c (reload_combine_note_store): Likewise.
10250 (move2add_valid_value_p): Likewise.
10251 * reload.c (regno_clobbered_p): Likewise.
10253 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10255 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
10257 * config/v850/v850.c (v850_reorg): Likewise.
10258 * reload.c (refers_to_regno_for_reload_p): Likewise.
10259 (find_equiv_reg): Likewise.
10260 * reload1.c (reload_reg_reaches_end_p): Likewise.
10262 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10264 * caller-save.c (add_used_regs): Use REG_NREGS instead of
10266 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
10267 * config/arm/arm.c (output_move_neon): Likewise.
10268 (arm_attr_length_move_neon): Likewise.
10269 (neon_split_vcombine): Likewise.
10270 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
10271 (c6x_mark_reg_written): Likewise.
10272 (c6x_dwarf_register_span): Likewise.
10273 * config/i386/i386.c (ix86_save_reg): Likewise.
10274 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
10275 (rws_access_reg): Likewise.
10276 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
10277 * mode-switching.c (create_pre_exit): Likewise.
10278 * ree.c (combine_reaching_defs): Likewise.
10279 (add_removable_extension): Likewise.
10280 * regcprop.c (find_oldest_value_reg): Likewise.
10281 (copyprop_hardreg_forward_1): Likewise.
10282 * reload.c (reload_inner_reg_of_subreg): Likewise.
10283 (push_reload): Likewise.
10284 (combine_reloads): Likewise.
10285 (find_dummy_reload): Likewise.
10286 (reload_adjust_reg_for_mode): Likewise.
10287 * reload1.c (find_reload_regs): Likewise.
10288 (forget_old_reloads_1): Likewise.
10289 (reload_reg_free_for_value_p): Likewise.
10290 (reload_adjust_reg_for_temp): Likewise.
10291 (emit_reload_insns): Likewise.
10292 (delete_output_reload): Likewise.
10293 * sel-sched.c (choose_best_reg_1): Likewise.
10294 (choose_best_pseudo_reg): Likewise.
10296 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10297 Alan Hayward <alan.hayward@arm.com>
10298 David Sherwood <david.sherwood@arm.com>
10300 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
10301 * target.def (slow_unaligned_access): New hook.
10302 * targhooks.h (default_slow_unaligned_access): Declare.
10303 * targhooks.c (default_slow_unaligned_access): New function.
10304 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
10305 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
10306 * doc/tm.texi: Regenerate.
10307 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
10308 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
10309 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
10311 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
10312 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
10314 (rs6000_slow_unaligned_access): New function.
10315 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
10316 (expand_block_compare): Likewise.
10317 (expand_strn_compare): Likewise.
10318 (rs6000_rtx_costs): Likewise.
10319 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
10320 (riscv_slow_unaligned_access): Likewise.
10321 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
10322 (riscv_slow_unaligned_access_p): ...this and make static.
10323 (riscv_option_override): Update accordingly.
10324 (riscv_slow_unaligned_access): New function.
10325 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
10326 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
10327 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
10328 (rs6000_slow_unaligned_access): New function.
10329 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
10330 (rs6000_rtx_costs): Likewise.
10331 * config/rs6000/rs6000-string.c (expand_block_compare)
10332 (expand_strn_compare): Use targetm.slow_unaligned_access instead
10333 of SLOW_UNALIGNED_ACCESS.
10334 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
10335 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
10336 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
10337 of SLOW_UNALIGNED_ACCESS.
10338 * expmed.c (simple_mem_bitfield_p): Likewise.
10339 * expr.c (alignment_for_piecewise_move): Likewise.
10340 (emit_group_load_1): Likewise.
10341 (emit_group_store): Likewise.
10342 (copy_blkmode_from_reg): Likewise.
10343 (emit_push_insn): Likewise.
10344 (expand_assignment): Likewise.
10345 (store_field): Likewise.
10346 (expand_expr_real_1): Likewise.
10347 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
10348 * lra-constraints.c (simplify_operand_subreg): Likewise.
10349 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10350 * gimple-ssa-store-merging.c: Likewise in block comment at start
10352 * tree-ssa-strlen.c: Include target.h.
10353 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
10354 of SLOW_UNALIGNED_ACCESS.
10355 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
10357 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10359 PR rtl-optimization/82185
10360 * expmed.c (emit_store_flag_int): Only test tem if it has been
10363 2017-09-12 Richard Biener <rguenther@suse.de>
10365 PR middle-end/82149
10366 * match.pd ((FTYPE) N CMP CST): Fix typo.
10368 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
10370 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
10372 (mips_near_type_p): Add 'short_call' attribute as a synonym
10374 * doc/extend.texi (short_call): Document new function attribute.
10376 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10379 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
10380 assertion check that in the condition.
10381 (get_atomic_generic_size): Likewise. Before testing if parameter
10382 has pointer type, if it has array type, call for C++
10383 default_conversion to perform array-to-pointer conversion.
10385 2017-09-12 Richard Biener <rguenther@suse.de>
10387 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
10388 for operations we cannot scalarize.
10390 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
10392 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
10393 vectors heap vectors. Clean up comments.
10394 Make visited_bbs a reference.
10395 (profitable_jump_thread_path): Make GC
10396 vectors heap vectors. Clean up comments.
10398 (convert_and_register_jump_thread_path): Make GC vectors heap
10400 (check_subpath_and_update_thread_path): Same. Clean up comments.
10401 Make visited_bbs a reference.
10402 (handle_phi): Abstract common code to to
10403 register_jump_thread_path_if_profitable.
10404 Rename VAR_BB to DEF_BB.
10406 Make GC vectors heap vectors.
10407 Make visited_bbs a reference.
10408 (handle_assignment): Same.
10409 (register_jump_thread_path_if_profitable): New.
10410 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
10412 Make GC vectors heap vectors. Clean up comments.
10413 Make visited_bbs a reference.
10414 (find_jump_threads_backwards): Make visited_bbs live in the stack.
10415 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
10418 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
10421 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
10422 words of E_DImode object are reachable by xtensa_uimm8x4 access.
10424 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
10426 Revert r251800 and r251799.
10428 2017-09-11 Martin Jambor <mjambor@suse.cz>
10431 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
10432 arguments in advance.
10433 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
10434 use it to find predecessor edges.
10435 (naive_outof_ssa): Collect vector of predecessors.
10437 2017-09-08 Jason Merrill <jason@redhat.com>
10439 PR c++/70029 - ICE with ref-qualifier and -flto
10440 * langhooks.h (struct lang_hooks_for_types): Add
10441 copy_lang_qualifiers.
10442 * attribs.c (build_type_attribute_qual_variant): Use it.
10443 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
10445 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
10446 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
10448 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
10451 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
10452 (*mulsi3_sp64): New instruction.
10453 (mulsi3): New expander.
10455 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
10457 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
10459 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10461 * sancov.c: Include memmodel.h.
10463 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
10466 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
10469 2017-09-07 Carl Love <cel@us.ibm.com>
10471 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
10472 the sldi instruction.
10474 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
10476 * sancov.c: Include tm_p.h.
10478 2017-09-07 Jakub Jelinek <jakub@redhat.com>
10481 * output.h (switch_to_other_text_partition): New declaration.
10482 * varasm.c (switch_to_other_text_partition): New function.
10483 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
10484 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
10485 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
10486 to the other text partition before emitting LCL label and switch back
10487 after emitting the word after it.
10489 2017-09-07 Richard Biener <rguenther@suse.de>
10491 * passes.def (pass_split_crit_edges): Remove instance before PRE.
10492 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
10493 critical edges here, after loop init.
10494 (pass_data_pre): Remove PROP_no_crit_edges flags.
10495 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
10496 for valueization of call args to avoid leaking VN_TOP.
10497 (visit_use): Assert we do not visit default defs.
10498 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
10499 Use error_mark_node to more easily detect leaking VN_TOP.
10500 All default-defs are varying, not VN_TOP. Mark them visited.
10501 (run_scc_vn): Make code match comment.
10503 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10505 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
10506 OPTION_MASK_FLOAT128_KEYWORD.
10507 (POWERPC_MASKS): Likewise.
10508 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
10509 support for the -mfloat128-type option, and make -mfloat128
10510 default on PowerPC Linux systems. Define or undefine
10511 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
10512 Define __float128 to be __ieee128 if IEEE 128-bit support is
10513 enabled, or undefine it.
10514 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
10515 Delete defining __FLOAT128_TYPE__.
10516 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
10517 -mfloat128-type option and make -mfloat128 default on PowerPC
10519 (TARGET_FLOAT128_TYPE): Likewise.
10520 (-mfloat128-type): Likewise.
10521 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10522 Delete the -mfloat128-type option and make -mfloat128 default on
10523 PowerPC Linux systems. Always use __ieee128 to be the keyword for
10524 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
10525 128-bit floating point is enabled. Change tests from using
10526 -mfloat128-type to -mfloat128.
10527 (rs6000_mangle_type): Use the correct mangling for the __float128
10528 type even if normal long double is restricted to 64-bits.
10529 (floatn_mode): Enable the _Float128 type by default on VSX Linux
10531 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
10532 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
10533 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
10535 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
10536 documentation for -mfloat128.
10538 2017-09-06 Olivier Hainque <hainque@adacore.com>
10540 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
10542 2017-09-06 Wish Wu <wishwu007@gmail.com>
10543 Jakub Jelinek <jakub@redhat.com>
10545 * asan.c (initialize_sanitizer_builtins): Add
10546 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
10547 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
10548 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
10549 BT_FN_VOID_UINT64_PTR variables.
10550 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
10551 (BT_FN_VOID_UINT16_UINT16): Likewise.
10552 (BT_FN_VOID_UINT32_UINT32): Likewise.
10553 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
10554 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
10555 (BT_FN_VOID_UINT64_PTR): Likewise.
10556 * common.opt (flag_sanitize_coverage): New variable.
10557 (fsanitize-coverage=trace-pc): Remove.
10558 (fsanitize-coverage=): Add.
10559 * flag-types.h (enum sanitize_coverage_code): New enum.
10560 * fold-const.c (fold_range_test): Disable non-short-circuit
10561 optimization if flag_sanitize_coverage.
10562 (fold_truth_andor): Likewise.
10563 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
10564 * opts.c (COVERAGE_SANITIZER_OPT): Define.
10565 (coverage_sanitizer_opts): New array.
10566 (get_closest_sanitizer_option): Add OPTS argument, handle also
10567 OPT_fsanitize_coverage_.
10568 (parse_sanitizer_options): Adjusted to also handle
10569 OPT_fsanitize_coverage_.
10570 (common_handle_option): Add OPT_fsanitize_coverage_.
10571 * sancov.c (instrument_comparison, instrument_switch): New function.
10572 (sancov_pass): Add trace-cmp support.
10573 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
10574 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
10575 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
10576 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
10577 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
10578 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
10579 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
10580 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
10581 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
10582 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
10584 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10586 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10587 error. Only quit immediately if parsing is complete.
10588 (BEGIN): Initialize fatal_err and parse_done.
10589 (begin fpu, end fpu): Check number of arguments.
10590 (begin arch, end arch): Likewise.
10591 (begin cpu, end cpu): Likewise.
10592 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10593 (optalias): Likewise.
10595 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10597 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10598 * config/arm/arm-isa.h: Delete. Move definitions to ...
10599 * arm-cpus.in: ... here. Use new feature and fgroup values.
10600 * config/arm/arm.c (arm_option_override): Use lower case for feature
10602 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10603 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10604 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10606 (print_isa_bits_for): New function.
10607 (gen_isa): New function.
10608 (gen_comm_data): Use print_isa_bits_for.
10609 (define feature): New keyword.
10610 (define fgroup): New keyword.
10611 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
10612 (arm-isa.h): Add rule to generate file.
10613 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10614 case for feature bit names.
10616 2017-09-06 Richard Biener <rguenther@suse.de>
10618 * tree-ssa-pre.c (NECESSARY): Remove.
10619 (create_expression_by_pieces): Do not touch pass-local flags.
10620 (insert_into_preds_of_block): Likewise.
10621 (do_pre_regular_insertion): Likewise.
10622 (eliminate_insert): Likewise.
10623 (eliminate_dom_walker::before_dom_children): Likewise.
10624 (fini_eliminate): Do not look at inserted_exprs.
10625 (mark_operand_necessary): Remove.
10626 (remove_dead_inserted_code): Replace with simple work-list
10627 algorithm based on inserted_exprs and SSA uses.
10628 (pass_pre::execute): Re-order fini_eliminate and
10629 remove_dead_inserted_code.
10631 2017-09-06 Olivier Hainque <hainque@adacore.com>
10633 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
10634 for VxWorks 7. Adjust surrounding comments.
10636 2017-09-06 Richard Biener <rguenther@suse.de>
10638 * gimple-ssa-strength-reduction.c
10639 (find_candidates_dom_walker::before_dom_children): Also allow
10642 2017-09-06 Richard Biener <rguenther@suse.de>
10644 PR tree-optimization/82108
10645 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
10646 for gap in the non-permutation SLP case.
10648 2017-09-06 Martin Jambor <mjambor@suse.cz>
10650 PR tree-optimization/82078
10651 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10652 add_access_to_work_queue...
10653 (build_accesses_from_assign): ...here.
10654 (propagate_all_subaccesses): Make sure racc is the group
10655 representative, if there is one.
10657 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10659 PR middle-end/82095
10660 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10663 2017-09-06 Richard Biener <rguenther@suse.de>
10665 * gimple-ssa-strength-reduction.c
10666 (find_candidates_dom_walker::before_doom_children): Use a
10667 type and not a mode check.
10669 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10672 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10674 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10676 2017-09-05 Jeff Law <law@redhat.com>
10678 PR tree-optimization/64910
10679 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10680 cases where we have 3 or more operands.
10682 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10684 PR middle-end/81768
10685 * omp-low.c (lower_omp_for): Recompute tree invariant if
10686 gimple_omp_for_initial/final is ADDR_EXPR.
10688 PR middle-end/81768
10689 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10690 into gimple val before gimplification fo the COND_EXPR.
10692 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10694 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10695 REGION_COPY argument.
10696 (thread_through_all_blocks): Remove unused argument to
10697 duplicate_thread_path.
10699 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10700 Alan Hayward <alan.hayward@arm.com>
10701 David Sherwood <david.sherwood@arm.com>
10703 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
10704 Take a scalar_mode rather than a machine_mode.
10705 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10706 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10707 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10708 (aarch64_gen_adjusted_ldpstp): Likewise.
10709 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10711 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10712 Alan Hayward <alan.hayward@arm.com>
10713 David Sherwood <david.sherwood@arm.com>
10715 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10716 Take a scalar_int_mode instead of a machine_mode.
10717 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10718 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10719 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10720 (aarch64_simd_attr_length_rglist): Delete.
10721 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10722 a scalar_int_mode instead of a machine_mode.
10723 (aarch64_add_offset): Likewise.
10724 (aarch64_internal_mov_immediate): Likewise
10725 (aarch64_add_constant_internal): Likewise.
10726 (aarch64_add_constant): Likewise.
10727 (aarch64_movw_imm): Likewise.
10728 (aarch64_rtx_arith_op_extract_p): Likewise.
10729 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10730 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10731 Remove assert that the mode isn't a vector.
10732 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10733 (aarch64_expand_mov_immediate): Update calls after above changes.
10734 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10735 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10736 (aarch64_move_imm): Likewise.
10737 (aarch64_can_const_movi_rtx_p): Likewise.
10738 (aarch64_strip_extend): Likewise.
10739 (aarch64_extr_rtx_p): Likewise.
10740 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10741 a CONST_INT when the mode parameter is VOIDmode.
10742 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10744 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10746 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10747 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10748 (bitwise_type_for_mode): Update accordingly.
10750 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10752 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10753 * stor-layout.c (mode_for_size_tree): Likewise.
10754 (mode_for_array): Update accordingly.
10755 (layout_decl): Likewise.
10756 (compute_record_mode): Likewise. Only set the mode once.
10758 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10760 * target.def (get_mask_mode): Change return type to opt_mode.
10762 * doc/tm.texi: Regenerate.
10763 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10764 * targhooks.c (default_get_mask_mode): Likewise.
10765 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10766 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10767 targetm.get_mask_mode.
10768 * tree.c (build_truth_vector_type): Likewise.
10770 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10772 * machmode.h (mode_for_vector): Return an opt_mode.
10773 * stor-layout.c (mode_for_vector): Likewise.
10774 (mode_for_int_vector): Update accordingly.
10775 (layout_type): Likewise.
10776 * config/i386/i386.c (emit_memmov): Likewise.
10777 (ix86_expand_set_or_movmem): Likewise.
10778 (ix86_expand_vector_init): Likewise.
10779 (ix86_get_mask_mode): Likewise.
10780 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10782 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10783 * expmed.c (extract_bit_field_1): Likewise.
10784 * expr.c (expand_expr_real_2): Likewise.
10785 * optabs-query.c (can_vec_perm_p): Likewise.
10786 (can_vec_mask_load_store_p): Likewise.
10787 * optabs.c (expand_vec_perm): Likewise.
10788 * targhooks.c (default_get_mask_mode): Likewise.
10789 * tree-vect-stmts.c (vectorizable_store): Likewise.
10790 (vectorizable_load): Likewise.
10791 (get_vectype_for_scalar_type_and_size): Likewise.
10793 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10795 * machmode.h (mode_for_int_vector): New function.
10796 * stor-layout.c (mode_for_int_vector): Likewise.
10797 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10798 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10799 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10800 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10801 (s390_expand_vcond): Likewise.
10803 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10805 * machmode.h (opt_machine_mode): New type.
10806 (opt_mode<T>): Allow construction from anything that can be
10808 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10809 (mode_for_size): Return an opt_machine_mode.
10810 * stor-layout.c (mode_for_size): Likewise.
10811 (mode_for_size_tree): Update call accordingly.
10812 (bitwise_mode_for_mode): Likewise.
10813 (make_fract_type): Likewise.
10814 (make_accum_type): Likewise.
10815 * caller-save.c (replace_reg_with_saved_mem): Update call
10817 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10818 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10819 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10820 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10821 * expmed.c (extract_bit_field_1): Likewise.
10822 * reload.c (get_secondary_mem): Likewise.
10823 * varasm.c (assemble_integer): Likewise.
10824 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10827 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10829 * machmode.h (decimal_float_mode_for_size): New function.
10830 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10831 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10832 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10833 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10834 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10835 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10837 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10839 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10840 (get_builtin_sync_mode): Likewise.
10841 (expand_ifn_atomic_compare_exchange): Likewise.
10842 (expand_builtin_atomic_clear): Likewise.
10843 (expand_builtin_atomic_test_and_set): Likewise.
10844 (fold_builtin_atomic_always_lock_free): Likewise.
10845 * calls.c (compute_argument_addresses): Likewise.
10846 (emit_library_call_value_1): Likewise.
10847 (store_one_arg): Likewise.
10848 * combine.c (combine_instructions): Likewise.
10849 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10850 * config/arm/arm.c (arm_function_value): Likewise.
10851 (aapcs_allocate_return_reg): Likewise.
10852 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10853 * config/i386/i386.c (construct_container): Likewise.
10854 (ix86_gimplify_va_arg): Likewise.
10855 (ix86_expand_sse_cmp): Likewise.
10856 (emit_memmov): Likewise.
10857 (emit_memset): Likewise.
10858 (expand_small_movmem_or_setmem): Likewise.
10859 (ix86_expand_pextr): Likewise.
10860 (ix86_expand_pinsr): Likewise.
10861 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10862 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10864 * config/mips/mips.c (mips_function_value_1) Likewise.
10865 (mips_block_move_straight): Likewise.
10866 (mips_expand_ins_as_unaligned_store): Likewise.
10867 * config/powerpcspe/powerpcspe.c
10868 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10869 (rs6000_darwin64_record_arg_flush): Likewise.
10870 * config/rs6000/rs6000.c
10871 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10872 (rs6000_darwin64_record_arg_flush): Likewise.
10873 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10874 (sparc_function_value_1): Likewise.
10875 * config/spu/spu.c (adjust_operand): Likewise.
10876 (spu_emit_branch_or_set): Likewise.
10877 (arith_immediate_p): Likewise.
10878 * emit-rtl.c (gen_lowpart_common): Likewise.
10879 * expr.c (expand_expr_real_1): Likewise.
10880 * function.c (assign_parm_setup_block): Likewise.
10881 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10882 * reload1.c (alter_reg): Likewise.
10883 * stor-layout.c (mode_for_vector): Likewise.
10884 (layout_type): Likewise.
10886 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10888 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10889 (spu_convert_move): Likewise.
10890 * lower-subreg.c (resolve_simple_move): Likewise.
10892 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10895 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10896 define_insn to a define_expand.
10897 (altivec_vsum2sws_direct): New define_insn.
10898 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10900 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10902 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10905 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10910 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10911 realign stack if stack alignment needed is less than incoming
10914 2017-09-05 Marek Polacek <polacek@redhat.com>
10917 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10920 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10922 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10924 2017-09-05 Richard Biener <rguenther@suse.de>
10926 PR tree-optimization/82084
10927 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10929 2017-09-05 Richard Biener <rguenther@suse.de>
10931 PR tree-optimization/82102
10932 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10934 2017-09-05 Martin Liska <mliska@suse.cz>
10936 PR tree-optimization/82032
10937 * tree-cfg.c (generate_range_test): New function.
10938 * tree-cfg.h (generate_range_test): Declared here.
10939 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10940 (cleanup_control_expr_graph): Use it.
10941 * tree-switch-conversion.c (try_switch_expansion): Remove
10943 (emit_case_nodes): Use generate_range_test.
10945 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10948 * config/i386/i386.md (*<btsc><mode>_mask): Add
10949 TARGET_USE_BT to insn constraint.
10950 (*btr<mode>_mask): Ditto.
10952 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10954 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10955 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10957 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10960 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10961 TARGET_NEON and TARGET_IWMMXT.
10962 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10963 TARGET_NEON and TARGET_IWMMXT.
10964 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10966 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10968 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10969 (ix86_rewrite_tls_address): Ditto.
10970 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10971 (ix86_rewrite_tls_address_1): Ditto.
10972 (ix86_rewrite_tls_address): Ditto.
10973 * config/i386/predicates.md (tls_address_pattern): New predicate.
10974 * config/i386/i386.md (TLS address splitter): New splitter.
10976 2017-09-04 Richard Biener <rguenther@suse.de>
10978 PR tree-optimization/82084
10979 * fold-const.h (can_native_encode_string_p): Declare.
10980 * fold-const.c (can_native_encode_string_p): Factor out from ...
10981 (native_encode_string): ... here.
10982 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10983 vectorizing stores from constants we later cannot handle.
10985 2017-09-04 Marek Polacek <polacek@redhat.com>
10988 * doc/invoke.texi: Update -Wtautological-compare documentation.
10990 2017-09-04 Jeff Law <law@redhat.com>
10992 PR tree-optimization/64910
10993 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10994 swap the first and last operand if the last is a constant.
10996 2017-09-04 Marek Polacek <polacek@redhat.com>
10999 * convert.c (do_narrow): When sanitizing signed integer overflows,
11000 bail out for signed types.
11001 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
11003 2017-09-04 Richard Biener <rguenther@suse.de>
11005 PR tree-optimization/82060
11006 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
11007 Move devirtualization after stmt folding and before EH/AB/noreturn
11008 cleanup to get the stmt refs canonicalized. Use a bool instead
11009 of gimple_modified_p since that doesn't work for NOPs. Schedule
11010 NOPs generated by folding for removal.
11012 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11013 Alan Hayward <alan.hayward@arm.com>
11014 David Sherwood <david.sherwood@arm.com>
11016 * coretypes.h (pad_direction): New enum.
11017 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
11018 (FUNCTION_ARG_PADDING): Likewise.
11019 * target.def (function_arg_padding): New hook.
11020 * targhooks.h (default_function_arg_padding): Declare.
11021 * targhooks.c (default_function_arg_padding): New function.
11022 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
11023 (TARGET_FUNCTION_ARG_PADDING): ...this.
11024 * doc/tm.texi: Regenerate.
11025 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
11026 instead of direction.
11027 (compute_argument_addresses): Likewise.
11028 (load_register_parameters): Likewise.
11029 (emit_library_call_value_1): Likewise.
11030 (store_one_arg): Use targetm.calls.function_arg_padding instead
11031 of FUNCTION_ARG_PADDING.
11032 (must_pass_in_stack_var_size_or_pad): Likewise.
11033 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
11034 (emit_group_store): Likewise.
11035 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
11036 instead of FUNCTION_ARG_PADDING.
11037 (emit_push_insn): Likewise, and propagate enum change throughout
11039 * function.h (direction): Delete.
11040 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
11042 * function.c (assign_parm_find_stack_rtl): Likewise.
11043 (assign_parm_setup_block_p): Likewise.
11044 (assign_parm_setup_block): Likewise.
11045 (gimplify_parameters): Likewise.
11046 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
11047 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
11049 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
11050 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11051 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
11052 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
11053 (aarch64_function_arg_padding): ...this new function.
11054 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
11055 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11056 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
11057 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11058 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
11059 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11060 (arm_pad_arg_upward): Replace with...
11061 (arm_function_arg_padding): ...this new function.
11062 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
11064 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
11065 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
11066 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11067 (ia64_hpux_function_arg_padding): Replace with...
11068 (ia64_function_arg_padding): ...this new function. Use pad_direction
11069 instead of direction. Check for TARGET_HPUX.
11070 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
11071 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11072 (iq2000_function_arg_padding): New function.
11073 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
11074 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
11075 (mips_function_arg_padding): ...this new function.
11076 (mips_pad_reg_upward): Update accordingly.
11077 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11078 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
11079 targetm.calls.function_arg_padding.
11080 (FUNCTION_ARG_PADDING): Delete.
11081 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
11082 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
11083 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11084 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
11085 (nios2_block_reg_padding): Return pad_direction instead of direction.
11086 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
11087 instead of direction.
11088 (nios2_function_arg_padding): Likewise. Make static.
11089 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11090 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
11091 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
11092 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
11093 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11094 (pa_function_arg_padding): Make static. Return pad_direction instead
11096 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
11097 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11098 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
11099 instead of direction. Use targetm.calls.function_arg_padding.
11100 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
11101 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
11102 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
11103 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
11104 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
11106 (function_arg_padding): Rename to...
11107 (rs6000_function_arg_padding): ...this. Make static. Return
11108 pad_direction instead of direction.
11109 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
11110 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
11111 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
11112 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
11113 instead of direction. Use targetm.calls.function_arg_padding.
11114 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
11115 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
11116 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
11117 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
11118 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11119 (function_arg_padding): Rename to...
11120 (rs6000_function_arg_padding): ...this. Make static. Return
11121 pad_direction instead of direction.
11122 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
11123 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
11124 * config/s390/s390.c (s390_function_arg_padding): New function.
11125 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11126 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
11127 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
11128 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
11129 (function_arg_padding): Rename to...
11130 (sparc_function_arg_padding): ...this. Make static. Return
11131 pad_direction instead of direction.
11132 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
11133 * config/spu/spu.c (spu_function_arg_padding): New function.
11134 (TARGET_FUNCTION_ARG_PADDING): Redefine.
11135 * system.h (FUNCTION_ARG_PADDING): Poison.
11137 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11138 Alan Hayward <alan.hayward@arm.com>
11139 David Sherwood <david.sherwood@arm.com>
11141 * target.def (modes_tieable_p): New hook.
11142 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
11143 (TARGET_MODES_TIEABLE_P): ...this.
11144 * doc/tm.texi.in: Regenerate.
11145 * hooks.h (hook_bool_mode_mode_true): Declare.
11146 * hooks.c (hook_bool_mode_mode_true): New function.
11147 * combine.c (subst): Use targetm.modes_tieable_p instead of
11149 * dse.c (find_shift_sequence): Likewise.
11150 * expmed.c (extract_low_bits): Likewise.
11151 * lower-subreg.c: Include target.h.
11152 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
11154 * rtlanal.c (rtx_cost): Likewise.
11155 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
11156 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
11157 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
11158 (TARGET_MODES_TIEABLE_P): Redefine.
11159 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
11160 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
11161 (TARGET_MODES_TIEABLE_P): Redefine.
11162 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
11163 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
11164 (arc_modes_tieable_p): New function.
11165 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
11166 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
11167 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
11168 (arm_modes_tieable_p): Make static.
11169 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
11170 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
11171 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
11172 (TARGET_MODES_TIEABLE_P): Redefine.
11173 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
11174 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
11175 (TARGET_MODES_TIEABLE_P): Redefine.
11176 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
11177 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
11178 (cr16_modes_tieable_p): New function.
11179 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
11180 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
11181 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
11182 (TRULY_NOOP_TRUNCATION): Update comment.
11183 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
11184 (TRULY_NOOP_TRUNCATION): Update comment.
11185 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
11186 (frv_modes_tieable_p): New function.
11187 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
11188 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
11189 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
11190 (TARGET_MODES_TIEABLE_P): Redefine.
11191 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
11192 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
11193 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
11194 (TARGET_MODES_TIEABLE_P): Redefine.
11195 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
11196 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
11197 (ia64_modes_tieable_p): New function.
11198 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
11199 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
11200 (iq2000_modes_tieable_p): New function.
11201 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
11202 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
11203 (lm32_modes_tieable_p): New function.
11204 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
11205 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
11206 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
11207 (TARGET_MODES_TIEABLE_P): Redefine.
11208 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
11209 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
11210 (m32r_modes_tieable_p): New function.
11211 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
11212 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
11213 (m68k_modes_tieable_p): New function.
11214 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
11215 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
11216 (mcore_modes_tieable_p): New function.
11217 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
11218 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
11220 (TARGET_MODES_TIEABLE_P): Redefine.
11221 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
11222 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
11223 * config/mips/mips.c (mips_modes_tieable_p): Make static.
11224 (TARGET_MODES_TIEABLE_P): Redefine.
11225 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
11226 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
11227 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
11228 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
11229 (mn10300_modes_tieable_p): ...this and make static.
11230 (TARGET_MODES_TIEABLE_P): Redefine.
11231 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
11232 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
11233 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
11234 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
11235 (msp430_modes_tieable_p): Make static.
11236 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
11237 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
11238 (TARGET_MODES_TIEABLE_P): Redefine.
11239 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
11240 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
11241 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
11242 (TARGET_MODES_TIEABLE_P): Redefine.
11243 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
11244 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
11245 * config/pa/pa.c (pa_modes_tieable_p): Make static.
11246 (TARGET_MODES_TIEABLE_P): Redefine.
11247 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
11248 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
11249 (pdp11_modes_tieable_p): New function.
11250 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
11251 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
11252 (rs6000_modes_tieable_p): New function.
11253 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
11254 * config/powerpcspe/powerpcspe.md: Update comment.
11255 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
11256 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
11257 (TARGET_MODES_TIEABLE_P): Redefine.
11258 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
11259 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
11260 (rl78_modes_tieable_p): New function.
11261 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
11262 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
11263 (rs6000_modes_tieable_p): New function.
11264 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
11265 * config/rs6000/rs6000.md: Update comment.
11266 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
11267 * config/rx/rx.c (rx_modes_tieable_p): New function.
11268 (TARGET_MODES_TIEABLE_P): Redefine.
11269 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
11270 * config/s390/s390.c (s390_modes_tieable_p): New function.
11271 (TARGET_MODES_TIEABLE_P): Redefine.
11272 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
11273 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
11274 (sh_modes_tieable_p): New function.
11275 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
11276 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
11277 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
11278 (sparc_modes_tieable_p): Make static.
11279 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
11280 * config/spu/spu.c (spu_modes_tieable_p): New function.
11281 (TARGET_MODES_TIEABLE_P): Redefine.
11282 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
11283 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
11284 (TARGET_MODES_TIEABLE_P): Redefine.
11285 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
11286 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
11287 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
11288 * config/v850/v850.c (v850_modes_tieable_p): New function.
11289 (TARGET_MODES_TIEABLE_P): Redefine.
11290 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
11291 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
11292 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
11293 (visium_modes_tieable_p): New function.
11294 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
11295 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
11296 (xtensa_modes_tieable_p): New function.
11297 * system.h (MODES_TIEABLE_P): Poison.
11299 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11300 Alan Hayward <alan.hayward@arm.com>
11301 David Sherwood <david.sherwood@arm.com>
11303 * target.def (hard_regno_mode_ok): New hook.
11304 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
11305 (TARGET_HARD_REGNO_MODE_OK): ...this.
11306 * doc/tm.texi.in: Regenerate.
11307 * hooks.h (hook_bool_uint_mode_true): Declare.
11308 * hooks.c (hook_bool_uint_mode_true): New function.
11309 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
11310 HARD_REGNO_MODE_OK.
11311 * genpreds.c (write_insn_preds_c): Add an include of target.h.
11312 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
11313 instead of HARD_REGNO_MODE_OK.
11314 * caller-save.c: Include target.h.
11315 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
11316 HARD_REGNO_MODE_OK.
11317 * combine.c (can_combine_p): Likewise.
11318 (combinable_i3pat): Likewise.
11319 (can_change_dest_mode): Likewise.
11320 * expr.c (init_expr_target): Likewise.
11321 (convert_move): Likewise.
11322 (convert_modes): Likewise.
11323 * ira.c (setup_prohibited_class_mode_regs): Likewise.
11324 (setup_prohibited_mode_move_regs): Likewise.
11325 * ira.h (target_ira): Likewise.
11326 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11327 * lra-constraints.c (process_alt_operands): Likewise.
11328 (split_reg): Likewise.
11329 * recog.c (peep2_find_free_register): Likewise.
11330 * ree.c (combine_reaching_defs): Likewise.
11331 * regcprop.c (maybe_mode_change): Likewise.
11332 * reginfo.c (init_reg_sets_1): Likewise.
11333 (choose_hard_reg_mode): Likewise.
11334 (simplifiable_subregs): Likewise.
11335 * regrename.c (check_new_reg_p): Likewise.
11336 * reload.c (find_valid_class): Likewise.
11337 (find_valid_class_1): Likewise.
11338 (reload_inner_reg_of_subreg): Likewise.
11339 (push_reload): Likewise.
11340 (combine_reloads): Likewise.
11341 (find_dummy_reload): Likewise.
11342 (find_reloads): Likewise.
11343 * reload1.c (find_reg): Likewise.
11344 (set_reload_reg): Likewise.
11345 (allocate_reload_reg): Likewise.
11346 (choose_reload_regs): Likewise.
11347 (reload_adjust_reg_for_temp): Likewise.
11348 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
11349 (simplify_subreg_regno): Likewise.
11350 * sel-sched.c (init_regs_for_mode): Likewise.
11351 * varasm.c (make_decl_rtl): Likewise.
11352 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
11353 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
11354 HARD_REGNO_MODE_OK.
11355 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
11356 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
11357 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11358 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
11359 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
11360 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11361 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
11362 (arc_mode_class): Delete.
11363 (HARD_REGNO_MODE_OK): Delete.
11364 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11365 (arc_hard_regno_mode_ok): Rename old array to...
11366 (arc_hard_regno_mode_ok_modes): ...this.
11367 (arc_conditional_register_usage): Update accordingly.
11368 (arc_mode_class): Make static.
11369 (arc_hard_regno_mode_ok): New function.
11370 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
11371 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
11372 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11373 (arm_hard_regno_mode_ok): Make static.
11374 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
11375 HARD_REGNO_MODE_OK.
11376 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
11377 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
11378 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
11380 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11381 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
11382 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
11383 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
11384 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
11385 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11386 * config/bfin/predicates.md (valid_reg_operand): Use
11387 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
11388 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
11389 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
11390 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11391 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
11392 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
11393 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11394 (cr16_hard_regno_mode_ok): Make static and return a bool.
11395 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
11396 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11397 (cris_hard_regno_mode_ok): New function.
11398 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
11399 (epiphany_mode_class): Delete.
11400 (HARD_REGNO_MODE_OK): Delete.
11401 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
11402 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11403 (hard_regno_mode_ok): Rename to...
11404 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
11405 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
11406 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
11407 HARD_REGNO_MODE_OK.
11408 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
11409 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
11410 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11411 (frv_hard_regno_mode_ok): Make static and return a bool.
11412 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
11413 HARD_REGNO_MODE_OK.
11414 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
11415 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
11416 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
11417 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
11419 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11420 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
11421 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
11422 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
11424 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11425 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
11426 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11427 (ia64_hard_regno_mode_ok): New function.
11428 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
11429 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11430 (iq2000_hard_regno_mode_ok): New function.
11431 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
11432 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11433 (lm32_hard_regno_mode_ok): New function.
11434 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
11435 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
11436 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
11437 instead of HARD_REGNO_MODE_OK.
11438 (m32c_hard_regno_ok): Rename to...
11439 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
11440 (m32c_cannot_change_mode_class): Update accordingly.
11441 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11442 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
11443 (m32r_mode_class): Delete.
11444 (HARD_REGNO_MODE_OK): Delete.
11445 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11446 (m32r_hard_regno_mode_ok): Rename to...
11447 (m32r_hard_regno_modes): ...this.
11448 (m32r_mode_class): Make static.
11449 (m32r_hard_regno_mode_ok): New function.
11450 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
11451 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
11452 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11453 (m68k_hard_regno_mode_ok): Make static.
11454 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
11455 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11456 (mcore_hard_regno_mode_ok): New function.
11457 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
11458 (HARD_REGNO_MODE_OK): Delete.
11459 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
11461 (microblaze_hard_regno_mode_ok_p): ...this and make static.
11462 (microblaze_hard_regno_mode_ok): New function.
11463 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11464 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
11465 (mips_hard_regno_mode_ok): Delete.
11466 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
11467 (mips_hard_regno_mode_ok_p): ...this and make static.
11468 (mips_hard_regno_mode_ok_p): Rename to...
11469 (mips_hard_regno_mode_ok_uncached): ...this.
11470 (mips_hard_regno_mode_ok): New function.
11471 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
11472 of HARD_REGNO_MODE_OK.
11473 (mips_option_override): Update after above name changes.
11474 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11475 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
11476 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
11477 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
11478 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
11479 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11480 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
11481 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
11482 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
11483 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11484 (msp430_hard_regno_mode_ok): Make static and return a bool.
11485 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
11486 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
11487 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
11489 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11490 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
11491 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
11492 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
11493 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
11494 (PA_HARD_REGNO_MODE_OK): ...this
11495 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
11496 (PA_HARD_REGNO_MODE_OK): ...this.
11497 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11498 (pa_hard_regno_mode_ok): New function.
11499 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
11500 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11501 (pdp11_hard_regno_mode_ok): New function.
11502 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
11503 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
11505 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
11507 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11508 (rs6000_hard_regno_mode_ok): Rename to...
11509 (rs6000_hard_regno_mode_ok_uncached): ...this.
11510 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11511 (rs6000_hard_regno_mode_ok): New function.
11512 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
11513 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
11514 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
11515 (riscv_hard_regno_mode_ok): ...this and make static.
11516 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11517 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
11518 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
11519 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11520 (rl78_hard_regno_mode_ok): Make static and return bool.
11521 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
11522 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
11524 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
11525 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11526 (rs6000_hard_regno_mode_ok): Rename to...
11527 (rs6000_hard_regno_mode_ok_uncached): ...this.
11528 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11529 (rs6000_hard_regno_mode_ok): New function.
11530 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
11531 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
11532 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11533 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
11534 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
11535 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
11536 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11537 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
11538 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
11539 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11540 (sh_hard_regno_mode_ok): Make static.
11541 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
11542 instead of HARD_REGNO_MODE_OK.
11543 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
11544 (sparc_mode_class): Delete.
11545 (HARD_REGNO_MODE_OK): Delete.
11546 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11547 (hard_regno_mode_classes): Make static.
11548 (sparc_mode_class): Likewise.
11549 (sparc_hard_regno_mode_ok): New function.
11550 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
11551 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
11552 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
11554 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11555 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
11556 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
11557 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
11558 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
11559 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11560 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
11561 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
11562 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11563 (visium_hard_regno_mode_ok): New function.
11564 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
11565 instead of HARD_REGNO_MODE_OK.
11566 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
11567 (HARD_REGNO_MODE_OK): Delete.
11568 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
11569 (xtensa_hard_regno_mode_ok_p): ...this and make static.
11570 (xtensa_option_override): Update accordingly.
11571 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11572 (xtensa_hard_regno_mode_ok): New function.
11573 * system.h (HARD_REGNO_MODE_OK): Poison.
11575 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11576 Alan Hayward <alan.hayward@arm.com>
11577 David Sherwood <david.sherwood@arm.com>
11579 * target.def (hard_regno_call_part_clobbered): New hook.
11580 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
11581 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
11582 * doc/tm.texi: Regenerate.
11583 * hooks.h (hook_bool_uint_mode_false): Declare.
11584 * hooks.c (hook_bool_uint_mode_false): New function.
11585 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11586 * cselib.c (cselib_process_insn): Use
11587 targetm.hard_regno_call_part_clobbered instead of
11588 HARD_REGNO_CALL_PART_CLOBBERED.
11589 * ira-conflicts.c (ira_build_conflicts): Likewise.
11590 * ira-costs.c (ira_tune_allocno_costs): Likewise.
11591 * lra-constraints.c (need_for_call_save_p): Likewise.
11592 * lra-lives.c: Include target.h.
11593 (check_pseudos_live_through_calls): Use
11594 targetm.hard_regno_call_part_clobbered instead of
11595 HARD_REGNO_CALL_PART_CLOBBERED.
11596 * regcprop.c: Include target.h.
11597 (copyprop_hardreg_forward_1): Use
11598 targetm.hard_regno_call_part_clobbered instead of
11599 HARD_REGNO_CALL_PART_CLOBBERED.
11600 * reginfo.c (choose_hard_reg_mode): Likewise.
11601 * regrename.c (check_new_reg_p): Likewise.
11602 * reload.c (find_equiv_reg): Likewise.
11603 * reload1.c (emit_reload_insns): Likewise.
11604 * sched-deps.c (deps_analyze_insn): Likewise.
11605 * sel-sched.c (init_regs_for_mode): Likewise.
11606 (mark_unavailable_hard_regs): Likewise.
11607 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
11608 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11609 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
11611 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11612 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11613 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
11615 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
11617 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11618 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11619 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
11621 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11622 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11623 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
11625 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11626 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
11628 * config/powerpcspe/powerpcspe.c
11629 (rs6000_hard_regno_call_part_clobbered): New function.
11630 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11631 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11632 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
11634 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11635 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11636 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
11638 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11639 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11640 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
11642 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11643 Alan Hayward <alan.hayward@arm.com>
11644 David Sherwood <david.sherwood@arm.com>
11646 * rtl.h (subreg_memory_offset): Declare.
11647 * emit-rtl.c (subreg_memory_offset): New function.
11648 * expmed.c (store_bit_field_1): Use it.
11649 * expr.c (undefined_operand_subword_p): Likewise.
11650 * simplify-rtx.c (simplify_subreg): Likewise.
11652 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11654 PR rtl-optimization/57448
11657 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11658 using atomic_load pattern.
11659 (expand_atomic_store): Likewise.
11661 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11664 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11665 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11666 handling. Use replace_call_with_value with NULL instead of
11667 gsi_replace, unlink_stmt_vdef and release_defs.
11669 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11672 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11674 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11677 * rtl.h (emit_library_call_value_1): Declare.
11678 (emit_library_call): Replace declaration with a series of overloads.
11679 Remove the parameter count argument.
11680 (emit_library_call_value): Likewise.
11681 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11682 with an "rtx_mode_t *".
11683 (emit_library_call_value): Delete.
11684 (emit_library_call): Likewise.
11685 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11686 (asan_emit_allocas_unpoison): Likewise.
11687 * builtins.c (expand_builtin_powi): Likewise.
11688 (expand_asan_emit_allocas_unpoison): Likewise.
11689 * cfgexpand.c (expand_main_function): Likewise.
11690 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11691 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11692 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11693 * config/arm/arm.c (arm_trampoline_init): Likewise.
11694 (arm_call_tls_get_addr): Likewise.
11695 (arm_expand_divmod_libfunc): Likewise.
11696 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11697 (smulsi3_highpart): Likewise.
11698 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11699 (c6x_expand_compare): Likewise.
11700 (c6x_expand_movmem): Likewise.
11701 * config/frv/frv.c (frv_trampoline_init): Likewise.
11702 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11703 (ix86_expand_divmod_libfunc): Likewise.
11704 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11705 (ia64_expand_compare): Likewise.
11706 (ia64_profile_hook): Likewise.
11707 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11708 (nonlocal_goto): Likewise.
11709 (restore_stack_nonlocal): Likewise.
11710 * config/m32r/m32r.c (block_move_call): Likewise.
11711 (m32r_trampoline_init): Likewise.
11712 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11713 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11714 (m68k_call_m68k_read_tp): Likewise.
11715 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11716 (microblaze_expand_divide): Likewise.
11717 * config/mips/mips.h (mips_args): Likewise.
11718 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11719 (MIPS_ICACHE_SYNC): Likewise.
11720 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11721 (nios2_trampoline_init): Likewise.
11722 * config/pa/pa.c (hppa_tls_call): Likewise.
11723 (pa_trampoline_init): Likewise.
11724 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11725 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11726 (expand_strn_compare): Likewise.
11727 (rs6000_generate_compare): Likewise.
11728 (rs6000_expand_float128_convert): Likewise.
11729 (output_profile_hook): Likewise.
11730 (rs6000_trampoline_init): Likewise.
11731 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11732 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11733 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11734 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11735 (rs6000_generate_compare): Likewise.
11736 (rs6000_expand_float128_convert): Likewise.
11737 (output_profile_hook): Likewise.
11738 (rs6000_trampoline_init): Likewise.
11739 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11740 * config/sh/sh.c (sh_trampoline_init): Likewise.
11741 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11742 (sparc_emit_float_lib_cmp): Likewise.
11743 (sparc32_initialize_trampoline): Likewise.
11744 (sparc64_initialize_trampoline): Likewise.
11745 (sparc_profile_hook): Likewise.
11746 * config/spu/spu.c (ea_load_store): Likewise.
11747 * config/spu/spu.md (floatunssidf2): Likewise.
11748 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11749 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11750 * config/visium/visium.c (expand_block_move_4): Likewise.
11751 (expand_block_move_2): Likewise.
11752 (expand_block_move_1): Likewise.
11753 (expand_block_set_4): Likewise.
11754 (expand_block_set_2): Likewise.
11755 (expand_block_set_1): Likewise.
11756 (visium_trampoline_init): Likewise.
11757 (visium_profile_hook): Likewise.
11758 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11759 (xtensa_setup_frame_addresses): Likewise.
11760 (xtensa_trampoline_init): Likewise.
11761 * except.c (sjlj_emit_function_enter): Likewise.
11762 (sjlj_emit_function_exit): Likewise.
11763 * explow.c (allocate_dynamic_stack_space): Likewise.
11764 (probe_stack_range): Likewise.
11765 * expr.c (convert_mode_scalar): Likewise.
11766 * optabs.c (expand_binop): Likewise.
11767 (expand_twoval_binop_libfunc): Likewise.
11768 (expand_unop): Likewise.
11769 (prepare_cmp_insn): Likewise.
11770 (prepare_float_lib_cmp): Likewise.
11771 (expand_float): Likewise.
11772 (expand_fix): Likewise.
11773 (expand_fixed_convert): Likewise.
11774 (maybe_emit_sync_lock_test_and_set): Likewise.
11775 (expand_atomic_compare_and_swap): Likewise.
11776 (expand_mem_thread_fence): Likewise.
11777 (expand_atomic_fetch_op): Likewise.
11779 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11781 * doc/generic.texi (OpenACC): Adjust URL.
11782 * doc/invoke.texi (C Dialect Options): Ditto.
11784 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11786 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11787 predicate for operand 1. Add (m,<S>) constraint.
11788 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11789 Prevent memory operand 1 with register operand 2.
11791 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11793 PR rtl-optimization/82024
11794 * combine.c (try_combine): If the combination result is a PARALLEL,
11795 and we only need to retain the SET in there that would be placed
11796 at I2, check that we can place that at I3 instead, before doing so.
11798 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11801 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11803 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11804 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11807 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11808 Jeff Law <law@redhat.com>
11810 * varasm.c (bss_initializer_p): Do not put constants into .bss
11811 (categorize_decl_for_section): Handle bss_initializer_p returning
11812 false when DECL_INITIAL is NULL.
11814 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11817 * config/s390/s390.c (s390_can_inline_p): New function.
11819 2017-09-01 Jeff Law <law@redhat.com>
11821 PR tree-optimization/82052
11822 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11823 Always initialize the returned slot after a hash table miss
11824 when INSERT is true.
11826 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11828 * config/s390/s390.md (mem_signal_fence): Remove.
11829 * doc/md.texi (mem_signal_fence): Remove.
11830 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11832 * target-insns.def (mem_signal_fence): Remove.
11834 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11837 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11840 * asan.c (create_odr_indicator): Strip name encoding from assembler
11841 name before appending it after __odr_asan_.
11843 2017-09-01 Martin Liska <mliska@suse.cz>
11845 PR tree-optimization/82059
11846 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11847 frequency only when an edge is redirected.
11849 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11851 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11852 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11853 (arc_conditional_register_usage): Remove ARC600 lp_count
11855 (arc_file_start): Emit Tag_ARC_CPU_variation.
11856 (arc_can_use_doloop_p): New conditions to use ZOLs.
11857 (hwloop_fail): New function.
11858 (hwloop_optimize): Likewise.
11859 (hwloop_pattern_reg): Likewise.
11860 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11861 (arc_reorg_loops): New function, calls reorg_loops.
11862 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11863 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11865 (arc_loop_hazard): Remove function, functionality moved into
11867 (arc_hazard): Remove arc_loop_hazard call.
11868 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11869 into hwloop_optimize.
11870 (arc_label_align): Remove ZOL handling.
11871 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11872 * config/arc/arc.md (doloop_begin): Remove pattern.
11873 (doloop_begin_i): Likewise.
11874 (doloop_end_i): Likewise.
11875 (doloop_fallback): Likewise.
11876 (doloop_fallback_m): Likewise.
11877 (doloop_end): Reimplement expand.
11878 (arc_lp): New pattern for LP instruction.
11879 (loop_end): New pattern.
11880 (loop_fail): Likewise.
11881 (decrement_and_branch_until_zero): Likewise.
11882 * config/arc/arc.opt (mlpc-width): New option.
11883 * doc/invoke.texi (mlpc-width): Document option.
11885 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11887 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11888 (arc_ccfsm_advance): Fix checking for delay slots.
11889 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11891 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11893 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11895 (movhi_insn): Update store instruction constraint which are saving
11896 6-bit short immediates.
11897 (movsi_insn): Consider also short scaled load operations.
11898 (zero_extendhisi2_i): Use Usd constraint instead of T.
11899 (extendhisi2_i): Add q constraint.
11900 (arc_clzsi2): Add type and length attributes.
11901 (arc_ctzsi2): Likewise.
11902 * config/arc/constraints.md (Usc): Update constraint, the
11903 assembler can parse two relocations for a single instruction.
11905 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11907 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11908 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11910 2017-08-31 Olivier Hainque <hainque@adacore.com>
11912 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11913 match as powerpc-wrs-vxworks*.
11915 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11917 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11918 register constraint for by-element operand.
11919 (aarch64_mls_elt_merge<mode>): Likewise.
11921 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11923 * config/arc/arc.c (arc_can_follow_jump): Check for short
11926 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11928 * config.gcc: Use g.opt for arc.
11929 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11930 functionality moved to ...
11931 (legitimate_scaled_address_p): New function, ...here.
11932 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11933 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11934 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11936 (arc_override_options): Handle G option.
11937 (arc_output_pic_addr_const): Correct function definition.
11938 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11939 (arc_decl_anon_ns_mem_p): Delete.
11940 (arc_in_small_data_p): Overhaul this function to take into
11941 consideration the value given via G option.
11942 (arc_rewrite_small_data_1): Renamed and corrected old
11943 arc_rewrite_small_data function.
11944 (arc_rewrite_small_data): New function.
11945 (small_data_pattern): Don't use pic_offset_table_rtx.
11946 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11947 * config/arc/simdext.md (movmisalignv2hi): Use
11948 prepare_move_operands function.
11950 (movmisalign*): Likewise.
11951 * doc/invoke.texi (ARC options): Document -G option.
11953 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11955 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11957 * config/arc/arc.c (arc_print_operand): Output scalled address for
11958 sdata whenever is possible.
11959 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11960 load/stores are available.
11961 (compact_sda_memory_operand): Check for the alignment required by
11962 code density instructions.
11963 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11965 * config/arc/constraints.md (Usd): Update constraint.
11966 (Us0): New constraint.
11967 (Usc): Update constraint.
11969 2017-08-31 Richard Biener <rguenther@suse.de>
11971 PR middle-end/82054
11972 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11973 function only once.
11975 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11977 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11978 Resize type_signature.
11980 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11981 Alan Hayward <alan.hayward@arm.com>
11982 David Sherwood <david.sherwood@arm.com>
11984 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11985 subregs whose inner modes can be stored in GPRs.
11986 (aarch64_classify_index): Likewise.
11988 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11989 Alan Hayward <alan.hayward@arm.com>
11990 David Sherwood <david.sherwood@arm.com>
11992 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11993 (V_INT_EQUIV): ...this.
11994 (v_cmp_result): Rename to...
11995 (v_int_equiv): ...this.
11996 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11997 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11998 (copysign<mode>3): Likewise.
11999 (aarch64_simd_bsl<mode>_internal): Likewise.
12000 (aarch64_simd_bsl<mode>): Likewise.
12001 (vec_cmp<mode><mode>): Likewise.
12002 (vcond<mode><mode>): Likewise.
12003 (vcond<v_cmp_mixed><mode>): Likewise.
12004 (vcondu<mode><v_cmp_mixed>): Likewise.
12005 (aarch64_cm<optab><mode>): Likewise.
12006 (aarch64_cmtst<mode>): Likewise.
12007 (aarch64_fac<optab><mode>): Likewise.
12008 (vec_perm_const<mode>): Likewise.
12009 (vcond_mask_<mode><v_cmp_result>): Rename to...
12010 (vcond_mask_<mode><v_int_equiv>): ...this.
12011 (vec_cmp<mode><v_cmp_result>): Rename to...
12012 (vec_cmp<mode><v_int_equiv>): ...this.
12014 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
12015 Alan Hayward <alan.hayward@arm.com>
12016 David Sherwood <david.sherwood@arm.com>
12018 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
12020 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
12021 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
12022 (UNSPEC_LD4_DREG): New unspecs.
12023 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
12024 (aarch64_ld2<mode>_dreg_be): Replace with...
12025 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
12027 (aarch64_ld3<mode>_dreg_le)
12028 (aarch64_ld3<mode>_dreg_be): Replace with...
12029 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
12031 (aarch64_ld4<mode>_dreg_le)
12032 (aarch64_ld4<mode>_dreg_be): Replace with...
12033 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
12036 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12038 PR tree-optimization/81987
12039 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
12040 insert an initializer in a location not dominated by the stride
12043 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
12045 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
12046 on the entire header of the finally block in the fallthru case.
12048 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
12050 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
12052 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
12054 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
12055 rs6000_emit_move_from_cr and call renamed function.
12056 (rs6000_emit_prologue): Call renamed functions.
12057 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
12058 movesi_from_cr, remove volatile CRs.
12060 2017-08-30 Jon Beniston <jon@beniston.com>
12061 Richard Biener <rguenther@suse.de>
12063 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
12064 of VECTOR_MODE_P check.
12065 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
12066 element vector types.
12068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12070 * df.h (df_read_modify_subreg_p): Remove in favor of...
12071 * rtl.h (read_modify_subreg_p): ...this new function. Take a
12072 const_rtx instead of an rtx.
12073 * cprop.c (local_cprop_find_used_regs): Update accordingly.
12074 * df-problems.c (df_word_lr_mark_ref): Likewise.
12075 * ira-lives.c (mark_pseudo_reg_live): Likewise.
12076 (mark_pseudo_reg_dead): Likewise.
12077 (mark_ref_dead): Likewise.
12078 * reginfo.c (init_subregs_of_mode): Likewise.
12079 * sched-deps.c (sched_analyze_1): Likewise.
12080 * df-scan.c (df_def_record_1): Likewise.
12081 (df_uses_record): Likewise.
12082 (df_read_modify_subreg_p): Remove in favor of...
12083 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
12084 const_rtx instead of an rtx.
12086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12087 Alan Hayward <alan.hayward@arm.com>
12088 David Sherwood <david.sherwood@arm.com>
12090 * rtl.h (partial_subreg_p): New function.
12091 * caller-save.c (save_call_clobbered_regs): Use it.
12092 * calls.c (expand_call): Likewise.
12093 * combine.c (combinable_i3pat): Likewise.
12094 (simplify_set): Likewise.
12095 (make_extraction): Likewise.
12096 (make_compound_operation_int): Likewise.
12097 (gen_lowpart_or_truncate): Likewise.
12098 (force_to_mode): Likewise.
12099 (make_field_assignment): Likewise.
12100 (reg_truncated_to_mode): Likewise.
12101 (record_truncated_value): Likewise.
12102 (move_deaths): Likewise.
12103 * cse.c (record_jump_cond): Likewise.
12104 (cse_insn): Likewise.
12105 * cselib.c (cselib_lookup_1): Likewise.
12106 * expmed.c (extract_bit_field_using_extv): Likewise.
12107 * function.c (assign_parm_setup_reg): Likewise.
12108 * ifcvt.c (noce_convert_multiple_sets): Likewise.
12109 * ira-build.c (create_insn_allocnos): Likewise.
12110 * lra-coalesce.c (merge_pseudos): Likewise.
12111 * lra-constraints.c (match_reload): Likewise.
12112 (simplify_operand_subreg): Likewise.
12113 (curr_insn_transform): Likewise.
12114 * lra-lives.c (process_bb_lives): Likewise.
12115 * lra.c (new_insn_reg): Likewise.
12116 (lra_substitute_pseudo): Likewise.
12117 * regcprop.c (mode_change_ok): Likewise.
12118 (maybe_mode_change): Likewise.
12119 (copyprop_hardreg_forward_1): Likewise.
12120 * reload.c (push_reload): Likewise.
12121 (find_reloads): Likewise.
12122 (find_reloads_subreg_address): Likewise.
12123 * reload1.c (alter_reg): Likewise.
12124 (eliminate_regs_1): Likewise.
12125 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12127 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
12129 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
12130 back to if statements, including unpack.
12132 2017-08-30 Martin Liska <mliska@suse.cz>
12134 PR inline-asm/82001
12135 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
12137 (func_checker::compare_asm_inputs_outputs): ... this function.
12138 (func_checker::compare_gimple_asm): Use the function to compare
12139 also ASM constrains.
12140 * ipa-icf-gimple.h: Rename the function.
12142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12143 Alan Hayward <alan.hayward@arm.com>
12144 David Sherwood <david.sherwood@arm.com>
12146 * coretypes.h (complex_mode): New type.
12147 * gdbhooks.py (build_pretty_printer): Handle it.
12148 * machmode.h (complex_mode): New class.
12149 (complex_mode::includes_p): New function.
12150 (is_complex_int_mode): Likewise.
12151 (is_complex_float_mode): Likewise.
12152 * genmodes.c (get_mode_class): Handle complex mode classes.
12153 * function.c (expand_function_end): Use is_complex_int_mode.
12155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12156 Alan Hayward <alan.hayward@arm.com>
12157 David Sherwood <david.sherwood@arm.com>
12159 * coretypes.h (scalar_mode_pod): New typedef.
12160 * gdbhooks.py (build_pretty_printer): Handle it.
12161 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
12162 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
12163 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
12164 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
12165 as_a <scalar_mode>.
12167 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12168 Alan Hayward <alan.hayward@arm.com>
12169 David Sherwood <david.sherwood@arm.com>
12171 * machmode.h (mode_for_vector): Take a scalar_mode instead
12173 * stor-layout.c (mode_for_vector): Likewise.
12174 * explow.c (promote_mode): Use as_a <scalar_mode>.
12175 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
12177 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12178 Alan Hayward <alan.hayward@arm.com>
12179 David Sherwood <david.sherwood@arm.com>
12181 * target.def (preferred_simd_mode): Take a scalar_mode
12182 instead of a machine_mode.
12183 * targhooks.h (default_preferred_simd_mode): Likewise.
12184 * targhooks.c (default_preferred_simd_mode): Likewise.
12185 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
12186 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
12187 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
12188 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
12189 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
12190 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
12191 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
12192 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
12194 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
12195 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
12196 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
12197 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
12198 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
12199 * doc/tm.texi: Regenerate.
12200 * optabs-query.c (can_vec_mask_load_store_p): Return false for
12203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12204 Alan Hayward <alan.hayward@arm.com>
12205 David Sherwood <david.sherwood@arm.com>
12207 * target.def (scalar_mode_supported_p): Take a scalar_mode
12208 instead of a machine_mode.
12209 * targhooks.h (default_scalar_mode_supported_p): Likewise.
12210 * targhooks.c (default_scalar_mode_supported_p): Likewise.
12211 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
12212 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12213 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
12214 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
12215 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
12216 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
12217 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
12218 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
12219 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
12220 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
12221 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
12222 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
12224 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
12225 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
12226 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12227 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12228 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
12230 * doc/tm.texi: Regenerate.
12232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12233 Alan Hayward <alan.hayward@arm.com>
12234 David Sherwood <david.sherwood@arm.com>
12236 * coretypes.h (opt_scalar_mode): New typedef.
12237 * gdbhooks.py (build_pretty_printers): Handle it.
12238 * machmode.h (mode_iterator::get_2xwider): Add overload for
12240 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
12242 * expr.c (convert_mode_scalar): Likewise.
12243 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12244 * optabs.c (expand_float): Likewise.
12245 (expand_fix): Likewise.
12246 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12249 Alan Hayward <alan.hayward@arm.com>
12250 David Sherwood <david.sherwood@arm.com>
12252 * optabs.c (expand_float): Explicitly check for scalars before
12253 using a branching expansion.
12254 (expand_fix): Likewise.
12256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12257 Alan Hayward <alan.hayward@arm.com>
12258 David Sherwood <david.sherwood@arm.com>
12260 * expr.c (convert_mode): Split scalar handling out into...
12261 (convert_mode_scalar): ...this new function. Treat the modes
12264 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12265 Alan Hayward <alan.hayward@arm.com>
12266 David Sherwood <david.sherwood@arm.com>
12268 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
12270 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
12272 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12273 Alan Hayward <alan.hayward@arm.com>
12274 David Sherwood <david.sherwood@arm.com>
12276 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
12277 rather than a machine_mode.
12278 (fixed_from_string): Likewise.
12279 (fixed_convert): Likewise.
12280 (fixed_convert_from_int): Likewise.
12281 (fixed_convert_from_real): Likewise.
12282 (real_convert_from_fixed): Likewise.
12283 * fixed-value.c (fixed_from_double_int): Likewise.
12284 (fixed_from_string): Likewise.
12285 (fixed_convert): Likewise.
12286 (fixed_convert_from_int): Likewise.
12287 (fixed_convert_from_real): Likewise.
12288 (real_convert_from_fixed): Likewise.
12289 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
12291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12292 Alan Hayward <alan.hayward@arm.com>
12293 David Sherwood <david.sherwood@arm.com>
12295 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
12296 of separate mode class checks. Do not allow vector modes here.
12297 (immed_wide_int_const): Use as_a <scalar_mode>.
12298 * explow.c (trunc_int_for_mode): Likewise.
12299 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
12300 (wi::shwi): Likewise.
12301 (wi::min_value): Likewise.
12302 (wi::max_value): Likewise.
12303 * dwarf2out.c (loc_descriptor): Likewise.
12304 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
12305 for CONST_WIDE_INT.
12307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12308 Alan Hayward <alan.hayward@arm.com>
12309 David Sherwood <david.sherwood@arm.com>
12311 * tree.h (SCALAR_TYPE_MODE): New macro.
12312 * expr.c (expand_expr_addr_expr_1): Use it.
12313 (expand_expr_real_2): Likewise.
12314 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
12315 (fold_convert_const_fixed_from_int): Likewise.
12316 (fold_convert_const_fixed_from_real): Likewise.
12317 (native_encode_fixed): Likewise
12318 (native_encode_complex): Likewise
12319 (native_encode_vector): Likewise.
12320 (native_interpret_fixed): Likewise.
12321 (native_interpret_real): Likewise.
12322 (native_interpret_complex): Likewise.
12323 (native_interpret_vector): Likewise.
12324 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
12325 (simd_clone_adjust_argument_types): Likewise.
12326 (simd_clone_init_simd_arrays): Likewise.
12327 (simd_clone_adjust): Likewise.
12328 * stor-layout.c (layout_type): Likewise.
12329 * tree.c (build_minus_one_cst): Likewise.
12330 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12331 * tree-inline.c (estimate_move_cost): Likewise.
12332 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
12333 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
12334 (vectorizable_reduction): Likewise.
12335 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
12336 (vect_recog_mixed_size_cond_pattern): Likewise.
12337 (check_bool_pattern): Likewise.
12338 (adjust_bool_pattern): Likewise.
12339 (search_type_for_mask_1): Likewise.
12340 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
12341 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12342 (vectorizable_load): Likewise.
12343 (vectorizable_store): Likewise.
12344 * ubsan.c (ubsan_encode_value): Likewise.
12345 * varasm.c (output_constant): Likewise.
12347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12348 Alan Hayward <alan.hayward@arm.com>
12349 David Sherwood <david.sherwood@arm.com>
12351 * coretypes.h (scalar_mode): New class.
12352 * machmode.h (scalar_mode): Likewise.
12353 (scalar_mode::includes_p): New function.
12354 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
12355 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
12356 * genmodes.c (get_mode_class): Handle remaining scalar modes.
12357 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
12358 * expmed.c (store_bit_field_1): Likewise.
12359 (extract_bit_field_1): Likewise.
12360 * expr.c (write_complex_part): Likewise.
12361 (read_complex_part): Likewise.
12362 (emit_move_complex_push): Likewise.
12363 (expand_expr_real_2): Likewise.
12364 * function.c (assign_parm_setup_reg): Likewise.
12365 (assign_parms_unsplit_complex): Likewise.
12366 * optabs.c (expand_binop): Likewise.
12367 * rtlanal.c (subreg_get_info): Likewise.
12368 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12369 * varasm.c (output_constant_pool_2): Likewise.
12371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12372 Alan Hayward <alan.hayward@arm.com>
12373 David Sherwood <david.sherwood@arm.com>
12375 * expmed.c (extract_high_half): Use scalar_int_mode and remove
12377 (expmed_mult_highpart_optab): Likewise.
12378 (expmed_mult_highpart): Likewise.
12380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12381 Alan Hayward <alan.hayward@arm.com>
12382 David Sherwood <david.sherwood@arm.com>
12384 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
12385 instead of a machine_mode.
12386 (builtin_memset_read_str): Likewise.
12387 * builtins.c (c_readstr): Likewise.
12388 (builtin_memcpy_read_str): Likewise.
12389 (builtin_strncpy_read_str): Likewise.
12390 (builtin_memset_read_str): Likewise.
12391 (builtin_memset_gen_str): Likewise.
12392 (expand_builtin_signbit): Use scalar_int_mode for local variables.
12393 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
12394 instead of a machine_mode.
12395 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
12397 (make_extraction): Likewise.
12398 (try_widen_shift_mode): Take and return scalar_int_modes instead
12400 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
12401 a scalar_int_mode instead of a machine_mode.
12402 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
12403 (avr_addr_space_pointer_mode): Likewise.
12404 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
12405 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
12406 (msp430_unwind_word_mode): Likewise.
12407 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
12408 (spu_addr_space_pointer_mode): Likewise.
12409 (spu_addr_space_address_mode): Likewise.
12410 (spu_libgcc_cmp_return_mode): Likewise.
12411 (spu_libgcc_shift_count_mode): Likewise.
12412 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
12413 (rl78_addr_space_pointer_mode): Likewise.
12414 (fl78_unwind_word_mode): Likewise.
12415 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
12417 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
12418 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
12419 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
12420 (mips_valid_pointer_mode): Likewise.
12421 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
12422 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
12423 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
12425 (ft32_addr_space_address_mode): Likewise.
12426 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
12427 scalar_int_mode instead of a machine_mode.
12428 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
12430 (m32c_addr_space_address_mode): Likewise.
12431 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
12432 (rs6000_eh_return_filter_mode): Likewise.
12433 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
12434 (rs6000_eh_return_filter_mode): Likewise.
12435 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
12436 (s390_libgcc_shift_count_mode): Likewise.
12437 (s390_unwind_word_mode): Likewise.
12438 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
12440 * target.def (mode_rep_extended): Likewise.
12441 (valid_pointer_mode): Likewise.
12442 (addr_space.valid_pointer_mode): Likewise.
12443 (eh_return_filter_mode): Return a scalar_int_mode rather than
12445 (libgcc_cmp_return_mode): Likewise.
12446 (libgcc_shift_count_mode): Likewise.
12447 (unwind_word_mode): Likewise.
12448 (addr_space.pointer_mode): Likewise.
12449 (addr_space.address_mode): Likewise.
12450 * doc/tm.texi: Regenerate.
12451 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
12453 (do_jump): Use scalar_int_mode for local variables.
12454 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
12455 rather than a machine_mode.
12456 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
12457 (scompare_loc_descriptor_wide): Likewise.
12458 (scompare_loc_descriptor_narrow): Likewise.
12459 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
12461 * except.c (sjlj_emit_dispatch_table): Likewise.
12462 (expand_builtin_eh_copy_values): Likewise.
12463 * explow.c (convert_memory_address_addr_space_1): Likewise.
12464 Take a scalar_int_mode rather than a machine_mode.
12465 (convert_memory_address_addr_space): Take a scalar_int_mode rather
12466 than a machine_mode.
12467 (memory_address_addr_space): Use scalar_int_mode for local variables.
12468 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
12469 rather than a machine_mode.
12470 * expmed.c (mask_rtx): Likewise.
12471 (init_expmed_one_conv): Likewise.
12472 (expand_mult_highpart_adjust): Likewise.
12473 (extract_high_half): Likewise.
12474 (expmed_mult_highpart_optab): Likewise.
12475 (expmed_mult_highpart): Likewise.
12476 (expand_smod_pow2): Likewise.
12477 (expand_sdiv_pow2): Likewise.
12478 (emit_store_flag_int): Likewise.
12479 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
12481 (extract_low_bits): Likewise.
12482 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
12484 * expr.c (pieces_addr::adjust): Likewise.
12485 (can_store_by_pieces): Likewise.
12486 (store_by_pieces): Likewise.
12487 (clear_by_pieces_1): Likewise.
12488 (expand_expr_addr_expr_1): Likewise.
12489 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
12490 (expand_expr_real_1): Likewise.
12491 (try_casesi): Likewise.
12492 * final.c (shorten_branches): Likewise.
12493 * fold-const.c (fold_convert_const_int_from_fixed): Change the
12494 type of "mode" to machine_mode.
12495 * internal-fn.c (expand_arith_overflow_result_store): Take a
12496 scalar_int_mode rather than a machine_mode.
12497 (expand_mul_overflow): Use scalar_int_mode for local variables.
12498 * loop-doloop.c (doloop_modify): Likewise.
12499 (doloop_optimize): Likewise.
12500 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
12501 than a machine_mode.
12502 (expand_doubleword_shift_condmove): Likewise.
12503 (expand_doubleword_shift): Likewise.
12504 (expand_doubleword_clz): Likewise.
12505 (expand_doubleword_popcount): Likewise.
12506 (expand_doubleword_parity): Likewise.
12507 (expand_absneg_bit): Use scalar_int_mode for local variables.
12508 (prepare_float_lib_cmp): Likewise.
12509 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
12510 rather than a machine_mode.
12511 (convert_memory_address_addr_space): Likewise.
12512 (get_mode_bounds): Likewise.
12513 (get_address_mode): Return a scalar_int_mode rather than a
12515 * rtlanal.c (get_address_mode): Likewise.
12516 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
12517 than a machine_mode.
12518 * targhooks.c (default_mode_rep_extended): Likewise.
12519 (default_valid_pointer_mode): Likewise.
12520 (default_addr_space_valid_pointer_mode): Likewise.
12521 (default_eh_return_filter_mode): Return a scalar_int_mode rather
12522 than a machine_mode.
12523 (default_libgcc_cmp_return_mode): Likewise.
12524 (default_libgcc_shift_count_mode): Likewise.
12525 (default_unwind_word_mode): Likewise.
12526 (default_addr_space_pointer_mode): Likewise.
12527 (default_addr_space_address_mode): Likewise.
12528 * targhooks.h (default_eh_return_filter_mode): Likewise.
12529 (default_libgcc_cmp_return_mode): Likewise.
12530 (default_libgcc_shift_count_mode): Likewise.
12531 (default_unwind_word_mode): Likewise.
12532 (default_addr_space_pointer_mode): Likewise.
12533 (default_addr_space_address_mode): Likewise.
12534 (default_mode_rep_extended): Take a scalar_int_mode rather than
12536 (default_valid_pointer_mode): Likewise.
12537 (default_addr_space_valid_pointer_mode): Likewise.
12538 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
12540 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
12541 rather than a machine_mode.
12542 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
12543 for local variables.
12544 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12545 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
12546 than a machine_mode.
12548 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12549 Alan Hayward <alan.hayward@arm.com>
12550 David Sherwood <david.sherwood@arm.com>
12552 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
12553 the mode argument to scalar_int_mode.
12554 (do_jump_by_parts_zero_rtx): Likewise.
12555 (do_jump_by_parts_equality_rtx): Likewise.
12556 (do_jump_by_parts_greater): Take a mode argument.
12557 (do_jump_by_parts_equality): Likewise.
12558 (do_jump_1): Update calls accordingly.
12560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12561 Alan Hayward <alan.hayward@arm.com>
12562 David Sherwood <david.sherwood@arm.com>
12564 * is-a.h (safe_dyn_cast): New function.
12565 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
12566 (jump_table_for_label): Likewise.
12567 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
12568 instead of an rtx_insn *.
12569 (shorten_branches): Use dyn_cast instead of LABEL_P and
12570 JUMP_TABLE_DATA_P. Use jump_table_for_label and
12571 rtx_jump_table_data::get_data_mode.
12572 (final_scan_insn): Likewise.
12574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12575 Alan Hayward <alan.hayward@arm.com>
12576 David Sherwood <david.sherwood@arm.com>
12578 * combine.c (try_combine): Use is_a <scalar_int_mode> when
12579 trying to combine a full-register integer set with a subreg
12582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12583 Alan Hayward <alan.hayward@arm.com>
12584 David Sherwood <david.sherwood@arm.com>
12586 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
12587 that is always either address_mode or pointer_mode.
12589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12590 Alan Hayward <alan.hayward@arm.com>
12591 David Sherwood <david.sherwood@arm.com>
12593 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
12594 when the two are known to be equal.
12596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12597 Alan Hayward <alan.hayward@arm.com>
12598 David Sherwood <david.sherwood@arm.com>
12600 * simplify-rtx.c (simplify_const_unary_operation): Use
12601 is_a <scalar_int_mode> instead of checking for a nonzero
12602 precision. Forcibly convert op_mode to a scalar_int_mode
12603 in that case. More clearly differentiate the operand and
12604 result modes and use the former when deciding what the value
12605 of a count-bits operation should be. Use is_int_mode instead
12606 of checking for a MODE_INT. Remove redundant check for whether
12607 this mode has a zero precision.
12609 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12610 Alan Hayward <alan.hayward@arm.com>
12611 David Sherwood <david.sherwood@arm.com>
12613 * optabs.c (widen_leading): Change the type of the mode argument
12614 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
12615 (widen_bswap): Likewise.
12616 (expand_parity): Likewise.
12617 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
12618 (expand_ffs): Likewise.
12619 (epand_unop): Check for scalar integer modes before calling the
12622 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12623 Alan Hayward <alan.hayward@arm.com>
12624 David Sherwood <david.sherwood@arm.com>
12626 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
12628 (expand_expr_real_1): Update call accordingly.
12630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12631 Alan Hayward <alan.hayward@arm.com>
12632 David Sherwood <david.sherwood@arm.com>
12634 * expmed.c (store_bit_field_using_insv): Add op0_mode and
12635 value_mode arguments. Use scalar_int_mode internally.
12636 (store_bit_field_1): Rename the new integer mode from imode
12637 to op0_mode and use it instead of GET_MODE (op0). Update calls
12638 to store_split_bit_field, store_bit_field_using_insv and
12639 store_fixed_bit_field.
12640 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
12641 Use scalar_int_mode internally. Use a bit count rather than a mode
12642 when calculating the largest bit size for get_best_mode.
12643 Update calls to store_split_bit_field and store_fixed_bit_field_1.
12644 (store_fixed_bit_field_1): Add mode and value_mode arguments.
12645 Remove assertion that OP0 has a scalar integer mode.
12646 (store_split_bit_field): Add op0_mode and value_mode arguments.
12647 Update calls to extract_fixed_bit_field.
12648 (extract_bit_field_using_extv): Add an op0_mode argument.
12649 Use scalar_int_mode internally.
12650 (extract_bit_field_1): Rename the new integer mode from imode to
12651 op0_mode and use it instead of GET_MODE (op0). Update calls to
12652 extract_split_bit_field, extract_bit_field_using_extv and
12653 extract_fixed_bit_field.
12654 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12655 to extract_split_bit_field and extract_fixed_bit_field_1.
12656 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12657 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12658 on the target mode.
12659 (extract_split_bit_field): Add an op0_mode argument. Update call
12660 to extract_fixed_bit_field.
12662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12663 Alan Hayward <alan.hayward@arm.com>
12664 David Sherwood <david.sherwood@arm.com>
12666 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12667 * explow.c (hard_function_value): Likewise.
12668 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12669 convert_to_mode call outside the loop.
12670 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12671 for the mode iterator. Require the mode specified by max_pieces
12673 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12675 (copy_blkmode_to_reg): Likewise.
12676 (set_storage_via_setmem): Likewise.
12677 * optabs.c (prepare_cmp_insn): Likewise.
12678 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12679 * stor-layout.c (finish_bitfield_representative): Likewise.
12681 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12682 Alan Hayward <alan.hayward@arm.com>
12683 David Sherwood <david.sherwood@arm.com>
12685 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12686 * expr.c (convert_move): Use them.
12687 (convert_modes): Likewise.
12688 (store_expr_with_bounds): Likewise.
12690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12691 Alan Hayward <alan.hayward@arm.com>
12692 David Sherwood <david.sherwood@arm.com>
12694 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12695 parameter for the mode of "x". Remove the "known_x", "known_mode"
12696 and "known_ret" arguments. Change the type of the mode argument
12697 to scalar_int_mode.
12698 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12699 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12700 (reg_num_sign_bit_copies_for_combine): Likewise.
12701 * rtlanal.c (nonzero_bits1): Likewise.
12702 (num_sign_bit_copies1): Likewise.
12703 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12704 (reg_num_sign_bit_copies_general): Likewise.
12705 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12706 (reg_nonzero_bits_general): Likewise.
12708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12709 Alan Hayward <alan.hayward@arm.com>
12710 David Sherwood <david.sherwood@arm.com>
12712 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12713 than in subroutines. Return 1 for non-integer modes.
12714 (cached_num_sign_bit_copies): Change the type of the mode parameter
12715 to scalar_int_mode.
12716 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12717 classes. Handle CONST_INT_P first and then check whether X also
12718 has a scalar integer mode. Check the same thing for inner registers
12719 of a SUBREG and for values that are being extended or truncated.
12721 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12722 Alan Hayward <alan.hayward@arm.com>
12723 David Sherwood <david.sherwood@arm.com>
12725 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12726 in subroutines. Return the mode mask for non-integer modes.
12727 (cached_nonzero_bits): Change the type of the mode parameter
12728 to scalar_int_mode.
12729 (nonzero_bits1): Likewise. Remove early exit for other mode
12730 classes. Handle CONST_INT_P first and then check whether X
12731 also has a scalar integer mode.
12733 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12734 Alan Hayward <alan.hayward@arm.com>
12735 David Sherwood <david.sherwood@arm.com>
12737 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12738 Return a scalar_int_mode and assert that the size is greater than
12740 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12741 (op_by_pieces_d::op_by_pieces_d): Likewise.
12742 (op_by_pieces_d::run): Likewise.
12743 (can_store_by_pieces): Likewise.
12745 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12746 Alan Hayward <alan.hayward@arm.com>
12747 David Sherwood <david.sherwood@arm.com>
12749 * combine.c (extract_left_shift): Add a mode argument and update
12751 (make_compound_operation_int): Change the type of the mode parameter
12752 to scalar_int_mode and update the call to extract_left_shift.
12754 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12755 Alan Hayward <alan.hayward@arm.com>
12756 David Sherwood <david.sherwood@arm.com>
12758 * combine.c (simplify_and_const_int): Change the type of the mode
12759 parameter to scalar_int_mode.
12760 (simplify_and_const_int_1): Likewise. Update recursive call.
12762 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12763 Alan Hayward <alan.hayward@arm.com>
12764 David Sherwood <david.sherwood@arm.com>
12766 * combine.c (simplify_compare_const): Check that the mode is a
12767 scalar_int_mode (rather than VOIDmode) before testing its
12769 (simplify_comparison): Move COMPARISON_P handling out of the
12770 loop and restrict the latter part of the loop to scalar_int_modes.
12771 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12772 and when considering SUBREG_REGs. Use is_int_mode instead of
12773 checking GET_MODE_CLASS against MODE_INT.
12775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12776 Alan Hayward <alan.hayward@arm.com>
12777 David Sherwood <david.sherwood@arm.com>
12779 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12780 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12781 shift_unit_mode and for modes involved in scalar shifts.
12783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12784 Alan Hayward <alan.hayward@arm.com>
12785 David Sherwood <david.sherwood@arm.com>
12787 * combine.c (force_int_to_mode): New function, split out from...
12788 (force_to_mode): ...here. Keep xmode up-to-date and use it
12789 instead of GET_MODE (x).
12791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12792 Alan Hayward <alan.hayward@arm.com>
12793 David Sherwood <david.sherwood@arm.com>
12795 * optabs-query.h (extraction_insn::struct_mode): Change type to
12796 opt_scalar_int_mode and update comment.
12797 (extraction_insn::field_mode): Change type to scalar_int_mode.
12798 (extraction_insn::pos_mode): Likewise.
12799 * combine.c (make_extraction): Update accordingly.
12800 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12801 (get_optab_extraction_insn): Likewise.
12802 * recog.c (simplify_while_replacing): Likewise.
12803 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12804 parameter to opt_scalar_int_mode.
12806 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12807 Alan Hayward <alan.hayward@arm.com>
12808 David Sherwood <david.sherwood@arm.com>
12810 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12811 to a scalar_int_mode instead of a machine_mode.
12812 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12813 (get_best_mode): Return a boolean and use a pointer argument to store
12814 the selected mode. Replace the limit mode parameter with a bit limit.
12815 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12816 for the values returned by bit_field_mode_iterator::next_mode.
12817 (store_bit_field): Update call to get_best_mode.
12818 (store_fixed_bit_field): Likewise.
12819 (extract_fixed_bit_field): Likewise.
12820 * expr.c (optimize_bitfield_assignment_op): Likewise.
12821 * fold-const.c (optimize_bit_field_compare): Likewise.
12822 (fold_truth_andor_1): Likewise.
12823 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12824 Update for new type of m_mode.
12825 (get_best_mode): As above.
12827 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12828 Alan Hayward <alan.hayward@arm.com>
12829 David Sherwood <david.sherwood@arm.com>
12831 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12832 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12833 (store_bit_field): Check is_a <scalar_int_mode> before calling
12834 strict_volatile_bitfield_p.
12835 (extract_bit_field): Likewise.
12837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12838 Alan Hayward <alan.hayward@arm.com>
12839 David Sherwood <david.sherwood@arm.com>
12841 * target.def (cstore_mode): Return a scalar_int_mode.
12842 * doc/tm.texi: Regenerate.
12843 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12844 * targhooks.h (default_cstore_mode): Likewise.
12845 * targhooks.c (default_cstore_mode): Likewise, using a forced
12847 * expmed.c (emit_cstore): Expect the target of the cstore to be
12850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12851 Alan Hayward <alan.hayward@arm.com>
12852 David Sherwood <david.sherwood@arm.com>
12854 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12856 (niter_desc): Likewise mode.
12857 (iv_analyze): Add a mode parameter.
12859 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12860 and change its type to scalar_int_mode.
12861 * loop-iv.c: Update commentary at head of file.
12862 (iv_constant): Pass the mode paraeter before the rtx it describes
12863 and change its type to scalar_int_mode. Remove VOIDmode handling.
12864 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12865 (iv_extend): Likewise.
12866 (shorten_into_mode): Likewise.
12867 (iv_add): Use scalar_int_mode.
12868 (iv_mult): Likewise.
12869 (iv_shift): Likewise.
12870 (canonicalize_iv_subregs): Likewise.
12871 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12872 it describes and change its mode to scalar_int_mode. Also change
12873 the type of the returned inner_mode to scalar_int_mode.
12874 (get_biv_step): Likewise, turning outer_mode from a pointer
12875 into a direct parameter. Update call to get_biv_step_1.
12876 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12877 iv_constant and get_biv_step.
12878 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12879 and change its type to scalar_int_mode. Don't initialise iv->mode
12880 to VOIDmode and remove later checks for its still being VOIDmode.
12881 Update calls to iv_analyze_op and iv_analyze_expr. Check
12882 is_a <scalar_int_mode> when changing the mode under consideration.
12883 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12884 Update call to iv_analyze_expr.
12885 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12886 inner register is not also a scalar_int_mode. Update call to
12888 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12889 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12890 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12891 separate mode class checks. Update calls to iv_analyze. Remove
12892 fix-up of VOIDmodes after iv_analyze_biv.
12893 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12894 don't have a scalar_int_mode. Update call to biv_p.
12896 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12897 Alan Hayward <alan.hayward@arm.com>
12898 David Sherwood <david.sherwood@arm.com>
12900 * cfgexpand.c (convert_debug_memory_address): Use
12901 as_a <scalar_int_mode>.
12902 * combine.c (expand_compound_operation): Likewise.
12903 (make_extraction): Likewise.
12904 (change_zero_ext): Likewise.
12905 (simplify_comparison): Likewise.
12906 * cse.c (cse_insn): Likewise.
12907 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12908 (mem_loc_descriptor): Likewise.
12909 (loc_descriptor): Likewise.
12910 * expmed.c (init_expmed_one_mode): Likewise.
12911 (synth_mult): Likewise.
12912 (emit_store_flag_1): Likewise.
12913 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12914 of a comparison with size.
12915 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12916 (reduce_to_bit_field_precision): Likewise.
12917 * function.c (expand_function_end): Likewise.
12918 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12919 * loop-doloop.c (doloop_modify): Likewise.
12920 * optabs.c (expand_binop): Likewise.
12921 (expand_unop): Likewise.
12922 (expand_copysign_absneg): Likewise.
12923 (prepare_cmp_insn): Likewise.
12924 (maybe_legitimize_operand): Likewise.
12925 * recog.c (const_scalar_int_operand): Likewise.
12926 * rtlanal.c (get_address_mode): Likewise.
12927 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12928 (simplify_cond_clz_ctz): Likewise.
12929 * tree-nested.c (get_nl_goto_field): Likewise.
12930 * tree.c (build_vector_type_for_mode): Likewise.
12931 * var-tracking.c (use_narrower_mode): Likewise.
12933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12934 Alan Hayward <alan.hayward@arm.com>
12935 David Sherwood <david.sherwood@arm.com>
12937 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12938 * builtins.c (expand_builtin_signbit): Use it.
12939 * cfgexpand.c (expand_debug_expr): Likewise.
12940 * dojump.c (do_jump): Likewise.
12941 (do_compare_and_jump): Likewise.
12942 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12943 * expmed.c (make_tree): Likewise.
12944 * expr.c (expand_expr_real_2): Likewise.
12945 (expand_expr_real_1): Likewise.
12946 (try_casesi): Likewise.
12947 * fold-const-call.c (fold_const_call_ss): Likewise.
12948 * fold-const.c (unextend): Likewise.
12949 (extract_muldiv_1): Likewise.
12950 (fold_single_bit_test): Likewise.
12951 (native_encode_int): Likewise.
12952 (native_encode_string): Likewise.
12953 (native_interpret_int): Likewise.
12954 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12955 * internal-fn.c (expand_addsub_overflow): Likewise.
12956 (expand_neg_overflow): Likewise.
12957 (expand_mul_overflow): Likewise.
12958 (expand_arith_overflow): Likewise.
12959 * match.pd: Likewise.
12960 * stor-layout.c (layout_type): Likewise.
12961 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12962 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12963 * tree-ssanames.c (get_range_info): Likewise.
12964 * tree-switch-conversion.c (array_value_type) Likewise.
12965 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12966 (vect_recog_divmod_pattern): Likewise.
12967 (vect_recog_mixed_size_cond_pattern): Likewise.
12968 * tree-vrp.c (extract_range_basic): Likewise.
12969 (simplify_float_conversion_using_ranges): Likewise.
12970 * tree.c (int_fits_type_p): Likewise.
12971 * ubsan.c (instrument_bool_enum_load): Likewise.
12972 * varasm.c (mergeable_string_section): Likewise.
12973 (narrowing_initializer_constant_valid_p): Likewise.
12974 (output_constant): Likewise.
12976 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12977 Alan Hayward <alan.hayward@arm.com>
12978 David Sherwood <david.sherwood@arm.com>
12980 * machmode.h (NARROWEST_INT_MODE): New macro.
12981 * expr.c (alignment_for_piecewise_move): Use it instead of
12982 GET_CLASS_NARROWEST_MODE (MODE_INT).
12983 (push_block): Likewise.
12984 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12986 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12989 Alan Hayward <alan.hayward@arm.com>
12990 David Sherwood <david.sherwood@arm.com>
12992 * postreload.c (move2add_valid_value_p): Change the type of the
12993 mode parameter to scalar_int_mode.
12994 (move2add_use_add2_insn): Add a mode parameter and use it instead
12996 (move2add_use_add3_insn): Likewise.
12997 (reload_cse_move2add): Update accordingly.
12999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13000 Alan Hayward <alan.hayward@arm.com>
13001 David Sherwood <david.sherwood@arm.com>
13003 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
13005 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
13006 * optabs.c (expand_unop): Likewise.
13008 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13009 Alan Hayward <alan.hayward@arm.com>
13010 David Sherwood <david.sherwood@arm.com>
13012 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
13013 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
13014 (popcount_loc_descriptor): Likewise.
13015 (bswap_loc_descriptor): Likewise.
13016 (rotate_loc_descriptor): Likewise.
13017 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
13018 calling the functions above.
13020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13021 Alan Hayward <alan.hayward@arm.com>
13022 David Sherwood <david.sherwood@arm.com>
13024 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
13026 (try_combine): Likewise.
13027 (simplify_if_then_else): Likewise.
13028 * cse.c (cse_insn): Likewise.
13029 * dwarf2out.c (mem_loc_descriptor): Likewise.
13030 * emit-rtl.c (gen_lowpart_common): Likewise.
13031 * simplify-rtx.c (simplify_truncation): Likewise.
13032 (simplify_binary_operation_1): Likewise.
13033 (simplify_const_relational_operation): Likewise.
13034 (simplify_ternary_operation): Likewise.
13035 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
13037 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13038 Alan Hayward <alan.hayward@arm.com>
13039 David Sherwood <david.sherwood@arm.com>
13041 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
13042 * reload.c (push_reload): Likewise.
13043 (find_reloads): Likewise.
13045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13046 Alan Hayward <alan.hayward@arm.com>
13047 David Sherwood <david.sherwood@arm.com>
13049 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
13050 (make_compound_operation_int): Likewise.
13051 (change_zero_ext): Likewise.
13052 * expr.c (convert_move): Likewise.
13053 (convert_modes): Likewise.
13054 * fwprop.c (forward_propagate_subreg): Likewise.
13055 * loop-iv.c (get_biv_step_1): Likewise.
13056 * optabs.c (widen_operand): Likewise.
13057 * postreload.c (move2add_valid_value_p): Likewise.
13058 * recog.c (simplify_while_replacing): Likewise.
13059 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13060 (simplify_binary_operation_1): Likewise. Remove redundant
13061 mode equality check.
13063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13064 Alan Hayward <alan.hayward@arm.com>
13065 David Sherwood <david.sherwood@arm.com>
13067 * combine.c (combine_simplify_rtx): Add checks for
13068 is_a <scalar_int_mode>.
13069 (simplify_if_then_else): Likewise.
13070 (make_field_assignment): Likewise.
13071 (simplify_comparison): Likewise.
13072 * ifcvt.c (noce_try_bitop): Likewise.
13073 * loop-invariant.c (canonicalize_address_mult): Likewise.
13074 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13077 Alan Hayward <alan.hayward@arm.com>
13078 David Sherwood <david.sherwood@arm.com>
13080 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
13081 is_a <scalar_int_mode> instead of != BLKmode.
13083 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13084 Alan Hayward <alan.hayward@arm.com>
13085 David Sherwood <david.sherwood@arm.com>
13087 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
13088 instead of != VOIDmode.
13089 * combine.c (if_then_else_cond): Likewise.
13090 (change_zero_ext): Likewise.
13091 * dwarf2out.c (mem_loc_descriptor): Likewise.
13092 (loc_descriptor): Likewise.
13093 * rtlanal.c (canonicalize_condition): Likewise.
13094 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
13096 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13097 Alan Hayward <alan.hayward@arm.com>
13098 David Sherwood <david.sherwood@arm.com>
13100 * simplify-rtx.c (simplify_binary_operation_1): Use
13101 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
13103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13104 Alan Hayward <alan.hayward@arm.com>
13105 David Sherwood <david.sherwood@arm.com>
13107 * wide-int.h (int_traits<unsigned char>) New class.
13108 (int_traits<unsigned short>) Likewise.
13109 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
13110 Use GET_MODE_UNIT_PRECISION and remove redundant test for
13112 * combine.c (set_nonzero_bits_and_sign_copies): Use
13113 is_a <scalar_int_mode>.
13114 (find_split_point): Likewise.
13115 (combine_simplify_rtx): Likewise.
13116 (simplify_logical): Likewise.
13117 (expand_compound_operation): Likewise.
13118 (expand_field_assignment): Likewise.
13119 (make_compound_operation): Likewise.
13120 (extended_count): Likewise.
13121 (change_zero_ext): Likewise.
13122 (simplify_comparison): Likewise.
13123 * dwarf2out.c (scompare_loc_descriptor): Likewise.
13124 (ucompare_loc_descriptor): Likewise.
13125 (minmax_loc_descriptor): Likewise.
13126 (mem_loc_descriptor): Likewise.
13127 (loc_descriptor): Likewise.
13128 * expmed.c (init_expmed_one_mode): Likewise.
13129 * lra-constraints.c (lra_constraint_offset): Likewise.
13130 * optabs.c (prepare_libcall_arg): Likewise.
13131 * postreload.c (move2add_note_store): Likewise.
13132 * reload.c (operands_match_p): Likewise.
13133 * rtl.h (load_extend_op): Likewise.
13134 * rtlhooks.c (gen_lowpart_general): Likewise.
13135 * simplify-rtx.c (simplify_truncation): Likewise.
13136 (simplify_unary_operation_1): Likewise.
13137 (simplify_binary_operation_1): Likewise.
13138 (simplify_const_binary_operation): Likewise.
13139 (simplify_const_relational_operation): Likewise.
13140 (simplify_subreg): Likewise.
13141 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13142 * var-tracking.c (adjust_mems): Likewise.
13143 (prepare_call_arguments): Likewise.
13145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13146 Alan Hayward <alan.hayward@arm.com>
13147 David Sherwood <david.sherwood@arm.com>
13149 * machmode.h (is_int_mode): New fuction.
13150 * combine.c (find_split_point): Use it.
13151 (combine_simplify_rtx): Likewise.
13152 (simplify_if_then_else): Likewise.
13153 (simplify_set): Likewise.
13154 (simplify_shift_const_1): Likewise.
13155 (simplify_comparison): Likewise.
13156 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
13157 * cse.c (notreg_cost): Likewise.
13158 (cse_insn): Likewise.
13159 * cselib.c (cselib_lookup_1): Likewise.
13160 * dojump.c (do_jump_1): Likewise.
13161 (do_compare_rtx_and_jump): Likewise.
13162 * dse.c (get_call_args): Likewise.
13163 * dwarf2out.c (rtl_for_decl_init): Likewise.
13164 (native_encode_initializer): Likewise.
13165 * expmed.c (emit_store_flag_1): Likewise.
13166 (emit_store_flag): Likewise.
13167 * expr.c (convert_modes): Likewise.
13168 (store_field): Likewise.
13169 (expand_expr_real_1): Likewise.
13170 * fold-const.c (fold_read_from_constant_string): Likewise.
13171 * gimple-ssa-sprintf.c (get_format_string): Likewise.
13172 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
13173 * optabs.c (expand_binop): Likewise.
13174 (expand_unop): Likewise.
13175 (expand_abs_nojump): Likewise.
13176 (expand_one_cmpl_abs_nojump): Likewise.
13177 * simplify-rtx.c (mode_signbit_p): Likewise.
13178 (val_signbit_p): Likewise.
13179 (val_signbit_known_set_p): Likewise.
13180 (val_signbit_known_clear_p): Likewise.
13181 (simplify_relational_operation_1): Likewise.
13182 * tree.c (vector_type_mode): Likewise.
13184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13185 Alan Hayward <alan.hayward@arm.com>
13186 David Sherwood <david.sherwood@arm.com>
13188 * machmode.h (smallest_mode_for_size): Fix formatting.
13189 (smallest_int_mode_for_size): New function.
13190 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
13191 instead of smallest_mode_for_size.
13192 * combine.c (make_extraction): Likewise.
13193 * config/arc/arc.c (arc_expand_movmem): Likewise.
13194 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
13195 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13196 * config/s390/s390.c (s390_expand_insv): Likewise.
13197 * config/sparc/sparc.c (assign_int_registers): Likewise.
13198 * config/spu/spu.c (spu_function_value): Likewise.
13199 (spu_function_arg): Likewise.
13200 * coverage.c (get_gcov_type): Likewise.
13201 (get_gcov_unsigned_t): Likewise.
13202 * dse.c (find_shift_sequence): Likewise.
13203 * expmed.c (store_bit_field_1): Likewise.
13204 * expr.c (convert_move): Likewise.
13205 (store_field): Likewise.
13206 * internal-fn.c (expand_arith_overflow): Likewise.
13207 * optabs-query.c (get_best_extraction_insn): Likewise.
13208 * optabs.c (expand_twoval_binop_libfunc): Likewise.
13209 * stor-layout.c (layout_type): Likewise.
13210 (initialize_sizetypes): Likewise.
13211 * targhooks.c (default_get_mask_mode): Likewise.
13212 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
13214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13215 Alan Hayward <alan.hayward@arm.com>
13216 David Sherwood <david.sherwood@arm.com>
13218 * machmode.h (opt_mode::else_blk): New function.
13219 (int_mode_for_mode): Declare.
13220 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
13221 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
13223 * cfgexpand.c (expand_debug_expr): Likewise.
13224 * combine.c (gen_lowpart_or_truncate): Likewise.
13225 (gen_lowpart_for_combine): Likewise.
13226 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
13227 * config/avr/avr.c (avr_to_int_mode): Likewise.
13228 (avr_out_plus_1): Likewise.
13229 (avr_out_plus): Likewise.
13230 (avr_out_round): Likewise.
13231 * config/i386/i386.c (ix86_split_to_parts): Likewise.
13232 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13233 (s390_expand_vcond): Likewise.
13234 * config/spu/spu.c (spu_split_immediate): Likewise.
13235 (spu_expand_mov): Likewise.
13236 * dse.c (get_stored_val): Likewise.
13237 * expmed.c (store_bit_field_1): Likewise.
13238 (convert_extracted_bit_field): Use int_mode_for_mode instead of
13240 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
13241 (extract_low_bits): Likewise.
13242 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
13243 handling rather than repeating the check.
13244 (emit_group_store): Likewise.
13245 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
13246 * optabs.c (expand_absneg_bit): Likewise.
13247 (expand_copysign_absneg): Likewise.
13248 (expand_copysign_bit): Likewise.
13249 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
13250 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
13251 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
13252 * var-tracking.c (prepare_call_arguments): Likewise.
13253 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
13254 int_mode_for_mode instead of mode_for_size.
13255 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13258 Alan Hayward <alan.hayward@arm.com>
13259 David Sherwood <david.sherwood@arm.com>
13261 * machmode.h (int_mode_for_size): New function.
13262 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
13263 instead of mode_for_size.
13264 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
13266 * combine.c (expand_field_assignment): Use int_mode_for_size
13267 instead of mode_for_size.
13268 (make_extraction): Likewise.
13269 (simplify_shift_const_1): Likewise.
13270 (simplify_comparison): Likewise.
13271 * dojump.c (do_jump): Likewise.
13272 * dwarf2out.c (mem_loc_descriptor): Likewise.
13273 * emit-rtl.c (init_derived_machine_modes): Likewise.
13274 * expmed.c (flip_storage_order): Likewise.
13275 (convert_extracted_bit_field): Likewise.
13276 * expr.c (copy_blkmode_from_reg): Likewise.
13277 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
13278 * internal-fn.c (expand_mul_overflow): Likewise.
13279 * lower-subreg.c (simple_move): Likewise.
13280 * optabs-libfuncs.c (init_optabs): Likewise.
13281 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13282 * tree.c (vector_type_mode): Likewise.
13283 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
13284 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
13285 * tree-vect-generic.c (expand_vector_parallel): Likewise.
13286 * tree-vect-stmts.c (vectorizable_load): Likewise.
13287 (vectorizable_store): Likewise.
13289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13290 Alan Hayward <alan.hayward@arm.com>
13291 David Sherwood <david.sherwood@arm.com>
13293 * coretypes.h (pod_mode): New type.
13294 (scalar_int_mode_pod): New typedef.
13295 * machmode.h (pod_mode): New class.
13296 (int_n_data_t::m): Change type to scalar_int_mode_pod.
13297 * genmodes.c (emit_mode_int_n): Update accordingly.
13298 * lower-subreg.h (target_lower_subreg): Change type to
13299 scalar_int_mode_pod.
13300 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
13301 scalar_int_mode_pod.
13303 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13304 Alan Hayward <alan.hayward@arm.com>
13305 David Sherwood <david.sherwood@arm.com>
13307 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
13308 machine_mode to scalar_int_mode.
13309 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
13310 (rs6000_option_override_internal): Remove cast to int.
13311 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
13312 machine_mode to scalar_int_mode.
13313 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
13314 (rs6000_option_override_internal): Remove cast to int.
13315 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
13316 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
13318 * config/s390/s390.c (s390_expand_builtin): Likewise.
13319 * coretypes.h (scalar_int_mode): New type.
13320 (opt_scalar_int_mode): New typedef.
13321 * machmode.h (scalar_int_mode): New class.
13322 (scalar_int_mode::includes_p): New function.
13323 (byte_mode): Change type to scalar_int_mode.
13324 (word_mode): Likewise.
13325 (ptr_mode): Likewise.
13326 * emit-rtl.c (byte_mode): Likewise.
13327 (word_mode): Likewise.
13328 (ptr_mode): Likewise.
13329 (init_derived_machine_modes): Update accordingly.
13330 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
13331 and MODE_PARTIAL_INT.
13332 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
13333 opt_scalar_int_mode.
13335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13336 Alan Hayward <alan.hayward@arm.com>
13337 David Sherwood <david.sherwood@arm.com>
13339 * target.def (libgcc_floating_mode_supported_p): Take a
13341 * doc/tm.texi: Regenerate.
13342 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
13344 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13345 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
13348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13349 Alan Hayward <alan.hayward@arm.com>
13350 David Sherwood <david.sherwood@arm.com>
13352 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
13353 * doc/tm.texi: Regenerate.
13354 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
13355 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
13356 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
13357 * targhooks.h (default_floatn_mode): Likewise.
13358 * targhooks.c (default_floatn_mode): Likewise.
13359 * tree.c (build_common_tree_nodes): Update accordingly.
13361 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13362 Alan Hayward <alan.hayward@arm.com>
13363 David Sherwood <david.sherwood@arm.com>
13365 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
13366 (mode_iterator::iterate_p): Likewise.
13367 (mode_iterator::get_wider): Likewise.
13368 * expr.c (init_expr_target): Use opt_scalar_float_mode.
13370 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13371 Alan Hayward <alan.hayward@arm.com>
13372 David Sherwood <david.sherwood@arm.com>
13374 * coretypes.h (opt_scalar_float_mode): New typedef.
13375 * machmode.h (float_mode_for_size): New function.
13376 * emit-rtl.c (double_mode): Delete.
13377 (init_emit_once): Use float_mode_for_size.
13378 * stor-layout.c (layout_type): Likewise.
13379 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
13381 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13382 Alan Hayward <alan.hayward@arm.com>
13383 David Sherwood <david.sherwood@arm.com>
13385 * output.h (assemble_real): Take a scalar_float_mode.
13386 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
13387 * config/arm/arm.md (consttable_4): Likewise.
13388 (consttable_8): Likewise.
13389 (consttable_16): Likewise.
13390 * config/mips/mips.md (consttable_float): Likewise.
13391 * config/s390/s390.c (s390_output_pool_entry): Likewise.
13392 * varasm.c (assemble_real): Take a scalar_float_mode.
13393 (output_constant_pool_2): Update accordingly.
13394 (output_constant): Likewise.
13396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13397 Alan Hayward <alan.hayward@arm.com>
13398 David Sherwood <david.sherwood@arm.com>
13400 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
13401 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
13402 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
13403 (native_encode_real): Likewise.
13404 (native_interpret_real): Likewise.
13405 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
13406 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13409 Alan Hayward <alan.hayward@arm.com>
13410 David Sherwood <david.sherwood@arm.com>
13412 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
13413 <scalar_float_mode>. Simplify.
13414 (gen_extend_conv_libfunc): Likewise.
13416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13417 Alan Hayward <alan.hayward@arm.com>
13418 David Sherwood <david.sherwood@arm.com>
13420 * coretypes.h (scalar_float_mode): New type.
13421 * machmode.h (mode_traits::from_int): Use machine_mode if
13422 USE_ENUM_MODES is defined.
13423 (is_a): New function.
13425 (dyn_cast): Likewise.
13426 (scalar_float_mode): New class.
13427 (scalar_float_mode::includes_p): New function.
13428 (is_float_mode): Likewise.
13429 * gdbhooks.py (MachineModePrinter): New class.
13430 (build_pretty_printer): Use it for scalar_float_mode.
13431 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
13432 (format_helper::format_helper): Turn into a template.
13433 * genmodes.c (get_mode_class): New function.
13434 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
13435 or machine_mode if none.
13436 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
13437 as_a <scalar_float_mode>.
13438 * dwarf2out.c (mem_loc_descriptor): Likewise.
13439 (insert_float): Likewise.
13440 (add_const_value_attribute): Likewise.
13441 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13442 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
13443 (expand_unop): Update accordingly.
13444 (expand_abs_nojump): Likewise.
13445 (expand_copysign_absneg): Take a scalar_float_mode.
13446 (expand_copysign_bit): Likewise.
13447 (expand_copysign): Update accordingly.
13449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13450 Alan Hayward <alan.hayward@arm.com>
13451 David Sherwood <david.sherwood@arm.com>
13453 * coretypes.h (opt_mode): New class.
13454 * machmode.h (opt_mode): Likewise.
13455 (opt_mode::else_void): New function.
13456 (opt_mode::require): Likewise.
13457 (opt_mode::exists): Likewise.
13458 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
13459 (GET_MODE_2XWIDER_MODE): Likewise.
13460 (mode_iterator::get_wider): Update accordingly.
13461 (mode_iterator::get_2xwider): Likewise.
13462 (mode_iterator::get_known_wider): Likewise, turning into a template.
13463 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
13464 forcing a wider mode to exist.
13465 * config/cr16/cr16.h (LONG_REG_P): Likewise.
13466 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13467 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
13468 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
13469 * lower-subreg.c (init_lower_subreg): Likewise.
13470 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
13471 on the final iteration.
13472 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
13473 a wider mode exists before asking for a move pattern.
13474 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
13475 forcing a wider mode to exist.
13476 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
13477 returning false if no such mode exists.
13478 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
13479 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
13480 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
13481 Avoid checking for a MODE_INT if we already know the mode is not a
13483 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
13484 forcing a wider mode to exist.
13485 (expmed_mult_highpart_optab): Likewise.
13486 (expmed_mult_highpart): Likewise.
13487 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
13489 * lto-streamer-in.c (lto_input_mode_table): Likewise.
13490 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
13491 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13492 * internal-fn.c (expand_mul_overflow): Update use of
13493 GET_MODE_2XWIDER_MODE.
13494 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13495 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
13496 GET_MODE_WIDER_MODE.
13497 (convert_plusminus_to_widen): Likewise.
13498 * tree-switch-conversion.c (array_value_type): Likewise.
13499 * var-tracking.c (emit_note_insn_var_location): Likewise.
13500 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13501 Return false inside rather than outside the loop if no wider mode
13503 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
13504 and GET_MODE_2XWIDER_MODE
13505 (can_compare_p): Use else_void.
13506 * gdbhooks.py (OptMachineModePrinter): New class.
13507 (build_pretty_printer): Use it for opt_mode.
13509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13510 Alan Hayward <alan.hayward@arm.com>
13511 David Sherwood <david.sherwood@arm.com>
13513 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
13514 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
13516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13517 Alan Hayward <alan.hayward@arm.com>
13518 David Sherwood <david.sherwood@arm.com>
13520 * machmode.h (mode_traits): New structure.
13521 (get_narrowest_mode): New function.
13522 (mode_iterator::start): Likewise.
13523 (mode_iterator::iterate_p): Likewise.
13524 (mode_iterator::get_wider): Likewise.
13525 (mode_iterator::get_known_wider): Likewise.
13526 (mode_iterator::get_2xwider): Likewise.
13527 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
13528 (FOR_EACH_MODE): Likewise.
13529 (FOR_EACH_MODE_FROM): Likewise.
13530 (FOR_EACH_MODE_UNTIL): Likewise.
13531 (FOR_EACH_WIDER_MODE): Likewise.
13532 (FOR_EACH_2XWIDER_MODE): Likewise.
13533 * builtins.c (expand_builtin_strlen): Use new mode iterators.
13534 * combine.c (simplify_comparison): Likewise
13535 * config/i386/i386.c (type_natural_mode): Likewise.
13536 * cse.c (cse_insn): Likewise.
13537 * dse.c (find_shift_sequence): Likewise.
13538 * emit-rtl.c (init_derived_machine_modes): Likewise.
13539 (init_emit_once): Likewise.
13540 * explow.c (hard_function_value): Likewise.
13541 * expmed.c (extract_fixed_bit_field_1): Likewise.
13542 (extract_bit_field_1): Likewise.
13543 (expand_divmod): Likewise.
13544 (emit_store_flag_1): Likewise.
13545 * expr.c (init_expr_target): Likewise.
13546 (convert_move): Likewise.
13547 (alignment_for_piecewise_move): Likewise.
13548 (widest_int_mode_for_size): Likewise.
13549 (emit_block_move_via_movmem): Likewise.
13550 (copy_blkmode_to_reg): Likewise.
13551 (set_storage_via_setmem): Likewise.
13552 (compress_float_constant): Likewise.
13553 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13554 * optabs-query.c (get_best_extraction_insn): Likewise.
13555 * optabs.c (expand_binop): Likewise.
13556 (expand_twoval_unop): Likewise.
13557 (expand_twoval_binop): Likewise.
13558 (widen_leading): Likewise.
13559 (widen_bswap): Likewise.
13560 (expand_parity): Likewise.
13561 (expand_unop): Likewise.
13562 (prepare_cmp_insn): Likewise.
13563 (prepare_float_lib_cmp): Likewise.
13564 (expand_float): Likewise.
13565 (expand_fix): Likewise.
13566 (expand_sfix_optab): Likewise.
13567 * postreload.c (move2add_use_add2_insn): Likewise.
13568 * reg-stack.c (reg_to_stack): Likewise.
13569 * reginfo.c (choose_hard_reg_mode): Likewise.
13570 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13571 * stor-layout.c (mode_for_size): Likewise.
13572 (smallest_mode_for_size): Likewise.
13573 (mode_for_vector): Likewise.
13574 (finish_bitfield_representative): Likewise.
13575 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
13576 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
13577 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13578 * var-tracking.c (prepare_call_arguments): Likewise.
13580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13581 Alan Hayward <alan.hayward@arm.com>
13582 David Sherwood <david.sherwood@arm.com>
13584 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
13585 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
13586 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
13587 * machmode.h (mode_size): Move earlier in file.
13588 (mode_precision): Likewise.
13589 (mode_inner): Likewise.
13590 (mode_nunits): Likewise.
13591 (mode_unit_size): Likewise.
13592 (unit_unit_precision): Likewise.
13593 (mode_wider): Likewise.
13594 (mode_2xwider): Likewise.
13595 (machine_mode): New class.
13596 (mode_to_bytes): New function.
13597 (mode_to_bits): Likewise.
13598 (mode_to_precision): Likewise.
13599 (mode_to_inner): Likewise.
13600 (mode_to_unit_size): Likewise.
13601 (mode_to_unit_precision): Likewise.
13602 (mode_to_nunits): Likewise.
13603 (GET_MODE_SIZE): Use mode_to_bytes.
13604 (GET_MODE_BITSIZE): Use mode_to_bits.
13605 (GET_MODE_PRECISION): Use mode_to_precision.
13606 (GET_MODE_INNER): Use mode_to_inner.
13607 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
13608 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
13609 (GET_MODE_NUNITS): Use mode_to_nunits.
13610 * system.h (ALWAYS_INLINE): New macro.
13611 * config/powerpcspe/powerpcspe-c.c
13612 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
13613 int for arg1_mode and arg2_mode.
13615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13616 Alan Hayward <alan.hayward@arm.com>
13617 David Sherwood <david.sherwood@arm.com>
13619 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13620 Prefix mode names with E_ in case statements.
13621 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13622 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
13623 (aarch64_split_simd_move): Likewise.
13624 (aarch64_gen_storewb_pair): Likewise.
13625 (aarch64_gen_loadwb_pair): Likewise.
13626 (aarch64_gen_store_pair): Likewise.
13627 (aarch64_gen_load_pair): Likewise.
13628 (aarch64_get_condition_code_1): Likewise.
13629 (aarch64_constant_pool_reload_icode): Likewise.
13630 (get_rsqrte_type): Likewise.
13631 (get_rsqrts_type): Likewise.
13632 (get_recpe_type): Likewise.
13633 (get_recps_type): Likewise.
13634 (aarch64_gimplify_va_arg_expr): Likewise.
13635 (aarch64_simd_container_mode): Likewise.
13636 (aarch64_emit_load_exclusive): Likewise.
13637 (aarch64_emit_store_exclusive): Likewise.
13638 (aarch64_expand_compare_and_swap): Likewise.
13639 (aarch64_gen_atomic_cas): Likewise.
13640 (aarch64_emit_bic): Likewise.
13641 (aarch64_emit_atomic_swap): Likewise.
13642 (aarch64_emit_atomic_load_op): Likewise.
13643 (aarch64_evpc_trn): Likewise.
13644 (aarch64_evpc_uzp): Likewise.
13645 (aarch64_evpc_zip): Likewise.
13646 (aarch64_evpc_ext): Likewise.
13647 (aarch64_evpc_rev): Likewise.
13648 (aarch64_evpc_dup): Likewise.
13649 (aarch64_gen_ccmp_first): Likewise.
13650 (aarch64_gen_ccmp_next): Likewise.
13651 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13652 (alpha_emit_xfloating_libcall): Likewise.
13653 (emit_insxl): Likewise.
13654 (alpha_arg_type): Likewise.
13655 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13656 (arc_preferred_simd_mode): Likewise.
13657 (arc_secondary_reload): Likewise.
13658 (get_arc_condition_code): Likewise.
13659 (arc_print_operand): Likewise.
13660 (arc_legitimate_constant_p): Likewise.
13661 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13662 * config/arc/arc.md (casesi_load): Likewise.
13663 (casesi_compact_jump): Likewise.
13664 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13665 (cc_use_register): Likewise.
13666 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13667 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13668 (arm_init_iwmmxt_builtins): Likewise.
13669 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13670 (neon_expand_vector_init): Likewise.
13671 (arm_attr_length_move_neon): Likewise.
13672 (maybe_get_arm_condition_code): Likewise.
13673 (arm_emit_vector_const): Likewise.
13674 (arm_preferred_simd_mode): Likewise.
13675 (arm_output_iwmmxt_tinsr): Likewise.
13676 (thumb1_output_casesi): Likewise.
13677 (thumb2_output_casesi): Likewise.
13678 (arm_emit_load_exclusive): Likewise.
13679 (arm_emit_store_exclusive): Likewise.
13680 (arm_expand_compare_and_swap): Likewise.
13681 (arm_evpc_neon_vuzp): Likewise.
13682 (arm_evpc_neon_vzip): Likewise.
13683 (arm_evpc_neon_vrev): Likewise.
13684 (arm_evpc_neon_vtrn): Likewise.
13685 (arm_evpc_neon_vext): Likewise.
13686 (arm_validize_comparison): Likewise.
13687 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13688 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13689 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13690 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13691 (c6x_preferred_simd_mode): Likewise.
13692 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13693 (epiphany_rtx_costs): Likewise.
13694 * config/epiphany/predicates.md (proper_comparison_operator):
13696 * config/frv/frv.c (condexec_memory_operand): Likewise.
13697 (frv_emit_move): Likewise.
13698 (output_move_single): Likewise.
13699 (output_condmove_single): Likewise.
13700 (frv_hard_regno_mode_ok): Likewise.
13701 (frv_matching_accg_mode): Likewise.
13702 * config/h8300/h8300.c (split_adds_subs): Likewise.
13703 (h8300_rtx_costs): Likewise.
13704 (h8300_print_operand): Likewise.
13705 (compute_mov_length): Likewise.
13706 (output_logical_op): Likewise.
13707 (compute_logical_op_length): Likewise.
13708 (compute_logical_op_cc): Likewise.
13709 (h8300_shift_needs_scratch_p): Likewise.
13710 (output_a_shift): Likewise.
13711 (compute_a_shift_length): Likewise.
13712 (compute_a_shift_cc): Likewise.
13713 (expand_a_rotate): Likewise.
13714 (output_a_rotate): Likewise.
13715 * config/i386/i386.c (classify_argument): Likewise.
13716 (function_arg_advance_32): Likewise.
13717 (function_arg_32): Likewise.
13718 (function_arg_64): Likewise.
13719 (function_value_64): Likewise.
13720 (ix86_gimplify_va_arg): Likewise.
13721 (ix86_legitimate_constant_p): Likewise.
13722 (put_condition_code): Likewise.
13723 (split_double_mode): Likewise.
13724 (ix86_avx256_split_vector_move_misalign): Likewise.
13725 (ix86_expand_vector_logical_operator): Likewise.
13726 (ix86_split_idivmod): Likewise.
13727 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13728 (ix86_build_const_vector): Likewise.
13729 (ix86_build_signbit_mask): Likewise.
13730 (ix86_match_ccmode): Likewise.
13731 (ix86_cc_modes_compatible): Likewise.
13732 (ix86_expand_branch): Likewise.
13733 (ix86_expand_sse_cmp): Likewise.
13734 (ix86_expand_sse_movcc): Likewise.
13735 (ix86_expand_int_sse_cmp): Likewise.
13736 (ix86_expand_vec_perm_vpermi2): Likewise.
13737 (ix86_expand_vec_perm): Likewise.
13738 (ix86_expand_sse_unpack): Likewise.
13739 (ix86_expand_int_addcc): Likewise.
13740 (ix86_split_to_parts): Likewise.
13741 (ix86_vectorize_builtin_gather): Likewise.
13742 (ix86_vectorize_builtin_scatter): Likewise.
13743 (avx_vpermilp_parallel): Likewise.
13744 (inline_memory_move_cost): Likewise.
13745 (ix86_tieable_integer_mode_p): Likewise.
13746 (x86_maybe_negate_const_int): Likewise.
13747 (ix86_expand_vector_init_duplicate): Likewise.
13748 (ix86_expand_vector_init_one_nonzero): Likewise.
13749 (ix86_expand_vector_init_one_var): Likewise.
13750 (ix86_expand_vector_init_concat): Likewise.
13751 (ix86_expand_vector_init_interleave): Likewise.
13752 (ix86_expand_vector_init_general): Likewise.
13753 (ix86_expand_vector_set): Likewise.
13754 (ix86_expand_vector_extract): Likewise.
13755 (emit_reduc_half): Likewise.
13756 (ix86_emit_i387_round): Likewise.
13757 (ix86_mangle_type): Likewise.
13758 (ix86_expand_round_sse4): Likewise.
13759 (expand_vec_perm_blend): Likewise.
13760 (canonicalize_vector_int_perm): Likewise.
13761 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13762 (expand_vec_perm_1): Likewise.
13763 (expand_vec_perm_interleave3): Likewise.
13764 (expand_vec_perm_even_odd_pack): Likewise.
13765 (expand_vec_perm_even_odd_1): Likewise.
13766 (expand_vec_perm_broadcast_1): Likewise.
13767 (ix86_vectorize_vec_perm_const_ok): Likewise.
13768 (ix86_expand_vecop_qihi): Likewise.
13769 (ix86_expand_mul_widen_hilo): Likewise.
13770 (ix86_expand_sse2_abs): Likewise.
13771 (ix86_expand_pextr): Likewise.
13772 (ix86_expand_pinsr): Likewise.
13773 (ix86_preferred_simd_mode): Likewise.
13774 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13775 * config/i386/sse.md (*andnot<mode>3): Likewise.
13776 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13777 (*<code><mode>3): Likewise.
13778 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13779 (ia64_expand_atomic_op): Likewise.
13780 (ia64_arg_type): Likewise.
13781 (ia64_mode_to_int): Likewise.
13782 (ia64_scalar_mode_supported_p): Likewise.
13783 (ia64_vector_mode_supported_p): Likewise.
13784 (expand_vec_perm_broadcast): Likewise.
13785 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13786 (iq2000_function_arg_advance): Likewise.
13787 (iq2000_function_arg): Likewise.
13788 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13789 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13790 (m68k_libcall_value): Likewise.
13791 (m68k_function_value): Likewise.
13792 (sched_attr_op_type): Likewise.
13793 * config/mcore/mcore.c (mcore_output_move): Likewise.
13794 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13796 (microblaze_function_arg): Likewise.
13797 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13798 (mips_print_operand): Likewise.
13799 (mips_mode_ok_for_mov_fmt_p): Likewise.
13800 (mips_vector_mode_supported_p): Likewise.
13801 (mips_preferred_simd_mode): Likewise.
13802 (mips_expand_vpc_loongson_even_odd): Likewise.
13803 (mips_expand_vec_unpack): Likewise.
13804 (mips_expand_vi_broadcast): Likewise.
13805 (mips_expand_vector_init): Likewise.
13806 (mips_expand_vec_reduc): Likewise.
13807 (mips_expand_msa_cmp): Likewise.
13808 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13809 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13810 (cc_flags_for_mode): Likewise.
13811 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13812 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13813 (nds32_output_casesi_pc_relative): Likewise.
13814 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13815 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13816 (nvptx_gen_unpack): Likewise.
13817 (nvptx_gen_pack): Likewise.
13818 (nvptx_gen_shuffle): Likewise.
13819 (nvptx_gen_wcast): Likewise.
13820 (nvptx_preferred_simd_mode): Likewise.
13821 * config/pa/pa.c (pa_secondary_reload): Likewise.
13822 * config/pa/predicates.md (base14_operand): Likewise.
13823 * config/powerpcspe/powerpcspe-c.c
13824 (altivec_resolve_overloaded_builtin): Likewise.
13825 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13827 (rs6000_preferred_simd_mode): Likewise.
13828 (output_vec_const_move): Likewise.
13829 (rs6000_expand_vector_extract): Likewise.
13830 (rs6000_split_vec_extract_var): Likewise.
13831 (reg_offset_addressing_ok_p): Likewise.
13832 (rs6000_legitimate_offset_address_p): Likewise.
13833 (rs6000_legitimize_address): Likewise.
13834 (rs6000_emit_set_const): Likewise.
13835 (rs6000_const_vec): Likewise.
13836 (rs6000_emit_move): Likewise.
13837 (spe_build_register_parallel): Likewise.
13838 (rs6000_darwin64_record_arg_recurse): Likewise.
13839 (swap_selector_for_mode): Likewise.
13840 (spe_init_builtins): Likewise.
13841 (paired_init_builtins): Likewise.
13842 (altivec_init_builtins): Likewise.
13843 (do_load_for_compare): Likewise.
13844 (rs6000_generate_compare): Likewise.
13845 (rs6000_expand_float128_convert): Likewise.
13846 (emit_load_locked): Likewise.
13847 (emit_store_conditional): Likewise.
13848 (rs6000_output_function_epilogue): Likewise.
13849 (rs6000_handle_altivec_attribute): Likewise.
13850 (rs6000_function_value): Likewise.
13851 (emit_fusion_gpr_load): Likewise.
13852 (emit_fusion_p9_load): Likewise.
13853 (emit_fusion_p9_store): Likewise.
13854 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13855 (fusion_gpr_mem_load): Likewise.
13856 (fusion_addis_mem_combo_load): Likewise.
13857 (fusion_addis_mem_combo_store): Likewise.
13858 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13859 (fusion_gpr_mem_load): Likewise.
13860 (fusion_addis_mem_combo_load): Likewise.
13861 (fusion_addis_mem_combo_store): Likewise.
13862 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13864 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13865 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13866 (rs6000_preferred_simd_mode): Likewise.
13867 (output_vec_const_move): Likewise.
13868 (rs6000_expand_vector_extract): Likewise.
13869 (rs6000_split_vec_extract_var): Likewise.
13870 (reg_offset_addressing_ok_p): Likewise.
13871 (rs6000_legitimate_offset_address_p): Likewise.
13872 (rs6000_legitimize_address): Likewise.
13873 (rs6000_emit_set_const): Likewise.
13874 (rs6000_const_vec): Likewise.
13875 (rs6000_emit_move): Likewise.
13876 (rs6000_darwin64_record_arg_recurse): Likewise.
13877 (swap_selector_for_mode): Likewise.
13878 (paired_init_builtins): Likewise.
13879 (altivec_init_builtins): Likewise.
13880 (rs6000_expand_float128_convert): Likewise.
13881 (emit_load_locked): Likewise.
13882 (emit_store_conditional): Likewise.
13883 (rs6000_output_function_epilogue): Likewise.
13884 (rs6000_handle_altivec_attribute): Likewise.
13885 (rs6000_function_value): Likewise.
13886 (emit_fusion_gpr_load): Likewise.
13887 (emit_fusion_p9_load): Likewise.
13888 (emit_fusion_p9_store): Likewise.
13889 * config/rx/rx.c (rx_gen_move_template): Likewise.
13890 (flags_from_mode): Likewise.
13891 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13892 (s390_slb_comparison): Likewise.
13893 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13894 (s390_vector_mode_supported_p): Likewise.
13895 (s390_cc_modes_compatible): Likewise.
13896 (s390_match_ccmode_set): Likewise.
13897 (s390_canonicalize_comparison): Likewise.
13898 (s390_emit_compare_and_swap): Likewise.
13899 (s390_branch_condition_mask): Likewise.
13900 (s390_rtx_costs): Likewise.
13901 (s390_secondary_reload): Likewise.
13902 (__SECONDARY_RELOAD_CASE): Likewise.
13903 (s390_expand_cs): Likewise.
13904 (s390_preferred_simd_mode): Likewise.
13905 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13906 * config/sh/sh.c (sh_print_operand): Likewise.
13907 (dump_table): Likewise.
13908 (sh_secondary_reload): Likewise.
13909 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13910 * config/sh/sh.md (casesi_worker_1): Likewise.
13911 (casesi_worker_2): Likewise.
13912 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13913 (fcc_comparison_operator): Likewise.
13914 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13915 (emit_soft_tfmode_cvt): Likewise.
13916 (sparc_preferred_simd_mode): Likewise.
13917 (output_cbranch): Likewise.
13918 (sparc_print_operand): Likewise.
13919 (sparc_expand_vec_perm_bmask): Likewise.
13920 (vector_init_bshuffle): Likewise.
13921 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13922 (spu_vector_mode_supported_p): Likewise.
13923 (spu_expand_insv): Likewise.
13924 (spu_emit_branch_or_set): Likewise.
13925 (spu_handle_vector_attribute): Likewise.
13926 (spu_builtin_splats): Likewise.
13927 (spu_builtin_extract): Likewise.
13928 (spu_builtin_promote): Likewise.
13929 (spu_expand_sign_extend): Likewise.
13930 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13931 (tilegx_simd_int): Likewise.
13932 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13933 (tilepro_simd_int): Likewise.
13934 * config/v850/v850.c (const_double_split): Likewise.
13935 (v850_print_operand): Likewise.
13936 (ep_memory_offset): Likewise.
13937 * config/vax/vax.c (vax_rtx_costs): Likewise.
13938 (vax_output_int_move): Likewise.
13939 (vax_output_int_add): Likewise.
13940 (vax_output_int_subtract): Likewise.
13941 * config/visium/predicates.md (visium_branch_operator): Likewise.
13942 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13943 (visium_print_operand_address): Likewise.
13944 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13945 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13946 (xtensa_expand_conditional_branch): Likewise.
13947 (xtensa_copy_incoming_a7): Likewise.
13948 (xtensa_output_literal): Likewise.
13949 * dfp.c (decimal_real_maxval): Likewise.
13950 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13952 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13953 Alan Hayward <alan.hayward@arm.com>
13954 David Sherwood <david.sherwood@arm.com>
13956 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13957 (mode_nunits_inline): Likewise.
13958 (mode_inner_inline): Likewise.
13959 (mode_unit_size_inline): Likewise.
13960 (mode_unit_precision_inline): Likewise.
13961 (emit_insn_modes_h): Likewise. Also emit a #define of the
13963 (emit_mode_wider): Add an E_ prefix to mode names.
13964 (emit_mode_complex): Likewise.
13965 (emit_mode_inner): Likewise.
13966 (emit_mode_adjustments): Likewise.
13967 (emit_mode_int_n): Likewise.
13968 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13969 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13970 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13971 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13972 (CRC32_BUILTIN, ENTRY): Likewise.
13973 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13974 (aarch64_pop_regs): Likewise.
13975 (aarch64_process_components): Likewise.
13976 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13977 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13978 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13979 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13980 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13981 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13983 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13984 (pcmpistr): Likewise.
13985 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13986 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13987 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13989 * config/rl78/rl78.c (mduc_regs): Likewise.
13990 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13991 (htm_expand_builtin): Likewise.
13992 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13993 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13994 * config/xtensa/xtensa.c (print_operand): Likewise.
13995 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13996 (NUM_MODE_VECTOR_INT): Likewise.
13997 * genoutput.c (null_operand): Likewise.
13998 (output_operand_data): Likewise.
13999 * genrecog.c (print_parameter_value): Likewise.
14000 * lra.c (debug_operand_data): Likewise.
14002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14004 * dwarf2out.c (scompare_loc_descriptor_wide)
14005 (scompare_loc_descriptor_narrow): New functions, split out from...
14006 (scompare_loc_descriptor): ...here.
14007 * expmed.c (emit_store_flag_int): New function, split out from...
14008 (emit_store_flag): ...here.
14010 2017-08-30 Richard Biener <rguenther@suse.de>
14012 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
14013 (dwarf2out_early_finish): Move setting of AT_pubnames from
14014 early debug output to early finish.
14016 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14018 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
14019 and -mdata-region to the assembler.
14021 2017-08-30 Richard Biener <rguenther@suse.de>
14023 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
14025 (gen_subprogram_die): Add DW_AT_object_pointer only early.
14026 (dwarf2out_early_global_decl): Only generate a DIE for the
14027 abstract origin if it doesn't already exist or is a declaration DIE.
14028 (resolve_addr): Do not add the linkage name twice when
14029 generating a stub DIE for the DW_TAG_GNU_call_site target.
14031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14033 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14034 Use machine_mode rather than int for arg1_mode.
14036 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
14039 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
14040 that the second argument of the built-in functions to unpack
14041 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
14042 switch statement instead a lot of if statements.
14043 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
14044 Allow 64-bit values to be in Altivec registers as well as
14045 traditional floating point registers.
14046 (pack<mode>, FMOVE128_VSX iterator): Likewise.
14048 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
14050 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
14051 MAX_REGS_PER_ADDRESS == 1.
14053 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
14055 * config/i386/i386.opt (flag_fentry): Do not init to -1.
14056 * config/i386/i386.c (ix86_option_override_internal): Simplify
14057 setting of opts->x_flag_entry.
14059 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14060 Jakub Jelinek <jakub@redhat.com>
14061 Richard Biener <rguenther@suse.de>
14063 PR tree-optimization/81503
14064 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
14065 folded constant fits in the target type; reorder tests for clarity.
14067 2017-08-29 Martin Liska <mliska@suse.cz>
14069 * passes.def: Include pass_lower_switch.
14070 * stmt.c (dump_case_nodes): Remove and move to
14071 tree-switch-conversion.
14072 (case_values_threshold): Likewise.
14073 (expand_switch_as_decision_tree_p): Likewise.
14074 (emit_case_decision_tree): Likewise.
14075 (expand_case): Likewise.
14076 (balance_case_nodes): Likewise.
14077 (node_has_low_bound): Likewise.
14078 (node_has_high_bound): Likewise.
14079 (node_is_bounded): Likewise.
14080 (emit_case_nodes): Likewise.
14081 (struct simple_case_node): New struct.
14082 (add_case_node): Remove.
14083 (emit_case_dispatch_table): Use vector instead of case_list.
14084 (reset_out_edges_aux): Remove.
14085 (compute_cases_per_edge): Likewise.
14086 (expand_case): Build list of simple_case_node.
14087 (expand_sjlj_dispatch_table): Use it.
14088 * tree-switch-conversion.c (struct case_node): Moved from
14089 stmt.c and adjusted.
14090 (emit_case_nodes): Likewise.
14091 (node_has_low_bound): Likewise.
14092 (node_has_high_bound): Likewise.
14093 (node_is_bounded): Likewise.
14094 (case_values_threshold): Likewise.
14095 (reset_out_edges_aux): Likewise.
14096 (compute_cases_per_edge): Likewise.
14097 (add_case_node): Likewise.
14098 (dump_case_nodes): Likewise.
14099 (balance_case_nodes): Likewise.
14100 (expand_switch_as_decision_tree_p): Likewise.
14101 (emit_jump): Likewise.
14102 (emit_case_decision_tree): Likewise.
14103 (try_switch_expansion): Likewise.
14104 (do_jump_if_equal): Likewise.
14105 (emit_cmp_and_jump_insns): Likewise.
14106 (fix_phi_operands_for_edge): New function.
14107 (record_phi_operand_mapping): Likewise.
14108 (class pass_lower_switch): New pass.
14109 (pass_lower_switch::execute): New function.
14110 (make_pass_lower_switch): Likewise.
14111 (conditional_probability):
14112 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
14113 * tree-pass.h: Add make_pass_lower_switch.
14115 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14118 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
14121 2017-08-29 Richard Biener <rguenther@suse.de>
14123 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
14124 we do not add a DW_AT_inline attribute twice.
14125 (gen_subprogram_die): Remove code setting DW_AT_inline on
14126 DECL_ABSTRACT_P nodes.
14128 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
14130 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
14131 calls to internal functions.
14132 (gimplify_modify_expr): Likewise.
14133 * tree-call-cdce.c (use_internal_fn): Likewise.
14134 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
14135 (convert_to_divmod): Set the nothrow flag.
14136 * tree-if-conv.c (predicate_mem_writes): Likewise.
14137 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14138 (vectorizable_call): Likewise.
14139 (vectorizable_store): Likewise.
14140 (vectorizable_load): Likewise.
14141 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
14142 (vect_recog_mask_conversion_pattern): Likewise.
14144 2017-08-29 Martin Liska <mliska@suse.cz>
14147 * gcc.c (driver_handle_option): Add new argument.
14148 * opts-common.c (handle_option): Pass
14149 target_option_override_hook.
14150 * opts-global.c (lang_handle_option): Add new option.
14151 (set_default_handlers): Add new argument.
14152 (decode_options): Likewise.
14153 * opts.c (target_handle_option): Likewise.
14154 (common_handle_option): Call target_option_override_hook.
14155 * opts.h (struct cl_option_handler_func): Add hook for
14156 target option override.
14157 (struct cl_option_handlers): Likewise.
14158 (set_default_handlers): Add new argument.
14159 (decode_options): Likewise.
14160 (common_handle_option): Likewise.
14161 (target_handle_option): Likewise.
14162 * toplev.c (toplev::main): Pass targetm.target_option.override
14165 2017-08-29 Richard Biener <rguenther@suse.de>
14166 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
14168 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
14169 life to the active subtree.
14171 2017-08-28 Jeff Law <law@redhat.com>
14173 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
14174 derive_equivalences.
14175 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
14176 Code moved into....
14177 (edge_info::derive_equivalences): New private member function
14179 * tree-ssa-dom.c (class edge_info): Changed from a struct
14180 to a class. Add ctor/dtor, methods and data members.
14181 (edge_info::edge_info): Renamed from allocate_edge_info.
14182 Initialize additional members.
14183 (edge_info::~edge_info): New.
14184 (free_dom_edge_info): Delete the edge info.
14185 (record_edge_info): Use new class & associated member functions.
14186 Tighten forms for testing for edge equivalences.
14187 (record_temporary_equivalences): Iterate over the simple
14188 equivalences rather than assuming there's only one per edge.
14189 (cprop_into_successor_phis): Iterate over the simple
14190 equivalences rather than assuming there's only one per edge.
14191 (optimize_stmt): Use operand_equal_p rather than pointer
14192 equality for mini-DSE code.
14194 2017-08-28 Nathan Sidwell <nathan@acm.org>
14196 * gcc.c (execute): Fold SIGPIPE handling into switch
14197 statement. Adjust internal error message.
14199 2017-08-28 Richard Biener <rguenther@suse.de>
14202 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
14203 Do nothing for removed DIEs.
14205 2017-08-28 Richard Biener <rguenther@suse.de>
14207 PR tree-optimization/81977
14208 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
14211 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
14214 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
14215 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
14216 using targetm.gen_mem_thread_fence.
14218 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
14221 * config/i386/i386.md (*<btsc><mode>): Change operand 2
14222 predicate to register_operand. Reorder operands.
14223 (*btr<mode>): Ditto.
14224 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
14225 (*btr<mode>_mask): Ditto.
14227 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
14229 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
14230 * config/rs6000/xmmintrin.h: New file.
14231 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
14233 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14236 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
14237 parameter and_insn and return it.
14238 (recombine_lvx_pattern): Insert a copy to ensure availability of
14239 the base register of the copied masking operation at the point of
14240 the instruction replacement.
14241 (recombine_stvx_pattern): Likewise.
14243 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
14245 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
14246 undocumented switches.
14247 (-mpower9-dform-vector): Likewise.
14248 (-mpower9-dform): Likewise.
14249 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
14250 comments to delete references to -mpower9-dform* switches.
14251 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
14252 Delete reference to -mpower9-dform* switches, test for
14253 -mpower9-vector instead.
14254 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
14255 (OTHER_P9_VECTOR_MASKS): Likewise.
14256 (POWERPC_MASKS): Likewise.
14257 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
14258 tests against -mpower9-dform* to -mpower9-vector. Delete code
14259 that checked for -mpower9-dform* consistancy with other options.
14260 Add test for -mpower9-misc to enable other power9 switches.
14261 (rs6000_init_hard_regno_mode_ok): Likewise.
14262 (rs6000_option_override_internal): Likewise.
14263 (rs6000_emit_prologue): Likewise.
14264 (rs6000_emit_epilogue): Likewise.
14265 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
14266 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
14267 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
14269 (emit_fusion_p9_store): Likewise.
14270 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
14271 resetting these macros if the assembler does not support ISA 3.0
14273 (TARGET_P9_DFORM_VECTOR): Likewise.
14274 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
14275 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
14277 2017-08-25 Alan Modra <amodra@gmail.com>
14279 PR rtl-optimization/81747
14280 * cse.c (cse_extended_basic_block): Don't attempt to record
14281 equivalences for degenerate conditional branches that branch
14282 to their fall-through.
14284 2017-08-24 Martin Sebor <msebor@redhat.com>
14286 PR middle-end/81908
14287 * gimple-fold.c (size_must_be_zero_p): New function.
14288 (gimple_fold_builtin_memory_op): Call it.
14290 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
14292 * config/rs6000/mm_malloc.h: New file.
14294 2017-08-24 Bin Cheng <bin.cheng@arm.com>
14296 PR tree-optimization/81913
14297 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
14298 analysis when either IVs in condition can wrap.
14300 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
14302 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
14303 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
14305 2017-08-24 Richard Biener <rguenther@suse.de>
14308 * targhooks.c (default_target_can_inline_p): Properly
14309 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
14310 is present and always compare.
14311 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
14312 infer -mfpmath=sse from TARGET_SSE_P.
14313 (ix86_can_inline_p): Properly use target_option_default_node when
14314 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
14316 2017-08-24 Richard Biener <rguenther@suse.de>
14319 * dwarf2out.c (output_die): Handle flag_generate_offload like
14321 (output_comp_unit): Likewise.
14322 (gen_array_type_die): Likewise.
14323 (dwarf2out_early_finish): Likewise.
14324 (note_variable_value_in_expr): Likewise.
14325 (dwarf2out_finish): Likewise. Adjust assert.
14326 * cgraphunit.c (symbol_table::compile): Move setting of
14327 flag_generate_offload earlier ...
14328 (symbol_table::finalize_compilation_unit): ... here, before
14329 early debug finalization.
14331 2017-08-24 Richard Biener <rguenther@suse.de>
14333 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
14334 and ipa-fnsummary.h.
14335 (ix86_can_inline_p): When ix86_fpmath flags do not match
14336 check whether the callee uses FP math at all.
14338 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
14340 PR middle-end/81931
14341 * tree-ssanames.c (get_nonzero_bits): Use element_precision
14342 instead of TYPE_PRECISION.
14344 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
14345 Alan Hayward <alan.hayward@arm.com>
14346 David Sherwood <david.sherwood@arm.com>
14348 * combine.c (make_extraction): Use subreg_offset_from_lsb.
14350 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
14352 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
14354 (ix86_frame::stack_realign_allocate): New field.
14355 (struct machine_frame_state): Modify comments.
14356 (machine_frame_state::sp_realigned_fp_end): New field.
14357 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
14358 layout calculation.
14359 (sp_valid_at): Add assertion to assure no attempt to access invalid
14360 offset of a realigned stack.
14361 (fp_valid_at): Likewise.
14362 (choose_baseaddr): Modify comments.
14363 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
14364 ix86_expand_prologue.
14365 (ix86_expand_prologue): Modify stack realignment and allocation.
14366 (ix86_expand_epilogue): Modify comments.
14367 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
14368 avx2_runtime, avx512f, and avx512f_runtime.
14370 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
14372 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
14373 (mstackrealign): Do not init to -1.
14374 * config/i386/i386.c (ix86_option_override_internal):
14375 Check opts_set, not opts when setting default value of
14376 opts->x_ix86_force_align_arg_pointer.
14378 2017-08-23 Richard Biener <rguenther@suse.de>
14380 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
14381 lang_hooks.decl_printable_name.
14382 * print-rtl-function.c (print_rtx_function): Likewise.
14383 * tree-pretty-print.c (dump_function_header): Likewise.
14385 2017-08-23 Richard Biener <rguenther@suse.de>
14388 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
14389 -g0 at compile-time.
14391 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14393 PR middle-end/19706
14394 * doc/sourcebuild.texi (Other hardware attributes):
14397 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14399 PR middle-end/19706
14400 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
14401 Fix single-use check.
14403 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
14405 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
14407 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14409 * doc/install.texi: Modify to add more details on running selected
14412 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14414 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
14415 is combined with -mabi=ms.
14416 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
14419 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14421 PR tree-optimization/81488
14422 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
14423 and cached_basis fields.
14424 (MAX_SPREAD): New constant.
14425 (alloc_cand_and_find_basis): Initialize new fields.
14426 (clear_visited): New function.
14427 (create_phi_basis_1): Rename from create_phi_basis, set visited
14428 and cached_basis fields.
14429 (create_phi_basis): New wrapper function.
14430 (phi_add_costs_1): Rename from phi_add_costs, add spread
14431 parameter, set visited field, short-circuit when limits reached.
14432 (phi_add_costs): New wrapper function.
14433 (record_phi_increments_1): Rename from record_phi_increments, set
14435 (record_phi_increments): New wrapper function.
14436 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
14437 (phi_incr_cost): New wrapper function.
14438 (all_phi_incrs_profitable_1): Rename from
14439 all_phi_incrs_profitable, set visited field.
14440 (all_phi_incrs_profitable): New wrapper function.
14442 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14443 Alan Hayward <alan.hayward@arm.com>
14444 David Sherwood <david.sherwood@arm.com>
14446 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
14447 that takes the outer and inner modes.
14448 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
14449 comparison as the canonical test for a paradoxical subreg.
14450 * combine.c (simplify_set): Use paradoxical_subreg_p.
14451 (make_extraction): Likewise.
14452 (force_to_mode): Likewise.
14453 (rtx_equal_for_field_assignment_p): Likewise.
14454 (gen_lowpart_for_combine): Likewise.
14455 (simplify_comparison): Likewise.
14456 * cse.c (equiv_constant): Likewise.
14457 * expmed.c (store_bit_field_1): Likewise.
14458 * final.c (alter_subreg): Likewise.
14459 * fwprop.c (propagate_rtx): Likewise.
14460 (forward_propagate_subreg): Likewise.
14461 * ira-conflicts.c (ira_build_conflicts): Likewise.
14462 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
14463 * lra-constraints.c (curr_insn_transform): Likewise.
14464 (split_reg): Likewise.
14465 * lra-eliminations.c (move_plus_up): Likewise.
14466 (lra_eliminate_regs_1): Likewise.
14467 * recog.c (general_operand): Likewise.
14468 * ree.c (combine_reaching_defs): Likewise.
14469 * reload.c (push_reload): Likewise.
14470 (find_reloads): Likewise.
14471 * reload1.c (elimination_effects): Likewise.
14472 (compute_reload_subreg_offset): Likewise.
14473 (choose_reload_regs): Likewise.
14474 * rtlanal.c (subreg_lsb_1): Likewise.
14475 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14476 (simplify_subreg): Likewise.
14477 * var-tracking.c (track_loc_p): Likewise.
14478 * emit-rtl.c (byte_lowpart_offset): Likewise.
14479 (paradoxical_subreg_p): Delete out-of-line definition.
14481 2017-08-22 Jeff Law <law@redhat.com>
14483 PR tree-optimization/81741
14484 PR tree-optimization/71947
14485 * tree-ssa-dom.c: Include tree-inline.h.
14486 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
14487 equivalences if one is more expensive to compute than the other.
14488 * tree-ssa-scopedtables.h (class const_or_copies): Make
14489 record_const_or_copy_raw method private.
14490 (class avail_exprs_stack): New method simplify_binary_operation.
14491 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
14492 avail_exprs_stack::simplify_binary_operation as needed.
14493 (avail_exprs_stack::simplify_binary_operation): New function.
14495 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
14497 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
14498 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
14499 (DOT_SYMBOLS): Likewise.
14500 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
14501 (RELOCATABLE_NEEDS_FIXUP): Likewise.
14502 (RS6000_ABI_NAME): Likewise.
14503 (TARGET_CMODEL): Likewise.
14504 (TOC_SECTION_ASM_OP): Likewise.
14505 (SET_CMODEL): New macro.
14506 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
14508 2017-08-22 Richard Biener <rguenther@suse.de>
14510 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
14511 to allow for free-lang-data replacements similar to verify_type_variant.
14513 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14514 Alan Hayward <alan.hayward@arm.com>
14515 David Sherwood <david.sherwood@arm.com>
14517 * config/aarch64/aarch64.md (casesi): Use DImode rather than
14518 VOIDmode for the LABEL_REF.
14520 2017-08-22 Richard Biener <rguenther@suse.de>
14522 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
14524 2017-08-22 Richard Biener <rguenther@suse.de>
14526 * common.opt (feliminate-dwarf2-dups): Ignore.
14527 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
14528 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
14529 same_die_p_wrap, compute_section_prefix,
14530 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
14531 (comdat_symbol_id, comdat_symbol_number): Likewise.
14532 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
14534 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
14535 (output_die): Mark unreachable path unreachable.
14536 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
14537 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
14538 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
14539 (dwarf2out_early_finish): Likewise.
14541 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
14543 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
14545 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
14548 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
14549 not VAR_P. Filter attribute warnings with OPT_Wattributes.
14550 (avr_attribute_table) <io, io_low, address>: Initialize
14551 .decl_required with true.
14553 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
14555 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
14556 undocumented debugging options.
14557 (-mvsx-scalar-double): Likewise.
14558 (-mallow-df-permute): Likewise.
14559 (-mvectorize-builtins): Likewise.
14560 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14561 (rs6000_builtin_vectorized_function): Likewise.
14562 (rs6000_builtin_md_vectorized_function): Likewise.
14563 (rs6000_opt_vars): Likewise.
14565 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
14568 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
14569 (*btrq_imm): Rename from *btrq.
14570 (*btcq_imm): Rename from *btcq.
14571 (btsc): New code attribute.
14572 (*<btsc><mode>): New insn pattern.
14573 (*btr<mode>): Ditto.
14574 (*<btsc><mode>_mask): New insn_and_split pattern.
14575 (*btr<mode>_mask): Ditto.
14577 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14578 Alan Hayward <alan.hayward@arm.com>
14579 David Sherwood <david.sherwood@arm.com>
14581 * function.c (pad_below): Simplify padding calculation.
14583 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14584 Alan Hayward <alan.hayward@arm.com>
14585 David Sherwood <david.sherwood@arm.com>
14587 * target.def (function_prologue): Remove frame size argument.
14588 (function_epilogue): Likewise.
14589 * doc/tm.texi: Regenerate.
14590 * final.c (final_start_function): Update call to function_prologue.
14591 (final_end_function): Update call to function_epilogue.
14592 (default_function_pro_epilogue): Remove frame size argument.
14593 * output.h (default_function_pro_epilogue): Likewise.
14594 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
14595 (arm_output_function_prologue): Likewise.
14596 * config/frv/frv.c (frv_function_prologue): Likewise.
14597 (frv_function_epilogue): Likewise.
14598 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
14599 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
14600 (ia64_output_function_epilogue): Likewise.
14601 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
14602 (m32r_output_function_epilogue): Likewise.
14603 * config/microblaze/microblaze.c (microblaze_function_prologue)
14604 (microblaze_function_epilogue): Likewise.
14605 * config/mips/mips.c (mips_output_function_prologue): Likewise.
14606 (mips_output_function_epilogue): Likewise.
14607 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
14608 (mmix_target_asm_function_epilogue): Likewise.
14609 * config/msp430/msp430.c (msp430_start_function): Likewise.
14610 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
14611 (nds32_asm_function_epilogue): Likewise.
14612 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
14613 * config/pa/pa.c (pa_output_function_prologue): Likewise.
14614 (pa_output_function_epilogue): Likewise.
14615 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
14616 (rs6000_output_function_epilogue): Likewise.
14617 * config/rl78/rl78.c (rl78_start_function): Likewise.
14618 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
14619 (rs6000_output_function_epilogue): Likewise.
14620 * config/rx/rx.c (rx_output_function_prologue): Likewise.
14621 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
14622 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
14623 (sparc_asm_function_epilogue): Likewise.
14625 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14627 * tree.h (type_has_mode_precision_p): New function.
14628 * convert.c (convert_to_integer_1): Use it.
14629 * expr.c (expand_expr_real_2): Likewise.
14630 (expand_expr_real_1): Likewise.
14631 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
14632 * match.pd: Likewise.
14633 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
14634 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
14635 * tree-tailcall.c (process_assignment): Likewise.
14636 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14637 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
14638 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
14639 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14640 (vectorizable_assignment): Likewise.
14641 (vectorizable_shift): Likewise.
14642 (vectorizable_operation): Likewise.
14643 * tree-vrp.c (register_edge_assert_for_2): Likewise.
14645 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
14647 * match.pd: Add pow (C, x) simplification.
14649 2017-08-21 Richard Biener <rguenther@suse.de>
14651 PR tree-optimization/81900
14652 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14653 for blocks with abnormal predecessors.
14654 (compute_antic): Do not set visited flag prematurely.
14656 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14659 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14661 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14663 * stor-layout.h (vector_type_mode): Move to...
14664 * tree.h (vector_type_mode): ...here.
14665 * stor-layout.c (vector_type_mode): Move to...
14666 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14668 2017-08-21 Richard Biener <rguenther@suse.de>
14670 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14671 register_external_die hooks.
14672 (debug_false_tree_charstarstar_uhwistar): Declare.
14673 (debug_nothing_tree_charstar_uhwi): Likewise.
14674 * debug.c (do_nothing_debug_hooks): Adjust.
14675 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14676 (debug_nothing_tree_charstar_uhwi): Likewise.
14677 * dbxout.c (dbx_debug_hooks): Adjust.
14678 (xcoff_debug_hooks): Likewise.
14679 * sdbout.c (sdb_debug_hooks): Likewise.
14680 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14681 * dwarf2out.c (macinfo_label_base): New global.
14682 (dwarf2out_register_external_die): New function for the
14683 register_external_die hook.
14684 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14685 (dwarf2_debug_hooks): Use them.
14686 (dwarf2_lineno_debug_hooks): Adjust.
14687 (struct die_struct): Add with_offset flag.
14688 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14689 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14690 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14691 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14692 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14693 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14694 defining section names for the early LTO debug variants.
14695 (reset_indirect_string): New helper.
14696 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14697 (print_dw_val): Add support for offsetted symbol references.
14698 (get_ultimate_context): Split out from is_cxx.
14699 (is_cxx): Use get_ultimate_context.
14700 (is_fortran): Add decl overload.
14701 (compute_comp_unit_symbol): Split out worker from
14702 compute_section_prefix.
14703 (compute_section_prefix): Call compute_comp_unit_symbol and
14704 set comdat_type_p here.
14705 (output_die): Skip DIE symbol output for the LTO added one.
14706 Handle DIE symbol references with offset.
14707 (output_comp_unit): Guard section name mangling properly.
14708 For LTO debug sections emit a symbol at the section beginning
14709 which we use to refer to its DIEs.
14710 (add_abstract_origin_attribute): For DIEs registered via
14711 dwarf2out_register_external_die directly refer to the early
14712 DIE rather than indirectly through the shadow one we created.
14713 Remove obsolete call to dwarf2out_abstract_function for
14714 non-function/block origins.
14715 (gen_array_type_die): When generating early LTO debug do
14716 not emit DW_AT_string_length.
14717 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14718 late when in LTO. As suggested place a gcc_unreachable for
14719 the DECL_ABSTRACT_P case.
14720 (gen_subprogram_die): Avoid another specification DIE
14721 for early built declarations/definitions for the late LTO case.
14722 (gen_variable_die): Add type references for late duplicated VLA dies
14724 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14725 we have the abstract instance already.
14726 (process_scope_var): Adjust decl DIE contexts in LTO which
14727 first puts them in limbo.
14728 (gen_decl_die): Do not generate type DIEs late apart from
14729 types for VLAs or for decls we do not yet have a DIE. Do not
14730 call dwarf2out_abstract_function late.
14731 (dwarf2out_early_global_decl): Make sure to create DIEs
14732 for abstract instances of a decl first.
14733 (dwarf2out_late_global_decl): Adjust comment.
14734 (output_macinfo_op): With multiple macro sections use
14735 macinfo_label_base to distinguish labels.
14736 (output_macinfo): Likewise. Update macinfo_label_base.
14737 Pass in the line info label.
14738 (note_variable_value_in_expr): When generating LTO resolve
14739 all variable values here by generating DIEs as needed.
14740 (init_sections_and_labels): Add early LTO debug flag parameter
14741 and generate different sections and names if set. Add generation
14742 counter for the labels so we can have multiple of them.
14743 (reset_dies): Helper to allow DIEs to be output multiple times.
14744 (dwarf2out_finish): When outputting DIEs to the fat part of an
14745 LTO object first reset DIEs.
14746 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14747 (modified_type_die): Check for decl_ultimate_origin being self
14749 (gen_type_die_with_usage): Likewise.
14750 (gen_typedef_die): Allow decl_ultimate_origin being self.
14751 (set_decl_abstract_flags): Remove.
14752 (set_block_abstract_flags): Likewise.
14753 (dwarf2out_abstract_function): Treat the early generated DIEs
14754 as the abstract copy and only add DW_AT_inline and
14755 DW_AT_artificial here and call set_decl_origin_self.
14756 If the DIE has an abstract origin don't do anything.
14757 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14758 if we have none yet (Go fails to build one, PR78628).
14759 (variably_modified_type_p): Prevent endless recursion for Ada
14760 cyclic pointer types.
14761 * lto-streamer-in.c: Include debug.h.
14762 (dref_queue): New global.
14763 (lto_read_tree_1): Stream in DIE references.
14764 (lto_input_tree): Register DIE references.
14765 (input_function): Stream DECL_DEBUG_ARGS.
14766 * lto-streamer-out.c: Include debug.h.
14767 (lto_write_tree_1): Output DIE references.
14768 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14769 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14770 (output_function): Stream DECL_DEBUG_ARGS.
14771 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14772 Stream DECL_ABSTRACT_ORIGIN.
14773 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14774 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14775 DECL_CONTEXT for file-scope decls.
14776 * lto-streamer.h (struct dref_entry): Declare.
14777 (dref_queue): Likewise.
14778 * cfgexpand.c (pass_expand::execute): Do not call the
14779 outlining_inline_function hook here.
14780 * lto-wrapper.c (debug_obj): New global.
14781 (tool_cleanup): Unlink it if required.
14782 (debug_objcopy): New function.
14783 (run_gcc): Handle early debug sections in the IL files by
14784 extracting them to separate files, partially linkin them and
14785 feeding the result back as result to the linker.
14786 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14787 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14788 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14789 sections into a separate segment.
14790 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14792 (darwin_asm_dwarf_section): Likewise.
14793 (darwin_asm_output_dwarf_offset): Likewise.
14794 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14796 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14797 Alan Hayward <alan.hayward@arm.com>
14798 David Sherwood <david.sherwood@arm.com>
14800 * read-md.h (md_reader::record_potential_iterator_use): Replace
14801 pointer argument with an rtx and an index.
14802 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14803 (apply_mode_iterator): Likewise.
14804 (apply_code_iterator): Likewise.
14805 (apply_int_iterator): Likewise.
14806 (apply_subst_iterator): Likewise.
14807 (record_iterator_use): Likewise.
14808 (record_attribute_use): Likewise.
14809 (md_reader::record_potential_iterator_use): Likewise. Update calls
14810 to record_iterator_use and apply_iterator.
14811 (iterator_use): Replace ptr with x and index.
14812 (attribute_use): Likewise.
14813 (apply_attribute_uses): Update calls to apply_iterator.
14814 (apply_iterators): Likewise. Update initialization of iterator_use.
14815 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14816 and record_potential_iterator_use.
14817 (rtx_reader::read_rtx_operand): Likewise.
14819 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14820 Alan Hayward <alan.hayward@arm.com>
14821 David Sherwood <david.sherwood@arm.com>
14823 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14826 2017-08-21 Richard Biener <rguenther@suse.de>
14828 PR middle-end/81884
14829 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14830 at struct end conservatively when comparing common bases.
14832 2017-08-21 Richard Biener <rguenther@suse.de>
14834 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14835 (mem_ref_in_stmt): Remove.
14836 (determine_max_movement): Use ref index to get at the reference.
14837 (invariantness_dom_walker::before_dom_children): Deal with
14838 lim data already initialized.
14839 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14841 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14843 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14844 (TARGET_ISA_ROUND): Ditto.
14845 (TARGET_ROUND): Ditto.
14846 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14847 * config/i386/i386.md: Ditto.
14848 * config/i386/sse.md: Ditto.
14849 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14850 with OPTION_MASK_ISA_SSE4_1.
14852 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14855 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14856 __builtin_ia32_lzcnt_u16.
14858 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14861 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14862 (rs6000_set_current_function): Rewrite function to use it.
14864 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14867 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14868 and TYPE_WARN_IF_NOT_ALIGN.
14869 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14870 (handle_warn_if_not_align): New.
14871 (place_union_field): Call handle_warn_if_not_align.
14872 (place_field): Call handle_warn_if_not_align.
14873 Copy TYPE_WARN_IF_NOT_ALIGN.
14874 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14875 (layout_type): Likewise.
14876 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14878 (tree_decl_common): Add warn_if_not_align.
14879 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14880 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14881 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14882 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14883 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14884 * doc/extend.texi: Document warn_if_not_aligned attribute.
14885 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14887 2017-08-17 Martin Liska <mliska@suse.cz>
14890 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14891 (get_data_dependence): Use it as pointer type.
14892 (distribute_loop): Likewise.
14894 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14896 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14897 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14898 (p8_vmrgow_<mode>_direct): New define_insn.
14899 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14900 handle endianness for vmrgew and vmrgow permute patterns.
14902 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14904 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14905 * config/rs6000/rs6000-cpus.def: Remove comment.
14906 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14907 (POWERPC_MASKS): Likewise.
14908 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14909 use of TARGET_VSX_TIMODE.
14910 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14911 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14912 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14913 (rs6000_option_override_internal): Remove dead code.
14914 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14915 (rs6000_legitimize_reload_address): Likewise.
14916 (rs6000_legitimate_address_p): Likewise.
14917 (rs6000_opt_masks): Delete "vsx-timode".
14918 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14919 from function comment.
14920 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14921 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14922 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14924 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14925 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14926 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14927 (VSX_TI): Likewise.
14929 (define_peephole2): Likewise.
14931 2017-08-17 Martin Sebor <msebor@redhat.com>
14934 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14935 past the end of an array.
14936 (test_pp_format): Add test cases.
14938 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14940 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14941 missing ECF_NOTHROW flags.
14943 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14946 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14947 operands residing in integer registers.
14948 (*vsx_le_perm_load_<mode>): Likewise.
14949 (*vsx_le_perm_store_<mode>): Likewise.
14950 (define_peephole2): Add peepholes to optimize the above.
14952 2017-08-17 Marek Polacek <polacek@redhat.com>
14954 PR middle-end/81814
14955 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14956 to mimic what shorten_compare did. Change the return type to bool.
14957 (fold_cond_expr_with_comparison): Update call to
14958 operand_equal_for_comparison_p.
14959 (fold_ternary_loc): Likewise.
14961 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14963 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14965 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14967 2017-08-17 Richard Biener <rguenther@suse.de>
14969 * tree-ssa-structalias.c (solve_graph): When propagating
14970 to successors update the graphs succ edges and avoid duplicate work.
14972 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14975 * config/i386/i386.c (ix86_option_override_internal): Save target
14976 specific options after ix86_stack_protector_guard_reg was changed.
14978 2017-08-17 Richard Biener <rguenther@suse.de>
14980 PR tree-optimization/81827
14981 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14982 (new_var_info): Initialize it conservatively.
14983 (get_call_vi): Mark register vars.
14984 (new_scalar_tmp_constraint_exp): Likewise.
14985 (handle_rhs_call): Likewise.
14986 (handle_const_call): Likewise.
14987 (create_function_info_for): Likewise.
14988 (solve_constraints): Sort varinfos to separate register from
14989 non-register vars to pack points-to solution bitmaps during
14992 2017-08-17 Marek Polacek <polacek@redhat.com>
14994 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14996 2017-08-17 Richard Biener <rguenther@suse.de>
14998 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14999 to true when overflow is undefined and we saturated the result.
15001 2017-08-17 Alan Modra <amodra@gmail.com>
15004 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
15005 Don't use store multiple if only one reg needs saving.
15006 (interesting_frame_related_regno): New function.
15007 (rs6000_frame_related): Don't emit frame info for regs that
15009 (rs6000_emit_epilogue): Likewise.
15011 2017-08-16 Nathan Sidwell <nathan@acm.org>
15013 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
15014 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
15015 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
15016 * tree.c (free_lang_data_in_type): Use else-if chain. Always
15017 clear TYPE_LANG_1. Remove obsolete member-function stripping.
15018 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
15019 (verify_type): Adjust for TYPE_BINFO move.
15020 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
15021 process TYPE_BINFO directly.
15022 (hash_tree): Likewise.
15023 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15025 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15028 2017-08-16 David Malcolm <dmalcolm@redhat.com>
15030 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
15032 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
15035 * config/i386/i386.md (*anddi_1_btr): Change predicates of
15036 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
15037 Add ix86_binary_operator_ok to insn constraint.
15038 (*iordi_1_bts): Ditto.
15039 (*xordi_1_btc): Ditto.
15040 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
15041 Update corresponding peephole2 pattern.
15045 2017-08-16 Bin Cheng <bin.cheng@arm.com>
15047 PR tree-optimization/81832
15048 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
15049 copy loop header which has IFN_LOOP_DIST_ALIAS call.
15051 2017-08-16 Marek Polacek <polacek@redhat.com>
15054 * fold-const.c (fold_indirect_ref_1): Restore original behavior
15055 regarding size_zero_node.
15057 2017-08-16 Martin Liska <mliska@suse.cz>
15060 * config.gcc: Respect previously set extra_objs in case
15063 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
15065 PR tree-optimization/81835
15066 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
15067 the phi SSA_NAME. Check that the condition in a COND_EXPR does
15068 not depend on the phi.
15070 2017-08-16 Alan Modra <amodra@gmail.com>
15072 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
15075 2017-08-16 Alan Modra <amodra@gmail.com>
15077 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
15078 (save_reg_p): ..into this. Update all callers.
15079 (first_reg_to_save): Simplify.
15081 2017-08-16 Alan Modra <amodra@gmail.com>
15083 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
15086 2017-08-15 Joseph Myers <joseph@codesourcery.com>
15090 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
15091 constant count if that count is less than 32.
15093 2017-08-15 Nathan Sidwell <nathan@acm.org>
15095 * gcc.c (execute): Emit friendlier message if inferior is killed
15096 by an external cause.
15098 2017-08-15 Richard Biener <rguenther@suse.de>
15100 PR tree-optimization/81790
15101 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
15102 CONSTRUCTORs from simplifying and VN.
15104 2017-08-14 Martin Sebor <msebor@redhat.com>
15106 * builtin-attrs.def: Add comments.
15108 2017-08-14 Martin Sebor <msebor@redhat.com>
15111 * doc/extend.texi (attribute nonstring): Document new attribute.
15113 2017-08-14 Martin Sebor <msebor@redhat.com>
15116 * tree-diagnostic.c (default_tree_printer): Handle %G.
15117 * gimple-pretty-print.h (percent_G_format): Declare new function.
15118 * gimple-pretty-print.c (percent_G_format): Define.
15119 * tree-pretty-print.c (percent_K_format): Add argument.
15121 2017-08-14 Martin Sebor <msebor@redhat.com>
15123 PR translation/79998
15124 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
15125 Remove a stray space.
15127 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
15130 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
15131 (*iordi_1_bts): Ditto.
15132 (*xordi_1_btc): Ditto.
15134 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15137 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
15138 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15140 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
15141 quoted strings, and make more translator-friendly.
15142 (darwin_rs6000_override_options): Likewise.
15143 (rs6000_option_override_internal): Likewise.
15144 (rs6000_return_in_memory): Fix overlong line.
15145 (init_cmulative_args): Use quoted strings, and make more
15146 translator-friendly.
15147 (rs6000_pass_by_reference): Fix overlong line.
15148 (def_builtin): Use quoted strings.
15149 (altivec_expand_predicate_builtin): Use quoted strings, and make
15150 more translator-friendly.
15151 (htm_expand_builtin): Use quoted strings.
15152 (cpu_expand_builtin): Use quoted strings, and make more
15153 translator-friendly.
15154 (altivec_expand_builtin): Likewise.
15155 (paired_expand_predicate_builtin): Likewise.
15156 (rs6000_invalid_builtin): Likewise.
15157 (builtin_function_type): Use quoted strings.
15158 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
15159 more translator-friendly.
15160 (rs6000_trampoline_init): Likewise.
15161 (rs6000_handle_altivec_attribute): Likewise.
15162 (rs6000_inner_target_options): Use quoted strings.
15163 (rs6000_disable_incompatible_switches): Likewise.
15164 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
15165 strings, and make more translator-friendly.
15166 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
15168 2017-08-14 Bin Cheng <bin.cheng@arm.com>
15170 PR tree-optimization/81799
15171 * tree-loop-distribution.c (version_loop_by_alias_check): Force
15172 cond_expr to simple gimple operand.
15174 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
15176 PR middle-end/46932
15177 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
15179 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
15183 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
15184 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
15185 TARGET_GASISR_PROLOGUES.
15186 * config/avr/avr.c (avr_option_override): Same.
15187 (avr_pass_pre_proep::execute): Same.
15189 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
15192 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
15193 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
15195 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
15197 * config/i386/i386.md (*load_tp_<mode>): Redefine as
15198 define_insn_and_split. Split to a memory load from 0 in
15199 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
15200 using PTR mode iterator.
15201 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
15202 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
15203 (*add_tp_<mode>): Redefine as define_insn_and_split.
15204 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
15205 address space. Merge with *add_tp_x32 using PTR mode iterator.
15206 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
15207 Split to an add with a memory load from 0 in
15208 DEFAULT_TLS_SEG_REG address space.
15210 2017-08-12 Andrew Pinski <apinski@cavium.com>
15212 * config/aarch64/aarch64-option-extensions.def (rdma):
15213 Fix feature string to what Linux prints out in /proc/cpuinfo.
15215 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
15218 * dwarf2out.c (modified_type_die): For C typedef types that have
15219 an ultimate origin, process the ultimate origin instead of the
15221 (gen_typedef_die): Assert that input DECLs have no ultimate
15223 (gen_type_die_with_usage): For typedef variants that have an
15224 ultimate origin, just call gen_decl_die on the original DECL.
15225 (process_scope_var): Avoid creating DIEs for local typedefs and
15226 concrete static variables.
15228 2017-08-12 Alan Modra <amodra@gmail.com>
15232 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
15233 match gnu-user.h startfile.
15234 (ENDFILE_LINUX_SPEC): Similarly.
15236 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
15239 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
15241 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
15243 2017-08-11 Tamar Christina <tamar.christina@arm.com>
15244 * config/aarch64/aarch64.md (mov<mode>): Change.
15245 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15246 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
15247 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
15249 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
15251 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
15252 for storage order barriers.
15254 2017-08-11 Martin Liska <mliska@suse.cz>
15257 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
15258 variables of void type.
15260 2017-08-11 Martin Liska <mliska@suse.cz>
15262 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
15263 TARGET_SUPPORTS_ALIASES.
15264 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
15265 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
15266 (optimize_weakref): Likewise.
15267 * symtab.c (symtab_node::noninterposable_alias): Likewise.
15268 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
15269 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
15271 2017-08-11 Martin Liska <mliska@suse.cz>
15274 * config/i386/i386.c (make_resolver_func): Do complete
15275 refactoring of the function.
15277 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
15280 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
15281 * config/i386/i386.c (ix86_stack_protect_guard): Use
15282 ix86_stack_protect_guard_symbol_str to generate varible declaration.
15283 * doc/invoke.texi (x86 Options): Document
15284 -mstack-protector-guard-symbol= option.
15286 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
15288 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
15289 * config/i386/i386.c (ix86_split_stack_guard): New function.
15290 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
15291 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
15292 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
15293 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
15294 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
15295 (split_stack_space_check): Call ix86_split_stack_guard.
15297 2017-08-10 Martin Sebor <msebor@redhat.com>
15299 * print-tree.c (print_node): Print location using the established
15301 Replace spaces with colons.
15302 (debug_raw, debug): Ditto.
15304 2017-08-10 Martin Sebor <msebor@redhat.com>
15307 * pretty-print.c (pp_format): Correct the handling of %s precision.
15309 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
15312 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15314 (ix86_finalize_stack_frame_flags): This. Also clear
15315 frame_pointer_needed if -fno-omit-frame-pointer is used without
15317 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15318 with ix86_finalize_stack_frame_flags.
15319 (ix86_expand_epilogue): Likewise.
15320 (ix86_expand_split_stack_prologue): Likewise.
15321 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
15323 2017-08-10 Martin Liska <mliska@suse.cz>
15326 * c-attribs.c (handle_target_attribute):
15327 Report warning for an empty string argument of target attribute.
15329 2017-08-09 Jakub Jelinek <jakub@redhat.com>
15332 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
15334 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
15335 or DECL_NONLOCAL labels.
15336 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
15337 or DECL_NONLOCAL labels here.
15339 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
15341 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
15342 to indicate when early gimple folding has been disabled.
15343 (rs6000_gimple_fold_builtin): Add debug content.
15344 (rs6000_invalid_builtin): Fix whitespace.
15345 (rs6000_expand_builtin): Fix whitespace.
15346 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
15348 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
15351 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
15352 SAVE_MULTIPLE if not all the registers that saves, should be saved.
15354 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
15356 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
15357 (qdf24xx): Likewise.
15358 * config/aarch64/aarch64.md: Include falkor.md.
15359 * config/aarch64/falkor.md: New.
15361 2017-08-09 Marek Polacek <polacek@redhat.com>
15364 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
15365 * diagnostic.c (emit_diagnostic): Add a comment.
15366 (emit_diagnostic_valist): New function.
15368 2017-08-09 Marek Polacek <polacek@redhat.com>
15371 * input.c (make_location): New overload.
15372 * input.h (make_location): Declare.
15374 2017-08-08 Alan Modra <amodra@gmail.com>
15375 H.J. Lu <hongjiu.lu@intel.com>
15378 * gcc.c (NO_PIE_SPEC): Delete.
15379 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
15381 (LINK_PIE_SPEC): ..to here.
15382 (LINK_COMMAND_SPEC): Support -no-pie.
15383 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
15384 chain of crtbegin*.o selection, update for PIE_SPEC changes and
15386 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
15387 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
15388 (ENDFILE_CRTEND_SPEC): Similarly.
15390 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
15393 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
15394 (mstack-protector-guard-offset=): Ditto.
15395 * config/i386/i386.c (ix86_option_override): Handle
15396 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
15398 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
15399 ix86_stack_protect_guard_offset variables.
15400 (TARGET_STACK_PROTECT_GUARD): Always define.
15401 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
15402 and -mstack-protector-guard-offset= options.
15404 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15406 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
15407 boundary case for the last candidate.
15409 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15411 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
15412 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
15414 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15416 PR middle-end/19706
15417 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
15418 * config/aarch64/aarch64-builtins.c
15419 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
15420 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
15421 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
15423 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15424 Andrew Pinski <pinskia@gmail.com>
15426 PR middle-end/19706
15427 * internal-fn.def (XORSIGN): New.
15428 * optabs.def (xorsign_optab): New.
15429 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
15430 (convert_expand_mult_copysign): New.
15431 (pass_optimize_widening_mul::execute): Call
15432 convert_expand_mult_copysign.
15434 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15436 PR tree-optimization/81354
15437 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
15438 Insert on edges rather than explicitly creating landing pads.
15439 (analyze_candidates_and_replace): Commit edge inserts.
15441 2017-08-08 Richard Biener <rguenther@suse.de>
15443 PR middle-end/81719
15444 * tree-ssa-loop-niter.c: Include tree-dfa.h.
15445 (expand_simple_operations): Also look through ADDR_EXPRs with
15446 MEM_REF bases treating them as POINTER_PLUS_EXPR.
15448 2017-08-08 Richard Biener <rguenther@suse.de>
15450 PR tree-optimization/81723
15451 * tree-vect-slp.c (struct bst_traits): New hash traits.
15452 (bst_fail): New global.
15453 (vect_build_slp_tree_2): New worker, split out from ...
15454 (vect_build_slp_tree): ... this now wrapping it with using
15455 bst_fail set to cache SLP tree build fails. Properly handle
15457 (vect_analyze_slp_instance): Allocate and free bst_fail.
15459 2017-08-08 Martin Liska <mliska@suse.cz>
15462 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
15463 LABEL_DECLs that can be from a different function.
15465 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15467 PR tree-optimization/81744
15468 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
15469 loop's number of iterations.
15471 2017-08-08 Martin Liska <mliska@suse.cz>
15473 * asan.c: Include header files.
15474 * attribs.c (build_decl_attribute_variant): New function moved
15476 (build_type_attribute_qual_variant): Likewise.
15477 (cmp_attrib_identifiers): Likewise.
15478 (simple_cst_list_equal): Likewise.
15479 (omp_declare_simd_clauses_equal): Likewise.
15480 (attribute_value_equal): Likewise.
15481 (comp_type_attributes): Likewise.
15482 (build_type_attribute_variant): Likewise.
15483 (lookup_ident_attribute): Likewise.
15484 (remove_attribute): Likewise.
15485 (merge_attributes): Likewise.
15486 (merge_type_attributes): Likewise.
15487 (merge_decl_attributes): Likewise.
15488 (merge_dllimport_decl_attributes): Likewise.
15489 (handle_dll_attribute): Likewise.
15490 (attribute_list_equal): Likewise.
15491 (attribute_list_contained): Likewise.
15492 * attribs.h (lookup_attribute): New function moved from tree.[ch].
15493 (lookup_attribute_by_prefix): Likewise.
15494 * bb-reorder.c: Include header files.
15495 * builtins.c: Likewise.
15496 * calls.c: Likewise.
15497 * cfgexpand.c: Likewise.
15498 * cgraph.c: Likewise.
15499 * cgraphunit.c: Likewise.
15500 * convert.c: Likewise.
15501 * dwarf2out.c: Likewise.
15502 * final.c: Likewise.
15503 * fold-const.c: Likewise.
15504 * function.c: Likewise.
15505 * gimple-expr.c: Likewise.
15506 * gimple-fold.c: Likewise.
15507 * gimple-pretty-print.c: Likewise.
15508 * gimple.c: Likewise.
15509 * gimplify.c: Likewise.
15510 * hsa-common.c: Likewise.
15511 * hsa-gen.c: Likewise.
15512 * internal-fn.c: Likewise.
15513 * ipa-chkp.c: Likewise.
15514 * ipa-cp.c: Likewise.
15515 * ipa-devirt.c: Likewise.
15516 * ipa-fnsummary.c: Likewise.
15517 * ipa-inline.c: Likewise.
15518 * ipa-visibility.c: Likewise.
15520 * lto-cgraph.c: Likewise.
15521 * omp-expand.c: Likewise.
15522 * omp-general.c: Likewise.
15523 * omp-low.c: Likewise.
15524 * omp-offload.c: Likewise.
15525 * omp-simd-clone.c: Likewise.
15526 * opts-global.c: Likewise.
15527 * passes.c: Likewise.
15528 * predict.c: Likewise.
15529 * sancov.c: Likewise.
15530 * sanopt.c: Likewise.
15531 * symtab.c: Likewise.
15532 * toplev.c: Likewise.
15533 * trans-mem.c: Likewise.
15534 * tree-chkp.c: Likewise.
15535 * tree-eh.c: Likewise.
15536 * tree-into-ssa.c: Likewise.
15537 * tree-object-size.c: Likewise.
15538 * tree-parloops.c: Likewise.
15539 * tree-profile.c: Likewise.
15540 * tree-ssa-ccp.c: Likewise.
15541 * tree-ssa-live.c: Likewise.
15542 * tree-ssa-loop.c: Likewise.
15543 * tree-ssa-sccvn.c: Likewise.
15544 * tree-ssa-structalias.c: Likewise.
15545 * tree-ssa.c: Likewise.
15546 * tree-streamer-in.c: Likewise.
15547 * tree-vectorizer.c: Likewise.
15548 * tree-vrp.c: Likewise.
15549 * tsan.c: Likewise.
15550 * ubsan.c: Likewise.
15551 * varasm.c: Likewise.
15552 * varpool.c: Likewise.
15553 * tree.c: Remove functions moved to attribs.[ch].
15554 * tree.h: Likewise.
15555 * config/aarch64/aarch64.c: Add attrs.h header file.
15556 * config/alpha/alpha.c: Likewise.
15557 * config/arc/arc.c: Likewise.
15558 * config/arm/arm.c: Likewise.
15559 * config/avr/avr.c: Likewise.
15560 * config/bfin/bfin.c: Likewise.
15561 * config/c6x/c6x.c: Likewise.
15562 * config/cr16/cr16.c: Likewise.
15563 * config/cris/cris.c: Likewise.
15564 * config/darwin.c: Likewise.
15565 * config/epiphany/epiphany.c: Likewise.
15566 * config/fr30/fr30.c: Likewise.
15567 * config/frv/frv.c: Likewise.
15568 * config/ft32/ft32.c: Likewise.
15569 * config/h8300/h8300.c: Likewise.
15570 * config/i386/winnt.c: Likewise.
15571 * config/ia64/ia64.c: Likewise.
15572 * config/iq2000/iq2000.c: Likewise.
15573 * config/lm32/lm32.c: Likewise.
15574 * config/m32c/m32c.c: Likewise.
15575 * config/m32r/m32r.c: Likewise.
15576 * config/m68k/m68k.c: Likewise.
15577 * config/mcore/mcore.c: Likewise.
15578 * config/microblaze/microblaze.c: Likewise.
15579 * config/mips/mips.c: Likewise.
15580 * config/mmix/mmix.c: Likewise.
15581 * config/mn10300/mn10300.c: Likewise.
15582 * config/moxie/moxie.c: Likewise.
15583 * config/msp430/msp430.c: Likewise.
15584 * config/nds32/nds32-isr.c: Likewise.
15585 * config/nds32/nds32.c: Likewise.
15586 * config/nios2/nios2.c: Likewise.
15587 * config/nvptx/nvptx.c: Likewise.
15588 * config/pa/pa.c: Likewise.
15589 * config/pdp11/pdp11.c: Likewise.
15590 * config/powerpcspe/powerpcspe.c: Likewise.
15591 * config/riscv/riscv.c: Likewise.
15592 * config/rl78/rl78.c: Likewise.
15593 * config/rx/rx.c: Likewise.
15594 * config/s390/s390.c: Likewise.
15595 * config/sh/sh.c: Likewise.
15596 * config/sol2.c: Likewise.
15597 * config/sparc/sparc.c: Likewise.
15598 * config/spu/spu.c: Likewise.
15599 * config/stormy16/stormy16.c: Likewise.
15600 * config/tilegx/tilegx.c: Likewise.
15601 * config/tilepro/tilepro.c: Likewise.
15602 * config/v850/v850.c: Likewise.
15603 * config/vax/vax.c: Likewise.
15604 * config/visium/visium.c: Likewise.
15605 * config/xtensa/xtensa.c: Likewise.
15607 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15610 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
15611 constraints since the -mupper-regs-* switches have been
15613 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
15614 into a vector from a double word element that was extracted from
15615 another vector, and eliminate extra XXPERMDI instructions.
15616 (vsx_concat_<mode>_2): Likewise.
15617 (vsx_concat_<mode>_3): Likewise.
15618 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
15619 concat to allow optimizing inserts from previous extracts.
15621 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
15623 * config/i386/i386.c (ix86_stack_protect_guard): Generate
15624 memory reference to a SSP offset in TLS address space.
15625 (ix86_print_operand) <case '@'>: Remove.
15626 (ix86_print_operand_punct_valid_p): Remove '@' code.
15627 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
15628 UNSPEC_SP_TLS_TEST.
15629 (stack_tls_protect_set_<mode>): Remove.
15630 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
15631 (stack_tls_protect_test_<mode>): Remove.
15632 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
15634 2017-08-07 Olivier Hainque <hainque@adacore.com>
15637 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
15639 2017-08-07 Douglas Rupp <rupp@adacore.com>
15641 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
15642 variable was referenced as multidir in command.
15644 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15647 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15650 2017-08-07 Martin Liska <mliska@suse.cz>
15652 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15653 * config/rl78/rl78.c: Add include of attribs.h.
15654 * config/sh/sh.c: Likewise.
15655 * config/v850/v850.c: Likewise.
15657 2017-08-07 Tom de Vries <tom@codesourcery.com>
15659 PR middle-end/78266
15660 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15662 2017-08-07 Martin Liska <mliska@suse.cz>
15664 * config/mips/mips.c: Include attribs.h.
15666 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15669 * doc/invoke.texi: Document change in behvaior for -Ofast for
15672 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15674 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15676 (aarch64_pop_regs): Likewise.
15677 (aarch64_gen_load_pair): Likewise.
15678 (aarch64_save_callee_saves): Likewise.
15679 (aarch64_restore_callee_saves): Likewise.
15681 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15683 * config/i386/i386.c: Revert the last change.
15685 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15688 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15690 (ix86_finalize_stack_frame_flags): This. Also clear
15691 frame_pointer_needed if -fno-omit-frame-pointer is used without
15693 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15694 with ix86_finalize_stack_frame_flags.
15695 (ix86_expand_epilogue): Likewise.
15696 (ix86_expand_split_stack_prologue): Likewise.
15698 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15701 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15702 to P_AES for Westmere.
15704 2017-08-07 Jonathan Yong <10walls@gmail.com>
15706 * config/i386/mingw.opt (fset-stack-executable): Removed.
15707 * config/i386/cygming.opt (fset-stack-executable): Moved
15709 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15711 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15713 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15715 2017-08-07 Marek Polacek <polacek@redhat.com>
15717 PR middle-end/81737
15718 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15720 2017-08-07 Martin Liska <mliska@suse.cz>
15722 * attribs.h (canonicalize_attr_name): New function.
15723 (cmp_attribs): Move from c-format.c and adjusted.
15724 (is_attribute_p): Moved from tree.h.
15725 * tree-inline.c: Add new includes.
15726 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15727 (private_is_attribute_p): Remove.
15728 (private_lookup_attribute): Likewise.
15729 (private_lookup_attribute_by_prefix): Simplify.
15730 (remove_attribute): Use is_attribute_p.
15731 * tree.h: Remove removed declarations.
15733 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15735 PR middle-end/81698
15736 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15737 instead of computing it in the function. Formatting fix.
15738 (expand_case): Don't rely on default_edge being the first edge,
15739 clear it if removing it, pass default_edge to
15740 emit_case_dispatch_table.
15741 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15744 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15746 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15747 insn in the function, emit NOP after the insn.
15749 2017-08-06 Tom de Vries <tom@codesourcery.com>
15751 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15754 2017-08-06 Tom de Vries <tom@codesourcery.com>
15756 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15759 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15761 PR tree-optimization/57371
15762 * match.pd: New pattern.
15764 2017-08-04 Marek Polacek <polacek@redhat.com>
15766 PR middle-end/81695
15767 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15768 perform the computation in offset_int.
15770 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15772 PR tree-optimization/81136
15773 * tree-vectorizer.h: Include tree-hash-traits.h.
15774 (vec_base_alignments): New typedef.
15775 (vec_info): Add a base_alignments field.
15776 (vect_record_base_alignments): Declare.
15777 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15779 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15780 (create_data_ref): Add an is_conditional_in_stmt argument.
15781 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15782 the is_conditional_in_stmt field.
15783 (data_ref_loc): Add an is_conditional_in_stmt field.
15784 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15785 (find_data_references_in_stmt): Update call to create_data_ref.
15786 (graphite_find_data_references_in_stmt): Likewise.
15787 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15788 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15789 (vect_record_base_alignment): New function.
15790 (vect_record_base_alignments): Likewise.
15791 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15792 for nested statements even if we fail to compute a misalignment.
15793 Use pooled base alignments for unconditional references.
15794 (vect_find_same_alignment_drs): Compare base addresses instead
15796 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15797 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15799 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15801 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15802 Add an explicit name for the enum. Use auto_vec for slp_instances
15803 and grouped_stores.
15804 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15806 (_bb_vec_info): Add a constructor and destructor.
15807 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15808 (destroy_loop_vec_info): Delete.
15809 (vect_destroy_datarefs): Likewise.
15810 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15811 (vec_info::vec_info): New function.
15812 (vec_info::~vec_info): Likewise.
15813 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15814 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15815 destroy_loop_vec_info.
15816 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15817 (_loop_vec_info::_loop_vec_info): ...this.
15818 (destroy_loop_vec_info): Replace with...
15819 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15820 the stmt_vec_infos. Leave handling of vec_info information to its
15821 destructor. Remove explicit vector releases.
15822 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15823 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15824 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15825 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15826 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15827 (destroy_bb_vec_info): Replace with...
15828 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15829 information to its destructor.
15830 (vect_slp_analyze_bb_1): Use new and delete instead of
15831 new_bb_vec_info and destroy_bb_vec_info.
15832 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15835 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15837 * tree-data-ref.h (subscript): Add access_fn field.
15838 (data_dependence_relation): Add could_be_independent_p.
15839 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15840 (same_access_functions): Move to tree-data-ref.c.
15841 * tree-data-ref.c (ref_contains_union_access_p): New function.
15842 (access_fn_component_p): Likewise.
15843 (access_fn_components_comparable_p): Likewise.
15844 (dr_analyze_indices): Add a reference to access_fn_component_p.
15845 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15847 (constant_access_functions): Likewise.
15848 (add_other_self_distances): Likewise.
15849 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15850 (initialize_data_dependence_relation): Use XCNEW and remove
15851 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15852 of access functions that have the same type. Allow the
15853 subsequence to end with different bases in some circumstances.
15854 Record the chosen access functions in SUB_ACCESS_FN.
15855 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15856 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15857 (subscript_dependence_tester_1): Likewise dra and drb.
15858 (build_classic_dist_vector): Update calls accordingly.
15859 (subscript_dependence_tester): Likewise.
15860 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15861 DDR_COULD_BE_INDEPENDENT_P.
15862 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15863 comp_alias_ddrs instead of may_alias_ddrs.
15864 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15866 (vect_analyze_data_ref_dependence): Use it if
15867 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15868 distance vectors if that fails.
15869 (dependence_distance_ge_vf): New function.
15870 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15871 LOOP_VINFO_MAY_ALIAS_DDRS.
15873 2017-08-04 Richard Biener <rguenther@suse.de>
15875 PR middle-end/81705
15876 * fold-const.c (fold_binary_loc): Properly restrict
15877 minus_var0 && minus_var1 case when associating undefined overflow
15880 2017-08-04 Tom de Vries <tom@codesourcery.com>
15882 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15884 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15886 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15887 Don't start diagnostic messages with a capital letter.
15888 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15890 (rs6000_invalid_builtin): Likewise.
15891 (rs6000_trampoline_init): Likewise.
15893 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15896 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15897 after setting changeable df flags.
15899 2017-08-03 Richard Biener <rguenther@suse.de>
15901 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15902 up if the use is in USE - X.
15904 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15906 * toplev.c (dumpfile.h): New include.
15907 (internal_error_reentered): New static function. Use it...
15908 (internal_error_function): ...here to handle reentered internal_error.
15910 2017-08-03 Richard Biener <rguenther@suse.de>
15912 PR middle-end/81148
15913 * fold-const.c (split_tree): Add minus_var and minus_con
15914 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15915 here but always use minus_*.
15916 (associate_trees): Assert we never associate with MINUS_EXPR
15917 and NULL first operand. Do not recurse for PLUS_EXPR operands
15918 when associating as MINUS_EXPR either.
15919 (fold_binary_loc): Track minus_var and minus_con.
15921 2017-08-03 Tom de Vries <tom@codesourcery.com>
15924 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15925 ACCEL_COMPILER, apply finish_options on
15926 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15928 2017-08-03 Tom de Vries <tom@codesourcery.com>
15931 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15932 function_entry_patch_area_size > 0.
15934 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15937 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15938 instead of 10??LU, perform unit multiplication in wide_int,
15939 don't change alloc_object_size_limit if the limit is larger
15942 PR tree-optimization/81655
15943 PR tree-optimization/81588
15944 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15945 the case when ranges[i].low and high are 1 for unsigned type with
15948 PR middle-end/81052
15949 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15950 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15952 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15954 * tree-vrp.h: Add include guard.
15956 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15959 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15960 (ud2): New insn pattern.
15961 * config/i386/i386.c (ix86_expand_epilogue):
15962 For naked functions, generate ud2 instead of trap insn.
15964 2017-08-02 Marek Polacek <polacek@redhat.com>
15967 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15969 2017-08-02 Tom de Vries <tom@codesourcery.com>
15970 Cesar Philippidis <cesar@codesourcery.com>
15972 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15973 Add missing edge probabilities.
15975 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15977 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15978 Correct endianness.
15980 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15982 PR middle-end/79499
15983 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15984 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15985 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15986 prologue_seq sequences - if any.
15988 2017-08-02 Richard Biener <rguenther@suse.de>
15990 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15991 via vectors if supported, integer extracts via punning if supported
15992 or otherwise vector extracts.
15994 2017-08-02 Richard Biener <rguenther@suse.de>
15996 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15998 (bitmap_insert_into_set): ... this.
16000 2017-08-02 Richard Biener <rguenther@suse.de>
16002 PR tree-optimization/81633
16004 2015-08-17 Alan Hayward <alan.hayward@arm.com>
16006 PR tree-optimization/71752
16007 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
16009 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
16011 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
16012 (machine_function::call_ms2sysv_pad_out): Remove field.
16013 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
16014 (ix86_compute_frame_layout): Likewise.
16016 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
16019 * config/i386/i386.c (ix86_set_func_type): Disallow naked
16020 attribute with interrupt attribute.
16022 2017-08-01 Andrew Pinski <apinski@cavium.com>
16024 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
16025 BIT_INSERT_EXPR's operand 1
16026 to see if the types precision matches.
16028 2017-08-01 Martin Liska <mliska@suse.cz>
16030 PR middle-end/70140
16031 * builtins.c (expand_builtin_memcpy_args): Remove.
16032 (expand_builtin_memcpy): Call newly added function
16033 expand_builtin_memory_copy_args.
16034 (expand_builtin_memcpy_with_bounds): Likewise.
16035 (expand_builtin_mempcpy): Remove last argument.
16036 (expand_builtin_mempcpy_with_bounds): Likewise.
16037 (expand_builtin_memory_copy_args): New function created from
16038 expand_builtin_mempcpy_args with small modifications.
16039 (expand_builtin_mempcpy_args): Remove.
16040 (expand_builtin_stpcpy): Remove unused argument.
16041 (expand_builtin): Likewise.
16042 (expand_builtin_with_bounds): Likewise.
16044 2017-08-01 Martin Liska <mliska@suse.cz>
16047 Make mempcpy more optimal (PR middle-end/70140).
16049 2017-08-01 Jakub Jelinek <jakub@redhat.com>
16052 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16053 __builtin_vec_cmpne verify both arguments are compatible vectors
16054 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
16055 verify arg1_type is a pointer or array type. For __builtin_vec_st,
16056 move computation of aligned to after checking the argument types.
16060 * config/rs6000/vsx.md (vextract_fp_from_shorth,
16061 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
16064 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
16065 Doug Rupp <rupp@adacore.com>
16066 Olivier Hainque <hainque@adacore.com>
16068 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
16069 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
16071 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
16072 for TARGET_OS_CPP_BUILTIN.
16073 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
16074 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
16076 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
16077 passing required abi options to the assembler for EABI configurations.
16078 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
16079 of .text.hot and .text.unlikely sections for kernel modules when
16080 using ARM style exceptions.
16081 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
16082 options. Add EXTRA_CC1_SPEC.
16083 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
16085 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
16087 (ARM_TARGET2_DWARF_FORMAT): Define.
16088 * config/arm/t-vxworks: Adjust multilib control to removal of the
16089 Diab command line options.
16091 2017-08-01 Martin Liska <mliska@suse.cz>
16093 PR gcov-profile/81561
16094 * gcov.c (unblock): Make unblocking safe as we need to preserve
16095 index correspondence of blocks and block_lists.
16097 2017-08-01 Richard Biener <rguenther@suse.de>
16099 PR tree-optimization/81181
16100 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
16101 (compute_antic): ... end of iteration here.
16103 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
16105 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
16106 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
16107 (ftree-slp-vectorize): Likewise.
16108 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
16109 can no longer be set independent of flag_tree_loop_vectorize.
16110 * omp-general.c (emp_max_vf): Likewise.
16111 * opts.c (enable_fdo_optimizations): Remove references to
16112 flag_tree_vectorize, these are now implicit.
16113 (common_handle_option): Remove handling for OPT_ftree_vectorize,
16114 and leave it for the options machinery.
16116 2017-08-01 Martin Liska <mliska@suse.cz>
16118 PR middle-end/70140
16119 * builtins.c (expand_builtin_memcpy_args): Remove.
16120 (expand_builtin_memcpy): Call newly added function
16121 expand_builtin_memory_copy_args.
16122 (expand_builtin_memcpy_with_bounds): Likewise.
16123 (expand_builtin_mempcpy): Remove last argument.
16124 (expand_builtin_mempcpy_with_bounds): Likewise.
16125 (expand_builtin_memory_copy_args): New function created from
16126 expand_builtin_mempcpy_args with small modifications.
16127 (expand_builtin_mempcpy_args): Remove.
16128 (expand_builtin_stpcpy): Remove unused argument.
16129 (expand_builtin): Likewise.
16130 (expand_builtin_with_bounds): Likewise.
16132 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
16135 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
16136 print "ds:" only for immediates in generic address space.
16138 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
16141 * config/i386/i386.c (ix86_funciton_naked): New prototype.
16142 (ix86_function_ok_for_sibcall): Return false for naked functions.
16144 2017-08-01 Richard Biener <rguenther@suse.de>
16146 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
16147 (compute_antic): Seed worklist with exit block predecessors.
16148 * cfganal.c (dfs_find_deadend): For a cycle return the source
16149 of the edge closing it.
16151 2017-08-01 Tamar Christina <tamar.christina@arm.com>
16153 * config/aarch64/aarch64.c
16154 (aarch64_can_const_movi_rtx_p): Move 0 check.
16156 2017-08-01 Bin Cheng <bin.cheng@arm.com>
16158 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
16159 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
16161 * match.pd: Ditto in address comparison pattern.
16163 2017-08-01 Bin Cheng <bin.cheng@arm.com>
16165 PR tree-optimization/81627
16166 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
16167 closed ssa form for store-store chain.
16169 2017-08-01 Bin Cheng <bin.cheng@arm.com>
16171 PR tree-optimization/81620
16172 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
16173 for store-store chain.
16175 2017-08-01 Jakub Jelinek <jakub@redhat.com>
16177 PR tree-optimization/81588
16178 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
16179 ranges[i].in_p, invert comparison code ccode. For >/>=,
16180 swap rhs1 and rhs2 and comparison code unconditionally,
16181 for </<= don't do that. Don't swap rhs1/rhs2 again if
16182 ranges[i].in_p, instead invert comparison code ccode if
16183 opcode or oe->rank is BIT_IOR_EXPR.
16186 * optabs.def (vec_extract_optab, vec_init_optab): Change from
16187 a direct optab to conversion optab.
16188 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
16189 with GET_MODE_INNER as last argument instead of optab_handler.
16190 * expmed.c (extract_bit_field_1): Likewise. Use vector from
16191 vector extraction if possible and optab is available.
16192 * expr.c (store_constructor): Use convert_optab_handler instead
16193 of optab_handler. Use vector initialization from smaller
16194 vectors if possible and optab is available.
16195 * tree-vect-stmts.c (vectorizable_load): Likewise.
16196 * doc/md.texi (vec_extract, vec_init): Document that the optabs
16197 now have two modes.
16198 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
16199 of vec_init from half-sized vectors with the same element mode.
16200 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
16201 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
16202 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
16203 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
16204 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
16205 after mode in gen_vec_extract* calls.
16206 (vec_extract<mode>): Renamed to ...
16207 (vec_extract<mode><ssescalarmodelower>): ... this.
16208 (vec_extract<mode><ssehalfvecmodelower>): New expander.
16209 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
16210 element mode after mode in gen_vec_init* calls.
16211 (VEC_INIT_HALF_MODE): New mode iterator.
16212 (vec_init<mode>): Renamed to ...
16213 (vec_init<mode><ssescalarmodelower>): ... this.
16214 (vec_init<mode><ssehalfvecmodelower>): New expander.
16215 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
16216 (vec_extractv2sfsf): ... this.
16217 (vec_initv2sf): Renamed to ...
16218 (vec_initv2sfsf): ... this.
16219 (vec_extractv2si): Renamed to ...
16220 (vec_extractv2sisi): ... this.
16221 (vec_initv2si): Renamed to ...
16222 (vec_initv2sisi): ... this.
16223 (vec_extractv4hi): Renamed to ...
16224 (vec_extractv4hihi): ... this.
16225 (vec_initv4hi): Renamed to ...
16226 (vec_initv4hihi): ... this.
16227 (vec_extractv8qi): Renamed to ...
16228 (vec_extractv8qiqi): ... this.
16229 (vec_initv8qi): Renamed to ...
16230 (vec_initv8qiqi): ... this.
16231 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
16232 (vec_init<mode>): Renamed to ...
16233 (vec_init<mode><VEC_base_l>): ... this.
16234 (vec_extract<mode>): Renamed to ...
16235 (vec_extract<mode><VEC_base_l>): ... this.
16236 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
16237 (vec_initv2sfsf): ... this.
16238 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
16239 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
16240 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
16241 element mode after mode in gen_vec_init* calls.
16242 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
16243 (vec_init<mode><Vel>): ... this.
16244 (vec_extract<mode>): Renamed to ...
16245 (vec_extract<mode><Vel>): ... this.
16246 * config/aarch64/iterators.md (Vel): New mode attribute.
16247 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
16248 Add element mode after mode in gen_vec_extract* calls.
16249 * config/s390/vector.md (non_vec_l): New mode attribute.
16250 (vec_extract<mode>): Renamed to ...
16251 (vec_extract<mode><non_vec_l>): ... this.
16252 (vec_init<mode>): Renamed to ...
16253 (vec_init<mode><non_vec_l>): ... this.
16254 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
16255 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
16257 * config/arm/iterators.md (V_elem_l): New mode attribute.
16258 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
16259 (vec_extract<mode><V_elem_l>): ... this.
16260 (vec_extractv2di): Renamed to ...
16261 (vec_extractv2didi): ... this.
16262 (vec_init<mode>): Renamed to ...
16263 (vec_init<mode><V_elem_l>): ... this.
16264 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
16265 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
16266 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
16267 Add element mode after gen_vec_extract* calls.
16268 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
16269 (vec_init<mode><unitmode>): ... this.
16270 (vec_extract<mode>): Renamed to ...
16271 (vec_extract<mode><unitmode>): ... this.
16272 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
16273 (vec_init<mode><unitmode>): ... this.
16274 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
16275 (vec_initv2sfsf): ... this.
16276 (vec_extractv2sf): Renamed to ...
16277 (vec_extractv2sfsf): ... this.
16278 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
16279 Add element mode after gen_vec_extract* calls.
16280 * config/mips/mips.md (unitmode): New mode iterator.
16281 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
16282 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
16283 * config/spu/spu.md (inner_l): New mode attribute.
16284 (vec_init<mode>): Renamed to ...
16285 (vec_init<mode><inner_l>): ... this.
16286 (vec_extract<mode>): Renamed to ...
16287 (vec_extract<mode><inner_l>): ... this.
16288 * config/sparc/sparc.md (veltmode): New mode iterator.
16289 (vec_init<VMALL:mode>): Renamed to ...
16290 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
16291 * config/ia64/vect.md (vec_initv2si): Renamed to ...
16292 (vec_initv2sisi): ... this.
16293 (vec_initv2sf): Renamed to ...
16294 (vec_initv2sfsf): ... this.
16295 (vec_extractv2sf): Renamed to ...
16296 (vec_extractv2sfsf): ... this.
16297 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
16298 (vec_init<mode>): Renamed to ...
16299 (vec_init<mode><VEC_base_l>): ... this.
16300 (vec_extract<mode>): Renamed to ...
16301 (vec_extract<mode><VEC_base_l>): ... this.
16302 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
16303 (vec_initv2sfsf): ... this.
16304 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
16305 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
16306 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
16307 gen_vec_init* calls.
16309 2017-08-01 Richard Biener <rguenther@suse.de>
16311 PR tree-optimization/81297
16312 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
16313 TREE_OVERFLOW from INTEGER_CSTs.
16315 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
16317 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
16319 2017-07-31 Carl Love <cel@us.ibm.com>
16321 * config/rs6000/rs6000-c: Add support for built-in functions
16322 vector signed char vec_xl_be (signed long long, signed char *);
16323 vector unsigned char vec_xl_be (signed long long, unsigned char *);
16324 vector signed int vec_xl_be (signed long long, signed int *);
16325 vector unsigned int vec_xl_be (signed long long, unsigned int *);
16326 vector signed long long vec_xl_be (signed long long, signed long long *);
16327 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
16328 vector signed short vec_xl_be (signed long long, signed short *);
16329 vector unsigned short vec_xl_be (signed long long, unsigned short *);
16330 vector double vec_xl_be (signed long long, double *);
16331 vector float vec_xl_be (signed long long, float *);
16332 * config/rs6000/altivec.h (vec_xl_be): Add #define.
16333 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
16334 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
16336 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
16337 (altivec_expand_builtin): Add switch statement to call
16338 altivec_expand_xl_be for each builtin.
16339 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
16340 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
16341 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
16342 __builtin_vsx_le_be_v16qi.
16343 * doc/extend.texi: Update the built-in documentation file for the
16344 new built-in functions.
16346 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16349 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
16351 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16353 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16355 * config.gcc: Add z14.
16356 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
16357 CPU model numbers for z13s and z14.
16358 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
16360 * config/s390/s390-opts.h (enum processor_type): Rename
16361 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16362 * config/s390/s390.c (processor_table): Add field for CPU name to
16363 be passed to Binutils.
16364 (s390_asm_output_machine_for_arch): Use the new field in
16365 processor_table for Binutils.
16366 (s390_expand_builtin): Replace arch12 with z14.
16367 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16368 (s390_get_sched_attrmask): Likewise.
16369 (s390_get_unit_mask): Likewise.
16370 * config/s390/s390.opt: Add z14 to processor_type enum.
16372 2017-07-31 Martin Jambor <mjambor@suse.cz>
16375 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
16376 regardless of optimization level.
16378 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
16379 Martin Liska <mliska@suse.cz>
16381 * predict.def: Remove old comment and adjust probability.
16382 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
16383 PREDICT statements.
16385 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16388 * config/i386/i386.c (ix86_function_naked): New function.
16389 (ix86_can_use_return_insn_p): Return false for naked functions.
16390 (ix86_expand_prologue): Skip prologue for naked functions.
16391 (ix86_expand_epilogue): Skip epilogue for naked functions
16392 and emit trap instruction.
16393 (ix86_warn_func_return): New function.
16394 (ix86_attribute_table): Add "naked" attribute specification.
16395 (TARGET_WARN_FUNC_RETURN): Define.
16396 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
16398 2017-07-31 Martin Liska <mliska@suse.cz>
16400 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
16401 (dump_gimple_bb_header): Always dump BB info.
16402 (pp_cfg_jump): Do not append info about BB when dumping a jump.
16404 2017-07-31 Martin Liska <mliska@suse.cz>
16407 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
16408 also with current_function_decl non-null equality.
16410 2017-07-31 Jakub Jelinek <jakub@redhat.com>
16413 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
16414 change type to the element type, instead add eltype variable and
16415 use it where we are interested in the element type.
16417 PR tree-optimization/81603
16418 * ipa-polymorphic-call.c
16419 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
16420 offset arithmetic in offset_int, bail out if the resulting bit offset
16421 doesn't fit into shwi.
16423 2017-07-31 Martin Liska <mliska@suse.cz>
16425 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
16426 (gimplify_save_expr): Fix comment.
16428 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
16431 * config/i386/i386.c (ix86_function_arg): Update arguments for
16433 (ix86_compute_frame_layout): Set the initial stack offset to
16434 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
16435 INCOMING_FRAME_SP_OFFSET.
16436 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
16437 stack before exception handler returns.
16438 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
16439 the 'ERROR_CODE' for exception handler.
16441 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
16443 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
16444 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
16445 (ASM_OUTPUT_REG_POP): Ditto.
16446 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
16447 instead of asm_fprintf to output pure string.
16449 2017-07-29 Jakub Jelinek <jakub@redhat.com>
16451 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
16452 to imported_module_or_decl hook.
16453 (debug_nothing_tree_tree_tree_bool): Remove.
16454 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
16455 * debug.c (do_nothing_debug_hooks): Use
16456 debug_nothing_tree_tree_tree_bool_bool instead of
16457 debug_nothing_tree_tree_tree_bool.
16458 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16459 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
16460 * sdbout.c (sdb_debug_hooks): Likewise.
16461 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
16462 (gen_namespace_die): Add DW_AT_export_symbols attribute if
16464 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
16465 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
16466 attribute, don't add anything.
16468 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16470 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
16471 (fold_build2_stat_loc): Likewise.
16472 (fold_build3_stat_loc): Likewise.
16473 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
16474 (fold_build1_loc): Remove macro.
16475 (fold_build2_loc): Likewise.
16476 (fold_build3_loc): Likewise.
16478 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16480 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
16481 (gimple_build_debug_bind_source_stat): Likewise.
16482 * gimple.h (gimple_build_debug_bind): Remove macro.
16483 (gimple_build_debug_bind_source): Likewise.
16485 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16487 * bitmap.c (bitmap_alloc): Adjust.
16488 (bitmap_gc_alloc): Likewise.
16489 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
16491 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16493 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
16494 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
16495 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
16496 (bitmap_gc_alloc_stat): Likewise.
16497 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
16499 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16501 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
16502 * rtl.h (shallow_copy_rtx): Remove macro.
16504 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16506 * emit-rtl.c (gen_raw_REG): Adjust.
16507 * gengenrtl.c (gendef): Likewise.
16508 * rtl.c (rtx_alloc_stat): Remove _stat from name.
16509 * rtl.h (rtx_alloc): Remove macro.
16511 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16513 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
16514 (build_tree_list_stat): Likewise.
16515 * tree.h (build_tree_list): Remove macro.
16516 (build_tree_list_vec): Likewise.
16518 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16520 * tree.c (make_vector_stat): Remove _stat from name.
16521 (build_vector_stat): Likewise.
16522 * tree.h (make_vector_stat): Remove macro.
16523 (build_vector_stat): Likewise.
16525 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16527 * tree.h (build_var_debug_value): Remove prototype.
16529 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16531 * tree.c (tree_cons_stat): Remove _stat from name.
16532 * tree.h (tree_cons): Remove macro.
16534 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16536 * tree.c (build_vl_exp_stat): Remove _stat from name.
16537 * tree.h (build_vl_exp): Remove macro.
16539 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16541 * tree.c (build_decl_stat): Remove _stat from name.
16542 * tree.h (build_decl): Remove macro.
16544 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16546 * gimple.c (gimple_build_with_ops_stat): Adjust.
16547 (gimple_alloc_stat): Remove _stat from name.
16548 * gimple.h (gimple_alloc): Remove macro.
16550 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16552 * tree.c (make_tree_vec_stat): Remove _stat from name.
16553 (grow_tree_vec_stat): Likewise.
16554 * tree.h (make_tree_vec_stat): Adjust prototype.
16555 (grow_tree_vec_stat): Likewise.
16556 (make_tree_vec): Remove macro.
16557 (grow_tree_vec): Likewise.
16559 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16561 * fold-const.c (fold_build1_stat_loc): Adjust.
16562 (fold_build2_stat_loc): Likewise.
16563 (fold_build3_stat_loc): Likewise.
16564 * tree.c (build0_stat): Remove _stat from name.
16565 (build1_stat): Likewise.
16566 (build2_stat): Likewise.
16567 (build3_stat): Likewise.
16568 (build4_stat): Likewise.
16569 (build5_stat): Likewise.
16570 * tree.h (build1_loc): Remove macro, and rename _stat function
16572 (build2_loc): Likewise.
16573 (build3_loc): Likewise.
16574 (build4_loc): Likewise.
16575 (build5_loc): Likewise.
16577 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16579 * tree.c (make_int_cst_stat): Remove _stat from name.
16580 * tree.h (make_int_cst_stat): Adjust prototype.
16581 (make_int_cst): Remove macro.
16583 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16585 * tree.c (make_tre_binfo_stat): Remove _stat from name.
16586 * tree.h (make_tree_binfo_stat): Adjust prototype.
16587 (make_tree_binfo): Remove.
16589 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16591 * tree.c (copy_node_stat): Rename to copy_node.
16592 (build_distinct_type_copy): Adjust.
16593 * tree.h (copy_node_stat): Adjust prototype.
16594 (copy_node): Remove macro.
16596 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16598 * tree.c (make_node_stat): rename to make_node.
16599 (build_tree_list_stat): Adjust.
16600 (build0_stat): Likewise.
16601 (build2_stat): Likewise.
16602 (build3_stat): Likewise.
16603 (build4_stat): Likewise.
16604 (build5_stat): Likewise.
16605 (build_decl_stat): Likewise.
16606 * tree.h (make_node_stat): Adjust prototype.
16607 (make_node): remove macro.
16609 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
16611 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
16612 (PPC_FEATURE2_SCV): Likewise.
16613 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
16615 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16617 * config/aarch64/aarch64.c
16618 (aarch64_internal_mov_immediate): Add new special pattern.
16619 * config/aarch64/aarch64.md (*movdi_aarch64):
16620 Add reg/32bit const mov case.
16622 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16623 Richard Sandiford <richard.sandiford@linaro.org>
16625 * config/aarch64/aarch64.md (mov<mode>): Generalize.
16626 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16627 Add integer and movi cases.
16628 (movi-split-hf-df-sf split, fp16): New.
16629 (enabled): Added TARGET_FP_F16INST.
16630 * config/aarch64/iterators.md (GPF_HF): New.
16631 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
16633 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16635 * config/aarch64/aarch64.c
16636 (aarch64_simd_container_mode): Add prototype.
16637 (aarch64_expand_mov_immediate): Add HI support.
16638 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
16639 (aarch64_can_const_movi_rtx_p): New.
16640 (aarch64_preferred_reload_class):
16641 Remove restrictions of using FP registers for certain SIMD operations.
16642 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
16643 (aarch64_valid_floating_const): Add integer move validation.
16644 (aarch64_simd_imm_scalar_p): Remove.
16645 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
16646 (aarch64_legitimate_constant_p): Expand list of supported cases.
16647 * config/aarch64/aarch64-protos.h
16648 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16649 (aarch64_reinterpret_float_as_int): New.
16650 (aarch64_simd_imm_scalar_p): Remove.
16651 * config/aarch64/constraints.md (Uvi): New.
16652 (Dd): Split into Ds and new Dd.
16653 * config/aarch64/aarch64.md (*movsi_aarch64):
16655 (*movdi_aarch64): Add SIMD mov case.
16657 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16659 * tree-predcom.c: (struct chain): Handle store-store chain in which
16660 stores for elimination only store loop invariant values.
16661 (execute_pred_commoning_chain): Ditto.
16662 (prepare_initializers_chain_store_elim): Ditto.
16663 (prepare_finalizers): Ditto.
16664 (is_inv_store_elimination_chain): New function.
16665 (initialize_root_vars_store_elim_1): New function.
16667 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16669 * tree-predcom.c: Revise general description of the pass.
16670 (enum chain_type): New enum type for store elimination.
16671 (struct chain): New field supporting store elimination.
16672 (struct component): Ditto.
16673 (dump_chain): Dump store-stores chain.
16674 (release_chain): Release resources.
16675 (split_data_refs_to_components): Compute and create component
16676 contains only stores for elimination.
16677 (get_chain_last_ref_at): New function.
16678 (make_invariant_chain): Initialization.
16679 (make_rooted_chain): Specify chain type in parameter and record it.
16680 (add_looparound_copies): Skip for store-stores chain.
16681 (determine_roots_comp): Compute type of chain and pass it to
16683 (initialize_root_vars_store_elim_2): New function.
16684 (finalize_eliminated_stores): New function.
16685 (remove_stmt): Handle store for elimination.
16686 (execute_pred_commoning_chain): Execute predictive commoning on
16687 store-store chains.
16688 (determine_unroll_factor): Skip unroll for store-stores chain.
16689 (prepare_initializers_chain_store_elim): New function.
16690 (prepare_initializers_chain): Hanlde store-store chain.
16691 (prepare_finalizers_chain, prepare_finalizers): New function.
16692 (tree_predictive_commoning_loop): Return integer value indicating
16693 if loop is unrolled or lcssa form is corrupted.
16694 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16696 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16698 * tree-predcom.c (initialize_root): Delete.
16699 (execute_pred_commoning_chain): Initialize root vars and replace
16700 reference of non-combined chain directly, rather than call above
16703 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16705 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16706 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16708 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16710 * tree-predcom.c (struct chain): New field init_seq.
16711 (release_chain): Release init_seq.
16712 (prepare_initializers_chain): Record intialization stmts in above
16714 (insert_init_seqs): New function.
16715 (tree_predictive_commoning_loop): Call insert_init_seqs.
16717 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16719 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16721 2017-07-28 Richard Biener <rguenther@suse.de>
16723 * match.pd: Remove superfluous :c.
16724 * genmatch.c (simplify::id): Add member.
16725 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16727 (current_id): New global.
16728 (dt_node::parent): Move from ...
16729 (dt_operand::parent): ... here. Add for_id member.
16730 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16731 (decision_tree::find_node): Relax order requirement when
16732 merging DT_TRUE nodes to ones inbetween the current simplify
16733 and the one we try to merge with. Add diagnostic whenever
16734 we need to enforce pattern order by not merging.
16735 (decision_tree::insert): Set current_id.
16736 (decision_tree::print_node): Dump parent node and for_id.
16737 (parser::last_id): Add member.
16738 (parser::push_simplify): Assign unique id.
16739 (parser::parser): Initialize last_id.
16741 2017-07-28 Martin Liska <mliska@suse.cz>
16744 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16745 gimple_build_debug_bind.
16747 2017-07-28 Richard Biener <rguenther@suse.de>
16749 PR tree-optimization/81502
16750 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16752 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16754 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16755 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16756 for BIT_FIELD_REF args.
16757 * fold-const.c (make_bit_field_ref): Likewise.
16758 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16760 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16763 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16764 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16765 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16766 Or it into SANITIZER_UNDEFINED.
16767 * ubsan.c: Include gimple-fold.h and varasm.h.
16768 (ubsan_expand_ptr_ifn): New function.
16769 (instrument_pointer_overflow): New function.
16770 (maybe_instrument_pointer_overflow): New function.
16771 (instrument_object_size): Formatting fix.
16772 (pass_ubsan::execute): Call instrument_pointer_overflow
16773 and maybe_instrument_pointer_overflow.
16774 * internal-fn.c (expand_UBSAN_PTR): New function.
16775 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16776 * sanitizer.def (__ubsan_handle_pointer_overflow,
16777 __ubsan_handle_pointer_overflow_abort): New builtins.
16778 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16779 * internal-fn.def (UBSAN_PTR): New internal function.
16780 * opts.c (sanitizer_opts): Add pointer-overflow.
16781 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16782 * fold-const.c (build_range_check): Compute pointer range check in
16783 integral type if pointer arithmetics would be needed. Formatting
16786 2017-07-28 Martin Liska <mliska@suse.cz>
16789 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16790 parameters that are of a variable-length.
16792 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16794 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16796 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16797 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16798 (CRT_CALL_STATIC_FUNCTION): Likewise.
16799 (ASM_DEFAULT_SPEC): New define.
16800 (ASM_SPEC32): Likewise.
16801 (ASM_SPEC64): Likewise.
16802 (ASM_SPEC_COMMON): Likewise.
16803 (ASM_SPEC): Likewise.
16804 (INVALID_64BIT): Likewise.
16805 (LINK_OS_DEFAULT_SPEC): Likewise.
16806 (LINK_OS_SPEC32): Likewise.
16807 (LINK_OS_SPEC64): Likewise.
16808 (POWERPC_LINUX): Likewise.
16809 (PTRDIFF_TYPE): Likewise.
16810 (RESTORE_FP_PREFIX): Likewise.
16811 (RESTORE_FP_SUFFIX): Likewise.
16812 (SAVE_FP_PREFIX): Likewise.
16813 (SAVE_FP_SUFFIX): Likewise.
16814 (SIZE_TYPE): Likewise.
16815 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16816 (TARGET_64BIT): Likewise.
16817 (TARGET_64BIT): Likewise.
16818 (TARGET_AIX): Likewise.
16819 (WCHAR_TYPE_SIZE): Likewise.
16820 (WCHAR_TYPE): Undefine.
16821 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16822 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16823 (CPP_OS_RTEMS_SPEC): Delete.
16824 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16825 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16827 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16829 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16831 PR tree-optimization/81578
16832 * tree-parloops.c (build_new_reduction): Bail out if
16833 reduction_code isn't one of the standard OpenMP reductions.
16834 Move the details printing after that decision.
16836 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16838 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16839 related to reload_in_progress.
16840 (splat_input_operand): Likewise.
16841 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16843 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16845 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16846 (TARGET_INSTANTIATE_DECLS): Likewise.
16847 (legitimate_indexed_address_p): Delete reload_in_progress code.
16848 (rs6000_debug_legitimate_address_p): Likewise.
16849 (rs6000_eliminate_indexed_memrefs): Likewise.
16850 (rs6000_emit_le_vsx_store): Likewise.
16851 (rs6000_emit_move_si_sf_subreg): Likewise.
16852 (rs6000_emit_move): Likewise.
16853 (register_to_reg_type): Likewise.
16854 (rs6000_pre_atomic_barrier): Likewise.
16855 (rs6000_machopic_legitimize_pic_address): Likewise.
16856 (rs6000_allocate_stack_temp): Likewise.
16857 (rs6000_address_for_fpconvert): Likewise.
16858 (rs6000_address_for_altivec): Likewise.
16859 (rs6000_secondary_memory_needed_rtx): Delete function.
16860 (rs6000_check_sdmode): Likewise.
16861 (rs6000_alloc_sdmode_stack_slot): Likewise.
16862 (rs6000_instantiate_decls): Likewise.
16863 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16864 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16865 Delete reload_in_progress.
16866 (*vec_reload_and_plus_<mptrsize>): Likewise.
16867 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16868 (vsx_div_v2di): Likewise.
16869 (vsx_udiv_v2di): Likewise.
16871 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16873 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16874 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16875 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16876 (rs6000_debug_reg_global): Delete print of LRA status.
16877 (rs6000_option_override_internal): Delete dead LRA related code.
16878 (rs6000_lra_p): Delete function.
16879 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16881 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16883 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16884 * config/riscv/rtems.h: New file.
16886 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16887 Sudakshina Das <sudi.das@arm.com>
16889 * config/aarch64/aarch64.md
16890 (define_split for and<mode>3nr_compare): Move
16891 non aarch64_logical_operand to a register.
16892 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16893 register immediate operand to a register.
16894 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16896 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16898 PR middle-end/81564
16899 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16901 2017-07-27 Richard Biener <rguenther@suse.de>
16903 PR tree-optimization/81573
16904 PR tree-optimization/81494
16905 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16906 multi defuse cycle case.
16908 2017-07-27 Richard Biener <rguenther@suse.de>
16910 PR tree-optimization/81571
16911 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16914 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16916 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16917 earlier and only if MASK_FPU is set. Adjust formatting.
16919 2017-07-27 Martin Liska <mliska@suse.cz>
16921 * opt-functions.awk: Add validation of value of Init.
16922 * optc-gen.awk: Pass new argument.
16924 2017-07-27 Martin Liska <mliska@suse.cz>
16926 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16927 Fix wrong condition.
16929 2017-07-27 Martin Liska <mliska@suse.cz>
16931 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16932 BBs and edges seen by autoFDO.
16934 2017-07-27 Richard Biener <rguenther@suse.de>
16936 PR tree-optimization/81502
16937 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16938 with incompatible but same sized type.
16939 (execute_update_addresses_taken): Likewise.
16941 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16943 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16944 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16946 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16949 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16950 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16951 Change s_operand to memory_operand.
16953 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16955 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16956 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16957 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16958 Emit instructions rather than returning an expression. Handle TFmode
16959 and KFmode by casting to TImode.
16960 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16961 (rs6000_emit_le_vsx_store): Likewise.
16962 * config/rs6000/vsx.md (VSX_TI): New iterator.
16963 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16964 (*vsx_le_undo_permute_<mode>): Likewise.
16965 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16966 emit the split sequence.
16967 (*vsx_le_perm_store_<mode>): Likewise.
16969 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16971 PR tree-optimization/81555
16972 PR tree-optimization/81556
16973 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16974 if true, force CHANGED for the recursive invocation.
16975 (reassociate_bb): Remember original length of ops array, pass
16976 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16978 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16979 attributes for noipa attribute. For naked attribute use
16980 lookup_attribute first before lookup_attribute_spec.
16981 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16983 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16984 for functions with noipa attribute.
16985 (cgraph_externally_visible_p): Return true for functions with noipa
16987 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16988 for functions with noipa attribute.
16989 * doc/extend.texi: Document noipa function attribute.
16990 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16991 also for functions with noipa attribute.
16992 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16994 2017-07-26 Andrew Pinski <apinski@cavium.com>
16996 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16997 vec_unalign_load_cost and vec_unalign_store_cost.
16999 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
17001 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
17002 -mvsx-small-integer option.
17003 (ISA_3_0_MASKS_IEEE): Likewise.
17004 (OTHER_VSX_VECTOR_MASKS): Likewise.
17005 (POWERPC_MASKS): Likewise.
17006 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
17007 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
17008 code, only testing for DImode being allowed in non-VSX floating
17010 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
17011 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
17013 (rs6000_option_override_internal): Delete -mvsx-small-integer.
17014 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
17015 TARGET_P8_VECTOR test.
17016 (rs6000_secondary_reload_simple_move): Likewise.
17017 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
17018 since TARGET_P9_VECTOR was already tested.
17019 (rs6000_opt_masks): Remove -mvsx-small-integer.
17020 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
17021 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17023 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
17024 test for TARGET_VEXTRACTUB was used, and that uses
17026 (p9 extract splitter): Likewise.
17027 (vsx_extract_<mode>_di_p9): Likewise.
17028 (vsx_extract_<mode>_store_p9): Likewise.
17029 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
17030 for TARGET_P9_VECTOR was used. Delete code that is now dead with
17031 the elimination of TARGET_VSX_SMALL_INTEGER.
17032 (vsx_extract_<mode>_p8): Likewise.
17033 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
17034 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
17035 (vsx_set_<mode>_p9): Likewise.
17036 (vsx_set_v4sf_p9): Likewise.
17037 (vsx_set_v4sf_p9_zero): Likewise.
17038 (vsx_insert_extract_v4sf_p9): Likewise.
17039 (vsx_insert_extract_v4sf_p9_2): Likewise.
17040 * config/rs6000/rs6000.md (sign extend splitter): Change
17041 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
17042 (floatsi<mode>2_lfiwax_mem): Likewise.
17043 (floatunssi<mode>2_lfiwzx_mem): Likewise.
17044 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
17045 since a test for TARGET_P9_VECTOR was used.
17046 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17047 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
17048 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17049 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
17050 TARGET_P8_VECTOR test.
17051 (fix_trunc<mode>si2_stfiwx): Likewise.
17052 (fix_trunc<mode>si2_internal): Likewise.
17053 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
17054 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17056 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17057 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
17058 TARGET_P8_VECTOR test.
17059 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17060 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
17061 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
17063 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
17064 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
17065 since a test for TARGET_P9_VECTOR was used.
17066 (splitter for loading small constants): Likewise.
17068 2017-07-26 Andrew Pinski <apinski@cavium.com>
17070 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
17073 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
17076 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
17077 (fp_valid_at): Likewise.
17079 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
17081 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
17082 (qdf24xx_addrcost_table): Likewise.
17083 (cortexa57_tunings): Update to use generic_branch_cost.
17084 (cortexa72_tunings): Likewise.
17085 (cortexa73_tunings): Likewise.
17086 (qdf24xx_tunings): Likewise.
17088 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
17090 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
17091 (thunderx2t99_branch_cost): Likewise.
17092 (cortexa35_tunings): Update to use generic_branch_cost.
17093 (cortexa53_tunings): Likewise.
17094 (cortexa57_tunings): Likewise.
17095 (cortexa72_tunings): Likewise.
17096 (cortexa73_tunings): Likewise.
17097 (thunderx2t99_tunings): Likewise.
17099 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
17101 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
17102 (sparc_option_override): Honour MASK_FSMULD.
17103 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
17104 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
17105 * config/sparc/sparc.opt (mfsmuld): New option.
17106 * doc/invoke.texi (mfsmuld): Document option.
17108 2017-07-26 Marek Polacek <polacek@redhat.com>
17110 PR middle-end/70992
17111 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
17113 2017-07-26 Richard Biener <rguenther@suse.de>
17115 * gimple-match-head.c (do_valueize): Return OP if valueize
17117 (get_def): New helper to get at the def stmt of a SSA name
17118 if valueize allows.
17119 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
17120 do_valueize to get at the def stmt.
17121 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
17123 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
17125 PR middle-end/46932
17126 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
17128 2017-07-26 Martin Liska <mliska@suse.cz>
17131 * function.c (expand_function_start): Make expansion of
17132 nonlocal_goto_save_area after parm_birth_insn.
17134 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
17136 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
17137 from all CPU target flags enable members.
17139 2017-07-26 Richard Biener <rguenther@suse.de>
17141 * genmatch.c (dt_simplify::gen): Make iterator vars const.
17142 (decision_tree::gen): Make 'type' const.
17143 (write_predicate): Likewise.
17145 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
17147 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
17148 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
17149 (rs6000_option_override_internal): Likewise.
17150 (rs6000_expand_vector_set): Likewise.
17151 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
17152 (TARGET_UPPER_REGS_SF): Likewise.
17153 (TARGET_UPPER_REGS_DI): Likewise.
17154 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
17155 (TARGET_DIRECT_MOVE_64BIT): Likewise.
17156 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
17157 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
17158 (Splitters for DI constants in Altivec registers): Likewise.
17159 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
17160 (vsx_set_v4sf_p9): Likewise.
17161 (vsx_set_v4sf_p9_zero): Likewise.
17162 (vsx_insert_extract_v4sf_p9): Likewise.
17163 (vsx_insert_extract_v4sf_p9_2): Likewise.
17165 2017-07-25 Carl Love <cel@us.ibm.com>
17167 * doc/extend.texi: Update the built-in documentation file for the
17168 existing built-in functions
17169 vector signed char vec_cnttz (vector signed char);
17170 vector unsigned char vec_cnttz (vector unsigned char);
17171 vector signed short vec_cnttz (vector signed short);
17172 vector unsigned short vec_cnttz (vector unsigned short);
17173 vector signed int vec_cnttz (vector signed int);
17174 vector unsigned int vec_cnttz (vector unsigned int);
17175 vector signed long long vec_cnttz (vector signed long long);
17176 vector unsigned long long vec_cnttz (vector unsigned long long);
17178 2017-07-25 Andrew Pinski <apinski@cavium.com>
17180 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
17181 accesses where the use is for the first operand of a BIT_INSERT.
17183 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
17186 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17187 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17189 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
17191 * config/i386/gstabs.h: Delete.
17192 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
17194 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
17196 * config/i386/i386.c (ix86_decompose_address): Do not check for
17197 register RTX when looking at index_reg or base_reg.
17198 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
17200 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
17202 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
17203 to update EH info here.
17205 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17207 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
17209 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17211 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
17213 2017-07-25 Torsten Duwe <duwe@suse.de>
17215 * common.opt: Introduce -fpatchable-function-entry
17216 command line option, and its variables function_entry_patch_area_size
17217 and function_entry_patch_area_start.
17218 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
17219 including a two-value parser.
17220 * target.def (print_patchable_function_entry): New target hook.
17221 * targhooks.h (default_print_patchable_function_entry): New function.
17222 * targhooks.c (default_print_patchable_function_entry): Likewise.
17223 * toplev.c (process_options): Switch off IPA-RA if
17224 patchable function entries are being generated.
17225 * varasm.c (assemble_start_function): Look at the
17226 patchable-function-entry command line switch and current
17227 function attributes and maybe generate NOP instructions by
17228 calling the print_patchable_function_entry hook.
17229 * doc/extend.texi: Document patchable_function_entry attribute.
17230 * doc/invoke.texi: Document -fpatchable_function_entry
17231 command line option.
17232 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
17234 * doc/tm.texi: Re-generate.
17236 2017-07-25 Jakub Jelinek <jakub@redhat.com>
17239 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
17240 TARGET_AVX512DQ rather than TARGET_AVX512BW.
17242 2017-07-25 Tamar Christina <tamar.christina@arm.com>
17244 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
17246 2017-07-25 Richard Biener <rguenther@suse.de>
17248 PR tree-optimization/81455
17249 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
17250 not walk in cycles when looking for guards.
17252 2017-07-25 Richard Biener <rguenther@suse.de>
17254 PR tree-optimization/81529
17255 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
17256 when optimizing backedge uses.
17258 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
17260 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
17262 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
17263 to dl_section_ref. On AIX, append an expression to subtract
17264 the size of the section length to dl_section_ref.
17266 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
17268 * configure.ac: If any of the config.* scripts fail, exit 1.
17269 * configure: Regenerate.
17271 2017-07-25 Richard Biener <rguenther@suse.de>
17273 PR middle-end/81546
17274 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
17275 of immediate uses, be more verbose on errors.
17277 2017-07-25 Richard Biener <rguenther@suse.de>
17279 PR tree-optimization/81510
17280 * tree-vect-loop.c (vect_is_simple_reduction): When the
17281 reduction stmt is not inside the loop bail out.
17283 2017-07-25 Richard Biener <rguenther@suse.de>
17285 PR tree-optimization/81303
17286 * tree-vect-loop-manip.c (vect_loop_versioning): Build
17287 profitability check against LOOP_VINFO_NITERSM1.
17289 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17291 * domwalk.c (cmp_bb_postorder): Simplify.
17292 (sort_bbs_postorder): New function. Use it...
17293 (dom_walker::walk): ...here to optimize common cases.
17295 2017-07-25 Martin Liska <mliska@suse.cz>
17298 * ipa-visibility.c (function_and_variable_visibility): Make the
17299 redirection just on target that supports aliasing.
17300 Fix GNU coding style.
17302 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
17305 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
17307 * config/arm/rtems.h: Likewise.
17308 * config/bfin/rtems.h: Likewise.
17309 * config/i386/rtemself.h: Likewise.
17310 * config/lm32/rtems.h: Likewise.
17311 * config/m32c/rtems.h: Likewise.
17312 * config/m68k/rtemself.h: Likewise.
17313 * config/microblaze/rtems.h: Likewise.
17314 * config/mips/rtems.h: Likewise.
17315 * config/moxie/rtems.h: Likewise.
17316 * config/nios2/rtems.h: Likewise.
17317 * config/powerpcspe/rtems.h: Likewise.
17318 * config/rs6000/rtems.h: Likewise.
17319 * config/rtems.h: Likewise.
17320 * config/sh/rtems.h: Likewise.
17321 * config/sh/rtemself.h: Likewise.
17322 * config/sparc/rtemself.h: Likewise.
17324 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17327 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
17328 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
17329 * tree-ssa-structalias.c (alias_get_name): Same.
17331 2017-07-25 Bin Cheng <bin.cheng@arm.com>
17334 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
17335 instructions if no du chain is found.
17337 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17339 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
17341 2017-07-25 Richard Biener <rguenther@suse.de>
17343 PR middle-end/81505
17344 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
17347 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
17349 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
17350 upper-regs options.
17351 (ISA_2_7_MASKS_SERVER): Likewise.
17352 (ISA_3_0_MASKS_IEEE): Likewise.
17353 (OTHER_P8_VECTOR_MASKS): Likewise.
17354 (OTHER_VSX_VECTOR_MASKS): Likewise.
17355 (POWERPC_MASKS): Likewise.
17356 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
17357 duplicate list of options.
17358 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
17359 explicit -mupper-regs options.
17360 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
17361 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
17362 alias for -mupper-regs-df.
17363 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17364 (rs6000_init_hard_regno_mode_ok): Likewise.
17365 (rs6000_option_override_internal): Likewise.
17366 (rs6000_opt_masks): Likewise.
17367 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
17368 options in terms of whether -mvsx or -mpower8-vector was used.
17369 (TARGET_UPPER_REGS_DI): Likewise.
17370 (TARGET_UPPER_REGS_SF): Likewise.
17371 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
17372 -mupper-regs-* options.
17374 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17376 * passes.c (emergency_dump_function): Print some empty lines and a
17377 header before the RTL dump.
17379 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17381 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
17383 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
17386 * config/aarch64/aarch64.c (aarch64_classify_symbol):
17387 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
17389 2017-07-24 Carl Love <cel@us.ibm.com>
17391 * config/rs6000/rs6000-c.c: Add support for built-in functions
17392 vector float vec_extract_fp32_from_shorth (vector unsigned short);
17393 vector float vec_extract_fp32_from_shortl (vector unsigned short);
17394 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
17395 vec_extract_fp_from_shortl): Add defines for the two builtins.
17396 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
17397 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
17399 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
17400 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
17401 * doc/extend.texi: Update the built-in documentation file for the
17402 new built-in function.
17404 2017-07-24 Jakub Jelinek <jakub@redhat.com>
17407 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
17409 * doc/generic.texi: Likewise.
17410 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17411 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17413 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
17415 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
17416 (aarch64_mls_elt_merge<mode>): Likewise.
17418 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
17420 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17421 having __cxa_atexit.
17423 2017-07-23 Michael Collison <michael.collison@arm.com>
17425 * config/arm/arm.c (arm_option_override): Deprecate
17426 use of -mstructure-size-boundary.
17427 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
17428 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
17430 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17433 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17434 Reduce cost estimate for direct moves.
17436 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
17439 * config/i386/i386.c (ix86_option_override_internal): Disable
17440 BMI, BMI2 and TBM instructions for -m16.
17442 2017-07-21 Carl Love <cel@us.ibm.com>
17444 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17445 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17446 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17447 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17448 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17449 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17450 VMULOSW): New enum "unspec" values.
17451 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17452 altivec_vmulosw): New patterns.
17453 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17454 VMULOSW): Add definitions.
17456 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
17458 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
17459 (qdf24xx): Likewise.
17460 * config/aarch64/aarch64-options-extensions.def (rdma); New.
17461 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
17462 (AARCH64_FL_V8_1): Renumber.
17463 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
17464 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
17465 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
17466 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
17467 rdma to feature modifiers list.
17469 2017-07-21 Yury Gribov <tetra2005@gmail.com>
17471 PR middle-end/56727
17472 * ipa-visibility (function_and_variable_visibility): Convert
17473 recursive PLT call to direct call if appropriate.
17475 2017-07-21 Andrew Pinski <apinski@cavium.com>
17477 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
17478 operand 1 to see if the types precision matches.
17479 * fold-const.c (operand_equal_p): Likewise.
17481 2017-07-21 Richard Biener <rguenther@suse.de>
17483 PR tree-optimization/81303
17484 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
17485 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
17486 (vect_peeling_hash_get_lowest_cost): Adjust.
17487 (vect_enhance_data_refs_alignment): Likewise. Use
17488 vect_get_peeling_costs_all_drs to compute the penalty for no
17489 peeling to match up costs.
17491 2017-07-21 Richard Biener <rguenther@suse.de>
17493 PR tree-optimization/81500
17494 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
17495 we didn't identify a reduction path.
17497 2017-07-21 Tom de Vries <tom@codesourcery.com>
17498 Cesar Philippidis <cesar@codesourcery.com>
17500 PR gcov-profile/81442
17501 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
17504 2017-07-21 Tom de Vries <tom@codesourcery.com>
17507 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
17509 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
17510 nvptx_override_options_after_change.
17512 2017-07-21 Ulrich Drepper <drepper@redhat.com>
17514 * dwarf2out.c (output_file_names): Avoid double testing for
17515 dwarf_version >= 5.
17517 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
17519 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
17521 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17523 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
17525 (try_crossjump_to_edge): Do not punt on partitioned functions.
17527 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17529 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
17530 Put all BBs reachable only via paths crossing cold region to cold
17532 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
17534 2016-07-21 Richard Biener <rguenther@suse.de>
17536 PR tree-optimization/81303
17537 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
17538 into account prologue and epilogue iterations when raising
17539 min_profitable_iters to sth at least covering one vector iteration.
17541 2017-07-21 Tamar Christina <tamar.christina@arm.com>
17543 * config/arm/arm.c (arm_test_cpu_arch_dat):
17546 2017-07-20 Nathan Sidwell <nathan@acm.org>
17548 Remove TYPE_METHODS.
17549 * tree.h (TYPE_METHODS): Delete.
17550 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
17551 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
17552 (dbxout_type_methods): Scan TYPE_FIELDS.
17553 (dbxout_type): Don't check TYPE_METHODS here.
17554 * function.c (use_register_for_decl): Always ignore register for
17555 class types when not optimizing.
17556 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
17557 * tree.c (free_lang_data_in_type): Stitch out member functions and
17558 templates from TYPE_FIELDS.
17559 (build_distinct_type_copy, verify_type_variant,
17560 verify_type): Member fns are on TYPE_FIELDS.
17561 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
17562 * tree-pretty-print.c (dump_generic_node): Likewise.
17564 2017-07-20 Jakub Jelinek <jakub@redhat.com>
17567 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
17568 V2TImode and V4TImode.
17569 (ix86_expand_vector_extract): Likewise.
17570 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
17571 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
17572 (ssescalarmode): Handle V4TImode and V2TImode.
17573 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
17574 (*vec_extractv2ti, *vec_extractv4ti): New insns.
17575 (VEXTRACTI128_MODE): New mode iterator.
17576 (splitter for *vec_extractv?ti first element): New.
17577 (VEC_INIT_MODE): New mode iterator.
17578 (vec_init<mode>): Consolidate 3 expanders into one using
17579 VEC_INIT_MODE mode iterator.
17581 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17583 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
17584 non_spilled_static_chain_regno_p.
17586 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17588 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
17590 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
17592 * bb-reorder.c (connect_traces): Allow copying of blocks within
17595 2017-07-20 Richard Biener <rguenther@suse.de>
17597 * gimple.h (gimple_phi_result): Add gphi * overload.
17598 (gimple_phi_result_ptr): Likewise.
17599 (gimple_phi_arg): Likewise. Adjust index assert to only
17600 allow actual argument accesses rather than all slots available
17602 (gimple_phi_arg_def): Add gphi * overload.
17603 * tree-phinodes.c (make_phi_node): Initialize only actual
17605 (resize_phi_node): Clear memory not covered by old node,
17606 do not initialize excess argument slots.
17607 (reserve_phi_args_for_new_edge): Initialize new argument slot
17610 2017-07-20 Bin Cheng <bin.cheng@arm.com>
17612 PR tree-optimization/81388
17614 2016-07-21 Bin Cheng <bin.cheng@arm.com>
17616 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
17617 by removing computation of may_be_zero.
17619 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17620 Tom de Vries <tom@codesourcery.com>
17622 PR middle-end/81030
17623 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
17624 when gimple level profile disagrees with what RTL expander did.
17626 2017-07-20 Richard Biener <rguenther@suse.de>
17628 PR tree-optimization/61171
17629 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
17630 (vect_analyze_stmt): Add slp instance parameter.
17631 (vectorizable_reduction): Likewise.
17632 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
17633 (vect_is_simple_reduction): Deal with chains not detected
17634 as SLP reduction chain, specifically not properly associated
17635 chains containing a mix of plus/minus.
17636 (get_reduction_op): Remove.
17637 (get_initial_defs_for_reduction): Simplify, pass in whether
17638 this is a reduction chain, pass in the SLP node for the PHIs.
17639 (vect_create_epilog_for_reduction): Get the SLP instance as
17641 (vectorizable_reduction): Get the SLP instance as arg.
17642 During analysis remember the SLP node with the PHIs in the
17643 instance. Simplify getting at the vectorized reduction PHIs.
17644 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
17645 through SLP instance.
17646 (vect_slp_analyze_operations): Likewise.
17647 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
17648 (vect_transform_stmt): Likewise.
17650 2017-07-20 Tom de Vries <tom@codesourcery.com>
17652 PR tree-optimization/81489
17653 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17654 read of phi arg location to before loop that modifies phi.
17656 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17658 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17661 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17663 PR middle-end/81331
17664 * except.c (execute): Fix ordering issue.
17666 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17668 PR rtl-optimization/81423
17669 * combine.c (make_compound_operation_int): Don't try to optimize
17670 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17672 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17674 PR rtl-optimization/81423
17675 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17676 with a constant that is -1 in the truncated to mode.
17678 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17680 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17681 (determine_unlikely_bbs): ... here.
17682 * predict.h (propagate_unlikely_bbs_forward): Declare.
17683 * cfgexpand.c (pass_expand::execute): Use it.
17684 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17686 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17687 propagate_unlikely_bbs_forward.
17689 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17691 PR middle-end/81331
17692 * except.c (maybe_add_nop_after_section_switch): New function.
17695 2017-07-19 Tom de Vries <tom@codesourcery.com>
17697 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17699 2017-07-19 Tom de Vries <tom@codesourcery.com>
17701 * gimple.h (gimple_phi_arg): Make assert more strict.
17703 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17705 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17706 * config/rs6000/mmintrin.h: New file.
17707 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17709 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17711 PR tree-optimization/81346
17712 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17714 2017-07-19 Tom de Vries <tom@codesourcery.com>
17716 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17718 2017-07-19 Tom de Vries <tom@codesourcery.com>
17720 * config/nvptx/nvptx-modes.def: Add V2DImode.
17721 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17722 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17723 (nvptx_output_mov_insn): Handle lack of mov.b128.
17724 (nvptx_print_operand): Handle 'H' and 'L' codes.
17725 (nvptx_vector_mode_supported): Allow V2DImode.
17726 (nvptx_preferred_simd_mode): New function.
17727 (nvptx_data_alignment): New function.
17728 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17729 nvptx_preferred_simd_mode.
17730 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17732 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17734 2017-07-19 Tom de Vries <tom@codesourcery.com>
17736 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17737 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17738 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17739 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17740 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17741 (mov<VECIM>_insn): New define_insn.
17742 (define_expand "mov<VECIM>): New define_expand.
17744 2017-07-19 Tom de Vries <tom@codesourcery.com>
17746 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17748 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17750 PR tree-optimization/81346
17751 * fold-const.h (fold_div_compare, range_check_type): Declare.
17752 * fold-const.c (range_check_type): New function.
17753 (build_range_check): Use range_check_type.
17754 (fold_div_compare): No longer static, rewritten into
17755 a match.pd helper function.
17756 (fold_comparison): Don't call fold_div_compare here.
17757 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17758 as helper function.
17760 2017-07-19 Nathan Sidwell <nathan@acm.org>
17762 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17763 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17764 * tree.c (find_decls_types_r, verify_type): Use
17765 TYPE_{MIN,MAX}_VALUE_RAW.
17766 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17767 (hash_tree): Likewise.
17768 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17770 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17773 2017-07-18 Tom de Vries <tom@codesourcery.com>
17775 PR middle-end/81464
17776 * omp-expand.c (expand_omp_for_static_chunk): Handle
17777 equal-argument loop exit phi.
17779 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17782 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17783 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17784 operand 2 predicate.
17785 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17786 operand 2 predicate.
17787 (ror,rol -> rorx splitters): Use const_int_operand as
17788 operand 2 predicate.
17790 2017-06-18 Richard Biener <rguenther@suse.de>
17792 PR tree-optimization/81410
17793 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17794 the gap in the ! slp_perm SLP case after each group.
17796 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17798 PR middle-end/81463
17799 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17802 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17804 PR middle-end/81462
17805 * predict.c (set_even_probabilities): Cleanup; do not affect
17806 probabilities that are already known.
17807 (combine_predictions_for_bb): Call even when count is set.
17809 2017-07-18 Nathan Sidwell <nathan@acm.org>
17811 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17814 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17817 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17818 optimization for loop niter analysis.
17820 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17823 * config/avr/avr.c (avr_optimize_casesi): Don't use
17824 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17826 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17828 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17829 body_cost_vec from _vect_peel_extended_info.
17830 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17831 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17834 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17836 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17838 2017-07-18 Richard Biener <rguenther@suse.de>
17840 PR tree-optimization/80620
17841 PR tree-optimization/81403
17842 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17843 info when re-using a VN table entry.
17845 2017-07-18 Richard Biener <rguenther@suse.de>
17847 PR tree-optimization/81418
17848 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17849 vectype_in. Verify that with lane-reducing reduction operations
17850 we have a single def-use cycle.
17852 2017-07-17 Carl Love <cel@us.ibm.com>
17854 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17856 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17857 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17858 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17859 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17860 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17861 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17862 VMULOSW): New enum "unspec" values.
17863 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17864 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17865 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17866 altivec_vmulosw): New patterns.
17867 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17868 VMULOSW): Add definitions.
17869 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17871 * config/alpha/alpha.c: Include predict.h.
17873 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17875 * tree-vrp.c (compare_assert_loc): Fix comparison function
17876 to return predictable results.
17878 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17880 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17882 (subdi3): Likewise.
17883 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17884 * doc/invoke.texi (mexpand-adddi): Update text.
17886 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17888 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17889 that also clobbers the CC register. The old expand code is moved
17891 (*arc_clzsi2): ... here.
17892 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17893 the CC register. The old expand code is moved to ...
17894 (arc_ctzsi2): ... here.
17896 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17898 * config/arc/arc.opt (mindexed-loads): Use initial value
17899 TARGET_INDEXED_LOADS_DEFAULT.
17900 (mauto-modify-reg): Use initial value
17901 TARGET_AUTO_MODIFY_REG_DEFAULT.
17902 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17903 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17904 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17905 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17907 2017-07-17 Martin Liska <mliska@suse.cz>
17910 * opts.c (finish_options): Do not allow -fgnu-tm
17911 w/ -fsanitize={kernel-,}address. Say sorry.
17913 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17916 * tree-loop-distribution.c (classify_partition): Only assert on
17917 numer of iterations.
17918 (merge_dep_scc_partitions): Delete prameter. Update function call.
17919 (distribute_loop): Remove code handling loop with unknown niters.
17920 (pass_loop_distribution::execute): Skip loop with unknown niters.
17922 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17925 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17926 function sort_partitions_by_post_order.
17928 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17930 PR tree-optimization/81374
17931 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17932 the max index of basic blocks, rather than number of basic blocks.
17934 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17936 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17938 (arc_legitimate_pic_operand_p): Likewise.
17939 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17941 (arc_needs_pcl_p): Likewise.
17942 (arc_legitimate_pc_offset_p): Likewise.
17943 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17944 function is also used in constrains.md.
17945 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17946 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17947 PLUS. Only return true/false in known cases, otherwise assert.
17948 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17949 is already called in arc_legitimate_constant_p.
17950 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17952 (LEGITIMATE_PIC_OPERAND_P): Use
17953 arc_raw_symbolic_reference_mentioned_p function.
17954 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17959 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17960 Andrew Burgess <andrew.burgess@embecosm.com>
17962 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17963 (arc_return_address_register): New function.
17964 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17965 (arc_handle_fndecl_attribute): Add naked attribute.
17966 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17967 (TARGET_WARN_FUNC_RETURN): Likewise.
17968 (arc_allocate_stack_slots_for_args): New function.
17969 (arc_warn_func_return): Likewise.
17970 (machine_function): Change type fn_type.
17971 (arc_compute_function_type): Consider new naked function type,
17972 change function return type.
17973 (arc_must_save_register): Adapt to handle new
17974 arc_compute_function_type's return type.
17975 (arc_expand_prologue): Likewise.
17976 (arc_expand_epilogue): Likewise.
17977 (arc_return_address_regs): Delete.
17978 (arc_return_address_register): New function.
17979 (arc_epilogue_uses): Use above function.
17980 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17981 (arc_function_type): Change encoding, add naked type.
17982 (ARC_INTERRUPT_P): Change to handle the new encoding.
17983 (ARC_FAST_INTERRUPT_P): Likewise.
17984 (ARC_NORMAL_P): Define.
17985 (ARC_NAKED_P): Likewise.
17986 (arc_compute_function_type): Delete prototype.
17987 * config/arc/arc.md (in_ret_delay_slot): Use
17988 arc_return_address_register function.
17989 (simple_return): Likewise.
17990 (p_return_i): Likewise.
17992 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17994 PR tree-optimization/81428
17995 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17996 can't be built for those types.
17998 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
18000 Remove stuff dead since r239246.
18002 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
18003 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
18004 (avr_inform_devices): Remove dead stuff.
18006 2017-07-17 Tamar Christina <tamar.christina@arm.com>
18008 * config/arm/arm_neon.h: Fix softp typo.
18010 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18012 PR tree-optimization/81365
18013 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
18014 aggregate moves onto bb predecessor edges, make sure there are no
18015 loads that could alias the lhs in between the start of bb and the
18018 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
18021 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
18022 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
18023 [LSHIFTRT, outer_code = TRUNCATE]: Same.
18025 2017-07-17 Jakub Jelinek <jakub@redhat.com>
18027 PR tree-optimization/81396
18028 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
18029 (init_symbolic_number): Initialize it to 1.
18030 (perform_symbolic_merge): Add n_ops from both operands into the new
18032 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
18033 without base_addr as useless if they need more than one operation.
18034 (bswap_replace): Handle !bswap case for NULL base_addr.
18036 2017-07-17 Tom de Vries <tom@codesourcery.com>
18039 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
18042 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
18044 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
18045 conditional builtin define __FIX_LEON3FT_B2BST.
18047 2017-07-17 Daniel Cederman <cederman@gaisler.com>
18049 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
18050 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
18053 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
18055 PR rtl-optimization/81424
18056 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
18057 to remove potential trapping from operands if -fnon-call-exceptions.
18059 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18061 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
18062 profile_proability for scalling.
18063 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
18065 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18067 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
18069 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18071 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
18072 fixpoint arithmetics.
18074 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18076 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
18077 fixpoint arithmetics.
18079 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18081 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
18082 fixpoint arithmetics.
18084 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
18086 * profile-count.h (profile_probability::from_reg_br_prob_note,
18087 profile_probability::to_reg_br_prob_note): New functions.
18088 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
18089 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
18090 * predict.c (probability_reliable_p): Update.
18091 (edge_probability_reliable_p): Update.
18092 (br_prob_note_reliable_p): Update.
18093 (invert_br_probabilities): Update.
18094 (add_reg_br_prob_note): New function.
18095 (combine_predictions_for_insn): Update.
18096 * asan.c (asan_clear_shadow): Update.
18097 * cfgbuild.c (compute_outgoing_frequencies): Update.
18098 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18099 (update_br_prob_note): Update.
18100 (rtl_verify_edges): Update.
18101 (purge_dead_edges): Update.
18102 (fixup_reorder_chain): Update.
18103 * emit-rtl.c (try_split): Update.
18104 * ifcvt.c (cond_exec_process_insns): Update.
18105 (cond_exec_process_if_block): Update.
18106 (dead_or_predicable): Update.
18107 * internal-fn.c (expand_addsub_overflow): Update.
18108 (expand_neg_overflow): Update.
18109 (expand_mul_overflow): Update.
18110 * loop-doloop.c (doloop_modify): Update.
18111 * loop-unroll.c (compare_and_jump_seq): Update.
18112 * optabs.c (emit_cmp_and_jump_insn_1): Update.
18113 * predict.h: Update.
18114 * reorg.c (mostly_true_jump): Update.
18116 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
18117 * config/alpha/alpha.c (emit_unlikely_jump): Update.
18118 * config/arc/arc.c: (emit_unlikely_jump): Update.
18119 * config/arm/arm.c: (emit_unlikely_jump): Update.
18120 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
18121 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
18122 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
18123 (ix86_print_operand): Update.
18124 (ix86_split_fp_branch): Update.
18125 (predict_jump): Update.
18126 * config/ia64/ia64.c (ia64_print_operand): Update.
18127 * config/mmix/mmix.c (mmix_print_operand): Update.
18128 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
18129 (rs6000_expand_split_stack_prologue): Update.
18130 * config/rs6000/rs6000.c: Update.
18131 * config/s390/s390.c (s390_expand_vec_strlen): Update.
18132 (s390_expand_vec_movstr): Update.
18133 (s390_expand_cs_tdsi): Update.
18134 (s390_expand_split_stack_prologue): Update.
18135 * config/sh/sh.c (sh_print_operand): Update.
18136 (expand_cbranchsi4): Update.
18137 (expand_cbranchdi4): Update.
18138 * config/sparc/sparc.c (output_v9branch): Update.
18139 * config/spu/spu.c (get_branch_target): Update.
18140 (ea_load_store_inline): Update.
18141 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
18142 * config/tilepro/tilepro.c: Update.
18144 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
18146 * gimplify.c (mostly_copy_tree_r): Revert latest change.
18147 (gimplify_save_expr): Likewise.
18149 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18151 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
18153 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18155 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
18157 * timevar.def (TV_IPA_FNSUMMARY): Define.
18159 2017-07-16 Daniel Cederman <cederman@gaisler.com>
18161 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
18162 to back store errata sensitive sequence from being generated.
18163 (sqrtdf2_fix): Likewise.
18165 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18167 * tree-ssa-threadupdate.c (compute_path_counts,
18168 update_joiner_offpath_counts): Use profile_probability.
18170 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
18173 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18175 * config/arm/arm-c.c (arm_cpu_builtins): Define
18176 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
18178 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
18180 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18181 array entries to represent __ieee128 versions of the
18182 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
18183 scalar_extract_sig, and scalar_insert_exp built-in functions.
18184 (altivec_resolve_overloaded_builtin): Add special case handling
18185 for the __builtin_scalar_insert_exp function, as represented by
18186 the P9V_BUILTIN_VEC_VSIEDP constant.
18187 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
18188 exponent support for __ieee128 argument.
18189 (VSESQP): Add scalar extract signature support for __ieee128
18191 (VSTDCNQP): Add scalar test negative support for __ieee128
18193 (VSIEQP): Add scalar insert exponent support for __int128 argument
18194 with __ieee128 result.
18195 (VSIEQPF): Add scalar insert exponent support for __ieee128
18196 argument with __ieee128 result.
18197 (VSTDCQP): Add scalar test data class support for __ieee128
18199 (VSTDCNQP): Add overload support for scalar test negative with
18200 __ieee128 argument.
18201 (VSTDCQP): Add overload support for scalar test data class
18202 __ieee128 argument.
18203 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
18204 UNSPEC_VSX_SXSIGDP.
18205 (UNSPEC_VSX_SIEXPQP): New constant.
18206 (xsxexpqp): New insn for VSX scalar extract exponent quad
18208 (xsxsigqp): New insn for VSX scalar extract significand quad
18210 (xsiexpqpf): New insn for VSX scalar insert exponent quad
18211 precision with floating point argument.
18212 (xststdcqp): New expand for VSX scalar test data class quad
18214 (xststdcnegqp): New expand for VSX scalar test negative quad
18216 (xststdcqp): New insn to match expansions for VSX scalar test data
18217 class quad precision and VSX scalar test negative quad precision.
18218 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
18219 special case operand checking to enforce that second operand of
18220 VSX scalar test data class with quad precision argument is a 7-bit
18222 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18223 prototypes and descriptions of __ieee128 versions of
18224 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
18225 scalar_test_data_class, and scalar_test_neg built-in functions.
18227 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18229 PR tree-optimization/81162
18230 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
18231 replace a negate with an add.
18233 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
18235 * doc/invoke.texi (arm/-mcpu): Document +crypto.
18237 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18239 * config/arm/arm-c.c (arm_cpu_builtins): Define
18240 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
18242 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18244 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
18245 (armv8-r): Set ARM Cortex-R52 as default CPU.
18246 * config/arm/arm-tables.opt: Regenerate.
18247 * config/arm/arm-tune.md: Regenerate.
18248 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18250 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
18251 extension for -mcpu=cortex-r52.
18253 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18255 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
18256 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
18257 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
18258 (fp-armv8): Define it as FP_ARMv8 only.
18259 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
18260 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
18262 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
18263 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
18264 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
18265 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
18266 than TARGET_FPU_ARMV8.
18267 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
18268 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
18269 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
18271 * config/arm/neon.md (neon_vrint): Likewise.
18272 (neon_vcvt): Likewise.
18273 (neon_<fmaxmin_op><mode>): Likewise.
18274 (<fmaxmin><mode>3): Likewise.
18275 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
18276 * config/arm/predicates.md (arm_cond_move_operator): Check against
18277 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
18279 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
18281 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
18282 to top of function.
18284 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18286 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
18287 loop in comment with memset.
18289 2017-07-14 Martin Liska <mliska@suse.cz>
18291 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
18292 * dwarf2out.c (is_java): Remove the function.
18293 (output_pubname): Remove usage of the function.
18294 (lower_bound_default): Remove usage of DW_LANG_Java.
18295 (gen_compile_unit_die): Likewise.
18296 * gcc.c: Remove compiler defaults for .java and .zip files.
18297 * gimple-expr.c (remove_suffix): Change as there's no longer
18298 extension than 4-letter one.
18299 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
18300 (gimplify_save_expr): Likewise.
18301 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
18302 as it's possible even for other languages than Java.
18303 * langhooks.h (struct lang_hooks): Remove Java from a comment.
18304 * lto-opts.c (lto_write_options): Remove reference to Java.
18305 * opts.c (strip_off_ending): Update file extension handling.
18306 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
18307 * tree-eh.c (lower_resx): Likewise.
18308 * tree.c (free_lang_data_in_type): Remove dead code.
18309 (find_decls_types_r): Likewise.
18310 (build_common_builtin_nodes): Remove Java from a comment.
18311 (verify_type): Remove dead code.
18312 * varasm.c (assemble_external): Remove Java from a comment.
18314 2017-07-14 Martin Liska <mliska@suse.cz>
18316 * opts.c (finish_options): Add quotes.
18317 (common_handle_option): Likewise.
18319 2017-07-14 Martin Liska <mliska@suse.cz>
18321 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
18322 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
18323 Remove N_SO_PASCAL.
18324 * dwarf2out.c (lower_bound_default): Do not handle
18326 (gen_compile_unit_die): Likewise.
18327 * gcc.c: Remove default extension binding for GNU Pascal.
18328 * stmt.c: Remove Pascal language from a comment.
18329 * xcoffout.c: Likewise.
18331 2017-07-13 David Malcolm <dmalcolm@redhat.com>
18334 * diagnostic-show-locus.c (fixit_cmp): New function.
18335 (layout::layout): Sort m_fixit_hints.
18336 (column_range::column_range): Assert that the values are valid.
18337 (struct char_span): New struct.
18338 (correction::overwrite): New method.
18339 (struct source_line): New struct.
18340 (line_corrections::add_hint): Add assertions. Reimplement memcpy
18341 calls in terms of classes source_line and char_span, and
18342 correction::overwrite.
18343 (selftest::test_overlapped_fixit_printing_2): New function.
18344 (selftest::diagnostic_show_locus_c_tests): Call it.
18346 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
18348 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
18349 early if there is no lhs.
18351 2017-07-13 Martin Liska <mliska@suse.cz>
18353 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
18354 (gen_reference_type_die): Likewise.
18355 * stor-layout.c: Remove Pascal-related comment.
18357 2017-07-13 Martin Liska <mliska@suse.cz>
18359 * opts.c (finish_options): Add quotes to error messages.
18360 (parse_sanitizer_options): Likewise.
18362 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
18364 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
18366 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
18368 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
18370 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18372 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
18374 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
18376 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18379 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
18380 provides the hardware capability bits, define the macro
18381 __BUILTIN_CPU_SUPPORTS__.
18382 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
18383 if GLIBC does not provide the hardware capability bits. Add a
18384 gcc_unreachable call if the built-in cpu function is neither
18385 __builtin_cpu_is nor __builtin_cpu_supports.
18386 (rs6000_get_function_versions_dispatcher): Change the warning
18387 that an old GLIBC is used which does not export the capability
18388 bits to be an error.
18389 * doc/extend.texi (target_clones attribute): Document the
18390 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
18391 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
18392 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
18393 the macros defined by GCC if the newer GLIBC is available.
18395 2017-07-12 Jeff Law <law@redhat.com>
18397 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
18398 remaining includes slightly.
18399 * config/riscv/riscv-builtins.c: Include profile-count.h.
18401 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18404 * config/avr/avr.c (avr_set_current_function): In diagnostic
18405 messages: Quote keywords and (parts of) identifiers.
18406 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
18409 2017-07-12 Carl Love <cel@us.ibm.com>
18411 * config/rs6000/rs6000-c.c: Add support for built-in functions
18412 vector bool char vec_revb (vector bool char);
18413 vector bool short vec_revb (vector short char);
18414 vector bool int vec_revb (vector bool int);
18415 vector bool long long vec_revb (vector bool long long);
18416 * doc/extend.texi: Update the built-in documentation file for the
18417 new built-in functions.
18419 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18421 * config/s390/s390.md: Remove movcc splitter.
18423 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18425 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
18426 load/store on condition.
18428 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18431 * config/avr/avr.c (avr_encode_section_info)
18432 [progmem && !TREE_READONLY]: Error if progmem object needs
18435 2017-07-11 Michael Collison <michael.collison@arm.com>
18437 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
18440 2017-07-11 Carl Love <cel@us.ibm.com>
18442 * config/rs6000/rs6000-c.c: Add support for builtins
18443 vector unsigned int vec_parity_lsbb (vector signed int);
18444 vector unsigned int vec_parity_lsbb (vector unsigned int);
18445 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
18446 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
18447 vector unsigned long long vec_parity_lsbb (vector signed long long);
18448 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
18449 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
18450 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
18451 * doc/extend.texi: Update the built-in documentation file for the
18452 new built-in functions.
18454 2017-07-11 David Malcolm <dmalcolm@redhat.com>
18456 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
18457 (layout::m_primary_loc): New field.
18458 (layout::layout): Initialize new field. Move location filtering
18459 logic from here to...
18460 (layout::maybe_add_location_range): ...this new method. Add
18461 support for filtering to just the lines already specified by other
18463 (layout::will_show_line_p): New method.
18464 (gcc_rich_location::add_location_if_nearby): New method.
18465 (selftest::test_add_location_if_nearby): New test function.
18466 (selftest::diagnostic_show_locus_c_tests): Call it.
18467 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
18470 2017-07-11 Tom de Vries <tom@codesourcery.com>
18472 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
18473 (bb_first_real_insn): New function.
18474 (nvptx_single): Add extra initialization of broadcasted condition
18477 2017-07-11 Nathan Sidwell <nathan@acm.org>
18479 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
18481 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
18483 * doc/extend.texi (AVR Function Attributes): Remove weblink to
18484 Binutils doc as TEXI will mess them up.
18485 * doc/invoke.texi (AVR Options): Same here.
18487 2017-07-11 Daniel Cederman <cederman@gaisler.com>
18489 * config/sparc/sparc.opt (mfix-ut700): New option.
18490 (mfix-gr712rc): Likewise.
18491 (sparc_fix_b2bst): New variable.
18492 * doc/invoke.texi (SPARC options): Document them.
18493 (ARM options): Fix warnings.
18494 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
18495 instructions to prevent sequences that can trigger the store-store
18496 errata for certain LEON3FT processors.
18497 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
18498 (sparc_option_override): Set sparc_fix_b2bst appropriately.
18499 * config/sparc/sparc.md (fix_b2bst): New attribute.
18500 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
18502 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
18505 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
18507 (*rsqrtsf2_sse): Ditto.
18509 (div<mode>3): Macroize insn from divdf3 and divsf3
18510 using MODEF mode iterator.
18512 2017-07-10 Martin Sebor <msebor@redhat.com>
18514 PR tree-optimization/80397
18515 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
18516 instead of testing for equality to INTEGER_TYPE.
18518 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
18520 * config.gcc: Remove uclibc from arc target spec.
18522 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
18524 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
18526 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18529 * lto-wrapper.c (remove_option): New function.
18530 (merge_and_complain): Merge PIC/PIE options more realistically.
18532 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18534 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
18538 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
18539 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
18540 * config.in: Regenerate.
18541 * configure: Regenerate.
18542 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
18543 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
18544 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
18545 (TARGET_GASISR_PROLOGUES): ...target mask.
18546 * common/config/avr/avr-common.c
18547 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
18548 Set -mgas-isr-prologues.
18549 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
18550 INSERT_PASS_BEFORE for it.
18551 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
18552 * config/avr/avr.c (avr_option_override)
18553 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
18554 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
18555 (avr_attribute_table) <no_gccisr>: Add new function attribute.
18556 (avr_set_current_function) <is_no_gccisr>: Init machine field.
18557 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
18559 (make_avr_pass_pre_proep): New function.
18560 (emit_push_sfr) <treg>: Add argument to function and use it
18561 instead of TMP_REG.
18562 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
18563 and set machine->gasisr.yes.
18564 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
18565 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
18566 __gcc_isr.n_pushed to .L__stack_usage.
18567 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
18568 (avr_asm_final_postscan_insn): ...this new static function.
18569 * config/avr/avr.h (machine_function)
18570 <is_no_gccisr, use_L__stack_usage>: New fields.
18571 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
18572 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
18573 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
18574 (gasisr, *gasisr): New expander and insn.
18575 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18576 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
18577 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
18579 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
18581 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
18584 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18586 Move jump-tables out of .text again.
18589 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
18590 (ASM_OUTPUT_ADDR_VEC): New function.
18591 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
18592 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
18593 INSN_ADDRESSes as asm comment.
18594 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
18595 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
18596 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
18597 * config/avr/avr.md (*tablejump): Adjust comment.
18598 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
18599 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
18601 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
18602 (avr_output_addr_vec): New proto.
18603 (avr_log_t) <insn_addresses>: New field.
18605 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
18608 * config/i386/i386.c (ix86_function_arg_advance): Set
18609 outgoing_args_on_stack to true if there are outgoing arguments
18611 (ix86_function_arg): Likewise.
18612 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
18613 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
18614 * config/i386/i386.h (machine_function): Add
18615 outgoing_args_on_stack.
18617 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
18619 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18620 supporting pthreds.
18621 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
18623 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
18625 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
18626 (REAL_H): Remove $(MACHMODE_H).
18627 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
18629 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
18630 $(MACHMODE_H) and double-int.h.
18631 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
18633 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
18636 2017-07-07 Andrew Pinski <apinski@cavium.com>
18638 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
18639 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
18641 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18643 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
18644 Add warning if GCC was not configured to link against a GLIBC that
18645 exports the hardware capability bits.
18646 (make_resolver_func): Make resolver function private and not a
18647 COMDAT function. Create the name with clone_function_name instead
18648 of make_unique_name.
18651 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18652 correct operand in doing the split.
18654 2017-07-07 Carl Love <cel@us.ibm.com>
18656 * config/rs6000/rs6000-c: Add support for built-in function
18657 vector unsigned short vec_pack_to_short_fp32 (vector float,
18659 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18660 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18661 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18662 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18663 (convert_4f32_8i16): Add define_expand.
18664 * doc/extend.texi: Update the built-in documentation file for the
18665 new built-in function.
18667 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18669 * config/sparc/m8.md: New file.
18670 * config/sparc/sparc.md: Include m8.md.
18672 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18674 * config/sparc/sparc.opt: New option -mvis4b.
18675 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18676 (sparc_option_override): Handle VIS4B.
18677 (enum sparc_builtins): Define
18678 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18679 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18680 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18681 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18682 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18683 (check_constant_argument): New function.
18684 (sparc_vis_init_builtins): Define builtins
18685 __builtin_vis_dictunpack{8,16,32},
18686 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18687 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18688 __builtin_vis_fpcmpde{8,16,32}shl and
18689 __builtin_vis_fpcmpur{8,16,32}shl.
18690 (sparc_expand_builtin): Check that the constant operands to
18691 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18692 constant and in range.
18693 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18695 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18696 (SPARC_IMM5_P): Likewise.
18697 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18698 (enabled): Handle vis4b.
18699 (UNSPEC_DICTUNPACK): New unspec.
18700 (UNSPEC_FPCMPSHL): Likewise.
18701 (UNSPEC_FPUCMPSHL): Likewise.
18702 (UNSPEC_FPCMPDESHL): Likewise.
18703 (UNSPEC_FPCMPURSHL): Likewise.
18704 (cpu_feature): New CPU feature `vis4b'.
18705 (dictunpack{8,16,32}): New insns.
18706 (FPCSMODE): New mode iterator.
18707 (fpcscond): New code iterator.
18708 (fpcsucond): Likewise.
18709 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18710 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18711 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18712 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18713 * config/sparc/constraints.md: Define constraints `q' for unsigned
18714 2-bit integer constants and `t' for unsigned 5-bit integer
18716 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18718 (imm5_operand_dictunpack16): Likewise.
18719 (imm5_operand_dictunpack32): Likewise.
18720 (imm2_operand): Likewise.
18721 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18722 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18723 ditunpack* and fpcmp*shl builtins.
18725 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18727 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18728 * config.in: Add HAVE_AS_SPARC6 define.
18729 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18731 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18733 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18734 (CPP_CPU_SPEC): Handle m8.
18735 (ASM_CPU_SPEC): Likewise.
18736 * config/sparc/sparc-opts.h (enum processor_type): Add
18738 * config/sparc/sparc.c (m8_costs): New struct.
18739 (sparc_option_override): Handle TARGET_CPU_m8.
18740 (sparc32_initialize_trampoline): Likewise.
18741 (sparc64_initialize_trampoline): Likewise.
18742 (sparc_issue_rate): Likewise.
18743 (sparc_register_move_cost): Likewise.
18744 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18745 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18746 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18747 (CPP_CPU_SPEC): Handle M8.
18748 (ASM_CPU_SPEC): Likewise.
18749 (AS_M8_FLAG): Define.
18750 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18751 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18752 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18754 * configure: Regenerate.
18755 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18758 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18760 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18762 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18763 ("*movdi_insn_sp32"): Do not set v3pipe.
18764 ("*movsi_insn"): Likewise.
18765 ("*movdi_insn_sp64"): Likewise.
18766 ("*movsf_insn"): Likewise.
18767 ("*movdf_insn_sp32"): Likewise.
18768 ("*movdf_insn_sp64"): Likewise.
18769 ("*zero_extendsidi2_insn_sp64"): Likewise.
18770 ("*sign_extendsidi2_insn"): Likewise.
18771 ("*mov<VM32:mode>_insn"): Likewise.
18772 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18773 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18774 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18775 ("<vlop:code><VL:mode>3"): Likewise.
18776 ("*not_<vlop:code><VL:mode>3"): Likewise.
18777 ("*nand<VL:mode>_vis"): Likewise.
18778 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18779 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18780 ("one_cmpl<VL:mode>2"): Likewise.
18781 ("faligndata<VM64:mode>_vis"): Likewise.
18782 ("alignaddrsi_vis"): Likewise.
18783 ("alignaddrdi_vis"): Likweise.
18784 ("alignaddrlsi_vis"): Likewise.
18785 ("alignaddrldi_vis"): Likewise.
18786 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18787 ("bmaskdi_vis"): Likewise.
18788 ("bmasksi_vis"): Likewise.
18789 ("bshuffle<VM64:mode>_vis"): Likewise.
18790 ("cmask8<P:mode>_vis"): Likewise.
18791 ("cmask16<P:mode>_vis"): Likewise.
18792 ("cmask32<P:mode>_vis"): Likewise.
18793 ("pdistn<P:mode>_vis"): Likewise.
18794 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18796 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18798 * config/sparc/sparc.md ("subtype"): New insn attribute.
18799 ("*wrgsr_sp64"): Set insn subtype.
18800 ("*rdgsr_sp64"): Likewise.
18801 ("alignaddrsi_vis"): Likewise.
18802 ("alignaddrdi_vis"): Likewise.
18803 ("alignaddrlsi_vis"): Likewise.
18804 ("alignaddrldi_vis"): Likewise.
18805 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18806 ("fexpand_vis"): Likewise.
18807 ("fpmerge_vis"): Likewise.
18808 ("faligndata<VM64:mode>_vis"): Likewise.
18809 ("bshuffle<VM64:mode>_vis"): Likewise.
18810 ("cmask8<P:mode>_vis"): Likewise.
18811 ("cmask16<P:mode>_vis"): Likewise.
18812 ("cmask32<P:mode>_vis"): Likewise.
18813 ("fchksm16_vis"): Likewise.
18814 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18815 ("fmean16_vis"): Likewise.
18816 ("fp<plusminus_insn>64_vis"): Likewise.
18817 ("<plusminus_insn>v8qi3"): Likewise.
18818 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18819 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18820 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18821 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18822 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18823 ("*movqi_insn"): Likewise.
18824 ("*movhi_insn"): Likewise.
18825 ("*movsi_insn"): Likewise.
18826 ("movsi_pic_gotdata_op"): Likewise.
18827 ("*movdi_insn_sp32"): Likewise.
18828 ("*movdi_insn_sp64"): Likewise.
18829 ("movdi_pic_gotdata_op"): Likewise.
18830 ("*movsf_insn"): Likewise.
18831 ("*movdf_insn_sp32"): Likewise.
18832 ("*movdf_insn_sp64"): Likewise.
18833 ("*zero_extendhisi2_insn"): Likewise.
18834 ("*zero_extendqihi2_insn"): Likewise.
18835 ("*zero_extendqisi2_insn"): Likewise.
18836 ("*zero_extendqidi2_insn"): Likewise.
18837 ("*zero_extendhidi2_insn"): Likewise.
18838 ("*zero_extendsidi2_insn_sp64"): Likewise.
18839 ("ldfsr"): Likewise.
18840 ("prefetch_64"): Likewise.
18841 ("prefetch_32"): Likewise.
18842 ("tie_ld32"): Likewise.
18843 ("tie_ld64"): Likewise.
18844 ("*tldo_ldub_sp32"): Likewise.
18845 ("*tldo_ldub1_sp32"): Likewise.
18846 ("*tldo_ldub2_sp32"): Likewise.
18847 ("*tldo_ldub_sp64"): Likewise.
18848 ("*tldo_ldub1_sp64"): Likewise.
18849 ("*tldo_ldub2_sp64"): Likewise.
18850 ("*tldo_ldub3_sp64"): Likewise.
18851 ("*tldo_lduh_sp32"): Likewise.
18852 ("*tldo_lduh1_sp32"): Likewise.
18853 ("*tldo_lduh_sp64"): Likewise.
18854 ("*tldo_lduh1_sp64"): Likewise.
18855 ("*tldo_lduh2_sp64"): Likewise.
18856 ("*tldo_lduw_sp32"): Likewise.
18857 ("*tldo_lduw_sp64"): Likewise.
18858 ("*tldo_lduw1_sp64"): Likewise.
18859 ("*tldo_ldx_sp64"): Likewise.
18860 ("*mov<VM32:mode>_insn"): Likewise.
18861 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18862 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18864 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18866 * config/sparc/sparc.md ("type"): New insn type viscmp.
18867 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18869 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18870 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18871 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18872 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18874 ("n7_vis_logical_11cycle"): Likewise.
18875 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18876 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18877 * config/sparc/niagara.md ("niag_vis"): Likewise.
18878 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18879 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18881 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18883 * config/sparc/sparc.md: New instruction type `bmask'.
18884 (bmaskdi_vis): Use the `bmask' type.
18885 (bmasksi_vis): Likewise.
18886 * config/sparc/ultra3.md (us3_array): Likewise.
18887 * config/sparc/niagara7.md (n7_array): Likewise.
18888 * config/sparc/niagara4.md (n4_array): Likewise.
18889 * config/sparc/niagara2.md (niag2_vis): Likewise.
18890 (niag3_vis): Likewise.
18891 * config/sparc/niagara.md (niag_vis): Likewise.
18893 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18895 * ipa-comdats.c: Remove optimize check from gate.
18896 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18897 for functions not optimized.
18898 (ipa_fn_summary_read): Skip optimize check.
18899 (ipa_fn_summary_write): Likewise.
18900 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18902 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18904 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18905 (check_callers): Give up on caller which is not optimized.
18906 (inline_small_functions): Likewise.
18907 (ipa_inline): Do not give up when not optimizing.
18908 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18909 away unoptimizes cdtors.
18910 (whole_program_function_and_variable_visibility): Do
18911 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18912 * ipa.c (process_references): Do not check optimize.
18913 (symbol_table::remove_unreachable_nodes): Update optimize check.
18914 (set_writeonly_bit): Update optimize check.
18915 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18916 (pass_ipa_single_use::gate): Remove.
18918 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18920 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18921 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18922 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18923 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18924 permute_load, permute_store, adjust_extract, adjust_splat,
18925 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18926 replace_swap_with_copy, dump_swap_insn_table,
18927 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18928 recombine_lvx_pattern, recombine_stvx_pattern,
18929 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18930 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18931 to file rs6000-p8swap.c.
18932 * config/rs6000/rs6000-p8swap.c: New file.
18933 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18934 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18935 and rs6000*-*-* targets.
18937 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18939 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18941 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18943 * lto-wrapper.c (merge_and_complain): Do not merge
18944 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18945 fsigned_zeros, ftrapping_math, fwrapv.
18946 (append_compiler_options): Do not track these options.
18947 (append_linker_options): Likewie
18949 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18951 * cgraphunit.c (cgraph_node::finalize_function): When
18952 !flag_toplevel_reorde set no_reorder flag.
18953 (varpool_node::finalize_decl): Likewise.
18954 (symbol_table::compile): Drop no toplevel reorder path.
18956 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18958 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18959 edges; zero probability is not better than uninitialized.
18961 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18963 * asan.h (asan_sanitize_allocas_p): Declare.
18964 * asan.c (asan_sanitize_allocas_p): New function.
18965 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18966 (handle_builtin_alloca): Likewise.
18967 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18968 if !asan_sanitize_allocas_p.
18969 * params.def (asan-instrument-allocas): Add new option.
18970 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18971 * opts.c (common_handle_option): Disable allocas sanitization for
18974 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18976 * asan.c: Include gimple-fold.h.
18977 (get_last_alloca_addr): New function.
18978 (handle_builtin_stackrestore): Likewise.
18979 (handle_builtin_alloca): Likewise.
18980 (asan_emit_allocas_unpoison): Likewise.
18981 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18982 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18983 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18984 (instrument_builtin_call): Pass gimple iterator to
18985 get_mem_refs_of_builtin_call.
18986 (last_alloca_addr): New global.
18987 * asan.h (asan_emit_allocas_unpoison): Declare.
18988 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18989 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18990 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18991 if function calls alloca.
18992 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18993 * gimple-fold.h (replace_call_with_value): Declare.
18994 * internal-fn.c: Include asan.h.
18995 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18996 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18998 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19000 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
19001 (C_SELFTEST_FLAGS): New.
19002 (CPP_SELFTEST_FLAGS): New.
19003 (SELFTEST_DEPS): New, from deps of s-selftest.
19004 (C_SELFTEST_DEPS): New, from deps of s-selftest.
19005 (CPP_SELFTEST_DEPS): New.
19006 (selftest): Add dependency on s-selftest-c++.
19007 (s-selftest): Rename to...
19008 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
19009 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
19010 than SELFTEST_FLAGS.
19011 (selftest-gdb): Rename to...
19012 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
19014 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
19015 (selftest-valgrind): Rename to...
19016 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
19018 (selftest-valgrind): Reintroduce as an alias for
19019 selftest-c-valgrind.
19020 (s-selftest-c++): New.
19021 (selftest-c++-gdb): New.
19022 (selftest-c++-valgrind): New.
19024 2017-07-06 Olivier Hainque <hainque@adacore.com>
19026 * gcc.c (process_command): When deciding if undefined variables
19027 should be ignored when processing specs, accept "gcc -v" as well.
19029 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
19031 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
19032 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
19034 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
19036 * config/arm/arm-cpus.in (armv8-r): Add new entry.
19037 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
19038 * config/arm/arm-tables.opt: Regenerate.
19039 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
19041 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
19043 2017-07-06 Carl Love <cel@us.ibm.com>
19045 * ChangeLog: Clean up from mid air collision
19047 2017-07-06 Carl Love <cel@us.ibm.com>
19049 * config/rs6000/rs6000-c.c: Add support for built-in functions
19050 vector signed int vec_subc (vector signed int, vector signed int);
19051 vector signed __int128 vec_subc (vector signed __int128,
19052 vector signed __int128);
19053 vector unsigned __int128 vec_subc (vector unsigned __int128,
19054 vector unsigned __int128);
19055 vector signed int vec_sube (vector signed int, vector signed int,
19056 vector signed int);
19057 vector unsigned int vec_sube (vector unsigned int,
19058 vector unsigned int,
19059 vector unsigned int);
19060 vector signed __int128 vec_sube (vector signed __int128,
19061 vector signed __int128,
19062 vector signed__int128);
19063 vector unsigned __int128 vec_sube (vector unsigned __int128,
19064 vector unsigned __int128,
19065 vector unsigned __int128);
19066 vector signed int vec_subec (vector signed int, vector signed int,
19067 vector signed int);
19068 vector unsigned int vec_subec (vector unsigned int,
19069 vector unsigned int,
19070 vector unsigned int);
19071 vector signed __int128 vec_subec (vector signed __int128,
19072 vector signed __int128,
19073 vector signed__int128);
19074 vector unsigned __int128 vec_subec (vector unsigned __int128,
19075 vector unsigned __int128,
19076 vector unsigned __int128);
19077 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
19078 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
19079 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
19080 BU_ALTIVEC_OVERLOAD_X definitions.
19081 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
19082 * doc/extend.texi: Update the built-in documentation file for the new
19083 built-in functions.
19085 2017-07-06 David Malcolm <dmalcolm@redhat.com>
19088 * diagnostic-show-locus.c (layout::layout): Use start and finish
19089 spelling location for the start and finish of each range.
19090 * genmatch.c (linemap_client_expand_location_to_spelling_point):
19091 Add unused aspect param.
19092 * input.c (expand_location_1): Add "aspect" param, and use it
19093 to access the correct part of the location.
19094 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
19096 (expand_location_to_spelling_point): Likewise.
19097 (linemap_client_expand_location_to_spelling_point): Add "aspect"
19098 param, and pass it to expand_location_1.
19100 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
19102 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
19103 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
19104 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
19105 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
19106 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
19107 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
19108 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
19109 _mm_maskz_getmant_ss): New intrinsics.
19110 (__builtin_ia32_getexpss128_mask): Changed to ...
19111 __builtin_ia32_getexpss128_round ... this.
19112 (__builtin_ia32_getexpsd128_mask): Changed to ...
19113 __builtin_ia32_getexpsd128_round ... this.
19114 * config/i386/i386-builtin-types.def
19115 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
19116 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
19117 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
19118 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
19119 __builtin_ia32_getmantss_mask_round): New builtins.
19120 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
19121 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
19122 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
19123 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
19124 * config/i386/sse.md
19125 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
19126 avx512f_sgetexp<mode><mask_scalar_name>
19127 <round_saeonly_scalar_name> ... this.
19128 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
19129 %0, %1, %2<round_saeonly_op3>}): Changed to ...
19130 vgetexp<ssescalarmodesuffix>
19131 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19132 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
19133 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
19134 avx512f_vgetmant<mode><mask_scalar_name>
19135 <round_saeonly_scalar_name> ... this.
19136 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
19137 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
19138 vgetmant<ssescalarmodesuffix>
19139 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
19140 %0<mask_scalar_operand4>, %1, %2
19141 <round_saeonly_scalar_mask_op4>, %3} ... this.
19142 * config/i386/subst.md (mask_scalar_operand4,
19143 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
19144 round_saeonly_scalar_nimm_predicate): New subst attributes.
19146 2017-07-06 Julia Koval <julia.koval@intel.com>
19148 * config/i386/i386.c (ix86_erase_embedded_rounding):
19149 Remove code for old rounding pattern.
19151 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
19153 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
19155 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
19157 * doc/sourcebuild.texi (Test Directives, Variants of
19158 dg-require-support): Add documentation for dg-require-stack-check.
19160 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
19162 * config/i386/subst.md (mask_scalar, round_scalar,
19163 round_saeonly_scalar): New meta-templates.
19164 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
19165 round_scalar_mask_operand3, round_scalar_mask_op3,
19166 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
19167 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
19168 round_saeonly_scalar_constraint,
19169 round_saeonly_scalar_prefix): New subst attribute.
19170 * config/i386/sse.md
19171 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
19172 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
19173 <round_scalar_name> ... this.
19174 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
19175 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
19176 <round_scalar_name> ... this.
19177 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
19178 <sse>_vm<code><mode>3<mask_scalar_name>
19179 <round_saeonly_scalar_name> ... this.
19180 (v<plusminus_mnemonic><ssescalarmodesuffix>
19181 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
19182 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
19183 v<plusminus_mnemonic><ssescalarmodesuffix>
19184 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19185 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
19186 (v<multdiv_mnemonic><ssescalarmodesuffix>
19187 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
19188 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
19189 v<multdiv_mnemonic><ssescalarmodesuffix>
19190 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19191 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
19192 (v<maxmin_float><ssescalarmodesuffix>
19193 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
19194 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
19195 v<maxmin_float><ssescalarmodesuffix>
19196 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
19197 %0<mask_scalar_operand3>, %1, %<iptr>2
19198 <round_saeonly_scalar_mask_op3>} ... this.
19200 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
19202 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
19203 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
19205 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
19206 Alan Hayward <alan.hayward@arm.com>
19207 David Sherwood <david.sherwood@arm.com>
19209 * combine.c (simplify_if_then_else): Remove "enum" before
19211 * compare-elim.c (can_eliminate_compare): Likewise.
19212 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19214 (aarch64_lookup_simd_builtin_type): Likewise.
19215 (aarch64_simd_builtin_type): Likewise.
19216 (aarch64_init_simd_builtin_types): Likewise.
19217 (aarch64_simd_expand_args): Likewise.
19218 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
19220 (aarch64_reverse_mask): Likewise.
19221 (aarch64_simd_emit_reg_reg_move): Likewise.
19222 (aarch64_gen_adjusted_ldpstp): Likewise.
19223 (aarch64_ccmp_mode_to_code): Likewise.
19224 (aarch64_operands_ok_for_ldpstp): Likewise.
19225 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
19226 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
19228 (aarch64_min_divisions_for_recip_mul): Likewise.
19229 (aarch64_reassociation_width): Likewise.
19230 (aarch64_get_condition_code_1): Likewise.
19231 (aarch64_simd_emit_reg_reg_move): Likewise.
19232 (aarch64_simd_attr_length_rglist): Likewise.
19233 (aarch64_reverse_mask): Likewise.
19234 (aarch64_operands_ok_for_ldpstp): Likewise.
19235 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
19236 (aarch64_gen_adjusted_ldpstp): Likewise.
19237 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
19239 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
19240 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19241 (arm_lookup_simd_builtin_type): Likewise.
19242 (arm_simd_builtin_type): Likewise.
19243 (arm_init_simd_builtin_types): Likewise.
19244 (arm_expand_builtin_args): Likewise.
19245 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
19246 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
19247 (ft32_setup_incoming_varargs): Likewise.
19248 (ft32_function_arg): Likewise.
19249 (ft32_function_arg_advance): Likewise.
19250 (ft32_pass_by_reference): Likewise.
19251 (ft32_arg_partial_bytes): Likewise.
19252 (ft32_valid_pointer_mode): Likewise.
19253 (ft32_addr_space_pointer_mode): Likewise.
19254 (ft32_addr_space_legitimate_address_p): Likewise.
19255 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
19257 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
19258 (ix86_emit_outlined_ms2sysv_restore): Likewise.
19259 (iamcu_alignment): Likewise.
19260 (canonicalize_vector_int_perm): Likewise.
19261 (ix86_noce_conversion_profitable_p): Likewise.
19262 (ix86_mpx_bound_mode): Likewise.
19263 (ix86_operands_ok_for_move_multiple): Likewise.
19264 * config/microblaze/microblaze-protos.h
19265 (microblaze_expand_conditional_branch_reg): Likewise.
19266 * config/microblaze/microblaze.c
19267 (microblaze_expand_conditional_branch_reg): Likewise.
19268 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19270 (rs6000_reassociation_width): Likewise.
19271 (rs6000_invalid_binary_op): Likewise.
19272 (fusion_p9_p): Likewise.
19273 (emit_fusion_p9_load): Likewise.
19274 (emit_fusion_p9_store): Likewise.
19275 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
19277 (riscv_hard_regno_mode_ok_p): Likewise.
19278 (riscv_address_insns): Likewise.
19279 (riscv_split_symbol): Likewise.
19280 (riscv_legitimize_move): Likewise.
19281 (riscv_function_value): Likewise.
19282 (riscv_hard_regno_nregs): Likewise.
19283 (riscv_expand_builtin): Likewise.
19284 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
19285 (riscv_build_integer): Likewise.
19286 (riscv_split_integer): Likewise.
19287 (riscv_legitimate_constant_p): Likewise.
19288 (riscv_cannot_force_const_mem): Likewise.
19289 (riscv_regno_mode_ok_for_base_p): Likewise.
19290 (riscv_valid_base_register_p): Likewise.
19291 (riscv_valid_offset_p): Likewise.
19292 (riscv_valid_lo_sum_p): Likewise.
19293 (riscv_classify_address): Likewise.
19294 (riscv_legitimate_address_p): Likewise.
19295 (riscv_address_insns): Likewise.
19296 (riscv_load_store_insns): Likewise.
19297 (riscv_force_binary): Likewise.
19298 (riscv_split_symbol): Likewise.
19299 (riscv_force_address): Likewise.
19300 (riscv_legitimize_address): Likewise.
19301 (riscv_move_integer): Likewise.
19302 (riscv_legitimize_const_move): Likewise.
19303 (riscv_legitimize_move): Likewise.
19304 (riscv_address_cost): Likewise.
19305 (riscv_subword): Likewise.
19306 (riscv_output_move): Likewise.
19307 (riscv_canonicalize_int_order_test): Likewise.
19308 (riscv_emit_int_order_test): Likewise.
19309 (riscv_function_arg_boundary): Likewise.
19310 (riscv_pass_mode_in_fpr_p): Likewise.
19311 (riscv_pass_fpr_single): Likewise.
19312 (riscv_pass_fpr_pair): Likewise.
19313 (riscv_get_arg_info): Likewise.
19314 (riscv_function_arg): Likewise.
19315 (riscv_function_arg_advance): Likewise.
19316 (riscv_arg_partial_bytes): Likewise.
19317 (riscv_function_value): Likewise.
19318 (riscv_pass_by_reference): Likewise.
19319 (riscv_setup_incoming_varargs): Likewise.
19320 (riscv_print_operand): Likewise.
19321 (riscv_elf_select_rtx_section): Likewise.
19322 (riscv_save_restore_reg): Likewise.
19323 (riscv_for_each_saved_reg): Likewise.
19324 (riscv_register_move_cost): Likewise.
19325 (riscv_hard_regno_mode_ok_p): Likewise.
19326 (riscv_hard_regno_nregs): Likewise.
19327 (riscv_class_max_nregs): Likewise.
19328 (riscv_memory_move_cost): Likewise.
19329 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
19330 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
19331 (rl78_addr_space_address_mode): Likewise.
19332 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19334 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19335 (rs6000_reassociation_width): Likewise.
19336 (rs6000_invalid_binary_op): Likewise.
19337 (fusion_p9_p): Likewise.
19338 (emit_fusion_p9_load): Likewise.
19339 (emit_fusion_p9_store): Likewise.
19340 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
19341 (ok_for_simple_move_operands): Likewise.
19342 (ok_for_simple_move_strict_operands): Likewise.
19343 (ok_for_simple_arith_logic_operands): Likewise.
19344 (visium_legitimize_reload_address): Likewise.
19345 (visium_select_cc_mode): Likewise.
19346 (output_cbranch): Likewise.
19347 (visium_split_double_move): Likewise.
19348 (visium_expand_copysign): Likewise.
19349 (visium_expand_int_cstore): Likewise.
19350 (visium_expand_fp_cstore): Likewise.
19351 * config/visium/visium.c (visium_pass_by_reference): Likewise.
19352 (visium_function_arg): Likewise.
19353 (visium_function_arg_advance): Likewise.
19354 (visium_libcall_value): Likewise.
19355 (visium_setup_incoming_varargs): Likewise.
19356 (visium_legitimate_constant_p): Likewise.
19357 (visium_legitimate_address_p): Likewise.
19358 (visium_legitimize_address): Likewise.
19359 (visium_secondary_reload): Likewise.
19360 (visium_register_move_cost): Likewise.
19361 (visium_memory_move_cost): Likewise.
19362 (prepare_move_operands): Likewise.
19363 (ok_for_simple_move_operands): Likewise.
19364 (ok_for_simple_move_strict_operands): Likewise.
19365 (ok_for_simple_arith_logic_operands): Likewise.
19366 (visium_function_value_1): Likewise.
19367 (rtx_ok_for_offset_p): Likewise.
19368 (visium_legitimize_reload_address): Likewise.
19369 (visium_split_double_move): Likewise.
19370 (visium_expand_copysign): Likewise.
19371 (visium_expand_int_cstore): Likewise.
19372 (visium_expand_fp_cstore): Likewise.
19373 (visium_split_cstore): Likewise.
19374 (visium_select_cc_mode): Likewise.
19375 (visium_split_cbranch): Likewise.
19376 (output_cbranch): Likewise.
19377 (visium_print_operand_address): Likewise.
19378 * expmed.c (flip_storage_order): Likewise.
19379 * expmed.h (emit_cstore): Likewise.
19380 (flip_storage_order): Likewise.
19381 * genrecog.c (validate_pattern): Likewise.
19382 * hsa-gen.c (gen_hsa_addr): Likewise.
19383 * internal-fn.c (expand_arith_overflow): Likewise.
19384 * ira-color.c (allocno_copy_cost_saving): Likewise.
19385 * lra-assigns.c (find_hard_regno_for_1): Likewise.
19386 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
19387 (process_invariant_for_inheritance): Likewise.
19388 * lra-eliminations.c (move_plus_up): Likewise.
19389 * omp-low.c (lower_oacc_reductions): Likewise.
19390 * simplify-rtx.c (simplify_subreg): Likewise.
19391 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
19392 (TARGET_CHKP_BOUND_MODE): Likewise..
19393 * targhooks.c (default_chkp_bound_mode): Likewise.
19394 (default_setup_incoming_vararg_bounds): Likewise.
19395 * targhooks.h (default_chkp_bound_mode): Likewise.
19396 (default_setup_incoming_vararg_bounds): Likewise.
19397 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
19398 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
19399 (have_whole_vector_shift): Likewise.
19400 * tree-vect-stmts.c (vectorizable_load): Likewise.
19401 * doc/tm.texi: Regenerate.
19403 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19405 Graceful degrade if Binutils PR21472 is not available.
19408 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
19409 .rodata in flash test fails.
19410 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
19411 * confgure: Regenerate.
19412 * config.in: Regenerate.
19413 * config/avr/avr.c (avr_asm_named_section)
19414 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
19415 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
19416 (avr_asm_init_sections): Same.
19418 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19420 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
19421 (fma<VH:mode>4_intrinsic): Likewise.
19422 (*fmsub<VCVTF:mode>4): Likewise.
19423 (*fmsub<VH:mode>4_intrinsic): Likewise.
19425 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19428 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
19429 Don't depend on "optimize > 0".
19430 (out_movhi_r_mr, out_movqi_mr_r): Same.
19431 (out_movhi_mr_r, out_movqi_r_mr): Same.
19432 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
19433 io_address_operand on "optimize > 0".
19435 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19437 * tree-loop-distribution.c: Add general explanantion on the pass.
19438 (generate_loops_for_partition): Mark distributed loop.
19439 (pg_add_dependence_edges): New parameter. Handle alias data
19440 dependence specially and record it in the parameter if asked.
19441 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
19442 (init_partition_graph_vertices, add_partition_graph_edge): New.
19443 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
19444 (free_partition_graph_vdata, build_partition_graph): New.
19445 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
19446 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
19447 (data_ref_segment_size, latch_dominated_by_data_ref): New.
19448 (compute_alias_check_pairs, version_loop_by_alias_check): New.
19449 (version_for_distribution_p, finalize_partitions): New.
19450 (distribute_loop): Handle alias data dependence specially. Factor
19451 out loop fusion code as functions and call these functions.
19453 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19455 * tree-loop-distribution.c (classify_partition): New parameter and
19456 better handle reduction statement.
19457 (rdg_build_partitions): Revise comment.
19458 (distribute_loop): Compute statements in all partitions and pass it
19459 to classify_partition.
19461 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19463 * tree-loop-distribution.c (enum partition_type): New.
19464 (struct partition): New field type.
19465 (partition_merge_into): Add parameter. Update partition type.
19466 (data_dep_in_cycle_p, update_type_for_merge): New functions.
19467 (build_rdg_partition_for_vertex): Compute partition type.
19468 (rdg_build_partitions): Dump partition type.
19469 (distribute_loop): Update calls to partition_merge_into.
19471 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19473 * tree-loop-distribution.c (struct ddr_hasher): New.
19474 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
19476 (classify_partition): Call get_data_dependence.
19477 (pg_add_dependence_edges): Ditto.
19478 (distribute_loop): Release data dependence hash table.
19480 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19482 * tree-loop-distribution.c (ref_base_address): Delete.
19483 (similar_memory_accesses): Rename ...
19484 (share_memory_accesses): ... to this. Check if partitions access
19485 the same memory reference.
19486 (distribute_loop): Call share_memory_accesses.
19488 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19490 * tree-loop-distribution.c (struct partition): New field recording
19491 its data reference.
19492 (partition_alloc, partition_free): Init and release data refs.
19493 (partition_merge_into): Merge data refs.
19494 (build_rdg_partition_for_vertex): Collect data refs for partition.
19495 (pg_add_dependence_edges): Change parameters from vector to bitmap.
19497 (distribute_loop): Remve data refs from vertice data of partition
19500 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19502 * tree-loop-distribution.c (params.h): Include header file.
19503 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
19504 (datarefs_vec): New global var.
19505 (create_rdg_vertices): Use datarefs_vec directly.
19506 (free_rdg): Don't free data references.
19507 (build_rdg): Update use. Don't free data references.
19508 (distribute_loop): Compute global variable for data references.
19509 Bail out if there are too many data references.
19511 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19513 * tree-loop-distribution.c (loop_nest): New global var.
19514 (build_rdg): Use loop directly, rather than loop nest.
19515 (pg_add_dependence_edges): Remove loop nest parameter. Use global
19517 (distribute_loop): Compute global variable loop nest. Update use.
19519 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19521 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
19522 (partition_merge_into): New parameter. Dump reason for fusion.
19523 (distribute_loop): Update use of partition_merge_into.
19525 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19527 * tree-loop-distribution.c (bb_top_order_index): New.
19528 (bb_top_order_index_size, bb_top_order_cmp): New.
19529 (stmts_from_loop): Use topological order.
19530 (pass_loop_distribution::execute): Compute and release topological
19531 order for basic blocks.
19533 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19535 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
19538 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19540 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
19541 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
19542 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
19543 * internal-fn.def (LOOP_DIST_ALIAS): New.
19544 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
19545 (fold_loop_internal_call): ... this.
19546 (vect_loop_dist_alias_call): New function.
19547 (set_uid_loop_bbs): Call fold_loop_internal_call.
19548 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
19551 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19554 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
19555 Require dead FLAGS_REG at the beginning of a peephole.
19557 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19560 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
19561 arguments in the call to __builtin_ia32_sbb_u32.
19562 (_subborrow_u64): Swap _X and _Y arguments in the call to
19563 __builtin_ia32_sbb_u64.
19565 2017-07-04 Jakub Jelinek <jakub@redhat.com>
19568 * tree-vrp.c (compare_assert_loc): Turn into a function template
19569 with stable template parameter. Only test if a->e is NULL,
19570 !a->e == !b->e has been verified already. Use e == NULL or
19571 e != NULL instead of e or ! e tests. If stable is true, don't use
19572 iterative_hash_expr, on the other side allow a or b or both NULL
19573 and sort the NULLs last.
19574 (process_assert_insertions): Sort using compare_assert_loc<false>
19575 instead of compare_assert_loc, later sort using
19576 compare_assert_loc<true> before calling process_assert_insertions_for
19577 in a loop. Use break instead of continue once seen NULL pointer.
19579 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19581 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19582 Cortex-R7 and Cortex-R8 processors.
19584 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19586 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
19587 uninitialized while src is not.
19589 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
19591 * common/config/arm/arm-common.c: Adjust include path for
19593 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
19594 (arm-cpu.h): Create in build directory. Adjust dependency rules.
19595 (arm-cpu-data.h): Likewise.
19596 (arm-cpu-cdata.h): Likewise.
19597 * config/arm/arm-cpu.h: Delete.
19598 * config/arm/arm-cpu-cdata.h: Delete.
19599 * config/arm/arm-cpu-data.h: Delete.
19601 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
19603 * config/arm/arm-cpus.in (cortex-a55): New.
19604 (cortex-a75): Likewise.
19605 (cortex-a75.cortex-a55): Likewise.
19606 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
19608 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
19609 * config/arm/arm-cpu-cdata.h: Regenerate.
19610 * config/arm/arm-cpu-data.h: Regenerate.
19611 * config/arm/arm-cpu.h: Regenerate.
19612 * config/arm/arm-tables.opt: Regenerate.
19613 * config/arm/arm-tune.md: Regenerate.
19615 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19617 * haifa-sched.c (sched_create_recovery_edges): Update profile.
19619 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19621 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
19624 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
19626 PR tree-optimization/81292
19627 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
19628 full_string_p, also call adjust_related_strinfos if the adjustment
19629 is simple, otherwise invalidate related strinfos.
19631 2017-07-04 Martin Liska <mliska@suse.cz>
19634 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
19635 newly created variable as DECL_IGNORED_P.
19637 2017-07-04 Martin Liska <mliska@suse.cz>
19640 * ipa-inline.c (inline_small_functions):
19641 Use xstrdup_for_dump.
19643 2017-07-04 Tom de Vries <tom@codesourcery.com>
19645 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
19647 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19650 * config/darwin.c (darwin_function_switched_text_sections):
19653 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19655 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19657 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19659 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19661 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19663 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19664 min_profitable_iters, and th as inclusive lower bounds.
19665 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19666 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19667 for min_profitable_iters and min_profitable_estimate.
19668 (vect_transform_loop): Treat th as an inclusive lower bound.
19669 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19671 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19674 * config/darwin.c (darwin_function_switched_text_sections):
19675 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19676 in two pieces, and suppress the use of buf.
19678 2017-07-03 Nathan Sidwell <nathan@acm.org>
19680 * hash-table.h (hash_table_mod1): Fix indentation.
19682 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19684 PR middle-end/81290
19685 * predict.c (force_edge_cold): Be more careful about propagation
19687 * profile-count.h (profile_probability::guessed,
19688 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19690 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19692 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19694 * doc/invoke.texi (rcpc architecture extension): Document it.
19696 2017-07-03 Richard Biener <rguenther@suse.de>
19698 PR tree-optimization/60510
19699 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19700 the scalar reduction PHI and use it.
19701 (vectorizable_reduction): Properly guard the single_defuse_cycle
19702 path for non-SLP reduction chains where we cannot use it.
19703 Rework reduc_def/index and vector type deduction. Rework
19704 vector operand gathering during reduction op code-gen.
19705 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19706 chains dissolve the chain and leave it to non-SLP reduction
19709 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19711 * tree-data-ref.h (dr_alignment): Declare.
19712 * tree-data-ref.c (dr_alignment): New function.
19713 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19714 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19716 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19718 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19720 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19721 and base_misalignment fields.
19722 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19723 * tree-data-ref.c: Include builtins.h.
19724 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19725 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19726 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19727 * tree-vect-data-refs.c: Include tree-cfg.h.
19728 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19729 fields instead of calculating an alignment here.
19730 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19731 innermost_loop_behavior fields.
19733 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19735 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19737 (DR_STEP_ALIGNMENT): New macro.
19738 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19739 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19740 (create_data_ref): Print it.
19741 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19742 to tell whether the step preserves vector (mis)alignment.
19743 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19744 Move the check for an integer step and generalise to all INTEGER_CST.
19745 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19746 Print the outer step alignment.
19748 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19750 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19751 with offset_alignment.
19752 (DR_ALIGNED_TO): Delete.
19753 (DR_OFFSET_ALIGNMENT): New macro.
19754 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19755 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19756 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19757 (create_data_ref): Likewise.
19758 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19759 (vect_analyze_data_refs): Likewise.
19760 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19761 creating dummy innermost behavior.
19763 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19765 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19766 with a "innermost_loop_behavior *" and refeence tree.
19767 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19768 (create_data_ref): Update call accordingly.
19769 * tree-predcom.c (find_looparound_phi): Likewise.
19771 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19773 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19774 fields with dr_wrt_vec_loop.
19775 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19776 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19777 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19778 (vect_dr_behavior): New function.
19779 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19780 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19781 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19782 track whether the step preserves the misalignment.
19783 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19784 Use vect_dr_behavior.
19785 (vect_setup_realignment): Update call accordingly.
19786 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19787 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19788 call to vect_create_addr_base_for_vector_ref.
19789 (vect_create_cond_for_align_checks): Likewise.
19790 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19791 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19792 (vect_recog_mask_conversion_pattern): Likewise.
19793 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19794 (new_stmt_vec_info): Remove redundant zeroing.
19796 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19798 * common/config/arm/arm-common.c (arm_be8_option): New function.
19799 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19800 (ISA_ARMv6): Add isa_bit_be8.
19801 * config/arm/arm.h (arm_be8_option): Add prototype.
19802 (BE8_SPEC_FUNCTION): New define.
19803 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19804 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19805 (mlittle-endian): Similarly.
19806 (mbe8, mbe32): New options.
19807 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19808 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19810 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19812 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19814 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19816 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19817 (cleanup_tree_cfg_bb): Use it.
19818 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19820 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19822 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19825 * loop-doloop.c (add_test): Update profile.
19827 2017-07-03 Martin Liska <mliska@suse.cz>
19830 * sanopt.c (rewrite_usage_of_param): New function.
19831 (sanitize_rewrite_addressable_params): Likewise.
19832 (pass_sanopt::execute): Call rewrite_usage_of_param.
19834 2017-07-03 Richard Biener <rguenther@suse.de>
19836 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19837 back to using VIEW_CONVERT_EXPR.
19839 2017-07-03 Martin Liska <mliska@suse.cz>
19842 * doc/extend.texi: Document when a resolver function is
19843 generated for target_clones.
19845 2017-07-03 Martin Liska <mliska@suse.cz>
19847 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19848 and shadow memory of auto variables which are subject of
19849 use-after-scope sanitization.
19850 (asan_expand_mark_ifn): Add do set only when is_poison.
19852 2016-07-03 Richard Biener <rguenther@suse.de>
19854 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19856 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19857 (vectorizable_reduction): Perform reduction PHI creation when
19858 visiting a reduction PHI and adjust and simplify code generation
19859 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19860 (vect_transform_loop): Visit reduction PHIs.
19861 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19862 defs into the SLP tree.
19863 (vect_build_slp_tree): Reduction defs terminate the recursion.
19864 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19866 (vect_get_vec_defs_for_stmt_copy): Export.
19867 (vect_get_vec_defs): Likewise.
19868 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19870 (vect_get_vec_defs_for_stmt_copy): Declare.
19871 (vect_get_vec_defs): Likewise.
19873 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19875 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19876 parameter with a "loop" parameter and use it instead of the
19877 loop containing DR_STMT. Don't check simple_iv when doing
19878 BB analysis. Describe the two analysis modes in the comment.
19880 2017-07-03 Tom de Vries <tom@codesourcery.com>
19882 PR tree-optimization/69468
19883 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19884 (find_same_succ_bb): Handle ignore_edge_flags.
19886 2017-07-03 Tom de Vries <tom@codesourcery.com>
19888 PR tree-optimization/81192
19889 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19891 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19893 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19895 2017-07-03 Tom de Vries <tom@codesourcery.com>
19897 PR tree-optimization/81192
19898 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19899 BB_SAME_SUCC (bb) == NULL.
19901 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19903 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19906 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19908 * dumpfile.c: Include profile-count.h
19909 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19911 (insert_cond_bb): Update profile.
19912 * tree-cfg.h (insert_cond_bb): Update prototype.
19913 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19914 * tree-dump.c: Do not include tree-cfg.
19916 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19918 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19920 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19922 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19925 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19927 * tree-complex.c (expand_complex_div_wide): update profile.
19929 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19930 Alan Hayward <alan.hayward@arm.com>
19931 David Sherwood <david.sherwood@arm.com>
19933 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19934 (CORETYPES_H): New define.
19935 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19936 (insn-modes-inline.h, s-modes-inline-h): New rules.
19937 (generated_files): Add insn-modes-inline.h.
19938 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19939 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19940 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19941 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19942 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19943 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19944 (build/gencodes.o, build/genconditions.o): Likewise.
19945 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19946 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19947 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19948 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19949 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19950 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19951 * coretypes.h: Include everything up to real.h for generators.
19952 Include insn-modes.h first. Include wide-int-print.h after
19953 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19954 * machmode.h: Don't include insn-modes.h here.
19955 * function-tests.c: Remove includes of signop.h, machmode.h,
19956 double-int.h and wide-int.h.
19958 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19960 * optc-save-gen.awk: Likewise.
19961 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19962 * godump.c: Remove include of wide-int-print.h.
19963 * pretty-print.h: Likewise.
19964 * wide-int-print.cc: Likewise.
19965 * wide-int.cc: Likewise.
19966 * hash-map-tests.c: Remove include of signop.h.
19967 * hash-set-tests.c: Likewise.
19968 * rtl-tests.c: Likewise.
19969 * mkconfig.sh: Remove include of machmode.h.
19970 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19972 (emit_insn_modes_inline_h): ...this new function. Emit the code
19973 into an insn-modes-inline.h header file, adding appropriate
19974 include guards and end comments.
19975 (emit_insn_modes_c_header): Remove include of machmode.h.
19976 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19978 (main): Handle -i flag and call emit_insn_modes_inline_h when
19981 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19983 * tree-ssa-strlen.c (strinfo): Rename the length field to
19984 nonzero_chars. Add a full_string_p field.
19985 (compare_nonzero_chars, zero_length_string_p): New functions.
19986 (get_addr_stridx): Add an offset_out parameter.
19987 Use compare_nonzero_chars.
19988 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19989 (new_strinfo): Update after above changes to strinfo.
19990 (set_endptr_and_length): Set full_string_p.
19991 (get_string_length): Update after above changes to strinfo.
19992 (unshare_strinfo): Update call to new_strinfo.
19993 (maybe_invalidate): Likewise.
19994 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19995 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19996 as a uhwi instead of an shwi. Update after above changes to
19997 strinfo and new_strinfo.
19998 (zero_length_string): Assert that chainsi contains full strings.
19999 Use zero_length_string_p. Update call to new_strinfo.
20000 (adjust_related_strinfos): Update after above changes to strinfo.
20001 Copy full_string_p from origsi.
20002 (adjust_last_stmt): Use zero_length_string_p.
20003 (handle_builtin_strlen): Update after above changes to strinfo and
20004 new_strinfo. Install the lhs as the string length if the previous
20005 entry didn't describe a full string.
20006 (handle_builtin_strchr): Update after above changes to strinfo
20008 (handle_builtin_strcpy): Likewise.
20009 (handle_builtin_strcat): Likewise.
20010 (handle_builtin_malloc): Likewise.
20011 (handle_pointer_plus): Likewise.
20012 (handle_builtin_memcpy): Likewise. Track nonzero characters
20013 that aren't necessarily followed by a nul terminator.
20014 (handle_char_store): Likewise.
20016 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20018 PR tree-optimization/80769
20019 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
20020 for malloc and calloc. Document the new invariant that all related
20021 strinfos have delayed lengths or none do.
20022 (verify_related_strinfos): Move earlier in file.
20023 (set_endptr_and_length): New function, split out from...
20024 (get_string_length): ...here. Also set the lengths of related
20026 (zero_length_string): Assert that chainsi has known (rather than
20028 (adjust_related_strinfos): Likewise.
20030 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
20032 PR tree-optimization/81136
20033 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
20034 assert that two references with the same misalignment have the same
20035 compile-time misalignment if those compile-time misalignments
20038 2017-07-01 Andi Kleen <ak@linux.intel.com>
20040 * print-tree.c (print_node): Print all attributes.
20042 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20044 * cfg.c (scale_bbs_frequencies): New function.
20045 * cfg.h (scale_bbs_frequencies): Declare it.
20046 * cfgloopanal.c (single_likely_exit): Cleanup.
20047 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
20049 (scale_loop_profile): Likewise.
20050 (loop_version): Likewise.
20051 (create_empty_loop_on_edge): Update.
20052 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
20053 scale_loop_frequencies, scale_loop_profile, loopify,
20054 loop_version): Update prototypes.
20055 * modulo-sched.c (sms_schedule): Update.
20056 * predict.c (unlikely_executed_edge_p): Also check probability.
20057 (probably_never_executed_edge_p): Fix typo.
20058 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20059 * tree-parloops.c (gen_parallel_loop): Update.
20060 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
20061 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20062 * tree-ssa-loop-split.c (split_loop): Update.
20063 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20064 * tree-vect-loop-manip.c (vect_do_peeling): Update.
20065 (vect_loop_versioning): Update.
20066 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20068 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20070 * trans-mem.c (split_bb_make_tm_edge): Update profile.
20072 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20074 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
20075 to keep profile consistent.
20077 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20079 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
20080 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
20081 * profile-count.h (max_safe_multiplier): Make unsigned.
20082 (profile_count::guessed_zero): New.
20084 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20086 * bb-reorder.c (fix_up_crossing_landing_pad,
20087 fix_crossing_conditional_branches): Use make_single_succ_edge
20088 to keep profile consistent.
20090 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
20092 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
20095 2017-07-01 Jakub Jelinek <jakub@redhat.com>
20098 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
20099 the right scopes, make sure cond_jump isn't preserved between multiple
20100 iterations. Search for fallthru edge whenever there are 3+ edges and
20101 use find_fallthru_edge for it.
20103 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20105 Patch by Alexander Monakov <amonakov@ispras.ru>
20106 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
20107 probabilities consistently.
20109 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20111 * pa.c (pa_expand_compare_and_swap_loop): Update call of
20112 emit_cmp_and_jump_insns.
20114 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20117 * tree-inline.c (expand_call_inline): Combine profile statuses.
20119 2017-06-30 Andrew Pinski <apinski@cavium.com>
20121 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
20122 fold_stmt returned true.
20124 2017-06-30 Nathan Sidwell <nathan@acm.org>
20126 * ggc.h (empty_string): Delete.
20127 * cfgexpand.c (expand_asm_stmt): Use plain "".
20128 * optabs.c (expand_asm_memory_barrier): Likewise.
20129 * stringpool.c (empty_string): Delete.
20130 (digit_vector, digit_string): Delete.
20131 (ggc_alloc_string): Use plain "", don't optimize single digit
20132 strings. Use ggc_alloc_atomic.
20134 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
20136 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
20137 comparison set and one other set, use the cost of the non-comparison
20140 2017-06-30 Nathan Sidwell <nathan@acm.org>
20142 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
20145 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
20147 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
20148 loops. Remove now unneeded calls to gimple_switch_set_label() that
20149 just set removed labels to NULL_TREE.
20151 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
20153 * tree-ssanames.c (set_range_info_raw): Abstract from ...
20154 (set_range_info): ...here. Only call set_range_info_raw if domain
20156 (set_nonzero_bits): Call set_range_info_raw.
20157 * tree-ssanames.h (set_range_info_raw): New.
20159 2017-06-30 Jakub Jelinek <jakub@redhat.com>
20162 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
20163 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
20164 of nonimmediate_operand and <store_mask_constraint> instead of m
20165 for the input operand. For V8FI iterator, always split if input
20166 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
20167 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
20168 <store_mask_predicate> instead of register_operand and
20169 <store_mask_constraint> instead of v for the input operand. Make
20170 sure both operands aren't MEMs for if not <mask_applied>.
20172 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
20174 * lto-wrapper.c (copy_file) Close both file descriptors before
20177 2017-06-30 Martin Liska <mliska@suse.cz>
20180 * multiple_target.c (create_dispatcher_calls): Make ifunc
20181 also for function that don't have calls or are not referenced.
20183 2017-06-30 Richard Biener <rguenther@suse.de>
20185 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
20186 analyze the first scalar stmt. Move vector type computation
20187 for the BB case here from ...
20188 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
20189 live operation processing in the SLP case properly.
20191 2017-06-30 Richard Biener <rguenther@suse.de>
20193 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
20195 2017-06-30 Martin Liska <mliska@suse.cz>
20198 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
20199 before BUILT_IN_UNWIND_RESUME when ASAN is used.
20201 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
20203 * doc/invoke.texi (AArch64): Add missing options and remove redundant
20206 2017-06-30 Richard Biener <rguenther@suse.de>
20208 PR tree-optimization/81249
20209 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
20210 condition reduction result to original scalar type.
20212 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20214 * profile-count.h (enum profile_quality): Fix typos and whitespace
20217 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20219 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
20220 type for branch probabilities.
20222 2017-06-29 Julian Brown <julian@codesourcery.com>
20223 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20225 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
20226 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
20227 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
20228 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
20230 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20232 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
20233 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
20234 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
20235 CC usage from generic code to here.
20236 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
20237 CC usage into the target macros.
20239 2017-06-29 Maya Rashish <coypu@sdf.org>
20241 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
20244 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20246 * arm/arm-builtins.c: Include profile-count.h
20247 * except.c (sjlj_emit_function_enter): Use
20248 profile_probability::unlikely.
20250 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20252 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
20253 and tocrel_offset be pointer args rather than implicitly using
20255 (legitimate_constant_pool_address_p, rs6000_emit_move,
20256 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
20257 tocrel_offset and use in toc_relative_expr_p call.
20258 (print_operand, print_operand_address): Use static tocrel_base_oac
20259 and tocrel_offset_oac.
20260 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
20263 2017-06-29 Maya Rashish <coypu@sdf.org>
20265 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
20267 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
20269 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
20270 objects, take into account only the alignment of 'op0' and 'mode1' if
20273 2017-06-29 Steve Ellcey <sellcey@cavium.com>
20275 * ccmp.c (ccmp_tree_comparison_p): New function.
20276 (ccmp_candidate_p): Update to use above function.
20277 (get_compare_parts): New function.
20278 (expand_ccmp_next): Update to use new functions.
20279 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
20281 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
20282 take mode as argument.
20283 * ccmp.h (expand_ccmp_expr): Add mode as argument.
20284 * expr.c (expand_expr_real_1): Pass mode as argument.
20286 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
20288 * combine.c (combine_instructions): Print insns to dump_file, together
20291 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20293 * asan.c (asan_emit_stack_protection): Update.
20294 (create_cond_insert_point): Update.
20295 * auto-profile.c (afdo_propagate_circuit): Update.
20296 * basic-block.h (struct edge_def): Turn probability to
20297 profile_probability.
20298 (EDGE_FREQUENCY): Update.
20299 * bb-reorder.c (find_traces_1_round): Update.
20300 (better_edge_p): Update.
20301 (sanitize_hot_paths): Update.
20302 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
20303 (make_single_succ_edge): Update.
20304 (check_bb_profile): Update.
20305 (dump_edge_info): Update.
20306 (update_bb_profile_for_threading): Update.
20307 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
20308 probabilitycount to 0.
20309 * cfgbuild.c (compute_outgoing_frequencies): Update.
20310 * cfgcleanup.c (try_forward_edges): Update.
20311 (outgoing_edges_match): Update.
20312 (try_crossjump_to_edge): Update.
20313 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
20314 (expand_gimple_tailcall): Update.
20315 (construct_init_block): Use make_single_succ_edge.
20316 (construct_exit_block): Use make_single_succ_edge.
20317 * cfghooks.c (verify_flow_info): Update.
20318 (redirect_edge_succ_nodup): Update.
20319 (split_edge): Update.
20320 (account_profile_record): Update.
20321 * cfgloopanal.c (single_likely_exit): Update.
20322 * cfgloopmanip.c (scale_loop_profile): Update.
20323 (set_zero_probability): Remove.
20324 (duplicate_loop_to_header_edge): Update.
20325 * cfgloopmanip.h (loop_version): Update prototype.
20326 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
20327 (force_nonfallthru_and_redirect): Update.
20328 (update_br_prob_note): Update.
20329 (rtl_verify_edges): Update.
20330 (purge_dead_edges): Update.
20331 (rtl_lv_add_condition_to_bb): Update.
20332 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
20333 * cgraphunit.c (init_lowered_empty_function): Update.
20334 (cgraph_node::expand_thunk): Update.
20335 * cilk-common.c: Include profile-count.h
20336 * dojump.c (inv): Remove.
20337 (jumpifnot): Update.
20338 (jumpifnot_1): Update.
20339 (do_jump_1): Update.
20341 (do_jump_by_parts_greater_rtx): Update.
20342 (do_compare_rtx_and_jump): Update.
20343 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
20344 do_jump_1. do_compare_rtx_and_jump): Update prototype.
20345 * dwarf2cfi.c: Include profile-count.h
20346 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
20347 (sjlj_emit_dispatch_table): Likewise.
20348 * explow.c: Include profile-count.h
20349 * expmed.c (emit_store_flag_force): Update.
20350 (do_cmp_and_jump): Update.
20351 * expr.c (compare_by_pieces_d::generate): Update.
20352 (compare_by_pieces_d::finish_mode): Update.
20353 (emit_block_move_via_loop): Update.
20354 (store_expr_with_bounds): Update.
20355 (store_constructor): Update.
20356 (expand_expr_real_2): Update.
20357 (expand_expr_real_1): Update.
20358 * expr.h (try_casesi, try_tablejump): Update prototypes.
20359 * gimple-pretty-print.c (dump_probability): Update.
20360 (dump_profile): New.
20361 (dump_gimple_label): Update.
20362 (dump_gimple_bb_header): Update.
20363 * graph.c (draw_cfg_node_succ_edges): Update.
20364 * hsa-gen.c (convert_switch_statements): Update.
20365 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
20366 (find_if_case_1): Update.
20367 (find_if_case_2): Update.
20368 * internal-fn.c (expand_arith_overflow_result_store): Update.
20369 (expand_addsub_overflow): Update.
20370 (expand_neg_overflow): Update.
20371 (expand_mul_overflow): Update.
20372 (expand_vector_ubsan_overflow): Update.
20373 * ipa-cp.c (good_cloning_opportunity_p): Update.
20374 * ipa-split.c (split_function): Use make_single_succ_edge.
20375 * ipa-utils.c (ipa_merge_profiles): Update.
20376 * loop-doloop.c (add_test): Update.
20377 (doloop_modify): Update.
20378 * loop-unroll.c (compare_and_jump_seq): Update.
20379 (unroll_loop_runtime_iterations): Update.
20380 * lra-constraints.c (lra_inheritance): Update.
20381 * lto-streamer-in.c (input_cfg): Update.
20382 * lto-streamer-out.c (output_cfg): Update.
20383 * mcf.c (adjust_cfg_counts): Update.
20384 * modulo-sched.c (sms_schedule): Update.
20385 * omp-expand.c (expand_omp_for_init_counts): Update.
20386 (extract_omp_for_update_vars): Update.
20387 (expand_omp_ordered_sink): Update.
20388 (expand_omp_for_ordered_loops): Update.
20389 (expand_omp_for_generic): Update.
20390 (expand_omp_for_static_nochunk): Update.
20391 (expand_omp_for_static_chunk): Update.
20392 (expand_cilk_for): Update.
20393 (expand_omp_simd): Update.
20394 (expand_omp_taskloop_for_outer): Update.
20395 (expand_omp_taskloop_for_inner): Update.
20396 * omp-simd-clone.c (simd_clone_adjust): Update.
20397 * optabs.c (expand_doubleword_shift): Update.
20398 (expand_abs): Update.
20399 (emit_cmp_and_jump_insn_1): Update.
20400 (expand_compare_and_swap_loop): Update.
20401 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
20402 * predict.c (predictable_edge_p): Update.
20403 (edge_probability_reliable_p): Update.
20404 (set_even_probabilities): Update.
20405 (combine_predictions_for_insn): Update.
20406 (combine_predictions_for_bb): Update.
20407 (propagate_freq): Update.
20408 (estimate_bb_frequencies): Update.
20409 (force_edge_cold): Update.
20410 * profile-count.c (profile_count::dump): Add missing space into dump.
20411 (profile_count::debug): Add newline.
20412 (profile_count::differs_from_p): Explicitly convert to unsigned.
20413 (profile_count::stream_in): Update.
20414 (profile_probability::dump): New member function.
20415 (profile_probability::debug): New member function.
20416 (profile_probability::differs_from_p): New member function.
20417 (profile_probability::differs_lot_from_p): New member function.
20418 (profile_probability::stream_in): New member function.
20419 (profile_probability::stream_out): New member function.
20420 * profile-count.h (profile_count_quality): Rename to ...
20421 (profile_quality): ... this one.
20422 (profile_probability): New.
20423 (profile_count): Update.
20424 * profile.c (compute_branch_probabilities): Update.
20425 * recog.c (peep2_attempt): Update.
20426 * sched-ebb.c (schedule_ebbs): Update.
20427 * sched-rgn.c (find_single_block_region): Update.
20428 (compute_dom_prob_ps): Update.
20429 (schedule_region): Update.
20430 * sel-sched-ir.c (compute_succs_info): Update.
20431 * stmt.c (struct case_node): Update.
20432 (do_jump_if_equal): Update.
20433 (get_outgoing_edge_probs): Update.
20434 (conditional_probability): Update.
20435 (emit_case_dispatch_table): Update.
20436 (expand_case): Update.
20437 (expand_sjlj_dispatch_table): Update.
20438 (emit_case_nodes): Update.
20439 * targhooks.c: Update.
20440 * tracer.c (better_p): Update.
20441 (find_best_successor): Update.
20442 * trans-mem.c (expand_transaction): Update.
20443 * tree-call-cdce.c: Update.
20444 * tree-cfg.c (gimple_split_edge): Upate.
20445 (move_sese_region_to_fn): Upate.
20446 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
20447 * tree-eh.c (lower_resx): Upate.
20448 (cleanup_empty_eh_move_lp): Upate.
20449 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20450 * tree-inline.c (copy_edges_for_bb): Update.
20451 (copy_cfg_body): Update.
20452 * tree-parloops.c (gen_parallel_loop): Update.
20453 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
20454 (gimple_gen_time_profiler): Update.
20455 * tree-ssa-dce.c (remove_dead_stmt): Update.
20456 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20457 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
20458 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
20459 (unloop_loops): Update.
20460 (try_peel_loop): Update.
20461 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20462 * tree-ssa-loop-split.c (connect_loops): Update.
20463 (split_loop): Update.
20464 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20465 (hoist_guard): Update.
20466 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
20467 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
20468 (value_replacement): Update.
20469 * tree-ssa-reassoc.c (branch_fixup): Update.
20470 * tree-ssa-tail-merge.c (replace_block_by): Update.
20471 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
20472 (create_edge_and_update_destination_phis): Update.
20473 (compute_path_counts): Update.
20474 (recompute_probabilities): Update.
20475 (update_joiner_offpath_counts): Update.
20476 (freqs_to_counts_path): Update.
20477 (duplicate_thread_path): Update.
20478 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
20479 (struct switch_conv_info): Update.
20480 (gen_inbound_check): Update.
20481 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20482 (vect_do_peeling): Update.
20483 (vect_loop_versioning): Update.
20484 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20485 (optimize_mask_stores): Update.
20486 * ubsan.c (ubsan_expand_null_ifn): Update.
20487 * value-prof.c (gimple_divmod_fixed_value): Update.
20488 (gimple_divmod_fixed_value_transform): Update.
20489 (gimple_mod_pow2): Update.
20490 (gimple_mod_pow2_value_transform): Update.
20491 (gimple_mod_subtract): Update.
20492 (gimple_mod_subtract_transform): Update.
20493 (gimple_ic): Update.
20494 (gimple_stringop_fixed_value): Update.
20495 (gimple_stringops_transform): Update.
20496 * value-prof.h: Update.
20498 2017-06-29 Carl Love <cel@us.ibm.com>
20500 * config/rs6000/rs6000-c.c: Add support for built-in functions
20501 vector signed int vec_signed (vector float);
20502 vector signed long long vec_signed (vector double);
20503 vector signed int vec_signed2 (vector double, vector double);
20504 vector signed int vec_signede (vector double);
20505 vector signed int vec_signedo (vector double);
20506 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
20507 instruction generator.
20508 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20509 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
20510 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
20512 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
20513 vunsignede_v2df): Add define_expands.
20514 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
20515 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
20516 VEC_UNSIGNEDO): Add definitions.
20517 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20518 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
20519 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
20520 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
20521 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
20522 * config/rs6000/altivec.h (vec_signed, vec_signed2,
20523 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
20524 vec_unsignede, vec_unsignedo): Add builtin defines.
20525 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
20527 * doc/extend.texi: Update the built-in documentation file for the
20528 new built-in functions.
20530 2017-06-29 Richard Biener <rguenther@suse.de>
20532 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
20533 reduction chains to LOOP_VINFO_REDUCTIONS.
20534 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
20535 SLP reductions after processing reduction chains.
20537 2017-06-29 Nathan Sidwell <nathan@acm.org>
20539 * builtins.c (fold_builtin_FUNCTION): Use
20540 lang_hooks.decl_printable_name.
20542 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
20544 PR middle-end/81194
20545 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
20546 with only one label.
20547 * stmt.c (expand_case): Assert NCASES is greater than one.
20549 2017-06-29 Richard Biener <rguenther@suse.de>
20551 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
20553 (group_case_labels): Likewise.
20554 (find_taken_edge): Push sanity checking on val to workers...
20555 (find_taken_edge_cond_expr): ... here
20556 (find_taken_edge_switch_expr): ... and here, handle cases
20557 with just a default label.
20558 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
20559 (group_case_labels): Likewise.
20560 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
20561 group_case_labels does anything cleanup the CFG again.
20563 2017-06-29 Bin Cheng <bin.cheng@arm.com>
20565 PR tree-optimization/81196
20566 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
20567 exit condition comparing two IVs.
20569 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
20571 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
20572 profile to the dummy entry at the end of the list of architectures.
20573 * config/arm/arm-cpu-cdata.h: Regenerated.
20575 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20576 Michael Collison <michael.collison@arm.com>
20579 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
20581 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
20582 (*aarch64_reg_<mode>3_minus_mask): New pattern.
20583 (*aarch64_<optab>_reg_di3_mask2): New pattern.
20584 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
20585 of shift when the shift amount is masked with constant equal to
20586 the size of the mode.
20587 * config/aarch64/predicates.md (subreg_lowpart_operator): New
20590 2017-06-29 Martin Liska <mliska@suse.cz>
20592 * config/i386/i386.opt: Change range from [1,5] to [0,5].
20594 2017-06-29 Yury Gribov <tetra2005@gmail.com>
20597 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
20600 (edge_growth_cache_entry::edge_growth_cache_entry): New
20602 (reset_edge_growth_cache): Update to use constructor.
20604 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20606 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
20607 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20608 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20610 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
20612 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
20613 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
20615 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
20617 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
20618 (*-linux-uclibc*): Add t-uclibc tmake_file.
20619 * config/t-musl: New.
20620 * config/t-uclibc: New.
20622 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
20624 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
20626 (gen_comm_data): Emit architectural setting of arch_prof.
20627 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
20629 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
20630 (armv8-m.base, armv8-m.main): Likewise.
20631 * arm-protos.h (arm_build_target): Add profile field.
20632 (arch_option): Likewise.
20633 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
20635 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
20636 arm_active_target.profile.
20638 2017-06-28 Richard Biener <rguenther@suse.de>
20640 PR middle-end/81227
20641 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
20642 TYPE_OVERFLOW_WRAPS.
20643 * match.pd (negate_expr_p): Likewise.
20644 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
20645 fold_build2, not fold_binary.
20647 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20649 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20650 Convert memory address to Pmode.
20651 (aarch64_print_operand): Assert MEM operands are always Pmode.
20653 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20656 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20657 Remove redundant if.
20658 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20659 * config/arm/aarch-common-protos.h
20660 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20661 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20663 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20666 * multiple_target.c (create_dispatcher_calls): Set the default
20667 clone to be static, not public.
20669 2017-06-28 Richard Biener <rguenther@suse.de>
20671 * tree-vect-loop.c (vectorizable_reduction): Move special
20672 cond reduction IV var creation ...
20673 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20674 parameter. Use STMT_VINFO_VECTYPE.
20675 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20678 2017-06-28 Martin Liska <mliska@suse.cz>
20681 * ipa-visibility.c (non_local_p): Handle visibility.
20683 2017-06-28 Martin Liska <mliska@suse.cz>
20686 * common.opt: Add IntegerRange to various options.
20687 * opt-functions.awk (integer_range_info): New function.
20688 * optc-gen.awk: Add integer_range_info to cl_options struct.
20689 * opts-common.c (decode_cmdline_option): Handle
20690 CL_ERR_INT_RANGE_ARG.
20691 (cmdline_handle_error): Likewise.
20692 * opts.c (print_filtered_help): Show valid interval in
20693 when --help is provided.
20694 * opts.h (struct cl_option): Add range_min and range_max fields.
20695 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20697 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20699 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20700 (x * C EQ/NE y * C): New transformation.
20702 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20704 * genmultilib (combination_space): Accept '+' in option names.
20706 2017-06-28 Martin Liska <mliska@suse.cz>
20709 * asan.c (instrument_derefs): Bail out inner references
20710 that are hard register variables.
20712 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20715 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20716 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20718 2017-06-28 Richard Biener <rguenther@suse.de>
20720 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20721 (vect_get_slp_defs): Adjust.
20722 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20724 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20726 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20727 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20728 (vectorizable_reduction): Adjust.
20729 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20731 (vect_get_slp_defs): Likewise.
20732 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20733 (vectorizable_bswap): Adjust.
20734 (vectorizable_call): Likewise.
20735 (vectorizable_conversion): Likewise.
20736 (vectorizable_assignment): Likewise.
20737 (vectorizable_shift): Likewise.
20738 (vectorizable_operation): Likewise.
20739 (vectorizable_store): Likewise.
20740 (vectorizable_condition): Likewise.
20741 (vectorizable_comparison): Likewise.
20743 2017-06-28 Michael Collison <michael.collison@arm.com>
20746 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20748 (arm_gen_movmemqi): Removed unused variable 'i'.
20749 Convert 'for' loop into 'while' loop.
20750 (arm_expand_prologue): Remove last unnecessary set of insn.
20751 (thumb_pop): Remove unused variable 'pushed_words'.
20752 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20754 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20756 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20757 * config/s390/s390-protos.h: Add prototype of
20758 s390_rel_address_ok_p.
20759 * config/s390/s390.c (s390_got_symbol): New function.
20760 (s390_rel_address_ok_p): New function.
20761 (legitimize_pic_address): Use s390_rel_address_ok_p.
20762 (s390_load_got): Use s390_got_symbol.
20763 (s390_option_override): Issue error if
20764 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20765 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20767 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20769 2017-06-27 Andrew Pinski <apinski@cavium.com>
20771 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20772 (X * copysign (1.0, X)): New pattern.
20773 (X * copysign (1.0, -X)): New pattern.
20774 (copysign (-1.0, CST)): New pattern.
20776 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20778 * genmultilib (combination_space): Remove variable.
20779 Validate reuse rules against regular expression for any sequence
20780 of multilib options in any order.
20782 2017-06-27 Michael Collison <michael.collison@arm.com>
20784 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20785 call aarch64_split_simd_combine.
20786 * (aarch64_combine_internal<mode>): Delete pattern.
20787 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20788 Allow register and subreg operands.
20790 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20792 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20793 specific need, just fallback on defaults.
20794 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20796 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20797 Olivier Hainque <hainque@adacore.com>
20799 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20801 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20802 targets. Pick a default if no particular attempt applied.
20803 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20806 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20808 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20809 (x86_64-wrs-vxworks7): Likewise.
20811 2017-06-27 Marek Polacek <polacek@redhat.com>
20814 * ubsan.c (instrument_null): Check get_base_address's result for null.
20816 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20818 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20820 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20822 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20823 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20824 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20825 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20826 New function types.
20827 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20828 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20829 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20830 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20831 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20832 BUILT_IN_FEUPDATEENV): New builtins.
20833 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20834 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20835 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20836 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20838 (builtin_structptr_types): Adjust size.
20839 * tree.c (builtin_structptr_types): Add four entries.
20841 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20842 Olivier Hainque <hainque@adacore.com>
20844 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20845 (TLS_SYM): New local macro, forcing reference to __tls__ on
20846 link command lines for VxWorks 7 RTPs, triggering initialization
20848 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20849 OS features TLS support, true for RTPs on VxWorks 7.
20850 * config/vxworks.c (vxworks_override_options): Setup emutls
20853 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20855 * predict.c (test_prediction_value_range): Use -1U instead of -1
20856 to avoid narrowing conversion warning.
20857 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20858 to avoid narrowing conversion warning.
20859 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20861 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20863 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20865 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20866 64bit configurations.
20867 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20868 (SIZE_TYPE): Likewise.
20869 * config/vxworks.c (vxworks_emutls_var_fields): Use
20870 long_unsigned_type_node instead of unsigned_type_node as the offset
20871 field type, which is "pointer" mode in emutls.c.
20873 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20876 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20878 PR middle-end/81207
20879 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20880 gimple_vuse copying separately from gimple_vdef copying.
20882 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20884 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20887 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20888 Olivier Hainque <hainque@adacore.com>
20890 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20891 for all vxworks7 targets.
20892 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20893 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20894 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20895 variations for VX6/VX7 and 32/64bits later on in ...
20896 (VXWORKS_LIB_SPEC): Leverage new macros.
20897 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20898 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20900 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20902 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20903 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20905 2017-06-26 Carl Love <cel@us.ibm.com>
20907 * config/rs6000/rs6000-c.c: Add support for built-in functions
20908 vector bool char vec_reve (vector bool char);
20909 vector signed char vec_reve (vector signed char);
20910 vector unsigned char vec_reve (vector unsigned char);
20911 vector bool int vec_reve (vector bool int);
20912 vector signed int vec_reve (vector signed int);
20913 vector unsigned int vec_reve (vector unsigned int);
20914 vector bool long long vec_reve (vector bool long long);
20915 vector signed long long vec_reve (vector signed long long);
20916 vector unsigned long long vec_reve (vector unsigned long long);
20917 vector bool short vec_reve (vector bool short);
20918 vector signed short vec_reve (vector signed short);
20919 vector double vec_reve (vector double);
20920 vector float vec_reve (vector float);
20921 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20922 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20923 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20924 (altivec_vreve): New pattern.
20925 * config/rs6000/altivec.h (vec_reve): New define.
20926 * doc/extend.texi (vec_rev): Update the built-in documentation file
20927 for the new built-in functions.
20929 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20931 PR tree-optimization/71815
20932 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20934 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20936 (slsr_process_phi): Likewise.
20937 (replace_uncond_cands_and_profitable_phis): Don't replace a
20938 multiply candidate with a stride of 1 (copy or cast).
20939 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20941 (lowest_cost_path): Likewise.
20942 (total_savings): Likewise.
20944 2017-06-26 Richard Biener <rguenther@suse.de>
20947 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20948 Use def_builtin_pure for all gather builtins.
20950 2017-06-26 Richard Biener <rguenther@suse.de>
20952 PR tree-optimization/81203
20953 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20954 non-dominating BBs.
20956 2017-06-26 Marek Polacek <polacek@redhat.com>
20959 * doc/invoke.texi: Document -Wmultistatement-macros.
20961 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20963 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20964 arm_neon_ok_no_float_abi effective target.
20966 2017-06-26 Richard Biener <rguenther@suse.de>
20968 PR tree-optimization/80928
20969 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20970 (copy_bbs): Set BB_DUPLICATED flag early.
20971 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20973 (execute_on_shrinking_pred): Likewise.
20974 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20975 BB_DUPLICATED blocks.
20976 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20977 iterate over all PHIs considering removal of *gsi.
20979 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20981 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20984 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20986 * config/rs6000/rs6000-string.c: (expand_block_clear,
20987 do_load_for_compare, select_block_compare_mode,
20988 compute_current_alignment, expand_block_compare,
20989 expand_strncmp_align_check, expand_strn_compare,
20990 expand_block_move, rs6000_output_load_multiple)
20991 Move functions related to string/block move/compare
20992 to a separate file.
20993 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20994 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20995 for this function which is now used in two files.
20996 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20997 * config.gcc: Add rs6000-string.o to extra_objs for
20998 targets powerpc*-*-* and rs6000*-*-*.
21000 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
21003 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
21004 32-bit, since indexed is not valid for DImode.
21005 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
21006 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
21007 (define_peephole2 for Altivec d-form load): Add 32-bit support.
21008 (define_peephole2 for Altivec d-form store): Likewise.
21011 * multiple_target.c (create_dispatcher_calls): Only create the
21012 dispatcher call if the function is the default clone of a
21013 versioned function.
21015 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
21017 PR middle-end/80902
21018 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
21019 a call, force the call to not be a tail call.
21021 2017-06-23 Jeff Law <law@redhat.com>
21023 * doc/contrib.texi: Add entry for Steven Pemberton's work on
21026 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
21028 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
21029 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
21030 handling for early expansion of vector shifts (sl,sr,sra,rl).
21031 (builtin_function_type): Add vector shift right instructions
21032 to the unsigned argument list.
21034 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
21036 rtl-optimizatoin/79286
21037 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
21038 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
21039 trap. PIC register plus a const unspec without offset can never trap.
21041 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
21043 * tree.h (builtin_structptr_type): New type.
21044 (builtin_structptr_types): Declare new array.
21045 * tree.c (builtin_structptr_types): New array.
21046 (free_lang_data, build_common_tree_nodes): Use it.
21048 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
21051 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
21054 2017-06-22 Matt Turner <mattst88@gmail.com>
21056 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
21057 Lake models to skylake case. Assume skylake for unknown
21058 models with clflushopt.
21060 2017-06-22 Jeff Law <law@redhat.com>
21062 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
21063 frame sizes that do not satisfy aarch64_uimm12_shift.
21065 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
21067 * profile-count.h (apply_probability,
21068 apply_scale, probability_in): Fix checks for zero.
21070 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21072 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
21073 * doc/cppdiropts.texi (-I @var{dir}): Document it.
21075 2016-06-22 Richard Biener <rguenther@suse.de>
21077 * tree-vect-loop.c (vect_model_reduction_cost): Handle
21078 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
21079 REDUC_MAX_EXPR support.
21080 (vectorizable_reduction): Likewise.
21081 (vect_create_epilog_for_reduction): Likewise.
21083 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
21085 * match.pd (A / (1 << B) -> A >> B): New.
21086 * generic-match-head.c: Include optabs-tree.h.
21087 * gimple-match-head.c: Likewise.
21088 * optabs-tree.h (target_supports_op_p): New.
21089 * optabs-tree.c (target_supports_op_p): New.
21091 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21093 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
21094 $gcc_cv_ld --help output.
21095 (gcc_cv_ld_demangle): Likewise.
21096 (gcc_cv_ld_eh_frame_hdr): Likewise.
21097 (gcc_cv_ld_pie): Likewise.
21098 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
21099 (gcc_cv_ld_buildid): Likewise.
21100 (gcc_cv_ld_sysroot): Likewise.
21101 (ld_bndplt_support): Likewise.
21102 (ld_pushpopstate_support): Likewise.
21103 * configure: Regenerate.
21104 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
21106 2017-06-21 Jakub Jelinek <jakub@redhat.com>
21109 * config/i386/sse.md (round<mode>2): Renumber match_dup and
21110 operands indexes to avoid gap between operands and match_dups.
21112 2017-06-21 Andrew Pinski <apinski@cavium.com>
21114 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
21115 Increment Arith_shift and Arith_shift_reg by 1.
21116 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
21118 * config/aarch64/aarch64.c (thunderx_tunings): Enable
21119 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
21120 (aarch64_strip_extend): Add new argument and test for it.
21121 (aarch64_cheap_mult_shift_p): New function.
21122 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
21123 add a cost if it is true.
21124 Update calls to aarch64_strip_extend.
21125 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
21127 2017-06-21 Andrew Pinski <apinski@cavium.com>
21129 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
21131 (thunderxt88): Likewise.
21132 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
21133 (thunderx_prefetch_tune): New variable.
21134 (thunderx2t99_prefetch_tune): Update for the correct values.
21135 (thunderxt88_tunings): New variable.
21136 (thunderx_tunings): Use thunderx_prefetch_tune instead of
21137 generic_prefetch_tune.
21138 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
21140 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21142 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
21143 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
21144 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
21145 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
21146 (aarch64_atomic_cas<mode>, GPI): Likewise.
21148 2017-06-21 Martin Liska <mliska@suse.cz>
21150 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
21151 statements on cold and hot labels.
21152 * predict.c (tree_estimate_probability_bb): Remove the
21153 prediction from this place.
21155 2017-06-21 Martin Liska <mliska@suse.cz>
21157 PR tree-optimization/79489
21158 * gimplify.c (maybe_add_early_return_predict_stmt): New
21160 (gimplify_return_expr): Call the function.
21161 * predict.c (tree_estimate_probability_bb): Remove handling
21163 * predict.def: Update comment about early return predictor.
21164 * gimple-predict.h (is_gimple_predict): New function.
21165 * predict.def: Change default value of early return to 66.
21166 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
21168 * passes.def: Put pass_strip_predict_hints to the beginning of
21171 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
21173 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
21174 FUNCTION_DECL declarations.
21175 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
21177 (dwaf2out_decl): Likewise.
21178 * godump.c (go_early_global_decl): Skip call to the real debug hook
21179 for FUNCTION_DECL declarations.
21180 * passes.c (rest_of_decl_compilation): Skip call to the
21181 early_global_decl debug hook for FUNCTION_DECL declarations, unless
21182 -fdump-go-spec is passed.
21184 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
21186 * config/i386/i386.c (struct builtin_isa): New field pure_p.
21187 Reorder for compactness.
21188 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
21189 (def_builtin_pure, def_builtin_pure2): New functions.
21190 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
21192 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
21194 * match.pd (nop_convert): New predicate.
21195 ((A +- CST1) +- CST2): Allow some NOP conversions.
21197 2017-06-21 Jakub Jelinek <jakub@redhat.com>
21200 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
21201 with ctors/dtors if GOVD_SHARED is set.
21203 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
21205 * config/aarch64/aarch64.md (movti_aarch64):
21206 Emit mov rather than orr.
21207 (movtf_aarch64): Likewise.
21208 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
21209 Emit mov rather than orr.
21211 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
21213 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
21214 Swap alternatives, make integer dup more expensive.
21216 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
21218 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
21219 Return true for non-tls symbols.
21221 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
21223 * config/aarch64/aarch64-cores.def (cortex-a55): New.
21224 (cortex-a75): Likewise.
21225 (cortex-a75.cortex-a55): Likewise.
21226 * config/aarch64/aarch64-tune.md: Regenerate.
21227 * doc/invoke.texi (-mtune): Document new values for -mtune.
21229 2017-06-21 Tom de Vries <tom@codesourcery.com>
21231 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
21232 stack_size feature.
21233 (Effective-Target Keywords, Other attributes): Suggest using
21234 dg-add-options stack_size feature to get stack limit in stack_size
21235 effective target documentation.
21237 2017-06-21 Julian Brown <julian@codesourcery.com>
21238 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21240 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
21241 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
21242 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
21244 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
21245 attribute type list for neon_multiply.
21246 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
21247 attribute type list for neon_multiply.
21248 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
21249 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
21250 attribute type list for neon_multiply.
21251 * config/arm/types.md (crypto_pmull): Add.
21252 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
21253 attribute type list.
21255 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
21257 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
21260 2017-06-20 Jakub Jelinek <jakub@redhat.com>
21262 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
21263 to make sure not to dereference a NULL cost_classes_ptr pointer.
21265 2017-06-20 Carl Love <cel@us.ibm.com>
21267 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21268 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21269 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21270 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21271 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21272 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21273 VMULOSW): New enum "unspec" values.
21274 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21275 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21276 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21277 altivec_vmulosw): New patterns.
21278 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21279 VMULOSW): Add definitions.
21281 2017-06-20 Julia Koval <julia.koval@intel.com>
21283 * config/i386/i386.c: Fix rounding expand for new pattern.
21284 * config/i386/subst.md: Fix pattern (parallel -> unspec).
21286 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21288 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
21289 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
21291 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21293 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
21296 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21298 * config/aarch64/aarch64-cores.def: Rearrange to sort by
21299 architecture, then by implementer ID.
21300 * config/aarch64/aarch64-tune.md: Regenerate.
21302 2017-06-20 Richard Biener <rguenther@suse.de>
21304 PR middle-end/81097
21305 * fold-const.c (split_tree): Fold to type before negating.
21307 2017-06-20 David Malcolm <dmalcolm@redhat.com>
21309 * diagnostic-show-locus.c
21310 (selftest::test_fixit_deletion_affecting_newline): New function.
21311 (selftest::diagnostic_show_locus_c_tests): Call it.
21313 2017-06-20 Andreas Schwab <schwab@suse.de>
21316 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
21319 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21321 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
21322 __ARM_FEATURE_COPROC according to support.
21324 2017-06-20 Jakub Jelinek <jakub@redhat.com>
21326 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21327 Rewritten to avoid overflow for > 32-bit pointers.
21330 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
21331 by removing enum keyword.
21332 (ubsan_type_descriptor): Likewise. Formatting fix.
21335 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
21336 splitter): Require TARGET_SSE2 in the condition.
21338 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
21341 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
21342 for doing vector set of SFmode on ISA 3.0.
21343 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
21344 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
21346 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
21347 SFmode value into a V4SF variable that was extracted from another
21348 V4SF variable without converting the element to double precision
21349 and back to single precision vector format.
21350 (vsx_insert_extract_v4sf_p9_2): Likewise.
21352 2017-06-19 Jakub Jelinek <jakub@redhat.com>
21354 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
21355 in UWHI to avoid undefined overflow.
21358 * ubsan.h (enum ubsan_encode_value_phase): New.
21359 (ubsan_encode_value): Change second argument to
21360 enum ubsan_encode_value_phase with default value of
21361 UBSAN_ENCODE_VALUE_GENERIC.
21362 * ubsan.c (ubsan_encode_value): Change second argument to
21363 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
21364 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
21365 create_tmp_var_raw instead of create_tmp_var and use a
21367 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
21368 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
21369 ubsan_encode_value callers.
21372 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
21373 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
21374 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
21376 2017-06-19 Richard Biener <rguenther@suse.de>
21378 PR middle-end/81118
21379 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
21380 estimates if we changed anything.
21382 2017-06-19 Richard Biener <rguenther@suse.de>
21384 PR tree-optimization/80887
21385 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
21386 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
21387 simplified lookups, then reset mprts_hook.
21388 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
21390 (try_to_simplify): Likewise.
21392 2017-06-19 Martin Liska <mliska@suse.cz>
21395 * gimplify.c (gimplify_switch_expr):
21396 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
21398 2017-06-19 Martin Liska <mliska@suse.cz>
21400 * doc/install.texi: Document that PGO runs in 4 stages.
21402 2017-06-19 Martin Liska <mliska@suse.cz>
21405 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
21406 to dispatcher function name.
21407 * multiple_target.c (replace_function_decl): New function.
21408 (create_dispatcher_calls): Redirect both edges and references.
21410 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
21412 * profile-count.c (profile_count::dump): Dump quality.
21413 (profile_count::differs_from_p): Update for unsigned val.
21414 * profile-count.h (profile_count_quality): New enum.
21415 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
21417 2017-06-19 Richard Biener <rguenther@suse.de>
21419 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
21420 struct function as arg.
21421 (estimate_numbers_of_iterations): Export overload with loop arg.
21422 (free_numbers_of_iterations_estimates_loop): Use an overload of
21423 free_numbers_of_iterations_estimates instead.
21424 * tree-cfg.c (remove_bb): Adjust.
21425 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
21426 * tree-parloops.c (gen_parallel_loop): Likewise.
21427 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
21429 (tree_unroll_loops_completely): Likewise.
21430 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
21431 Use an overload instead and export.
21432 (estimated_loop_iterations): Adjust.
21433 (max_loop_iterations): Likewise.
21434 (likely_max_loop_iterations): Likewise.
21435 (estimate_numbers_of_iterations): Take struct function as arg
21437 (loop_exits_before_overflow): Adjust.
21438 (free_numbers_of_iterations_estimates_loop): Use an overload.
21439 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
21440 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
21442 2017-06-19 Richard Biener <rguenther@suse.de>
21445 * ipa-prop.c (find_constructor_constant_at_offset): Handle
21446 RANGE_EXPR conservatively.
21448 2017-06-16 Carl Love <cel@us.ibm.com>
21450 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21451 definitions for vec_float, vec_float2, vec_floato,
21452 vec_floate built-ins.
21453 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
21454 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
21456 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
21457 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
21458 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
21459 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
21460 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
21461 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
21462 vec_floato): Add builtin defines.
21463 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
21464 Update the built-in documentation file for the new built-in
21467 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21469 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
21470 (mthumb): Mark as the negative of -marm.
21472 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21474 * doc/invoke.texi (ARM Options, -mcpu): Document supported
21476 (ARM Options, -mtune): Document that this accepts the same
21477 extension options as -mcpu.
21478 (ARM Options, -mfpu): Document addition of -mfpu=auto.
21480 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21482 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
21483 permitted extensions.
21485 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21487 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
21488 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
21489 (armv8-m.main): Add option +nodsp.
21490 * config/arm/arm-cpu-cdata.h: Regenerated.
21492 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21494 * config/arm/t-fuchsia: New file.
21495 * config.gcc (arm*-*-fuchsia*): Use it.
21497 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21499 * config/arm/t-symbian: Rewrite for new option infrastructure.
21501 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21503 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
21504 (MULTILIB_REQUIRED): Likewise.
21506 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21508 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
21509 (MULTILIB_RESUE): Likewise.
21510 (MULTILIB_MATCHES): Likewise.
21511 (MULTLIB_REQUIRED): Likewise.
21513 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21515 * config/arm/t-rtems: Rewrite for new option framework.
21517 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21519 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
21520 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
21521 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
21522 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
21523 * config/arm/t-multilib: ... here.
21524 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
21525 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
21526 armv7-a and armv8*-a when A-profile libraries have not been built.
21527 * config/arm/t-rmprofile: Rewrite.
21529 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21531 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
21532 with a backslash. Remove the backslash after substituting unescaped
21534 * doc/fragments.texi (MULTILIB_REUSE): Document it.
21536 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21538 * config.gcc: (arm*-*-*): When building a-profile libraries, force
21539 the driver to pass through the default setting of -mfloat-abi.
21540 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
21542 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
21543 (all_feat_combs): New rule.
21544 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
21546 * config/arm/t-aprofile: Rewrite.
21548 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21550 * config/arm/arm.h (FPUTYPE_AUTO): Define.
21551 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
21552 fpu is not specified by the user/command-line.
21553 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
21554 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
21555 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
21556 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
21557 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
21558 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
21560 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21562 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
21563 * config/arm/t-arm-elf: Rewritten.
21565 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21567 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
21568 have some floating-point instructions.
21569 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
21570 (TARGET_MAYBE_HARD_FLOAT): New macro.
21571 * config/arm/arm-builtins.c (arm_init_builtins): Use
21572 TARGET_MAYBE_HARD_FLOAT.
21573 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
21575 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21577 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
21578 (configargs.h): Include it.
21579 (arm_print_hint_for_fpu_option): New function.
21580 (arm_parse_fpu_option): New function.
21581 (candidate_extension): New class.
21582 (arm_canon_for_multilib): New function.
21583 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
21584 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
21585 (ARCH_CANONICAL_SPECS): New macro.
21586 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
21588 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21590 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
21591 are set after handling multilib fragments. Set target_cpu_default2
21594 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21596 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
21598 (arm*-*-*): Set target_cpu_default2 to a quoted string.
21599 * config/arm/parsecpu.awk (check_cpu): Validate any extension
21601 (check_arch): Likewise.
21602 * config/arm/arm.c (arm_configure_build_target): Handle
21603 TARGET_CPU_DEFAULT being a string constant. Scan any feature
21604 options in the default.
21606 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21608 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
21609 when an option is an alias of another.
21610 * config/arm/parsecpu.awk (optalias): New parser token.
21611 (gen_comm_data): Mark non-alias options as such. Emit entries
21612 for extension aliases.
21613 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
21614 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
21615 (armv6kz, armv6zk, armv6t2): Likewise.
21616 (armv7): Make vfpv3-d16 an alias.
21617 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
21619 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
21620 Sort in canonical order.
21621 (armv8-a): Sort in canonical order.
21622 (armv8.1-a, armv8.2-a): Likewise.
21623 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
21625 (cortex-a9): Sort in canonical order.
21626 * config/arm/arm.c (selftests.h): Include it.
21627 (arm_test_cpu_arch_data): New function.
21628 (arm_run_self_tests): New function.
21629 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
21630 (targetm): Move declaration to the end of the file.
21631 * arm-cpu-cdata.h: Regenerated.
21633 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21635 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
21636 call to target_mode_check describing the type of option passed.
21637 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
21638 (arm_target_thumb_only): Use arm_parse_arch_option_name or
21639 arm_parse_cpu_option_name to match parameters against list of
21641 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
21642 arm_arch_core_flags data structure.
21643 * config/arm/arm-cpu_cdata.h: Regenerated.
21645 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21647 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21649 (arm_print_hint_for_cpu_option): Likewise.
21650 (arm_print_hint_for_arch_option): Likewise.
21651 (arm_parse_cpu_option_name): Likewise.
21652 (arm_parse_arch_option_name): Likewise.
21653 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21654 of entries in the all_fpus list.
21655 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21656 (arm_parse_cpu_option_name): Declare.
21657 (arm_parse_arch_option_name): Declare.
21658 (arm_parse_option_features): Declare.
21659 (arm_intialize_isa): Declare.
21660 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21662 (gen_comm_data): ... here. Make definitions non-static.
21663 * config/arm/arm-cpu-data.h: Regenerated.
21664 * config/arm/arm-cpu-cdata.h: Regenerated.
21666 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21668 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21669 (cpu_arch_extension): New structure.
21670 (cpu_arch_option, arch_option, cpu_option): New structures.
21671 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21672 architecture types.
21673 (gen_data): Generate new format data tables.
21674 * config/arm/arm.c (cpu_tune): New structure.
21675 (cpu_option, processors): Delete.
21676 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21677 (arm_print_hint_for_cpu_option): ... this and ...
21678 (arm_print_hint_for_arch_option): ... this.
21679 (arm_parse_arch_cpu_name): Delete. Replace with ...
21680 (arm_parse_cpu_option_name): ... this and ...
21681 (arm_parse_arch_option_name): ... this.
21682 (arm_unrecognized_feature): Change type of target parameter to
21684 (arm_parse_arch_cpu_features): Delete. Replace with ...
21685 (arm_parse_option_features): ... this.
21686 (arm_configure_build_target): Rework to use new configuration data
21688 (arm_print_tune_info): Rework for new configuration data tables.
21689 * config/arm/arm-cpu-data.h: Regenerated.
21690 * config/arm/arm-cpu.h: Regenerated.
21692 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21694 * Makefile.in (OBJS): Move sbitmap.o from here ...
21695 (OBJS-libcommon): ... to here.
21697 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21699 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21700 (ISA_ALL_CRYPTO): New macro.
21701 (ISA_ALL_SIMD): New macro
21702 (ISA_ALL_FP): New macro.
21703 * config/arm/arm.c (fpu_bitlist): Update initializer.
21704 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21706 (arm9e): Add fpu. Add option for nofp
21707 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21708 (arm926ej-s, arm1026ej-s): Likewise.
21709 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21710 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21711 neon-fp16, neon-vfpv4, nofp and nosimd.
21712 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21713 (cortex-a8): Add fpu. Add option for nofp.
21714 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21715 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21716 (cortex-r4f): Add fpu.
21717 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21718 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21720 (cortex-r8): Likewise.
21721 (cortex-m4): Add fpu. Add option for nofp.
21722 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21723 (cortex-a17.cortex-a7): Likewise.
21724 (cortex-a32): Add fpu. Add options for crypto and nofp.
21725 (cortex-a35, cortex-a53): Likewise.
21726 (cortex-a57): Add fpu. Add option for crypto.
21727 (cortex-a72, cortex-a73): Likewise.
21728 (exynos-m1): Likewise.
21729 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21730 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21731 (cortex-m33): Add fpu. Add option for nofp.
21732 * config/arm/arm-cpu-cdata.h: Regenerated
21733 * config/arm/arm-cpu-data.h: Regenerated.
21735 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21737 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21738 (armv5te, armv5tej): Likewise.
21739 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21740 (armv7): Add options fp and vfpv3-d16.
21741 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21742 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21744 (armv7ve): Likewise.
21745 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21746 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21747 (armv8-a): Add nocrypto option.
21748 (armv8.1-a, armv8.2-a): Likewise.
21749 (armv8-m.main): add options fp, fp.dp and nofp.
21751 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21753 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21755 (armv8-a+crc): Delete.
21756 (armv8.1-a): Add options simd, crypto and nofp.
21757 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21758 (armv8.2-a+fp16): Delete.
21759 (armv8-m.main): Add option dsp.
21760 (armv8-m.main+dsp): Delete.
21761 (cortex-a8): Add fpu. Add nofp option.
21762 (cortex-a9): Add fpu. Add nofp and nosimd options.
21763 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21764 link to main cpu and architecture data structures.
21765 (gen_comm_data): Only put isa attributes from the main architecture
21767 (option): New statement for architecture and CPU entries.
21768 * arm.c (struct cpu_option): New structure.
21769 (struct processors): Add entry for options.
21770 (arm_unrecognized_feature): New function.
21771 (arm_parse_arch_cpu_name): Ignore any characters after the first
21773 (arm_parse_arch_cpu_feature): New function.
21774 (arm_configure_build_target): Separate out any CPU and architecture
21775 features and parse separately. Don't error out if -mfpu=auto is
21776 used with only an architecture string.
21777 (arm_print_asm_arch_directives): New function.
21778 (arm_file_start): Call it.
21779 * config/arm/arm-cpu-cdata.h: Regenerated.
21780 * config/arm/arm-cpu-data.h: Likewise.
21781 * config/arm/arm-tables.opt: Likewise.
21783 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21785 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21786 assembler when it is not -mfpu=auto.
21788 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21790 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21791 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21792 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21793 (ASM_CPU_SPEC): Rewrite.
21794 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21795 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21796 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21797 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21798 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21799 copied string is NUL-terminated. Also strip any characters prefixed
21801 (arm_rewrite_selected_arch): New function.
21802 (arm_rewrite_march): New function.
21804 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21806 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21807 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21808 (march, mcpu, mtune): Convert to string-based options.
21809 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21810 (arm_parse_arch_cpu_name): New function.
21811 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21812 identify selected architecture or CPU.
21813 (arm_option_save): New function.
21814 (TARGET_OPTION_SAVE): Redefine.
21815 (arm_option_restore): Restore string options.
21816 (arm_option_print): Print string options.
21818 2017-06-16 Martin Sebor <msebor@redhat.com>
21820 PR tree-optimization/80933
21821 PR tree-optimization/80934
21822 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21823 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21824 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21825 (gimple_fold_builtin): Call them.
21827 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21829 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21830 as unlikely; update profile.
21832 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21834 * predict.c (force_edge_cold): Handle declaring edges impossible
21837 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21839 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21841 (try_unroll_loop_completely): Fix reporting.
21843 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21845 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21847 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21850 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21851 if given a non-constant argument for an intrinsic which requires a
21854 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21856 * profile.c (compare_freqs): New function.
21857 (branch_prob): Sort edge list.
21858 (find_spanning_tree): Assume that the list is priority sorted.
21860 2017-06-16 Richard Biener <rguenther@suse.de>
21862 PR tree-optimization/81090
21863 * passes.def (pass_record_bounds): Remove.
21864 * tree-pass.h (make_pass_record_bounds): Likewise.
21865 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21866 make_pass_record_bounds): Likewise.
21867 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21868 not free niter estimates at the beginning but at the end.
21869 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21871 2017-06-16 Richard Biener <rguenther@suse.de>
21873 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21874 initializer to workaround ICE in host GCC 4.8.
21876 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21878 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21880 (clone_inlined_nodes): Update.
21882 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21884 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21885 prefetch settings, and enable prefetching by default at -O3.
21887 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21889 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21890 Set flag_prefetch_loop_arrays according to tuning data.
21892 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21894 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21895 New tune structure.
21896 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21897 [Unrelated to main purpose of the patch] Place the pointer field last
21898 to enable type checking errors when tune structure are wrongly merged.
21899 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21900 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21901 (thunderx2t99_prefetch_tune): New tune constants.
21902 (tune_params *_tunings): Update all tunings (no functional change).
21903 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21904 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21905 from tunings structures.
21907 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21910 * ubsan.c (instrument_null): Add T argument, use it instead
21911 of computing it based on IS_LHS.
21912 (instrument_object_size): Likewise.
21913 (pass_ubsan::execute): Adjust instrument_null and
21914 instrument_object_size callers to pass gimple_get_lhs or
21915 gimple_assign_rhs1 result to it. Use instrument_null instead of
21916 calling get_base_address and instrument_mem_ref. Handle
21917 aggregate call arguments for object-size sanitization.
21919 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21921 PR tree-optimization/81089
21922 * tree-vrp.c (is_masked_range_test): Validate operands of
21925 2017-06-15 Martin Sebor <msebor@redhat.com>
21928 * dumpfile.c (dump_register): Avoid calling memset to initialize
21929 a class with a default ctor.
21930 * gcc.c (struct compiler): Remove const qualification.
21931 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21932 * hash-table.h: Ditto.
21933 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21935 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21936 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21938 * params.h (struct param_info): Make struct members non-const.
21939 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21940 with default initialization.
21941 * vec.h (vec_copy_construct, vec_default_construct): New helper
21943 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21944 with vec_copy_construct.
21945 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21946 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21947 * doc/invoke.texi (-Wclass-memaccess): Document.
21949 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21951 * emit-rtl.h (is_leaf): Update comment about local
21952 register allocator.
21954 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21957 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21958 for a variable to have a section before checking if the section has a
21960 Set section to.persistent if persistent attribute is set.
21961 Warn if .persistent attribute is used on an automatic variable.
21963 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21965 PR rtl-optimization/80474
21966 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21968 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21970 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21973 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21975 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21976 (casesi): Emit bounds checking as RTL.
21977 (casesi_internal_mips16_<mode>): Remove bounds checking.
21979 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21981 * config/xtensa/xtensa.c (xtensa_option_override): Append
21982 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21983 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21984 xtensa_doloop_hooks): Define unconditionally.
21985 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21987 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21988 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21989 for it in xtensa_option_override.
21990 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21991 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21993 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21995 * doc/cppopts.texi: Document '-' special value to -MF.
21997 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21999 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
22000 (cortex_a53_fconst): Likewise.
22001 (cortex_a53_fpmul): Likewise.
22002 (cortex_a53_f_load_64): Likewise.
22003 (cortex_a53_f_load_many): Likewise.
22004 (cortex_a53_advsimd_alu): Likewise.
22005 (cortex_a53_advsimd_alu_q): Likewise.
22006 (cortex_a53_advsimd_mul): Likewise.
22007 (cortex_a53_advsimd_mul_q): Likewise.
22008 (fpmac bypass): Add new bypass for fpmac-fpmac case.
22009 Add missing fmul, r2f_cvt and fconst cases.
22011 2017-06-14 Richard Biener <rguenther@suse.de>
22013 PR middle-end/81088
22014 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
22016 (fold_binary_loc): When associating do not treat pre-existing
22017 TREE_OVERFLOW on literal constants as a reason to allow
22018 TREE_OVERFLOW on associated literal constants.
22020 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22022 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
22023 (MASK_FEATURES): New macro.
22024 * config/sparc/sparc.c (sparc_option_override): Remove the special
22025 handling of -mfpu and generalize it to all MASK_FEATURES switches.
22027 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22029 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
22030 a division of 0 if non-call exceptions are enabled.
22032 2017-06-14 Andrew Pinski <apinski@cavium.com>
22033 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22036 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
22037 Improve vector initialization code gen for only variable case.
22039 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
22041 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
22043 2017-06-14 Richard Biener <rguenther@suse.de>
22045 PR tree-optimization/81083
22046 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
22049 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22051 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
22052 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
22053 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
22054 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
22055 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
22056 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
22058 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22060 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
22061 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
22063 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22065 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
22067 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22069 * config/rs6000/t-rtems: Don't handle SPE.
22071 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22073 * config/rs6000/t-linux: Don't handle SPE.
22075 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22077 * config/rs6000/eabispe.h: Delete file.
22079 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22081 * config/rs6000/t-spe: Delete file.
22083 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22085 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
22086 (rs6000_legitimate_offset_address_p): Return false for anything in
22087 V2SImode or V2SFmode.
22089 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
22091 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
22092 except V2SF and V2SI. Rearrange the vector modes, and add comments.
22093 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
22095 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
22096 (rs6000_legitimate_offset_address_p): Ditto.
22097 (rs6000_emit_move): Ditto.
22098 (rs6000_init_builtins): Remove V4HI_type_node.
22100 2017-06-13 Martin Liska <mliska@suse.cz>
22103 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
22104 (gate_asan): Likewise.
22105 * asan.h (asan_no_sanitize_address_p): Remove the function.
22106 (sanitize_flags_p): New function.
22107 * builtins.def: Fix coding style.
22108 * common.opt: Use renamed enum value.
22109 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
22110 * doc/extend.texi: Document no_sanitize attribute.
22111 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
22112 to SANITIZE_UNDEFINED_NONDEFAULT.
22113 * gcc.c (sanitize_spec_function): Use the renamed enum value.
22114 * gimple-fold.c (optimize_atomic_compare_exchange_p):
22115 Use sanitize_flags_p.
22116 * gimplify.c (gimplify_function_tree): Likewise.
22117 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
22118 * opts.c (parse_no_sanitize_attribute): New function.
22119 (common_handle_option): Use renamed enum value.
22120 * opts.h (parse_no_sanitize_attribute): Declare.
22121 * tree.c (sanitize_flags_p): New function.
22122 * tree.h: Declared here.
22123 * tsan.c: Use sanitize_flags_p.
22124 * ubsan.c (ubsan_expand_null_ifn): Likewise.
22125 (instrument_mem_ref): Likewise.
22126 (instrument_bool_enum_load): Likewise.
22127 (do_ubsan_in_current_function): Remove the function.
22128 (pass_ubsan::execute): Use sanitize_flags_p.
22129 * ubsan.h: Remove do_ubsan_in_current_function
22130 * tree-cfg.c (print_no_sanitize_attr_value): New function.
22131 (dump_function_to_file): Use it here.
22133 2017-06-13 Martin Jambor <mjambor@suse.cz>
22135 PR tree-optimization/80803
22136 PR tree-optimization/81063
22137 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
22138 (propagate_subaccesses_across_link): Enqueue subtree whenever
22139 necessary instead of relying on the caller.
22141 2017-06-13 Martin Jambor <mjambor@suse.cz>
22143 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
22144 that have a first_link.
22145 (sort_and_splice_var_accesses): Do not check first_link before
22147 (subtree_mark_written_and_enqueue): Likewise.
22148 (propagate_all_subaccesses): Likewise and do not stop at first
22149 parent with a first_link.
22151 2017-06-13 Martin Jambor <mjambor@suse.cz>
22153 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
22156 2017-06-13 Yury Gribov <tetra2005@gmail.com>
22158 * match.pd: New pattern.
22160 2017-06-13 Yury Gribov <tetra2005@gmail.com>
22162 * tree-vrp.c (is_masked_range_test): New function.
22163 (register_edge_assert_for): Determine ranges for
22166 2017-06-13 Yury Gribov <tetra2005@gmail.com>
22168 PR tree-optimization/67328
22169 * fold-const.c (maskable_range_p): New function.
22170 (build_range_check): Generate bittests if possible.
22172 2017-06-13 Martin Liska <mliska@suse.cz>
22174 * gimple-pretty-print.c (dump_probability): Add new argument.
22175 (dump_edge_probability): Dump both probability and count.
22176 (dump_gimple_label): Likewise.
22177 (dump_gimple_bb_header): Likewise.
22179 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
22182 * config/avr/avr-devices.c: Fix indentation.
22183 * config/avr/gen-avr-mmcu-specs.c: Dito.
22185 2017-06-13 Richard Biener <rguenther@suse.de>
22187 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
22188 instead get vector type from stmt_info.
22189 (vectorizable_reduction): Adjust. Remove dead code.
22191 2017-06-13 Richard Biener <rguenther@suse.de>
22193 PR middle-end/81065
22194 * fold-const.c (extract_muldiv_1): Remove bogus distribution
22195 case of C * (x * C2 + C3).
22196 (fold_addr_of_array_ref_difference): Properly fold index difference.
22198 2017-06-12 David S. Miller <davem@davemloft.net>
22201 * config/sparc/sparc.md (return expander): Emit frame blockage if
22202 function uses alloca.
22204 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
22206 * combine.c (make_field_assignment): Check len rather than the mode
22207 precision when calling force_to_mode.
22209 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
22211 Support multilibs and devices that see flash in RAM address range.
22214 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
22215 (avr_mcu_t) <flash_pm_offset>: New field.
22216 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
22217 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
22218 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
22219 (AVR_TINY_PM_OFFSET): Remove macro.
22220 * config/avr/avr.opt (-mshort-calls): New option.
22221 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
22222 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
22223 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
22224 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
22225 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
22226 instead of avr_arch->have_jmp_call.
22227 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
22228 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
22229 avr_arch->flash_pm_offset to define.
22230 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
22231 new field flash_pm_offset. Add entry for avrxmega3.
22232 (avr_texinfo): Add entry for avrxmega3.
22233 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
22234 attiny212, attiny214,
22235 attiny412, attiny414, attiny416, attiny417,
22236 attiny814, attiny816, attiny817,
22237 attiny1614, attiny1616, attiny1617,
22238 attiny3214, attiny3216, attiny3217.
22239 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
22240 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
22241 (avr_print_operand_address) [AVR_TINY]: Same.
22242 (avr_asm_init_sections) <readonly_data_section>: Only patch
22243 callback if avr_arch->flash_pm_offset = 0.
22244 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
22245 for rodata if avr_arch->flash_pm_offset != 0.
22246 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
22247 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
22248 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
22249 (m_options): Append opt_rcall.
22250 (m_dirnames): Append dir_rcall.
22251 * config/avr/t-multilib: Regenerate.
22253 * configure.ac [target=avr]: Check whether avrxmega3 default
22254 linker description file works as needed.
22255 * configure: Regenerate.
22256 * doc/avr-mmcu.texi: Regenerate.
22257 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
22258 <__AVR_ARCH__>: Document avrxmega3 and 103.
22259 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
22260 <__AVR_SHORT_CALLS__>: Document it.
22261 <__AVR_PM_BASE_ADDRESS__>: Document it.
22262 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
22263 (AVR Variable Attributes) <progmem>: Document this is
22264 not needed for avrxmega3.
22265 (AVR Named Address Spaces) <__flash>: Dito.
22267 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
22269 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
22271 2017-06-12 Doug Rupp <rupp@adacore.com>
22273 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
22274 Append vxworks-stdint.h to the tm_file list.
22275 * config/vxworks-stdint.h: New file.
22277 2017-06-12 Martin Liska <mliska@suse.cz>
22279 PR tree-optimization/81041
22280 * tree-profile.c (gimple_gen_ic_func_profiler):
22281 Create an extra BB in profile-generate
22282 (gimple_gen_time_profiler): Likewise.
22284 2017-06-12 Jakub Jelinek <jakub@redhat.com>
22286 PR tree-optimization/81003
22287 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
22288 (update_range_test): Use it instead of force_gimple_operand_gsi.
22290 2017-06-12 Richard Biener <rguenther@suse.de>
22292 PR tree-optimization/81053
22293 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
22294 with backedge value not defined in loop. Simplify def stmt
22297 2017-06-11 Tom de Vries <tom@codesourcery.com>
22300 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
22302 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
22303 nvptx_cannot_force_const_mem.
22305 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22307 * opts.c (finish_options): Move test for flag_split_stack after
22308 it has been initialized.
22310 2017-06-11 Jason Merrill <jason@redhat.com>
22312 * tree.h (id_equal): New.
22313 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
22314 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
22315 instead of strcmp of IDENTIFIER_POINTER.
22317 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22319 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
22320 (mark_all_inlined_calls_cdtor): Fix formating.
22321 (inline_transform): Rescale profile before inlining.
22323 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22325 * cgraph.h (cgraph_edge::clone): Update prototype.
22326 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
22327 (cgraph_node::create_clone): Update.
22328 (cgraph_node::create_version_clone): Update.
22329 * tree-inline.c (copy_bb): Update.
22330 (expand_call_inline): Update.
22332 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22334 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
22335 factored out from ...
22336 (rs6000_emit_prologue): ... here.
22338 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22340 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
22341 factored out from ...
22342 (rs6000_emit_prologue): ... here.
22344 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22346 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
22348 (handle_missing_profiles): Fix computation of tp_first_run.
22349 (counts_to_freqs): Do not touch freqs when count is 0.
22351 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22353 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
22356 2017-06-10 Tom de Vries <tom@codesourcery.com>
22358 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22359 attributes): Document signal effective target.
22361 2017-06-10 Tom de Vries <tom@codesourcery.com>
22363 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22364 Document effective target stack_size.
22366 2017-06-09 David Malcolm <dmalcolm@redhat.com>
22368 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
22369 to the edit_context if they can be auto-applied.
22371 2017-06-9 Ian Lance Taylor <iant@golang.org>
22373 * opts.c (finish_options): If -fsplit-stack, disable implicit
22374 -forder-blocks-and-partition.
22375 * doc/invoke.texi (Optimize Options): Document that when using
22376 -fsplit-stack -forder-blocks-and-partition is not implicitly
22379 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22381 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
22382 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
22383 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
22384 * builtins.def (abort, trap, unreachable): Declare cold.
22385 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
22386 * tree-core.h (ECF_COLD): New.
22387 * tree.c (set_call_expr_flags): Handle ECF_COLD.
22388 (build_common_builtin_nodes): Mark unreachable and abort as cold.
22390 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22392 * predict.c (unlikely_executed_stmt_p): Cleanup.
22394 2017-06-09 Richard Biener <rguenther@suse.de>
22396 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
22397 model if the ref is always written to.
22399 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22401 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
22403 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22405 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
22408 2017-06-09 Tom de Vries <tom@codesourcery.com>
22411 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
22412 "target cannot support label values" when encountering LABEL_REF.
22414 2017-06-09 Martin Liska <mliska@suse.cz>
22416 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
22417 (gimple_gen_ic_func_profiler): Emit direct comparison
22418 of __gcov_indirect_call_callee with NULL.
22419 (gimple_gen_time_profiler): Change probability from
22420 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
22422 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22424 * profile.c (edge_gcov_counts): Turn to pointer.
22425 (compute_branch_probabilities, compute_branch_probabilities): Update.
22426 (branch_prob): Do not clear edge_gcov_count.
22427 * profile.h (edge_gcov_counts): Turn to pointer.
22428 (edge_gcov_count): Update.
22430 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22432 * gimple.h (gimple_check_failed): Mark cold.
22434 2017-06-09 Richard Biener <rguenther@suse.de>
22436 PR tree-optimization/66623
22437 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
22438 refactor check_reduction into two parts, properly computing
22439 whether we have to check reduction validity for outer loop
22442 2017-06-09 Richard Biener <rguenther@suse.de>
22444 PR tree-optimization/79483
22445 * graphite-scop-detection.c (order): New global.
22446 (get_order): Compute bb to order mapping that satisfies code
22447 generation constraints.
22448 (cmp_pbbs): New helper.
22449 (build_scops): Start domwalk at entry block, sort generated
22452 2017-06-09 Richard Biener <rguenther@suse.de>
22454 PR middle-end/81007
22455 * ipa-polymorphic-call.c
22456 (ipa_polymorphic_call_context::restrict_to_inner_class):
22457 Skip FIELD_DECLs with error_mark_node type.
22458 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
22461 2017-06-09 Martin Liska <mliska@suse.cz>
22463 * predict.c (struct branch_predictor): New struct.
22464 (test_prediction_value_range): New test.
22465 (predict_c_tests): New function.
22466 * selftest-run-tests.c (selftest::run_tests): Run the function.
22467 * selftest.h: Declare new tests.
22469 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
22472 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
22473 gen_add3_insn did not fail.
22474 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
22475 r0, construct that number in a temporary reg and add that reg to r0.
22476 If asked to put the result in r0 as well, fail.
22478 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
22480 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22481 for early expansion of vec_eqv.
22483 2017-06-08 Jakub Jelinek <jakub@redhat.com>
22485 PR middle-end/81005
22486 * ubsan.c (instrument_null): Avoid pointless code temporary.
22487 (pass_ubsan::execute): Instrument aggregate arguments of calls.
22489 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
22493 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
22496 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
22499 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22501 * predict.c (unlikely_executed_edge_p): Move ahead.
22502 (probably_never_executed_edge_p): Use it.
22504 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22506 PR middle-end/79988
22507 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
22508 gimple_call_builtin_p call.
22510 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22512 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
22513 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
22514 rtl_check_failed_type2, rtl_check_failed_code1,
22515 rtl_check_failed_code2, rtl_check_failed_code_mode,
22516 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
22517 rtvec_check_failed_bounds, rtl_check_failed_flag,
22518 _fatal_insn_not_found, _fatal_insn): Likewise.
22519 * tree.h (tree_contains_struct_check_failed,
22520 tree_check_failed, tree_not_check_failed,
22521 tree_class_check_failed, tree_range_check_failed,
22522 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
22523 tree_vec_elt_check_failed, phi_node_elt_check_failed,
22524 tree_operand_check_failed, omp_clause_check_failed,
22525 omp_clause_operand_check_failed, omp_clause_range_check_failed):
22528 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22530 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
22531 flag_branch_probabilities.
22532 * ipa-inline.c (edge_badness): Likewise.
22533 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
22534 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
22535 * predict.c (maybe_hot_frequency_p): Likewise.
22536 (probably_never_executed): Likewise.
22537 * sched-ebb.c (schedule_ebbs): Likewise.
22538 * sched-rgn.c (find_single_block_region): Likewise.
22539 * tracer.c (tail_duplicate): Likewise.
22541 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22543 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
22544 longer requires x_flag_profile_use.
22546 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22548 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
22549 instead of flag_reorder_blocks_and_partition.
22550 * dbxout.c (dbxout_function_end): Likewise.
22551 * dwarf2out.c (gen_subprogram_die): Likewise.
22552 * haifa-sched.c (sched_create_recovery_edges): Likewise.
22553 * hw-doloop.c (reorg_loops): Likewise.
22554 * varasm.c (assemble_start_function,
22555 assemble_end_function): Likewise.
22556 (decide_function_section): Do not check for
22557 flag_reorder_blocks_and_partition.
22559 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22561 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22563 (chkp_get_hard_register_fake_addr_expr): Ditto.
22564 (chkp_build_addr_expr): Add check for hard reg case.
22565 (chkp_parse_array_and_component_ref): Ditto.
22566 (chkp_find_bounds_1): Ditto.
22567 (chkp_process_stmt): Don't generate bounds store for
22570 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22572 * predict.c (maybe_hot_bb_p): Do not check profile status.
22573 (maybe_hot_edge_p): Likewise.
22574 (probably_never_executed): Check for zero counts even if profile
22576 (unlikely_executed_edge_p): New function.
22577 (unlikely_executed_stmt_p): New function.
22578 (unlikely_executed_bb_p): New function.
22579 (set_even_probabilities): Use unlikely predicates.
22580 (combine_predictions_for_bb): Likewise.
22581 (predict_paths_for_bb): Likewise.
22582 (predict_paths_leading_to_edge): Likewise.
22583 (determine_unlikely_bbs): New function.
22584 (estimate_bb_frequencies): Use it.
22585 (compute_function_frequency): Use zero counts even if profile is
22587 * profile-count.h: Fix typo.
22589 2017-08-08 Julia Koval <julia.koval@intel.com>
22591 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
22592 _mm512_mask_cvtsepi16_storeu_epi8,
22593 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
22594 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
22595 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
22596 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
22597 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
22598 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
22599 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
22600 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
22601 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
22602 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
22603 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
22604 __builtin_ia32_pmovuswb256mem_mask,
22605 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
22606 __builtin_ia32_pmovwb512mem_mask): New builtins.
22608 2017-08-08 Julia Koval <julia.koval@intel.com>
22610 PR target/73350,80862
22611 * config/i386/subst.md (round): Fix round pattern.
22612 * config/i386/i386.c (ix86_erase_embedded_rounding):
22613 Fix erasing rounding for the fixed pattern.
22615 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22617 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
22619 2017-06-08 Martin Liska <mliska@suse.cz>
22621 PR gcov-profile/80911
22622 * gcov.c (block_info::block_info): New constructor.
22624 2017-06-07 Carl Love <cel@us.ibm.com>
22626 * config/rs6000/rs6000-c: The return type of the following
22627 built-in functions was implemented as int not long long. Fix sign
22628 of return value for the unsigned version of vec_mulo and vec_mule.
22629 vector unsigned long long vec_bperm (vector unsigned long long,
22630 vector unsigned char)
22631 vector signed long long vec_mule (vector signed int,
22633 vector unsigned long long vec_mule (vector unsigned int,
22634 vector unsigned int)
22635 vector signed long long vec_mulo (vector signed int,
22637 vector unsigned long long vec_mulo (vector unsigned int,
22638 vector unsigned int)
22639 * doc/extend.texi: Fix the documentation for the built-in
22642 2017-06-07 Carl Love <cel@us.ibm.com>
22645 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
22648 2017-06-07 Carl Love <cel@us.ibm.com>
22650 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22651 support, Generate doublehv for signed int/float for BE case only.
22653 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22655 * doc/invoke.texi (mcx16): Rewrite.
22657 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22659 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22660 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22661 *mov<mode>_softfloat, and an anonymous splitter): Use
22662 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22664 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22666 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22668 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22669 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22670 (rs6000_debug_reg_global): Adjust.
22671 (rs6000_init_hard_regno_mode_ok): Adjust.
22672 (rs6000_dbx_register_number): Adjust.
22673 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22674 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22675 Remove SPE_ACC and SPEFSCR.
22676 (REG_ALLOC_ORDER): Ditto.
22677 (FRAME_POINTER_REGNUM): Change to 111.
22678 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22679 (REG_CLASS_NAMES): Ditto.
22680 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22681 (REGISTER_NAMES): Ditto.
22682 (ADDITIONAL_REG_NAMES): Ditto.
22683 (rs6000_reg_names): Ditto.
22684 * config/rs6000/rs6000.md: Renumber some register number
22687 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22689 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22691 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22692 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22694 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22695 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22696 Delete the SPE high registers.
22697 (REG_ALLOC_ORDER): Ditto.
22698 (enum reg_class): Remove SPE_HIGH_REGS.
22699 (REG_CLASS_NAMES): Ditto.
22700 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22701 (REGISTER_NAMES): Ditto.
22702 (rs6000_reg_names): Ditto.
22703 * doc/tm.texi.in: Remove SPE as example.
22704 * doc/tm.texi: Regenerate.
22706 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22708 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22709 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22710 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22711 * config/rs6000/rs6000.md (type): Remove "brinc".
22713 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22715 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22716 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22717 * config/rs6000/linuxspe.h: Delete file.
22718 * config/rs6000/rs6000.md: Don't include spe.md.
22719 * config/rs6000/spe.h: Delete file.
22720 * config/rs6000/spe.md: Delete file.
22721 * config/rs6000/t-rs6000: Remove spe.md.
22723 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22725 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22726 (reg_or_none500mem_operand): Delete.
22727 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22728 instead of reg_or_none500mem_operand.
22730 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22732 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22733 handling of SPE flags.
22734 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22736 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22738 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22740 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22741 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22742 paired_divv2sf3): Similar.
22743 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22744 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22745 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22747 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22748 Rename the paired_* instruction patterns.
22749 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22751 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22752 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22753 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22754 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22755 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22756 PAIRED_VECTOR_MODE.
22757 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22758 (spe_func_has_64bit_regs_p): Delete.
22759 (spe_expand_predicate_builtin): Delete.
22760 (spe_expand_evsel_builtin): Delete.
22761 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22762 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22763 (invalid_e500_subreg): Delete.
22764 (rs6000_legitimize_address): Always force_reg op2 as well, for
22765 paired single memory accesses.
22766 (rs6000_member_type_forces_blk): Delete.
22767 (rs6000_spe_function_arg): Delete.
22768 (rs6000_expand_unop_builtin): Delete SPE handling.
22769 (rs6000_expand_binop_builtin): Ditto.
22770 (spe_expand_stv_builtin): Delete.
22771 (bdesc_2arg_spe): Delete.
22772 (spe_expand_builtin): Delete.
22773 (spe_expand_predicate_builtin): Delete.
22774 (spe_expand_evsel_builtin): Delete.
22775 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22776 (spe_init_builtins): Delete.
22777 (spe_func_has_64bit_regs_p): Delete.
22778 (savres_routine_name): Delete "info" parameter. Adjust callers.
22779 (rs6000_emit_stack_reset): Ditto.
22780 (rs6000_dwarf_register_span): Delete.
22781 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22782 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22783 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22785 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22787 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22788 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22789 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22790 mulv2sf3, divv2sf3): Delete expanders.
22792 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22794 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22796 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22798 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22799 * config/rs6000/rs6000.c: Ditto.
22801 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22803 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22804 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22805 comparison_operator.
22807 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22809 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22810 * config/rs6000/rs6000.opt: Ditto.
22811 * config/rs6000/t-rtems: Ditto.
22813 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22815 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22816 TARGET_E500_SINGLE by 0, simplify.
22817 * config/rs6000/rs6000.c: Ditto.
22818 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22819 (spe_build_register_parallel): Delete.
22820 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22821 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22822 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22823 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22824 (E500_CONVERT): Delete.
22825 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22827 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22829 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22830 * config/rs6000/dfp.md: Ditto.
22831 (negdd2, *negdd2_fpr): Merge.
22832 (absdd2, *absdd2_fpr): Merge.
22833 (negtd2, *negtd2_fpr): Merge.
22834 (abstd2, *abstd2_fpr): Merge.
22835 * config/rs6000/e500.h: Delete file.
22836 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22837 TARGET_FPRS by 1 and simplify.
22838 * config/rs6000/rs6000-c.c: Ditto.
22839 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22840 TARGET_DF_SPE by 0.
22841 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22843 * config/rs6000/rs6000.md: Ditto.
22844 (floatdidf2, *floatdidf2_fpr): Merge.
22845 (move_from_CR_gt_bit): Delete.
22846 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22847 (E500_CR_IOR_COMPARE): Delete.
22848 (All patterns that require !TARGET_FPRS): Delete.
22849 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22851 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22853 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22855 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22857 * graphds.c (add_edge): Intitialize edge's attached data.
22858 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22859 pointer parameter. Call pointed function on each edge during
22860 graph traversing. Skip traversing the edge when the function
22862 (graphds_dfs, graphds_scc): Ditto.
22863 (for_each_edge): New parameter. Pass the new parameter to callback
22865 * graphds.h (skip_edge_callback): New function pointer type.
22866 (graphds_dfs, graphds_scc): New function pointer parameter.
22867 (graphds_edge_callback, for_each_edge): New parameter.
22869 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22871 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22872 out code checking if runtime alias check is possible to below ...
22873 Call the new function.
22874 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22875 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22877 2017-06-07 Marek Polacek <polacek@redhat.com>
22880 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22881 TYPE_OVERFLOW_WRAPS check.
22883 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22885 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22886 if versioning is required.
22887 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22888 peeling with the check for versioning.
22890 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22892 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22893 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22894 Set true to new parameter if new ssa variable is defined.
22895 (vect_gen_vector_loop_niters): Refactor. Set range information
22896 for the new vector loop bound variable.
22897 (vect_do_peeling): Ditto.
22899 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22901 * tree-affine.c (ssa.h): Include header file.
22902 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22903 has wrapping overflow behavior.
22905 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22907 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22909 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22911 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22912 (tree_to_aff_combination): ... here.
22914 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22916 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22917 reg_pressure model function.
22918 (ivopts_global_cost_for_size): Delete.
22919 (determine_set_costs, iv_ca_recount_cost): Call new model function
22920 ivopts_estimate_reg_pressure.
22922 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22924 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22925 expensive than udiv. Remove floating point cases from mod.
22927 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22929 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22930 Increase idiv cost.
22932 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22934 * config/aarch64/aarch64.md
22935 (copysignsf3): Fix mask generation.
22937 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22939 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22941 (class dump_manager): Add register_dumps method.
22942 * dumpfile.c: Include langhooks.h.
22943 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22944 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22945 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22946 (dump_manager::dump_register): Start with 512 entries instead of 32.
22947 (dump_manager::register_dumps): New method.
22948 * toplev.c (general_init): Instead of invoking register_dumps
22949 langhook, invoke register_dumps method on the dump manager.
22950 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22953 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22955 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22956 Say that # requires an associated define_split to exist, and that
22957 the define_split must be suitable for use after register allocation.
22959 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22961 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22962 (compute_outgoing_frequencies): Also initialize zero counts.
22963 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22964 around loops; preserve more of profile when nothing changes.
22966 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22968 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22970 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22971 * config/arm/arm-cpu-cdata.h: Regenerate.
22972 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22973 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22974 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22975 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22976 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22978 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22980 * config/arm/t-rmprofile: Likewise.
22981 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22983 2017-06-06 David S. Miller <davem@davemloft.net>
22986 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22987 blockage if function uses alloca.
22989 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22991 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22992 New "uid" fields to hold pretty-print IDs of group and ref.
22993 Memory references are now identified as <group_id>:<ref_id>
22994 instead of using [random] addresses.
22995 (dump_mem_details): Simplify, no functional change.
22996 (dump_mem_ref): Simplify and make output more concise.
22997 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22998 (find_or_create_group): Initialize group uid.
22999 (record_ref): Initialize ref uid. Improve debug output.
23000 (prune_group_by_reuse, should_issue_prefetch_p,)
23001 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
23002 (mark_nontemporal_store, determine_loop_nest_reuse):
23003 Improve debug output.
23005 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23007 * dbgcnt.def (prefetch): New debug counter.
23008 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
23009 (schedule_prefetches): Stop issueing prefetches if debug counter
23012 2017-06-06 Tom de Vries <tom@codesourcery.com>
23014 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
23015 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
23017 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23019 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
23020 Use aarch64_reg_or_zero predicate for operand 4.
23021 (aarch64_compare_and_swap<mode> define_insn_and_split):
23022 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
23023 (aarch64_store_exclusive<mode>): Likewise for operand 2.
23025 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23027 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
23028 (arm_compute_save_core_reg_mask): This.
23029 (thumb1_compute_save_reg_mask): Rename into ...
23030 (thumb1_compute_save_core_reg_mask): This.
23031 (arm_compute_save_reg0_reg12_mask): Adapt comment.
23032 (arm_compute_frame_layout): Likewise.
23034 2017-06-06 Richard Biener <rguenther@suse.de>
23036 PR tree-optimization/80974
23037 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
23038 keep or clear leaders SSA info.
23040 2017-06-06 Tom de Vries <tom@codesourcery.com>
23042 * config/nvptx/nvptx.c (split_mode_p): New function.
23043 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
23045 2017-06-06 Tom de Vries <tom@codesourcery.com>
23047 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
23049 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23052 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
23055 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23057 * shrink-wrap.c (handle_simple_exit): Update profile.
23058 (try_shrink_wrapping): Upate profile.
23060 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23062 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
23063 (tree_guess_outgoing_edge_probabilities): New.
23064 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
23065 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
23067 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23069 * ipa-split.c (split_function): Initialize return bb profile.
23071 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
23073 * profile.c (compute_branch_probabilities): Also initialize
23074 EXIT_BLOCK profile.
23076 2017-06-06 Richard Biener <rguenther@suse.de>
23078 PR tree-optimization/80928
23079 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
23080 (vect_analyze_loop_operations): Properly guard analysis for
23082 (vect_transform_loop): Likewise.
23083 (vect_analyze_loop_2): Also reset SLP type on PHIs.
23084 (vect_model_induction_cost): Do not cost for pure SLP.
23085 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
23086 of induction in inner loop vectorization.
23087 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
23088 (vect_get_and_check_slp_defs): Handle vect_induction_def.
23089 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
23091 (vect_analyze_slp_cost_1): Cost induction.
23092 (vect_detect_hybrid_slp_stmts): Handle PHIs.
23093 (vect_get_slp_vect_defs): Likewise.
23094 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
23095 (vect_transform_stmt): Handle SLP reductions.
23096 * tree-vectorizer.h (vectorizable_induction): Adjust.
23098 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23100 * config/rs6000/rs6000.c (make_resolver_func): Update
23101 init_lowered_empty_function call.
23103 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
23105 * doc/invoke.texi: Document the -fprofile-abs-path option.
23106 * common.opt (fprofile-abs-path): New option.
23107 * gcov-io.h (gcov_write_filename): Declare.
23108 * gcov-io.c (gcov_write_filename): New function.
23109 * coverage.c (coverage_begin_function): Use gcov_write_filename.
23110 * profile.c (output_location): Likewise.
23112 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23114 * shring-wrap.c: Revert accidental commit.
23116 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
23118 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
23120 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23122 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
23124 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
23125 profile in callgraph edge.
23126 * profile-count.h (apply_probability): If THIS is 0, then result is 0
23127 (apply_scale): Likewise.
23128 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
23129 Also scale profile when inlining function with zero profile.
23130 (initialize_cfun): Update exit block profile even when it is zero.
23131 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
23132 when profile is read.
23134 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23136 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
23137 (CLONE_*): New constants to define the processors we can generate
23138 code for with the target_clone attribute.
23139 (rs6000_clone_map): New array to identify which clone processors
23140 the current program is running on.
23141 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
23142 target_clone attribute.
23143 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
23144 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
23145 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
23146 (cpu_expand_builtin): Add support for target_clone attribute.
23147 (rs6000_valid_attribute_p): Allow "default" attribute.
23148 (get_decl_name): New debug function to simplify printing the
23149 current function name in debugging statements.
23150 (rs6000_clone_priority): New functions to support the target_clone
23151 attribute, and be able to generate code to switch between ISA 2.05
23152 through ISA 3.0 (power6 through power9).
23153 (rs6000_compare_version_priority): Likewise.
23154 (rs6000_get_function_versions_dispatcher): Likewise.
23155 (make_resolver_func): Likewise.
23156 (add_condition_to_bb): Likewise.
23157 (dispatch_function_versions): Likewise.
23158 (rs6000_generate_version_dispatcher_body): Likewise.
23159 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
23160 (fusion_gpr_load_p): Fix a spacing issue.
23161 * doc/extend.texi (Common Function Attributes): Document that the
23162 PowerPC supports the target_clone attribute.
23164 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
23166 * config/arm/arm.h: explain F symbol found in description of ARM
23167 register allocation in its legend.
23169 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23171 * config/mips/frame-header-opt.c: Include profile-count.h.
23172 * config/riscv/riscv.c: Include profile-count.h
23174 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
23176 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
23178 (sm_set_flag_if_changed): Add bbs field.
23179 (execute_sm_if_changed_flag_set): Pass BBS.
23180 (execute_sm): Update.
23182 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23184 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
23187 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23189 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
23190 (peephole2): New peephole2 to emit the above.
23191 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
23193 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23195 * config/aarch64/aarch64.c (define_peephole2 above
23196 *sub_<shift>_<mode>): New peephole.
23198 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23200 * config/i386/i386.c (make_resolver_func): Update.
23201 * Makefile.in: Add profile-count.h and profile-count.o
23202 * auto-profile.c (afdo_indirect_call): Update to new API.
23203 (afdo_set_bb_count): Update.
23204 (afdo_propagate_edge): Update.
23205 (afdo_propagate_circuit): Update.
23206 (afdo_calculate_branch_prob): Update.
23207 (afdo_annotate_cfg): Update.
23208 * basic-block.h: Include profile-count.h
23209 (struct edge_def): Turn count to profile_count.
23210 (struct basic_block_def): Likewie.
23211 (REG_BR_PROB_BASE): Move to profile-count.h
23212 (RDIV): Move to profile-count.h
23213 * bb-reorder.c (max_entry_count): Turn to profile_count.
23214 (find_traces): Update.
23215 (rotate_loop):Update.
23216 (connect_traces):Update.
23217 (sanitize_hot_paths):Update.
23218 * bt-load.c (migrate_btr_defs): Update.
23219 * cfg.c (RDIV): Remove.
23220 (init_flow): Use alloc_block.
23221 (alloc_block): Uninitialize count.
23222 (unchecked_make_edge): Uninitialize count.
23223 (check_bb_profile): Update.
23224 (dump_edge_info): Update.
23225 (dump_bb_info): Update.
23226 (update_bb_profile_for_threading): Update.
23227 (scale_bbs_frequencies_int): Update.
23228 (scale_bbs_frequencies_gcov_type): Update.
23229 (scale_bbs_frequencies_profile_count): New.
23230 * cfg.h (update_bb_profile_for_threading): Update.
23231 (scale_bbs_frequencies_profile_count): Declare.
23232 * cfgbuild.c (compute_outgoing_frequencies): Update.
23233 (find_many_sub_basic_blocks): Update.
23234 * cfgcleanup.c (try_forward_edges): Update.
23235 (try_crossjump_to_edge): Update.
23236 * cfgexpand.c (expand_gimple_tailcall): Update.
23237 (construct_exit_block): Update.
23238 * cfghooks.c (verify_flow_info): Update.
23239 (dump_bb_for_graph): Update.
23240 (split_edge): Update.
23241 (make_forwarder_block): Update.
23242 (duplicate_block): Update.
23243 (account_profile_record): Update.
23244 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
23245 (get_estimated_loop_iterations): Update.
23246 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
23247 (single_likely_exit): Update.
23248 * cfgloopmanip.c (scale_loop_profile): Update.
23250 (set_zero_probability): Update.
23251 (lv_adjust_loop_entry_edge): Update.
23252 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23253 (purge_dead_edges): Update.
23254 (rtl_account_profile_record): Update.
23255 * cgraph.c (cgraph_node::create): Uninitialize count.
23256 (symbol_table::create_edge): Uninitialize count.
23257 (cgraph_update_edges_for_call_stmt_node): Update.
23258 (cgraph_edge::dump_edge_flags): Update.
23259 (cgraph_node::dump): Update.
23260 (cgraph_edge::maybe_hot_p): Update.
23261 * cgraph.h: Include profile-count.h
23262 (create_clone), create_edge, create_indirect_edge): Update.
23263 (cgraph_node): Turn count to profile_count.
23264 (cgraph_edge0: Likewise.
23265 (make_speculative, clone): Update.
23266 (create_edge): Update.
23267 (init_lowered_empty_function): Update.
23268 * cgraphclones.c (cgraph_edge::clone): Update.
23269 (duplicate_thunk_for_node): Update.
23270 (cgraph_node::create_clone): Update.
23271 * cgraphunit.c (cgraph_node::analyze): Update.
23272 (cgraph_node::expand_thunk): Update.
23273 * final.c (dump_basic_block_info): Update.
23274 * gimple-streamer-in.c (input_bb): Update.
23275 * gimple-streamer-out.c (output_bb): Update.
23276 * graphite.c (print_global_statistics): Update.
23277 (print_graphite_scop_statistics): Update.
23278 * hsa-brig.c: Include basic-block.h.
23279 * hsa-dump.c: Include basic-block.h.
23280 * hsa-gen.c (T sum_slice): Update.
23281 (convert_switch_statements):Update.
23282 * hsa-regalloc.c: Include basic-block.h.
23283 * ipa-chkp.c (chkp_produce_thunks): Update.
23284 * ipa-cp.c (struct caller_statistics): Update.
23285 (init_caller_stats): Update.
23286 (gather_caller_stats): Update.
23287 (ipcp_cloning_candidate_p): Update.
23288 (good_cloning_opportunity_p): Update.
23289 (get_info_about_necessary_edges): Update.
23290 (dump_profile_updates): Update.
23291 (update_profiling_info): Update.
23292 (update_specialized_profile): Update.
23293 (perhaps_add_new_callers): Update.
23294 (decide_about_value): Update.
23295 (ipa_cp_c_finalize): Update.
23296 * ipa-devirt.c (struct odr_type_warn_count): Update.
23297 (struct decl_warn_count): Update.
23298 (struct final_warning_record): Update.
23299 (possible_polymorphic_call_targets): Update.
23300 (ipa_devirt): Update.
23301 * ipa-fnsummary.c (redirect_to_unreachable): Update.
23302 * ipa-icf.c (sem_function::merge): Update.
23303 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
23304 * ipa-inline.c (compute_uninlined_call_time): Update.
23305 (compute_inlined_call_time): Update.
23306 (want_inline_small_function_p): Update.
23307 (want_inline_self_recursive_call_p): Update.
23308 (edge_badness): Update.
23309 (lookup_recursive_calls): Update.
23310 (recursive_inlining): Update.
23311 (inline_small_functions): Update.
23312 (dump_overall_stats): Update.
23313 (dump_inline_stats): Update.
23314 * ipa-profile.c (ipa_profile_generate_summary): Update.
23315 (ipa_propagate_frequency): Update.
23316 (ipa_profile): Update.
23317 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23318 * ipa-utils.c (ipa_merge_profiles): Update.
23319 * loop-doloop.c (doloop_modify): Update.
23320 * loop-unroll.c (report_unroll): Update.
23321 (unroll_loop_runtime_iterations): Update.
23322 * lto-cgraph.c (lto_output_edge): Update.
23323 (lto_output_node): Update.
23324 (input_node): Update.
23325 (input_edge): Update.
23326 (merge_profile_summaries): Update.
23327 * lto-streamer-in.c (input_cfg): Update.
23328 * lto-streamer-out.c (output_cfg): Update.
23329 * mcf.c (create_fixup_graph): Update.
23330 (adjust_cfg_counts): Update.
23331 (sum_edge_counts): Update.
23332 * modulo-sched.c (sms_schedule): Update.
23333 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
23334 * predict.c (maybe_hot_count_p): Update.
23335 (probably_never_executed): Update.
23336 (dump_prediction): Update.
23337 (combine_predictions_for_bb): Update.
23338 (propagate_freq): Update.
23339 (handle_missing_profiles): Update.
23340 (counts_to_freqs): Update.
23341 (rebuild_frequencies): Update.
23342 (force_edge_cold): Update.
23343 * predict.h: Include profile-count.h
23344 (maybe_hot_count_p, counts_to_freqs): UPdate.
23345 * print-rtl-function.c: Do not include cfg.h
23346 * print-rtl.c: Include basic-block.h
23347 * profile-count.c: New file.
23348 * profile-count.h: New file.
23349 * profile.c (is_edge_inconsistent): Update.
23350 (correct_negative_edge_counts): Update.
23351 (is_inconsistent): Update.
23352 (set_bb_counts): Update.
23353 (read_profile_edge_counts): Update.
23354 (compute_frequency_overlap): Update.
23355 (compute_branch_probabilities): Update; Initialize and deinitialize
23357 (branch_prob): Update.
23358 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
23359 (edge_gcov_count): New.
23360 (bb_gcov_count): New.
23361 * shrink-wrap.c (try_shrink_wrapping): Update.
23362 * tracer.c (better_p): Update.
23363 * trans-mem.c (expand_transaction): Update.
23364 (ipa_tm_insert_irr_call): Update.
23365 (ipa_tm_insert_gettmclone_call): Update.
23366 * tree-call-cdce.c: Update.
23367 * tree-cfg.c (gimple_duplicate_sese_region): Update.
23368 (gimple_duplicate_sese_tail): Update.
23369 (gimple_account_profile_record): Update.
23370 (execute_fixup_cfg): Update.
23371 * tree-inline.c (copy_bb): Update.
23372 (copy_edges_for_bb): Update.
23373 (initialize_cfun): Update.
23374 (freqs_to_counts): Update.
23375 (copy_cfg_body): Update.
23376 (expand_call_inline): Update.
23377 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23378 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
23379 (try_unroll_loop_completely): Update.
23380 (try_peel_loop): Update.
23381 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23382 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
23383 * tree-ssa-loop-split.c (connect_loops): Update.
23384 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
23385 * tree-ssa-reassoc.c (branch_fixup): Update.
23386 * tree-ssa-tail-merge.c (replace_block_by): Update.
23387 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
23388 (compute_path_counts): Update.
23389 (update_profile): Update.
23390 (recompute_probabilities): Update.
23391 (update_joiner_offpath_counts): Update.
23392 (estimated_freqs_path): Update.
23393 (freqs_to_counts_path): Update.
23394 (clear_counts_path): Update.
23395 (ssa_fix_duplicate_block_edges): Update.
23396 (duplicate_thread_path): Update.
23397 * tree-switch-conversion.c (case_bit_test_cmp): Update.
23398 (struct switch_conv_info): Update.
23399 * tree-tailcall.c (decrease_profile): Update.
23400 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23401 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23402 * value-prof.c (check_counter): Update.
23403 (gimple_divmod_fixed_value): Update.
23404 (gimple_mod_pow2): Update.
23405 (gimple_mod_subtract): Update.
23406 (gimple_ic_transform): Update.
23407 (gimple_stringop_fixed_value): Update.
23408 * value-prof.h (gimple_ic): Update.
23410 2017-06-02 Carl Love <cel@us.ibm.com>
23412 * config/rs6000/rs6000-c: Add support for built-in functions
23413 vector double vec_doublee (vector signed int);
23414 vector double vec_doublee (vector unsigned int);
23415 vector double vec_doublee (vector float);
23416 vector double vec_doubleh (vector signed int);
23417 vector double vec_doubleh (vector unsigned int);
23418 vector double vec_doubleh (vector float);
23419 vector double vec_doublel (vector signed int);
23420 vector double vec_doublel (vector unsigned int);
23421 vector double vec_doublel (vector float);
23422 vector double vec_doubleo (vector signed int);
23423 vector double vec_doubleo (vector unsigned int);
23424 vector double vec_doubleo (vector float);.
23425 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
23426 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
23428 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
23429 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
23430 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
23432 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
23433 vec_doublel, vec_doubleh.
23434 * doc/extend.texi: Update the built-in documentation file for the
23435 new built-in functions.
23437 2017-06-02 David Malcolm <dmalcolm@redhat.com>
23440 * ipa-inline-analysis.c (free_growth_caches): Set
23441 edge_removal_hook_holder to NULL after removing it.
23443 2017-06-02 Sudakshina Das <sudi.das@arm.com>
23445 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
23446 comparision with zero.
23448 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23449 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23450 for early expansion of vec_min and vec_max builtins.
23451 (builtin_function_type): Add min/max unsigned variants to those
23452 identified as having unsigned arguments.
23454 2017-06-02 Olivier Hainque <hainque@adacore.com>
23456 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
23458 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23460 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
23461 Use VALL_F16 iterator rather than VALL.
23463 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23465 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
23466 Emit CBNZ inside loop when doing a strong exchange and comparing
23467 against zero. Generate the CC flags after the loop.
23469 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
23471 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
23472 (dl_section_ref): New.
23473 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
23474 On AIX, append an expression to subtract the size of the
23475 section length to dl_section_ref.
23477 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23479 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23480 for early expansion of vector absolute builtins.
23482 2017-06-02 Richard Biener <rguenther@suse.de>
23484 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
23485 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
23487 2017-06-02 Richard Biener <rguenther@suse.de>
23489 PR tree-optimization/80948
23490 * tree-tailcall.c (find_tail_calls): Track stmts to move in
23491 stmt order as well.
23493 2017-06-02 Richard Biener <rguenther@suse.de>
23495 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
23497 * tree-vect-stmts.c (process_use): Do not mark backedge defs
23498 for inductions as relevant.
23500 2017-06-02 Richard Biener <rguenther@suse.de>
23502 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
23503 (vectorizable_induction): ... this. Remove dead code.
23505 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
23507 * builtins. (expand_builtin_alloca): Remove second parameter and
23508 infer its value from the first parameter instead.
23509 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
23511 2017-06-02 Jakub Jelinek <jakub@redhat.com>
23513 PR rtl-optimization/80903
23514 * loop-doloop.c (add_test): Unshare sequence.
23516 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
23518 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
23520 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
23522 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
23524 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
23525 xlogue_layout::get_instance, logue_layout::xlogue_layout,
23526 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
23527 (xlogue_layout::get_stub_rtx): Make static.
23528 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
23529 (xlogue_layout::compute_stub_managed_regs): Rename to...
23530 (xlogue_layout::count_stub_managed_regs): ...this.
23531 (xlogue_layout::is_stub_managed_reg): New function.
23532 (xlogue_layout::m_stub_names): Rename to...
23533 (xlogue_layout::s_stub_names): ...this, make static.
23534 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
23535 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
23536 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
23537 xlogue_layout::s_stub_names): Instantiate statics.
23538 (stub_managed_regs): Remove.
23539 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
23540 (disable_call_ms2sysv_xlogues): Rename to...
23541 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
23542 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
23544 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
23545 change after reload_completed.
23546 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
23548 (ix86_expand_prologue): Likewise.
23549 (ix86_expand_epilogue): Likewise.
23550 (ix86_expand_split_stack_prologue): Likewise.
23551 (ix86_compute_frame_layout): Remove frame parameter ...
23552 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
23553 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
23555 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
23556 (ix86_frame): Move from here ...
23557 * config/i386/i386.h (ix86_frame): ... to here.
23558 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
23559 complete ix86_frame data structure instead. Remove some_ld_name.
23561 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23563 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
23564 symbols that hold a DECL_VALUE_EXPR.
23566 2017-06-01 Martin Jambor <mjambor@suse.cz>
23568 PR tree-optimization/80898
23569 * tree-sra.c (process_subtree_disqualification): Removed.
23570 (disqualify_candidate): Do not acll
23571 process_subtree_disqualification.
23572 (subtree_mark_written_and_enqueue): New function.
23573 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
23574 RHS has been disqualified and re-queue LHS if necessary. Apart
23575 from that, ignore disqualified RHS.
23577 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23579 * config/s390/s390.c (s390_emit_epilogue): Disable early return
23580 address fetch for z10 or later.
23582 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23584 * config/arc/arc.md (tst_movb): Add guard when splitting.
23586 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23588 * config/arc/arc.c (arc_can_eliminate): Test against
23589 arc_frame_pointer_needed.
23591 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23593 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
23594 to prevent store reordering.
23595 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
23596 (type): Add block type.
23597 (stack_tie): Define special instruction to be used in
23600 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23602 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
23603 constraint. It is not valid for the pattern.
23604 (noncommutative_binary_comparison): Likewise.
23606 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23608 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
23611 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23613 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
23614 be used by the reg-alloc.
23616 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23618 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
23619 reg-alloc when having mul64 or mul32x16 instructions.
23620 (mulsidi3): Likewise.
23621 (umulsidi3): Likewise.
23622 (mulsi32x16): New pattern.
23623 (mulsi64): Likewise.
23624 (mulsidi64): Likewise.
23625 (umulsidi64): Likewise.
23626 (MUL32x16_REG): Define.
23627 (mul64_600): Use MUL32x16_REG.
23628 (mac64_600): Likewise.
23629 (umul64_600): Likewise.
23630 (umac64_600): Likewise.
23632 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23634 * config/arc/arc.md (mulsi3_700): Make it commutative.
23636 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
23638 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
23640 (*sign_extendsidi2_insn): Likewise for movstosw.
23642 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23644 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
23645 the type of the input discriminant value. Convert the
23646 discriminant value of signedness vary.
23648 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23650 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23651 Add to -Wall section.
23653 2017-06-01 Richard Biener <rguenther@suse.de>
23655 PR middle-end/66313
23656 * fold-const.c (fold_plusminus_mult_expr): If the factored
23657 factor may be zero use a wrapping type for the inner operation.
23658 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23659 and handle moved defs.
23660 (process_assignment): Properly guard the unary op case. Return a
23661 tri-state indicating that moving the stmt before the call may allow
23662 to continue. Pass through to_move.
23663 (find_tail_calls): Handle moving unrelated defs before
23666 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23669 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23670 splitter result in the canonical way.
23672 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23674 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23675 also for 32bit target. Update insn attributes.
23676 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23678 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23680 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23681 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23682 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23683 (_mm_maskz_min_ss): New intrinsics.
23685 2017-05-31 Martin Liska <mliska@suse.cz>
23687 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23688 Change comment style to one we normally use.
23689 (vectorizable_reduction): Likewise.
23690 (vectorizable_induction): Likewise.
23691 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23692 (vectorizable_call): Likewise.
23693 (vectorizable_simd_clone_call): Likewise.
23694 (vectorizable_conversion): Likewise.
23695 (vectorizable_assignment): Likewise.
23696 (vectorizable_shift): Likewise.
23697 (vectorizable_operation): Likewise.
23698 (vectorizable_store): Likewise.
23699 (vectorizable_load): Likewise.
23700 * tree-vectorizer.h: Likewise.
23702 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23704 * passes.c (emergency_dump_function): New.
23705 * tree-pass.h (emergency_dump_function): Declare.
23706 * plugin.c (plugins_internal_error_function): Remove.
23707 * plugin.h (plugins_internal_error_function): Remove declaration.
23708 * toplev.c (internal_error_function): New static function. Use it...
23709 (general_init): ...here.
23711 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23713 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23714 (arc_rtx_costs): Add costs for new patterns.
23715 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23716 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23718 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23720 * tree-ssa-strlen.c (get_next_strinfo): New function.
23721 (get_stridx_plus_constant): Use it.
23722 (zero_length_string): Likewise.
23723 (adjust_related_strinfos): Likewise.
23724 (adjust_last_stmt): Likewise.
23726 2017-05-31 Richard Biener <rguenther@suse.de>
23729 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23730 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23732 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23734 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23735 loop_vinfo argument and use of dependence distance vectors.
23736 Check instead whether the two references differ only in their
23737 initial value and assume that they have the same alignment if the
23738 difference is a multiple of the vector alignment.
23739 (vect_analyze_data_refs_alignment): Update call accordingly.
23741 2017-05-31 Martin Liska <mliska@suse.cz>
23744 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23746 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23748 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23749 (create_intersect_range_checks): Move from ...
23750 * tree-data-ref.c (create_intersect_range_checks_index)
23751 (create_intersect_range_checks): ... to here.
23752 (create_runtime_alias_checks): New function factored from ...
23753 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23754 here. Call above function.
23755 * tree-data-ref.h (create_runtime_alias_checks): New function.
23757 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23759 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23760 segment length for dr_b and compute it in wide_int.
23762 2017-05-31 Richard Biener <rguenther@suse.de>
23764 PR tree-optimization/80906
23765 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23766 and pass through iv_map.
23767 (copy_bb_and_scalar_dependences): Adjust.
23768 (translate_pending_phi_nodes): Likewise.
23769 (copy_loop_close_phi_args): Handle code-generating IVs instead
23772 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23774 * diagnostic-color.c (color_dict): Add "type-diff".
23775 (parse_gcc_colors): Update comment.
23776 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23777 -fdiagnostics-show-template-tree and -fno-elide-type.
23778 (GCC_COLORS): Add type-diff to example.
23780 (-fdiagnostics-show-template-tree): New.
23781 (-fno-elide-type): New.
23782 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23783 the pp_format_decoder callback. Call any m_format_postprocessor's
23785 (pretty_printer::pretty_printer): Initialize
23786 m_format_postprocessor.
23787 (pretty_printer::~pretty_printer): Delete any
23788 m_format_postprocessor.
23789 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23790 (class format_postprocessor): New class.
23791 (struct pretty_printer::format_decoder): Document the new parameters.
23792 (struct pretty_printer::m_format_postprocessor): New field.
23793 * tree-diagnostic.c (default_tree_printer): Update for new
23794 bool and const char ** params.
23795 * tree-diagnostic.h (default_tree_printer): Likewise.
23797 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23799 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23800 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23801 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23802 rs6000_gen_cell_microcode code.
23803 (rs6000_final_prescan_insn): Delete.
23804 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23805 "warn-cell-microcode" entries.
23806 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23807 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23808 throughout. Change cc_reg_not_micro_cr0_operand to
23809 cc_reg_not_cr0_operand throughout.
23810 (*extendhi<mode>2_noload): Delete.
23811 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23812 (mwarn-cell-microcode): Delete.
23813 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23814 -mgen-cell-microcode and -mwarn-cell-microcode.
23816 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23819 * config/i386/constraints.md (Yd): New constraint.
23821 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23822 and (?Yd, r) alternatives. Update insn attributes.
23823 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23824 and (?*Yd, r) alternatives. Update insn attributes.
23825 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23827 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23829 * gimplify.c (gimplify_modify_expr): Don't create a
23830 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23833 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23835 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23837 2017-05-30 Richard Biener <rguenther@suse.de>
23839 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23840 and reduc_def fields.
23841 (STMT_VINFO_REDUC_TYPE): New define.
23842 (STMT_VINFO_REDUC_DEF): Likewise.
23843 (vect_force_simple_reduction): Adjust prototype.
23844 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23845 (vect_is_simple_reduction): Remove check_reduction argument.
23846 (vect_force_simple_reduction): Adjust and set
23847 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23848 (vectorizable_reduction): Do not re-do reduction analysis
23849 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23850 * tree-parloops.c (gather_scalar_reductions): Adjust.
23852 2017-05-30 Richard Biener <rguenther@suse.de>
23854 PR middle-end/80901
23855 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23858 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23860 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23861 Introduce unknown_misalignment parameter and remove vf.
23862 (vect_peeling_hash_get_lowest_cost):
23863 Pass unknown_misalignment parameter.
23864 (vect_enhance_data_refs_alignment):
23865 Fix unsupportable data ref treatment.
23867 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23869 * tree-vect-data-refs.c (vect_get_data_access_cost):
23870 Workaround for SLP handling.
23871 (vect_enhance_data_refs_alignment):
23872 Compute costs for doing no peeling at all, compare to the best
23873 peeling costs so far and avoid peeling if cheaper.
23875 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23877 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23878 Return peeling info and set costs to zero for unlimited cost
23880 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23881 with unknown misalignment. Compute and costs for unknown
23882 misalignment, compare them to the costs for known misalignment
23883 and choose the cheapest for peeling.
23885 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23887 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23888 (vect_get_peeling_costs_all_drs): Create function.
23889 (vect_peeling_hash_get_lowest_cost):
23890 Use vect_get_peeling_costs_all_drs.
23891 (vect_peeling_supportable): Create function.
23892 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23894 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23896 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23897 DR_HAS_NEGATIVE_STEP.
23898 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23899 (vect_enhance_data_refs_alignment): Use.
23900 (vect_duplicate_ssa_name_ptr_info): Use.
23901 * tree-vectorizer.h (dr_misalignment): Use.
23902 (known_alignment_for_access_p): Use.
23904 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23907 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23909 (has_section_name): New function.
23911 2017-05-30 Martin Liska <mliska@suse.cz>
23914 * auto-profile.c (get_function_decl_from_block): Fix
23917 2017-05-30 Richard Biener <rguenther@suse.de>
23919 PR middle-end/80876
23920 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23922 2017-05-30 Martin Liska <mliska@suse.cz>
23924 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23925 * dumpfile.h (struct dump_file_info): Remove ctors.
23927 2017-05-30 Martin Liska <mliska@suse.cz>
23929 * predict.def: Fix GNU coding style.
23931 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23933 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23934 Mark 'to' argument with ATTRIBUTE_UNUSED.
23936 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23938 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23939 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23940 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23943 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23945 * doc/install.texi (Options specification): Restore entry of
23946 --enable-sjlj-exceptions.
23948 2017-05-27 Michael Eager <eager@eagercon.com>
23951 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23953 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23955 * config/microblaze/microblaze.h
23956 (FIXED_REGISTERS): Update in macro.
23957 (CALL_USED_REGISTERS): Update in macro.
23959 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23961 * doc/install.texi: Add links to macOS binary distributions.
23963 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23967 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23969 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23971 2017-05-26 Martin Liska <mliska@suse.cz>
23973 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23975 2017-05-26 Martin Liska <mliska@suse.cz>
23977 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23978 always leading ';; '.
23979 (dump_bb_info): Likewise.
23980 (brief_dump_cfg): Likewise.
23981 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23982 * dumpfile.c: Remove usage of TDF_VERBOSE.
23983 * dumpfile.h (enum dump_kind): Likewise.
23984 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23985 * print-tree.c (debug_verbose): Remove.
23986 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23987 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23988 * tree-diagnostic.c (default_tree_printer): Replace
23989 TDF_DIAGNOSTIC with TDF_SLIM.
23991 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23993 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23994 in parameter loop, rather than loop_vinfo.
23995 (create_intersect_range_checks): Ditto.
23996 (vect_create_cond_for_alias_checks): Update call to above functions.
23998 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24000 PR tree-optimization/80815
24001 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
24002 for merging runtime alias checks. Handle negative DR_STEPs.
24004 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24006 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
24008 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
24009 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
24010 out code pruning runtime alias checks.
24011 * tree-data-ref.c (prune_runtime_alias_test_list): New function
24012 factored out from above.
24013 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
24015 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
24017 (prune_runtime_alias_test_list): New decalaration.
24019 2017-05-26 Bin Cheng <bin.cheng@arm.com>
24021 * tree-vect-data-refs.c (compare_tree): Rename and move ...
24022 * tree-data-ref.c (data_ref_compare_tree): ... to here.
24023 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
24024 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
24025 (operator==, comp_dr_with_seg_len_pair): Ditto.
24026 (vect_prune_runtime_alias_test_list): Ditto.
24028 2017-05-26 Martin Liska <mliska@suse.cz>
24031 * params.def: Bound partial-inlining-entry-probability param.
24033 2017-05-26 Marek Polacek <polacek@redhat.com>
24036 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
24039 2017-05-26 Richard Biener <rguenther@suse.de>
24041 PR tree-optimization/80842
24042 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
24045 2017-05-26 Richard Biener <rguenther@suse.de>
24047 PR tree-optimization/80844
24048 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
24050 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
24052 * doc/md.texi (Machine Constraints): Update x86 family
24053 machine constraints section to match 'config/i386/constraints.md'.
24055 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
24057 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
24059 2017-05-25 Nathan Sidwell <nathan@acm.org>
24061 * doc/invoke.texi (--enable-languages): Update documentation.
24063 2017-05-25 Martin Liska <mliska@suse.cz>
24065 * dumpfile.c: Add TDF_FOLDING.
24066 * dumpfile.h (enum dump_kind): Likewise.
24067 * genmatch.c (dt_simplify::gen_1): Use it.
24069 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24071 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
24073 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24075 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
24076 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
24078 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
24080 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
24081 * match.pd (X == C): Rewrite it here.
24082 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
24083 with_certain_nonzero_bits2): New predicates.
24084 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
24086 2017-05-24 Nathan Sidwell <nathan@acm.org>
24088 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
24091 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
24094 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
24096 * config/powerpcspe: New port. Files are copied from the rs6000
24097 port, with "rs6000" in filenames replaced by "powerpcspe".
24099 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
24101 PR rtl-optimization/80754
24102 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
24104 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
24106 * config/sparc/sparc.md (length): Return the correct value for -mflat
24107 sibcalls to match output_sibcall.
24109 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
24113 * config/rs6000/rs6000.c (struct machine_function): Add new field
24115 (rs6000_get_separate_components): Init that field, use it.
24116 (rs6000_components_for_bb): Use the field.
24118 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
24120 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
24122 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
24124 PR middle-end/80823
24125 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
24127 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24130 * config/s390/s390.c (s390_check_qrst_address): Check incoming
24131 address against address_operand predicate.
24132 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
24134 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
24136 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
24137 parameters passed indirectly.
24139 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
24141 * config/i386/i386.md (*movdi_internal): Remove SSE4
24142 alternative 18 (?r, *v). Update insn attributes.
24143 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
24144 Update insn attributes.
24145 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
24146 Update insn attributes.
24147 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
24148 alternative 1 (r, v). Remove isa attribute.
24149 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
24150 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
24151 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
24153 2017-05-23 Tom de Vries <tom@codesourcery.com>
24155 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
24158 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24160 * cgraphunit.c (symbol_table::process_new_functions): Update.
24161 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
24162 (inline_generate_summary): Rename to ...
24163 (ipa_fn_summary_generate): ... this one.
24164 (inline_read_summary): Rename to ...
24165 (ipa_fn_summary_read): ... this one.
24166 (inline_write_summary): Rename to ...
24167 (ipa_fn_summary_write): ... this one.
24168 (inline_free_summary): Rename to ...
24169 (ipa_free_fn_summary): ... this one.
24170 (pass_data_local_fn_summary, pass_local_fn_summary,
24171 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
24172 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
24173 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
24174 make_pass_ipa_fn_summary): New.
24175 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
24176 inline_write_summary, inline_free_summary): Remove.
24177 (ipa_free_fn_summary) : New.
24178 * ipa-inline.c (ipa_inline): Update.
24179 (pass_ipa_inline): Do not generate summaries.
24180 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
24182 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
24183 and add pass_ipa_fn_summary.
24184 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
24186 (make_pass_inline_parameters): Remove.
24188 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
24190 * omp-low.c (struct omp_context): Remove "default_kind" member.
24193 * omp-offload.c (execute_oacc_device_lower): Remove the
24194 parallelism dimensions function attributes for unparallelized
24195 OpenACC kernels constructs.
24197 2017-05-23 Martin Liska <mliska@suse.cz>
24199 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
24201 (cgraph_edge::make_speculative): Likewise.
24202 (cgraph_edge::resolve_speculation): Likewise.
24203 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
24204 (cgraph_node::dump): Likewise.
24205 * cgraph.h: Likewise.
24206 * cgraphunit.c (analyze_functions): Likewise.
24207 (symbol_table::compile): Likewise.
24208 * ipa-cp.c (print_all_lattices): Likewise.
24209 (determine_versionability): Likewise.
24210 (initialize_node_lattices): Likewise.
24211 (ipcp_verify_propagated_values): Likewise.
24212 (estimate_local_effects): Likewise.
24213 (update_profiling_info): Likewise.
24214 (create_specialized_node): Likewise.
24215 (perhaps_add_new_callers): Likewise.
24216 (decide_about_value): Likewise.
24217 (decide_whether_version_node): Likewise.
24218 (identify_dead_nodes): Likewise.
24219 (ipcp_store_bits_results): Likewise.
24220 * ipa-devirt.c (dump_targets): Likewise.
24221 (ipa_devirt): Likewise.
24222 * ipa-icf.c (sem_item::dump): Likewise.
24223 (sem_function::equals): Likewise.
24224 (sem_variable::equals): Likewise.
24225 (sem_item_optimizer::read_section): Likewise.
24226 (sem_item_optimizer::execute): Likewise.
24227 (congruence_class::dump): Likewise.
24228 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
24229 (dump_inline_summary): Likewise.
24230 (estimate_node_size_and_time): Likewise.
24231 (inline_analyze_function): Likewise.
24232 * ipa-inline-transform.c (inline_call): Likewise.
24233 * ipa-inline.c (report_inline_failed_reason): Likewise.
24234 (want_early_inline_function_p): Likewise.
24235 (edge_badness): Likewise.
24236 (update_edge_key): Likewise.
24237 (inline_small_functions): Likewise.
24238 * ipa-profile.c (ipa_profile): Likewise.
24239 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
24240 (ipa_make_edge_direct_to_target): Likewise.
24241 (remove_described_reference): Likewise.
24242 (ipa_impossible_devirt_target): Likewise.
24243 (propagate_controlled_uses): Likewise.
24244 (ipa_print_node_params): Likewise.
24245 (ipcp_transform_function): Likewise.
24246 * ipa-pure-const.c (pure_const_read_summary): Likewise.
24247 (propagate_pure_const): Likewise.
24248 * ipa-reference.c (generate_summary): Likewise.
24249 (read_write_all_from_decl): Likewise.
24250 (propagate): Likewise.
24251 (ipa_reference_read_optimization_summary): Likewise.
24252 * ipa-utils.c (ipa_merge_profiles): Likewise.
24253 * ipa.c (walk_polymorphic_call_targets): Likewise.
24254 (symbol_table::remove_unreachable_nodes): Likewise.
24255 (ipa_single_use): Likewise.
24256 * passes.c (execute_todo): Likewise.
24257 * predict.c (drop_profile): Likewise.
24258 * symtab.c (symtab_node::get_dump_name): New function.
24259 (symtab_node::dump_name): Likewise.
24260 (symtab_node::dump_asm_name): Likewise.
24261 (symtab_node::dump_references): Likewise.
24262 (symtab_node::dump_referring): Likewise.
24263 (symtab_node::dump_base): Likewise.
24264 (symtab_node::debug_symtab): Likewise.
24265 * tree-sra.c (convert_callers_for_node): Likewise.
24266 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
24267 * value-prof.c (init_node_map): Likewise.
24269 2017-05-23 Martin Liska <mliska@suse.cz>
24271 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
24272 and symtab_node::debug_symtab to symbol_table::debug.
24273 * cgraphunit.c (analyze_functions): Use the renamed function.
24274 (symbol_table::compile): Likewise.
24275 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
24276 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
24277 * passes.c (execute_todo): Likewise.
24278 * symtab.c (symbol_table::dump): New function.
24279 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
24281 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24283 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
24284 that nonconst implies exec.
24286 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24288 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24289 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24290 (inline_edge_summary_vec): Turn into ...
24291 (ipa_call_summaries): ... this one.
24292 (redirect_to_unreachable, edge_set_predicate,
24293 evaluate_properties_for_edge, inline_summary_alloc,
24294 reset_ipa_call_summary, reset_inline_summary,
24295 inline_summary_t::duplicate): Update.
24296 (inline_edge_duplication_hook): Turn to ...
24297 (ipa_call_summary_t::duplicate): ... this one.
24298 (inline_edge_removal_hook): Turn to ...
24299 (ipa_call_summary_t::remove): ... this one.
24300 (dump_inline_edge_summary): Turn to ...
24301 (dump_ipa_call_summary): ... this one.
24302 (estimate_function_body_sizes): Update.
24303 (inline_update_callee_summaries): Update.
24304 (remap_edge_change_prob): Update.
24305 (remap_edge_summaries): Update.
24306 (inline_merge_summary): Update.
24307 (do_estimate_edge_time): Update.
24308 (inline_generate_summary): Update.
24309 (inline_read_section): Update.
24310 (inline_read_summary): Update.
24311 (inline_free_summary): Update.
24312 * ipa-inline.c (can_inline_edge_p): Update.
24313 (compute_inlined_call_time): Update.
24314 (want_inline_small_function_p): Update.
24315 (edge_badness): Update.
24316 (early_inliner): Update.
24317 * ipa-inline.h (inline_edge_summary): Turn to ...
24318 (ipa_call_summary): ... this one.
24319 (ipa_call_summary_t): New class.
24320 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24321 (ipa_call_summaries): New.
24322 (inline_edge_summary): Remove.
24323 (estimate_edge_growth): Update.
24324 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24325 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24326 * ipa-split.c (execute_split_functions): Update.
24327 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24329 2017-05-23 Tom de Vries <tom@codesourcery.com>
24331 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24332 attributes): Document rdrand effective target.
24334 2017-05-23 Tom de Vries <tom@codesourcery.com>
24336 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24337 attributes): Sort alphabetically.
24339 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
24341 * config/avr/genmultilib.awk: Use gsub instead of gensub.
24343 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
24346 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
24347 V2DF/V2DI splat into two separate patterns, one that handles
24348 registers, and the other that only handles memory. Drop support
24349 for splatting from a GPR on ISA 2.07 and then splitting the
24350 splat into direct move and splat.
24351 (vsx_splat_<mode>_reg): Likewise.
24352 (vsx_splat_<mode>_mem): Likewise.
24354 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
24356 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
24358 2017-05-22 Jakub Jelinek <jakub@redhat.com>
24360 PR middle-end/80809
24361 * omp-low.c (finish_taskreg_remap): New function.
24362 (finish_taskreg_scan): If unit size of ctx->record_type
24363 is non-constant, unshare the size expression and replace
24364 decls in it with possible outer var refs.
24366 PR middle-end/80809
24367 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
24368 GOVD_SHARED rather than GOVD_PRIVATE with it.
24369 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
24370 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
24372 PR middle-end/80853
24373 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
24374 as last argument to build_outer_var_ref for pointer bases of array
24375 section reductions.
24377 2017-05-19 Martin Sebor <msebor@redhat.com>
24379 * print-tree.c (print_node): Print DECL_READ_P flag.
24381 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24383 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
24384 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
24385 * cgraph.c: Likewise.
24386 * cgraphunit.c: Likewise.
24387 * gengtype.c: Likewise.
24388 * ipa-cp.c: Likewise.
24389 * ipa-devirt.c: Likewise.
24390 * ipa-icf.c: Likewise.
24391 * ipa-predicate.c: Likewise.
24392 * ipa-profile.c: Likewise.
24393 * ipa-prop.c: Likewise.
24394 * ipa-split.c: Likewise.
24396 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
24397 edge_predicate_pool, dump_inline_hints,
24398 inline_summary::account_size_time, redirect_to_unreachable,
24399 edge_set_predicate, set_hint_predicate,
24400 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
24401 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
24402 inline_summary_t::remove, remap_hint_predicate_after_duplication,
24403 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
24404 ipa_call_summary_t::remove, initialize_growth_caches,
24405 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
24406 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
24407 mark_modified, unmodified_parm_1, unmodified_parm,
24408 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
24409 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
24410 compute_bb_predicates, will_be_nonconstant_expr_predicate,
24411 will_be_nonconstant_predicate, record_modified_bb_info,
24412 get_minimal_bb, record_modified, param_change_prob,
24413 phi_result_unknown_predicate, predicate_for_phi_result,
24414 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
24415 estimate_function_body_sizes, compute_inline_parameters,
24416 compute_inline_parameters_for_curren, pass_data_inline_parameters,
24417 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
24418 inline_update_callee_summaries, remap_edge_change_prob,
24419 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
24420 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
24421 inline_analyze_function, inline_summary_t::insert,
24422 inline_generate_summary, read_ipa_call_summary, inline_read_section,
24423 inline_read_summary, write_ipa_call_summary, inline_write_summary,
24424 inline_free_summary): Move to ipa-fnsummary.h
24425 (predicate_t): Remove.
24426 * ipa-fnsummary.c: New file.
24427 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
24428 (enum inline_hints_vals, inline_hints, agg_position_info,
24429 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
24430 inline_summaries, ipa_call_summary, ipa_call_summary_t,
24431 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
24432 dump_inline_summary, dump_inline_hints, inline_generate_summary,
24433 inline_read_summary, inline_write_summary, inline_free_summary,
24434 inline_analyze_function, initialize_inline_failed,
24435 inline_merge_summary, inline_update_overall_summary,
24436 compute_inline_parameters): Move to ipa-fnsummary.h
24437 * ipa-fnsummary.h: New file.
24438 * ipa-inline-transform.h: Include ipa-inline.h.
24439 * ipa-inline.c: LIkewise.
24441 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24443 * ipa-inline.c (edge_badness): Use inlined_time instead of
24444 inline_summaries->get.
24446 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24448 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
24450 2017-05-22 Nathan Sidwell <nathan@acm.org>
24452 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
24453 (fdump-lang): Document 'raw' option.
24454 * dumpfile.h (TDI_tu): Delete.
24455 * dumpfile.c (dump_files): Remove translation-unit.
24456 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
24458 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
24460 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
24461 command option from $(AWK) call.
24462 * config/avr/genmultilib.awk: Simplify and rewrite so that it
24463 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
24464 [FORMAT]: Remove handling of variable.
24465 * config/avr/t-multilib: Regenerate.
24467 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24469 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
24471 (dump_inline_summary): Do not print self_time.
24472 (estimate_function_body_sizes): Do not set self_time.
24473 (compute_inline_parameters): Likewise.
24474 (inline_read_section, inline_write_summary): Do not stream self_time.
24475 * ipa-inline.h (inline_summary): Drop self_time.
24477 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24479 * ipa-inline-analysis.c (account_size_time): Rename to ...
24480 (inline_summary::account_size_time): ... this one.
24481 (reset_ipa_call_summary): Turn to ...
24482 (ipa_call_summary::reset): ... this one.
24483 (reset_inline_summary): Turn to ...
24484 (inline_summary::reset): ... this one.
24485 (inline_summary_t::remove): Update.
24486 (inline_summary_t::duplicate): Update.
24487 (ipa_call_summary_t::remove): Update.
24488 (dump_inline_summary): Update.
24489 (estimate_function_body_sizes): Update.
24490 (compute_inline_parameters): Update.
24491 (estimate_node_size_and_time): Update.
24492 (inline_merge_summary): Update.
24493 (inline_update_overall_summary): Update.
24494 (inline_read_section): Update.
24495 (inline_write_summary): Update.
24496 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
24497 add account_size_time and reset member functions.
24498 (ipa_call_summary): Add reset function.
24499 * ipa-predicate.h (predicate::operator &): Constify.
24501 2017-05-22 Richard Biener <rguenther@suse.de>
24503 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
24505 2017-05-19 Jason Merrill <jason@redhat.com>
24507 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
24509 2017-05-19 Marek Polacek <polacek@redhat.com>
24512 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
24513 TYPE_OVERFLOW_WRAPS checks.
24515 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
24517 * tree-core.h (enum omp_clause_default_kind): Add
24518 "OMP_CLAUSE_DEFAULT_PRESENT".
24519 * tree-pretty-print.c (dump_omp_clause): Handle it.
24520 * gimplify.c (enum gimplify_omp_var_data): Add
24521 "GOVD_MAP_FORCE_PRESENT".
24522 (gimplify_adjust_omp_clauses_1): Map it to
24523 "GOMP_MAP_FORCE_PRESENT".
24524 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
24526 * gimplify.c (oacc_default_clause): Clarify.
24528 2017-05-19 Nathan Sidwell <nathan@acm.org>
24530 LANG_HOOK_REGISTER_DUMPS
24531 * toplev.c (general_init): Call register dump lang hook.
24532 * doc/invoke.texi: Document -fdump-lang option family.
24533 * dumpfile.c (dump_files): Remove class dump here.
24534 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
24535 * dumpfile.h (tree_dump_index): Remove TDI_class.
24536 * langhooks-def.h (lhd_register_dumps): Declare.
24537 (LANG_HOOKS_REGISTER_DUMPS): Define.
24538 (LANG_HOOKS_INITIALIZER): Add it.
24539 * langhooks.c (lhd_register_dumps): Define.
24540 * langhooks.h (struct lang_hooks): Add register_dumps.
24542 2017-05-19 Nathan Sidwell <nathan@acm.org>
24544 * context.h (context::set_passes): New.
24545 * context.c (context::context): Do not create pass manager.
24546 * toplev.c (general_init): Create pass manager here.
24548 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
24550 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
24551 use this splitter if two add or or instructions would also work for
24552 the constant we want to generate.
24554 2017-05-19 Richard Biener <rguenther@suse.de>
24557 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
24558 predicate evaluation.
24560 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24562 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
24564 * ipa-inline.c (want_inline_small_function_p): Do not cast to
24567 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24569 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24570 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24571 (inline_edge_summary_vec): Turn into ...
24572 (ipa_call_summaries): ... this one.
24573 (redirect_to_unreachable, edge_set_predicate,
24574 evaluate_properties_for_edge, inline_summary_alloc,
24575 reset_ipa_call_summary, reset_inline_summary,
24576 inline_summary_t::duplicate): Update.
24577 (inline_edge_duplication_hook): Turn to ...
24578 (ipa_call_summary_t::duplicate): ... this one.
24579 (inline_edge_removal_hook): Turn to ...
24580 (ipa_call_summary_t::remove): ... this one.
24581 (dump_inline_edge_summary): Turn to ...
24582 (dump_ipa_call_summary): ... this one.
24583 (estimate_function_body_sizes): Update.
24584 (inline_update_callee_summaries): Update.
24585 (remap_edge_change_prob): Update.
24586 (remap_edge_summaries): Update.
24587 (inline_merge_summary): Update.
24588 (do_estimate_edge_time): Update.
24589 (inline_generate_summary): Update.
24590 (inline_read_section): Update.
24591 (inline_read_summary): Update.
24592 (inline_free_summary): Update.
24593 * ipa-inline.c (can_inline_edge_p): Update.
24594 (compute_inlined_call_time): Update.
24595 (want_inline_small_function_p): Update.
24596 (edge_badness): Update.
24597 (early_inliner): Update.
24598 * ipa-inline.h (inline_edge_summary): Turn to ...
24599 (ipa_call_summary): ... this one.
24600 (ipa_call_summary_t): New class.
24601 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24602 (ipa_call_summaries): New.
24603 (inline_edge_summary): Remove.
24604 (estimate_edge_growth): Update.
24605 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24606 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24607 * ipa-split.c (execute_split_functions): Update.
24608 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24610 2017-05-19 Richard Biener <rguenther@suse.de>
24612 PR middle-end/80764
24613 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
24615 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
24617 * config/rs6000/rs6000.c (struct machine_function): Add field
24618 fpr_is_wrapped_separately.
24619 (rs6000_get_separate_components): Use 64 components. Handle the
24620 new FPR components.
24621 (rs6000_components_for_bb): Handle the FPR components.
24622 (rs6000_emit_prologue_components): Handle the FPR components.
24623 (rs6000_emit_epilogue_components): Handle the FPR components.
24624 (rs6000_set_handled_components): Handle the FPR components.
24625 (rs6000_emit_prologue): Don't output prologue code for those FPRs
24626 that are already separately shrink-wrapped.
24627 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
24628 that are already separately shrink-wrapped.
24630 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24633 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
24636 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
24637 (define_peephole2 for Altivec d-form load): Add peepholes to catch
24638 cases where the register allocator uses a move and an offsettable
24639 memory operation to/from a FPR register on ISA 2.06/2.07.
24640 (define_peephole2 for Altivec d-form store): Likewise.
24642 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
24645 * config/i386/mmx.md (*mov<mode>_internal): Enable
24646 alternatives 11, 12, 13 and 14 also for 32bit targets.
24647 Remove alternatives 15, 16, 17 and 18.
24648 * config/i386/sse.md (vec_concatv2di): Change
24649 alternative (!x, *y) to (x, ?!*Yn).
24651 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24653 * dumpfile.h (enum dump_kind): Remove stray comma.
24655 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24657 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24658 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24659 predicate::num_conditions
24660 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24661 (CHANGED): turn into predicate::changed.
24662 (agg_position_info): Move to ipa-predicate.h
24663 (add_condition, predicate::add_clause, predicate::operator &=,
24664 predicate::or_with, predicate::evaluate, predicate::probability,
24665 dump_condition, dump_clause, predicate::dump,
24666 predicate::remap_after_duplication, predicate::remap_after_inlining,
24667 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24668 (evaluate_conditions_for_known_args): Update.
24669 (set_cond_stmt_execution_predicate): Update.
24670 * ipa-inline.h: Include ipa-predicate.h
24671 (condition, inline_param_summary, conditions, agg_position_info,
24672 predicate): Move to ipa-predicate.h
24673 * ipa-predicate.c: New file.
24674 * ipa-predicate.h: New file.
24676 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24678 * final.c (leaf_function_p): Check we are not in a sequence.
24680 2017-05-18 Martin Liska <mliska@suse.cz>
24682 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24683 * dumpfile.c (dump_register): Use new enum dump_kind.
24684 (get_dump_file_name): Likewise.
24685 (dump_enable_all): Likewise.
24686 (dump_switch_p_1): Likewise.
24687 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24688 * dumpfile.h (enum dump_kind): New enum type.
24689 (struct dump_file_info): Create constructor and
24690 format fields and comments.
24691 * passes.c (pass_manager::register_one_dump_file):
24693 * statistics.c (statistics_early_init): Likewise.
24694 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24695 TDF_TREE with TDF_SLIM.
24696 (gather_memory_references_ref): Likewise.
24698 2017-05-18 Martin Liska <mliska@suse.cz>
24700 * vec.h (struct vnull): Use it.
24702 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24704 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24705 (true_predicate, false_predicate, true_predicate_p,
24706 false_predicate_p): Remove.
24707 (single_cond_predicate, not_inlined_predicate): Turn to member function
24709 (add_condition): Update.
24710 (add_clause): Turn to...
24711 (predicate::add_clause): ... this one; update; allow passing NULL
24713 (and_predicates): Turn to ...
24714 (predicate::operator &=): ... this one.
24715 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24716 (or_predicates): Turn to ...
24717 (predicate::or_with): ... this one.
24718 (evaluate_predicate): Turn to ...
24719 (predicate::evaluate): ... this one.
24720 (predicate_probability): Turn to ...
24721 (predicate::probability): ... this one.
24722 (dump_condition): Update.
24723 (dump_predicate): Turn to ...
24724 (predicate::dump): ... this one.
24725 (account_size_time): Update.
24726 (edge_set_predicate): Update.
24727 (set_hint_predicate): UPdate.
24728 (evaluate_conditions_for_known_args): Update.
24729 (evaluate_properties_for_edge): Update.
24730 (remap_predicate_after_duplication): Turn to...
24731 (predicate::remap_after_duplication): ... this one.
24732 (remap_hint_predicate_after_duplication): Update.
24733 (inline_summary_t::duplicate): UPdate.
24734 (dump_inline_edge_summary): Update.
24735 (dump_inline_summary): Update.
24736 (set_cond_stmt_execution_predicate): Update.
24737 (set_switch_stmt_execution_predicate): Update.
24738 (compute_bb_predicates): Update.
24739 (will_be_nonconstant_expr_predicate): Update.
24740 (will_be_nonconstant_predicate): Update.
24741 (phi_result_unknown_predicate): Update.
24742 (predicate_for_phi_result): Update.
24743 (array_index_predicate): Update.
24744 (estimate_function_body_sizes): Update.
24745 (estimate_node_size_and_time): Update.
24746 (estimate_ipcp_clone_size_and_time): Update.
24747 (remap_predicate): Rename to ...
24748 (predicate::remap_after_inlining): ... this one.
24749 (remap_hint_predicate): Update.
24750 (inline_merge_summary): Update.
24751 (inline_update_overall_summary): Update.
24752 (estimate_size_after_inlining): Update.
24753 (read_predicate): Rename to ...
24754 (predicate::stream_in): ... this one.
24755 (read_inline_edge_summary): Update.
24756 (write_predicate): Rename to ...
24757 (predicate::stream_out): ... this one.
24758 (write_inline_edge_summary): Update.
24759 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24760 (clause_t): Turn to uint32_t
24761 (predicate): Turn to class; implement constructor and operators
24763 (size_time_entry): Update.
24764 (inline_summary): Update.
24765 (inline_edge_summary): Update.
24767 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24769 * fold-const.c (fold_binary_loc): Move transformation...
24770 * match.pd (C - X CMP X): ... here.
24772 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24774 * config/sparc/sparc.c (sparc_option_override): Set function
24775 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24776 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24778 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24780 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24782 2017-05-18 Marek Polacek <polacek@redhat.com>
24785 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24786 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24787 gimple_assign_load_p.
24789 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24791 PR middle-end/80692
24792 * real.c (do_compare): Give decimal_do_compare preference over
24793 comparing just the signs.
24795 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24797 * doc/md.texi (Canonicalization of Instructions): Describe the
24798 canonical form of instructions that inherently set a condition
24801 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24803 PR middle-end/80775
24804 * tree-cfg.c: Move deletion of unreachable case statements to after
24805 the merging of consecutive case labels.
24807 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24809 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24810 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24811 restoring of callee-saved registers.
24813 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24815 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24816 * config/visium/visium.c (single_set_and_flags): Likewise.
24817 * config/visium/visium.md (Substitutions): Likewise.
24819 2017-05-17 Martin Liska <mliska@suse.cz>
24821 * cfg.c: Introduce dump_flags_t type and
24822 use it instead of int type.
24824 * cfghooks.c: Likewise.
24825 * cfghooks.h (struct cfg_hooks): Likewise.
24826 * cfgrtl.c: Likewise.
24827 * cfgrtl.h: Likewise.
24828 * cgraph.c (cgraph_node::get_body): Likewise.
24829 * coretypes.h: Likewise.
24830 * domwalk.c: Likewise.
24831 * domwalk.h: Likewise.
24832 * dumpfile.c (struct dump_option_value_info): Likewise.
24833 (dump_enable_all): Likewise.
24834 (dump_switch_p_1): Likewise.
24835 (opt_info_switch_p): Likewise.
24836 * dumpfile.h (enum tree_dump_index): Likewise.
24837 (struct dump_file_info): Likewise.
24838 * genemit.c: Likewise.
24839 * generic-match-head.c: Likewise.
24840 * gengtype.c (open_base_files): Likewise.
24841 * gimple-pretty-print.c: Likewise.
24842 * gimple-pretty-print.h: Likewise.
24843 * graph.c (print_graph_cfg): Likewise.
24844 * graphite-scop-detection.c (dot_all_sese): Likewise.
24845 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24846 * loop-unroll.c (report_unroll): Likewise.
24847 * passes.c (pass_manager::register_one_dump_file): Likewise.
24848 * print-tree.c: Likewise.
24849 * statistics.c: Likewise.
24850 * tree-cfg.c: Likewise.
24851 * tree-cfg.h: Likewise.
24852 * tree-dfa.c: Likewise.
24853 * tree-dfa.h: Likewise.
24854 * tree-dump.c (dump_function): Likewise.
24855 * tree-dump.h (struct dump_info): Likewise.
24856 * tree-pretty-print.c: Likewise.
24857 * tree-pretty-print.h: Likewise.
24858 * tree-ssa-live.c: Likewise.
24859 * tree-ssa-live.h: Likewise.
24860 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24861 * tree-vect-loop.c: Likewise.
24862 * tree-vect-slp.c: Likewise.
24864 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24865 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24867 PR tree-optimization/80457
24868 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24869 of all arguments to a statement as scalar_to_vec operations.
24870 (vectorizable_call): Adjust call to vect_model_simple_cost for
24872 (vectorizable_conversion): Likewise.
24873 (vectorizable_assignment): Likewise.
24874 (vectorizable_shift): Likewise.
24875 (vectorizable_operation): Likewise.
24876 (vectorizable_comparison): Likewise.
24877 (vect_is_simple_cond): Record the def types for operands.
24878 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24879 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24880 for statement argument count.
24882 2017-05-16 Carl Love <cel@us.ibm.com>
24884 * config/rs6000/rs6000-c: Add support for built-in functions
24885 vector unsigned long long vec_bperm (vector unsigned long long,
24886 vector unsigned char)
24887 vector signed long long vec_mule (vector signed int,
24889 vector unsigned long long vec_mule (vector unsigned int,
24890 vector unsigned int)
24891 vector signed long long vec_mulo (vector signed int,
24893 vector unsigned long long vec_mulo (vector unsigned int,
24894 vector unsigned int)
24895 vector signed char vec_sldw (vector signed char,
24896 vector signed char,
24898 vector unsigned char vec_sldw (vector unsigned char,
24899 vector unsigned char,
24901 vector signed short vec_sldw (vector signed short,
24902 vector signed short,
24904 vector unsigned short vec_sldw (vector unsigned short,
24905 vector unsigned short,
24907 vector signed int vec_sldw (vector signed int,
24910 vector unsigned int vec_sldw (vector unsigned int,
24911 vector unsigned int,
24913 vector signed long long vec_sldw (vector signed long long,
24914 vector signed long long,
24916 vector unsigned long long vec_sldw (vector unsigned long long,
24917 vector unsigned long long,
24919 * config/rs6000/rs6000-c: Add support for built-in functions
24920 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24921 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24922 * doc/extend.texi: Update the built-in documentation for the
24923 new built-in functions.
24925 2017-05-16 Marek Polacek <polacek@redhat.com>
24929 * tree.c (save_expr): Don't fold the expression.
24931 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24933 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24934 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24935 and (?*y,m). Update insn attributes.
24937 2017-05-16 Martin Liska <mliska@suse.cz>
24939 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24940 flags argument of print_gimple_stmt, print_gimple_expr,
24941 print_generic_stmt and print_generic_expr.
24942 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24943 * coretypes.h: Likewise.
24944 * except.c (dump_eh_tree): Likewise.
24945 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24946 * gimple-pretty-print.h: Likewise.
24947 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24948 (backprop::push_to_worklist): Likewise.
24949 (backprop::pop_from_worklist): Likewise.
24950 (backprop::process_use): Likewise.
24951 (backprop::intersect_uses): Likewise.
24952 (note_replacement): Likewise.
24953 * gimple-ssa-store-merging.c
24954 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24955 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24956 (pass_store_merging::execute): Likewise.
24957 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24958 (ssa_base_cand_dump_callback): Likewise.
24959 (dump_incr_vec): Likewise.
24960 (replace_refs): Likewise.
24961 (replace_mult_candidate): Likewise.
24962 (create_add_on_incoming_edge): Likewise.
24963 (create_phi_basis): Likewise.
24964 (insert_initializers): Likewise.
24965 (all_phi_incrs_profitable): Likewise.
24966 (introduce_cast_before_cand): Likewise.
24967 (replace_one_candidate): Likewise.
24968 * gimplify.c (gimplify_expr): Likewise.
24969 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24970 (set_rename): Likewise.
24971 (rename_uses): Likewise.
24972 (copy_loop_phi_nodes): Likewise.
24973 (add_close_phis_to_merge_points): Likewise.
24974 (copy_loop_close_phi_args): Likewise.
24975 (copy_cond_phi_args): Likewise.
24976 (graphite_copy_stmts_from_block): Likewise.
24977 (translate_pending_phi_nodes): Likewise.
24978 * graphite-poly.c (print_pdr): Likewise.
24979 (dump_gbb_cases): Likewise.
24980 (dump_gbb_conditions): Likewise.
24981 (print_scop_params): Likewise.
24982 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24983 (build_cross_bb_scalars_use): Likewise.
24984 (gather_bbs::before_dom_children): Likewise.
24985 * hsa-dump.c (dump_hsa_immed): Likewise.
24986 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24987 (get_replacement_map): Likewise.
24988 * ipa-inline-analysis.c (dump_condition): Likewise.
24989 (estimate_function_body_sizes): Likewise.
24990 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24991 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24992 * ipa-prop.c (ipa_dump_param): Likewise.
24993 (ipa_print_node_jump_functions_for_edge): Likewise.
24994 (ipa_modify_call_arguments): Likewise.
24995 (ipa_modify_expr): Likewise.
24996 (ipa_dump_param_adjustments): Likewise.
24997 (ipa_dump_agg_replacement_values): Likewise.
24998 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24999 * ipa-pure-const.c (check_stmt): Likewise.
25000 (pass_nothrow::execute): Likewise.
25001 * ipa-split.c (execute_split_functions): Likewise.
25002 * omp-offload.c (dump_oacc_loop_part): Likewise.
25003 (dump_oacc_loop): Likewise.
25004 * trans-mem.c (tm_log_emit): Likewise.
25005 (tm_memopt_accumulate_memops): Likewise.
25006 (dump_tm_memopt_set): Likewise.
25007 (dump_tm_memopt_transform): Likewise.
25008 * tree-cfg.c (gimple_verify_flow_info): Likewise.
25009 (print_loop): Likewise.
25010 * tree-chkp-opt.c (chkp_print_addr): Likewise.
25011 (chkp_gather_checks_info): Likewise.
25012 (chkp_get_check_result): Likewise.
25013 (chkp_remove_check_if_pass): Likewise.
25014 (chkp_use_outer_bounds_if_possible): Likewise.
25015 (chkp_reduce_bounds_lifetime): Likewise.
25016 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
25017 (chkp_mark_completed_bounds): Likewise.
25018 (chkp_register_incomplete_bounds): Likewise.
25019 (chkp_mark_invalid_bounds): Likewise.
25020 (chkp_maybe_copy_and_register_bounds): Likewise.
25021 (chkp_build_returned_bound): Likewise.
25022 (chkp_get_bound_for_parm): Likewise.
25023 (chkp_build_bndldx): Likewise.
25024 (chkp_get_bounds_by_definition): Likewise.
25025 (chkp_generate_extern_var_bounds): Likewise.
25026 (chkp_get_bounds_for_decl_addr): Likewise.
25027 * tree-chrec.c (chrec_apply): Likewise.
25028 * tree-data-ref.c (dump_data_reference): Likewise.
25029 (dump_subscript): Likewise.
25030 (dump_data_dependence_relation): Likewise.
25031 (analyze_overlapping_iterations): Likewise.
25032 * tree-inline.c (expand_call_inline): Likewise.
25033 (tree_function_versioning): Likewise.
25034 * tree-into-ssa.c (dump_defs_stack): Likewise.
25035 (dump_currdefs): Likewise.
25036 (dump_names_replaced_by): Likewise.
25037 (dump_update_ssa): Likewise.
25038 (update_ssa): Likewise.
25039 * tree-object-size.c (pass_object_sizes::execute): Likewise.
25040 * tree-parloops.c (build_new_reduction): Likewise.
25041 (try_create_reduction_list): Likewise.
25042 (ref_conflicts_with_region): Likewise.
25043 (oacc_entry_exit_ok_1): Likewise.
25044 (oacc_entry_exit_single_gang): Likewise.
25045 * tree-pretty-print.h: Likewise.
25046 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
25047 (get_scalar_evolution): Likewise.
25048 (add_to_evolution): Likewise.
25049 (get_loop_exit_condition): Likewise.
25050 (analyze_evolution_in_loop): Likewise.
25051 (analyze_initial_condition): Likewise.
25052 (analyze_scalar_evolution): Likewise.
25053 (instantiate_scev): Likewise.
25054 (number_of_latch_executions): Likewise.
25055 (gather_chrec_stats): Likewise.
25056 (final_value_replacement_loop): Likewise.
25057 (scev_const_prop): Likewise.
25058 * tree-sra.c (dump_access): Likewise.
25059 (disqualify_candidate): Likewise.
25060 (create_access): Likewise.
25061 (reject): Likewise.
25062 (maybe_add_sra_candidate): Likewise.
25063 (create_access_replacement): Likewise.
25064 (analyze_access_subtree): Likewise.
25065 (analyze_all_variable_accesses): Likewise.
25066 (sra_modify_assign): Likewise.
25067 (initialize_constant_pool_replacements): Likewise.
25068 (find_param_candidates): Likewise.
25069 (decide_one_param_reduction): Likewise.
25070 (replace_removed_params_ssa_names): Likewise.
25071 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
25072 * tree-ssa-copy.c (dump_copy_of): Likewise.
25073 (copy_prop_visit_cond_stmt): Likewise.
25074 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
25075 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
25076 (record_equivalences_from_stmt): Likewise.
25077 * tree-ssa-dse.c (compute_trims): Likewise.
25078 (delete_dead_call): Likewise.
25079 (delete_dead_assignment): Likewise.
25080 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
25081 (forward_propagate_into_cond): Likewise.
25082 (pass_forwprop::execute): Likewise.
25083 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
25084 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
25086 (move_computations_worker): Likewise.
25087 (execute_sm): Likewise.
25088 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
25089 (remove_exits_and_undefined_stmts): Likewise.
25090 (remove_redundant_iv_tests): Likewise.
25091 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
25092 (adjust_iv_update_pos): Likewise.
25093 * tree-ssa-math-opts.c (bswap_replace): Likewise.
25094 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
25095 (value_replacement): Likewise.
25096 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
25097 * tree-ssa-pre.c (print_pre_expr): Likewise.
25098 (get_representative_for): Likewise.
25099 (create_expression_by_pieces): Likewise.
25100 (insert_into_preds_of_block): Likewise.
25101 (eliminate_insert): Likewise.
25102 (eliminate_dom_walker::before_dom_children): Likewise.
25103 (eliminate): Likewise.
25104 (remove_dead_inserted_code): Likewise.
25105 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
25106 * tree-ssa-reassoc.c (get_rank): Likewise.
25107 (eliminate_duplicate_pair): Likewise.
25108 (eliminate_plus_minus_pair): Likewise.
25109 (eliminate_not_pairs): Likewise.
25110 (undistribute_ops_list): Likewise.
25111 (eliminate_redundant_comparison): Likewise.
25112 (update_range_test): Likewise.
25113 (optimize_range_tests_var_bound): Likewise.
25114 (optimize_vec_cond_expr): Likewise.
25115 (rewrite_expr_tree): Likewise.
25116 (rewrite_expr_tree_parallel): Likewise.
25117 (linearize_expr): Likewise.
25118 (break_up_subtract): Likewise.
25119 (linearize_expr_tree): Likewise.
25120 (attempt_builtin_powi): Likewise.
25121 (attempt_builtin_copysign): Likewise.
25122 (transform_stmt_to_copy): Likewise.
25123 (transform_stmt_to_multiply): Likewise.
25124 (dump_ops_vector): Likewise.
25125 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
25126 (print_scc): Likewise.
25127 (set_ssa_val_to): Likewise.
25128 (visit_reference_op_store): Likewise.
25129 (visit_use): Likewise.
25130 (sccvn_dom_walker::before_dom_children): Likewise.
25131 (run_scc_vn): Likewise.
25132 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
25134 (expr_hash_elt::print): Likewise.
25135 (const_and_copies::pop_to_marker): Likewise.
25136 (const_and_copies::record_const_or_copy_raw): Likewise.
25137 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
25138 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
25139 (dump_predicates): Likewise.
25140 (find_uninit_use): Likewise.
25141 (warn_uninitialized_phi): Likewise.
25142 (pass_late_warn_uninitialized::execute): Likewise.
25143 * tree-ssa.c (verify_vssa): Likewise.
25144 (verify_ssa): Likewise.
25145 (maybe_optimize_var): Likewise.
25146 * tree-vrp.c (dump_value_range): Likewise.
25147 (dump_all_value_ranges): Likewise.
25148 (dump_asserts_for): Likewise.
25149 (register_edge_assert_for_2): Likewise.
25150 (vrp_visit_cond_stmt): Likewise.
25151 (vrp_visit_switch_stmt): Likewise.
25152 (vrp_visit_stmt): Likewise.
25153 (vrp_visit_phi_node): Likewise.
25154 (simplify_cond_using_ranges_1): Likewise.
25155 (fold_predicate_in): Likewise.
25156 (evrp_dom_walker::before_dom_children): Likewise.
25157 (evrp_dom_walker::push_value_range): Likewise.
25158 (evrp_dom_walker::pop_value_range): Likewise.
25159 (execute_early_vrp): Likewise.
25161 2017-05-16 Richard Biener <rguenther@suse.de>
25163 * dwarf2out.c (loc_list_from_tree_1): Do not create
25164 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
25166 2017-05-16 Richard Biener <rguenther@suse.de>
25168 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
25170 (note_variable_value_in_expr): If we resolved the decl ref
25171 do not push to the stack.
25173 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
25175 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
25176 operations in fast-math mode.
25177 (vaddq_f16): Likewise.
25178 (vmul_f16): Likewise.
25179 (vmulq_f16): Likewise.
25180 (vsub_f16): Likewise.
25181 (vsubq_f16): Likewise.
25182 * config/arm/neon.md (add<mode>3): New.
25184 (fma:<VH:mode>3): New. Also remove outdated comment.
25187 2017-05-16 Martin Liska <mliska@suse.cz>
25191 * ipa-devirt.c (warn_types_mismatch): Fix typo.
25192 (odr_types_equivalent_p): Likewise.
25194 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
25196 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
25198 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
25201 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
25202 non-interunit SSE move alternatives with '?'.
25203 (zero-extendsidi peephole2): New peephole to skip intermediate
25204 general register in SSE zero-extend sequence.
25206 2017-05-15 Jeff Law <law@redhat.com>
25208 * reorg.c (relax_delay_slots): Create a new variable to hold
25209 the temporary target rather than clobbering TARGET_LABEL.
25211 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
25212 missing argument to extract_bit_field call.
25213 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
25215 2017-05-15 Martin Liska <mliska@suse.cz>
25218 * gcc.c (process_command): Do not allow empty argument of -o option.
25220 2017-05-15 Renlin Li <renlin.li@arm.com>
25222 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
25223 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
25224 * config/aarch64/constraints.md (Usf): Add long call check.
25225 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
25226 (call_value): Likewise.
25227 (sibcall): Likewise.
25228 (sibcall_value): Likewise.
25230 (call_value_insn): New.
25231 (sibcall_insn): Update rtx pattern.
25232 (sibcall_value_insn): Likewise.
25233 (call_internal): Remove.
25234 (call_value_internal): Likewise.
25235 (sibcall_internal): Likewise.
25236 (sibcall_value_internal): Likewise.
25237 (call_reg): Likewise.
25238 (call_symbol): Likewise.
25239 (call_value_reg): Likewise.
25240 (call_value_symbol): Likewise.
25242 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
25245 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
25247 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
25249 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
25250 compatible with CCGOCmode and with CCZmode.
25252 2017-05-14 Martin Sebor <msebor@redhat.com>
25254 PR middle-end/77671
25255 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
25256 (gimple_fold_builtin_snprintf): Same.
25257 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
25258 (gimple_fold_builtin_snprintf): Same.
25259 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
25260 of character types.
25261 (is_call_safe): New function.
25262 (try_substitute_return_value): Call it.
25263 (try_simplify_call): New function.
25264 (pass_sprintf_length::handle_gimple_call): Call it.
25266 2017-05-14 Martin Sebor <msebor@redhat.com>
25268 PR middle-end/80669
25269 * builtins.c (expand_builtin_stpncpy): Simplify.
25271 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
25273 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
25274 * config/i386/i386.h
25275 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
25276 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
25277 (struct machine_function): Add new members call_ms2sysv,
25278 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
25279 (struct machine_frame_state): New fields sp_realigned and
25280 sp_realigned_offset.
25281 * config/i386/i386.c
25282 (enum xlogue_stub): New enum.
25283 (enum xlogue_stub_sets): New enum.
25284 (class xlogue_layout): New class.
25285 (struct ix86_frame): New fields stack_realign_allocate_offset,
25286 stack_realign_offset and outlined_save_offset. Modify comments to
25287 detail stack layout when using out-of-line stubs.
25288 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
25289 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
25290 -mcall-ms2sysv-xlogues.
25291 (stub_managed_regs): New static variable.
25292 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
25293 registers managed by out-of-line stub.
25294 (disable_call_ms2sysv_xlogues): New function.
25295 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
25296 m->call_ms2sysv when appropriate and compute frame layout for
25298 (sp_valid_at, fp_valid_at): New inline functions.
25299 (choose_basereg): New function.
25300 (choose_baseaddr): Add align parameter, use choose_basereg and modify
25302 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
25303 Use align parameter of choose_baseaddr to generated aligned SSE movs
25305 (pro_epilogue_adjust_stack): Modify to track
25306 machine_frame_state::sp_realigned.
25307 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
25308 (ix86_nsaved_sseregs): Likewise.
25309 (ix86_emit_save_regs): Likewise.
25310 (ix86_emit_save_regs_using_mov): Likewise.
25311 (ix86_emit_save_sse_regs_using_mov): Likewise.
25312 (get_scratch_register_on_entry): Likewise.
25313 (gen_frame_set): New function.
25314 (gen_frame_load): Likewise.
25315 (gen_frame_store): Likewise.
25316 (emit_outlined_ms2sysv_save): Likewise.
25317 (emit_outlined_ms2sysv_restore): Likewise.
25318 (ix86_expand_prologue): Modify stack re-alignment code and call
25319 emit_outlined_ms2sysv_save when appropriate.
25320 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
25321 parameter rtx_insn *insn, which allows the function to be used to only
25322 generate the notes.
25323 (ix86_expand_epilogue): Modify validity checks of frame and stack
25324 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
25325 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
25326 * config/i386/predicates.md
25327 (save_multiple): New predicate.
25328 (restore_multiple): Likewise.
25329 * config/i386/sse.md
25330 (save_multiple<mode>): New pattern.
25331 (save_multiple_realign<mode>): Likewise.
25332 (restore_multiple<mode>): Likewise.
25333 (restore_multiple_and_return<mode>): Likewise.
25334 (restore_multiple_leave_return<mode>): Likewise.
25335 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
25337 2017-05-14 Julia Koval <julia.koval@intel.com>
25339 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
25340 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
25341 (__builtin_ia32_xsetbv): New builtins.
25342 * config/i386/i386.c (ix86_expand_special_args_builtin):
25344 (ix86_expand_builtin): Special expand for new intrinsics.
25345 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
25346 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
25347 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
25349 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25351 * cfganal.c (inverted_post_order_compute): Change argument type
25353 * cfganal.h (inverted_post_order_compute): Adjust prototype.
25354 * df-core.c (rest_of_handle_df_initialize): Adjust.
25355 (rest_of_handle_df_finish): Likewise.
25356 (df_analyze_1): Likewise.
25357 (df_analyze): Likewise.
25358 (loop_inverted_post_order_compute): Change argument to be a vec *.
25359 (df_analyze_loop): Adjust.
25360 (df_get_n_blocks): Likewise.
25361 (df_get_postorder): Likewise.
25362 * df.h (struct df_d): Change field to be a vec.
25363 * lcm.c (compute_laterin): Adjust.
25364 (compute_available): Likewise.
25365 * lra-lives.c (lra_create_live_ranges_1): Likewise.
25366 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
25367 * tree-ssa-pre.c (compute_antic): Likewise.
25369 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25371 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
25372 (depth_first_search::depth_first_search): Change structure init
25373 function to this constructor.
25374 (depth_first_search::add_bb): Rename function to this member.
25375 (depth_first_search::execute): Likewise.
25376 (flow_dfs_compute_reverse_finish): Adjust.
25378 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25380 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
25381 (longest_simple_path): Likewise.
25382 * shrink-wrap.c (spread_components): Likewise.
25383 (disqualify_problematic_components): Likewise.
25384 (emit_common_heads_for_components): Likewise.
25385 (emit_common_tails_for_components): Likewise.
25386 (insert_prologue_epilogue_for_components): Likewise.
25388 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25390 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
25393 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25395 * df-core.c (df_set_blocks): Start using auto_bitmap.
25396 (df_compact_blocks): Likewise.
25397 * df-problems.c (df_rd_confluence_n): Likewise.
25398 * df-scan.c (df_insn_rescan_all): Likewise.
25399 (df_process_deferred_rescans): Likewise.
25400 (df_update_entry_block_defs): Likewise.
25401 (df_update_exit_block_uses): Likewise.
25402 (df_entry_block_bitmap_verify): Likewise.
25403 (df_exit_block_bitmap_verify): Likewise.
25404 (df_scan_verify): Likewise.
25405 * lra-constraints.c (lra_constraints): Likewise.
25406 (undo_optional_reloads): Likewise.
25407 (lra_undo_inheritance): Likewise.
25408 * lra-remat.c (calculate_gen_cands): Likewise.
25409 (do_remat): Likewise.
25410 * lra-spills.c (assign_spill_hard_regs): Likewise.
25411 (spill_pseudos): Likewise.
25412 * tree-ssa-pre.c (bitmap_set_and): Likewise.
25413 (bitmap_set_subtract_values): Likewise.
25415 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25417 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
25418 management with auto_bitmap.
25419 (fix_inter_tick): Likewise.
25420 (fix_recovery_deps): Likewise.
25421 * ira.c (add_store_equivs): Likewise.
25422 (find_moveable_pseudos): Likewise.
25423 (split_live_ranges_for_shrink_wrap): Likewise.
25424 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
25425 (rtx_reuse_manager::seen_def_p): Likewise.
25426 (rtx_reuse_manager::set_seen_def): Likewise.
25427 * print-rtl.h (class rtx_reuse_manager): Likewise.
25429 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25431 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
25433 (migrate_btr_def): Likewise.
25434 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
25435 * df-core.c (loop_post_order_compute): Likewise.
25436 (loop_inverted_post_order_compute): Likewise.
25437 * hsa-common.h: Likewise.
25438 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
25439 * init-regs.c (initialize_uninitialized_regs): Likewise.
25440 * ipa-inline.c (resolve_noninline_speculation): Likewise.
25441 (inline_small_functions): Likewise.
25442 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
25443 * ira.c (combine_and_move_insns): Likewise.
25444 (build_insn_chain): Likewise.
25445 * loop-invariant.c (find_invariants): Likewise.
25446 * lower-subreg.c (propagate_pseudo_copies): Likewise.
25447 * predict.c (tree_predict_by_opcode): Likewise.
25448 (predict_paths_leading_to): Likewise.
25449 (predict_paths_leading_to_edge): Likewise.
25450 (estimate_loops_at_level): Likewise.
25451 (estimate_loops): Likewise.
25452 * shrink-wrap.c (try_shrink_wrapping): Likewise.
25453 (spread_components): Likewise.
25454 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
25455 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
25456 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
25457 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
25458 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
25459 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
25460 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
25461 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
25462 (mark_threaded_blocks): Likewise.
25463 (thread_through_all_blocks): Likewise.
25464 * tree-ssa.c (verify_ssa): Likewise.
25465 (execute_update_addresses_taken): Likewise.
25466 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
25468 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25470 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
25472 (post_order_compute): Likewise.
25473 (inverted_post_order_compute): Likewise.
25474 (pre_and_rev_post_order_compute_fn): Likewise.
25476 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25478 * genrecog.c (int_set::int_set): Explicitly construct our
25479 auto_vec base class.
25480 * vec.h (auto_vec::auto_vec): New constructor.
25482 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25484 * bitmap.h (class auto_bitmap): New constructor taking
25485 bitmap_obstack * argument.
25487 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25489 * bitmap.h (class auto_bitmap): Change type of m_bits to
25490 bitmap_head, and adjust ctor / dtor and member operators.
25492 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
25494 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
25495 when returned register mode doesn't match original mode.
25497 2017-05-12 Jeff Law <law@redhat.com>
25498 Jakub Jelinek <jakub@redhat.com>
25500 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
25501 we look for cc setter after the compare-elim changes.
25502 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
25503 within the vector to match what compare-elim now expects.
25504 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
25505 (xorsi3_flags, one_cmplsi2_flags): Likewise.
25507 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
25508 after the compare-elim changes.
25509 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
25510 the vector to match what compare-elim now expects.
25511 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
25512 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
25513 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
25514 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
25515 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
25517 * config/visium/visium.c (single_set_and_flags): Fix where
25518 we look for cc setter after the compare-elim changes.
25519 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
25520 with the vector to match what compare-elim now expects.
25521 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
25522 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
25523 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
25524 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
25525 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
25526 (neg<mode>2_insn_set_overflow): Likewise.
25528 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
25530 PR middle-end/79794
25531 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
25532 maybe_expand_insn call, set ops[0].target. If still set after call,
25533 set alt_rtl. Add extra arg to recursive calls.
25534 (extract_bit_field): Add alt_rtl argument. Pass to
25536 * expmed.h (extract_bit_field): Fix prototype.
25537 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
25538 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
25539 to extract_bit_field_calls.
25540 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
25541 Pass alt_rtl to extract_bit_field calls.
25542 * calls.c (store_unaligned_arguments_into_psuedos)
25543 load_register_parameters): Pass extra NULL to extract_bit_field calls.
25544 * optabs.c (maybe_legitimize_operand): Clear op->target when call
25546 * optabs.h (struct expand_operand): Add target bitfield.
25548 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25550 * compare-elim.c (try_eliminate_compare): Canonicalize
25551 operation with embedded compare to
25552 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
25553 (set (reg) (operation)].
25555 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
25557 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25560 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
25561 cost of adding a carry flag for ADC instruction.
25562 [case MINUS]: Ignore the cost of subtracting a carry flag
25563 for SBB instruction.
25565 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
25567 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
25569 * config/rs6000/bmiintrin.h: New file.
25570 * config/rs6000/bmi2intrin.h: New file.
25571 * config/rs6000/x86intrin.h: New file.
25573 2017-05-12 Jeff Law <law@redhat.com>
25575 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
25578 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
25580 PR middle-end/80707
25581 * tree-cfg.c: Remove cfg edges of unreachable case statements.
25583 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25585 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25586 early expansion of vector divide builtins.
25587 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
25588 builtins identified as having unsigned arguments.
25590 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25592 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
25593 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
25594 expansion of vector logical operations (and, andc, or, xor,
25597 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25599 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
25600 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
25602 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25604 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25605 early GIMPLE expansion of vector multiplies.
25607 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25609 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
25610 TARGET_HAVE_MOVT conditional.
25611 (movt splitter): Likewise.
25613 2017-05-12 Richard Biener <rguenther@suse.de>
25615 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
25616 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25617 Fold all stmts not inplace.
25619 2017-05-12 Richard Biener <rguenther@suse.de>
25621 PR tree-optimization/80713
25622 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
25623 inserted_exprs bit for not removed stmts.
25625 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
25627 PR middle-end/69921
25628 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
25629 parallelized" attribute for parallelized OpenACC kernels.
25630 * omp-offload.c (execute_oacc_device_lower): Use it.
25632 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
25633 Set "oacc kernels" attribute.
25634 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
25635 parameter. Adjust all users.
25636 (oacc_fn_attrib_kernels_p): Remove function.
25637 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
25638 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
25639 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
25640 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
25641 assert "oacc kernels" attribute is set.
25643 2017-05-11 Carl Love <cel@us.ibm.com>
25645 * config/rs6000/rs6000-c: Add support for built-in functions
25646 vector unsigned char vec_popcnt (vector signed char)
25647 vector unsigned char vec_popcnt (vector unsigned char)
25648 vector unsigned short vec_popcnt (vector signed short)
25649 vector unsigned short vec_popcnt (vector unsigned short)
25650 vector unsigned int vec_popcnt (vector signed int)
25651 vector unsigned int vec_popcnt (vector unsigned int)
25652 vector unsigned long long vec_popcnt (vector signed long long)
25653 vector unsigned long long vec_popcnt (vector unsigned long long)
25654 vector signed long long vec_slo (vector signed long long,
25655 vector signed char)
25656 vector signed long long vec_slo (vector signed long long,
25657 vector unsigned char)
25658 vector unsigned long long vec_slo (vector unsigned long long,
25659 vector signed char)
25660 vector unsigned long long vec_slo (vector unsigned long long,
25661 vector unsigned char)
25662 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25663 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25664 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25665 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25666 * doc/extend.texi: Update the built-in documentation file for the
25667 new built-in functions.
25669 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25671 * attribs.h (sorted_attr_string): Move machine independent
25672 functions for target clone support from the i386 port to common
25673 code. Rename ix86_function_versions to common_function_versions.
25674 Rename make_name to make_unique_name.
25675 (common_function_versions): Likewise.
25676 (make_unique_name): Likewise.
25677 (make_dispatcher_decl): Likewise.
25678 (is_function_default_version): Likewise.
25679 * attribs.c (attr_strcmp): Likewise.
25680 (sorted_attr_string): Likewise.
25681 (common_function_versions): Likewise.
25682 (make_unique_name): Likewise.
25683 (make_dispatcher_decl): Likewise.
25684 (is_function_default_version): Likewise.
25685 * config/i386/i386.c (attr_strcmp): Likewise.
25686 (sorted_attr_string): Likewise.
25687 (ix86_function_versions): Likewise.
25688 (make_name): Likewise.
25689 (make_dispatcher_decl): Likewise.
25690 (is_function_default_version): Likewise.
25691 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25693 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25696 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25697 Account for direct move costs for vec_construct of integer
25700 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25703 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25704 (UNSPEC_STX_ATOMIC): Ditto.
25705 (loaddi_via_sse): New insn.
25706 (storedi_via_sse): Ditto.
25707 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25708 Update corresponding peephole2 patterns.
25709 (atomic_storedi_fpu): Ditto.
25711 2017-05-11 Julia Koval <julia.koval@intel.com>
25713 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25714 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25716 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25717 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25718 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25720 2017-05-11 Nathan Sidwell <nathan@acm.org>
25722 * graphite-poly.c: Include dumpfile.h.
25724 * dumpfle.h (dump_function): Declare here ...
25725 * tree-dump.h (dump_function): ... not here.
25726 * dumpfile.c: #include tree-cfg.h.
25727 (dump_function): Move here from ...
25728 * tree-dump.c (dump_function): ... here.
25729 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25730 * graphite-poly.c: Don't include tree-dump.h.
25731 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25732 * print-tree.c: Likewise.
25733 * stor-layout.c: Likewise.
25734 * tree-nested.c: Likewise.
25736 * dumpfile.c (dump_start): Use TDF_FLAGS.
25737 (dump_enable_all): Fix TDF_KIND check thinko.
25739 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25741 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25742 array entries to represent two legal parameterizations of the
25743 overloaded __builtin_cmpb function, as represented by the
25744 P6_OV_BUILTIN_CMPB constant.
25745 (altivec_resolve_overloaded_builtin): Add special case handling
25746 for the __builtin_cmpb function, as represented by the
25747 P6_OV_BUILTIN_CMPB constant.
25748 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25749 (BU_P6_64BIT_2): New macro.
25750 (BU_P6_OVERLOAD_2): New macro
25751 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25752 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25753 (CMPB): Add overload support to represent both 32-bit and 64-bit
25754 compare-bytes function.
25755 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25756 support for TARGET_CMPB.
25757 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25758 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25759 documentation of the __builtin_cmpb overloaded built-in function.
25761 2017-05-11 Richard Biener <rguenther@suse.de>
25763 PR tree-optimization/80705
25764 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25765 bases are not vectorizable.
25767 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25769 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25770 when counting register pressure.
25772 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25774 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25775 (iv_ca_more_deps): Renamed to ...
25776 (iv_ca_compare_deps): ... this.
25777 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25779 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25781 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25783 (determine_group_iv_costs): ... here.
25784 (find_inv_vars_cb): Record inv var if it's not recorded before.
25786 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25788 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25789 (get_shiftadd_cost): Ditto.
25791 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25793 * tree-ssa-address.c: Include header file.
25794 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25796 (add_to_parts): Refactor.
25797 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25798 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25801 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25803 PR tree-optimization/53090
25804 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25806 (extract_cond_operands): Detect condition with IV on both sides
25807 and return COMP_IV_EXPR_2.
25808 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25809 (rewrite_use_compare): Simplify by removing call to function
25810 extract_cond_operands.
25812 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25814 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25815 (extract_cond_operands): Detect condition comparing against non-
25816 invariant bound and return appropriate enum value.
25817 (find_interesting_uses_cond): Update use of extract_cond_operands.
25818 Handle its return value accordingly.
25819 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25821 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25823 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25824 nonlinear iv_use computation in loop invariant sensitive way.
25826 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25828 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25829 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25831 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25833 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25834 (dump_cand): Support iv_cand.inv_exprs.
25835 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25837 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25840 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25842 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25844 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25845 as local function. Include necessary header files.
25846 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25848 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25850 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25852 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25854 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25855 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25856 RSHIFT_EXPR and BIT_NOT_EXPR.
25858 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25860 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25861 (adjust_setup_cost): New parameter supporting round up adjustment.
25862 (struct address_cost_data): Delete.
25863 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25864 (split_address_cost, ptr_difference_cost): Delete.
25865 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25866 (struct ainc_cost_data): New struct.
25867 (get_address_cost_ainc): New function.
25868 (get_address_cost, get_computation_cost): Reimplement.
25869 (determine_group_iv_cost_address): Record inv_expr for all uses of
25871 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25872 (iv_ca_has_deps): Reimplemented to ...
25873 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25875 (iv_ca_extend): Call iv_ca_more_deps.
25877 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25879 * tree-ssa-address.c (struct mem_address): Move to header file.
25880 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25881 * tree-ssa-address.h (struct mem_address): Move from C file.
25882 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25884 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25886 * tree-affine.h (aff_combination_type): New interface.
25887 (aff_combination_zero_p): Remove static.
25888 (aff_combination_const_p): New interface.
25889 (aff_combination_singleton_var_p): New interfaces.
25891 2017-05-11 Richard Biener <rguenther@suse.de>
25893 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25894 Skip unreachable blocks and destinations.
25895 (eliminate): Move stmt removal and fixup ...
25896 (fini_eliminate): ... here. Skip inserted exprs.
25897 (pass_pre::execute): Move fini_pre after fini_eliminate.
25898 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25899 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25900 PRE to get rid of dead code that has invalid SSA form and
25901 split critical edges again.
25903 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25905 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25907 2017-05-11 Richard Biener <rguenther@suse.de>
25909 * passes.c (execute_function_todo): Verify loops if they are
25910 said to be up-to-date.
25911 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25912 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25914 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25917 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25918 handle calling assemble_external ourself.
25921 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25922 modes with zero size. Enhance comment.
25924 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25926 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25927 built-ins for vec_xl and vec_xst with short and char pointer
25930 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25932 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25933 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25934 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25935 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25936 (_mm_maskz_min_round_ss): New intrinsics.
25937 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25938 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25939 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25940 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25941 (__builtin_ia32_minss_mask_round): New builtins.
25942 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25943 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25944 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25946 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25947 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25949 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25952 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25954 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25955 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25956 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25957 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25958 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25959 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25960 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25961 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25962 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25963 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25964 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25965 (__builtin_ia32_mulss_mask_round): New builtins.
25966 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25967 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25968 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25970 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25971 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25973 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25976 2017-05-10 Julia Koval <julia.koval@intel.com>
25978 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25979 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25980 (_mm256_setr_m128i): New intrinsics.
25982 2017-05-10 Julia Koval <julia.koval@intel.com>
25984 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25985 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25986 (_mm_maskz_rcp14_ss): New intrinsics.
25987 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25988 (__builtin_ia32_rcp14ss_mask): New builtins.
25989 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25991 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25993 PR tree-optimization/51513
25994 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25995 (assert_unreachable_fallthru_edge_p): Use it.
25996 (group_case_labels_stmt): Likewise.
25997 * tree-cfg.h: Prototype it.
25998 * stmt.c: Include cfghooks.h and tree-cfg.h.
25999 (emit_case_dispatch_table) <gap_label>: New local variable.
26000 Use it to fill dispatch table gaps.
26001 Test for default_label before updating probabilities.
26002 (expand_case) <default_label>: Remove unneeded initialization.
26003 Test for unreachable default case statement and remove its edge.
26004 Set default_label accordingly.
26005 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
26007 2017-05-10 Carl Love <cel@us.ibm.com>
26009 * config/rs6000/rs6000-c: Add support for built-in functions
26010 vector signed char vec_neg (vector signed char)
26011 vector signed short int vec_neg (vector short int)
26012 vector signed int vec_neg (vector signed int)
26013 vector signed long long vec_neg (vector signed long long)
26014 vector float vec_neg (vector float)
26015 vector double vec_neg (vector double)
26016 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
26018 * config/rs6000/altivec.h: Add define for vec_neg
26019 * doc/extend.texi: Update the built-in documentation for the
26020 new built-in functions.
26022 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26024 PR tree-optimization/77644
26025 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
26027 2017-05-10 Nathan Sidwell <nathan@acm.org>
26029 * dumpfile.h (TDI_lang_all): New.
26030 (TDF_KIND): New. Renumber others
26031 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
26033 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
26035 (get_dump_file_name): Adjust suffix generation.
26036 (dump_enable_all): Use TDF_KIND.
26037 * doc/invoke.texi (-fdump-lang-all): Document.
26039 * dumpfile.h: Tabify.
26041 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
26044 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
26045 Move member access before delete.
26047 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
26049 * tree-inline.c (expand_call_inline): Split block at stmt
26052 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
26055 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
26056 are now unused after splitting mov{sf,sd}_hardfloat.
26057 (f32_lr2): Likewise.
26058 (f32_lm): Likewise.
26059 (f32_lm2): Likewise.
26060 (f32_li): Likewise.
26061 (f32_li2): Likewise.
26062 (f32_lv): Likewise.
26063 (f32_sr): Likewise.
26064 (f32_sr2): Likewise.
26065 (f32_sm): Likewise.
26066 (f32_sm2): Likewise.
26067 (f32_si): Likewise.
26068 (f32_si2): Likewise.
26069 (f32_sv): Likewise.
26070 (f32_dm): Likewise.
26071 (f32_vsx): Likewise.
26072 (f32_av): Likewise.
26073 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
26074 For movsf, order stores so the VSX stores occur before the GPR
26075 store which encourages the register allocator to use a traditional
26076 FPR instead of a GPR. For movsd, order the stores so that the GPR
26077 store comes before the VSX stores to allow the power6 to work.
26078 This is due to the power6 not having a 32-bit integer store
26079 instruction from a FPR.
26080 (movsf_hardfloat): Likewise.
26081 (movsd_hardfloat): Likewise.
26083 2017-05-09 Martin Sebor <msebor@redhat.com>
26085 PR translation/80280
26086 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
26089 PR translation/80280
26090 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
26091 data member added in r247778.
26092 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
26094 2017-05-09 Nathan Sidwell <nathan@acm.org>
26096 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
26098 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
26101 2017-05-09 Marek Polacek <polacek@redhat.com>
26103 * doc/invoke.texi: Fix typo.
26105 2017-05-09 Richard Biener <rguenther@suse.de>
26107 * tree-vrp.c (vrp_val_is_max): Adjust comment.
26108 (vrp_val_is_min): Likewise.
26109 (set_value_range_to_value): Likewise.
26110 (set_value_range_to_nonnegative): Likewise.
26111 (gimple_assign_nonzero_p): Likewise.
26112 (gimple_stmt_nonzero_p): Likewise.
26113 (vrp_int_const_binop): Likewise. Remove unreachable case.
26114 (adjust_range_with_scev): Adjust comments.
26115 (compare_range_with_value): Likewise.
26116 (extract_range_from_phi_node): Likewise.
26117 (test_for_singularity): Likewise.
26119 2017-05-09 Richard Biener <rguenther@suse.de>
26121 * tree-vrp.c (get_single_symbol): Add assert that we don't
26122 get overflowed constants as invariant part.
26123 (compare_values_warnv): Add comment before the TREE_NO_WARNING
26124 checks. Use wi::cmp instead of recursing for integer constants.
26125 (compare_values): Just ignore whether we assumed undefined
26126 overflow instead of failing the compare.
26127 (extract_range_for_var_from_comparison_expr): Add comment before the
26128 TREE_NO_WARNING sets.
26129 (test_for_singularity): Likewise.
26130 (extract_range_from_comparison): Do not disable optimization
26131 when we assumed undefined overflow.
26132 (extract_range_basic): Remove init of unused var.
26134 2017-05-09 Richard Biener <rguenther@suse.de>
26136 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
26137 (extract_range_from_multiplicative_op_1): Adjust.
26138 (extract_range_from_binary_expr_1): Use int_const_binop.
26140 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
26143 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
26144 rs6000_store_data_bypass_p in seven define_bypass directives and
26145 in several comments.
26146 * config/rs6000/rs6000-protos.h: Add prototype for
26147 rs6000_store_data_bypass_p function.
26148 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
26149 function implements slightly different (rs6000-specific) semantics
26150 than store_data_bypass_p, returning false rather than aborting
26151 with assertion error when arguments do not satisfy the
26152 requirements of store data bypass.
26153 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
26154 rs6000_store_data_bypass_p.
26156 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
26158 * config/xtensa/xtensa-protos.h
26159 (xtensa_initial_elimination_offset): New declaration.
26160 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26161 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
26162 macro definition, add case for FRAME_POINTER_REGNUM when
26163 FRAME_GROWS_DOWNWARD.
26164 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
26165 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
26166 xtensa_initial_elimination_offset.
26168 2017-05-08 Nathan Sidwell <nathan@acm.org>
26170 * doc/invoke.texi: Alphabetize -fdump options.
26172 2017-05-08 Martin Sebor <msebor@redhat.com>
26174 PR translation/80280
26175 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
26177 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
26179 * target.def (compute_frame_layout): New optional target hook.
26180 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
26181 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
26182 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
26184 * reload1.c (verify_initial_elim_offsets): Likewise.
26185 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
26186 (use_simple_return_p): Call arm_compute_frame_layout if needed.
26187 (arm_get_frame_offsets): Split up into this ...
26188 (arm_compute_frame_layout): ... and this function.
26190 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
26192 * config/aarch64/constraints.md (Usa): New constraint.
26193 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
26195 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
26197 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
26198 with_multilib_list after it has been checked.
26200 2017-05-08 Richard Biener <rguenther@suse.de>
26202 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
26203 (bitmap_set_subtract_values): Likewise.
26205 2017-05-08 Richard Biener <rguenther@suse.de>
26207 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
26208 (gimple_assign_nonzero): ... this and remove strict_overflow_p
26210 (gimple_stmt_nonzero_warnv_p): Rename to ...
26211 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
26213 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
26214 (extract_range_basic): Adjust, do not disable propagation on
26215 strict overflow sensitive simplification.
26216 (vrp_visit_cond_stmt): Likewise.
26218 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
26220 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
26221 body size unconditionally.
26223 2017-05-07 Jeff Law <law@redhat.com>
26226 2017-05-06 Jeff Law <law@redhat.com>
26227 PR tree-optimization/78496
26228 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
26231 PR tree-optimization/78496
26232 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
26233 (simplify_stmt_using_ranges): Call it.
26234 (vrp_dom_walker::before_dom_children): Extract equivalences
26235 from an ASSERT_EXPR with an equality comparison against a
26238 2017-05-06 Jeff Law <law@redhat.com>
26240 PR tree-optimization/78496
26241 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
26244 PR tree-optimization/78496
26245 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
26246 (simplify_stmt_using_ranges): Call it.
26247 (vrp_dom_walker::before_dom_children): Extract equivalences
26248 from an ASSERT_EXPR with an equality comparison against a
26251 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
26253 * lra-constraints.c (lra_copy_reg_equiv): New function.
26254 (split_reg): Use it to copy equivalence information from the
26255 original register to the spill register.
26257 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
26259 PR rtl-optimization/75964
26260 * simplify-rtx.c (simplify_const_relational_operation): Remove
26261 invalid handling of comparisons of integer ABS.
26263 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
26265 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
26266 initialize to zero.
26267 (init_regs): Remove declaration.
26268 (function_arg_advance_32): Initialize error_p as boolean variable.
26270 2017-05-05 Nathan Sidwell <nathan@acm.org>
26272 * store-motion.c (remove_reachable_equiv_notes): Reformat long
26273 lines. Use for (;;).
26275 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26277 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
26278 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
26279 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
26280 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
26281 VF=2 that require versioning.
26283 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26285 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
26288 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26290 * diagnostic.h (diagnostic_override_option_index): Convert from
26291 macro to inline function.
26293 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26295 * diagnostic.c (last_module_changed_p): New function.
26296 (set_last_module): New function.
26297 (diagnostic_report_current_module): Convert macro usage to
26298 the above functions.
26299 * diagnostic.h (diagnostic_context::last_module): Strengthen
26300 from const line_map * to const line_map_ordinary *.
26301 (diagnostic_last_module_changed): Delete macro.
26302 (diagnostic_set_last_module): Delete macro.
26304 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26306 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
26307 with diagnostic_report_diagnostic.
26308 (diagnostic_n_impl_richloc): Likewise.
26309 * diagnostic.h (report_diagnostic): Delete macro.
26310 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
26311 with diagnostic_report_diagnostic.
26312 * substring-locations.c (format_warning_va): Likewise.
26314 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26316 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
26317 save/restor of format_spec. Move option-printing code to...
26318 (print_option_information): ...this new function, and
26319 reimplement by simply printing to the pretty_printer,
26320 rather than appending to the format string.
26322 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26324 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
26325 handling logic into...
26326 (update_effective_level_from_pragmas): ...this new function.
26328 2017-05-04 Andrew Waterman <andrew@sifive.com>
26330 * config/riscv/riscv.opt (mstrict-align): New option.
26331 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
26332 (SLOW_UNALIGNED_ACCESS): Define.
26333 (riscv_slow_unaligned_access): Declare.
26334 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
26336 (riscv_slow_unaligned_access): New variable.
26337 (rocket_tune_info): Set slow_unaligned_access to true.
26338 (optimize_size_tune_info): Set slow_unaligned_access to false.
26339 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
26340 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
26341 (riscv_option_override): Set riscv_slow_unaligned_access.
26342 * doc/invoke.texi: Add -mstrict-align to RISC-V.
26344 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
26346 * config/riscv/riscv.md: Unify indentation.
26348 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26353 * config/rs6000/rs6000.md (u code attribute): Add FIX and
26355 (extendsi<mode>2): Add support for doing sign extension via
26356 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
26357 don't have ISA 3.0 instructions.
26358 (extendsi<mode>2 splitter): Likewise.
26359 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
26360 generate the normal insns since SImode can now go in vector
26361 registers. Disallow the special UNSPECs needed for previous
26362 machines to hide SImode being used. Add new insns
26363 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
26364 (fix_trunc<mode>si2_stfiwx): Likewise.
26365 (fix_trunc<mode>si2_internal): Likewise.
26366 (fixuns_trunc<mode>si2): Likewise.
26367 (fixuns_trunc<mode>si2_stfiwx): Likewise.
26368 (fctiw<u>z_<mode>_smallint): Likewise.
26369 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
26370 of floating point to 32-bit integer from doing a direct move to
26371 the GPR registers to do a store.
26372 (fctiwz_<mode>): Break long line.
26374 2017-05-05 Bin Cheng <bin.cheng@arm.com>
26376 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
26377 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
26378 (addr_list, addr_offset_valid_p): New.
26379 (split_address_groups): Check offset validity with above function.
26380 (gt-tree-ssa-loop-ivopts.h): Include header file.
26382 2017-05-05 Nathan Sidwell <nathan@acm.org>
26384 * config.gcc (arm*-*-*): Add missing 'fi'.
26386 2017-05-05 Steve Ellcey <sellcey@cavium.com>
26388 * doc/invoke.texi (-fopt-info): Explicitly say order of options
26389 included in -fopt-info does not matter.
26390 * doc/optinfo.texi (-fopt-info): Fix description of default
26391 behavour. Explicitly say order of options included in -fopt-info
26394 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26396 * config.gcc: Allow combinations of aprofile and rmprofile values for
26397 --with-multilib-list.
26398 * config/arm/t-multilib: New file.
26399 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
26400 variables. Remove setting of ISA and floating-point ABI in
26401 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
26402 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
26403 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
26404 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
26406 * config/arm/t-rmprofile: Likewise except for the matches changes.
26407 * doc/install.texi (--with-multilib-list): Document the combination of
26408 aprofile and rmprofile values and warn about pitfalls in doing that.
26410 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26412 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
26413 (movdi_aarch64): Likewise.
26415 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26417 PR tree-optimization/80632
26418 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
26420 (build_arrays): Initialize it for virtual phis.
26421 (fix_phi_nodes): Use it for virtual phis.
26423 PR tree-optimization/80558
26424 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
26425 [x, y] op z into [x op, y op z] for op & or | if conditions
26428 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
26429 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26432 * config/arm/arm.md (use_literal_pool): Remove.
26433 (64-bit immediate split): No longer takes cost into consideration
26434 if arm_disable_literal_pool is enabled.
26435 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
26436 used when arm_disable_literal_pool is enabled.
26437 (arm_max_const_double_inline_cost): Remove use of
26438 arm_disable_literal_pool.
26439 (push_minipool_fix): Add assert.
26440 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
26441 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
26442 (no_literal_pool_sf_immediate): New.
26444 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26446 PR tree-optimization/80613
26447 * tree-ssa-dce.c (propagate_necessity): Remove cases for
26448 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
26450 2017-05-05 Richard Biener <rguenther@suse.de>
26452 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
26454 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
26456 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
26457 of this flag from insn conditions due to removal from r247495.
26459 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26461 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
26463 (arm_early_store_addr_dep_ptr): Likewise.
26464 * config/arm/aarch-common-protos.h
26465 (arm_early_load_addr_dep_ptr): Add prototype.
26466 (arm_early_store_addr_dep_ptr): Likewise.
26467 * config/arm/cortex-a53.md: Add new bypasses.
26469 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26471 * tree.c (next_type_uid): Change type to unsigned.
26472 (type_hash_canon): Decrement back next_type_uid if
26473 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
26474 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
26477 2017-05-04 Martin Sebor <msebor@redhat.com>
26479 * builtins.c: Fix a trivial typo in a comment.
26481 PR middle-end/79234
26482 * builtins.c (check_sizes): Adjust to handle reading past the end.
26483 Avoid printing excessive upper bound of ranges. Use %E to print
26484 tree nodes instead of converting them to %wu.
26485 (expand_builtin_memchr): New function.
26486 (compute_dest_size): Rename...
26487 (compute_objsize): ...to this.
26488 (expand_builtin_memcpy): Adjust.
26489 (expand_builtin_mempcpy): Adjust.
26490 (expand_builtin_strcat): Adjust.
26491 (expand_builtin_strcpy): Adjust.
26492 (check_strncat_sizes): Adjust.
26493 (expand_builtin_strncat): Adjust.
26494 (expand_builtin_strncpy): Adjust and simplify.
26495 (expand_builtin_memset): Adjust.
26496 (expand_builtin_bzero): Adjust.
26497 (expand_builtin_memcmp): Adjust.
26498 (expand_builtin): Handle memcmp.
26499 (maybe_emit_chk_warning): Check strncat just once.
26501 2017-05-04 Martin Sebor <msebor@redhat.com>
26503 PR preprocessor/79214
26504 PR middle-end/79222
26505 PR middle-end/79223
26506 * builtins.c (check_sizes): Add inlining context and issue
26507 warnings even when -Wno-system-headers is set.
26508 (check_strncat_sizes): Same.
26509 (expand_builtin_strncat): Same.
26510 (expand_builtin_memmove): New function.
26511 (expand_builtin_stpncpy): Same.
26512 (expand_builtin): Handle memmove and stpncpy.
26514 2017-05-04 Bin Cheng <bin.cheng@arm.com>
26516 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
26517 which is not used any more.
26519 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26521 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
26523 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26525 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
26526 (cortexa53_tunings): Likewise.
26527 (cortexa57_tunings): Likewise.
26528 (cortexa72_tunings): Likewise.
26529 (cortexa73_tunings): Likewise.
26531 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26533 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
26534 Set loop alignment to 8.
26536 2017-05-04 Martin Sebor <msebor@redhat.com>
26538 PR translation/80280
26539 * builtins.c (expand_builtin_object_size): Add missing quoting to
26540 %D and like directives.
26541 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
26542 (hsa_type_for_tree_type): Same.
26543 (verify_function_arguments): Same.
26544 * symtab.c (symbol_table::change_decl_assembler_name): Same.
26545 * varasm.c (get_section): Same.
26548 2017-05-04 Martin Sebor <msebor@redhat.com>
26550 PR translation/80280
26551 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
26553 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26555 * config/aarch64/aarch64.c (generic_addrcost_table):
26556 Change HI/TI mode setting.
26558 2017-05-04 Martin Jambor <mjambor@suse.cz>
26560 PR tree-optimization/80622
26561 * tree-sra.c (comes_initialized_p): New function.
26562 (build_accesses_from_assign): Only set write lazily when
26563 comes_initialized_p is false.
26564 (analyze_access_subtree): Use comes_initialized_p.
26565 (propagate_subaccesses_across_link): Assert !comes_initialized_p
26566 instead of testing for PARM_DECL.
26568 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26570 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
26571 constraint on operand 0 to allow more general addressing modes.
26572 Adjust output template.
26573 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
26575 * config/aarch64/aarch64-protos.h
26576 (aarch64_address_valid_for_prefetch_p): Declare prototype.
26577 * config/aarch64/constraints.md (Dp): New address constraint.
26578 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
26581 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
26583 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
26584 update use of estimate_ipcp_clone_size_and_time.
26585 (estimate_local_effects): Update use of
26586 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
26587 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
26588 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
26589 Return nonspecialized time.
26591 2017-05-04 Richard Biener <rguenther@suse.de>
26593 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
26594 for the last VUSE which def dominates the PHI. Directly call
26596 (get_continuation_for_phi_1): Remove.
26598 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
26600 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
26601 to explain the use of truncating division. Cap the number of
26602 iterations to the maximum given by nb_iterations_upper_bound,
26605 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26607 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
26608 * configure: Regenerate.
26609 * config.in: Regenerate.
26610 * config/i386/driver-mingw32.c: new file.
26611 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
26612 * config.host: Link driver-mingw32.o on MinGW host.
26613 * doc/install.texi: Document new --enable-mingw-wildcard configure
26616 2017-05-04 Marek Polacek <polacek@redhat.com>
26618 PR tree-optimization/80612
26619 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
26621 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26622 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
26624 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
26625 (movt splitter): Likewise.
26626 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
26627 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
26628 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
26629 block for Thumb-1 with MOVT.
26630 (thumb2_legitimate_address_p): Move code block ...
26631 (can_avoid_literal_pool_for_label_p): ... into this new function.
26632 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
26634 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
26635 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
26636 "M-profile targets with the MOVT instruction".
26638 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26640 * config/arm/arm-builtins.c (arm_init_builtins): Rename
26641 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
26642 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
26644 2017-05-04 Martin Liska <mliska@suse.cz>
26646 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
26647 variable cond_code.
26649 2017-05-04 Richard Biener <rguenther@suse.de>
26651 * tree.c (array_at_struct_end_p): Handle arrays at struct
26652 end with flexarrays more conservatively. Refactor and treat
26653 arrays of arrays or aggregates more strict. Fix
26654 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26655 * tree.c (array_at_struct_end_p): Adjust prototype.
26656 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26657 * gimple-fold.c (get_range_strlen): Likewise.
26658 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26660 2017-05-04 Richard Biener <rguenther@suse.de>
26662 PR tree-optimization/31130
26663 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26665 (supports_overflow_infinity): Likewise.
26666 (is_negative_overflow_infinity): Likewise.
26667 (is_positive_overflow_infinity): Likewise.
26668 (is_overflow_infinity): Likewise.
26669 (stmt_overflow_infinity): Likewise.
26670 (overflow_infinity_range_p): Likewise.
26671 (usable_range_p): Remove as always returning true.
26672 (make_overflow_infinity): Remove.
26673 (negative_overflow_infinity): Likewise.
26674 (positive_overflow_infinity): Likewise.
26675 (avoid_overflow_infinity): Likewise.
26676 (set_value_range): Adjust accordingly.
26677 (set_value_range_to_nonnegative): Likewise, remove now unused
26678 overflow_infinity arg.
26679 (vrp_operand_equal_p): Adjust.
26680 (update_value_range): Likewise.
26681 (range_int_cst_singleton_p): Likewise.
26682 (operand_less_p): Likewise.
26683 (compare_values_warnv): Likewise.
26684 (extract_range_for_var_from_comparison_expr): Likewise.
26685 (vrp_int_const_binop): Likewise.
26686 (zero_nonzero_bits_from_vr): Likewise.
26687 (extract_range_from_multiplicative_op_1): Likewise.
26688 (extract_range_from_binary_expr_1): Likewise.
26689 (extract_range_from_unary_expr): Likewise.
26690 (extract_range_from_comparison): Likewise.
26691 (extract_range_basic): Likewise.
26692 (adjust_range_with_scev): Likewise.
26693 (compare_ranges): Likewise.
26694 (compare_range_with_value): Likewise.
26695 (dump_value_range): Likewise.
26696 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26698 (simplify_cond_using_ranges): Adjust.
26700 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26702 * brig-builtins.def: Added a builtin for class_f64.
26703 * builtin-types.def: Added a builtin type needed by class_f64.
26705 2017-05-03 Jason Merrill <jason@redhat.com>
26707 * timevar.def: Add TV_CONSTEXPR.
26709 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26711 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26713 2017-05-03 Martin Jambor <mjambor@suse.cz>
26715 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26716 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26717 * ipa-cp.c (ipcp_propagate_stage): Do not call
26718 ipa_update_after_lto_read.
26719 * ipa-inline.c (ipa_inline): Likewise.
26721 2017-05-03 Martin Jambor <mjambor@suse.cz>
26723 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26724 tag. Added a default constructor and a destructor.
26725 (ipa_edge_args_sum_t): New class;
26726 (ipa_edge_args_sum): Declare.
26727 (ipa_edge_args_vector): Remove declaration.
26728 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26729 (ipa_free_edge_args_substructures): Remove declaration.
26730 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26731 (ipa_edge_args_info_available_for_edge_p): Likewise.
26732 * ipa-prop.c (ipa_edge_args_vector): Removed.
26733 (edge_removal_hook_holder): Likewise.
26734 (edge_duplication_hook_holder): Likewise.
26735 (ipa_edge_args_sum): New variable.
26736 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26737 ipa_edge_args_vector.
26738 (ipa_free_edge_args_substructures): Likewise.
26739 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26740 ipa_edge_args_vector.
26741 (ipa_edge_removal_hook): Turned into method
26742 ipa_edge_args_sum_t::remove.
26743 (ipa_edge_duplication_hook): Turned into method
26744 ipa_edge_args_sum_t::duplicate.
26745 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26746 registering edge hooks.
26747 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26748 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26749 ipa_edge_args_sum instead of ipa_edge_args_vector.
26750 * ipa-profile.c (ipa_profile): Likewise.
26752 2017-05-03 Martin Jambor <mjambor@suse.cz>
26754 * symbol-summary.h (function_summary): New method exists.
26755 (function_summary::symtab_removal): Deallocate through release.
26756 (call_summary): New class.
26757 (gt_ggc_mx): New overload.
26758 (gt_pch_nx): Likewise.
26759 (gt_pch_nx): Likewise.
26761 2017-05-03 Jeff Law <law@redhat.com>
26763 PR tree-optimization/78496
26764 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26765 from simplify_cond_using_ranges. Split off code to walk
26766 backwards through casts into ...
26767 (simplify_cond_using_ranges_2): New function.
26768 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26769 (execute_vrp): After identifying jump threads, call
26770 simplify_cond_using_ranges_2.
26772 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26775 * ipa-inline.h (inline_summary): Add ctor.
26776 (create_ggc): Do not use ggc_cleared_alloc.
26778 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26779 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26781 * gcc.c (handle_braces): Support escaping in switch matching
26783 * doc/invoke.texi (Spec Files): Document it.
26784 Remove superfluous @code markup in items.
26786 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26788 * diagnostic-show-locus.c (struct column_range): New struct.
26789 (get_affected_columns): New function.
26790 (get_printed_columns): New function.
26791 (struct correction): New struct.
26792 (correction::ensure_capacity): New function.
26793 (correction::ensure_terminated): New function.
26794 (struct line_corrections): New struct.
26795 (line_corrections::~line_corrections): New dtor.
26796 (line_corrections::add_hint): New function.
26797 (layout::print_trailing_fixits): Reimplement in terms of the new
26799 (selftest::test_overlapped_fixit_printing): New function.
26800 (selftest::diagnostic_show_locus_c_tests): Call it.
26802 2017-05-03 Nathan Sidwell <nathan@acm.org>
26804 Canonicalize canonical type hashing
26805 * tree.h (type_hash_canon_hash): Declare.
26806 * tree.c (type_hash_list, attribute_hash_list): Move into
26807 type_hash_canon_hash.
26808 (build_type_attribute_qual_variant): Break out hash code calc into
26809 type_hash_canon_hash.
26810 (type_hash_canon_hash): New. Generic type hash computation.
26811 (build_range_type_1, build_array_type_1, build_function_type,
26812 build_method_type_directly, build_offset_type, build_complex_type,
26813 make_vector_type): Call it.
26815 2017-05-03 Richard Biener <rguenther@suse.de>
26817 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26818 When all DRs have unknown misaligned do not always peel
26819 when there is a store but apply the same costing model as if
26820 there were only loads.
26822 2017-05-03 Richard Biener <rguenther@suse.de>
26825 PR tree-optimization/80492
26826 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26827 compare_base_decls returning dont-know properly.
26829 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26831 * config/arm/iterators.md (CCSI): New mode iterator.
26832 (arch): New mode attribute.
26833 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26834 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26835 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26836 code iterator for success result mode.
26837 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26838 the corresponding new insn generators.
26840 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26843 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26844 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26846 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26848 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26849 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26850 (DDR_A): Wrap DDR argument in brackets.
26851 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26852 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26853 (DDR_REVERSED_P): Likewise.
26855 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26857 PR tree-optimization/79472
26858 * tree-switch-conversion.c (struct switch_conv_info): Add
26859 contiguous_range and default_case_nonstandard fields.
26860 (collect_switch_conv_info): Compute contiguous_range and
26861 default_case_nonstandard fields, don't clear final_bb if
26862 contiguous_range and only the default case doesn't have the required
26864 (check_all_empty_except_final): Set default_case_nonstandard instead
26865 of failing if contiguous_range and the default case doesn't have empty
26867 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26868 and only the default case doesn't have the required constants. Skip
26870 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26871 if default_case_nonstandard.
26872 (build_constructors): Build constant 1 just once. Assert that default
26873 values aren't inserted in between cases if contiguous_range. Skip
26875 (build_arrays): Skip virtual phis.
26876 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26877 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26878 Handle virtual phis.
26879 (gen_inbound_check): Handle default_case_nonstandard case.
26880 (process_switch): Adjust check_final_bb caller. Call
26881 gather_default_values with the first non-default case instead of
26882 default case if default_case_nonstandard.
26884 2017-05-02 Nathan Sidwell <nathan@acm.org>
26886 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26887 check. Fix formatting.
26889 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26891 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26892 errors when comparing specialized and unspecialized times.
26894 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26896 * diagnostic-show-locus.c
26897 (layout::should_print_annotation_line_p): Make private.
26898 (layout::print_annotation_line): Make private.
26899 (layout::annotation_line_showed_range_p): Make private.
26900 (layout::show_ruler): Make private.
26901 (layout::print_source_line): Make private. Pass in line and
26902 line_width, rather than calling location_get_source_line. Drop
26904 (layout::print_leading_fixits): New method.
26905 (layout::print_any_fixits): Rename to...
26906 (layout::print_trailing_fixits): ...this, and make private.
26907 Don't print newline fixits.
26908 (diagnostic_show_locus): Move logic for printing one row into...
26909 (layout::print_line): ...this new function. Move the
26910 location_get_source_line call and error-handling from
26911 print_source_line to here. Call print_leading_fixits, and rename
26912 print_any_fixits to print_trailing_fixits.
26913 (selftest::test_fixit_insert_containing_newline): Update now that
26914 newlines are partially supported.
26915 (selftest::test_fixit_insert_containing_newline_2): New test.
26916 (selftest::test_fixit_replace_containing_newline): Update comments.
26917 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26918 * edit-context.c (class added_line): New class.
26919 (class edited_line): Describe newline handling in comment.
26920 (edited_line::actually_edited_p): New method.
26921 (edited_line::print_content): Delete redundant decl.
26922 (edited_line::m_predecessors): New field.
26923 (edited_file::print_content): Call edited_line::print_content.
26924 (edited_file::print_diff): Update to support newlines.
26925 (edited_file::print_diff_hunk): Likewise.
26926 (edited_file::print_run_of_changed_lines): New function.
26927 (edited_file::print_diff_line): Convert to...
26928 (print_diff_line): ...this.
26929 (edited_file::get_effective_line_count): New function.
26930 (edited_line::edited_line): Initialize new field m_predecessors.
26931 (edited_line::~edited_line): Clean up m_predecessors.
26932 (edited_line::apply_fixit): Handle newlines.
26933 (edited_line::get_effective_line_count): New function.
26934 (edited_line::print_content): New function.
26935 (edited_line::print_diff_lines): New function.
26936 (selftest::test_applying_fixits_insert_containing_newline): New
26938 (selftest::test_applying_fixits_replace_containing_newline): New
26940 (selftest::insert_line): New function.
26941 (selftest::test_applying_fixits_multiple_lines): Add example of
26943 (selftest::edit_context_c_tests): Call the new tests.
26945 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26947 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26948 parameter cand. Update dump information.
26949 (get_computation_cost): Update uses.
26951 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26953 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26954 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26955 (get_computation_at, rewrite_use_address): Update use of
26956 get_computation_aff.
26958 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26960 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26961 (get_computation): Delete.
26962 (get_computation_cost): Implement like get_computation_cost_at.
26963 Use get_computation_at.
26964 (get_computation_cost_at): Delete.
26965 (rewrite_use_nonlinear_expr): Use get_computation_at.
26966 (rewrite_use_compare, remove_unused_ivs): Ditto.
26968 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26970 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26972 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26974 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26975 (ivopts_global_cost_for_size): Rename parameter and update uses.
26976 (iv_ca_recount_cost): Update uses.
26977 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26978 candidates seperately in n_invs and n_cands.
26979 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26981 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26983 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26984 (find_inv_vars_cb): New.
26985 (find_depends): Renamed to ...
26986 (find_inv_vars): ... this.
26987 (add_candidate_1, force_var_cost): Call find_inv_vars.
26988 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26990 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26992 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26993 inv_vars. Add inv_exprs.
26994 (struct iv_cand): Rename depends_on to inv_vars.
26995 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26996 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26997 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26998 (dump_cand): Dump inv_vars.
26999 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
27000 (record_invariant, find_depends, add_candidate_1): Ditto.
27001 (set_group_iv_cost, force_var_cost): Ditto.
27002 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
27003 (get_computation_cost_at, get_computation_cost): Ditto.
27004 (determine_group_iv_cost_generic): Ditto.
27005 (determine_group_iv_cost_address): Ditto.
27006 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
27007 (determine_group_iv_costs): Ditto.
27008 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
27009 (iv_ca_set_remove_invariants): Renamed to ...
27010 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
27011 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
27012 (iv_ca_set_add_invariants): Renamed to ...
27013 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
27014 (iv_ca_set_cp): Use iv_ca_set_add_invs.
27015 (iv_ca_has_deps): Support inv_vars and inv_exprs.
27016 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
27017 (create_new_ivs): Remove useless dump.
27019 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27021 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
27023 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
27024 (iv_ca_set_no_cp, create_new_iv): Ditto.
27026 2017-05-02 Bin Cheng <bin.cheng@arm.com>
27028 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27030 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
27032 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
27033 function tree_check2.
27035 2017-05-02 Martin Liska <mliska@suse.cz>
27037 * doc/gcov.texi: Add missing preposition.
27038 * gcov.c (function_info::function_info): Properly fill up
27039 all member variables.
27041 2017-05-02 Tamar Christina <tamar.christina@arm.com>
27043 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
27045 2017-05-02 Tamar Christina <tamar.christina@arm.com>
27047 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
27049 2017-05-02 Martin Liska <mliska@suse.cz>
27052 * lto-streamer-in.c (lto_read_tree_1): Remove
27053 LTO_STREAMER_DEBUG.
27054 * lto-streamer.c (struct tree_hash_entry): Likewise.
27055 (struct tree_entry_hasher): Likewise.
27056 (tree_entry_hasher::hash): Likewise.
27057 (tree_entry_hasher::equal): Likewise.
27058 (lto_streamer_init): Likewise.
27059 (lto_orig_address_map): Likewise.
27060 (lto_orig_address_get): Likewise.
27061 (lto_orig_address_remove): Likewise.
27062 * lto-streamer.h: Likewise.
27063 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
27064 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
27066 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
27068 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
27069 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
27070 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
27071 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
27072 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
27073 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
27074 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
27075 (mm_maskz_sub_ss): New intrinsics.
27076 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27077 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27078 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
27079 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
27080 (__builtin_ia32_subss_mask_round): New builtins.
27081 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27082 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27083 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
27085 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
27086 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27088 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27091 2017-05-02 Martin Jambor <mjambor@suse.cz>
27093 PR tree-optimization/78687
27094 * tree-sra.c (access): New field parent.
27095 (process_subtree_disqualification): New function.
27096 (disqualify_candidate): Call it.
27097 (build_accesses_from_assign): Reset write flag if creating an
27099 (build_access_subtree): Fill in parent field and also prpagate
27100 down grp_write flag.
27101 (create_artificial_child_access): New parameter set_grp_write, set
27102 grp_write to its value.
27103 (propagate_subaccesses_across_link): Also propagate grp_write flag
27105 (propagate_all_subaccesses): Push the closest parent back to work
27106 queue if add_access_to_work_queue returned true.
27108 2017-05-02 Richard Biener <rguenther@suse.de>
27110 * common.opt (fstrict-overflow): Alias negative to fwrapv.
27111 * doc/invoke.texi (fstrict-overflow): Remove all traces of
27112 -fstrict-overflow documentation.
27113 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
27114 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
27115 flag_strict_overflow.
27116 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
27117 * lto-opts.c (lto_write_options): Do not stream it.
27118 * lto-wrapper.c (merge_and_complain): Do not handle it.
27119 * opts.c (default_options_table): Do not set -fstrict-overflow.
27120 (finish_options): Likewise do not clear it when sanitizing.
27121 * simplify-rtx.c (simplify_const_relational_operation): Do not
27122 test flag_strict_overflow.
27124 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
27126 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
27127 using enabled attribute.
27128 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
27129 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
27130 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
27131 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
27132 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
27133 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
27134 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
27135 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
27136 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
27137 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
27139 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
27141 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
27143 2017-05-02 Richard Biener <rguenther@suse.de>
27145 PR tree-optimization/80591
27147 2017-04-10 Richard Biener <rguenther@suse.de>
27149 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27152 2017-05-02 Richard Biener <rguenther@suse.de>
27154 PR tree-optimization/80549
27155 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
27156 (cleanup_tree_cfg_noloop): Create forwarders to known loop
27157 headers if they do not have a preheader.
27159 2017-05-02 Martin Liska <mliska@suse.cz>
27162 * common.opt: Fix typo.
27163 * doc/invoke.texi: Likewise.
27165 2017-05-01 Jan Beulich <jbeulich@suse.com>
27167 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
27168 swapping, add (x,x,m,x,n) alternative.
27170 2017-05-01 Nathan Sidwell <nathan@acm.org>
27172 * calls.c (combine_pending_stack_adjustment_and_call): Remove
27173 unnecessary unadjusted_alignment check.
27175 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
27178 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
27180 * gimplify.c (gimplify_cilk_detach): New function.
27181 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
27182 * tree-core.h: Document EXPR_CILK_SPAWN.
27183 * tree.h (EXPR_CILK_SPAWN): Define.
27185 2017-05-01 David Malcolm <dmalcolm@redhat.com>
27187 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
27188 to use new fixit_hint representation, using the "replace" logic.
27189 (get_line_span_for_fixit_hint): Likewise.
27190 (layout::print_any_fixits): Likewise.
27191 (selftest::test_one_liner_many_fixits): Rename to...
27192 (selftest::test_one_liner_many_fixits_1): ...this, and update
27193 comment and expected output to reflect that the multiple fix-it
27194 hints are now consolidated into one insertion.
27195 (selftest::test_one_liner_many_fixits_2): New test.
27196 (selftest::test_diagnostic_show_locus_one_liner): Update for
27198 (selftest::test_fixit_consolidation): Update for fix-it API
27200 * diagnostic.c (print_parseable_fixits): Likewise.
27201 * edit-context.c (edited_line::m_line_events): Convert from
27202 auto_vec <line_event *> to auto_vec <line_event>.
27203 (class line_event): Convert from abstract base class to a concrete
27204 class, taking over the role of replace_event.
27205 (class insert_event): Delete.
27206 (class replace_event): Rename to class line_event. Convert to
27208 (edit_context::add_fixits): Reimplement.
27209 (edit_context::apply_insert): Delete.
27210 (edit_context::apply_replace): Rename to...
27211 (edit_context::apply_fixit): ...this. Convert to half-open range.
27212 (edited_file::apply_insert): Delete.
27213 (edited_file::apply_replace): Rename to...
27214 (edited_file::apply_fixit): ...this.
27215 (edited_line::~edited_line): Drop deletion of events.
27216 (edited_line::apply_insert): Delete.
27217 (edited_line::apply_replace): Rename to...
27218 (edited_line::apply_fixit): ...this. Convert to half-open range.
27219 Update for change to type of m_line_events.
27220 * edit-context.h (edit_context::apply_insert): Delete.
27221 (edit_context::apply_replace): Rename to...
27222 (edit_context::apply_fixit): ...this.
27224 2017-05-01 Martin Sebor <msebor@redhat.com>
27226 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
27229 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
27232 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
27233 __get_cpuid_max returns 0.
27234 (__get_cpuid_count): Ditto.
27236 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
27238 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
27239 replacement expression is another instance of one of its arguments.
27241 2017-05-01 Jakub Jelinek <jakub@redhat.com>
27244 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
27245 check for stack push/pop autoinc.
27246 * config/i386/i386.c (ix86_agi_dependent): Return false
27247 if the only reason why modified_in_p returned true is that
27248 addr is SP based and set_insn is a push or pop.
27250 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
27252 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
27255 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
27258 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
27259 (account_size_time): Use two predicates - exec_pred and
27261 (evaluate_conditions_for_known_args): Compute both clause and
27263 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
27264 (inline_summary_t::duplicate): Update.
27265 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
27267 (compute_inline_parameters): Likewise.
27268 (estimate_edge_size_and_time): Update caluclation of time.
27269 (estimate_node_size_and_time): Compute both time and nonspecialized
27271 (estimate_ipcp_clone_size_and_time): Update.
27272 (inline_merge_summary): Update.
27273 (do_estimate_edge_time): Update.
27274 (do_estimate_edge_size): Update.
27275 (do_estimate_edge_hints): Update.
27276 (inline_read_section, inline_write_summary): Stream both new predicates.
27277 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
27279 (compute_inlined_call_time): Cleanup.
27280 (big_speedup_p): Update.
27281 (edge_badness): Update.
27282 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
27283 (size_time_entry): Replace predicate by exec_predicate and
27284 nonconst_predicate.
27285 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
27286 (estimate_edge_time): Return also nonspec_time.
27287 (reset_edge_growth_cache): Update.
27289 2017-04-29 Jakub Jelinek <jakub@redhat.com>
27291 PR rtl-optimization/80491
27292 * ifcvt.c (noce_process_if_block): When looking for x setter
27293 with missing else_bb, don't check only the insn right before
27294 cond_earliest, but look for the last insn that x is modified in
27295 within the same bb.
27297 PR rtl-optimization/80491
27298 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
27300 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
27302 PR tree-optimization/80487
27303 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
27305 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27307 PR tree-optimization/79697
27308 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
27309 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
27310 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
27312 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
27313 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
27315 2017-04-28 Martin Sebor <msebor@redhat.com>
27317 PR tree-optimization/80523
27318 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
27319 (init_target_to_host_charmap, target_to_host, target_strtol10): New
27321 (maybe_warn, format_directive, parse_directive): Use new functions.
27322 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
27324 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
27326 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
27328 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
27330 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
27331 target_header_dir): Set correctly.
27332 * configure: Regenerated.
27333 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27334 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27335 instead of SYSTEM_HEADER_DIR.
27337 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
27339 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
27340 (estimate_local_effects): Likewise.
27341 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
27342 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
27343 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
27344 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
27345 do_estimate_edge_time, estimate_edge_time): Likewise.
27346 * ipa-inline-analysis.c (estimate_node_size_and_time,
27347 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
27348 (estimate_time_after_inlining): Remove.
27350 2017-04-28 Martin Liska <mliska@suse.cz>
27352 * doc/gcov.texi: Enhance documentation of gcov.
27354 2017-04-28 Martin Liska <mliska@suse.cz>
27356 * doc/gcov.texi: Sort options in alphabetic order.
27357 * doc/gcov-dump.texi: Likewise.
27358 * doc/gcov-tool.texi: Likewise.
27359 * gcov.c (print_usage): Likewise.
27360 * gcov-dump.c (print_usage): Likewise.
27361 * gcov-tool.c (print_merge_usage_message): Likewise.
27362 (print_rewrite_usage_message): Likewise.
27363 (print_overlap_usage_message): Likewise.
27365 2017-04-28 Martin Liska <mliska@suse.cz>
27367 PR gcov-profile/53915
27368 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
27370 2017-04-28 Martin Liska <mliska@suse.cz>
27372 PR gcov-profile/79891
27373 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
27374 is marked by compiler as living on a line.
27375 (get_cycles_count): Remove usage of the union.
27376 (output_intermediate_file): Likewise.
27377 (find_source): Fix GNU coding style.
27378 (accumulate_line_counts): Remove old non-all block mode.
27379 (output_lines): Remove usage of the union.
27380 * profile.c (output_location): Include all BBs, even if
27381 belonging to a same line (and file) as a previous BB.
27383 2017-04-28 Martin Liska <mliska@suse.cz>
27385 * gcov.c (process_args): Handle new argument 'w'.
27386 (read_graph_file): Assign ID to BBs.
27387 (output_branch_count): Display BB # if verbose flag is set.
27388 (output_lines): Likewise for arcs.
27389 (print_usage): Add '--verbose' option help.
27390 * doc/gcov.texi: Document --verbose (-w) option.
27392 2017-04-28 Martin Liska <mliska@suse.cz>
27394 * gcov.c (struct block_location_info): New struct.
27395 (process_file): Fill up the new structure.
27396 (read_graph_file): Replace usage of encoding by the newly added
27398 (add_line_counts): Likewise.
27399 (accumulate_line_counts): Remove usage of the union.
27400 (function_info::function_info): New function.
27401 (function_info::~function_info): Likewise.
27402 (process_file): Call delete instead of release_function.
27403 (release_function): Release the function.
27404 (release_structures): Call delete instead of release_function.
27405 (solve_flow_graph): Replace usage of num_blocks.
27406 (find_exception_blocks): Likewise.
27407 (output_lines): Fix GNU coding style.
27409 2017-04-28 Martin Liska <mliska@suse.cz>
27412 * coverage.c (coverage_remove_note_file): New function.
27413 * coverage.h: Declare the function.
27414 * toplev.c (finalize): Clean if an error has been seen.
27416 2017-04-28 Martin Liska <mliska@suse.cz>
27418 PR gcov-profile/80031
27419 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
27420 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
27421 * gcov.c (read_graph_file): Read just number of blocks.
27422 * profile.c (branch_prob): Do not stream 0 flags per a basic
27425 2017-04-28 Martin Liska <mliska@suse.cz>
27427 * gcov-dump.c (tag_*): Add new argument to declarations.
27428 (dump_gcov_file): Likewise.
27429 (tag_blocks): Add and use new argument depth.
27430 (tag_arcs): Likewise.
27431 (tag_lines): Likewise.
27432 (tag_counters): Likewise.
27433 (tag_summary): Likewise.
27434 (dump_working_sets): Use depth to do a proper indentation.
27436 2017-04-28 Jakub Jelinek <jakub@redhat.com>
27439 * cgraph.h (symtab_node::debug_symtab): No longer inline.
27440 * symtab.c (symtab_node::debug_symtab): Move definition here.
27442 2017-04-28 Richard Biener <rguenther@suse.de>
27444 * lto-streamer.h (LTO_major_version): Bump to 7.
27446 2017-04-28 Richard Biener <rguenther@suse.de>
27448 * tree-vrp.c (assert_info): New struct.
27449 (add_assert_info): New helper.
27450 (register_edge_assert_for_2): Refactor to add asserts to a vector
27452 (register_edge_assert_for_1): Likewise.
27453 (register_edge_assert_for): Likewise.
27454 (finish_register_edge_assert_for): New helper actually registering
27455 asserts where live on edge.
27456 (find_conditional_asserts): Adjust.
27457 (find_switch_asserts): Likewise.
27458 (evrp_dom_walker::try_find_new_range): Generalize.
27459 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
27461 2017-04-27 Marek Polacek <polacek@redhat.com>
27464 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
27465 arg10 and arg11 to itype.
27467 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
27469 * doc/extend.texi (Object Size Checking): Improve grammar.
27471 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
27474 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
27475 that the logic for permitting reciprocal estimates matches that
27478 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27481 * tree.c (type_cache_hasher::equal): Only compare
27482 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
27483 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
27484 non-aggregate element types.
27485 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
27486 about the flag on ARRAY_TYPEs in the comment, formatting fix.
27488 2017-04-27 Richard Biener <rguenther@suse.de>
27490 PR middle-end/80533
27491 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
27492 stripping ARRAY_REFs from MEM_EXPR make sure we're not
27493 keeping a reference to a trailing array.
27495 2017-04-27 Richard Biener <rguenther@suse.de>
27497 PR middle-end/80539
27498 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
27499 being in loop-closed SSA form conservatively.
27500 (chrec_fold_multiply_poly_poly): Likewise.
27502 2017-04-27 Tamar Christina <tamar.christina@arm.com>
27504 PR middle-end/79665
27505 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
27506 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
27508 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27511 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
27512 (aarch64_function_arg_alignment): Return unsigned int again, but still
27513 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
27514 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27515 Don't emit -Wpsabi note.
27516 (aarch64_function_arg_boundary): Likewise.
27517 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27520 2017-04-26 Nathan Sidwell <nathan@acm.org>
27522 * tree.h (crc32_unsigned_n): Declare.
27523 (crc32_unsigned, crc32_unsigned): Make inline.
27524 * tree.c (crc32_unsigned_bits): Replace with ...
27525 (crc32_unsigned_n): ... this.
27526 (crc32_unsigned, crc32_byte): Remove.
27527 (crc32_string): Remove unnecessary braces.
27529 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27531 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
27532 * ipa-inline-analysis.c (MAX_TIME): Remove.
27533 (account_size_time): Use sreal for time.
27534 (dump_inline_summary): Update.
27535 (estimate_function_body_sizes): Update.
27536 (estimate_edge_size_and_time): Update.
27537 (estimate_calls_size_and_time): Update.
27538 (estimate_node_size_and_time): Update.
27539 (inline_merge_summary): Update.
27540 (inline_update_overall_summary): Update.
27541 (estimate_time_after_inlining): Update.
27542 (inline_read_section): Update.
27543 (inline_write_summary): Update.
27544 * ipa-inline.c (compute_uninlined_call_time): Update.
27545 (compute_inlined_call_time): Update.
27546 (recursive_inlining): Update.
27547 (inline_small_functions): Update.
27548 (dump_overall_stats): Update.
27549 * ipa-inline.h: Include sreal.h.
27550 (size_time_entry): Turn time to sreal.
27551 (inline_summary): Turn self_time nad time to sreal.
27553 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27555 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
27557 (sreal::stream_out, sreal::stream_in): New.
27558 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
27560 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27562 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
27565 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
27568 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
27570 (dimode_scalar_chain::compute_convert_gain): Ditto.
27571 (dimode_scalar_chain::make_vector_copies): Ditto.
27572 (dimode_scalar_chain::convert_reg): Ditto.
27573 (dimode_scalar_chain::convert_insn): Ditto.
27574 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
27575 (VI248_AVX512BW_1): New mode iterator.
27576 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
27577 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
27580 2017-04-25 Martin Sebor <msebor@redhat.com>
27582 PR tree-optimization/80497
27583 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
27584 constants are representable in HOST_WIDE_INT.
27585 (parse_directive): Ditto.
27587 2017-04-25 Martin Sebor <msebor@redhat.com>
27590 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
27591 (new_zero_array): Adjust signature.
27592 (dom_info::dom_init): Used unsigned rather that size_t.
27593 (dom_info::dom_info): Same.
27595 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27596 Jakub Jelinek <jakub@redhat.com>
27599 * config/arm/arm.c: Include gimple.h.
27600 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27601 returns negative, increment ncrn only if it returned positive.
27602 (arm_needs_doubleword_align): Return int instead of bool,
27603 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
27604 members, but if there is any such non-FIELD_DECL
27605 > PARM_BOUNDARY aligned decl, return -1 instead of false.
27606 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27607 returns negative, increment nregs only if it returned positive.
27608 (arm_setup_incoming_varargs): Likewise.
27609 (arm_function_arg_boundary): Emit -Wpsabi note if
27610 arm_needs_doubleword_align returns negative, return
27611 DOUBLEWORD_ALIGNMENT only if it returned positive.
27613 2017-04-25 Marek Polacek <polacek@redhat.com>
27616 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27617 first argument to type.
27619 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
27622 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
27623 type checks to test for compatibility instead of equality.
27625 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27626 Jakub Jelinek <jakub@redhat.com>
27629 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
27631 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
27632 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
27633 the alignment computation, but return their maximum in warn_alignment.
27634 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27635 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
27637 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
27638 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27641 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27643 * config/arc/simdext.md (dmpyh): Fix typo.
27645 2017-04-25 Richard Biener <rguenther@suse.de>
27647 PR tree-optimization/80492
27648 * alias.c (compare_base_decls): Handle registers with asm
27649 specification conservatively.
27650 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27651 compare_base_decls returning dont-know properly.
27653 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27655 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27656 (legitimate_offset_address_p): New function.
27657 (arc_legitimate_address_p): Use above function.
27659 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27661 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27663 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27665 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27666 ACCH registers whenever they are available.
27668 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27670 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27671 double regs fix when not used.
27673 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27675 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27677 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27678 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27680 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27682 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27683 when emitting short ADD instructions.
27685 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27687 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27689 (cmpsi_cc_c_insn): Likewise.
27690 (cbranchsi4_scratch): Compute proper instruction length using
27691 compact_hreg_operand.
27692 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27694 2017-04-25 Richard Biener <rguenther@suse.de>
27696 PR middle-end/80509
27697 * passes.c (pass_manager::pass_manager): Initialize
27698 m_name_to_pass_map.
27700 2017-04-25 Richard Biener <rguenther@suse.de>
27702 PR tree-optimization/79201
27703 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27705 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27708 * config/s390/vector.md: Split MEM->GPR vector moves for
27709 non-s_operand addresses.
27711 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27714 * config/s390/predicates.md (reload_const_wide_int_operand): New
27716 * config/s390/s390.md ("movti"): Remove d/P alternative.
27717 ("movti_bigconst"): New pattern definition.
27719 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27722 * s390-protos.h (s390_expand_cs_hqi): Removed.
27723 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27724 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27725 modes as well as CCZ1mode and CCZmode.
27726 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27727 signature of s390_emit_compare_and_swap.
27728 (s390_expand_cs_hqi): Likewise, make static.
27729 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27730 compare-and-swap, in some cases.
27731 (s390_expand_cs): Wrapper function.
27732 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27734 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27735 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27736 patterns for small and large integers. Forbid symref memory operands.
27737 Move expander to s390.c. Require cc register.
27738 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27739 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27740 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27741 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27742 symref memory operands. Remove CC mode and call s390_match_ccmode
27744 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27746 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27748 * config/s390/s390.md (define_peephole2): New peephole to help
27749 combining the load-and-test pattern with volatile memory.
27751 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27753 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27754 with CCZmode for TARGET_Z196.
27756 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27758 PR rtl-optimization/80501
27759 * combine.c (make_compound_operation_int): Set subreg_code to SET
27760 even for AND with mask of the sign bit of mode.
27762 PR rtl-optimization/80500
27763 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27764 sum's initial value.
27766 2017-04-25 Julian Brown <julian@codesourcery.com>
27767 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27769 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27771 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27773 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27775 2017-04-25 Julian Brown <julian@codesourcery.com>
27776 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27778 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27779 (thunderx2t99_sha): New Reservation.
27781 2017-04-25 Julian Brown <julian@codesourcery.com>
27782 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27784 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27785 type for 1-element load.
27787 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27789 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27791 2017-04-24 Martin Jambor <mjambor@suse.cz>
27793 PR tree-optimization/80293
27794 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27795 char arrays not totally scalarizable if it is false.
27796 (analyze_all_variable_accesses): Pass correct value in the new
27797 parameter. Add a statistics counter.
27799 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27801 PR middle-end/79931
27802 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27804 2017-04-24 Richard Biener <rguenther@suse.de>
27806 PR tree-optimization/80494
27807 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27808 out for complex types.
27810 2017-04-24 Richard Biener <rguenther@suse.de>
27812 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27813 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27814 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27815 (DFS): Adjust and never fail.
27816 (sccvn_dom_walker::fail): Remove.
27817 (sccvn_dom_walker::before_dom_children): Adjust.
27818 (run_scc_vn): Likewise and never fail.
27819 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27820 (pass_fre::execute): Likewise.
27822 2017-04-24 Richard Biener <rguenther@suse.de>
27824 PR tree-optimization/79725
27825 * tree-ssa-sink.c (statement_sink_location): Return whether
27826 failure reason was zero uses. Move that check later.
27827 (sink_code_in_bb): Deal with zero uses by removing the stmt
27830 2017-04-24 Richard Biener <rguenther@suse.de>
27833 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27834 pointer-based references.
27836 2017-04-24 Richard Biener <rguenther@suse.de>
27839 * pass_manager.h (pass_manager::operator new): Remove.
27840 (pass_manager::operator delete): Likewise.
27841 * passes.c (pass_manager::operator new): Remove.
27842 (pass_manager::operator delete): Likewise.
27843 (pass_manager::pass_manager): Zero individual pass members.
27845 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27848 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27849 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27850 Check "XEXP (src, 1)" operand here.
27851 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27852 Check "XEXP (src, 1)" operand here.
27853 (dimode_scalar_chain::make_vector_copies): Detect count register
27854 of a shift instruction. Zero extend count register from QImode
27855 to DImode to satisfy vector shift pattern count operand predicate.
27856 Substitute vector shift count operand with a DImode copy.
27857 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27860 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27862 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27863 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27864 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27865 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27866 (UNSPEC_NOREX_MEM): Remove definition.
27868 2017-04-21 Richard Biener <rguenther@suse.de>
27870 PR tree-optimization/79547
27871 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27872 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27873 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27874 without any constraints.
27876 2017-04-21 Richard Biener <rguenther@suse.de>
27878 PR tree-optimization/78847
27879 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27881 2017-04-21 Richard Biener <rguenther@suse.de>
27883 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27884 (build_distinct_type_copy): Likewise.
27885 (build_variant_type_copy): Likewise.
27886 * tree.c (build_qualified_type): Pass down mem-stat info.
27887 (build_distinct_type_copy): Likewise.
27888 (build_variant_type_copy): Likewise.
27890 2017-04-21 Richard Biener <rguenther@suse.de>
27892 PR tree-optimization/80237
27893 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27895 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27896 for a simplified result.
27898 2016-04-21 Richard Biener <rguenther@suse.de>
27900 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27901 sth as strict as a simple_iv but a chrec without symbols and an
27902 operand defined in the loop we are peeling (and not some subloop).
27903 (propagate_constants_for_unrolling): Propagate all constants.
27905 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27908 * config/i386/i386.c (print_reg): Remove assert for disalowed
27909 regno values, call output_operand_lossage instead.
27911 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27914 * config/i386/constraints.md (Yc): New register constraint.
27915 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27916 Use Yc constraint for alternative 2 of operand 0. Remove
27917 preferred_for_speed attribute.
27919 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27921 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27922 lastprivate clauses in SIMT case.
27924 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27926 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27928 2017-04-20 Richard Biener <rguenther@suse.de>
27930 PR tree-optimization/57796
27931 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27932 as N scalar stores.
27933 (vect_model_load_cost): Cost gathers as N scalar loads.
27935 2017-04-20 Richard Biener <rguenther@suse.de>
27937 * ggc-page.c (ggc_allocated_p): Rename to ...
27938 (safe_lookup_page_table_entry): ... this and return the lookup
27940 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27942 2017-04-20 Richard Biener <rguenther@suse.de>
27944 PR tree-optimization/80453
27945 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27946 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27947 from the conditions.
27948 (vn_phi_eq): Pass them down.
27949 (vn_phi_lookup): Record them.
27950 (vn_phi_insert): Likewise.
27952 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27954 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27955 uninitialized variable warning to avoid buffer overrun.
27957 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27960 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27961 is suppressed for '{ 0 }' in C.
27963 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27965 * BASE-VER: Set to 8.0.0.
27967 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27969 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27970 priority .init_array and .fini_array section with SECTION_NOTYPE
27973 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27975 PR middle-end/80423
27976 * tree.h (build_array_type): Add typeless_storage default argument.
27977 * tree.c (type_cache_hasher::equal): Also compare
27978 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27979 (build_array_type): Add typeless_storage argument, set
27980 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27982 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27983 (build_array_type): Likewise. Add typeless_storage argument.
27985 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27986 Jakub Jelinek <jakub@redhat.com>
27988 PR tree-optimization/80426
27989 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27990 operation on symbolic operands, also compute the overflow for the
27991 invariant part when the operation degenerates into a negation.
27993 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27996 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27997 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
28000 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
28002 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
28005 * config/avr/avr.c (tree.h): Include it.
28006 (cgraph.h): Include it.
28007 (avr_encode_section_info): Don't warn for uninitialized progmem
28008 variable if it's just an alias.
28010 2017-04-19 Richard Biener <rguenther@suse.de>
28013 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
28014 when needed by AutoPGO.
28016 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
28019 * doc/lto.texi: Remove an extra 'that'.
28021 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
28023 PR rtl-optimization/80429
28024 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
28025 are only used in debug insns.
28027 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
28028 Vladimir Makarov <vmakarov@redhat.com>
28030 * config/sparc/predicates.md (input_operand): Add comment. Return
28031 true for any memory operand when LRA is in progress.
28032 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
28034 2017-04-18 Jeff Law <law@redhat.com>
28037 * mips.md ({return,simple_return}_internal): Do not overwrite
28040 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28042 PR tree-optimization/80443
28043 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
28044 instead of adding 1, subtract -1 and similarly instead of subtracting
28047 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
28049 PR rtl-optimization/80357
28050 * haifa-sched.c (tmp_bitmap): New variable.
28051 (model_recompute): Handle duplicate use records.
28052 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
28053 (free_global_sched_pressure_data): Free it.
28055 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
28058 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
28059 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28060 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28061 instead of SYSTEM_HEADER_DIR.
28063 2017-04-18 Jeff Law <law@redhat.com>
28065 PR middle-end/80422
28066 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
28067 predecessors after walking up the insn chain.
28069 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28072 * dwarf2out.c (modified_type_die): Try harder not to emit internal
28073 sizetype type into debug info.
28075 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28078 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
28079 unneeded test for TARGET_UPPER_REGS_SF.
28080 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
28082 2017-04-18 Jakub Jelinek <jakub@redhat.com>
28085 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
28086 instead of gsi_after_labels.
28088 2017-04-18 Jeff Law <law@redhat.com>
28090 * regcprop.c (maybe_mode_change): Avoid creating copies of the
28094 2017-04-13 Jeff Law <law@redhat.com>
28095 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
28096 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
28098 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
28101 * config/avr/avr.c (intl.h): Include it.
28102 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
28104 2017-04-18 Martin Liska <mliska@suse.cz>
28106 PR gcov-profile/78783
28107 * gcov-tool.c (gcov_output_files): Validate that destination
28108 file is either removed by the tool or by a user.
28110 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
28111 Guy Benyei <guybe@mellanox.com>
28113 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
28114 block, and do not negate it, the stored id is already negative.
28116 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
28118 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
28120 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28123 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
28124 masks of options that should be turned off if the VSX vector
28125 options are turned off.
28126 (OTHER_P8_VECTOR_MASKS): Likewise.
28127 (OTHER_VSX_VECTOR_MASKS): Likewise.
28128 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
28129 rs6000_disable_incompatible_switches to validate no type switches
28131 (rs6000_incompatible_switch): New function to disallow turning on
28132 other vector options if -mno-vsx, -mno-power8-vector, or
28133 -mno-power9-vector are specified.
28135 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28137 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
28139 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28141 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
28142 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
28143 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
28144 (ARG_POINTER_CFA_OFFSET): Likewise.
28146 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
28148 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
28149 conditions to take advantage of various optimizations.
28151 2017-04-13 Jeff Law <law@redhat.com>
28153 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
28154 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
28155 (zero_extendsidi2_dext): Likewise.
28157 2017-04-13 Jakub Jelinek <jakub@redhat.com>
28160 * fold-const.c (fold_ternary_loc): Revert
28161 use op0 instead of fold_convert_loc (loc, type, arg0) part of
28164 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
28166 PR rtl-optimization/80343
28167 * lra-remat.c (update_scratch_ops): Assign original hard reg to
28168 new scratch pseudo.
28170 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
28173 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
28174 to ubsan_encode_value.
28176 2017-04-13 Jeff Law <law@redhat.com>
28178 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
28179 appearing in DEBUG_INSNs.
28181 2017-04-13 Martin Liska <mliska@suse.cz>
28183 PR gcov-profile/80413
28184 * gcov-io.c (gcov_write_string): Copy to buffer just when
28185 allocated size is greater than zero.
28187 2017-04-13 Jakub Jelinek <jakub@redhat.com>
28190 * dwarf2out.c (decls_for_scope): Ignore declarations of
28191 current_function_decl in BLOCK_NONLOCALIZED_VARS.
28193 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
28196 * ipa-visibility.c (non_local_p): Fix typos.
28197 (localize_node): When localizing symbol in same comdat group,
28198 dissolve the group only when we know external symbols are going
28200 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
28202 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28204 PR tree-optimization/79390
28205 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
28206 order does not result in usable sequence, retry with reversed operand
28212 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
28213 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
28214 op0 instead of fold_convert_loc (loc, type, arg0).
28216 2017-04-12 Jeff Law <law@redhat.com>
28218 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
28219 has a delay slot in the generated code.
28221 * config/cris/cris.md (cris_preferred_reload_class): Return
28222 GENNONACR_REGS rather than GENERAL_REGS.
28224 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28227 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
28228 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
28229 signedness of the result type.
28231 2017-04-12 Richard Biener <rguenther@suse.de>
28232 Jeff Law <law@redhat.com>
28234 PR tree-optimization/80359
28235 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
28236 trim stores to TARGET_MEM_REFs.
28238 2017-04-12 Richard Biener <rguenther@suse.de>
28240 PR tree-optimization/79390
28241 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
28242 threading case even more.
28244 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
28247 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
28248 for quad_address_p for TImode, instead of just not indexed_address.
28250 2017-04-12 Richard Biener <rguenther@suse.de>
28251 Bernd Edlinger <bernd.edlinger@hotmail.de>
28253 PR middle-end/79671
28254 * alias.c (component_uses_parent_alias_set_from): Handle
28255 TYPE_TYPELESS_STORAGE.
28256 (get_alias_set): Likewise.
28257 * tree-core.h (tree_type_common): Add typeless_storage flag.
28258 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
28259 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
28260 for types containing members with TYPE_TYPELESS_STORAGE.
28261 (place_field): Likewise.
28262 (layout_type): Likewise for ARRAY_TYPE.
28263 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
28264 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
28265 TYPE_TYPELESS_STORAGE.
28266 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
28268 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28271 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
28272 first argument to type.
28274 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28278 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
28279 CONST0_RTX (mode) rather than const0_rtx where appropriate.
28280 (rs6000_expand_binop_builtin): Likewise.
28281 (rs6000_expand_ternop_builtin): Likewise; also add missing
28282 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
28283 vshasigma built-ins.
28284 * doc/extend.texi: Document that vec_xxpermdi's third argument
28285 must be a constant.
28287 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
28289 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
28290 Use shift_const cost parameter when calculating gain of STV shifts.
28292 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
28294 PR rtl-optimization/70478
28295 * lra-constraints.c (process_alt_operands): Check memory for
28296 disfavoring memory insn operand.
28298 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28300 PR middle-end/80100
28301 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
28302 left shift in unsigned HOST_WIDE_INT type.
28304 PR rtl-optimization/80385
28305 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
28306 (not (neg X)) into (plus X -1) for complex or non-integral modes.
28309 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
28310 if they have any depend clauses.
28312 2017-04-11 Martin Liska <mliska@suse.cz>
28315 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
28316 * ipa-split.c (split_function): Create a local comdat symbol
28317 if caller is in a comdat group.
28319 2017-04-11 Martin Liska <mliska@suse.cz>
28322 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
28325 2017-04-11 Martin Sebor <msebor@redhat.com>
28327 PR middle-end/80364
28328 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
28329 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
28331 (directive::set_width, directive::set_precision, format_character):
28333 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
28336 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
28339 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
28340 conflict, set target->arch_name instead of target->cpu_name.
28342 2017-04-11 Richard Biener <rguenther@suse.de>
28344 PR tree-optimization/80374
28345 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
28346 build_zero_cst, remove fold_convertible_p check again.
28348 2017-04-11 Martin Liska <mliska@suse.cz>
28351 * ubsan.c (instrument_object_size): Do not instrument register
28354 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28357 * config/i386/i386-builtin-types.def
28358 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
28359 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
28360 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
28361 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
28362 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
28363 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
28364 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
28365 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
28366 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
28367 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
28368 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
28369 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
28370 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
28371 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
28372 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
28373 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
28374 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
28375 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
28376 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
28377 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
28378 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
28379 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
28380 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
28381 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
28382 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
28383 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
28384 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
28385 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
28386 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
28387 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
28388 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
28389 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
28390 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
28391 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
28392 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
28393 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
28394 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
28395 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
28396 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
28397 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
28398 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
28399 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
28400 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
28401 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
28402 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
28403 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
28405 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
28406 flag to second_arg_count, handle 4 argument function type _COUNT
28407 aliases, handle second_arg_count on second argument rather than last.
28409 2017-04-10 Jeff Law <law@redhat.com>
28411 PR tree-optimization/80374
28412 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
28413 record anything if we can not convert integer_zero_node to the
28416 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28419 * config/rs6000/rs6000.c (rs6000_option_override_internal):
28420 Enhance special handling given to the TARGET_P9_MINMAX option in
28421 relation to certain other options.
28423 2017-04-10 Bin Cheng <bin.cheng@arm.com>
28425 PR tree-optimization/80153
28426 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
28427 remove POINTER_PLUS_EXPR's base part directly, rather than through
28430 2017-04-10 Richard Biener <rguenther@suse.de>
28431 Bin Cheng <bin.cheng@arm.com>
28433 PR tree-optimization/80153
28434 * tree-affine.c (aff_combination_to_tree): Get base pointer from
28435 the first element of pointer type aff_tree. Build result expr in
28437 (add_elt_to_tree): Convert to type unconditionally. Remove other
28438 fold_convert calls.
28439 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
28440 (rewrite_use_nonlinear_expr): Check invariant using iv information.
28442 2017-04-10 Richard Biener <rguenther@suse.de>
28444 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28447 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
28449 PR rtl-optimization/70478
28450 * lra-constraints.c (curr_small_class_check): New.
28451 (update_and_check_small_class_inputs): New.
28452 (process_alt_operands): Update curr_small_class_check. Disfavor
28453 alternative insn memory operands. Check available regs for small
28456 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28459 * config/mips/mips.opt (-mvirt): Update description.
28460 * doc/invoke.texi (-mvirt): Likewise.
28462 2017-04-10 Richard Biener <rguenther@suse.de>
28464 PR middle-end/80362
28465 * fold-const.c (fold_binary_loc): Look at unstripped ops when
28466 looking for NEGATE_EXPR in -A / -B to A / B folding.
28468 2017-04-10 Martin Liska <mliska@suse.cz>
28470 PR gcov-profile/80224
28471 * gcov.c (print_usage): Fix usage string.
28472 (get_gcov_intermediate_filename): Remove.
28473 (output_gcov_file): Use both for normal and intermediate format.
28474 (generate_results): Do not initialize special file for
28475 intermediate format.
28477 2017-04-10 Richard Biener <rguenther@suse.de>
28479 PR tree-optimization/80304
28480 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
28483 2017-04-10 Nathan Sidwell <nathan@acm.org>
28486 * config/rs6000/rs6000.c (rs6000_vector_type): New.
28487 (rs6000_init_builtins): Use it.
28489 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28491 * config/arm/arm.md (<mrc>): Add mode to SET source.
28492 (<mrrc>): Likewise.
28494 2017-04-10 Richard Biener <rguenther@suse.de>
28496 PR middle-end/80344
28497 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
28499 2017-04-10 Jakub Jelinek <jakub@redhat.com>
28502 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
28503 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
28504 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
28505 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
28506 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
28507 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
28508 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
28509 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
28510 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
28511 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
28512 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
28513 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
28514 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
28515 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
28516 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
28517 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
28518 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
28519 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
28520 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
28521 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
28522 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
28523 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
28524 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
28526 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
28528 PR rtl-optimization/70478
28529 * lra-constraints.c: Reverse the last patch.
28531 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
28533 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
28534 Add comment for WCHAR_T.
28536 2017-04-08 Martin Liska <mliska@suse.cz>
28539 2017-04-07 Martin Liska <mliska@suse.cz>
28542 * ipa-split.c (split_function): Add function part to a same comdat
28545 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28548 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
28550 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
28552 * rs6000/rs6000.c (vec_load_pendulum): Rename...
28553 (vec_pairing): ...to this.
28554 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
28555 (rs6000_sched_init): Adjust for name change.
28556 (struct rs6000_sched_context): Likewise.
28557 (rs6000_init_sched_context): Likewise.
28558 (rs6000_set_sched_context): Likewise.
28560 2017-04-07 Jakub Jelinek <jakub@redhat.com>
28566 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
28568 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
28569 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
28570 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
28572 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
28574 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
28576 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28578 PR rtl-optimization/70703
28579 * ira-color.c (update_conflict_hard_regno_costs): Use
28580 int64_t instead of HOST_WIDE_INT.
28582 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28584 PR rtl-optimization/70478
28585 * lra-constraints.c (process_alt_operands): Disfavor alternative
28586 insn memory operands.
28588 2017-04-07 Jeff Law <law@redhat.com>
28590 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
28591 CALL and NOTE_INSN_CALL_ARG_LOCATION.
28593 2017-04-07 Martin Liska <mliska@suse.cz>
28596 * config/aarch64/aarch64.c (aarch64_process_target_attr):
28597 Show error message instead of an ICE.
28599 2017-04-07 Martin Liska <mliska@suse.cz>
28602 * ipa-split.c (split_function): Add function part to a same comdat
28605 2017-04-07 Richard Biener <rguenther@suse.de>
28607 PR middle-end/80341
28608 * tree.c (get_unwidened): Also handle ! for_type case for
28610 * convert.c (do_narrow): Split out from ...
28611 (convert_to_integer_1): ... here. Do not pass final truncation
28612 type to get_unwidened for TRUNC_DIV_EXPR.
28614 2017-04-07 Richard Biener <rguenther@suse.de>
28616 * tree-affine.c (wide_int_ext_for_comb): Take type rather
28618 (aff_combination_const): Adjust.
28619 (aff_combination_scale): Likewise.
28620 (aff_combination_add_elt): Likewise.
28621 (aff_combination_add_cst): Likewise.
28622 (aff_combination_convert): Likewise.
28623 (add_elt_to_tree): Likewise. Remove unused argument.
28624 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
28626 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
28628 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
28630 * config/arm/arm.c (arm_default_short_enums): Use
28631 ARM_DEFAULT_SHORT_ENUMS.
28632 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
28634 2017-04-06 Jakub Jelinek <jakub@redhat.com>
28637 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
28638 members with redundant out-of-class redeclaration.
28640 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28643 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
28644 * config/i386/i386.md (*zero_extendsidi2):
28645 Add (?*x,*x) and (?*v,*v) alternatives.
28647 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28650 * config/i386/i386.c (ix86_expand_builtin)
28651 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28652 mode from insn data. Convert operands to insn operand mode.
28653 Copy operands that don't satisfy insn predicate to a register.
28655 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28657 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28660 2017-04-06 Richard Biener <rguenther@suse.de>
28662 PR tree-optimization/80334
28663 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28664 preserve alignment of accesses.
28666 2017-04-06 Richard Biener <rguenther@suse.de>
28668 PR tree-optimization/80262
28669 * tree-sra.c (build_ref_for_offset): Preserve address-space
28671 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28672 Drop useless address-space information on MEM_REF offsets.
28674 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28676 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28678 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28680 PR rtl-optimization/70703
28681 * ira-color.c (update_conflict_hard_regno_costs): Use
28682 HOST_WIDE_INT instead of long.
28684 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28687 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28688 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28690 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28691 for x86_64 target. Handle -m3dnowa option.
28693 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28695 PR rtl-optimization/70703
28696 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28697 (update_conflict_hard_regno_costs): Use long instead of unsigned
28698 arithmetic for cost calculation.
28700 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28701 Bernd Edlinger <bernd.edlinger@hotmail.de>
28704 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28707 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28710 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28711 ptr_mode with Pmode throughout.
28712 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28713 into probe_stack_range and use DImode.
28715 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28718 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28719 call_eh_return is true.
28721 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28723 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28724 Initialize last_match_fntype_index.
28726 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28729 * tree-nvr.c: Include internal-fn.h.
28730 (pass_return_slot::execute): Ignore internal calls without
28733 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28734 Richard Biener <rguenther@suse.de>
28737 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28738 captures used multiple times, except for the last use.
28739 * generic-match-head.c: Include gimplify.h.
28741 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28743 PR tree-optimization/79390
28744 * target.h (struct noce_if_info): Declare.
28745 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28746 * target.def (noce_conversion_profitable_p): New target hook.
28747 * ifcvt.h (struct noce_if_info): New type, moved from ...
28748 * ifcvt.c (struct noce_if_info): ... here.
28749 (noce_conversion_profitable_p): Renamed to ...
28750 (default_noce_conversion_profitable_p): ... this. No longer
28752 (noce_try_store_flag_constants, noce_try_addcc,
28753 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28754 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28755 instead of noce_conversion_profitable_p.
28756 * config/i386/i386.c: Include ifcvt.h.
28757 (ix86_option_override_internal): Don't override
28758 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28759 (ix86_noce_conversion_profitable_p): New function.
28760 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28761 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28762 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28763 * doc/tm.texi: Regenerated.
28765 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28767 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28770 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28773 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28774 instructions for small multiply cores.
28776 2017-04-04 Jeff Law <law@redhat.com>
28778 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28780 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28781 that are not set by the loop over the elements.
28783 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28786 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28787 int mode, convert_modes it to mode as unsigned, otherwise use
28788 lowpart_subreg to mode rather than SImode.
28789 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28790 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28791 Use DImode instead of SImode for the shift count operand.
28792 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28795 2017-04-04 Richard Biener <rguenther@suse.de>
28797 PR middle-end/80281
28798 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28799 arithmetic done for the negate or the plus. Simplify.
28800 (A - (-B) -> A + B): Likewise.
28801 * fold-const.c (split_tree): Make sure to not negate pointers.
28803 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28805 PR rtl-optimization/60818
28806 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28807 a compare of comparisons with the thing compared if this results
28808 in a different machine mode.
28810 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28812 * alias.c (base_alias_check): Fix typo in comment.
28813 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28814 * cgraphunit.c (symbol_table::compile): Likewise.
28815 * collect2.c (maybe_run_lto_and_relink): Likewise.
28816 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28817 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28818 * config/avr/avr.c (avr_map_op_t): Likewise.
28819 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28820 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28821 * config/epiphany/epiphany.md (movcc): Likewise.
28822 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28823 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28825 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28826 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28827 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28828 * config/sh/sh.c (sh_rtx_costs): Likewise.
28829 * fold-const.c (fold_truth_andor): Likewise.
28830 * genautomata.c (collapse_flag): Likewise.
28831 * gengtype.h (struct type::u::s): Likewise.
28832 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28833 * input.c (FORMAT_AMOUNT): Likewise.
28834 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28835 (known_aggs_to_agg_replacement_list): Likewise.
28836 * ipa-inline-analysis.c: Likewise.
28837 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28838 * ipa-polymorphic-call.c
28839 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28840 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28841 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28843 * modulo-sched.c (apply_reg_moves): Likewise.
28844 * omp-expand.c (build_omp_regions_1): Likewise.
28845 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28846 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28847 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28848 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28849 * value-prof.c: Likewise.
28850 * var-tracking.c (val_reset): Likewise.
28852 2017-04-03 Richard Biener <rguenther@suse.de>
28854 PR tree-optimization/80275
28855 * fold-const.c (split_address_to_core_and_offset): Handle
28858 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28860 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28861 descriptors is at least equal to that of functions.
28863 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28865 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28867 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28870 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28871 (mov<IMOD4:mode>): New expander.
28872 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28874 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28876 PR rtl-optimization/79405
28877 * fwprop.c (propagations_left): New variable.
28878 (forward_propagate_into): Decrement it.
28879 (fwprop_init): Initialize it.
28880 (fw_prop): If the variable has reached zero, stop propagating.
28881 (fwprop_addr): Ditto.
28883 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28886 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28887 a FUNCTION_DECL, pass it as decl instead of origin to
28890 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28892 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28895 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28898 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28899 TARGET_VSX_SMALL_INTEGER.
28901 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28903 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28904 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28906 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28908 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28909 extraction from odd-numbered MSA register.
28911 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28913 PR middle-end/80173
28914 * expmed.c (store_bit_field_1): Don't attempt to create
28915 a word subreg out of hard registers wider than word if they
28916 have HARD_REGNO_NREGS of 1 for their mode.
28918 PR middle-end/80163
28919 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28920 conversions to integer types wider than word and pointer.
28923 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28924 (rtx_equal_for_cselib_p): Pass 0 to it.
28925 * cselib.c (cselib_hasher::equal): Likewise.
28926 (rtx_equal_for_cselib_1): Add depth argument. If depth
28927 is 128, don't look up VALUE locs and punt. Increment
28928 depth in recursive calls when walking VALUE locs.
28930 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28932 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28933 (make_gcov_file_name): Use the canonical path name for generating
28935 (read_line): Fix handling of files with ascii null bytes.
28937 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28939 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28940 to initialise a vector register instead
28941 of using a const_int.
28943 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28945 PR translation/80189
28946 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28947 diagnostic messages.
28949 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28952 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28953 (dfp_diex_<mode>): Update mode of operand 1.
28954 * doc/extend.texi (dxex, dxexq): Document change to return type.
28955 (diex, diexq): Document change to argument type.
28957 2017-03-30 Martin Jambor <mjambor@suse.cz>
28960 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28961 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28962 it reflects the signature changes performed at the callee side.
28963 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28964 to cgraph_build_function_type_skip_args.
28965 (build_function_decl_skip_args): Adjust call to the above function.
28967 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28970 * config/i386/sse.md
28971 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28972 register as dest whenever it is a MEM not rtx_equal_p to the
28973 corresponding dup operand, and when forcing into reg move the
28974 reg into the memory afterwards.
28975 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28976 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28977 for the force_reg mode.
28978 (avx512vl_vextractf128<mode>): Use register as dest either
28979 always when a MEM, or when it is a MEM not rtx_equal_p to the
28980 corresponding dup operand, or even not when it is a CONST_VECTOR
28981 depending on the mode and lo vs. hi.
28982 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28984 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28985 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28986 Likewise. Require that operands[2] is even.
28987 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28988 Remove extraneous parens. Require that operands[2] is a multiple
28990 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28991 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28992 disallow memory then.
28994 2017-03-30 Richard Biener <rguenther@suse.de>
28996 PR tree-optimization/77498
28997 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28998 to non-constants over backedges.
29000 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
29002 PR rtl-optimization/80233
29003 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
29004 as last_combined_insn. Do not test for BARRIER_P separately.
29006 2017-03-29 Andreas Schwab <schwab@suse.de>
29009 * calls.c (prepare_call_address): Convert funexp to Pmode before
29010 copying to temp reg.
29012 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29014 PR tree-optimization/80158
29015 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
29016 Handle possible future case of more than one alternate
29018 (replace_rhs_if_not_dup): Likewise.
29019 (replace_one_candidate): Likewise.
29021 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
29023 PR rtl-optimization/80193
29024 * ira.c (ira): Do not check allocation for LRA.
29026 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
29028 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
29029 (nvptx_output_simt_exit): Declare.
29030 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
29031 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
29032 (init_softstack_frame): Move initialization of crtl->is_leaf to...
29033 (nvptx_declare_function_name): ...here. Emit declaration of local
29034 memory space buffer for omp_simt_enter insn.
29035 (nvptx_output_unisimt_switch): New.
29036 (nvptx_output_softstack_switch): New.
29037 (nvptx_output_simt_enter): New.
29038 (nvptx_output_simt_exit): New.
29039 * config/nvptx/nvptx.h (struct machine_function): New fields
29040 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
29041 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
29042 (UNSPECV_SIMT_EXIT): Ditto.
29043 (omp_simt_enter_insn): New insn.
29044 (omp_simt_enter): New expansion.
29045 (omp_simt_exit): New insn.
29046 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
29048 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
29049 (expand_GOMP_SIMT_ENTER_ALLOC): New.
29050 (expand_GOMP_SIMT_EXIT): New.
29051 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
29052 (GOMP_SIMT_ENTER_ALLOC): Ditto.
29053 (GOMP_SIMT_EXIT): Ditto.
29054 * target-insns.def (omp_simt_enter): New insn.
29055 (omp_simt_exit): Ditto.
29056 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
29058 (lower_rec_simd_input_clauses): Implement SIMT privatization.
29059 (lower_rec_input_clauses): Likewise.
29060 (lower_lastprivate_clauses): Handle SIMT privatization.
29062 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
29063 (ompdevlow_adjust_simt_enter): New.
29064 (find_simtpriv_var_op): New.
29065 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
29066 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
29068 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
29069 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
29070 (copy_decl_for_dup_finish): Ditto.
29072 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
29074 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
29077 * config/i386/i386.c (ix86_option_override_internal): Always
29078 allow -mpreferred-stack-boundary=3 for 64-bit targets.
29080 2017-03-28 Bin Cheng <bin.cheng@arm.com>
29082 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
29084 2017-03-28 Bin Cheng <bin.cheng@arm.com>
29086 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
29087 mark new edge's irreducible flag accordign to it.
29088 (vect_do_peeling): Check loop preheader edge's irreducible flag
29089 and pass it to function slpeel_add_loop_guard.
29091 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
29093 PR tree-optimization/80218
29094 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
29095 Update block frequencies and counts.
29097 2017-03-28 Richard Biener <rguenther@suse.de>
29099 PR tree-optimization/78644
29100 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
29101 of a simplification result we may not use it at all.
29103 2017-03-28 Richard Biener <rguenther@suse.de>
29106 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
29107 without arguments, generate default definition of a SSA name.
29109 2017-03-28 Richard Biener <rguenther@suse.de>
29111 PR middle-end/80222
29112 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
29113 TYPE_REF_CAN_ALIAS_ALL references.
29114 * fold-const.c (fold_indirect_ref_1): Likewise.
29116 2017-03-28 Martin Liska <mliska@suse.cz>
29119 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
29120 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
29122 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
29123 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
29125 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
29126 (EXTRA_SPECS): Define.
29127 (SUBTARGET_EXTRA_SPECS): Likewise.
29128 (SUBTARGET_CPP_SPEC): Likewise.
29129 * config/arc/elf.h (EXTRA_SPECS): Renamed to
29130 SUBTARGET_EXTRA_SPECS.
29131 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
29133 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
29135 * config/arc/simdext.md (vst64_insn): Update pattern.
29136 (vld32wh_insn): Likewise.
29137 (vld32wl_insn): Likewise.
29138 (vld64_insn): Likewise.
29139 (vld32_insn): Likewise.
29141 2017-03-28 Marek Polacek <polacek@redhat.com>
29144 * fold-const.c (fold_comparison): Use protected_set_expr_location
29145 instead of SET_EXPR_LOCATION.
29147 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
29149 * tree.c (add_expr): Avoid name lookup warning.
29151 2017-03-27 Jeff Law <law@redhat.com>
29153 PR tree-optimization/80216
29154 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
29155 function name. Limit recursion depth.
29156 (record_temporary_equivalences): Corresponding changes.
29158 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
29160 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
29163 2017-03-27 Jakub Jelinek <jakub@redhat.com>
29166 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
29168 * cfgcleanup.c (reg_note_cfa_p): New array.
29169 (insns_have_identical_cfa_notes): New function.
29170 (old_insns_match_p): Don't cross-jump in between /f
29171 and non-/f instructions. If both i1 and i2 are frame related,
29172 verify all CFA notes, their order and content.
29174 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
29177 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
29178 HImode and SImode with zero extend to DImode to one insn.
29179 (bswap<mode>2_extenddi): Likewise.
29180 (bswapsi2_extenddi): Likewise.
29181 (bswaphi2_extendsi): Likewise.
29182 (bswaphi2): Combine bswap HImode and SImode into one insn.
29183 Separate memory insns from swapping register.
29184 (bswapsi2): Likewise.
29185 (bswap<mode>2): Likewise.
29186 (bswaphi2_internal): Delete, no longer used.
29187 (bswapsi2_internal): Likewise.
29188 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
29189 store, and gpr<-gpr swap insns.
29190 (bswap<mode>2_store): Likewise.
29191 (bswaphi2_reg): Register only splitter, combine with the splitter.
29192 (bswaphi2 splitter): Likewise.
29193 (bswapsi2_reg): Likewise.
29194 (bswapsi2 splitter): Likewise.
29195 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
29196 the insns into load, store, and register/register insns.
29197 (bswapdi2_ldbrx): Likewise.
29198 (bswapdi2_load): Likewise.
29199 (bswapdi2_store): Likewise.
29200 (bswapdi2_reg): Likewise.
29202 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
29204 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
29205 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
29207 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
29210 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
29212 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29213 special handling for target option conflicts between dform
29214 options (-mpower9-dform, -mpower9-dform-vector,
29215 -mpower9-dform-scalar) and -mno-direct-move.
29217 2017-03-27 Richard Biener <rguenther@suse.de>
29219 PR tree-optimization/80181
29220 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
29222 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29224 * config/arc/predicates.md (move_double_src_operand): Replace the
29225 call to move_double_src_operand with a call to address_operand.
29227 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29229 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
29230 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
29231 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
29233 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29235 * config/arc/predicates.md (long_immediate_loadstore_operand):
29236 Consider scaled addresses cases.
29238 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29240 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
29241 restored when in interrupt.
29242 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
29243 doesn't have delay slot.
29245 2017-03-27 Richard Biener <rguenther@suse.de>
29248 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
29249 inlined thunk clones.
29251 2017-03-27 Jakub Jelinek <jakub@redhat.com>
29254 * asan.c (instrument_derefs): Copy over last operand from
29255 original COMPONENT_REF to the new COMPONENT_REF with
29256 DECL_BIT_FIELD_REPRESENTATIVE.
29257 * ubsan.c (instrument_object_size): Likewise.
29259 2017-03-27 Richard Biener <rguenther@suse.de>
29261 PR tree-optimization/80170
29262 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
29263 sure DR/SCEV didnt fold in constants we do not see when looking
29264 at the reference base alignment.
29266 2017-03-27 Richard Biener <rguenther@suse.de>
29268 PR middle-end/80171
29269 * gimple-fold.c (fold_ctor_reference): Properly guard against
29270 NULL return value from canonicalize_constructor_val.
29272 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
29275 * config/i386/i386.c (ix86_expand_builtin)
29276 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
29277 flags reg setting and flags reg using instructions.
29278 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
29279 clobbering instructions to zero extend op2.
29281 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
29283 * doc/install.texi (Configuration) <--with-aix-soname>:
29284 Update link to AIX ld.
29286 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
29288 PR rtl-optimization/80160
29289 PR rtl-optimization/80159
29290 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
29291 reg_alternate_class into account.
29293 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
29296 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
29297 to consider in curr_insn_transform.
29299 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29301 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
29302 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
29303 and emit_mode_inner.
29305 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29307 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
29308 argument to the overloaded builtin variants. Use the new flag to
29309 deprecate certain builtin variants.
29310 * config/s390/s390-builtin-types.def: Add new builtin types.
29311 * config/s390/s390-builtins.h: Support new flags field for
29312 overloaded builtins.
29313 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
29314 (s390_macro_to_expand): Enable vector float data type.
29315 (s390_cpu_cpp_builtins_internal): Indicate support of the new
29316 builtins by incrementing the __VEC__ version number.
29317 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
29319 (s390_resolve_overloaded_builtin): Emit error messages depending
29320 on the builtin flags.
29321 * config/s390/s390.c (s390_expand_builtin): Support additional
29322 flags argument. Change error message to match the messages
29323 emitted in s390-c.c.
29324 * config/s390/s390.md: New UNSPEC_* constants.
29325 (op_type): Add new instruction types.
29326 * config/s390/vecintrin.h: Add new builtins and test data class
29328 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
29329 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
29330 (VEC_INEXACT, VEC_NOINEXACT): New constants.
29331 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
29332 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
29333 ("vec_mergel<mode>"): V_HW -> VEC_HW.
29335 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
29336 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
29337 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
29338 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
29340 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
29341 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
29342 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
29343 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
29345 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
29346 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
29347 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
29348 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
29349 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
29350 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
29351 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
29353 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
29354 ("vec_scatter_element<V_HW_4:mode>_DI")
29355 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
29356 ("vec_fpint<mode>", "vflls")
29357 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
29358 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
29359 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
29360 ("*vec_cmphe<mode>_cc"): ... these.
29362 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
29363 mode constant instead of magic value.
29365 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29367 * config/s390/s390.c (s390_expand_vec_compare): Support other
29368 vector floating point modes than just V2DF.
29369 (s390_expand_vcond): Likewise.
29370 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
29371 (s390_cannot_change_mode_class): Prevent mode changes between TF
29372 and V1TF in vector registers.
29373 * config/s390/s390.md (DF, SF): New mode attributes.
29374 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
29375 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
29376 SFmode support for VRs.
29377 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
29379 (VFT, VF_HW): New mode iterators.
29380 (vw, sdx): New mode attributes.
29381 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
29382 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
29383 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
29384 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
29385 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
29386 also the new vector floating point modes. Renaming to ...
29388 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
29389 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
29390 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
29391 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
29392 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
29393 ("vec_unordered<mode>"): ... these.
29395 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
29396 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
29397 ("*vec_extendv2df"): New insn definitions.
29399 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29401 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
29402 ("mulditi3_2", "*muldi3_sign"): New patterns.
29403 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
29404 rename the pattern definition.
29406 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29408 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
29410 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
29412 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29414 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
29415 instruction if possible.
29416 * config/s390/vector.md (vec_halfnumelts): New mode
29418 ("*vec_vllezlf<mode>"): New pattern.
29420 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29422 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
29423 ("popcountv4si2", "popcountv2di2"): Rename to ...
29424 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
29425 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
29427 ("popcount<mode>2_vxe"): New pattern.
29429 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29431 * common/config/s390/s390-common.c (processor_flags_table): Add
29433 * config.gcc: Add arch12.
29434 * config/s390/driver-native.c (s390_host_detect_local_cpu):
29435 Default to arch12 for unknown CPU model numbers.
29436 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
29437 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
29438 PROCESSOR_max sanity check.
29439 * config/s390/s390-opts.h (enum processor_type): Add
29441 * config/s390/s390.c (processor_table): Add arch12.
29442 (s390_expand_builtin): Add check for B_VXE flag.
29443 (s390_issue_rate): Add PROCESSOR_ARCH12.
29444 (s390_get_sched_attrmask): Likewise.
29445 (s390_get_unit_mask): Likewise.
29446 (s390_sched_score): Enable z13 scheduling for arch12.
29447 (s390_sched_reorder): Likewise.
29448 (s390_sched_variable_issue): Likewise.
29449 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
29451 (s390_tune_attr): Use z13 scheduling also for arch12.
29452 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
29453 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
29454 (TARGET_VXE_P): New macros.
29455 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
29456 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
29457 * config/s390/s390.opt: Add arch12 as processor_type.
29459 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29461 * config/s390/s390.md
29462 ("fixuns_truncdddi2", "fixuns_trunctddi2")
29463 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
29464 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
29466 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
29467 Rename expanders to ...
29469 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
29470 ("fixuns_truncdddi2_emu"): ... these.
29472 ("fixuns_trunc<mode>si2_emu"): New expander.
29474 ("*fixuns_truncdfdi2_z13"): Rename to ...
29475 ("*fixuns_truncdfdi2_vx"): ... this.
29477 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29479 * config/s390/2964.md: Remove the single element vector compare
29480 instructions which are no longer used.
29481 * config/s390/s390.c (s390_select_ccmode): Remove handling of
29483 (s390_canonicalize_comparison): Remove handling of DFmode
29485 (s390_expand_vec_compare_scalar): Remove function.
29486 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
29487 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
29489 ("*cmp<mode>_ccs"): Add wfcdb instruction.
29491 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29493 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
29495 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
29496 will anyway by matched by mov<mode>_64dfp.
29498 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29500 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
29501 vlef/vstef. Add missing operand to vleif.
29503 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29505 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
29506 pair for all vector types with 64 bit elements.
29507 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
29508 * config/s390/vector.md (V_HW_64): ... here.
29509 (V_128_NOSINGLE): New mode iterator.
29510 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
29511 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
29512 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
29513 ("*vec_load_pairv2di"): Change to ...
29514 ("*vec_load_pair<mode>"): ... this one.
29516 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29518 * config/s390/constraints.md: Add comments.
29519 (jKK): Reject element sizes > 8 bytes.
29520 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
29522 * config/s390/s390.md: Add the s_operand checks formerly in
29523 s390_split_ok_p to various splitters where they are still
29525 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
29526 for 128 bit vectors. Plus two splitters.
29528 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29530 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
29533 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29536 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
29537 error if the boundary argument is not constant.
29539 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29541 PR rtl-optimization/80112
29542 * loop-doloop.c (doloop_condition_get): Don't check condition
29543 if cmp isn't SET with IF_THEN_ELSE src.
29545 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29547 PR tree-optimization/80158
29548 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
29549 replacing a candidate statement, also replace it for the
29550 candidate's alternate interpretation.
29551 (replace_rhs_if_not_dup): Likewise.
29552 (replace_one_candidate): Likewise.
29554 2017-03-24 Richard Biener <rguenther@suse.de>
29556 PR tree-optimization/80167
29557 * graphite-isl-ast-to-gimple.c
29558 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
29560 (translate_isl_ast_to_gimple::get_rename): Likewise.
29562 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29564 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
29565 handling of certain combinations of target options, including the
29566 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
29567 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
29569 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29572 * config/arm/arm.md (*load_multiple): Add reload_completed to
29573 matching condition.
29575 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29576 Richard Biener <rguenth@suse.de>
29578 PR tree-optimization/79908
29579 PR tree-optimization/80136
29580 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29581 been cast away, gimplify_and_add suffices.
29583 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
29585 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
29587 2017-03-23 Richard Biener <rguenther@suse.de>
29589 PR tree-optimization/80032
29590 * gimplify.c (gimple_push_cleanup): Forced unconditional
29591 cleanups still have to go to the conditional_cleanups
29594 2017-03-22 Jakub Jelinek <jakub@redhat.com>
29596 PR tree-optimization/80072
29597 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
29599 (next_operand_entry_id): Change type to unsigned int.
29600 (sort_by_operand_rank): Make sure to return the right return value
29601 even if unsigned fields are bigger than INT_MAX.
29602 (struct oecount): Change cnt and id type to unsigned int.
29603 (oecount_hasher::equal): Formatting fix.
29604 (oecount_cmp): Make sure to return the right return value
29605 even if unsigned fields are bigger than INT_MAX.
29606 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
29609 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
29610 TREE_READONLY on result if writing it more than once.
29613 * doc/invoke.texi (-fsanitize=thread): Document that with
29614 -fnon-call-exceptions atomics are not able to throw
29618 * tsan.c: Include tree-eh.h.
29619 (instrument_builtin_call): Call maybe_clean_eh_stmt or
29620 maybe_clean_or_replace_eh_stmt where needed.
29621 (instrument_memory_accesses): Add cfg_changed argument.
29622 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
29623 if it returned true.
29624 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
29626 PR rtl-optimization/63191
29627 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
29628 wrapper function, moved the whole old content into ...
29629 (ix86_delegitimize_address_1): ... this. New inline function.
29630 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
29631 true as last argument instead of ix86_delegitimize_address.
29633 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29635 * config/aarch64/aarch64.c (generic_branch_cost): Copy
29636 cortexa57_branch_cost.
29638 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29640 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
29642 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29645 * doc/md.texi (Constraints): Document wA constraint.
29646 * config/rs6000/constraints.md (wA): New.
29647 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
29648 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29649 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29650 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29652 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29655 * gimplify.c (is_oacc_declared): New function.
29656 (oacc_default_clause): Use it to set default flags for acc declared
29657 variables inside parallel regions.
29658 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29659 declared variables.
29660 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29661 declare attribute to any decl as necessary.
29663 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29666 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29667 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29668 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29669 (arm_arch_lpae): This.
29670 * config/arm/arm.c (arm_arch7ve): Rename into ...
29671 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29672 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29675 2017-03-22 Martin Liska <mliska@suse.cz>
29678 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29679 error message instead of an ICE.
29681 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29683 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29685 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29687 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29689 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29692 2017-03-21 Martin Sebor <msebor@redhat.com>
29694 * doc/extend.texi: Use "cannot" instead of "can't."
29695 * doc/hostconfig.texi: Same.
29696 * doc/install.texi: Same.
29697 * doc/invoke.texi: Same.
29698 * doc/loop.texi: Same.
29699 * doc/md.texi: Same.
29700 * doc/objc.texi: Same.
29701 * doc/rtl.texi: Same.
29702 * doc/tm.texi: Same.
29703 * doc/tm.texi.in: Same.
29704 * doc/trouble.texi: Same.
29706 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29709 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29710 (collect_checksum_attributes): Set it.
29711 (die_checksum_ordered): Use it.
29713 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29715 PR tree-optimization/79908
29716 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29717 change: For a VA_ARG whose LHS has been cast away, use
29718 force_gimple_operand to construct the side effects.
29720 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29722 PR translation/80001
29723 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29724 more amenable to translation.
29725 (oacc_loop_auto_partitions): Likewise.
29727 2017-03-21 Marek Polacek <polacek@redhat.com>
29728 Martin Sebor <msebor@redhat.com>
29730 PR tree-optimization/80109
29731 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29732 on INTEGRAL_TYPE_P.
29734 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29735 Segher Boessenkool <segher@kernel.crashing.org>
29738 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29739 check reg_used_between_p between insn and one of succ or succ2
29740 depending on if succ is artificial insn not inserted into insn
29743 2017-03-21 Martin Liska <mliska@suse.cz>
29745 PR gcov-profile/80081
29746 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29747 * doc/gcc.texi: Include gcov-dump stuff.
29748 * doc/gcov-dump.texi: New file.
29750 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29752 PR rtl-optimization/79150
29753 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29754 conditional jump, if the jump is the last insn of the loop.
29756 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29757 Richard Biener <rguenth@suse.de>
29759 PR tree-optimization/79908
29760 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29761 been cast away, use force_gimple_operand to construct the side
29764 2017-03-21 Martin Liska <mliska@suse.cz>
29766 PR libfortran/79956
29767 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29770 2017-03-21 Brad Spengler <spender@grsecurity.net>
29773 * plugin.c (htab_hash_plugin): New function.
29774 (add_new_plugin): Use it and adjust.
29775 (parse_plugin_arg_opt): Adjust.
29776 (init_one_plugin): Likewise.
29778 2017-03-21 Richard Biener <rguenther@suse.de>
29780 PR tree-optimization/80032
29781 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29782 if set force the cleanup to happen unconditionally.
29783 (gimplify_target_expr): Push inserted clobbers with force_uncond
29784 to avoid them being removed by control-dependent DCE.
29786 2017-03-21 Richard Biener <rguenther@suse.de>
29788 PR tree-optimization/80122
29789 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29790 va_arg_pack_len when the inlined call stmt requires pack
29792 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29794 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29797 * tsan.c (instrument_builtin_call): If the memory model argument
29798 is not a constant, assume it is valid.
29801 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29804 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29806 PR rtl-optimization/79910
29807 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29808 if its dest is used by an insn before I2 (other than the combined
29809 insns themselves, which are properly handled already).
29811 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29814 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29816 * combine.c (record_used_regs): New static function.
29817 (try_combine): Handle situations where there is an additional
29818 instruction between I2 and I3 which needs to have a LOG_LINK
29822 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29824 * combine.c (try_combine): Delete redundant i1 test. Call
29825 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29827 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29830 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29831 alternatives 13/14.
29833 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29835 PR tree-optimization/80054
29836 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29837 the optimization if a PHI or any of its arguments is not dominated
29838 by the candidate's basis. Use gphi* rather than gimple* as
29840 (replace_profitable_candidates): Clean up a gimple* variable that
29841 should be a gphi* variable.
29843 2017-03-20 Martin Sebor <msebor@redhat.com>
29846 * doc/extend.texi (attribute constructor): Document present limitation.
29848 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29851 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29852 __POWER9_VECTOR__ #ifdef control, change template definition to
29853 use Power9-specific built-in function.
29854 (vec_any_eq): Likewise.
29855 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29856 to control outcomes from this test.
29857 (vector_ae_<mode>p): For VEC_F modes, likewise.
29859 2017-03-20 Ian Lance Taylor <iant@google.com>
29861 * config/i386/i386.c (ix86_function_regparm): Save an extra
29862 register for -fsplit-stack with DECL_STATIC_CHAIN.
29864 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29867 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29868 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29870 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29872 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29874 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29877 2017-03-20 Marek Polacek <polacek@redhat.com>
29880 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29882 2017-03-20 Richard Biener <rguenther@suse.de>
29884 PR tree-optimization/80113
29885 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29886 allocate extra SSA name for PHI def.
29887 (add_close_phis_to_outer_loops): Likewise.
29888 (add_close_phis_to_merge_points): Likewise.
29889 (copy_loop_close_phi_args): Likewise.
29890 (copy_cond_phi_nodes): Likewise.
29892 2017-03-20 Martin Liska <mliska@suse.cz>
29894 PR middle-end/79753
29895 * tree-chkp.c (chkp_build_returned_bound): Do not build
29896 returned bounds for a LHS that's not a BOUNDED_P type.
29898 2017-03-20 Martin Liska <mliska@suse.cz>
29902 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29903 COMPLEX_CST and VECTOR_CST.
29905 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29908 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29909 target operand. A new splitter adds the clobber statement in case
29910 the target operand is dead anyway.
29912 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29914 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29915 to age-old versions of binutils and glibc.
29917 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29919 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29921 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29923 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29925 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29927 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29928 requirement for binutils 2.13.
29930 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29932 * combine.c (try_combine): Delete redundant i1 test. Call
29933 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29935 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29937 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29938 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29940 <riscv64-*-elf>: Re-arrange section
29941 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29942 <riscv32-*-linux>: Likewise.
29943 <riscv64-*-elf>: Likewise
29944 <riscv64-*-linux>: Likewise.
29946 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29949 * aarch64.opt(verbose-cost-dump): Fix typo.
29951 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29954 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29955 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29957 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29959 * reload.c (find_reloads): When reloading a nonoffsettable address,
29960 use RELOAD_OTHER for it and its address reloads.
29962 PR rtl-optimization/79910
29963 * combine.c (record_used_regs): New static function.
29964 (try_combine): Handle situations where there is an additional
29965 instruction between I2 and I3 which needs to have a LOG_LINK
29968 2017-03-17 Jeff Law <law@redhat.com>
29970 PR tree-optimization/71437
29971 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29972 conditional in the hash table first.
29973 (vrp_dom_walker::before_dom_children): Extract condition from
29974 ASSERT_EXPR. Record condition, its inverion and any implied
29975 conditions as well.
29977 2017-03-17 Marek Polacek <polacek@redhat.com>
29978 Markus Trippelsdorf <markus@trippelsdorf.de>
29980 PR tree-optimization/80079
29981 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29984 2017-03-17 Richard Biener <rguenther@suse.de>
29986 PR middle-end/80075
29987 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29988 Properly verify the LHS before the RHS possibly claims to be
29990 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29993 2017-03-17 Martin Jambor <mjambor@suse.cz>
29995 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29996 (List of -O2 options): Likewise.
29997 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
30000 2017-03-17 Tom de Vries <tom@codesourcery.com>
30002 * gcov-dump.c (print_usage): Print bug_report_url.
30004 2017-03-17 Richard Biener <rguenther@suse.de>
30006 PR middle-end/80050
30007 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
30008 (parser::peek): Likewise.
30010 2017-03-17 Richard Biener <rguenther@suse.de>
30012 PR tree-optimization/80048
30013 * sese.c (free_sese_info): Properly release rename_map and
30014 copied_bb_map elements.
30016 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
30018 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
30019 Add linked-list forward and backlinks. Insert on
30020 construction, remove on destruction.
30021 (class pass_store_merging): Add m_stores_head field.
30022 (pass_store_merging::terminate_and_process_all_chains):
30023 Iterate over m_stores_head list.
30024 (pass_store_merging::terminate_all_aliasing_chains):
30026 (pass_store_merging::execute): Check for debug stmts first.
30027 Push new chains onto the m_stores_head stack.
30029 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
30032 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
30033 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
30034 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
30036 2017-03-16 Jeff Law <law@redhat.com>
30038 PR tree-optimization/71437
30039 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
30040 member function. Implementation moved into after_dom_children
30041 member function and into the threader's thread_outgoing_edges
30043 (dom_opt_dom_walker::after_dom_children): Simplify by moving
30044 some code into new thread_outgoing_edges.
30045 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
30046 definition. Simplify marker handling (do it here). Assume we always
30047 have the available expression and the const/copies tables.
30048 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
30050 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
30051 * tree-vrp.c (equiv_stack): No longer file scoped.
30052 (vrp_dom_walker): New class.
30053 (vrp_dom_walker::before_dom_children): New member function.
30054 (vrp_dom_walker::after_dom_children): Likewise.
30055 (identify_jump_threads): Setup domwalker. Use it rather than
30056 walking edges in a random order by hand. Simplify setup/finalization.
30057 (finalize_jump_threads): Remove.
30058 (vrp_finalize): Do not call identify_jump_threads here.
30059 (execute_vrp): Do it here instead and call thread_through_all_blocks
30062 PR tree-optimization/71437
30063 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
30065 (simplify_stmt_for_jump_threading): Add basic_block argument. All
30067 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
30068 (dom_opt_dom_walker::thread_across_edge): Remove
30069 handle_dominating_asserts argument. All callers changed.
30070 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
30071 changes. Remove calls to lhs_of_dominating_assert. Other
30072 uses of handle_dominating_asserts turn into unconditional code
30073 (simplify_control_stmt_condition_1): Likewise.
30074 (simplify_control_stmt_condition): Likewise.
30075 (thread_through_normal_block, thread_across_edge): Likewise.
30076 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
30077 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
30078 object if it is not an SSA_NAME.
30079 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
30080 before calling into the VRP specific simplifiers.
30081 (identify_jump_threads): Remove handle_dominating_asserts
30084 2017-03-16 Jakub Jelinek <jakub@redhat.com>
30087 * tree-diagnostic.c (default_tree_printer): No longer static.
30088 * tree-diagnostic.h (default_tree_printer): New prototype.
30090 2017-03-16 Tamar Christina <tamar.christina@arm.com>
30092 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
30093 Change ins into fmov.
30095 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30097 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
30098 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
30099 Use h_con constraint for operand 1.
30100 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
30101 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
30103 2017-03-15 Jeff Law <law@redhat.com>
30105 PR tree-optimization/71437
30106 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
30107 (record_temporary_equivalences): Use it.
30109 PR tree-optimization/71437
30110 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
30111 tree-ssa-scopedtables.
30112 (lookup_avail_expr, build_and_record_new_cond): Likewise.
30113 (record_conditions, record_cond, vuse_eq): Likewise.
30114 (record_edge_info): Adjust to API tweak of record_conditions.
30115 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
30116 (record_temporary_equivalences, optimize_stmt): Likewise.
30117 (eliminate_redundant_computations): Likewise.
30118 (record_equivalences_from_stmt): Likewise.
30119 * tree-ssa-scopedtables.c: Include options.h and params.h.
30120 (vuse_eq): New function, moved from tree-ssa-dom.c
30121 (build_and_record_new_cond): Likewise.
30122 (record_conditions): Likewise. Accept vector of conditions rather
30123 than edge_equivalence structure for first argument.
30124 for the first argument.
30125 (avail_exprs_stack::lookup_avail_expr): New member function, moved
30126 from tree-ssa-dom.c.
30127 (avail_exprs_stack::record_cond): Likewise.
30128 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
30129 from tree-ssa-dom.c.
30130 (avail_exprs_stack): Add new member functions lookup_avail_expr
30132 (record_conditions): Declare.
30134 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
30137 * lra-constraints.c (process_alt_operands): Increase reject for
30138 reloading an input/output operand.
30140 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
30143 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
30144 insns to convert from signed/unsigned char/short to IEEE 128-bit
30146 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
30148 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
30151 * config/i386/i386.c (ix86_vector_duplicate_value): Create
30152 subreg of inner mode for values already in registers.
30154 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
30156 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
30157 iteration reg is used after the loop.
30159 2017-03-14 Martin Sebor <msebor@redhat.com>
30161 PR tree-optimization/79800
30162 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
30163 precision in negative-positive range.
30164 (format_floating): Call non-const overload with adjusted precision.
30166 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30169 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
30172 2017-03-14 Martin Sebor <msebor@redhat.com>
30174 PR middle-end/80020
30175 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
30176 * builtins.def (aligned_alloc): Use it.
30179 * Makefile.in (GTFILES): Add calls.c.
30180 * calls.c: Include "gt-calls.h".
30182 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
30184 PR rtl-optimization/79728
30185 * regs.h (struct target_regs): New field
30186 x_contains_allocatable_regs_of_mode.
30187 (contains_allocatable_regs_of_mode): New macro.
30188 * reginfo.c (init_reg_sets_1): Initialize it, and change
30189 contains_reg_of_mode so it includes global regs as well.
30190 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
30191 rather than contains_regs_of_mode.
30193 2017-03-14 Martin Liska <mliska@suse.cz>
30195 * doc/invoke.texi: Document options that can't be combined with
30196 -fcheck-pointer-bounds.
30198 2017-03-14 Martin Liska <mliska@suse.cz>
30200 PR middle-end/79831
30201 * doc/invoke.texi (-Wchkp): Document the option.
30203 2017-03-14 Martin Liska <mliska@suse.cz>
30205 * Makefile.in: Install gcov-dump.
30207 2017-03-14 Martin Liska <mliska@suse.cz>
30209 * multiple_target.c (expand_target_clones): Bail out for
30210 an invalid attribute.
30212 2017-03-14 Richard Biener <rguenther@suse.de>
30214 * alias.c (struct alias_set_entry): Pack properly.
30215 * cfgloop.h (struct loop): Likewise.
30216 * cse.c (struct set): Likewise.
30217 * ipa-utils.c (struct searchc_env): Likewise.
30218 * loop-invariant.c (struct invariant): Likewise.
30219 * lra-remat.c (struct cand): Likewise.
30220 * recog.c (struct change_t): Likewise.
30221 * rtl.h (struct address_info): Likewise.
30222 * symbol-summary.h (function_summary): Likewise.
30223 * tree-loop-distribution.c (struct partition): Likewise.
30224 * tree-object-size.c (struct object_size_info): Likewise.
30225 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
30226 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
30227 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
30228 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
30229 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
30230 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
30231 (struct _stmt_vec_info): Likewise.
30233 2017-03-14 Martin Liska <mliska@suse.cz>
30236 * multiple_target.c (create_dispatcher_calls): Check that
30237 a target can create a function dispatcher.
30239 2017-03-14 Martin Liska <mliska@suse.cz>
30242 * multiple_target.c (expand_target_clones): Drop local.local
30243 flag for default implementation.
30245 2017-03-14 Richard Biener <rguenther@suse.de>
30247 PR tree-optimization/80030
30248 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
30250 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
30252 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
30253 gcc_fallthrough() instead of __attribute__((fallthrough));
30255 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
30257 * doc/gcc.texi: Remove "up" link to (DIR).
30258 * doc/gccint.texi: Ditto.
30260 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
30262 * doc/install.texi (Specific) <avr>: Remove reference to
30265 2017-03-13 Jeff Law <law@redhat.com>
30267 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
30268 attribute rather than comments.
30270 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
30271 match_scratch operand is highest.
30273 2017-03-13 Martin Liska <mliska@suse.cz>
30275 PR middle-end/78339
30276 * ipa-pure-const.c (warn_function_noreturn): If the declarations
30277 is a CHKP clone, use original declaration.
30279 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30281 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
30282 (arc_conditional_register_usage): Use a different allocation order
30283 when optimizing for size.
30284 * common/config/arc/arc-common.c (arc_option_optimization_table):
30285 Section anchors default on when optimizing for size.
30287 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30289 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
30291 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30293 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
30294 * config/arc/arc.md (cpu_facility): Add cd variant.
30295 (*movqi_insn): Add code density variant.
30296 (*movhi_insn): Likewise.
30297 (*movqi_insn): Likewise.
30298 (*addsi3_mixed): Likewise.
30299 (subsi3_insn): Likewise.
30301 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30303 * config/arc/arc.md (movsi_cond_exec): Update constraint.
30305 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30307 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
30308 expressions with MINUS and UNARY ops.
30310 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30313 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
30315 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
30316 between vec_select and vector argument.
30317 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
30318 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
30319 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
30320 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
30321 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
30322 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
30324 2017-03-13 Richard Biener <rguenther@suse.de>
30327 * params.def (vect-max-peeling-for-alignment): Fix typo.
30329 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30331 * doc/install.texi (Specific) <mips-*-*>: Remove description of
30332 issue that only occurred with binutils below 2.18.
30334 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30336 * doc/install.texi (Specific) <cris-axis-elf>: No longer
30337 refer to binutils 2.11/2.12 minimum.
30339 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30341 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
30342 ftp.kernel.org and simplify binutils requirement.
30344 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
30346 * doc/invoke.texi (Warning Options): Fix spelling of link-time
30348 (Optimize Options): Ditto. Also remove redundancy.
30350 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30352 PR translation/79848
30353 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
30355 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
30356 to G_ to avoid double translation.
30358 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30360 PR translation/79923
30361 * auto-profile.c (get_combined_location): Convert leading
30362 character of diagnostics to lower case and remove trailing period.
30363 (read_profile): Likewise for various diagnostics.
30364 * config/arm/arm.c (arm_option_override): Remove trailing period
30365 from various diagnostics.
30366 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
30367 (msp430_expand_delay_cycles): Likewise.
30369 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30372 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
30373 full command-line argument, rather than just "str".
30374 (aarch64_validate_march): Likewise.
30375 (aarch64_validate_mtune): Likewise.
30377 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
30379 PR rtl-optimization/78911
30380 * lra-assigns.c (must_not_spill_p): New function.
30381 (spill_for): Use it.
30383 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30385 PR tree-optimization/79981
30386 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
30387 ATOMIC_COMPARE_EXCHANGE ifn result.
30388 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
30389 IFN_ATOMIC_COMPARE_EXCHANGE.
30391 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30394 * opts.c (parse_sanitizer_options): Add missing question mark to
30395 "did you mean" message.
30397 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30399 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
30401 (VMULEUH_UNS): Likewise.
30402 (VMULOUB_UNS): Likewise.
30403 (VMULOUH_UNS): Likewise.
30404 * config/rs6000/rs6000.c (builtin_function_type): Remove
30405 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
30407 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30410 * read-rtl-function.c (function_reader::read_rtx_operand): Update
30411 x with result of extra_parsing_for_operand_code_0.
30412 (function_reader::extra_parsing_for_operand_code_0): Convert
30413 return type from void to rtx, returning x. When reading
30414 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
30415 larger size containing struct block_symbol.
30417 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
30419 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
30420 -mfloat128-hardware without -m64.
30422 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
30425 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
30426 entries to the case statement that marks unsigned arguments to
30427 overloaded functions.
30429 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30431 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
30432 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
30434 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
30437 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
30438 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
30440 2017-03-10 Martin Liska <mliska@suse.cz>
30444 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
30445 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
30448 2017-03-10 Olivier Hainque <hainque@adacore.com>
30450 * tree-switch-conversion (array_value_type): Start by resetting
30451 candidate type to it's main variant.
30453 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30455 PR rtl-optimization/79909
30456 * combine.c (try_combine): Use simplify_replace_rtx on individual
30457 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
30458 of the whole CALL_INSN_FUNCTION_USAGE.
30460 PR tree-optimization/79972
30461 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
30462 get_range_info on SSA_NAMEs. Formatting fixes.
30464 2017-03-10 Richard Biener <rguenther@suse.de>
30465 Jakub Jelinek <jakub@redhat.com>
30467 PR tree-optimization/77975
30468 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
30469 edge to be constant.
30470 (get_val_for): For constant x return it. Formatting fix.
30471 (loop_niter_by_eval): Avoid pointless looping if the next iteration
30472 would use the same bases as the current one.
30474 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30476 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
30477 instead of vec_select for V1TImode.
30478 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
30480 (VSX_LE_128): Add V1TI to this mode iterator.
30481 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
30482 (*vsx_le_perm_store_<mode>): Likewise.
30483 (pre-reload splitter for VSX stores): Likewise.
30484 (post-reload splitter for VSX stores): Likewise.
30485 (*vsx_xxpermdi2_le_<mode>): Likewise.
30486 (*vsx_lxvd2x2_le_<mode>): Likewise.
30487 (*vsx_stxvd2x2_le_<mode>): Likewise.
30489 2017-03-09 Michael Eager <eager@eagercon.com>
30491 Correct failures with --enable-checking=yes,rtl.
30493 * config/microblaze/microblaze.c (microblaze_expand_shift):
30494 Replace GET_CODE test with CONST_INT_P and INTVAL test with
30495 test for const0_rtx.
30496 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
30497 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
30499 2017-03-09 Richard Biener <rguenther@suse.de>
30501 PR tree-optimization/79977
30502 * graphite-scop-detection.c (scop_detection::merge_sese):
30503 Handle the case of extra exits to blocks dominating the entry.
30505 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
30507 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
30510 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
30512 PR rtl-optimization/79949
30513 * lra-constraints.c (process_alt_operands): Check memory when
30514 trying to predict a cycle. Print about the overall increase.
30516 2017-03-09 Richard Biener <rguenther@suse.de>
30518 PR middle-end/79971
30519 * gimple-expr.c (useless_type_conversion_p): Preserve
30520 TYPE_SATURATING for fixed-point types.
30522 2017-03-09 Richard Biener <rguenther@suse.de>
30525 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
30526 alignment of BLKmode params.
30528 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30531 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
30532 (VALL_NO_V2Q): Likewise.
30534 * config/aarch64/aarch64-simd.md
30535 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
30537 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
30538 VALL_NO_V2Q mode iterator.
30539 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
30541 2017-03-09 Martin Liska <mliska@suse.cz>
30543 PR tree-optimization/79631
30544 * tree-chkp-opt.c (chkp_is_constant_addr): Call
30545 tree_int_cst_sign_bit just for INTEGER constants.
30547 2017-03-09 Martin Liska <mliska@suse.cz>
30551 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
30554 2017-03-09 Marek Polacek <polacek@redhat.com>
30557 * tree.c (inchash::add_expr): Handle TREE_VEC.
30559 2017-03-09 Martin Liska <mliska@suse.cz>
30562 (chkp_narrow_size_and_offset): New function.
30563 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
30564 (void chkp_parse_bit_field_ref): New function.
30565 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
30566 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
30568 2017-03-09 Martin Liska <mliska@suse.cz>
30571 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
30572 (chkp_find_bounds_1): Remove gcc_unreachable.
30574 2017-03-09 Jakub Jelinek <jakub@redhat.com>
30577 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
30578 BUILT_IN_SYNC*, determine the access type from the size suffix and
30579 always build a MEM_REF with that type. Handle forgotten
30580 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
30583 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
30584 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
30585 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
30586 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
30587 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
30588 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
30589 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
30590 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
30591 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
30592 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
30593 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
30594 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
30595 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
30596 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
30597 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
30598 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
30599 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
30600 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
30601 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
30602 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
30603 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
30604 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
30605 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
30606 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
30607 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
30608 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
30609 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
30610 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
30611 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
30612 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
30613 definitions outside of __OPTIMIZE__ guarded section.
30616 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
30617 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
30618 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
30619 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
30622 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30624 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
30625 ("vfenez<mode>"): Add missing constraints.
30627 2017-03-08 Martin Sebor <msebor@redhat.com>
30630 * config/nds32/nds32.c (nds32_option_override):
30631 Fix misspelled diagnostic.
30633 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30636 * gimplify.c (gimplify_omp_for): Replace index var in outer
30637 taskloop statement with an artificial variable and add
30638 OMP_CLAUSE_PRIVATE clause for it.
30640 2017-03-08 Richard Biener <rguenther@suse.de>
30642 PR tree-optimization/79955
30643 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
30644 for accesses that are completely outside of the variable.
30646 2017-03-08 Andrew Haley <aph@redhat.com>
30648 PR tree-optimization/79943
30649 * tree-ssa-loop-split.c (compute_new_first_bound): When
30650 calculating the new upper bound, (END-BEG) should be added, not
30653 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30655 * config/avr/avr.md (setmemhi): Make sure match_dup
30656 operand number comes before match_scratch.
30658 2017-03-08 Richard Biener <rguenther@suse.de>
30660 PR tree-optimization/79920
30661 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30662 with ncopies == 1 to ...
30663 (vect_transform_slp_perm_load): ... here. Properly compute
30664 all element loads by iterating VF times over the group. Do
30665 not handle ncopies (computed in a broken way) in
30666 vect_create_mask_and_perm.
30668 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30671 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30672 is a uniform vector, use uniform_vector_p return value instead of
30673 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30675 2017-03-07 Marek Polacek <polacek@redhat.com>
30677 PR middle-end/79809
30678 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30679 (alloca_call_type): Likewise.
30681 2017-03-07 Martin Liska <mliska@suse.cz>
30683 * gcov.c (process_args): Put comment to correct location.
30685 2017-03-07 Martin Liska <mliska@suse.cz>
30687 PR middle-end/68270
30688 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30689 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30690 (chkp_narrow_bounds_for_field): Likewise.
30691 (chkp_parse_array_and_component_ref): Pass one more argument to
30694 2017-03-07 Richard Biener <rguenther@suse.de>
30696 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30699 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30701 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30702 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30704 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30707 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30708 to end of description.
30709 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30711 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30713 PR rtl-optimization/79901
30714 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30716 (*avx512f_<code><mode>3<mask_name>): ... this.
30717 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30718 iterator instead of VI8_AVX2_AVX512BW.
30720 PR rtl-optimization/79901
30721 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30722 min/max expander, expand it using expand_vec_cond_expr.
30725 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30728 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30731 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30732 to void * for PCH reasons.
30733 * dwarf2out.c (output_loc_operands, output_die): Cast
30734 v.val_vec.array to unsigned char *.
30736 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30739 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30742 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30744 PR rtl-optimization/79571
30745 * lra-constraints.c (process_alt_operands): Calculate static
30746 reject and subtract it from overall when only addresses will be
30749 2017-03-06 Julia Koval <julia.koval@intel.com>
30752 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30753 incoming stack boundary to 128 for 64-bit targets.
30755 2017-03-06 Richard Biener <rguenther@suse.de>
30757 PR tree-optimization/79894
30758 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30759 to NULL after folding it.
30761 2017-03-06 Richard Biener <rguenther@suse.de>
30763 PR tree-optimization/79824
30764 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30765 check disabling peeling for gaps.
30767 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30769 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30770 attributes): Document gettimeofday.
30772 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30774 * config/s390/s390.c (s390_option_override_internal): Set
30775 PARAM_MIN_VECT_LOOP_BOUND
30777 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30779 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30780 * config/s390/s390.md: Likewise.
30782 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30785 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30786 (<avx2_avx512>_perm<mode>): Rename to ...
30787 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30789 (<avx512>_perm<mode>_mask): Rename to ...
30790 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30792 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30793 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30794 instead of VI8F_256_512.
30795 (avx512f_perm<mode>): New define_expand.
30796 (avx512f_perm<mode>_mask): Likewise.
30797 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30798 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30800 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30802 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30803 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30805 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30807 2017-03-06 Martin Liska <mliska@suse.cz>
30810 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30811 when having a SSA NAME w/o VAR_DECL assigned to it.
30813 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30815 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30816 msa_dpsub_<su>_d): Fix MODE for vec_select.
30818 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30820 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30822 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30824 2017-03-06 Richard Biener <rguenther@suse.de>
30826 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30827 * plugin.c (register_plugin_info): Likewise.
30828 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30830 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30832 * config/i386/sse.md (sse_storehps, sse_storelps,
30833 avx_<castmode><avxsizesuffix>_<castmode>,
30834 avx512f_<castmode><avxsizesuffix>_<castmode>,
30835 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30836 in condition that at least one operand is not a MEM.
30838 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30840 PR middle-end/79805
30841 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30842 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30844 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30845 gimple_call_nothrow_p flag based on whether original builtin can throw.
30846 If it can, emit following stmts on the fallthrough edge.
30847 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30848 don't create new bb if inserting just debug stmts on the edge, try to
30849 insert them on the fallthru bb or just reset debug stmts.
30851 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30854 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30855 restore recog_data (including the operand rtxes inside it) around
30856 the call to get_insn_template.
30858 2017-03-03 Martin Sebor <msebor@redhat.com>
30860 PR tree-optimization/79699
30861 * context.c (context::~context): Free MPFR caches to avoid
30862 a memory leak on program exit.
30864 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30866 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30867 Use wide_int::ulow () instead of .elt (0).
30869 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30871 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30872 (*pushxf): Limit oF constraint to 32bit targets and add oC
30873 constraint for 64bit targets.
30874 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30875 (*pushdf): Change rmF constraint to rmC.
30877 2017-03-03 Martin Liska <mliska@suse.cz>
30879 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30880 Remove unused variable.
30882 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30885 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30886 is a memory operand, increase num_memory.
30887 (ix86_expand_args_builtin): Likewise.
30889 2017-03-03 Jan Hubicka <jh@suse.cz>
30892 * ipa-devirt.c (maybe_record_node): Properly handle
30893 __cxa_pure_virtual visibility.
30895 2017-03-03 Martin Liska <mliska@suse.cz>
30897 PR tree-optimization/79803
30898 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30900 (pass_loop_prefetch::execute): Disabled optimization if an
30901 assumption about L1 cache size is not met.
30903 2017-03-03 Martin Liska <mliska@suse.cz>
30905 PR rtl-optimization/79574
30906 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30907 (hash_scan_set): Likewise.
30908 (dump_hash_table): Likewise.
30909 (hoist_code): Likewise.
30911 2017-03-03 Richard Biener <rguenther@suse.de>
30913 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30914 in place of uhigh ().
30915 (fixed_convert_from_real): Likewise.
30917 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30920 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30922 2017-03-03 Richard Biener <rguenther@suse.de>
30924 PR middle-end/79818
30925 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30926 TYPE_OVERFLOW_UNDEFINED check.
30928 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30930 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30932 (vector_ae_<mode>_p): Likewise.
30933 (vector_nez_<mode>_p): Likewise.
30934 (vector_ne_v2di_p): Likewise.
30935 (vector_ae_v2di_p): Likewise.
30936 (vector_ne_<mode>_p): Likewise.
30937 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30939 (vsx_tsqrt<mode>2_fe): Likewise.
30941 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30944 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30946 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30948 PR rtl-optimization/79780
30949 * cprop.c (one_cprop_pass): When second and further conditional trap
30950 in a single basic block is turned into an unconditional trap, turn it
30951 into a deleted note to avoid RTL verification failures.
30953 2017-03-02 Richard Biener <rguenther@suse.de>
30955 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30957 2017-03-02 Richard Biener <rguenther@suse.de>
30959 PR tree-optimization/79345
30961 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30962 param and abort the walk, returning -1 if it is hit.
30963 (walk_aliased_vdefs): Take a limit param and pass it on.
30964 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30965 defaulting to 0 and return a signed int.
30966 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30967 (check_defs): New helper.
30968 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30969 about uninitialized memory.
30970 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30971 bogus uninitialized warning.
30972 (fixed_convert_from_real): Likewise.
30974 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30976 PR tree-optimization/66768
30977 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30978 iv_use if base object can't be determined.
30980 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30982 PR tree-optimization/79345
30983 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30984 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30985 (get_pattern_stats): Initialize it.
30986 * genemit.c (gen_expand): Verify match_scratch numbers come after
30987 match_operand/match_dup numbers.
30988 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30989 match_scratch numbers.
30990 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30992 * config/s390/s390.md (trunctdsd2): Likewise.
30994 2017-03-02 Richard Biener <rguenther@suse.de>
30996 * wide-int.h (wide_int_storage::operator=): Implement in terms
30999 2017-03-02 Richard Biener <rguenther@suse.de>
31001 PR tree-optimization/79777
31002 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
31003 the to insert expression to sth existing.
31005 2017-03-01 Martin Sebor <msebor@redhat.com>
31007 PR middle-end/79692
31008 * gimple-ssa-sprintf.c
31009 (directive::known_width_and_precision): New function.
31010 (format_integer): Use it.
31011 (get_mpfr_format_length): Consider the full range of precision
31012 when computing %g output with the # flag. Set the likely byte
31013 count to 3 rather than 1 when precision is indeterminate.
31014 (format_floating): Correct the lower bound of precision.
31016 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31018 * doc/invoke.texi: Document default code model for 64-bit Linux.
31020 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31023 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
31024 udiv rather than div since input pattern is unsigned.
31026 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
31028 * config/i386/i386.c (print_reg): Warn for values of
31029 unsupported size in integer register.
31031 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
31034 * config/rs6000/predicates.md (current_file_function_operand): Do
31035 not allow self calls to be local if the function is replaceable.
31037 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
31040 * config/rs6000/altivec.h (vec_ctz and others): Change the
31041 preprocessor macro that controls conditional compilation from
31042 _ARCH_PWR9 to __POWER9_VECTOR__.
31043 (vec_all_ne): Change parameterization of __altivec_scalar_pred
31044 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
31045 control (instead of _ARCH_PWR9 control) so that template
31046 definition uses power9-specific function.
31047 (vec_any_eq): Likewise.
31048 (vec_all_ne): Change macro definition to use a power9-specific
31049 expansion under #ifdef __POWER9_VECTOR__ control (instead of
31050 _ARCH_PWR9 control).
31051 (vec_any_eq) Likewise.
31052 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
31053 expansion for CMPNEF to remove support for xvcmpnesp instruction.
31054 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
31055 support for xvcmpnedp instruction.
31056 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
31057 macro expansion so that Power9 implementation of vec_all_ne does
31058 not use the AltiVec predicate framework.
31059 (VCMPNEH_P): Likewise.
31060 (VCMPNEW_P): Likewise.
31061 (VCMPNED_P): Likewise.
31062 (VCMPNEFP_P): Likewise.
31063 (VCMPNEDP_P): Likewise.
31064 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
31065 implementation of vec_any_eq to not use AltiVec predicate
31067 (VCMPAEH_P): Likewise.
31068 (VCMPAEW_P): Likewise.
31069 (VCMPAED_P): Likewise.
31070 (VCMPAEFP_P): Likewise.
31071 (VCMPAEDP_P): Likewise.
31072 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
31073 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
31074 not use the AltiVec predicate framework.
31075 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
31076 of vec_any_eq to not use AltiVec predicate framework.
31077 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
31078 support for predefined __POWER9_VECTOR__ macro to indicate that
31079 Power9 instruction selection is enabled.
31080 (altivec_overloaded_builtins): Remove extraneous
31081 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
31082 function argument types RS6000_BTI_bool_V16QI and
31083 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
31084 entry for overloaded function argument types RS6000_BTI_bool_V4SI
31085 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
31086 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
31087 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
31088 Power9 for implementations of vec_cmpne. Change the signature for
31089 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
31090 (representing vec_all_ne) to remove the previously described first
31091 argument of type RS6000_BTI_INTSI, as this was an artifact of
31092 reliance on the AltiVec predicate framework, which is no longer
31093 used in the implementation of these functions. Add
31094 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
31095 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
31096 since, unlike the AltiVec predicate framework implementation, we
31097 do not share function descriptors between vec_alle and vec_anyeq.
31098 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
31099 set of modes that receive special treatment even when
31100 TARGET_P9_VECTOR is true. The special treatment emits code that
31101 does not depend on Power9 instructions.
31102 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
31103 define_expand to not rely on AltiVec predicate framework.
31104 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
31106 (vector_ne_v2di_p): Change this define_expand to not rely on
31107 AltiVec predicate framework.
31108 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
31110 (vector_ne_<mode>_p): Change this define_expand to not rely on
31111 AltiVec predicate framework.
31112 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
31114 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
31115 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
31116 define_insn pattern.
31117 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
31118 define_insn pattern because the xvcmpne<VSs>. instruction is not
31120 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
31121 instruction is not supported.
31123 2017-03-01 Jakub Jelinek <jakub@redhat.com>
31125 * config/nvptx/nvptx.c: Include intl.h.
31127 2017-03-01 Martin Jambor <mjambor@suse.cz>
31130 * ipa-prop.h (ipa_bits): Removed field known.
31131 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
31132 to pointers. Adjusted their comments to warn about their sharing.
31133 (ipcp_transformation_summary): Change bits to a vector of pointers.
31134 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
31135 (ipa_get_ipa_bits_for_value): Declare.
31136 * tree-vrp.h (value_range): Mark as GTY((for_user)).
31137 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
31138 (ipa_bits_hash_table): Likewise.
31139 (ipa_vr_ggc_hash_traits): Likewise.
31140 (ipa_vr_hash_table): Likewise.
31141 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
31142 being pointers and vr_known being removed.
31143 (ipa_set_jf_unknown): Likewise.
31144 (ipa_get_ipa_bits_for_value): New function.
31145 (ipa_set_jfunc_bits): Likewise.
31146 (ipa_get_value_range): New overloaded functions.
31147 (ipa_set_jfunc_vr): Likewise.
31148 (ipa_compute_jump_functions_for_edge): Use the above functions to
31149 construct bits and vr parts of jump functions.
31150 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
31151 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
31153 (ipcp_grow_transformations_if_necessary): Also allocate
31154 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
31156 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
31157 them. Fix too long lines.
31158 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
31159 vr_known being removed.
31160 (ipa_read_jump_function): Use new setter functions to construct bits
31161 and vr parts of jump functions or set them to NULL.
31162 (write_ipcp_transformation_info): Adjust for bits being pointers.
31163 (read_ipcp_transformation_info): Likewise.
31164 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
31166 Include gt-ipa-prop.h.
31167 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
31169 (ipcp_store_bits_results): Likewise.
31170 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
31171 Do not write to existing jump functions but use a temporary instead.
31173 2017-03-01 Jakub Jelinek <jakub@redhat.com>
31176 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
31177 attempt to use its first operand as BIT_FIELD_REF base.
31179 2017-03-01 Richard Biener <rguenther@suse.de>
31181 PR middle-end/79721
31182 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
31183 interpolating formula in wrapping arithmetic.
31184 (chrec_apply): Convert chrec_evaluate return value to wanted type.
31186 2017-03-01 Jakub Jelinek <jakub@redhat.com>
31188 PR tree-optimization/79734
31189 * tree-vect-generic.c (expand_vector_condition): Optimize
31190 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
31191 Handle VEC_COND_EXPR where comparison has different inner width from
31192 type's inner width.
31194 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
31196 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
31197 markup, and similar issues. Remove @opindex entries for things
31198 that aren't options. Add missing -mmpy-option entries.
31200 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31202 PR tree-optimization/79737
31203 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
31204 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
31205 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
31206 instead of byte_size. Formatting fix.
31207 (shift_bytes_in_array_right): Formatting fix.
31209 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
31212 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
31213 condition on optimize for the leaf function test.
31215 2017-02-28 Martin Liska <mliska@suse.cz>
31218 * read-rtl-function.c (function_reader::handle_unknown_directive):
31219 Bail out when one uses -flto.
31221 2017-02-28 Martin Liska <mliska@suse.cz>
31223 * common.opt: Replace space with tabular for options of <number>
31225 * config/i386/i386.opt: Show <number> value for
31226 -mlarge-data-threshold.
31227 * opts.c (print_filtered_help): Do not display number in hexadecimal
31230 2017-02-28 Martin Liska <mliska@suse.cz>
31232 * common.opt: Fix --help=option -Q for options which are of
31235 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
31237 * config/i386/i386.c (print_reg): Error out for values
31238 of 8-bit size in invalid integer register.
31240 2017-02-28 Martin Sebor <msebor@redhat.com>
31242 PR tree-optimization/79691
31243 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
31245 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31248 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
31249 gcc_unreachable with output_operand_lossage.
31251 2017-02-28 Richard Biener <rguenther@suse.de>
31253 PR tree-optimization/79740
31254 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
31256 (visit_nary_op): Insert the nary into the hashtable if we
31257 pattern-matched sth.
31258 * tree-ssa-pre.c (eliminate_insert): Robustify.
31260 2017-02-28 Richard Biener <rguenther@suse.de>
31262 PR middle-end/79731
31263 * fold-const.c (decode_field_reference): Reject out-of-bound
31266 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31268 * config/i386/i386.c: Include intl.h.
31269 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
31270 instead of just cond ? "..." : "...".
31271 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
31272 * coverage.c (read_counts_file): Likewise.
31273 * omp-offload.c: Include intl.h.
31274 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
31275 of just cond ? "..." : "...".
31276 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
31277 of just cond ? "..." : "...".
31279 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
31282 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
31284 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
31285 'tune for' CPU name.
31286 * config/arm/arm-cpu-data.h: Regenerated.
31288 2017-02-28 Richard Biener <rguenther@suse.de>
31290 PR tree-optimization/79732
31291 * tree-inline.c (expand_call_inline): Do not shadow var.
31293 2017-02-28 Richard Biener <rguenther@suse.de>
31295 PR tree-optimization/79723
31296 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
31297 address-space properly.
31299 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
31301 * doc/optinfo.texi (Optimization groups): Fix option used for
31303 * doc/invoke.texi (-fopt-info): Document "omp".
31304 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
31305 (OPTGROUP_ALL): Add OPTGROUP_OMP.
31306 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
31307 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
31308 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
31310 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
31312 * dumpfile.c (optgroup_options): Instead of "openmp", associate
31313 OPTGROUP_OMP with "omp".
31315 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
31318 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
31319 for arithmetic shift of unsigned V2DI.
31321 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
31323 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
31324 arc/linux.h headers.
31325 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
31326 (LINK_SPEC): Likewise.
31327 (ARC_TLS_EXTRA_START_SPEC): Likewise.
31328 (EXTRA_SPECS): Likewise.
31329 (STARTFILE_SPEC): Likewise.
31330 (ENDFILE_SPEC): Likewise.
31331 (LIB_SPEC): Likewise.
31332 (TARGET_SDATA_DEFAULT): Likewise.
31333 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
31334 (MULTILIB_DEFAULTS): Likewise.
31335 (DWARF2_UNWIND_INFO): Likewise.
31336 * config/arc/big.h: New file.
31337 * config/arc/elf.h: Likewise.
31338 * config/arc/linux.h: Likewise.
31339 * config/arc/t-uClibc: Remove.
31341 2017-02-27 Bin Cheng <bin.cheng@arm.com>
31343 PR tree-optimization/77536
31344 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
31345 (tree_transform_and_unroll_loop): Use above function to compute the
31346 estimated niter of unrolled loop and use it when scaling profile.
31347 Also use count info rather than frequency if it's non-zero.
31348 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
31349 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
31350 (vect_transform_loop): Call above function.
31352 2017-02-27 Richard Biener <rguenther@suse.de>
31354 PR tree-optimization/45397
31355 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
31356 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
31357 (visit_nary_op): Add pattern matching for CSEing sign-changed
31358 or truncated operations with wider ones.
31360 2017-02-27 Richard Biener <rguenther@suse.de>
31362 PR tree-optimization/79690
31363 * tree-vect-stmts.c (vectorizable_store): Use vector type
31364 built from the DR with address-space.
31366 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
31368 * doc/invoke.texi (Optimize Options): Refine the description
31369 of asan-use-after-return.
31371 2017-02-25 Alan Modra <amodra@gmail.com>
31373 PR rtl-optimization/79584
31374 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
31375 base, not ad->base_term, the reg within base. Remove assertion
31376 that ad->base == ad->base_term. Replace gen_int_mode using
31377 bogus mode with const0_rtx.
31379 2017-02-25 Jakub Jelinek <jakub@redhat.com>
31381 PR middle-end/79396
31382 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
31383 FMA_EXPR like tcc_binary or tcc_unary.
31385 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
31388 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
31390 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
31391 (output_loc_operands): Handle DW_OP_call_ref and
31392 DW_OP_GNU_variable_value.
31393 (struct variable_value_struct): New type.
31394 (struct variable_value_hasher): Likewise.
31395 (variable_value_hash): New variable.
31396 (string_types): Remove.
31397 (copy_loc_descr): New function.
31398 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
31399 (prepend_loc_descr_to_each): New function.
31400 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
31401 instead of add_loc_descr_to_each if the first argument is single
31402 location list and the second has multiple.
31403 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
31404 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
31405 when looking for variable value which doesn't have other location info.
31406 (loc_list_from_tree): Formatting fix.
31407 (gen_array_type_die): Simplify DW_AT_string_length handling.
31408 (adjust_string_types): Remove.
31409 (gen_subprogram_die): Don't call adjust_string_types nor test/set
31410 string_types. Call resolve_variable_values.
31411 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
31412 (resolve_addr_in_expr): Likewise. Add A argument.
31413 (copy_deref_exprloc): Remove deref argument. Adjust for the
31414 original expression being DW_OP_GNU_variable_value with optionally
31415 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
31416 optionally after it.
31417 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
31418 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
31419 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
31420 (variable_value_hasher::hash, variable_value_hasher::equal): New
31422 (resolve_variable_value_in_expr, resolve_variable_value,
31423 resolve_variable_values, note_variable_value_in_expr,
31424 note_variable_value): New functions.
31425 (dwarf2out_early_finish): Call note_variable_value on all toplevel
31428 2017-02-24 Jakub Jelinek <jakub@redhat.com>
31431 * opts.h (handle_generated_option): Add GENERATED_P argument.
31432 * opts-common.c (handle_option): Adjust function comment.
31433 (handle_generated_option): Add GENERATED_P argument, pass it to
31435 (control_warning_option): Pass false to handle_generated_option
31437 * opts.c (maybe_default_option): Pass true to handle_generated_option
31439 * optc-gen.awk: Likewise.
31441 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31443 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
31444 a REG, look at the REG it is a SUBREG of.
31445 (splitter for cmpeqsi_t): Ditto.
31447 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31449 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
31450 the special USEs with the pattern of the insn, not the insn itself.
31452 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
31455 * doc/invoke.texi: Document -mload-store-pairs.
31457 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31458 Sandra Loosemore <sandra@codesourcery.com>
31460 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
31461 argument isn't a CONST_INT.
31462 (nios2_alternate_compare_const): Assert op is a CONST_INT.
31463 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
31464 (nios2_validate_compare): Bypass alternate compare logic if *op2
31465 is not a CONST_INT.
31466 (ldstwm_operation_p): Return false if first_base is not a REG or
31467 if first_offset is not a CONST_INT.
31469 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31471 * config/cris/cris.md: Use correct operand in a define_peephole2.
31473 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31475 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
31477 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31479 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
31480 this_insn if it is an INSN or JUMP_INSN.
31481 (force_offsettable): Look at base, not at addr.
31482 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
31483 on things that aren't necessarily CONST_INTs.
31485 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
31487 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
31488 -mfpmath=sse is the default also for x86-32 targets with SSE2
31489 instruction set when @option{-ffast-math} is enabled
31491 2017-02-24 Jeff Law <law@redhat.com>
31493 PR rtl-optimizatoin/79286
31494 * ira.c (update_equiv_regs): Drop may_trap_p exception to
31497 2017-02-24 Richard Biener <rguenther@suse.de>
31499 PR tree-optimization/79389
31500 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
31503 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
31505 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
31506 function comment to reflect reality.
31507 (loop_exits_before_overflow): Fix typo in function description.
31509 2017-02-24 Richard Biener <rguenther@suse.de>
31511 PR tree-optimization/79389
31512 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
31513 properly that a threading opportunity exists. Detect conditional
31514 copy/constant propagation opportunities.
31516 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
31518 * config/visium/visium.md (type): Add trap.
31519 (b): New mode attribute.
31520 (*btst): Rename into...
31521 (*btst<mode>): ...this and adjust.
31522 (*cbranchsi4_btst_insn): Rename into...
31523 (*cbranch<mode>4_btst_insn): ...this and adjust.
31524 (trap): New define_insn.
31526 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31528 PR tree-optimization/79389
31529 * ifcvt.c (struct noce_if_info): Add rev_cond field.
31530 (noce_reversed_cond_code): New function.
31531 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
31532 reversed_comparison_code. Formatting fix.
31533 (noce_try_store_flag): Test rev_cond != NULL in addition to
31534 reversed_comparison_code.
31535 (noce_try_store_flag_constants): Likewise.
31536 (noce_try_store_flag_mask): Likewise.
31537 (noce_try_addcc): Use rev_cond if non-NULL instead of
31538 reversed_comparison_code.
31539 (noce_try_cmove_arith): Likewise. Formatting fixes.
31540 (noce_try_minmax, noce_try_abs): Clear rev_cond.
31541 (noce_find_if_block): Initialize rev_cond.
31542 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
31543 instead of false as last argument never attempt to reverse it
31546 2017-02-23 Bin Cheng <bin.cheng@arm.com>
31548 PR tree-optimization/79663
31549 * tree-predcom.c (combine_chains): Process refs in reverse order
31550 only for ZERO length chains, and add explaining comment.
31552 2017-02-23 Jeff Law <law@redhat.com>
31554 PR tree-optimization/79578
31555 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
31556 in call to operand_equal_p.
31558 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
31561 * config/i386/cpuid.h: Fix another undefined behavior.
31563 2017-02-23 Richard Biener <rguenther@suse.de>
31565 PR tree-optimization/79683
31566 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
31567 vector types for data-refs.
31569 2017-02-23 Martin Liska <mliska@suse.cz>
31571 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
31573 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31575 PR middle-end/79665
31576 * internal-fn.c (get_range_pos_neg): Moved to ...
31577 * tree.c (get_range_pos_neg): ... here. No longer static.
31578 * tree.h (get_range_pos_neg): New prototype.
31579 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
31580 are known to be in between 0 and signed maximum inclusive, try to
31581 expand both unsigned and signed divmod and use the cheaper one from
31584 2017-02-22 Jeff Law <law@redhat.com>
31586 PR tree-optimization/79578
31587 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
31588 to compare base operands.
31590 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
31593 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
31594 gpc_reg_operand instead of fpr_reg_operand.
31596 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
31598 * config/mips/mips.c (mips_return_in_memory): Force FP
31599 vector types to be returned in memory for o32 ABI.
31601 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31603 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
31604 instead of DW_TAG_member for static data member declarations and don't
31605 set no_linkage_name for static inline data members.
31606 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
31609 2017-02-22 Martin Liska <mliska@suse.cz>
31611 * doc/invoke.texi: Replace inequality signs with square brackets
31614 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
31617 * lra-constraints.c (simplify_operand_subreg): Handle
31618 WORD_REGISTER_OPERATIONS targets.
31620 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31623 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
31624 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
31625 elimination by swapping fld*.
31627 2017-02-22 Richard Biener <rguenther@suse.de>
31629 PR tree-optimization/79673
31630 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
31631 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
31632 irrelevant address-space qualifiers and avoiding a
31633 ADDR_SPACE_CONVERT_EXPR from fold_convert.
31635 2017-02-22 Richard Biener <rguenther@suse.de>
31637 PR tree-optimization/79666
31638 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
31639 to not symbolically negate if that may introduce undefined
31642 2017-02-22 Martin Liska <mliska@suse.cz>
31645 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
31646 * data-streamer-out.c (streamer_write_gcov_count_stream):
31648 * value-prof.c (stream_out_histogram_value): Make assert more
31649 precise based on type of counter.
31651 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31654 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31655 Use nonimmediate_operand instead of memory_operand for operand 1.
31656 (float-extend standard_x87sse_constant_load splitter): Ditto.
31658 2017-02-21 Jeff Law <law@redhat.com>
31660 PR tree-optimization/79621
31661 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31662 blocks with edges to themselves.
31664 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31667 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31668 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31669 Use gimple_call_builtin_p.
31672 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31673 on temporarily removed DEBUG_INSNs.
31675 PR tree-optimization/79649
31676 * tree-loop-distribution.c (classify_partition): Give up on
31677 non-generic address space loads/stores.
31679 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31681 * doc/loop.texi (Loop manipulation): Remove nonexistent
31682 tree_ssa_loop_version from the documentation.
31683 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31685 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31688 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31689 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31690 * config/rs6000/rs6000.c: Include except.h.
31691 (rs6000_expand_split_stack_prologue): Call
31692 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31694 2017-02-21 Martin Jambor <mjambor@suse.cz>
31697 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31698 have been analyzed.
31700 2017-02-21 Martin Jambor <mjambor@suse.cz>
31702 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31703 for backward compatibility only.
31704 * doc/invoke.texi (Option Summary): Remove all references to
31705 -fipa-cp-alignment.
31707 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31711 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31713 * lra-constraints.c (curr_insn_transform): Handle
31714 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31716 2017-02-21 Martin Liska <mliska@suse.cz>
31718 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31720 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31722 PR translation/79638
31723 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31725 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31728 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31729 (arm_function_ok_for_sibcall): Return false for an indirect call by
31730 descriptor if all the argument registers are used.
31731 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31732 alignment of the function.
31734 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31736 PR tree-optimization/61441
31737 * simplify-rtx.c (simplify_const_unary_operation): For
31738 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31739 the sNaN unmodified.
31741 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31743 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31744 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31745 instead of SYSTEM_HEADER_DIR.
31747 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31748 Martin Liška <mliska@suse.cz>
31750 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31751 Fix typos and grammar, use active voice, and clarify.
31753 2017-02-20 Marek Polacek <polacek@redhat.com>
31755 PR middle-end/79537
31756 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31759 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31761 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31764 * config/i386/i386.c (ix86_expand_builtin): Handle
31765 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31766 ix86_builtins_isa[fcode].isa as a requirement of those
31767 flags and any other flag in the bitmask.
31768 (ix86_init_mmx_sse_builtins): Use 0 instead of
31769 ~OPTION_MASK_ISA_64BIT as mask.
31770 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31771 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31772 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31773 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31775 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31778 * lra-constraints.c (split_reg): Check requested split mode
31779 is supported by the register.
31781 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31783 * lra-constraints.c (simplify_operand_subreg): Remove early
31786 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31789 * lra-constraints.c (curr_insn_transform): Tighten condition
31790 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31792 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31795 * lra-constraints.c (curr_insn_transform): Handle
31796 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31798 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31801 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31803 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31805 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31808 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31811 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31813 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31816 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31818 * final.c (last_columnnum, override_columnnum): New variables.
31819 (final_start_function): Set last_columnnum, pass it to begin_prologue
31820 hook and pass 0 to dwarf2out_begin_prologue.
31821 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31822 to source_line debug hook.
31823 (notice_source_line): Compute last_columnnum and for debug_column_info
31824 return true on column changes.
31825 * debug.h (struct gcc_debug_hooks): Add column argument to
31826 source_line and begin_prologue hooks.
31827 (debug_nothing_int_charstar_int_bool): Remove prototype.
31828 (debug_nothing_int_int_charstar,
31829 debug_nothing_int_int_charstar_int_bool): New prototypes.
31830 (dwarf2out_begin_prologue): Add column argument.
31831 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31832 begin_prologue hooks.
31833 (debug_nothing_int_charstar_int_bool): Remove.
31834 (debug_nothing_int_int_charstar,
31835 debug_nothing_int_int_charstar_int_bool): New functions.
31836 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31837 through to dwarf2out_source_line.
31838 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31839 (dwarf2out_source_line): Add column argument, emit it if requested.
31840 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31842 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31843 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31844 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31845 through to dwarf2out_begin_prologue.
31846 (vmsdbgout_source_line): Add column argument, pass it through to
31847 dwarf2out_source_line.
31848 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31849 dbxout_source_line caller.
31850 (dbxout_source_line): Add column argument.
31852 * common.opt (gno-column-info, gcolumn-info): New options.
31853 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31854 (check_die): Also test for multiple DW_AT_decl_column attributes.
31855 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31856 DW_AT_decl_column if requested.
31857 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31859 (gen_variable_die): Likewise.
31860 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31861 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31864 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31865 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31866 (ix86_handle_option): Handle OPT_m3dnowa.
31867 * doc/invoke.texi (-m3dnowa): Document.
31868 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31869 -m3dnowa instead of -m3dnow -march=athlon.
31872 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31873 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31875 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31878 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31879 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31880 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31881 generator for vsx_xxpermdi_<mode>_be.
31882 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31883 force big-endian semantics.
31884 (vsx_xxpermdi_<mode>_be): New define_expand with same
31885 implementation as previous version of vsx_xxpermdi_<mode>.
31887 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31889 PR tree-optimization/79327
31890 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31891 variable, its initialization and use.
31893 2017-02-17 Julia Koval <julia.koval@intel.com>
31895 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31896 (OPTION_MASK_ISA_PKU_UNSET): New.
31897 (ix86_handle_option): Handle -mrdpid.
31898 * config/i386/cpuid.h (bit_RDPID): New.
31899 * config/i386/driver-i386.c (host_detect_local_cpu):
31900 Detect RDPID feature.
31901 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31902 * config/i386/i386-c.c (ix86_target_macros_internal):
31904 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31905 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31906 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31907 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31908 * config/i386/i386.md (define_insn "rdpid"): New.
31909 * config/i386/i386.opt Add -mrdpid.
31910 * config/i386/immintrin.h (_rdpid_u32): New.
31912 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31914 PR rtl-optimization/79541
31915 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31916 instead of transforming it into USE.
31918 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31920 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31921 If HONOR_SNANS (SFmode) force the input to a register.
31922 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31923 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31924 an frsp or similar insn.
31926 2017-02-17 Martin Liska <mliska@suse.cz>
31928 PR rtl-optimization/79577
31929 * params.def (selsched-max-sched-times): Increase minimum to 1.
31931 2017-02-17 Martin Liska <mliska@suse.cz>
31933 PR rtl-optimization/79574
31934 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31936 2017-02-17 Martin Liska <mliska@suse.cz>
31938 PR tree-optimization/79529
31939 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31940 ssa_defined_default_def_p to handle cases which are implicitly
31942 * tree-ssa.c (ssa_defined_default_def_p): New function.
31943 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31944 which are implicitly defined.
31945 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31947 2017-02-17 Richard Biener <rguenther@suse.de>
31949 PR middle-end/79576
31950 * params.def (max-ssa-name-query-depth): Limit to 10.
31952 2017-02-17 Richard Biener <rguenther@suse.de>
31954 PR tree-optimization/79552
31955 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31958 2017-02-17 Richard Biener <rguenther@suse.de>
31961 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31963 2017-02-17 Marek Polacek <polacek@redhat.com>
31965 PR middle-end/79536
31966 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31967 (fold_negate_expr): New wrapper.
31969 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31971 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31972 Correct terminology and de-emphasize pre-standard behavior.
31974 2017-02-16 Alan Modra <amodra@gmail.com>
31976 PR rtl-optimization/79286
31977 * ira.c (def_dominates_uses): New function.
31978 (update_equiv_regs): Don't create an equivalence for insns that
31979 may trap where the register def does not dominate the use.
31981 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31983 PR rtl-optimization/78127
31984 * lra.c (lra): Call lra_eliminate before finish the loop after
31987 2017-02-16 Richard Biener <rguenther@suse.de>
31989 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31991 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31992 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31993 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31994 (isl_val_int_from_wi): New function.
31995 (extract_affine_gmp): Rename to ...
31996 (extract_affine_wi): ... this, take a widest_int.
31997 (extract_affine_int): Just wrap extract_affine_wi.
31998 (add_param_constraints): Use isl_val_int_from_wi.
31999 (add_loop_constraints): Likewise, and extract_affine_wi.
32001 2017-02-15 Jeff Law <law@redhat.com>
32003 PR middle-end/79521
32004 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
32005 ira_init_register_move_cost_if_necessary.
32007 2017-02-15 Martin Sebor <msebor@redhat.com>
32009 PR middle-end/32003
32010 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
32011 removed in a prior commit.
32013 2017-02-15 Bin Cheng <bin.cheng@arm.com>
32015 PR tree-optimization/79347
32016 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
32017 counters during peeling.
32019 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
32021 * Makefile.in (site.exp): Remove "set ISLVER".
32023 2017-02-15 Jakub Jelinek <jakub@redhat.com>
32026 * real.c (real_from_integer): Call real_convert even for decimal.
32028 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
32031 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
32033 2017-02-14 Andrew Pinski <apinski@cavium.com>
32035 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
32036 cores and change the partno/implementer to be correct.
32037 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
32038 the 'B" as the implementer.
32039 * config/aarch64/aarch64-tune.md: Regenerate.
32041 2017-02-14 Carl Love <cel@us.ibm.com>
32043 * config/rs6000/rs6000.c: Add case statement entry to make the
32044 xvcvuxdsp built-in argument unsigned.
32045 * config/rs6000/vsx.md: Fix the source and return operand types so they
32046 match the instruction definitions from the ISA document. Fix typo
32047 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
32050 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
32053 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
32054 member early_clobber_alts.
32055 * lra-lives.c (reg_early_clobber_p): New.
32056 (process_bb_lives): Use it.
32057 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
32058 (debug_operand_data): Initialize early_clobber_alts.
32059 (setup_operand_alternative): Set up early_clobber_alts.
32060 (collect_non_operand_hard_regs): Ditto. Pass early clobber
32061 alternatives to new_insn_reg.
32062 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
32064 (lra_update_insn_regno_info): Pass the new arg.
32066 2017-02-14 Jakub Jelinek <jakub@redhat.com>
32068 PR middle-end/79505
32069 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
32070 (new_oacc_loop_raw): Don't clear already cleared fields.
32073 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
32074 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
32075 _mm512_prefetch_i64gather_ps): New inline functions and macros.
32077 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
32080 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
32082 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
32085 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
32086 the extra instruction to the right place to store 128-bit constant
32089 2017-02-14 Martin Sebor <msebor@redhat.com>
32091 PR middle-end/79448
32092 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
32093 warning for strings of unknown length.
32095 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
32097 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
32099 2017-02-14 Jeff Law <law@redhat.com>
32102 * ira-costs.c (scan_one_insn): Initialize register move costs
32103 for pseudos seen in USE/CLOBBER insns.
32105 PR tree-optimization/79095
32106 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
32107 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
32108 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
32109 if the operands are known to be not equal, then the resulting range
32111 (intersect_ranges): If the new range is ~[0,0] and the old range is
32112 wide, then prefer ~[0,0].
32113 * tree-vrp.c (overflow_comparison_p_1): New function.
32114 (overflow_comparison_p): New function.
32115 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
32116 if NAME is used in an overflow test.
32117 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
32118 overflow check that can be expressed as an equality test, then adjust
32119 ops to be that equality test.
32121 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32123 * config/s390/s390-builtin-types.def: Remove flags argument.
32124 * config/s390/s390.c (s390_init_builtins): Likewise.
32126 2017-02-14 Martin Liska <mliska@suse.cz>
32128 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
32129 vector. Fix trailing white spaces.
32131 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
32133 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
32136 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32138 PR rtl-optimization/68664
32139 * config/arm/arm.c (arm_sched_can_speculate_insn):
32140 New function. Declare prototype.
32141 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
32143 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32145 PR rtl-optimization/68664
32146 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
32148 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
32150 2017-02-14 Amit Pawar <amit.pawar@amd.com>
32152 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
32153 max skip bytes for function, loop and jump.
32155 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32157 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
32158 ABS_EXPR for gimple dump.
32160 2017-02-14 Jakub Jelinek <jakub@redhat.com>
32163 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
32165 PR tree-optimization/79408
32166 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
32167 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
32168 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
32169 also if rhs1 is INTEGER_CST.
32171 2017-02-14 Richard Biener <rguenther@suse.de>
32173 PR middle-end/79432
32174 * tree-into-ssa.c (insert_phi_nodes): When the function can
32175 have abnormal edges rewrite SSA names with broken use-def
32176 dominance out of SSA and register them for PHI insertion.
32178 2017-02-13 Martin Sebor <msebor@redhat.com>
32180 PR middle-end/79496
32181 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
32182 clearing info.nowrite flag when snprintf size argument is a range.
32184 2017-02-13 Jakub Jelinek <jakub@redhat.com>
32186 * cprop.c (cprop_jump): Add missing space in string literal.
32187 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
32188 (get_constraint_for_component_ref): Likewise.
32189 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
32190 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
32191 * lra-constraints.c (process_alt_operands): Likewise.
32192 * ipa-inline.c (inline_small_functions): Likewise.
32193 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
32194 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
32195 * trans-mem.c (diagnose_tm_1_op): Likewise.
32196 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
32197 (grid_parallel_clauses_gridifiable): Likewise.
32199 * config/nvptx/mkoffload.c (process): Add space in between
32202 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
32203 "MOD4_SSE_REGS" and "ALL_REGS".
32205 * spellcheck.c (test_data): Add , in between "foo" and "food".
32207 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32210 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
32211 boundary crossing check and subsequent code generation agree.
32213 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32215 * config/aarch64/aarch64.c (has_memory_op): Delete.
32216 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
32219 2017-02-13 Jakub Jelinek <jakub@redhat.com>
32221 PR rtl-optimization/79388
32222 PR rtl-optimization/79450
32223 * combine.c (distribute_notes): When removing TEM_INSN for which
32224 corresponding dest has last value recorded, invalidate that last
32227 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32229 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
32230 of explicit '@'. Add missing assembly comment marker on branch costs
32233 2017-02-13 Nathan Sidwell <nathan@acm.org>
32235 * gengtype-lex.l (<in_struct>): Add '/'.
32237 2017-02-13 Martin Liska <mliska@suse.cz>
32240 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
32242 2017-02-13 Richard Biener <rguenther@suse.de>
32244 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
32246 * configure: Re-generate.
32247 * config.in: Likewise.
32248 * graphite-dependences.c: Simplify as if
32249 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
32250 * graphite-isl-ast-to-gimple.c: Likewise.
32251 * graphite-optimize-isl.c: Likewise.
32252 * graphite-poly.c: Likewise.
32253 * graphite-sese-to-poly.c: Likewise.
32254 * graphite.h: Likewise.
32255 * toplev.c: Include isl/version.h and use isl_version () for
32256 printing the ISL version.
32257 * doc/install.texi: Update ISL requirement.
32259 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
32261 * doc/standards.texi (Standards): Update reference to
32264 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
32266 * doc/extend.texi (Named Address Spaces): sourceware.org now
32268 * doc/install.texi (Binaries): Ditto.
32271 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
32273 * doc/cpp.texi: Replace "stringify"/"stringification" with C
32274 standard terminology "stringize"/"stringizing" throughout.
32275 * doc/cppinternals.texi: Likewise.
32277 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
32279 * doc/extend.texi: Fix some spelling mistakes and typos.
32280 * doc/invoke.texi: Likewise.
32282 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
32285 * params.def (inline-min-speedup) Change from 10 to 8.
32287 2017-02-11 Jakub Jelinek <jakub@redhat.com>
32289 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
32292 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
32295 * ipa-inline-analysis.c (get_minimal_bb): New function.
32296 (record_modified): Use it.
32297 (remap_edge_change_prob): Handle also ancestor functions.
32299 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
32301 * doc/contrib.texi (Contributors): Remove broken link into
32302 the Mauve CVS repository.
32304 2017-02-11 Jakub Jelinek <jakub@redhat.com>
32306 PR middle-end/79454
32307 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
32308 result computation whenever lhs doesn't have vector mode, not
32309 just when it has BLKmode.
32311 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
32313 * doc/makefile.texi (profiledbootstrap): Refer to the
32314 installation instructions only in textual form.
32316 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32319 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
32321 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
32323 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
32324 (Specific): Update mingw-w64 reference.
32326 (Specific): Remove broken link to Renesas RX processor.
32328 2017-02-10 Richard Biener <rguenther@suse.de>
32330 * toplev.c (process_options): Do not mention obsolete graphite
32331 options when printing sorry message about missing graphite support.
32332 Mention -floop-nest-optimize.
32334 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
32336 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
32337 (vtst_p16): Likewise.
32338 (vtstq_p8): Likewise.
32339 (vtstq_p16): Likewise.
32341 (vtstq_p64): Likewise.
32342 * config/arm/arm_neon.h (vgetq_lane_p64): New.
32343 (vset_lane_p64): New.
32344 (vsetq_lane_p64): New.
32346 2017-02-10 Jakub Jelinek <jakub@redhat.com>
32348 PR tree-optimization/79411
32349 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
32350 stmt operands are SSA_NAMEs used in abnormal phis.
32351 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
32354 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32357 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
32360 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32362 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
32364 2017-02-09 Jakub Jelinek <jakub@redhat.com>
32366 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
32370 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
32371 not arbitrary TREE_CONSTANT.
32374 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
32375 "omp declare target link" attribute unless is_global_var.
32376 * omp-offload.c (find_link_var_op): Likewise.
32378 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
32379 Chung-Lin Tang <cltang@codesourcery.com>
32381 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
32383 (gimplify_adjust_omp_clauses): Don't delete TILE.
32384 (gimplify_omp_for): Deal with TILE.
32385 * internal-fn.c (expand_GOACC_TILE): New function.
32386 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
32388 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
32389 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
32391 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
32392 avoid DIV for outermost collapse var.
32393 (expand_oacc_for): Insert tile element loop as needed. Adjust.
32394 Remove out of date comments, fix whitespace.
32395 * omp-general.c (omp_extract_for_data): Deal with tiling.
32396 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
32397 adjust OLF_DIM_BASE value.
32398 (struct omp_for_data): Add tiling field.
32399 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
32400 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
32401 for auto loops. Remove default auto determining, moved to
32402 oacc_loop_fixed_partitions.
32403 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
32404 stmts, add e_mask field.
32405 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
32406 (oacc_thread_numbers): Use oacc_dim_call.
32407 (oacc_xform_tile): New.
32408 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
32409 (finish_oacc_loop): Adjust for ifns vector.
32410 (oacc_loop_discover_walk): Append loop abstraction sites to list,
32411 add case for GOACC_TILE fns.
32412 (oacc_loop_xform_loop): Delete.
32413 (oacc_loop_process): Iterate over call list directly, and add
32414 handling for GOACC_TILE fns.
32415 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
32417 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
32418 vector partitioning to outer loops. Assign 2 partitions to loops
32419 when available. Add TILE handling.
32420 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
32421 (execite_oacc_device_lower): Process GOACC_TILE fns,
32422 ignore unknown specs.
32423 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
32424 * tree.c (omp_clause_num_ops): Adjust TILE ops.
32425 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
32427 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
32429 * configure.ac (ACX_BUGURL): Update.
32430 * configure: Regenerate.
32432 2017-02-09 Richard Biener <rguenther@suse.de>
32434 PR tree-optimization/69823
32435 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
32436 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
32438 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32440 * config/arc/arc-c.def: Add __NPS400__ definition.
32441 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
32442 (TARGET_NPS400): Define.
32444 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32446 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
32448 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
32449 pointer, arch_info.
32450 (arc_cpu_types): Fill the arch_info field with a pointer into the
32451 arc_arch_types table.
32452 (arc_selected_cpu): Declare.
32453 * config/arc/arc.c (arc_selected_cpu): Make global.
32454 (arc_selected_arch): Delete.
32455 (arc_base_cpu): Delete.
32456 (arc_override_options): Remove references to deleted variables,
32457 update access to arch information.
32458 (ARC_OPT): Update access to arch information.
32459 (ARC_OPTX): Likewise.
32460 * config/arc/arc.h (arc_base_cpu): Remove declaration.
32461 (TARGET_ARC600): Update access to arch information.
32462 (TARGET_ARC601): Likewise.
32463 (TARGET_ARC700): Likewise.
32464 (TARGET_EM): Likewise.
32465 (TARGET_HS): Likewise.
32466 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
32469 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
32472 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
32473 condition/operands for integer GE/LE/GEU/LEU operations.
32475 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
32477 PR translation/79397
32478 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
32481 2017-02-08 Martin Jambor <mjambor@suse.cz>
32484 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
32485 whether allocation happened.
32486 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
32487 nothing was allocated.
32489 2017-02-08 Jakub Jelinek <jakub@redhat.com>
32491 PR tree-optimization/79408
32492 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
32493 constant, but SSA_NAME with a known integer range, use the minimum
32494 of that range instead of op1 to determine if modulo can be replaced
32495 with its first operand.
32497 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32499 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
32501 2017-02-08 Richard Biener <rguenther@suse.de>
32503 PR tree-optimization/71824
32504 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32505 Check all loops contained in the merged region.
32507 2017-02-07 Andrew Pinski <apinski@cavium.com>
32509 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
32511 2017-02-07 Andrew Pinski <apinski@cavium.com>
32513 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
32514 (thunderxt88): Likewise.
32515 (thunderxt81): Disable LSE and change v8.1 to v8.
32516 (thunderxt83): Likewise.
32518 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32519 Richard Biener <rguenther@suse.de>
32521 PR middle-end/79399
32522 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
32523 type from int to size_t.
32524 * ira-costs.c (struct_costs_size): Change type from int to size_t.
32526 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32528 PR rtl-optimization/79386
32529 * cprop.c (bypass_conditional_jumps): Initialize
32530 bypass_last_basic_block already before splitting bbs after
32531 unconditional traps...
32532 (bypass_conditional_jumps): ... rather than here.
32535 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
32536 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
32537 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
32538 fix -masm=intel patterns.
32540 2017-02-07 Richard Biener <rguenther@suse.de>
32542 PR tree-optimization/79256
32543 PR middle-end/79278
32544 * builtins.c (get_object_alignment_2): Use min_align_of_type
32545 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
32546 and ADJUST_FIELD_ALIGN.
32548 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
32550 * doc/tm.texi: Regenerate.
32551 * stor-layout.c (layout_decl): Adjust.
32552 (update_alignment_for_field): Likewise.
32553 (place_field): Likewise.
32554 (min_align_of_type): Likewise.
32555 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
32556 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
32557 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
32558 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
32559 * config/frv/frv.c (frv_adjust_field_align): Likewise.
32560 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
32561 * config/i386/i386.c (x86_field_alignment): Likewise.
32562 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
32563 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
32564 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
32565 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
32566 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
32567 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
32571 2017-01-30 Richard Biener <rguenther@suse.de>
32573 PR tree-optimization/79256
32574 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32575 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32578 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
32580 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
32581 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
32582 builtins to SImode and emit a zero-extend, if necessary.
32584 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32586 * docs/invoke.texi (RISC-V Options): Alphabetize.
32588 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32590 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
32593 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32595 * config/riscv/riscv.c: New file.
32596 * common/config/riscv/riscv-common.c: Likewise.
32597 * config.gcc: Likewise.
32598 * config/riscv/constraints.md: Likewise.
32599 * config/riscv/elf.h: Likewise.
32600 * config/riscv/generic.md: Likewise.
32601 * config/riscv/linux.h: Likewise.
32602 * config/riscv/multilib-generator: Likewise.
32603 * config/riscv/peephole.md: Likewise.
32604 * config/riscv/pic.md: Likewise.
32605 * config/riscv/predicates.md: Likewise.
32606 * config/riscv/riscv-builtins.c: Likewise.
32607 * config/riscv/riscv-c.c: Likewise.
32608 * config/riscv/riscv-ftypes.def: Likewise.
32609 * config/riscv/riscv-modes.def: Likewise.
32610 * config/riscv/riscv-opts.h: Likewise.
32611 * config/riscv/riscv-protos.h: Likewise.
32612 * config/riscv/riscv.h: Likewise.
32613 * config/riscv/riscv.md: Likewise.
32614 * config/riscv/riscv.opt: Likewise.
32615 * config/riscv/sync.md: Likewise.
32616 * config/riscv/t-elf-multilib: Likewise.
32617 * config/riscv/t-linux: Likewise.
32618 * config/riscv/t-linux-multilib: Likewise.
32619 * config/riscv/t-riscv: Likewise.
32620 * configure.ac: Likewise.
32621 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
32622 Waterman as RISC-V maintainers.
32623 * doc/install.texi: Add RISC-V entries.
32624 * doc/invoke.texi: Add RISC-V options section.
32625 * doc/md.texi: Add RISC-V constraints section.
32626 * configure: Regenerated.
32628 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
32631 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
32632 false values to be constant vectors with all 0 or all 1 bits set.
32633 (vcondu<mode><mode>): Likewise.
32634 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
32636 (fpmask_comparison_operator): Update comment.
32637 (vecint_comparison_operator): New predicate.
32638 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
32639 vector conditionals when the true and false values are constant
32640 vectors with all 0 bits or all 1 bits set.
32642 2017-02-06 Martin Sebor <msebor@redhat.com>
32644 PR tree-optimization/79376
32645 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
32647 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32649 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32650 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32651 to simplify split condition.
32653 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32655 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32656 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32659 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32661 PR rtl-optimization/68664
32662 * target.def (can_speculate_insn): New hook.
32663 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32664 * doc/tm.texi: Regenerate.
32665 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32666 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32667 (rs6000_sched_can_speculate_insn): New function.
32669 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32671 PR tree-optimization/79284
32672 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32673 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32674 vectorizable_mask_load_store, vectorizable_operation,
32675 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32676 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32677 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32678 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32679 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32680 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32681 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32682 is_gimple_assign (stmt). Replace another such test with
32683 is_gimple_assign (stmt).
32685 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32688 * config/avr/avr.c (rtl-iter.h): Include it.
32689 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32690 (avr_legitimate_combined_insn): ...and implementation.
32692 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32694 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32695 * config/s390/s390.c (s390_const_operand_ok)
32696 (s390_canonicalize_comparison, s390_extract_part)
32697 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32698 (s390_contiguous_bitmask_p, s390_rtx_costs)
32699 (legitimize_pic_address): Likewise.
32700 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32701 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32702 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32703 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32704 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32706 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32708 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32709 REGNO($0) == REGNO($1).
32711 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32713 * config/s390/linux.h(SIZE_TYPE): Add comment.
32715 2017-02-06 Julian Brown <julian@codesourcery.com>
32716 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32717 Virendra Pathak <virendra.pathak@broadcom.com>
32719 * config/aarch64/aarch64-cores.def: Change the scheduler
32721 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32722 * config/aarch64/thunderx2t99.md: New file.
32724 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32726 * doc/standards.texi (Go Language): Update link to language
32729 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32731 * tree-eh.c (lower_resx): Sanitize profile.
32732 (cleanup_empty_eh_move_lp): Likewise.
32734 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32737 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32739 * cfgloopmanip.h (loop_version): Update prototype.
32740 * modulo-sched.c (sms_schedule): Update call of loop_version.
32741 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32742 * tree-parloops.c (gen_parallel_loop): Likewise.
32743 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32744 * tree-ssa-loop-split.c (split_loop): Likewise.
32745 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32746 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32748 2017-02-05 Martin Liska <mliska@suse.cz>
32751 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32753 (print_operand_address): Initialize a struct to zero.
32755 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32757 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32758 garbage collector only in textual form.
32760 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32762 * doc/extend.texi (x86 specific memory model extensions for
32763 transactional memory): Simplify a phrase.
32765 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32768 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32769 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32770 (atomic_storedi_1): Likewise.
32772 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32774 PR tree-optimization/79338
32775 * tree-parloops.c (gather_scalar_reductions): Don't call
32776 vect_analyze_loop_form for loop->inner before destroying loop's
32779 2017-02-03 Martin Sebor <msebor@redhat.com>
32781 PR tree-optimization/79327
32782 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32783 when precision has resulted in leading zeros.
32784 (format_integer): Adjust the likely counter to assume an unknown
32785 argument that may be zero is non-zero.
32787 2017-02-03 Jason Merrill <jason@redhat.com>
32790 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32791 avoid copying non-taken branch.
32793 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32795 PR tree-optimization/79340
32796 * tree-vect-loop.c (vectorizable_reduction): Release
32797 vec_defs elements after safe_splicing them into other vectors.
32800 PR tree-optimization/79327
32801 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32802 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32804 (format_integer): Use wide_int_to_tree instead of build_int_cst
32805 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32806 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32807 of shortest and longest sequence.
32809 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32811 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32812 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32814 2017-02-03 Walter Lee <walt@tilera.com>
32817 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32818 after initial stackframe link reg save.
32819 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32821 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32824 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32825 wu for stxssp alternative.
32827 2017-02-03 Martin Sebor <msebor@redhat.com>
32829 PR tree-optimization/79352
32830 * gimple-fold.c (get_range_strlen): Add argument.
32831 (get_range_strlen): Change return type to bool.
32832 (get_maxval_strlen): Pass in a dummy argument.
32833 * gimple-fold.h (get_range_strlen): Change return type to bool.
32834 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32835 * tree.h (array_at_struct_end_p): Add argument.
32836 * tree.c (array_at_struct_end_p): Handle it.
32838 2017-02-03 Martin Liska <mliska@suse.cz>
32841 * multiple_target.c (create_dispatcher_calls): Redirect edge
32842 from a caller of a dispatcher.
32843 (expand_target_clones): Make the clones local.
32844 (ipa_target_clone): Do both target clones and resolvers.
32845 (ipa_dispatcher_calls): Remove the pass.
32846 (pass_dispatcher_calls::gate): Likewise.
32847 (make_pass_dispatcher_calls): Likewise.
32848 * passes.def (pass_target_clone): Put as very first IPA early
32851 2017-02-03 Martin Liska <mliska@suse.cz>
32853 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32854 in case of a function with ifunc attribute.
32856 2017-02-03 Martin Liska <mliska@suse.cz>
32858 * cgraph.c (cgraph_node::dump): Dump function version info.
32859 * symtab.c (symtab_node::dump_base): Add missing new line.
32861 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32863 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32864 (ifcombine_ifandif): Use it.
32866 2017-02-03 Martin Liska <mliska@suse.cz>
32868 * doc/invoke.texi: Document default value for
32869 use-after-scope-direct-emission-threshold.
32871 2017-02-03 Martin Liska <mliska@suse.cz>
32873 PR tree-optimization/79339
32874 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32875 (format_floating): Likewise.
32877 2017-02-03 Martin Liska <mliska@suse.cz>
32880 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32882 (ipa_node_params_t::remove): Likewise.
32883 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32884 initialization from removed ipa_node_params_t::insert.
32885 (ipa_node_params::~ipa_node_params): Move from removed
32886 ipa_node_params_t::release.
32887 * symbol-summary.h (symbol_summary::m_released): New member.
32888 Do not release a summary twice. Do not allow to call finalizer
32889 for types of a summary that live in GGC memory.
32891 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32893 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32896 2017-02-02 Martin Sebor <msebor@redhat.com>
32898 PR middle-end/79275
32899 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32900 (format_string): Tighten up the range of output for non-constant
32901 strings and correct the expected range for wide non-constant strings.
32903 2017-02-02 Martin Sebor <msebor@redhat.com>
32905 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32907 PR middle-end/32003
32908 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32910 (-fdump-tree-@var): Add to index and document how to come up
32911 with pass-specific option and dump file names.
32912 (-fdump-passes): Clarify where to look for output.
32914 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32916 PR middle-end/77445
32917 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32918 statistics of the analyzed path; allow threading for speed when
32919 any of BBs along the path are optimized for speed.
32921 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32923 PR middle-end/78468
32924 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32925 settings of the virtual registers.
32928 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32930 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32931 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32934 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32936 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32937 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32939 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32941 * config/s390/s390.md: Add missing comments with the expanded
32943 * config/s390/vector.md: Likewise.
32944 * config/s390/vx-builtins.md: Likewise.
32946 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32949 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32950 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32951 conditions on a single line.
32953 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32955 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32956 __S390_VX__ to __VX__.
32958 2017-02-01 Andrew Pinski <apinski@cavium.com>
32960 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32961 stmt_info to record_stmt_cost.
32962 (vect_get_known_peeling_cost): Pass stmt_info if known to
32964 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32965 cpu_vector_cost field into
32966 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32967 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32968 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32969 splitting of scalar_stmt_cost and vec_stmt_cost.
32970 (thunderx_vector_cost): Likewise.
32971 (cortexa57_vector_cost): LIkewise.
32972 (exynosm1_vector_cost): Likewise.
32973 (xgene1_vector_cost): Likewise.
32974 (thunderx2t99_vector_cost): Improve after the splitting of the two
32976 (aarch64_builtin_vectorization_cost): Update for the splitting of
32977 scalar_stmt_cost and vec_stmt_cost.
32979 2017-02-01 Torvald Riegel <triegel@redhat.com>
32980 Richard Henderson <rth@redhat.com>
32982 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32983 conditional on existance of a fast atomic load.
32984 * optabs-query.c (can_atomic_load_p): New function.
32985 * optabs-query.h (can_atomic_load_p): Declare it.
32986 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32987 no fast atomic load is available for the particular size of access.
32988 (expand_atomic_compare_and_swap): Likewise.
32989 (expand_atomic_load): Likewise.
32990 (expand_atomic_store): Likewise.
32991 (expand_atomic_fetch_op): Likewise.
32992 * testsuite/lib/target-supports.exp
32993 (check_effective_target_sync_int_128): Remove x86 because it provides
32994 no fast atomic load.
32995 (check_effective_target_sync_int_128_runtime): Likewise.
32997 2017-02-01 Richard Biener <rguenther@suse.de>
32999 * graphite.c: Include tree-vectorizer.h for find_loop_location.
33000 (graphite_transform_loops): Provide opt-info for optimized nests.
33001 * tree-parloop.c (parallelize_loops): Provide opt-info for
33002 parallelized loops.
33004 2017-02-01 Richard Biener <rguenther@suse.de>
33006 PR middle-end/79315
33007 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
33008 was not set before.
33010 2017-02-01 Richard Biener <rguenther@suse.de>
33012 PR tree-optimization/71824
33013 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
33014 Verify the loops are valid in the merged SESE region.
33015 (scop_detection::can_represent_loop_1): Check analyzing the
33016 evolution of the number of iterations in the region succeeds.
33018 2017-01-31 Ian Lance Taylor <iant@golang.org>
33020 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
33021 REG_ARGS_SIZE note to 32-bit push insns and call insn.
33023 2017-01-31 David Malcolm <dmalcolm@redhat.com>
33025 PR preprocessor/79210
33026 * input.c (get_substring_ranges_for_loc): Replace line_width
33027 assertion with error-handling.
33029 2017-01-31 Richard Biener <rguenther@suse.de>
33031 PR tree-optimization/77318
33032 * graphite-sese-to-poly.c (extract_affine): Fix assert.
33033 (create_pw_aff_from_tree): Take loop parameter.
33034 (add_condition_to_pbb): Pass loop of the condition to
33035 create_pw_aff_from_tree.
33037 2017-01-31 Jakub Jelinek <jakub@redhat.com>
33039 * config/s390/s390.c (s390_asan_shadow_offset): New function.
33040 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
33042 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
33046 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
33048 (convert_int_to_float128): Likewise.
33049 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
33050 (convert_int_to_float128): Likewise.
33051 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
33052 (UNSPEC_IEEE128_CONVERT): Likewise.
33053 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
33054 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
33055 Use local variables for IBM extended format.
33056 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
33057 (fix_trunc<mode>si2_fprs): Likewise.
33058 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
33059 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
33060 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
33061 to know that we can now have integers of all sizes in vector
33063 (fix<uns>_<mode>di2_hw): Likewise.
33064 (float<uns>_<mode>si2_hw): Likewise.
33065 (fix_<mode>si2_hw): Likewise.
33066 (fixuns_<mode>si2_hw): Likewise.
33067 (float<uns>_<mode>di2_hw): Likewise.
33068 (float_<mode>di2_hw): Likewise.
33069 (float_<mode>si2_hw): Likewise.
33070 (floatuns_<mode>di2_hw): Likewise.
33071 (floatuns_<mode>si2_hw): Likewise.
33072 (xscvqp<su>wz_<mode>): Delete, no longer used.
33073 (xscvqp<su>dz_<mode>): Likewise.
33074 (xscv<su>dqp_<mode>): Likewise.
33075 (ieee128_mfvsrd_64bit): Likewise.
33076 (ieee128_mfvsrd_32bit): Likewise.
33077 (ieee128_mfvsrwz): Likewise.
33078 (ieee128_mtvsrw): Likewise.
33079 (ieee128_mtvsrd_64bit): Likewise.
33080 (ieee128_mtvsrd_32bit): Likewise.
33082 2017-01-31 Martin Liska <mliska@suse.cz>
33085 * ipa-prop.c (ipa_free_all_node_params): Call release method
33086 instead of ~sumbol_summary to not to trigger double times
33089 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
33091 PR tree-optimization/71691
33092 * bitmap.h (class auto_bitmap): New.
33093 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
33094 is_maybe_undefined instead of ssa_undefined_value_p.
33096 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33098 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
33099 __S390_ARCH_LEVEL__ to __ARCH__.
33101 2017-01-31 Jakub Jelinek <jakub@redhat.com>
33103 PR tree-optimization/79267
33104 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
33105 if should_remove_lhs_p is true.
33107 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
33110 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
33111 (add_alignment_attribute): New.
33112 (base_type_die): Add alignment attribute.
33113 (subrange_type_die): Likewise.
33114 (modified_type_die): Likewise.
33115 (gen_array_type_die): Likewise.
33116 (gen_descr_array_type_die: Likewise.
33117 (gen_enumeration_type_die): Likewise.
33118 (gen_subprogram_die): Likewise.
33119 (gen_variable_die): Likewise.
33120 (gen_field_die): Likewise.
33121 (gen_ptr_to_mbr_type_die): Likewise.
33122 (gen_struct_or_union_type_die): Likewise.
33123 (gen_subroutine_type_die): Likewise.
33124 (gen_typedef_die): Likewise.
33125 (base_type_cmp): Compare alignment attribute.
33127 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33130 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
33131 (setb_unsigned) New pattern for setb with CCUNS.
33132 * config/rs6000/rs6000.c (expand_block_compare): Use a different
33133 subfc./subfe sequence to avoid overflow problems. Generate a
33134 shorter sequence with cmpld/setb for power9.
33135 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
33136 for generating subfc. instruction.
33137 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
33138 now uses this instruction.
33140 2017-01-30 Ian Lance Taylor <iant@google.com>
33143 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
33144 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
33146 2017-01-30 Martin Sebor <msebor@redhat.com>
33148 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
33149 Move constant to the right of a relational operator.
33150 (get_mpfr_format_length, format_character, format_string): Ditto.
33151 (should_warn_p, maybe_warn): Same.
33153 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
33155 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
33158 * asan.c (get_translation_unit_decl): Remove function.
33159 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
33161 2017-01-30 Martin Liska <mliska@suse.cz>
33163 PR gcov-profile/79259
33164 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
33165 -fprofile-generate.
33167 2017-01-30 Martin Liska <mliska@suse.cz>
33170 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
33171 Initialize variables with NULL value.
33173 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
33176 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
33178 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
33180 2017-01-30 Richard Biener <rguenther@suse.de>
33182 PR tree-optimization/79276
33183 * tree-vrp.c (process_assert_insertions): Properly adjust common
33184 when removing a duplicate.
33186 2017-01-30 Richard Biener <rguenther@suse.de>
33188 PR tree-optimization/79256
33189 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
33190 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
33192 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
33194 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
33197 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
33198 ("*r<noxa>sbg_<mode>_sll_bitmask")
33199 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
33200 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
33201 Use contiguous_bitmask_nowrap_operand.
33203 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33206 * config/rs6000/altivec.h (vec_xl): Revise #define.
33207 (vec_xst): Likewise.
33209 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
33211 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
33213 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
33215 PR rtl-optimization/79194
33216 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
33217 traps before call to bypass_conditional_jumps.
33219 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
33221 PR tree-optimization/71374
33222 * lra-constraints.c (check_conflict_input_operands): New.
33223 (match_reload): Use it.
33225 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
33228 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
33229 account to calculate conflict_set.
33231 2017-01-27 Bin Cheng <bin.cheng@arm.com>
33233 PR rtl-optimization/78559
33234 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
33235 other_insn in combine.
33237 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
33239 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
33240 uint16_type_node for BT_UINT16.
33242 2017-01-27 David Malcolm <dmalcolm@redhat.com>
33244 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
33245 "RTL Tests" to menu.
33246 (GIMPLE Tests): New node.
33247 (RTL Tests): New node.
33249 2017-01-27 Richard Biener <rguenther@suse.de>
33251 PR tree-optimization/79245
33252 * tree-loop-distribution.c (distribute_loop): Apply cost
33253 modeling also to detected patterns.
33255 2017-01-27 Richard Biener <rguenther@suse.de>
33257 PR tree-optimization/71433
33258 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
33259 (compare_assert_loc): New function.
33260 (process_assert_insertions): Sort and optimize assert locations
33261 to remove duplicates and push down identical assertions on
33262 edges to their destination block.
33264 2017-01-27 Richard Biener <rguenther@suse.de>
33266 PR tree-optimization/79244
33267 * tree-vrp.c (remove_range_assertions): Forcefully propagate
33268 out SSA names even if abnormal.
33270 2017-01-27 Jakub Jelinek <jakub@redhat.com>
33272 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
33273 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
33274 instead of MPFR_RNDN.
33276 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
33279 * arm.c (arm_option_override): Don't call build_target_option_node
33280 until after doing all option overrides.
33281 (arm_valid_target_attribute_tree): Likewise.
33283 2017-01-27 Martin Liska <mliska@suse.cz>
33285 * doc/invoke.texi (-fprofile-arcs): Document profiling support
33286 for {cd}tors and C++ {cd}tors.
33288 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
33290 * config/s390/s390.md ("*setmem_long_and")
33291 ("*setmem_long_and_31z"): Use zero_extend instead of and.
33293 2017-01-26 Martin Sebor <msebor@redhat.com>
33295 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
33298 2017-01-26 Martin Sebor <msebor@redhat.com>
33300 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
33301 HAVE_DFmode before using XFmode or DFmode.
33302 (parse_directive): Avoid using the z length modifier to avoid
33303 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
33305 PR middle-end/78703
33306 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
33307 to accept adjustment as an array.
33308 (get_int_range): New function.
33309 (struct directive): Make width and prec arrays.
33310 (directive::set_width, directive::set_precision): Call get_int_range.
33311 (format_integer, format_floating): Handle width and precision ranges.
33312 (format_string, parse_directive): Same.
33314 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33317 * dwarf2out.c (generate_skeleton_bottom_up): For children with
33318 comdat_type_p set, just clone them, but keep the children in the
33322 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
33323 which have direct callers with -fvar-tracking-assignments enabled
33325 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
33326 inside of type units.
33328 2017-01-26 Martin Sebor <msebor@redhat.com>
33330 PR middle-end/78703
33331 * gimple-ssa-sprintf.c (struct result_range): Add likely and
33333 (struct format_result): Replace number_chars, number_chars_min,
33334 and number_chars_max with a single member of struct result_range.
33336 (format_result::operator+=): Adjust.
33337 (struct fmtresult): Remove bounded. Handle likely and unlikely
33339 (fmtresult::adjust_for_width_or_precision): New function.
33340 (fmtresult:type_max_digits): New function.
33341 (bytes_remaining): Handle likely and unlikely counters.
33342 (min_bytes_remaining): Remove.
33343 (format_percent): Simplify.
33344 (format_integer, format_floating): Set likely and unlikely counters.
33345 (get_string_length, format_character, format_string): Same.
33346 (format_plain, should_warn_p): New function.
33347 (maybe_warn): Call should_warn_p. Update diagnostic messages
33348 and handle those for all directives, including plain strings.
33349 (format_directive): Handle likely and unlikely counters.
33350 Remove unnecessary quoting from diagnostics. Add an informational
33352 (add_bytes): Remove.
33353 (pass_sprintf_length::compute_format_length): Simplify.
33354 (try_substitute_return_value): Handle likely and unlikely counters.
33356 2017-01-26 Carl Love <cel@us.ibm.com>
33358 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
33359 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
33361 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
33364 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
33365 endianess for subregs into account.
33366 * lra-constraints.c (lra_constraints): Do risky transformations
33367 always on the first iteration.
33368 * lra-lives.c (check_pseudos_live_through_calls): Add arg
33369 last_call_used_reg_set.
33370 (process_bb_lives): Define and use last_call_used_reg_set.
33371 * lra.c (lra): Always continue after lra_constraints on the first
33374 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
33376 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
33378 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
33380 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33382 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
33383 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
33384 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
33385 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
33386 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
33387 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
33388 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
33389 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
33390 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
33392 2017-01-26 Marek Polacek <polacek@redhat.com>
33395 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
33396 for the third operand.
33398 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33400 PR middle-end/79236
33401 * omp-low.c (struct omp_context): Add simt_stmt field.
33402 (scan_omp_for): Return omp_context *.
33403 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
33404 context to the _simt_ SIMD stmt.
33405 (lower_omp_for): For combined SIMD with sibling _simt_
33406 SIMD, make sure to use the same decls in _looptemp_
33407 clauses as in the sibling.
33409 2017-01-26 David Sherwood <david.sherwood@arm.com>
33411 PR middle-end/79212
33412 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
33415 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33418 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
33419 emit fld b; fld a; if possible.
33421 * brig-builtins.def: Update copyright years.
33422 * config/arm/arm_acle_builtins.def: Update copyright years.
33424 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
33427 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
33428 constraint instead of o for the stxsd instruction.
33430 2017-01-25 Carl Love <cel@us.ibm.com>
33432 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
33433 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
33435 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
33437 * doc/invoke.texi (C++ Dialect Options): Fix typo.
33439 2017-01-25 Richard Biener <rguenther@suse.de>
33441 PR tree-optimization/69264
33442 * target.def (vector_alignment_reachable): Improve documentation.
33443 * doc/tm.texi: Regenerate.
33444 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
33446 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
33447 earlier changes with respect to TYPE_USER_ALIGN.
33448 (vector_alignment_reachable_p): Likewise. Improve dumping.
33450 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33453 * config/arm/arm.md (xordi3): Force constant operand into a register
33456 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33458 * doc/invoke.texi (-fstore-merging): Correct default optimization
33459 levels at which it is enabled.
33460 (-O): Move -fstore-merging from list to...
33463 2017-01-25 Richard Biener <rguenther@suse.de>
33466 * omp-expand.c: Include debug.h.
33467 (expand_omp_taskreg): Make sure to generate early debug before
33468 outlining anything from a function.
33469 (expand_omp_target): Likewise.
33470 (grid_expand_target_grid_body): Likewise.
33472 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
33475 * asan.c (get_translation_unit_decl): New function.
33476 (asan_add_global): Extract modules file name from globals
33477 TRANSLATION_UNIT_DECL name.
33479 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
33482 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
33483 for long calls with APCS frame and VFP.
33485 2017-01-24 David Malcolm <dmalcolm@redhat.com>
33487 * cfg.c (original_copy_tables_initialized_p): New function.
33488 * cfg.h (original_copy_tables_initialized_p): New decl.
33489 * cfgrtl.c (relink_block_chain): Guard the call to
33490 free_original_copy_tables with a call to
33491 original_copy_tables_initialized_p.
33492 * cgraph.h (symtab_node::native_rtl_p): New decl.
33493 * cgraphunit.c (symtab_node::native_rtl_p): New function.
33494 (symtab_node::needed_p): Don't assert for early assembly output
33495 for __RTL functions.
33496 (cgraph_node::finalize_function): Set "force_output" for __RTL
33498 (cgraph_node::analyze): Bail out early for __RTL functions.
33499 (analyze_functions): Update assertion to support __RTL functions.
33500 (cgraph_node::expand): Bail out early for __RTL functions.
33501 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
33503 * function.h (struct function): Update comment for field
33505 * gimple-expr.c: Include "tree-pass.h".
33506 (gimple_has_body_p): Return false for __RTL functions.
33507 * Makefile.in (OBJS): Add run-rtl-passes.o.
33508 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
33510 (gcc::pass_manager::get_clean_slate): New accessor.
33511 * passes.c: Include "insn-addr.h".
33512 (should_skip_pass_p): Add logging. Update logic for running
33513 "expand" to be compatible with both __GIMPLE and __RTL. Guard
33514 property-provider override so it is only done for gimple passes.
33516 (skip_pass): New function.
33517 (execute_one_pass): Call skip_pass when skipping passes.
33518 * read-md.c (md_reader::read_char): Support filtering
33519 the input to a subset of line numbers.
33520 (md_reader::md_reader): Initialize fields
33521 m_first_line and m_last_line.
33522 (md_reader::read_file_fragment): New function.
33523 * read-md.h (md_reader::read_file_fragment): New decl.
33524 (md_reader::m_first_line): New field.
33525 (md_reader::m_last_line): New field.
33526 * read-rtl-function.c (function_reader::create_function): Only
33527 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
33528 curr_properties. Set DECL_INITIAL to a dummy block.
33529 (read_rtl_function_body_from_file_range): New function.
33530 * read-rtl-function.h (read_rtl_function_body_from_file_range):
33532 * run-rtl-passes.c: New file.
33533 * run-rtl-passes.h: New file.
33535 2017-01-24 Jeff Law <law@redhat.com>
33537 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
33540 2017-01-24 Bin Cheng <bin.cheng@arm.com>
33542 PR tree-optimization/79159
33543 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
33544 (record_nonwrapping_iv): Improve boundary using above function if no
33545 value range information.
33547 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
33548 Martin Jambor <mjambor@suse.cz>
33550 * brig-builtins.def: New file.
33551 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
33552 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
33553 (DEF_HSAIL_SAT_BUILTIN): Likewise.
33554 (DEF_HSAIL_INTR_BUILTIN): Likewise.
33555 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
33556 * builtin-types.def (BT_INT8): New.
33557 (BT_INT16): Likewise.
33558 (BT_UINT8): Likewise.
33559 (BT_UINT16): Likewise.
33560 (BT_FN_ULONG): Likewise.
33561 (BT_FN_UINT_INT): Likewise.
33562 (BT_FN_UINT_ULONG): Likewise.
33563 (BT_FN_UINT_LONG): Likewise.
33564 (BT_FN_UINT_PTR): Likewise.
33565 (BT_FN_ULONG_PTR): Likewise.
33566 (BT_FN_INT8_FLOAT): Likewise.
33567 (BT_FN_INT16_FLOAT): Likewise.
33568 (BT_FN_UINT32_FLOAT): Likewise.
33569 (BT_FN_UINT16_FLOAT): Likewise.
33570 (BT_FN_UINT8_FLOAT): Likewise.
33571 (BT_FN_UINT64_FLOAT): Likewise.
33572 (BT_FN_UINT16_UINT32): Likewise.
33573 (BT_FN_UINT32_UINT16): Likewise.
33574 (BT_FN_UINT16_UINT16_UINT16): Likewise.
33575 (BT_FN_INT_PTR_INT): Likewise.
33576 (BT_FN_UINT_PTR_UINT): Likewise.
33577 (BT_FN_LONG_PTR_LONG): Likewise.
33578 (BT_FN_ULONG_PTR_ULONG): Likewise.
33579 (BT_FN_VOID_UINT64_UINT64): Likewise.
33580 (BT_FN_UINT8_UINT8_UINT8): Likewise.
33581 (BT_FN_INT8_INT8_INT8): Likewise.
33582 (BT_FN_INT16_INT16_INT16): Likewise.
33583 (BT_FN_INT_INT_INT): Likewise.
33584 (BT_FN_UINT_FLOAT_UINT): Likewise.
33585 (BT_FN_FLOAT_UINT_UINT): Likewise.
33586 (BT_FN_ULONG_UINT_UINT): Likewise.
33587 (BT_FN_ULONG_UINT_PTR): Likewise.
33588 (BT_FN_ULONG_ULONG_ULONG): Likewise.
33589 (BT_FN_UINT_UINT_UINT): Likewise.
33590 (BT_FN_VOID_UINT_PTR): Likewise.
33591 (BT_FN_UINT_UINT_PTR: Likewise.
33592 (BT_FN_UINT32_UINT64_PTR): Likewise.
33593 (BT_FN_INT_INT_UINT_UINT): Likewise.
33594 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
33595 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
33596 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
33597 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
33598 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
33599 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
33600 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
33601 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
33602 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
33603 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
33604 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
33605 * doc/frontends.texi: List BRIG FE.
33606 * doc/install.texi (Testing): Add BRIG tesring requirements.
33607 * doc/invoke.texi (Overall Options): Mention BRIG.
33608 * doc/standards.texi (Standards): Doucment BRIG HSA version.
33610 2017-01-24 Richard Biener <rguenther@suse.de>
33612 PR translation/79208
33613 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
33615 2017-01-24 Martin Jambor <mjambor@suse.cz>
33618 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
33619 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
33620 and known_contexts.
33622 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
33624 PR middle-end/79123
33625 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
33626 casts from signed to unsigned really don't have a range.
33628 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
33630 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
33631 GMP_RNDx for compatiblity.
33633 2017-01-24 Martin Liska <mliska@suse.cz>
33636 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
33637 that would prevent us to call alloca with -1 as argument.
33639 2017-01-24 Jakub Jelinek <jakub@redhat.com>
33641 * dwarf2out.c (output_compilation_unit_header, output_file_names):
33642 Avoid -Wformat-security warning.
33644 2017-01-23 Andrew Pinski <apinski@cavium.com>
33646 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33649 2017-01-23 Martin Sebor <msebor@redhat.com>
33651 PR middle-end/78703
33652 * gimple-ssa-sprintf.c (warn_level): New global.
33653 (format_integer): Use it here and throughout the rest of the file.
33654 Use the same switch to compute sign as base.
33655 (maybe_warn): New function.
33656 (format_directive): Factor out warnings into maybe_warn.
33657 Add debugging output. Use warn_level.
33658 (add_bytes): Use warn_level.
33659 (pass_sprintf_length::compute_format_length): Add debugging output.
33660 (try_substitute_return_value): Same.
33661 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33663 PR middle-end/78703
33664 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33665 (struct fmtresult, format_integer, format_floating): Adjust.
33666 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33667 (get_string_length, format_string,format_directive): Same.
33668 (pass_sprintf_length::compute_format_length): Same.
33669 (try_substitute_return_value): Simplify slightly.
33671 PR middle-end/78703
33672 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33673 (fmtresult::operator+=): Outlined.
33674 (struct fmtresult): Add ctors.
33675 (struct conversion_spec): Rename...
33676 (struct directive): ...to this. Add and remove data members.
33677 (directive::set_width, directive::set_precision): New functions.
33678 (format_percent): Use fmtresult ctor.
33679 (get_width_and_precision): Remove.
33680 (format_integer): Make naming changes. Avoid computing width and
33682 (format_floating): Same. Adjust indentation.
33683 (format_character, format_none): New functions.
33684 (format_string): Moved character handling to format_character.
33685 (format_directive): Remove arguments, change return type.
33686 (parse_directive): New function.
33687 (pass_sprintf_length::compute_format_length): Move directive
33688 parsing to parse_directive.
33690 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33692 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33693 (assign_assembler_name_if_needed): ... this.
33694 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33695 (assign_assembler_name_if_needed): ... this.
33696 (free_lang_data_in_cgraph): Adjust callers.
33697 * cgraphunit.c (cgraph_node::analyze): Likewise.
33698 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33701 2017-01-23 Richard Biener <rguenther@suse.de>
33703 PR tree-optimization/79088
33704 PR tree-optimization/79188
33705 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33706 resetting loop bounds after last path deletion. Reset loop
33707 bounds of the target loop, make code match the comments.
33708 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33709 Make sure loops need no fixups.
33711 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33713 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33714 exponent support with double type for first argument.
33715 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33716 type returned by __builtin_vec_extract_sig,
33717 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33718 functions from "vector int" to "vector unsigned int" or from
33719 "vector long long int" to "vector unsigned long long int".
33720 Changed type returned by __builtin_vec_extract_exp,
33721 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33722 functions from "vector int" to "vector unsigned int" or from
33723 "vector long long int" to "vector unsigned long long int".
33724 Changed return type of __builtin_vec_test_data_class,
33725 __builtin_vec_test_data_class_sp, and
33726 __builtin_vec_test_data_class_dp from "vector int" to
33727 "vector bool int" or from "vector long long int" to "vector bool
33728 long long int" and changed second argument type from "unsigned
33729 int" to "int". Added new overloaded function forms "vector float
33730 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33731 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33732 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33733 double, vector unsigned long long int)" and "vector double
33734 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33735 long int)". Changed return type of
33736 __builtin_scalar_test_data_class and
33737 __builtin_scalar_test_data_class_sp and
33738 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33739 int" and changed second argument from "unsigned int" to "int".
33740 Changed type returned by __builtin_scalar_test_neg,
33741 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33742 from "int" to "bool int". Added new overloaded function form
33743 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33744 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33745 exponent double-precision with floating point first argument.
33746 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33747 documentation of scalar_test_data_class, scalar_test_neg,
33748 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33749 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33750 vec_test_data_class built-in functions to reflect refinements in
33751 their type signatures.
33753 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33755 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33757 (aarch64_elf_asm_destructor): Likewise.
33759 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33761 PR rtl-optimization/78634
33762 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33763 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33764 * ifcvt.c (noce_try_cmove): Add missing cost check.
33766 PR rtl-optimization/71724
33767 * combine.c (if_then_else_cond): Look for situations where it is
33768 beneficial to undo the work of one of the recursive calls.
33770 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33772 PR tree-optimization/70754
33773 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33774 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33775 combined stmt before it if not NULL.
33776 (combine_chains): Process refs reversely and compute dominance point
33779 2017-01-23 Martin Liska <mliska@suse.cz>
33781 PR tree-optimization/79196
33782 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33783 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33785 (strlen_optimize_stmt): Call the renamed function.
33787 2017-01-23 Michael Matz <matz@suse.de>
33789 PR tree-optimization/78384
33790 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33792 2017-01-23 Richard Biener <rguenther@suse.de>
33794 PR tree-optimization/79186
33795 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33796 both incoming edges before moving an assert.
33798 2017-01-23 Martin Jambor <mjambor@suse.cz>
33800 * ipa-prop.c (load_from_param_1): Removed.
33801 (load_from_unmodified_param): Bits from load_from_param_1 put back
33803 (load_from_param): Removed.
33804 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33805 with stmt. Reverted back to use of load_from_unmodified_param.
33807 2017-01-23 Martin Jambor <mjambor@suse.cz>
33810 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33811 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33812 field a pointer to garbage collected vector, mark lattices and
33813 ipcp_orig_node with GTY((skip)).
33814 (ipa_get_param_count): Adjust to descriptors being a pointer.
33815 (ipa_get_param): Likewise.
33816 (ipa_get_type): Likewise.
33817 (ipa_get_param_move_cost): Likewise.
33818 (ipa_set_param_used): Likewise.
33819 (ipa_get_controlled_uses): Likewise.
33820 (ipa_set_controlled_uses): Likewise.
33821 (ipa_is_param_used): Likewise.
33822 (ipa_node_params_t): Move into garbage collector. New methods insert
33824 (ipa_node_params_sum): Annotate wth GTY(()).
33825 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33827 (ipa_load_from_parm_agg): Adjust declaration.
33828 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33829 * ipa-profile.c (ipa_profile): Likewise.
33830 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33831 (ipa_populate_param_decls): Make descriptors parameter garbage
33833 (ipa_dump_param): Adjust to descriptors being a pointer.
33834 (ipa_alloc_node_params): Likewise.
33835 (ipa_initialize_node_params): Likewise.
33836 (load_from_param_1): Make descriptors parameter garbage collected.
33837 (load_from_unmodified_param): Likewise.
33838 (load_from_param): Likewise.
33839 (ipa_load_from_parm_agg): Likewise.
33840 (ipa_node_params::~ipa_node_params): Removed.
33841 (ipa_free_all_node_params): Remove call to delete operator.
33842 (ipa_node_params_t::insert): New.
33843 (ipa_node_params_t::remove): Likewise.
33844 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33845 copy known_csts and known_contexts vectors.
33846 (ipa_read_node_info): Adjust to descriptors being a pointer.
33847 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33849 (ipcp_transform_function): Make descriptors variable garbage
33852 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33854 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33855 * config/i386/avx512dqintrin.h: Ditto.
33856 * config/i386/avx512fintrin.h: Ditto.
33857 * config/i386/i386.c: Handle new builtins.
33858 * config/i386/i386-builtin.def: Add new builtins.
33859 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33860 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33862 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33863 Martin Liska <mliska@suse.cz>
33865 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33866 * asan.c (asan_expand_poison_ifn): Support stores and use
33867 appropriate ASAN report function.
33868 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33869 * internal-fn.def (ASAN_POISON_USE): Declare.
33870 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33871 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33872 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33873 ASAN_POISON calls w/o LHS.
33874 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33875 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33876 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33877 * gimplify.c (asan_poison_variables): Add attribute
33878 use_after_scope_memory to variables that really needs to live
33880 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33881 having the attribute.
33883 2017-01-23 Martin Liska <mliska@suse.cz>
33885 * asan.c (create_asan_shadow_var): New function.
33886 (asan_expand_poison_ifn): Likewise.
33887 * asan.h (asan_expand_poison_ifn): New declaration.
33888 * internal-fn.c (expand_ASAN_POISON): Likewise.
33889 * internal-fn.def (ASAN_POISON): New builtin.
33890 * sanopt.c (pass_sanopt::execute): Expand
33891 asan_expand_poison_ifn.
33892 * tree-inline.c (copy_decl_for_dup_finish): Make function
33894 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33895 * tree-ssa.c (is_asan_mark_p): New function.
33896 (execute_update_addresses_taken): Rewrite local variables
33897 (identified just by use-after-scope as addressable) into SSA.
33899 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33901 * doc/install.texi (Specific): opensource.apple.com uses https
33902 now. Remove trailing slash.
33904 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33906 * README.Portability: Remove note on an Irix compatibility issue.
33908 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33910 * gcov.c (INCLUDE_ALGORITHM): Define.
33911 (INCLUDE_VECTOR): Define.
33912 No longer include <vector> and <algorithm> directly.
33914 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33916 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33918 * doc/invoke.texi (Code Gen Options): Ditto.
33920 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33923 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33925 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33927 rtl-optimization/79125
33928 * cprop.c (local_cprop_pass): Handle cases where we make an
33929 unconditional trap.
33931 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33935 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33936 read from, for big endian.
33938 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33940 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33941 register pauth builtins for LP64 only.
33943 2017-01-20 Marek Polacek <polacek@redhat.com>
33946 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33949 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33951 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33953 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33954 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33955 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33957 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33960 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33961 in big-endian mode when they are not a single duplicated value.
33963 2017-01-20 Richard Biener <rguenther@suse.de>
33965 * BASE-VER: Bump to 7.0.1.
33967 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33969 * omp-low.c (omplow_simd_context): New struct. Use it...
33970 (lower_rec_simd_input_clauses): ...here and...
33971 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33972 references to idx, lane, max_vf, is_simt.
33974 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33976 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33979 2017-01-20 Martin Jambor <mjambor@suse.cz>
33981 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33982 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33984 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33985 (GTFILES): Rename hsa.c to hsa-common.c.
33986 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33987 * hsa-dump.c: Likewise.
33988 * hsa-gen.c: Likewise.
33989 * hsa-regalloc.c: Likewise.
33990 * ipa-hsa.c: Likewise.
33991 * omp-expand.c: Likewise.
33992 * omp-low.c: Likewise.
33993 * toplev.c: Likewise.
33995 2017-01-20 Marek Polacek <polacek@redhat.com>
33998 * doc/invoke.texi: Document -Wduplicated-branches.
33999 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
34000 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
34001 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
34002 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
34003 return 0 only when not OEP_LEXICOGRAPHIC.
34004 (fold_build_cleanup_point_expr): Use the expression
34005 location when building CLEANUP_POINT_EXPR.
34006 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
34007 * tree.c (add_expr): Handle error_mark_node.
34009 2017-01-20 Martin Liska <mliska@suse.cz>
34012 * tree-profile.c (init_ic_make_global_vars): Do not call
34014 (gimple_init_gcov_profiler): Likewise.
34016 2017-01-20 Martin Liska <mliska@suse.cz>
34019 * cgraph.h (maybe_create_reference): Remove argument and
34021 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
34023 * ipa-cp.c (create_specialized_node): Likewise.
34024 * symtab.c (symtab_node::maybe_create_reference): Handle
34025 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
34027 2017-01-20 Martin Liska <mliska@suse.cz>
34029 * read-rtl-function.c (function_reader::create_function): Use
34030 build_decl instread of build_decl_stat.
34032 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
34034 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
34035 * config/i386/avx512dqintrin.h: Ditto.
34036 * config/i386/avx512fintrin.h: Ditto.
34037 * config/i386/i386-builtin-types.def: Add new types.
34038 * config/i386/i386.c: Handle new types.
34039 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
34040 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
34041 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
34042 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
34043 (__builtin_ia32_kshiftridi): New.
34044 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
34046 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
34050 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
34051 define to rs6000_init_stack_protect_guard.
34052 (rs6000_init_stack_protect_guard): New function.
34054 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
34055 Yunqiang Su <yunqiang.su@imgtec.com>
34057 * config.gcc (supported_defaults): Add madd4.
34058 (with_madd4): Add validation.
34059 (all_defaults): Add madd4.
34060 * config/mips/mips.opt (mmadd4): New option.
34061 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
34063 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
34065 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
34066 (ISA_HAS_FUSED_MADD4): Likewise.
34067 * doc/invoke.texi (-mmadd4): Document the new option.
34068 * doc/install.texi (--with-madd4): Document the new option.
34070 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34072 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
34073 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
34074 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
34075 (aarch64_init_pauth_hint_builtins): New.
34076 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
34077 (aarch64_expand_builtin): Expand new builtins.
34079 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34081 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
34082 * combine-stack-adj.c (no_unhandled_cfa): Handle
34083 REG_CFA_TOGGLE_RA_MANGLE.
34084 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
34085 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
34086 info for return address signing.
34087 (aarch64_expand_epilogue): Likewise.
34089 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34091 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
34092 * config/aarch64/aarch64-protos.h
34093 (aarch64_return_address_signing_enabled): New declaration.
34094 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
34096 (aarch64_expand_prologue): Sign return address before it's pushed onto
34098 (aarch64_expand_epilogue): Authenticate return address fetched from
34100 (aarch64_override_options): Sanity check for ILP32 and ISA level.
34101 (aarch64_attributes): New function attributes for "sign-return-address".
34102 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
34103 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
34104 ("*do_return"): Generate combined instructions according to key index.
34105 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
34106 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
34108 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
34109 * config/aarch64/aarch64.opt (msign-return-address=): New.
34110 * doc/extend.texi (AArch64 Function Attributes): Documents
34111 "sign-return-address=".
34112 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
34114 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
34116 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
34117 overall option summary.
34119 2017-01-19 Jiong Wang <jiong.wang@arm.com>
34121 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
34122 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
34123 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
34124 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
34126 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
34128 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
34129 -mpower9-minmax by default for -mcpu=power9.
34130 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
34131 128-bit floating point.
34133 2017-01-20 Alan Modra <amodra@gmail.com>
34135 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
34136 optimizing for size.
34138 2017-01-20 Alan Modra <amodra@gmail.com>
34141 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
34142 for strcmp and strncmp from corresponding builtin decl.
34144 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
34146 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
34147 instead of i386/rtems-64.h.
34148 * config/i386/rtems-64.h: Remove.
34150 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
34154 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
34156 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
34158 2017-01-19 Tamar Christina <tamar.christina@arm.com>
34160 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
34161 Change int to HOST_WIDE_INT.
34162 * config/aarch64/aarch64-protos.h
34163 (aarch64_simd_gen_const_vector_dup): Likewise.
34164 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
34166 2017-01-19 David Malcolm <dmalcolm@redhat.com>
34168 * langhooks-def.h (lhd_type_for_size): New decl.
34169 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
34170 * langhooks.c (lhd_type_for_size): New function, taken from
34173 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
34175 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
34176 define_bypass for CR latency.
34177 (power9-cracked-alu): Update bypass latency and remove power9-branch.
34178 (power9-alu2): Add define_bypass for CR latency.
34180 (power9-mul): Update insn latency.
34181 (power9-mul-compare): Update insn latency, bypass latency and remove
34184 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34186 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
34188 * config/aarch64/aarch64.md
34189 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
34190 aarch64_nopcrelative_literal_loads.
34191 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
34193 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
34195 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
34196 TARGET_LOONGSON_3A.
34197 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
34199 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
34202 * config.gcc (supported_defaults): Add lxc1-sxc1.
34203 (with_lxc1_sxc1): Add validation.
34204 (all_defaults): Add lxc1-sxc1.
34205 * config/mips/mips.opt (mlxc1-sxc1): New option.
34206 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
34208 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
34209 __mips_no_lxc1_sxc1.
34210 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
34211 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
34212 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
34214 2017-01-19 Richard Biener <rguenther@suse.de>
34216 PR tree-optimization/72488
34217 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
34218 sure to restore SSA info.
34219 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
34221 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
34223 PR rtl-optimization/79121
34224 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
34225 of the inner type when shifting an extended value.
34227 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34230 * symtab.c (symtab_node::equal_address_to): Fix comparing of
34231 interposable aliases.
34233 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
34236 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
34237 Use the evmergelohi instruction.
34238 (mov_si<mode>_e500_subreg4_2_le): Likewise.
34239 (mov_sitf_e500_subreg8_2_be): Likewise.
34240 (mov_sitf_e500_subreg12_2_le): Likewise.
34241 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
34242 (mov_si<mode>_e500_subreg4_2_be): Likewise.
34243 (mov_sitf_e500_subreg8_2_le): Likewise.
34244 (mov_sitf_e500_subreg12_2_be): Likewise.
34246 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34248 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
34249 attribute from vecsimple to vecperm.
34250 (altivec_vbpermq2): Likewise.
34252 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34255 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
34257 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34258 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
34259 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
34260 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
34261 case where N arg is SIZE_MAX.
34262 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
34263 (cmpstrsi): Add pattern.
34265 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34267 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
34268 __builtin_vec_revb builtins.
34269 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
34270 built-in functions to support generation of the ISA 3.0 XXBR<x>
34271 vector byte reverse instructions.
34272 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
34273 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
34274 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
34275 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
34276 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
34277 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
34278 (P9V_BUILTIN_VEC_REVB): Likewise.
34279 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
34280 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
34281 (p9_xxbrq_v16qi): Likewise.
34282 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
34283 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
34284 (p9_xxbrh_v8hi): Likewise.
34285 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
34286 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
34287 vec_revb built-in functions.
34289 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
34291 PR rtl-optimization/78952
34292 * config/i386/i386.md (any_extract): New code iterator.
34293 (*insvqi_2): Use any_extract for source operand.
34294 (*insvqi_3): Use any_shiftrt for source operand.
34296 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
34298 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
34300 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
34302 2017-01-18 Matthias Klose <doko@ubuntu.com>
34304 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
34306 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34308 * config/rs6000/altivec.h (vec_bperm): Change #define.
34309 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
34310 (altivec_vbpermq2): New define_insn.
34311 (altivec_vbpermd): Likewise.
34312 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
34313 function interface.
34314 (VBPERMD): Likewise.
34315 (VBPERM): New polymorphic function interface.
34316 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
34317 Add entries for P9V_BUILTIN_VEC_VBPERM.
34318 * doc/extend.texi: Add interfaces for vec_bperm.
34320 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34322 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
34323 first letter of error messages.
34324 (s390_resolve_overloaded_builtin): Likewise.
34325 * config/s390/s390.c (s390_expand_builtin): Likewise.
34326 (s390_invalid_arg_for_unprototyped_fn): Likewise.
34327 (s390_valid_target_attribute_inner_p): Likewise.
34328 * config/s390/s390.md ("tabort"): Likewise.
34330 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
34332 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
34333 (ISA_AVOID_DIV_HILO): New macro.
34334 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
34335 (ISA_HAS_DDIV): Likewise.
34337 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34339 * doc/invoke.texi (fabi-version): Correct number of occurrences.
34341 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34343 * doc/invoke.texi (fabi-version): Spelling fix.
34345 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34348 * doc/invoke.texi (fabi-version): Mention mangling fix for
34351 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34354 * doc/invoke.texi (fabi-version): Document discriminator mangling.
34356 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
34359 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
34360 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
34362 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
34364 (stack_protect_test): Ditto.
34365 * config/rs6000/rs6000.opt (mstack-protector-guard=,
34366 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
34368 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
34369 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
34370 -mstack-protector-guard-offset=.
34371 (RS/6000 and PowerPC Options): Ditto.
34373 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34375 * config/i386/i386.h (MASK_CLASS_P): New define.
34376 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
34377 there are no registers from different register sets also when
34378 mask registers are used. Update function comment.
34379 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
34380 to (*k/*r) and (*k/*km) alternatives.
34382 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
34384 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
34385 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
34386 (EH_RETURN_HANDLER_RTX): New define.
34387 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
34388 Force frame pointer in EH return functions.
34389 (aarch64_expand_epilogue): Add barrier for eh_return.
34390 (aarch64_final_eh_return_addr): Remove.
34391 (aarch64_eh_return_handler_rtx): New function.
34392 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
34394 (aarch64_eh_return_handler_rtx): New prototype.
34396 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34398 * config/rs6000/altivec.h (vec_rlmi): New #define.
34399 (vec_vrlnm): Likewise.
34400 (vec_rlnm): Likewise.
34401 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
34402 (UNSPEC_VRLNM): Likewise.
34403 (VIlong): New mode iterator.
34404 (altivec_vrl<VI_char>mi): New define_insn.
34405 (altivec_vrl<VI_char>nm): Likewise.
34406 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
34408 (VRLDNM): Likewise.
34409 (RLNM): New polymorphic function entry.
34410 (VRLWMI): New monomorphic function entry.
34411 (VRLDMI): Likewise.
34412 (RLMI): New polymorphic function entry.
34413 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
34414 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
34415 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
34418 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34421 * dwarf2out.c (field_byte_offset): Restore the
34422 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
34423 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
34424 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
34427 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34430 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
34432 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34435 * dwarf2out.c (add_data_member_location_attribute): For constant
34436 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
34437 instead of DW_AT_data_member_location, DW_AT_bit_offset and
34438 DW_AT_byte_size attributes.
34440 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34442 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
34443 after forcing to constant memory when the code model is medium.
34445 2017-01-17 Julia Koval <julia.koval@intel.com>
34448 * config/i386/avx512fintrin.h
34449 (_mm512_i32gather_ps): Change __addr type to void const*.
34450 (_mm512_mask_i32gather_ps): Ditto.
34451 (_mm512_i32gather_pd): Ditto.
34452 (_mm512_mask_i32gather_pd): Ditto.
34453 (_mm512_i64gather_ps): Ditto.
34454 (_mm512_mask_i64gather_ps): Ditto.
34455 (_mm512_i64gather_pd): Ditto.
34456 (_mm512_mask_i64gather_pd): Ditto.
34457 (_mm512_i32gather_epi32): Ditto.
34458 (_mm512_mask_i32gather_epi32): Ditto.
34459 (_mm512_i32gather_epi64): Ditto.
34460 (_mm512_mask_i32gather_epi64): Ditto.
34461 (_mm512_i64gather_epi32): Ditto.
34462 (_mm512_mask_i64gather_epi32): Ditto.
34463 (_mm512_i64gather_epi64): Ditto.
34464 (_mm512_mask_i64gather_epi64): Ditto.
34465 (_mm512_i32scatter_ps): Change __addr type to void*.
34466 (_mm512_mask_i32scatter_ps): Ditto.
34467 (_mm512_i32scatter_pd): Ditto.
34468 (_mm512_mask_i32scatter_pd): Ditto.
34469 (_mm512_i64scatter_ps): Ditto.
34470 (_mm512_mask_i64scatter_ps): Ditto.
34471 (_mm512_i64scatter_pd): Ditto.
34472 (_mm512_mask_i64scatter_pd): Ditto.
34473 (_mm512_i32scatter_epi32): Ditto.
34474 (_mm512_mask_i32scatter_epi32): Ditto.
34475 (_mm512_i32scatter_epi64): Ditto.
34476 (_mm512_mask_i32scatter_epi64): Ditto.
34477 (_mm512_i64scatter_epi32): Ditto.
34478 (_mm512_mask_i64scatter_epi32): Ditto.
34479 (_mm512_i64scatter_epi64): Ditto.
34480 (_mm512_mask_i64scatter_epi64): Ditto.
34481 * config/i386/avx512pfintrin.h
34482 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
34483 (_mm512_mask_prefetch_i32gather_ps): Ditto.
34484 (_mm512_mask_prefetch_i64gather_pd): Ditto.
34485 (_mm512_mask_prefetch_i64gather_ps): Ditto.
34486 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
34487 (_mm512_prefetch_i32scatter_ps): Ditto.
34488 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
34489 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
34490 (_mm512_prefetch_i64scatter_pd): Ditto.
34491 (_mm512_prefetch_i64scatter_ps): Ditto.
34492 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
34493 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
34494 * config/i386/avx512vlintrin.h
34495 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
34496 (_mm_mmask_i32gather_ps): Ditto.
34497 (_mm256_mmask_i32gather_pd): Ditto.
34498 (_mm_mmask_i32gather_pd): Ditto.
34499 (_mm256_mmask_i64gather_ps): Ditto.
34500 (_mm_mmask_i64gather_ps): Ditto.
34501 (_mm256_mmask_i64gather_pd): Ditto.
34502 (_mm_mmask_i64gather_pd): Ditto.
34503 (_mm256_mmask_i32gather_epi32): Ditto.
34504 (_mm_mmask_i32gather_epi32): Ditto.
34505 (_mm256_mmask_i32gather_epi64): Ditto.
34506 (_mm_mmask_i32gather_epi64): Ditto.
34507 (_mm256_mmask_i64gather_epi32): Ditto.
34508 (_mm_mmask_i64gather_epi32): Ditto.
34509 (_mm256_mmask_i64gather_epi64): Ditto.
34510 (_mm_mmask_i64gather_epi64): Ditto.
34511 (_mm256_i32scatter_ps): Change __addr type to void*.
34512 (_mm256_mask_i32scatter_ps): Ditto.
34513 (_mm_i32scatter_ps): Ditto.
34514 (_mm_mask_i32scatter_ps): Ditto.
34515 (_mm256_i32scatter_pd): Ditto.
34516 (_mm256_mask_i32scatter_pd): Ditto.
34517 (_mm_i32scatter_pd): Ditto.
34518 (_mm_mask_i32scatter_pd): Ditto.
34519 (_mm256_i64scatter_ps): Ditto.
34520 (_mm256_mask_i64scatter_ps): Ditto.
34521 (_mm_i64scatter_ps): Ditto.
34522 (_mm_mask_i64scatter_ps): Ditto.
34523 (_mm256_i64scatter_pd): Ditto.
34524 (_mm256_mask_i64scatter_pd): Ditto.
34525 (_mm_i64scatter_pd): Ditto.
34526 (_mm_mask_i64scatter_pd): Ditto.
34527 (_mm256_i32scatter_epi32): Ditto.
34528 (_mm256_mask_i32scatter_epi32): Ditto.
34529 (_mm_i32scatter_epi32): Ditto.
34530 (_mm_mask_i32scatter_epi32): Ditto.
34531 (_mm256_i32scatter_epi64): Ditto.
34532 (_mm256_mask_i32scatter_epi64): Ditto.
34533 (_mm_i32scatter_epi64): Ditto.
34534 (_mm_mask_i32scatter_epi64): Ditto.
34535 (_mm256_i64scatter_epi32): Ditto.
34536 (_mm256_mask_i64scatter_epi32): Ditto.
34537 (_mm_i64scatter_epi32): Ditto.
34538 (_mm_mask_i64scatter_epi32): Ditto.
34539 (_mm256_i64scatter_epi64): Ditto.
34540 (_mm256_mask_i64scatter_epi64): Ditto.
34541 (_mm_i64scatter_epi64): Ditto.
34542 (_mm_mask_i64scatter_epi64): Ditto.
34543 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
34544 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
34545 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
34546 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
34547 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
34548 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
34549 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
34550 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
34551 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
34552 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
34553 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
34554 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
34555 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
34556 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
34557 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
34558 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
34559 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
34560 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
34561 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
34562 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
34563 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
34564 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
34565 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
34566 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
34567 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
34568 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
34569 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
34570 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
34571 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
34572 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
34573 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
34574 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
34575 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
34576 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
34577 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
34578 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
34579 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
34580 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
34581 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
34582 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
34583 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
34584 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
34585 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
34586 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
34587 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
34588 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
34589 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
34590 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
34591 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
34592 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
34593 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
34594 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
34595 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
34596 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
34597 definitions accordingly.
34599 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
34600 Kuan-Lin Chen <kuanlinchentw@gmail.com>
34603 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
34606 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
34609 * ira-conflicts.c (ira_build_conflicts): Update total conflict
34610 hard regs for inner regno.
34612 2017-01-17 Martin Liska <mliska@suse.cz>
34615 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
34616 assumption and add comment.
34618 2017-01-17 Nathan Sidwell <nathan@acm.org>
34620 * ipa-visibility.c (localize_node): New function, broken out of ...
34621 (function_and_variable_visibility): ... here. Call it.
34623 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34625 PR middle-end/77445
34626 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
34627 correctly set frequency of oudgoing edge.
34628 (duplicate_thread_path): Fix profile updating.
34630 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34633 * configure.ac: Add GCC_BASE_VER.
34634 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
34635 version from BASE-VER file.
34636 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
34637 (gcc.o): Depend on $(BASEVER).
34638 * common.opt (dumpfullversion): New option.
34639 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
34640 * doc/invoke.texi: Document -dumpfullversion.
34641 * doc/install.texi: Document --with-gcc-major-version-only.
34642 * configure: Regenerated.
34644 2017-01-17 Richard Biener <rguenther@suse.de>
34646 PR tree-optimization/71433
34647 * tree-vrp.c (register_new_assert_for): Merge same asserts
34648 on all incoming edges.
34649 (process_assert_insertions_for): Handle insertions at the
34652 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34654 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34655 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34657 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34660 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34663 2017-01-17 Alan Modra <amodra@gmail.com>
34666 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34667 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34668 symbolic stack limit when pic.
34670 2017-01-16 Martin Sebor <msebor@redhat.com>
34672 PR tree-optimization/78608
34673 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34675 2017-01-16 Jeff Law <law@redhat.com>
34678 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34679 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34680 for several include directories that may be relative to sysroot.
34681 * config/i386/x-mingw32 (gplus_includedir): Define.
34682 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34683 (native_system_includedir): Likewise.
34684 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34685 override if TARGET_SYSTEM_ROOT is defined.
34686 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34688 PR tree-optimization/79090
34689 PR tree-optimization/33562
34690 PR tree-optimization/61912
34691 PR tree-optimization/77485
34692 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34693 and computed trims into the dump file.
34695 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34697 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34699 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34702 * gimplify.c (gimplify_init_constructor): If want_value and
34703 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34707 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34708 sequence. Formatting fixes.
34709 (doloop_optimize): Formatting fixes.
34712 * gcc.c (debug_level_greater_than_spec_func): New function.
34713 (static_spec_functions): Add debug-level-gt spec function.
34714 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34716 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34717 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34718 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34719 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34720 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34721 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34723 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34725 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34726 QImode fixups to general and mask registers only.
34728 2017-01-16 Carl Love <cel@us.ibm.com>
34730 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34731 for built-in functions
34732 vector signed char vec_nabs (vector signed char)
34733 vector signed short vec_nabs (vector signed short)
34734 vector signed int vec_nabs (vector signed int)
34735 vector signed long long vec_nabs (vector signed long long)
34736 vector float vec_nabs (vector float)
34737 vector double vec_nabs (vector double)
34738 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34740 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34741 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34742 * doc/extend.texi: Update the documentation file for the new built-in
34745 2017-01-16 Martin Sebor <msebor@redhat.com>
34747 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34750 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34752 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34753 UNSPEC_VSX__XXSPLTD to require special splat handling.
34755 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34758 * system.h: Poison strndup.
34760 2017-01-16 Alan Modra <amodra@gmail.com>
34763 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34766 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34768 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34770 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34772 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34773 call recog here. Assert that INSN_CODE (insn) is non-negative.
34775 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34778 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34780 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34781 in the currently scheduled RTL fragment.
34783 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34785 PR rtl-optimization/78751
34786 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34789 2017-01-14 Jeff Law <law@redhat.com>
34791 PR tree-optimization/79090
34792 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34793 variable length stores.
34794 (compute_trims): Delete dead assignment to *trim_tail.
34795 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34798 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34800 PR rtl-optimization/78626
34801 PR rtl-optimization/78727
34802 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34803 of a block, and split such blocks after everything else is finished.
34805 2017-01-14 Alan Modra <amodra@gmail.com>
34808 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34809 target legitimate_combined_insn.
34810 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34811 (rs6000_legitimate_combined_insn): New function.
34812 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34814 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34815 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34816 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34818 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34820 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34822 2017-01-13 Jeff Law <law@redhat.com>
34824 PR tree-optimization/33562
34825 PR tree-optimization/61912
34826 PR tree-optimization/77485
34827 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34829 (delete_dead_assignment): Likewise.
34830 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34831 statement to delete_dead_call and delete_dead_assignment.
34833 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34836 * substring-locations.c (format_warning_va): Strengthen case 1 so
34837 that both endpoints of the substring must be within the format
34838 range for just the substring to be printed.
34840 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34842 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34843 * config/i386/i386.c (ix86_target_string): Add missing options
34844 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34845 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34846 flags_other and ix86_target_other to flags2_other. Display unknown
34848 (ix86_valid_target_attribute_inner_p): Add missing options and
34849 reorder options by implied ISAs, as in ix86_target_string.
34851 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34853 * hash-table.h (hash_table::too_empty_p): New function.
34854 (hash_table::expand): Use it.
34855 (hash_table::traverse): Likewise.
34856 (hash_table::empty_slot): Use sizeof (value_type) instead of
34857 sizeof (PTR) to convert bytes to elements. Shrink the table
34858 if the current size is excessive for the current number of
34861 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34863 * ira-costs.c (record_reg_classes): Break from the inner loop
34864 early once alt_fail is known to be true. Update outer loop
34865 handling accordingly.
34867 2017-01-13 Jeff Law <law@redhat.com>
34869 * tree-ssa-dse.c (decrement_count): New function.
34870 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34871 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34872 when we know the partially dead statement is a mem* function.
34874 PR tree-optimization/61912
34875 PR tree-optimization/77485
34876 * tree-ssa-dse.c: Include expr.h.
34877 (maybe_trim_constructor_store): New function.
34878 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34880 PR tree-optimization/33562
34881 PR tree-optimization/61912
34882 PR tree-optimization/77485
34883 * doc/invoke.texi: Document new dse-max-object-size param.
34884 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34885 * tree-ssa-dse.c: Include params.h.
34886 (dse_store_status): New enum.
34887 (initialize_ao_ref_for_dse): New, partially extracted from
34889 (valid_ao_ref_for_dse, normalize_ref): New.
34890 (setup_live_bytes_from_ref, compute_trims): Likewise.
34891 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34892 (maybe_trim_partially_dead_store): Likewise.
34893 (maybe_trim_complex_store): Likewise.
34894 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34895 Track what bytes live from the original store. Return tri-state
34896 for dead, partially dead or live.
34897 (dse_dom_walker): Add constructor, destructor and new private members.
34898 (delete_dead_call, delete_dead_assignment): New extracted from
34900 (dse_optimize_stmt): Make a member of dse_dom_walker.
34901 Use initialize_ao_ref_for_dse.
34903 PR tree-optimization/33562
34904 PR tree-optimization/61912
34905 PR tree-optimization/77485
34906 * sbitmap.h (bitmap_count_bits): Prototype.
34907 (bitmap_clear_range, bitmap_set_range): Likewise.
34908 * sbitmap.c (bitmap_clear_range): New function.
34909 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34911 2017-01-13 Martin Liska <mliska@suse.cz>
34914 * function.c (set_cfun): Add new argument force.
34915 * function.h (set_cfun): Likewise.
34916 * ipa-inline-transform.c (inline_call): Use the function when
34917 strict alising from is dropped for function we inline to.
34919 2017-01-13 Richard Biener <rguenther@suse.de>
34921 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34922 for dumping GIMPLE INTEGER_CSTs.
34924 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34926 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34927 to 201112L since C++17.
34929 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34932 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34933 if -fsanitize=kernel-address is present.
34935 2017-01-13 Richard Biener <rguenther@suse.de>
34937 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34938 as _Literal ( type ) number in case usual suffixes do not
34939 preserve all information.
34941 2017-01-13 Richard Biener <rguenther@suse.de>
34943 PR tree-optimization/77283
34944 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34945 and ssa-iterators.h.
34946 (is_feasible_trace): Implement a cost model based on joiner
34949 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34952 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34953 char or short to __float128/_Float128 directly.
34955 2017-01-12 Martin Sebor <msebor@redhat.com>
34957 to -Wformat-overflow.
34958 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34959 (min_bytes_remaining): Same.
34960 (get_string_length): Same.
34961 (format_string): Same.
34962 (format_directive): Same.
34964 (pass_sprintf_length::handle_gimple_call): Same.
34966 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34968 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34969 info.nowrite calls with no lhs that can't throw. Return bool
34970 whether gsi_remove has been called or not.
34971 (pass_sprintf_length::handle_gimple_call): Return bool whether
34972 try_substitute_return_value called gsi_remove. Formatting fix.
34973 (pass_sprintf_length::execute): Don't use gsi_remove if
34974 handle_gimple_call returned true.
34977 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34978 be removed due to side-effects, don't remove following barrier nor
34979 turn the successor edge into fallthru edge.
34981 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34984 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34985 element-reversing loads and stores as not swappable.
34987 2017-01-12 Nathan Sidwell <nathan@acm.org>
34988 Nicolai Stange <nicstange@gmail.com>
34990 * combine.c (try_combine): Don't ignore result of overlap checking
34991 loop. Combine overlap & asm check into single loop.
34993 2017-01-12 Richard Biener <rguenther@suse.de>
34995 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34996 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34998 2017-01-12 Richard Biener <rguenther@suse.de>
35000 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
35001 and TS_TARGET_OPTION directly derive from TS_BASE.
35002 * tree-core.h (tree_optimization_option): Derive from tree_base.
35003 (tree_target_option): Likewise.
35005 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
35007 * config/i386/i386.c (memory_address_length): Increase len
35008 only when rip_relative_addr_p returns false.
35010 2017-01-11 Julia Koval <julia.koval@intel.com>
35012 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
35013 (OPTION_MASK_ISA_SGX_SET): New.
35014 (ix86_handle_option): Handle OPT_msgx.
35015 * config.gcc: Added sgxintrin.h.
35016 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
35017 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
35018 * config/i386/i386.c (ix86_target_string): Add -msgx.
35020 (ix86_option_override_internal): Handle new options.
35021 (ix86_valid_target_attribute_inner_p): Add sgx.
35022 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
35023 * config/i386/i386.opt: Add msgx.
35024 * config/i386/sgxintrin.h: New file.
35025 * config/i386/x86intrin.h: Add sgxintrin.h.
35027 2017-01-11 Jakub Jelinek <jakub@redhat.com>
35030 * fold-const.c (maybe_nonzero_address): Return 1 for function
35032 (tree_single_nonzero_warnv_p): Don't handle function local objects
35036 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
35039 2017-01-11 David Malcolm <dmalcolm@redhat.com>
35042 * opts.c: Include "spellcheck.h"
35043 (struct string_fragment): New struct.
35044 (struct edit_distance_traits<const string_fragment &>): New
35046 (get_closest_sanitizer_option): New function.
35047 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
35049 2017-01-11 Jakub Jelinek <jakub@redhat.com>
35051 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
35053 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
35054 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
35055 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
35056 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
35057 for initial die_offset if dwarf_split_debug_info.
35058 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
35059 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
35061 (output_skeleton_debug_sections): Formatting fix. Use
35062 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
35063 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
35065 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
35067 * config/arm/cortex-a53.md: Add bypasses for
35068 cortex_a53_r2f_cvt.
35069 (cortex_a53_r2f): Only use for transfers.
35070 (cortex_a53_f2r): Likewise.
35071 (cortex_a53_r2f_cvt): Add reservation for conversions.
35072 (cortex_a53_f2r_cvt): Likewise.
35074 2017-01-11 Tamar Christina <tamar.christina@arm.com>
35076 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
35077 to all inlined functions, change static to extern.
35079 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
35082 * config/arm/arm.c (legitimize_pic_address): Handle reference to
35084 (arm_assemble_integer): Likewise.
35086 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
35088 * config.gcc: Use new awk script to check CPU, FPU and architecture
35089 parameters for --with-... options.
35090 * config/arm/parsecpu.awk: New file
35091 * config/arm/arm-cpus.in: New file.
35092 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
35094 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
35096 * config/arm/t-arm: Update dependency rules.
35097 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
35098 of processing .def files.
35099 * config/arm/genopt.sh: Deleted.
35100 * config/arm/gentune.sh: Deleted.
35101 * config/arm/arm-cores.def: Deleted.
35102 * config/arm/arm-arches.def: Deleted.
35103 * config/arm/arm-fpus.def: Deleted.
35104 * config/arm/arm-tune.md: Regenerated.
35105 * config/arm/arm-tables.opt: Regenerated.
35106 * config/arm/arm-cpu.h: New generated file.
35107 * config/arm/arm-cpu-data.h: New generated file.
35108 * config/arm/arm-cpu-cdata.h: New generated file.
35110 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
35113 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
35115 (input_varpool_node): Unpack dynamically_initialized bit.
35117 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
35119 PR rtl-optimization/79032
35120 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
35121 the alignment of the adjusted memory reference against that of MODE,
35122 instead of the alignment of the original memory reference.
35124 2017-01-11 Martin Jambor <mjambor@suse.cz>
35126 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
35128 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
35129 decorated functions.
35131 2017-01-11 Richard Biener <rguenther@suse.de>
35133 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
35134 set range/nonnull info for PHI results. Do not set it on
35135 stmts marked for removal.
35137 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
35139 * expr.c (store_field): In the bitfield case, fetch the return value
35140 from the registers before applying a single big-endian adjustment.
35141 Always do a final load for a BLKmode value not larger than a word.
35143 2017-01-10 David Malcolm <dmalcolm@redhat.com>
35146 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35147 that we correctly handle column numbers greater than
35148 LINE_MAP_MAX_COLUMN_NUMBER.
35150 2017-01-10 Martin Sebor <msebor@redhat.com>
35152 PR middle-end/78245
35153 * gimple-ssa-sprintf.c (get_destination_size): Call
35154 {init,fini}object_sizes.
35155 * tree-object-size.c (addr_object_size): Adjust.
35156 (pass_through_call): Adjust.
35157 (pass_object_sizes::execute): Adjust.
35158 * tree-object-size.h (fini_object_sizes): Declare.
35160 2017-01-10 Martin Sebor <msebor@redhat.com>
35162 PR tree-optimization/78775
35163 * builtins.c (get_size_range): Move...
35164 * calls.c: ...to here.
35165 (alloc_max_size): Accept zero argument.
35166 (operand_signed_p): Remove.
35167 (maybe_warn_alloc_args_overflow): Call get_size_range.
35168 * calls.h (get_size_range): Declare.
35170 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
35172 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
35173 from TI's devices.csv file as of September 2016.
35174 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
35176 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
35178 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
35179 * doc/invoke.texi: Likewise.
35180 * doc/md.texi: Likewise.
35181 * doc/objc.texi: Likewise.
35183 2017-01-10 Joshua Conner <joshconner@google.com>
35185 * config/arm/fuchsia-elf.h: New file.
35186 * config/fuchsia.h: New file.
35187 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
35188 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
35190 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
35192 2016-01-10 Richard Biener <rguenther@suse.de>
35194 PR tree-optimization/79034
35195 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35196 Propagate out degenerate PHIs in the joiner.
35198 2017-01-10 Martin Liska <mliska@suse.cz>
35200 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
35201 (sort_congruence_classes_by_decl_uid): Likewise.
35202 (sort_congruence_class_groups_by_decl_uid): Likewise.
35203 (sem_item_optimizer::merge_classes): Sort class, groups in these
35204 classes and members in the groups by DECL_UID of declarations.
35205 This would make merge operations stable.
35207 2017-01-10 Martin Liska <mliska@suse.cz>
35209 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
35210 usage of m_classes_vec.
35211 (sem_item_optimizer::~sem_item_optimizer): Likewise.
35212 (sem_item_optimizer::get_group_by_hash): Likewise.
35213 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
35214 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
35215 (sem_item_optimizer::verify_classes): Likewise.
35216 (sem_item_optimizer::process_cong_reduction): Likewise.
35217 (sem_item_optimizer::dump_cong_classes): Likewise.
35218 (sem_item_optimizer::merge_classes): Likewise.
35219 * ipa-icf.h (congruence_class_hash): Rename from
35220 congruence_class_group_hash. Remove declaration of m_classes_vec.
35222 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
35224 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
35225 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
35226 * config.gcc: Add avx512vpopcntdqintrin.h.
35227 * config/i386/avx512vpopcntdqintrin.h: New.
35228 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
35229 * config/i386/i386-builtin-types.def: Add new types.
35230 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
35231 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
35232 __builtin_ia32_vpopcountq_v8di_mask): New.
35233 * config/i386/i386-c.c (ix86_target_macros_internal): Define
35234 __AVX512VPOPCNTDQ__.
35235 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
35236 (PTA_AVX512VPOPCNTDQ): Define.
35237 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
35238 TARGET_AVX512VPOPCNTDQ_P): Define.
35239 * config/i386/i386.opt: Add mavx512vpopcntdq.
35240 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
35241 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
35243 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35245 PR middle-end/77484
35246 * predict.def (PRED_CALL): Set to 67.
35248 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
35250 * expr.c (store_field): In the bitfield case, if the value comes from
35251 a function call and is of an aggregate type returned in registers, do
35252 not modify the field mode; extract the value in all cases if the mode
35253 is BLKmode and the size is not larger than a word.
35255 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
35258 * config/i386/cpuid.h: Fix undefined behavior.
35260 2017-01-04 Jeff Law <law@redhat.com>
35262 PR tree-optimization/79007
35263 PR tree-optimization/67955
35264 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
35265 conservative for pt.null when flag_non_call_exceptions is on.
35267 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35269 PR translation/79019
35270 PR translation/79020
35271 * params.def (PARAM_INLINE_MIN_SPEEDUP,
35272 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
35273 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
35275 * config/avr/avr.opt (maccumulate-args): Likewise.
35276 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
35277 * common.opt (freport-bug): Likewise.
35278 * cif-code.def (CIF_FINAL_ERROR): Likewise.
35279 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
35280 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
35281 translatable string.
35282 * config/i386/i386.c (function_value_32): Likewise.
35283 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
35284 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
35286 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
35287 * common/config/msp430/msp430-common.c (msp430_handle_option):
35289 * symtab.c (symtab_node::verify_base): Likewise.
35290 * opts.c (set_debug_level): Likewise.
35291 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
35292 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
35293 missing whitespace to translatable strings.
35294 * config/avr/avr.md (bswapsi2): Fix typo in comment.
35295 * config/sh/superh.h: Likewise.
35296 * config/i386/xopintrin.h: Likewise.
35297 * config/i386/znver1.md: Likewise.
35298 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
35299 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
35300 * double-int.h (struct double_int): Likewise.
35301 * double-int.c (div_and_round_double): Likewise.
35302 * wide-int.cc: Likewise.
35303 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
35304 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
35305 * cfgcleanup.c (crossjumps_occured): Renamed to ...
35306 (crossjumps_occurred): ... this.
35307 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
35310 PR tree-optimization/78899
35311 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
35312 returning bool return struct loop *, NULL for failure and the new
35314 (versionable_outer_loop_p): Don't version outer loop if it has
35315 dont_vectorized bit set.
35316 (tree_if_conversion): When versioning outer loop, ensure
35317 tree_if_conversion is performed also on the inner loop of the
35318 non-vectorizable outer loop copy.
35319 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
35320 LOOP_VECTORIZED in inner loop of the scalar outer loop and
35321 prevent vectorization of it.
35322 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
35323 the outer loop vectorization of the non-scalar version is attempted
35324 before vectorization of the inner loop in scalar version. If
35325 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
35326 vectorization of its inner loop.
35327 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
35328 has 2 inner loops, rename also on edges from bb whose single pred
35329 is outer_loop->header. Fix typo in function comment.
35331 2017-01-09 Martin Sebor <msebor@redhat.com>
35334 * asan.c (asan_emit_stack_protection): Increase local buffer size
35335 to avoid snprintf truncation warning.
35337 2017-01-09 Andrew Pinski <apinski@cavium.com>
35339 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
35340 to reference thunderx2t99 for the tuning structure
35341 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
35343 (thunderx2t99_extra_costs): This.
35344 * config/aarch64/aarch64-tune.md: Regenerate.
35345 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
35346 (vulcan_addrcost_table): This.
35347 (vulcan_regmove_cost): Rename to ...
35348 (thunderx2t99_regmove_cost): This.
35349 (vulcan_vector_cost): Rename to ...
35350 (thunderx2t99_vector_cost): this.
35351 (vulcan_branch_cost): Rename to ...
35352 (thunderx2t99_branch_cost): This.
35353 (vulcan_tunings): Rename to ...
35354 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
35355 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
35357 2017-01-09 Martin Jambor <mjambor@suse.cz>
35361 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
35362 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
35363 (propagate_vr_accross_jump_function): Use the above function for all
35364 value range computations for pass-through jump functions and type
35365 converasion from explicit value range values.
35366 (ipcp_propagate_stage): Do not attempt to deduce types of formal
35367 parameters from TYPE_ARG_TYPES.
35368 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
35369 (ipa_write_node_info): Stream type of the actual argument.
35370 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
35372 2017-01-09 Martin Liska <mliska@suse.cz>
35375 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
35376 (lookup_compiler): Do not show error message with have_E.
35378 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35380 PR tree-optimization/78938
35381 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
35382 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
35383 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
35384 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
35387 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35389 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
35392 2017-01-09 Richard Biener <rguenther@suse.de>
35394 PR tree-optimization/78997
35395 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
35396 name condition properly.
35398 2017-01-09 Richard Biener <rguenther@suse.de>
35401 * dwarf2out.c (is_cxx): New overload with context.
35402 (is_naming_typedef_decl): Use it.
35404 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35406 * invoke.texi (Option Summary): Correct spacing in option lists
35407 and add line breaks to fix over-long lines.
35409 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35411 PR middle-end/17660
35413 * extend.texi (Common Variable Attributes): Add xref to GCC
35414 Internals manual to explain mode attribute keywords.
35416 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35419 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
35420 and Preprocessor Options.
35421 (Options for Linking): Document -pthread here....
35422 (RS/6000 and PowerPC Options): ...not here.
35423 (Solaris 2 Options): ...or here.
35424 * doc/cppopts.texi: Document -pthread.
35426 2017-01-08 Martin Sebor <msebor@redhat.com>
35428 PR middle-end/77708
35429 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
35430 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
35431 New member functions.
35432 (format_directive): Used them.
35434 (pass_sprintf_length::handle_gimple_call): Same.
35435 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
35436 to avoid truncation for any argument.
35437 (extract_affine_mul): Same.
35438 * tree.c (get_file_function_name): Same.
35440 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35442 PR middle-end/77484
35443 * predict.def (PRED_INDIR_CALL): Set to 86.
35445 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35447 PR preprocessor/54124
35448 * doc/cppopts.texi: Reformat -d subtable to list the full name
35449 of the options. Add cross-reference to the docs for the general
35450 compiler -d options.
35451 * doc/invoke.texi (Developer Options): Add cross-reference to the
35452 preprocessor-specific -d option documentation.
35454 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35456 PR preprocessor/13498
35457 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
35458 redudant material, and reflect new command-line options.
35459 (System Headers): Likewise.
35461 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35463 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
35464 -isystem, and -idirafter. Copy-edit.
35465 * doc/cppopts.texi: Copy-edit. Remove contradiction about
35466 default for -ftrack-macro-expansion. Delete obsolete and
35467 badly-formatted implementation details about -fdebug-cpp output.
35468 * doc/cppwarnopts.texi: Copy-edit.
35470 2017-01-07 David Malcolm <dmalcolm@redhat.com>
35473 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35474 that the transition from a max line width >= 1<<10 to narrower
35475 lines works correctly.
35477 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
35479 * doc/options.texi (PerFunction): New.
35480 * opt-functions.awk (switch_flags): Map both Optimization and
35481 PerFunction to CL_OPTIMIZATION.
35482 * opth-gen.awk: Test for PerFunction flag along with
35484 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
35485 it only when the latter is present. Skip those that don't in
35486 the hash function generator.
35487 * common.opt (fvar-tracking): Mark as PerFunction instead of
35489 (fvar-tracking-assignments): Likewise.
35490 (fvar-tracking-assignments-toggle): Likewise.
35491 (fvar-tracking-uninit): Likewise.
35493 2017-01-07 Jakub Jelinek <jakub@redhat.com>
35495 PR translation/79018
35496 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
35499 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
35502 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
35503 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
35504 TARGET_LONG_JUMP_TABLE_OFFSETS.
35505 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
35506 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
35507 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
35508 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
35509 * config/m68k/m68k.md (tablejump expander): Likewise.
35510 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
35511 TARGET_LONG_JUMP_TABLE_OFFSETS.
35512 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
35513 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
35515 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
35516 David Holsgrove <david.holsgrove@xilinx.com>
35518 * common/config/microblaze/microblaze-common.c
35519 (TARGET_EXCEPT_UNWIND_INFO): Remove.
35520 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
35522 * config/microblaze/microblaze.c (microblaze_must_save_register)
35523 (microblaze_expand_epilogue, microblaze_return_addr): Handle
35525 (microblaze_eh_return): New function.
35526 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
35527 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
35528 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
35529 * config/microblaze/microblaze.md (eh_return): New pattern.
35531 2017-01-06 Jakub Jelinek <jakub@redhat.com>
35533 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
35534 GCC_DIAGNOSTIC_STRINGIFY): Define.
35536 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
35538 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35540 * config/arm/arm.md (<mcrr>): New.
35542 * config/arm/arm.c (arm_arch5te): New.
35543 (arm_option_override): Set arm_arch5te.
35544 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
35546 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
35547 (arm_mcrr_qualifiers): ... this. New.
35548 (MRRC_QUALIFIERS): Define to...
35549 (arm_mrrc_qualifiers): ... this. New.
35550 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
35552 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
35553 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
35554 (MRRCI, mrrc, MRRC): New.
35555 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
35556 VUNSPEC_MRRC2): New.
35558 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35560 * config/arm/arm.md (<mcr>): New.
35562 * config/arm/arm.c (arm_coproc_builtin_available): Add
35563 support for mcr, mrc, mcr2 and mrc2.
35564 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
35565 (arm_mcr_qualifiers): ... this. New.
35566 (MRC_QUALIFIERS): Define to ...
35567 (arm_mrc_qualifiers): ... this. New.
35568 (MCR_QUALIFIERS): Define to ...
35569 (arm_mcr_qualifiers): ... this. New.
35570 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
35572 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
35573 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
35574 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
35575 VUNSPEC_MRC2): New.
35577 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35579 * config/arm/arm.md (*ldc): New.
35583 * config/arm/arm.c (arm_coproc_builtin_available): Add
35584 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
35585 (arm_coproc_ldc_stc_legitimate_address): New.
35586 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
35587 'qualifier_const_pointer'.
35588 (LDC_QUALIFIERS): Define to...
35589 (arm_ldc_qualifiers): ... this. New.
35590 (STC_QUALIFIERS): Define to...
35591 (arm_stc_qualifiers): ... this. New.
35592 * config/arm/arm-protos.h
35593 (arm_coproc_ldc_stc_legitimate_address): New.
35594 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
35595 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
35596 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
35597 stc2, stcl, stc2l): New.
35598 * config/arm/constraints.md (Uz): New.
35599 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
35600 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
35601 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
35602 VUNSPEC_STC2L): New.
35604 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35606 * config/arm/arm.md (<cdp>): New.
35607 * config/arm/arm.c (neon_const_bounds): Rename this ...
35608 (arm_const_bounds): ... this.
35609 (arm_coproc_builtin_available): New.
35610 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
35611 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
35612 (CDP_QUALIFIERS): Define to...
35613 (arm_cdp_qualifiers): ... this. New.
35615 (arm_expand_builtin_args): Add case for 6 arguments.
35616 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
35617 (arm_const_bounds): ... this.
35618 (arm_coproc_builtin_available): New.
35619 * config/arm/arm_acle.h (__arm_cdp): New.
35621 * config/arm/arm_acle_builtins.def (cdp): New.
35623 * config/arm/iterators.md (CDPI,CDP,cdp): New.
35624 * config/arm/neon.md: Rename all 'neon_const_bounds' to
35625 'arm_const_bounds'.
35626 * config/arm/types.md (coproc): New.
35627 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
35628 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
35629 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
35630 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
35632 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35634 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
35635 (UBINOP_QUALIFIERS): New.
35637 (acle_builtin_data): New. Change comment.
35638 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
35639 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
35640 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
35641 arm_acle_builtins.def.
35642 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
35643 (arm_init_acle_builtins): New.
35644 (CRC32_BUILTIN): Remove.
35645 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
35646 crc32cb, crc32ch and crc32cw.
35647 (arm_init_crc32_builtins): Remove.
35648 (arm_init_builtins): Use arm_init_acle_builtins rather
35649 than arm_init_crc32_builtins.
35650 (arm_expand_acle_builtin): New.
35651 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35652 * config/arm/arm_acle_builtins.def: New.
35654 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35656 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35657 (arm_builtin_datum): ... this.
35658 (arm_init_neon_builtin): Rename to ...
35659 (arm_init_builtin): ... this. Add a new parameters PREFIX
35660 and USE_SIG_IN_NAME.
35661 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35662 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35663 'arm_builtin_datum'.
35664 (arm_init_vfp_builtins): Likewise.
35665 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35666 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35667 (arm_expand_neon_args): Rename to ...
35668 (arm_expand_builtin_args): ... this. Rename builtin_arg
35669 enum values and differentiate between ARG_BUILTIN_MEMORY
35670 and ARG_BUILTIN_NEON_MEMORY.
35671 (arm_expand_neon_builtin_1): Rename to ...
35672 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35673 values, arm_expand_builtin_args and add bool parameter NEON.
35674 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35675 (arm_expand_vfp_builtin): Likewise.
35676 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35678 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35680 PR middle-end/77484
35681 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35682 * predict.c (tree_estimate_probability_bb): Reverse direction of
35683 polymorphic call predictor.
35685 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35687 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35688 (determine_pass_name_match): ...this new function and...
35689 (should_skip_pass_p): ...this new function.
35691 2017-01-06 Nathan Sidwell <nathan@acm.org>
35693 * ipa-visibility.c (function_and_variable_visibility): Reformat
35694 comments and long lines. Remove extrneous if.
35695 * symtab.c (symtab_node::make_decl_local): Fix code format.
35696 (symtab_node::set_section_for_node): Fix comment typo.
35698 2017-01-06 Martin Liska <mliska@suse.cz>
35701 * lra-constraints.c: Rename invariant to lra_invariant.
35702 * predict.c (set_even_probabilities): Initialize e to NULL.
35704 2017-01-05 Martin Sebor <msebor@redhat.com>
35706 PR tree-optimization/78910
35707 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35708 (format_integer): Correct off-by-one error in the handling
35709 of precision with negative numbers in signed conversions..
35711 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35713 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35715 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35717 PR tree-optimization/71016
35718 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35719 factor_out_conditional_conversion. Formatting fix.
35720 (factor_out_conditional_conversion): Add cond_stmt argument.
35721 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35722 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35725 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35727 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35728 read-rtl-function.o, and selftest-rtl.o.
35729 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35730 (selftest::aarch64_test_loading_full_dump): New function.
35731 (selftest::aarch64_run_selftests): New function.
35732 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35733 selftest::aarch64_run_selftests.
35734 * config/i386/i386.c
35735 (selftest::ix86_test_loading_dump_fragment_1): New function.
35736 (selftest::ix86_test_loading_call_insn): New function.
35737 (selftest::ix86_test_loading_full_dump): New function.
35738 (selftest::ix86_test_loading_unspec): New function.
35739 (selftest::ix86_run_selftests): Call the new functions.
35740 * emit-rtl.c (maybe_set_max_label_num): New function.
35741 * emit-rtl.h (maybe_set_max_label_num): New decl.
35742 * function.c (instantiate_decls): Guard call to
35743 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35744 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35746 * gensupport.c (gen_reader::gen_reader): Pass "false"
35747 for new "compact" param of rtx_reader.
35748 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35749 rather than an empty string for NULL strings.
35750 * read-md.c: Potentially include config.h rather than bconfig.h.
35751 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35752 (have_error): New global, copied from errors.c.
35753 (md_reader::read_name): Rename to...
35754 (md_reader::read_name_1): ...this, adding "out_loc" param,
35755 and converting "missing name or number" to returning false, rather
35757 (md_reader::read_name): Reimplement in terms of read_name_1.
35758 (md_reader::read_name_or_nil): New function.
35759 (md_reader::read_string): Handle "(nil)" by returning NULL.
35760 (md_reader::md_reader): Add new param "compact".
35761 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35762 (md_reader::read_file): New method.
35763 * read-md.h (md_reader::md_reader): Add new param "compact".
35764 (md_reader::read_file): New method.
35765 (md_reader::is_compact): New accessor.
35766 (md_reader::read_name): Convert return type from void to file_location.
35767 (md_reader::read_name_or_nil): New decl.
35768 (md_reader::read_name_1): New decl.
35769 (md_reader::m_compact): New field.
35770 (noop_reader::noop_reader): Pass "false" for new "compact" param
35772 (rtx_reader::rtx_reader): Add new "compact" param.
35773 (rtx_reader::read_rtx_operand): Make virtual and convert return
35774 type from void to rtx.
35775 (rtx_reader::read_until): New decl.
35776 (rtx_reader::handle_any_trailing_information): New virtual function.
35777 (rtx_reader::postprocess): New virtual function.
35778 (rtx_reader::finalize_string): New virtual function.
35779 (rtx_reader::m_in_call_function_usage): New field.
35780 (rtx_reader::m_reuse_rtx_by_id): New field.
35781 * read-rtl-function.c: New file.
35782 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35783 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35784 (selftest::verify_three_block_rtl_cfg): New decl.
35785 * read-rtl-function.h: New file.
35786 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35787 For host, include function.h, memmodel.h, and emit-rtl.h.
35788 (one_time_initialization): New function.
35789 (struct compact_insn_name): New struct.
35790 (compact_insn_names): New array.
35791 (find_code): Handle insn codes in compact dumps.
35792 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35793 (bind_subst_iter_and_attr): Likewise.
35794 (add_condition_to_string): Likewise.
35795 (add_condition_to_rtx): Likewise.
35796 (apply_attribute_uses): Likewise.
35797 (add_current_iterators): Likewise.
35798 (apply_iterators): Likewise.
35799 (initialize_iterators): Guard usage of apply_subst_iterator with
35800 #ifdef GENERATOR_FILE.
35801 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35802 (md_reader::read_mapping): Likewise.
35803 (add_define_attr_for_define_subst): Likewise.
35804 (add_define_subst_attr): Likewise.
35805 (read_subst_mapping): Likewise.
35806 (check_code_iterator): Likewise.
35807 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35809 (one_time_initialization): New function.
35810 (rtx_reader::read_until): New method.
35811 (read_flags): New function.
35812 (parse_reg_note_name): New function.
35813 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35814 Handle reuse_rtx ids.
35815 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35816 Add parsing support for RTL dumps, mirroring the special-cases in
35817 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35818 values, and calling handle_any_trailing_information.
35819 (rtx_reader::read_rtx_operand): Convert return type from void
35820 to rtx, returning return_rtx. Handle case 'e'. Call
35821 finalize_string on XSTR and XTMPL fields.
35822 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35823 "(nil)" values were omitted. Call the postprocess vfunc on the
35825 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35826 class ctor. Initialize m_in_call_function_usage. Call
35827 one_time_initialization.
35828 * rtl-tests.c (selftest::test_uncond_jump): Call
35829 set_new_first_and_last_insn.
35830 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35831 * selftest-rtl.c: New file.
35832 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35833 (selftest::get_insn_by_uid): New decl.
35834 * selftest-run-tests.c (selftest::run_tests): Call
35835 read_rtl_function_c_tests.
35836 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35837 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35840 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35842 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35843 operands in a special way. Assert that pos+len <= mode precision.
35845 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35847 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35848 3 argument Alias with unlimited for the negative form.
35849 (fno-vect-cost-model): Removed.
35851 2017-01-05 Martin Liska <mliska@suse.cz>
35853 * hsa-gen.c (gen_hsa_divmod): New function.
35854 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35856 2017-01-05 Martin Liska <mliska@suse.cz>
35859 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35862 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35864 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35865 small constant length operands.
35867 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35869 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35870 between loop iterations.
35872 2017-01-05 Martin Liska <mliska@suse.cz>
35875 * gimplify.c (gimplify_decl_expr): Compare to
35876 asan_poisoned_variables instread of checking flags.
35877 (gimplify_target_expr): Likewise.
35878 (gimplify_expr): Likewise.
35879 (gimplify_function_tree): Conditionally initialize
35880 asan_poisoned_variables.
35882 2017-01-04 Jeff Law <law@redhat.com>
35884 PR tree-optimizatin/78812
35885 * rtl.h (contains_mem_rtx_p): Prototype.
35886 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35887 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35888 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35889 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35890 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35892 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35894 * input.c (assert_char_at_range): Default-initialize actual_range.
35896 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35898 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35899 to match the caller.
35901 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35903 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35904 insns after final jump in test to emit dummy move.
35906 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35908 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35909 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35911 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35913 * multiple_target.c (create_dispatcher_calls): Init e_next.
35914 * tree-ssa-loop-split.c (split_loop): Init border.
35915 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35918 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35923 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35924 (altivec_register_operand): Do not return true if the operand
35925 contains a SUBREG mixing SImode and SFmode.
35926 (vsx_register_operand): Likewise.
35927 (vsx_reg_sfsubreg_ok): New predicate.
35928 (vfloat_operand): Do not return true if the operand contains a
35929 SUBREG mixing SImode and SFmode.
35930 (vint_operand): Likewise.
35931 (vlogical_operand): Likewise.
35932 (gpc_reg_operand): Likewise.
35933 (int_reg_operand): Likewise.
35934 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35935 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35936 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35938 (rs6000_emit_move_si_sf_subreg): New helper function.
35939 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35940 fixup SUBREGs involving SImode and SFmode.
35941 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35942 numbers for the new peephole2 optimization.
35943 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35944 the GLIBC math library that do AND/IOR/XOR operations on single
35945 precision floating point.
35946 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35947 target macros to say whether we need to avoid SUBREGs mixing
35949 (TARGET_ALLOW_SF_SUBREG): Likewise.
35950 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35951 (UNSPEC_SI_FROM_SF): Likewise.
35952 (iorxor): Change spacing.
35953 (and_ior_xor): New iterator for AND, IOR, and XOR.
35954 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35955 (movdi_from_sf_zero_ext): Likewise.
35956 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35957 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35958 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35959 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35960 (fms<mode>4): Likewise.
35961 (fnma<mode>4): Likewise.
35962 (fnms<mode>4): Likewise.
35963 (nfma<mode>4): Likewise.
35964 (nfms<mode>4): Likewise.
35966 2017-01-04 Marek Polacek <polacek@redhat.com>
35969 * doc/invoke.texi: Document -Wpointer-compare.
35971 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35973 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35976 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35977 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35980 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35983 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35984 documentation of the powerpc_popcntb_ok attribute.
35985 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35986 code to issue warning messages if a requested CPU configuration is
35987 not supported by the binary (assembler and loader) toolchain.
35988 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35989 made to define a built-in function that has been disabled.
35990 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35991 made to define a built-in function that has been disabled.
35992 (altivec_init_builtins): Add comment explaining why definition
35993 of the DST built-in functions is not preceded by an assertion
35994 check. Add assertions to prevent ICE if attempts are made to
35995 define an altivec predicate or an abs* built-in function that has
35997 (htm_init_builtins): Add comment explaining why definition of the
35998 htm built-in functions is not preceded by an assertion check.
36000 2017-01-04 Jeff Law <law@redhat.com>
36002 PR tree-optimizatin/67955
36003 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
36004 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
36005 the points-to solution does not include pt_null. Use DECL_PT_UID
36008 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
36010 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
36011 Use gen_int_mode instead of gen_lopwart for const_int operands.
36013 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36015 PR tree-optimization/71563
36016 * match.pd: Simplify X << Y into X if Y is known to be 0 or
36017 out of range value - has low bits known to be zero.
36019 2017-01-04 Alan Modra <amodra@gmail.com>
36021 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
36022 * configure: Regenerate.
36023 * config.in: Regenerate.
36025 2017-01-04 Jakub Jelinek <jakub@redhat.com>
36028 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
36029 a substring of the message, but strcmp with the whole message. Ifdef
36030 ENABLE_NLS, translate the message first using dgettext.
36032 2017-01-03 Jeff Law <law@redhat.com>
36034 PR tree-optimizatin/78856
36035 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
36036 (mark_threaded_blocks): Remove code to truncate thread paths that
36037 cross multiple loop headers. Instead invalidate the cached loop
36038 iteration information and handle case of a thread path walking
36039 into an irreducible region.
36041 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
36044 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
36045 assertions. Add support for doing the signbit if the IEEE 128-bit
36046 floating point value is in a GPR.
36047 * config/rs6000/rs6000.md (Fsignbit): Delete.
36048 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
36049 Update the length attribute if the value is in a GPR.
36050 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
36051 the sign or zero extension instruction, since the value is always 0/1.
36052 (signbit<mode>2_dm2): Delete using <Fsignbit>.
36055 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
36056 extracting SImode to a GPR register so that we can generate a
36057 store, limit the vector to be in a traditional Altivec register
36058 for the vextuwrx instruction.
36060 2017-01-03 Ian Lance Taylor <iant@google.com>
36062 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
36064 2017-01-03 Martin Sebor <msebor@redhat.com>
36066 PR tree-optimization/78696
36067 * gimple-ssa-sprintf.c (format_floating): Correct handling of
36068 precision. Use MPFR for %f for greater fidelity. Correct handling
36070 (pass_sprintf_length::compute_format_length): Set width and precision
36071 specified by asrerisk to void_node for vararg functions.
36072 (try_substitute_return_value): Adjust dump output.
36074 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
36076 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
36078 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
36080 * doc/invoke.texi (SPARC options): Document -mlra as the default.
36081 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
36082 -mlra/-mno-lra was passed to the compiler.
36084 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
36086 PR rtl-optimization/65618
36087 * emit-rtl.c (try_split): Move initialization of "before" and
36088 "after" to just before the call to emit_insn_after_setloc.
36090 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
36092 * doc/md.texi (Standard Names): Remove reference to Java frontend.
36094 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
36096 * dwarf2out.c (gen_enumeration_type_die): When
36097 -gno-strict-dwarf, add a DW_AT_encoding attribute.
36099 2017-01-03 Jakub Jelinek <jakub@redhat.com>
36101 PR tree-optimization/78965
36102 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
36103 Change first argument from const call_info & to call_info &. For %n
36104 set info.nowrite to false.
36106 PR middle-end/78901
36107 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
36108 possibly throwing calls.
36110 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
36111 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
36112 and fns handling, rather than in a separate case SSA_NAME.
36114 2017-01-02 Jeff Law <law@redhat.com>
36116 * config/darwin-driver.c (darwin_driver_init): Const-correctness
36117 fixes for first_period and second_period variables.
36119 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
36122 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
36123 (*insvqi_1): New insn pattern.
36124 (*insvqi_1_mem_rex64): Ditto.
36125 (*insvqi_2): Ditto.
36126 (*insvqi_3): Rename from *insvqi.
36128 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
36130 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
36132 * doc/cfg.texi (Edges): Remove reference to Java.
36133 (Maintaining the CFG): Ditto.
36135 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36137 PR middle-end/77674
36138 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
36139 transparent aliases.
36141 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36143 PR middle-end/77484
36144 * predict.def (PRED_CALL): Update hitrate.
36145 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
36146 * predict.c (tree_estimate_probability_bb): Split CALL predictor
36147 into direct/indirect/polymorphic variants.
36149 2017-01-01 Jakub Jelinek <jakub@redhat.com>
36151 Update copyright years.
36153 * gcc.c (process_command): Update copyright notice dates.
36154 * gcov-dump.c (print_version): Ditto.
36155 * gcov.c (print_version): Ditto.
36156 * gcov-tool.c (print_version): Ditto.
36157 * gengtype.c (create_file): Ditto.
36158 * doc/cpp.texi: Bump @copying's copyright year.
36159 * doc/cppinternals.texi: Ditto.
36160 * doc/gcc.texi: Ditto.
36161 * doc/gccint.texi: Ditto.
36162 * doc/gcov.texi: Ditto.
36163 * doc/install.texi: Ditto.
36164 * doc/invoke.texi: Ditto.
36166 Copyright (C) 2017 Free Software Foundation, Inc.
36168 Copying and distribution of this file, with or without modification,
36169 are permitted in any medium without royalty provided the copyright
36170 notice and this notice are preserved.