1 2017-02-06 Martin Sebor <msebor@redhat.com>
3 PR tree-optimization/79376
4 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
6 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
8 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
9 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
10 to simplify split condition.
12 2017-02-06 Jakub Jelinek <jakub@redhat.com>
14 * omp-expand.c (oxpand_omp_atomic_fetch_op,
15 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
18 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
20 PR rtl-optimization/68664
21 * target.def (can_speculate_insn): New hook.
22 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
23 * doc/tm.texi: Regenerate.
24 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
25 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
26 (rs6000_sched_can_speculate_insn): New function.
28 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30 PR tree-optimization/79284
31 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
33 vectorizable_mask_load_store, vectorizable_operation,
34 vect_is_simple_cond, get_same_sized_vectype): Use it instead
35 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
36 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
39 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
40 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
41 is_gimple_assign (stmt). Replace another such test with
42 is_gimple_assign (stmt).
44 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
47 * config/avr/avr.c (rtl-iter.h): Include it.
48 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
49 (avr_legitimate_combined_insn): ...and implementation.
51 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
53 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
54 * config/s390/s390.c (s390_const_operand_ok)
55 (s390_canonicalize_comparison, s390_extract_part)
56 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
57 (s390_contiguous_bitmask_p, s390_rtx_costs)
58 (legitimize_pic_address): Likewise.
59 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
60 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
61 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
62 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
63 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
65 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
67 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
68 REGNO($0) == REGNO($1).
70 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
72 * config/s390/linux.h(SIZE_TYPE): Add comment.
74 2017-02-06 Julian Brown <julian@codesourcery.com>
75 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
76 Virendra Pathak <virendra.pathak@broadcom.com>
78 * config/aarch64/aarch64-cores.def: Change the scheduler
80 * config/aarch64/aarch64.md: Include thunderx2t99.md.
81 * config/aarch64/thunderx2t99.md: New file.
83 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
85 * doc/standards.texi (Go Language): Update link to language
88 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
90 * tree-eh.c (lower_resx): Sanitize profile.
91 (cleanup_empty_eh_move_lp): Likewise.
93 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
96 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
98 * cfgloopmanip.h (loop_version): Update prototype.
99 * modulo-sched.c (sms_schedule): Update call of loop_version.
100 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
101 * tree-parloops.c (gen_parallel_loop): Likewise.
102 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
103 * tree-ssa-loop-split.c (split_loop): Likewise.
104 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
105 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
107 2017-02-05 Martin Liska <mliska@suse.cz>
110 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
112 (print_operand_address): Initialize a struct to zero.
114 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
116 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
117 garbage collector only in textual form.
119 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
121 * doc/extend.texi (x86 specific memory model extensions for
122 transactional memory): Simplify a phrase.
124 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
127 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
128 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
129 (atomic_storedi_1): Likewise.
131 2017-02-04 Jakub Jelinek <jakub@redhat.com>
133 PR tree-optimization/79338
134 * tree-parloops.c (gather_scalar_reductions): Don't call
135 vect_analyze_loop_form for loop->inner before destroying loop's
138 2017-02-03 Martin Sebor <msebor@redhat.com>
140 PR tree-optimization/79327
141 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
142 when precision has resulted in leading zeros.
143 (format_integer): Adjust the likely counter to assume an unknown
144 argument that may be zero is non-zero.
146 2017-02-03 Jason Merrill <jason@redhat.com>
149 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
150 avoid copying non-taken branch.
152 2017-02-03 Jakub Jelinek <jakub@redhat.com>
154 PR tree-optimization/79340
155 * tree-vect-loop.c (vectorizable_reduction): Release
156 vec_defs elements after safe_splicing them into other vectors.
159 PR tree-optimization/79327
160 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
161 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
163 (format_integer): Use wide_int_to_tree instead of build_int_cst
164 + to_?hwi. If argmin is NULL, just set argmin and argmax to
165 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
166 of shortest and longest sequence.
168 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
170 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
171 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
173 2017-02-03 Walter Lee <walt@tilera.com>
176 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
177 after initial stackframe link reg save.
178 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
180 2017-02-03 Jakub Jelinek <jakub@redhat.com>
183 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
184 wu for stxssp alternative.
186 2017-02-03 Martin Sebor <msebor@redhat.com>
188 PR tree-optimization/79352
189 * gimple-fold.c (get_range_strlen): Add argument.
190 (get_range_strlen): Change return type to bool.
191 (get_maxval_strlen): Pass in a dummy argument.
192 * gimple-fold.h (get_range_strlen): Change return type to bool.
193 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
194 * tree.h (array_at_struct_end_p): Add argument.
195 * tree.c (array_at_struct_end_p): Handle it.
197 2017-02-03 Martin Liska <mliska@suse.cz>
200 * multiple_target.c (create_dispatcher_calls): Redirect edge
201 from a caller of a dispatcher.
202 (expand_target_clones): Make the clones local.
203 (ipa_target_clone): Do both target clones and resolvers.
204 (ipa_dispatcher_calls): Remove the pass.
205 (pass_dispatcher_calls::gate): Likewise.
206 (make_pass_dispatcher_calls): Likewise.
207 * passes.def (pass_target_clone): Put as very first IPA early
210 2017-02-03 Martin Liska <mliska@suse.cz>
212 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
213 in case of a function with ifunc attribute.
215 2017-02-03 Martin Liska <mliska@suse.cz>
217 * cgraph.c (cgraph_node::dump): Dump function version info.
218 * symtab.c (symtab_node::dump_base): Add missing new line.
220 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
222 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
223 (ifcombine_ifandif): Use it.
225 2017-02-03 Martin Liska <mliska@suse.cz>
227 * doc/invoke.texi: Document default value for
228 use-after-scope-direct-emission-threshold.
230 2017-02-03 Martin Liska <mliska@suse.cz>
232 PR tree-optimization/79339
233 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
234 (format_floating): Likewise.
236 2017-02-03 Martin Liska <mliska@suse.cz>
239 * ipa-prop.c (ipa_node_params_t::insert): Remove current
241 (ipa_node_params_t::remove): Likewise.
242 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
243 initialization from removed ipa_node_params_t::insert.
244 (ipa_node_params::~ipa_node_params): Move from removed
245 ipa_node_params_t::release.
246 * symbol-summary.h (symbol_summary::m_released): New member.
247 Do not release a summary twice. Do not allow to call finalizer
248 for types of a summary that live in GGC memory.
250 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
252 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
255 2017-02-02 Martin Sebor <msebor@redhat.com>
258 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
259 (format_string): Tighten up the range of output for non-constant
260 strings and correct the expected range for wide non-constant strings.
262 2017-02-02 Martin Sebor <msebor@redhat.com>
264 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
267 * doc/invoke.texi (-fdump-rtl-): Remove pass-specific options from
269 (-fdump-tree-@var): Add to index and document how to come up
270 with pass-specific option and dump file names.
271 (-fdump-passes): Clarify where to look for output.
273 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
276 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
277 statistics of the analyzed path; allow threading for speed when
278 any of BBs along the path are optimized for speed.
280 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
283 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
284 settings of the virtual registers.
287 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
289 * explow.c (get_dynamic_stack_size): Take known alignment of stack
290 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
293 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
295 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
296 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
298 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
300 * config/s390/s390.md: Add missing comments with the expanded
302 * config/s390/vector.md: Likewise.
303 * config/s390/vx-builtins.md: Likewise.
305 2017-02-02 Jakub Jelinek <jakub@redhat.com>
308 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
309 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
310 conditions on a single line.
312 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
314 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
315 __S390_VX__ to __VX__.
317 2017-02-01 Andrew Pinski <apinski@cavium.com>
319 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
320 stmt_info to record_stmt_cost.
321 (vect_get_known_peeling_cost): Pass stmt_info if known to
323 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
324 cpu_vector_cost field into
325 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
326 field into vec_int_stmt_cost and vec_fp_stmt_cost.
327 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
328 splitting of scalar_stmt_cost and vec_stmt_cost.
329 (thunderx_vector_cost): Likewise.
330 (cortexa57_vector_cost): LIkewise.
331 (exynosm1_vector_cost): Likewise.
332 (xgene1_vector_cost): Likewise.
333 (thunderx2t99_vector_cost): Improve after the splitting of the two
335 (aarch64_builtin_vectorization_cost): Update for the splitting of
336 scalar_stmt_cost and vec_stmt_cost.
338 2017-02-01 Torvald Riegel <triegel@redhat.com>
339 Richard Henderson <rth@redhat.com>
341 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
342 conditional on existance of a fast atomic load.
343 * optabs-query.c (can_atomic_load_p): New function.
344 * optabs-query.h (can_atomic_load_p): Declare it.
345 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
346 no fast atomic load is available for the particular size of access.
347 (expand_atomic_compare_and_swap): Likewise.
348 (expand_atomic_load): Likewise.
349 (expand_atomic_store): Likewise.
350 (expand_atomic_fetch_op): Likewise.
351 * testsuite/lib/target-supports.exp
352 (check_effective_target_sync_int_128): Remove x86 because it provides
354 (check_effective_target_sync_int_128_runtime): Likewise.
356 2017-02-01 Richard Biener <rguenther@suse.de>
358 * graphite.c: Include tree-vectorizer.h for find_loop_location.
359 (graphite_transform_loops): Provide opt-info for optimized nests.
360 * tree-parloop.c (parallelize_loops): Provide opt-info for
363 2017-02-01 Richard Biener <rguenther@suse.de>
366 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
369 2017-02-01 Richard Biener <rguenther@suse.de>
371 PR tree-optimization/71824
372 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
373 Verify the loops are valid in the merged SESE region.
374 (scop_detection::can_represent_loop_1): Check analyzing the
375 evolution of the number of iterations in the region succeeds.
377 2017-01-31 Ian Lance Taylor <iant@golang.org>
379 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
380 REG_ARGS_SIZE note to 32-bit push insns and call insn.
382 2017-01-31 David Malcolm <dmalcolm@redhat.com>
384 PR preprocessor/79210
385 * input.c (get_substring_ranges_for_loc): Replace line_width
386 assertion with error-handling.
388 2017-01-31 Richard Biener <rguenther@suse.de>
390 PR tree-optimization/77318
391 * graphite-sese-to-poly.c (extract_affine): Fix assert.
392 (create_pw_aff_from_tree): Take loop parameter.
393 (add_condition_to_pbb): Pass loop of the condition to
394 create_pw_aff_from_tree.
396 2017-01-31 Jakub Jelinek <jakub@redhat.com>
398 * config/s390/s390.c (s390_asan_shadow_offset): New function.
399 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
401 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
405 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
407 (convert_int_to_float128): Likewise.
408 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
409 (convert_int_to_float128): Likewise.
410 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
411 (UNSPEC_IEEE128_CONVERT): Likewise.
412 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
413 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
414 Use local variables for IBM extended format.
415 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
416 (fix_trunc<mode>si2_fprs): Likewise.
417 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
418 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
419 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
420 to know that we can now have integers of all sizes in vector
422 (fix<uns>_<mode>di2_hw): Likewise.
423 (float<uns>_<mode>si2_hw): Likewise.
424 (fix_<mode>si2_hw): Likewise.
425 (fixuns_<mode>si2_hw): Likewise.
426 (float<uns>_<mode>di2_hw): Likewise.
427 (float_<mode>di2_hw): Likewise.
428 (float_<mode>si2_hw): Likewise.
429 (floatuns_<mode>di2_hw): Likewise.
430 (floatuns_<mode>si2_hw): Likewise.
431 (xscvqp<su>wz_<mode>): Delete, no longer used.
432 (xscvqp<su>dz_<mode>): Likewise.
433 (xscv<su>dqp_<mode>): Likewise.
434 (ieee128_mfvsrd_64bit): Likewise.
435 (ieee128_mfvsrd_32bit): Likewise.
436 (ieee128_mfvsrwz): Likewise.
437 (ieee128_mtvsrw): Likewise.
438 (ieee128_mtvsrd_64bit): Likewise.
439 (ieee128_mtvsrd_32bit): Likewise.
441 2017-01-31 Martin Liska <mliska@suse.cz>
444 * ipa-prop.c (ipa_free_all_node_params): Call release method
445 instead of ~sumbol_summary to not to trigger double times
448 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
450 PR tree-optimization/71691
451 * bitmap.h (class auto_bitmap): New.
452 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
453 is_maybe_undefined instead of ssa_undefined_value_p.
455 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
457 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
458 __S390_ARCH_LEVEL__ to __ARCH__.
460 2017-01-31 Jakub Jelinek <jakub@redhat.com>
462 PR tree-optimization/79267
463 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
464 if should_remove_lhs_p is true.
466 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
469 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
470 (add_alignment_attribute): New.
471 (base_type_die): Add alignment attribute.
472 (subrange_type_die): Likewise.
473 (modified_type_die): Likewise.
474 (gen_array_type_die): Likewise.
475 (gen_descr_array_type_die: Likewise.
476 (gen_enumeration_type_die): Likewise.
477 (gen_subprogram_die): Likewise.
478 (gen_variable_die): Likewise.
479 (gen_field_die): Likewise.
480 (gen_ptr_to_mbr_type_die): Likewise.
481 (gen_struct_or_union_type_die): Likewise.
482 (gen_subroutine_type_die): Likewise.
483 (gen_typedef_die): Likewise.
484 (base_type_cmp): Compare alignment attribute.
486 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
489 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
490 (setb_unsigned) New pattern for setb with CCUNS.
491 * config/rs6000/rs6000.c (expand_block_compare): Use a different
492 subfc./subfe sequence to avoid overflow problems. Generate a
493 shorter sequence with cmpld/setb for power9.
494 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
495 for generating subfc. instruction.
496 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
497 now uses this instruction.
499 2017-01-30 Ian Lance Taylor <iant@google.com>
502 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
503 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
505 2017-01-30 Martin Sebor <msebor@redhat.com>
507 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
508 Move constant to the right of a relational operator.
509 (get_mpfr_format_length, format_character, format_string): Ditto.
510 (should_warn_p, maybe_warn): Same.
512 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
514 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
517 * asan.c (get_translation_unit_decl): Remove function.
518 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
520 2017-01-30 Martin Liska <mliska@suse.cz>
522 PR gcov-profile/79259
523 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
526 2017-01-30 Martin Liska <mliska@suse.cz>
529 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
530 Initialize variables with NULL value.
532 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
535 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
537 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
539 2017-01-30 Richard Biener <rguenther@suse.de>
541 PR tree-optimization/79276
542 * tree-vrp.c (process_assert_insertions): Properly adjust common
543 when removing a duplicate.
545 * gcc.dg/torture/pr79276.c: New testcase.
547 2017-01-30 Richard Biener <rguenther@suse.de>
549 PR tree-optimization/79256
550 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
551 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
553 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
555 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
558 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
559 ("*r<noxa>sbg_<mode>_sll_bitmask")
560 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
561 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
562 Use contiguous_bitmask_nowrap_operand.
564 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
567 * config/rs6000/altivec.h (vec_xl): Revise #define.
570 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
572 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
574 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
576 PR rtl-optimization/79194
577 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
578 traps before call to bypass_conditional_jumps.
580 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
582 PR tree-optimization/71374
583 * lra-constraints.c (check_conflict_input_operands): New.
584 (match_reload): Use it.
586 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
589 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
590 account to calculate conflict_set.
592 2017-01-27 Bin Cheng <bin.cheng@arm.com>
594 PR rtl-optimization/78559
595 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
596 other_insn in combine.
598 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
600 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
601 uint16_type_node for BT_UINT16.
603 2017-01-27 David Malcolm <dmalcolm@redhat.com>
605 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
607 (GIMPLE Tests): New node.
608 (RTL Tests): New node.
610 2017-01-27 Richard Biener <rguenther@suse.de>
612 PR tree-optimization/79245
613 * tree-loop-distribution.c (distribute_loop): Apply cost
614 modeling also to detected patterns.
616 2017-01-27 Richard Biener <rguenther@suse.de>
618 PR tree-optimization/71433
619 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
620 (compare_assert_loc): New function.
621 (process_assert_insertions): Sort and optimize assert locations
622 to remove duplicates and push down identical assertions on
623 edges to their destination block.
625 2017-01-27 Richard Biener <rguenther@suse.de>
627 PR tree-optimization/79244
628 * tree-vrp.c (remove_range_assertions): Forcefully propagate
629 out SSA names even if abnormal.
631 2017-01-27 Jakub Jelinek <jakub@redhat.com>
633 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
634 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
635 instead of MPFR_RNDN.
637 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
640 * arm.c (arm_option_override): Don't call build_target_option_node
641 until after doing all option overrides.
642 (arm_valid_target_attribute_tree): Likewise.
644 2017-01-27 Martin Liska <mliska@suse.cz>
646 * doc/invoke.texi (-fprofile-arcs): Document profiling support
647 for {cd}tors and C++ {cd}tors.
649 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
651 * config/s390/s390.md ("*setmem_long_and")
652 ("*setmem_long_and_31z"): Use zero_extend instead of and.
654 2017-01-26 Martin Sebor <msebor@redhat.com>
656 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
659 2017-01-26 Martin Sebor <msebor@redhat.com>
661 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
662 HAVE_DFmode before using XFmode or DFmode.
663 (parse_directive): Avoid using the z length modifier to avoid
664 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
667 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
668 to accept adjustment as an array.
669 (get_int_range): New function.
670 (struct directive): Make width and prec arrays.
671 (directive::set_width, directive::set_precision): Call get_int_range.
672 (format_integer, format_floating): Handle width and precision ranges.
673 (format_string, parse_directive): Same.
675 2017-01-26 Jakub Jelinek <jakub@redhat.com>
678 * dwarf2out.c (generate_skeleton_bottom_up): For children with
679 comdat_type_p set, just clone them, but keep the children in the
683 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
684 which have direct callers with -fvar-tracking-assignments enabled
686 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
687 inside of type units.
689 2017-01-26 Martin Sebor <msebor@redhat.com>
692 * gimple-ssa-sprintf.c (struct result_range): Add likely and
694 (struct format_result): Replace number_chars, number_chars_min,
695 and number_chars_max with a single member of struct result_range.
697 (format_result::operator+=): Adjust.
698 (struct fmtresult): Remove bounded. Handle likely and unlikely
700 (fmtresult::adjust_for_width_or_precision): New function.
701 (fmtresult:type_max_digits): New function.
702 (bytes_remaining): Handle likely and unlikely counters.
703 (min_bytes_remaining): Remove.
704 (format_percent): Simplify.
705 (format_integer, format_floating): Set likely and unlikely counters.
706 (get_string_length, format_character, format_string): Same.
707 (format_plain, should_warn_p): New function.
708 (maybe_warn): Call should_warn_p. Update diagnostic messages
709 and handle those for all directives, including plain strings.
710 (format_directive): Handle likely and unlikely counters.
711 Remove unnecessary quoting from diagnostics. Add an informational
714 (pass_sprintf_length::compute_format_length): Simplify.
715 (try_substitute_return_value): Handle likely and unlikely counters.
717 2017-01-26 Carl Love <cel@us.ibm.com>
719 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
720 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
722 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
725 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
726 endianess for subregs into account.
727 * lra-constraints.c (lra_constraints): Do risky transformations
728 always on the first iteration.
729 * lra-lives.c (check_pseudos_live_through_calls): Add arg
730 last_call_used_reg_set.
731 (process_bb_lives): Define and use last_call_used_reg_set.
732 * lra.c (lra): Always continue after lra_constraints on the first
735 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
737 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
739 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
741 2017-01-26 Jakub Jelinek <jakub@redhat.com>
743 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
744 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
745 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
746 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
747 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
748 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
749 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
750 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
751 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
753 2017-01-26 Marek Polacek <polacek@redhat.com>
756 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
757 for the third operand.
759 2017-01-26 Jakub Jelinek <jakub@redhat.com>
762 * omp-low.c (struct omp_context): Add simt_stmt field.
763 (scan_omp_for): Return omp_context *.
764 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
765 context to the _simt_ SIMD stmt.
766 (lower_omp_for): For combined SIMD with sibling _simt_
767 SIMD, make sure to use the same decls in _looptemp_
768 clauses as in the sibling.
770 2017-01-26 David Sherwood <david.sherwood@arm.com>
773 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
776 2017-01-26 Jakub Jelinek <jakub@redhat.com>
779 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
780 emit fld b; fld a; if possible.
782 * brig-builtins.def: Update copyright years.
783 * config/arm/arm_acle_builtins.def: Update copyright years.
785 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
788 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
789 constraint instead of o for the stxsd instruction.
791 2017-01-25 Carl Love <cel@us.ibm.com>
793 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
794 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
796 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
798 * doc/invoke.texi (C++ Dialect Options): Fix typo.
800 2017-01-25 Richard Biener <rguenther@suse.de>
802 PR tree-optimization/69264
803 * target.def (vector_alignment_reachable): Improve documentation.
804 * doc/tm.texi: Regenerate.
805 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
807 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
808 earlier changes with respect to TYPE_USER_ALIGN.
809 (vector_alignment_reachable_p): Likewise. Improve dumping.
811 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
814 * config/arm/arm.md (xordi3): Force constant operand into a register
817 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
819 * doc/invoke.texi (-fstore-merging): Correct default optimization
820 levels at which it is enabled.
821 (-O): Move -fstore-merging from list to...
824 2017-01-25 Richard Biener <rguenther@suse.de>
827 * omp-expand.c: Include debug.h.
828 (expand_omp_taskreg): Make sure to generate early debug before
829 outlining anything from a function.
830 (expand_omp_target): Likewise.
831 (grid_expand_target_grid_body): Likewise.
833 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
836 * asan.c (get_translation_unit_decl): New function.
837 (asan_add_global): Extract modules file name from globals
838 TRANSLATION_UNIT_DECL name.
840 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
843 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
844 for long calls with APCS frame and VFP.
846 2017-01-24 David Malcolm <dmalcolm@redhat.com>
848 * cfg.c (original_copy_tables_initialized_p): New function.
849 * cfg.h (original_copy_tables_initialized_p): New decl.
850 * cfgrtl.c (relink_block_chain): Guard the call to
851 free_original_copy_tables with a call to
852 original_copy_tables_initialized_p.
853 * cgraph.h (symtab_node::native_rtl_p): New decl.
854 * cgraphunit.c (symtab_node::native_rtl_p): New function.
855 (symtab_node::needed_p): Don't assert for early assembly output
857 (cgraph_node::finalize_function): Set "force_output" for __RTL
859 (cgraph_node::analyze): Bail out early for __RTL functions.
860 (analyze_functions): Update assertion to support __RTL functions.
861 (cgraph_node::expand): Bail out early for __RTL functions.
862 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
864 * function.h (struct function): Update comment for field
866 * gimple-expr.c: Include "tree-pass.h".
867 (gimple_has_body_p): Return false for __RTL functions.
868 * Makefile.in (OBJS): Add run-rtl-passes.o.
869 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
871 (gcc::pass_manager::get_clean_slate): New accessor.
872 * passes.c: Include "insn-addr.h".
873 (should_skip_pass_p): Add logging. Update logic for running
874 "expand" to be compatible with both __GIMPLE and __RTL. Guard
875 property-provider override so it is only done for gimple passes.
877 (skip_pass): New function.
878 (execute_one_pass): Call skip_pass when skipping passes.
879 * read-md.c (md_reader::read_char): Support filtering
880 the input to a subset of line numbers.
881 (md_reader::md_reader): Initialize fields
882 m_first_line and m_last_line.
883 (md_reader::read_file_fragment): New function.
884 * read-md.h (md_reader::read_file_fragment): New decl.
885 (md_reader::m_first_line): New field.
886 (md_reader::m_last_line): New field.
887 * read-rtl-function.c (function_reader::create_function): Only
888 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
889 curr_properties. Set DECL_INITIAL to a dummy block.
890 (read_rtl_function_body_from_file_range): New function.
891 * read-rtl-function.h (read_rtl_function_body_from_file_range):
893 * run-rtl-passes.c: New file.
894 * run-rtl-passes.h: New file.
896 2017-01-24 Jeff Law <law@redhat.com>
898 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
901 2017-01-24 Bin Cheng <bin.cheng@arm.com>
903 PR tree-optimization/79159
904 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
905 (record_nonwrapping_iv): Improve boundary using above function if no
906 value range information.
908 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
909 Martin Jambor <mjambor@suse.cz>
911 * brig-builtins.def: New file.
912 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
913 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
914 (DEF_HSAIL_SAT_BUILTIN): Likewise.
915 (DEF_HSAIL_INTR_BUILTIN): Likewise.
916 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
917 * builtin-types.def (BT_INT8): New.
918 (BT_INT16): Likewise.
919 (BT_UINT8): Likewise.
920 (BT_UINT16): Likewise.
921 (BT_FN_ULONG): Likewise.
922 (BT_FN_UINT_INT): Likewise.
923 (BT_FN_UINT_ULONG): Likewise.
924 (BT_FN_UINT_LONG): Likewise.
925 (BT_FN_UINT_PTR): Likewise.
926 (BT_FN_ULONG_PTR): Likewise.
927 (BT_FN_INT8_FLOAT): Likewise.
928 (BT_FN_INT16_FLOAT): Likewise.
929 (BT_FN_UINT32_FLOAT): Likewise.
930 (BT_FN_UINT16_FLOAT): Likewise.
931 (BT_FN_UINT8_FLOAT): Likewise.
932 (BT_FN_UINT64_FLOAT): Likewise.
933 (BT_FN_UINT16_UINT32): Likewise.
934 (BT_FN_UINT32_UINT16): Likewise.
935 (BT_FN_UINT16_UINT16_UINT16): Likewise.
936 (BT_FN_INT_PTR_INT): Likewise.
937 (BT_FN_UINT_PTR_UINT): Likewise.
938 (BT_FN_LONG_PTR_LONG): Likewise.
939 (BT_FN_ULONG_PTR_ULONG): Likewise.
940 (BT_FN_VOID_UINT64_UINT64): Likewise.
941 (BT_FN_UINT8_UINT8_UINT8): Likewise.
942 (BT_FN_INT8_INT8_INT8): Likewise.
943 (BT_FN_INT16_INT16_INT16): Likewise.
944 (BT_FN_INT_INT_INT): Likewise.
945 (BT_FN_UINT_FLOAT_UINT): Likewise.
946 (BT_FN_FLOAT_UINT_UINT): Likewise.
947 (BT_FN_ULONG_UINT_UINT): Likewise.
948 (BT_FN_ULONG_UINT_PTR): Likewise.
949 (BT_FN_ULONG_ULONG_ULONG): Likewise.
950 (BT_FN_UINT_UINT_UINT): Likewise.
951 (BT_FN_VOID_UINT_PTR): Likewise.
952 (BT_FN_UINT_UINT_PTR: Likewise.
953 (BT_FN_UINT32_UINT64_PTR): Likewise.
954 (BT_FN_INT_INT_UINT_UINT): Likewise.
955 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
956 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
957 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
958 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
959 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
960 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
961 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
962 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
963 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
964 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
965 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
966 * doc/frontends.texi: List BRIG FE.
967 * doc/install.texi (Testing): Add BRIG tesring requirements.
968 * doc/invoke.texi (Overall Options): Mention BRIG.
969 * doc/standards.texi (Standards): Doucment BRIG HSA version.
971 2017-01-24 Richard Biener <rguenther@suse.de>
974 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
976 2017-01-24 Martin Jambor <mjambor@suse.cz>
979 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
980 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
983 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
986 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
987 casts from signed to unsigned really don't have a range.
989 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
991 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
992 GMP_RNDx for compatiblity.
994 2017-01-24 Martin Liska <mliska@suse.cz>
997 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
998 that would prevent us to call alloca with -1 as argument.
1000 2017-01-24 Jakub Jelinek <jakub@redhat.com>
1002 * dwarf2out.c (output_compilation_unit_header, output_file_names):
1003 Avoid -Wformat-security warning.
1005 2017-01-23 Andrew Pinski <apinski@cavium.com>
1007 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
1010 2017-01-23 Martin Sebor <msebor@redhat.com>
1013 * gimple-ssa-sprintf.c (warn_level): New global.
1014 (format_integer): Use it here and throughout the rest of the file.
1015 Use the same switch to compute sign as base.
1016 (maybe_warn): New function.
1017 (format_directive): Factor out warnings into maybe_warn.
1018 Add debugging output. Use warn_level.
1019 (add_bytes): Use warn_level.
1020 (pass_sprintf_length::compute_format_length): Add debugging output.
1021 (try_substitute_return_value): Same.
1022 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
1025 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
1026 (struct fmtresult, format_integer, format_floating): Adjust.
1027 (fmtresult::fmtresult): Set max correctly in two argument ctor.
1028 (get_string_length, format_string,format_directive): Same.
1029 (pass_sprintf_length::compute_format_length): Same.
1030 (try_substitute_return_value): Simplify slightly.
1033 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
1034 (fmtresult::operator+=): Outlined.
1035 (struct fmtresult): Add ctors.
1036 (struct conversion_spec): Rename...
1037 (struct directive): ...to this. Add and remove data members.
1038 (directive::set_width, directive::set_precision): New functions.
1039 (format_percent): Use fmtresult ctor.
1040 (get_width_and_precision): Remove.
1041 (format_integer): Make naming changes. Avoid computing width and
1043 (format_floating): Same. Adjust indentation.
1044 (format_character, format_none): New functions.
1045 (format_string): Moved character handling to format_character.
1046 (format_directive): Remove arguments, change return type.
1047 (parse_directive): New function.
1048 (pass_sprintf_length::compute_format_length): Move directive
1049 parsing to parse_directive.
1051 2017-01-23 Jakub Jelinek <jakub@redhat.com>
1053 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
1054 (assign_assembler_name_if_needed): ... this.
1055 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
1056 (assign_assembler_name_if_needed): ... this.
1057 (free_lang_data_in_cgraph): Adjust callers.
1058 * cgraphunit.c (cgraph_node::analyze): Likewise.
1059 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
1062 2017-01-23 Richard Biener <rguenther@suse.de>
1064 PR tree-optimization/79088
1065 PR tree-optimization/79188
1066 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
1067 resetting loop bounds after last path deletion. Reset loop
1068 bounds of the target loop, make code match the comments.
1069 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
1070 Make sure loops need no fixups.
1072 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
1074 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
1075 exponent support with double type for first argument.
1076 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
1077 type returned by __builtin_vec_extract_sig,
1078 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
1079 functions from "vector int" to "vector unsigned int" or from
1080 "vector long long int" to "vector unsigned long long int".
1081 Changed type returned by __builtin_vec_extract_exp,
1082 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
1083 functions from "vector int" to "vector unsigned int" or from
1084 "vector long long int" to "vector unsigned long long int".
1085 Changed return type of __builtin_vec_test_data_class,
1086 __builtin_vec_test_data_class_sp, and
1087 __builtin_vec_test_data_class_dp from "vector int" to
1088 "vector bool int" or from "vector long long int" to "vector bool
1089 long long int" and changed second argument type from "unsigned
1090 int" to "int". Added new overloaded function forms "vector float
1091 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
1092 "vector float __builtin_vec_insert_exp_sp (vector float, vector
1093 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
1094 double, vector unsigned long long int)" and "vector double
1095 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
1096 long int)". Changed return type of
1097 __builtin_scalar_test_data_class and
1098 __builtin_scalar_test_data_class_sp and
1099 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
1100 int" and changed second argument from "unsigned int" to "int".
1101 Changed type returned by __builtin_scalar_test_neg,
1102 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
1103 from "int" to "bool int". Added new overloaded function form
1104 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
1105 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
1106 exponent double-precision with floating point first argument.
1107 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
1108 documentation of scalar_test_data_class, scalar_test_neg,
1109 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
1110 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
1111 vec_test_data_class built-in functions to reflect refinements in
1112 their type signatures.
1114 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
1116 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
1118 (aarch64_elf_asm_destructor): Likewise.
1120 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
1122 PR rtl-optimization/78634
1123 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
1124 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
1125 * ifcvt.c (noce_try_cmove): Add missing cost check.
1127 PR rtl-optimization/71724
1128 * combine.c (if_then_else_cond): Look for situations where it is
1129 beneficial to undo the work of one of the recursive calls.
1131 2017-01-23 Bin Cheng <bin.cheng@arm.com>
1133 PR tree-optimization/70754
1134 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
1135 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
1136 combined stmt before it if not NULL.
1137 (combine_chains): Process refs reversely and compute dominance point
1140 2017-01-23 Martin Liska <mliska@suse.cz>
1142 PR tree-optimization/79196
1143 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
1144 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
1146 (strlen_optimize_stmt): Call the renamed function.
1148 2017-01-23 Michael Matz <matz@suse.de>
1150 PR tree-optimization/78384
1151 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
1153 2017-01-23 Richard Biener <rguenther@suse.de>
1155 PR tree-optimization/79186
1156 * tree-vrp.c (register_new_assert_for): Make sure we've seen
1157 both incoming edges before moving an assert.
1159 2017-01-23 Martin Jambor <mjambor@suse.cz>
1161 * ipa-prop.c (load_from_param_1): Removed.
1162 (load_from_unmodified_param): Bits from load_from_param_1 put back
1164 (load_from_param): Removed.
1165 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
1166 with stmt. Reverted back to use of load_from_unmodified_param.
1168 2017-01-23 Martin Jambor <mjambor@suse.cz>
1171 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
1172 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
1173 field a pointer to garbage collected vector, mark lattices and
1174 ipcp_orig_node with GTY((skip)).
1175 (ipa_get_param_count): Adjust to descriptors being a pointer.
1176 (ipa_get_param): Likewise.
1177 (ipa_get_type): Likewise.
1178 (ipa_get_param_move_cost): Likewise.
1179 (ipa_set_param_used): Likewise.
1180 (ipa_get_controlled_uses): Likewise.
1181 (ipa_set_controlled_uses): Likewise.
1182 (ipa_is_param_used): Likewise.
1183 (ipa_node_params_t): Move into garbage collector. New methods insert
1185 (ipa_node_params_sum): Annotate wth GTY(()).
1186 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
1188 (ipa_load_from_parm_agg): Adjust declaration.
1189 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
1190 * ipa-profile.c (ipa_profile): Likewise.
1191 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
1192 (ipa_populate_param_decls): Make descriptors parameter garbage
1194 (ipa_dump_param): Adjust to descriptors being a pointer.
1195 (ipa_alloc_node_params): Likewise.
1196 (ipa_initialize_node_params): Likewise.
1197 (load_from_param_1): Make descriptors parameter garbage collected.
1198 (load_from_unmodified_param): Likewise.
1199 (load_from_param): Likewise.
1200 (ipa_load_from_parm_agg): Likewise.
1201 (ipa_node_params::~ipa_node_params): Removed.
1202 (ipa_free_all_node_params): Remove call to delete operator.
1203 (ipa_node_params_t::insert): New.
1204 (ipa_node_params_t::remove): Likewise.
1205 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
1206 copy known_csts and known_contexts vectors.
1207 (ipa_read_node_info): Adjust to descriptors being a pointer.
1208 (ipcp_modif_dom_walker): Make m_descriptors field garbage
1210 (ipcp_transform_function): Make descriptors variable garbage
1213 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
1215 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
1216 * config/i386/avx512dqintrin.h: Ditto.
1217 * config/i386/avx512fintrin.h: Ditto.
1218 * gcc/config/i386/i386.c: Handle new builtins.
1219 * config/i386/i386-builtin.def: Add new builtins.
1220 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
1221 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
1223 2017-01-23 Jakub Jelinek <jakub@redhat.com>
1224 Martin Liska <mliska@suse.cz>
1226 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
1227 * asan.c (asan_expand_poison_ifn): Support stores and use
1228 appropriate ASAN report function.
1229 * internal-fn.c (expand_ASAN_POISON_USE): New function.
1230 * internal-fn.def (ASAN_POISON_USE): Declare.
1231 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
1232 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
1233 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
1234 ASAN_POISON calls w/o LHS.
1235 * tree-ssa.c (execute_update_addresses_taken): Create clobber
1236 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
1237 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
1238 * gimplify.c (asan_poison_variables): Add attribute
1239 use_after_scope_memory to variables that really needs to live
1241 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
1242 having the attribute.
1244 2017-01-23 Martin Liska <mliska@suse.cz>
1246 * asan.c (create_asan_shadow_var): New function.
1247 (asan_expand_poison_ifn): Likewise.
1248 * asan.h (asan_expand_poison_ifn): New declaration.
1249 * internal-fn.c (expand_ASAN_POISON): Likewise.
1250 * internal-fn.def (ASAN_POISON): New builtin.
1251 * sanopt.c (pass_sanopt::execute): Expand
1252 asan_expand_poison_ifn.
1253 * tree-inline.c (copy_decl_for_dup_finish): Make function
1255 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
1256 * tree-ssa.c (is_asan_mark_p): New function.
1257 (execute_update_addresses_taken): Rewrite local variables
1258 (identified just by use-after-scope as addressable) into SSA.
1260 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
1262 * doc/install.texi (Specific): opensource.apple.com uses https
1263 now. Remove trailing slash.
1265 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
1267 * README.Portability: Remove note on an Irix compatibility issue.
1269 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
1271 * gcov.c (INCLUDE_ALGORITHM): Define.
1272 (INCLUDE_VECTOR): Define.
1273 No longer include <vector> and <algorithm> directly.
1275 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
1277 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
1279 * doc/invoke.texi (Code Gen Options): Ditto.
1281 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
1284 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
1286 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
1288 rtl-optimization/79125
1289 * cprop.c (local_cprop_pass): Handle cases where we make an
1292 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
1296 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
1297 read from, for big endian.
1299 2017-01-20 Jiong Wang <jiong.wang@arm.com>
1301 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
1302 register pauth builtins for LP64 only.
1304 2017-01-20 Marek Polacek <polacek@redhat.com>
1307 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
1310 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
1312 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
1314 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
1315 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
1316 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
1318 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1321 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
1322 in big-endian mode when they are not a single duplicated value.
1324 2017-01-20 Richard Biener <rguenther@suse.de>
1326 * BASE-VER: Bump to 7.0.1.
1328 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
1330 * omp-low.c (omplow_simd_context): New struct. Use it...
1331 (lower_rec_simd_input_clauses): ...here and...
1332 (lower_rec_input_clauses): ...here to hold common data. Adjust all
1333 references to idx, lane, max_vf, is_simt.
1335 2017-01-20 Graham Markall <graham.markall@embecosm.com>
1337 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
1340 2017-01-20 Martin Jambor <mjambor@suse.cz>
1342 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
1343 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
1345 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
1346 (GTFILES): Rename hsa.c to hsa-common.c.
1347 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
1348 * hsa-dump.c: Likewise.
1349 * hsa-gen.c: Likewise.
1350 * hsa-regalloc.c: Likewise.
1351 * ipa-hsa.c: Likewise.
1352 * omp-expand.c: Likewise.
1353 * omp-low.c: Likewise.
1354 * toplev.c: Likewise.
1356 2017-01-20 Marek Polacek <polacek@redhat.com>
1359 * doc/invoke.texi: Document -Wduplicated-branches.
1360 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
1361 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
1362 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
1363 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
1364 return 0 only when not OEP_LEXICOGRAPHIC.
1365 (fold_build_cleanup_point_expr): Use the expression
1366 location when building CLEANUP_POINT_EXPR.
1367 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
1368 * tree.c (add_expr): Handle error_mark_node.
1370 2017-01-20 Martin Liska <mliska@suse.cz>
1373 * tree-profile.c (init_ic_make_global_vars): Do not call
1375 (gimple_init_gcov_profiler): Likewise.
1377 2017-01-20 Martin Liska <mliska@suse.cz>
1380 * cgraph.h (maybe_create_reference): Remove argument and
1382 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
1384 * ipa-cp.c (create_specialized_node): Likewise.
1385 * symtab.c (symtab_node::maybe_create_reference): Handle
1386 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
1388 2017-01-20 Martin Liska <mliska@suse.cz>
1390 * read-rtl-function.c (function_reader::create_function): Use
1391 build_decl instread of build_decl_stat.
1393 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
1395 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
1396 * config/i386/avx512dqintrin.h: Ditto.
1397 * config/i386/avx512fintrin.h: Ditto.
1398 * config/i386/i386-builtin-types.def: Add new types.
1399 * gcc/config/i386/i386.c: Handle new types.
1400 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
1401 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
1402 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
1403 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
1404 (__builtin_ia32_kshiftridi): New.
1405 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
1407 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
1411 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
1412 define to rs6000_init_stack_protect_guard.
1413 (rs6000_init_stack_protect_guard): New function.
1415 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
1416 Yunqiang Su <yunqiang.su@imgtec.com>
1418 * config.gcc (supported_defaults): Add madd4.
1419 (with_madd4): Add validation.
1420 (all_defaults): Add madd4.
1421 * config/mips/mips.opt (mmadd4): New option.
1422 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
1424 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
1426 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
1427 (ISA_HAS_FUSED_MADD4): Likewise.
1428 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
1429 * gcc/doc/install.texi (--with-madd4): Document the new option.
1431 2017-01-19 Jiong Wang <jiong.wang@arm.com>
1433 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
1434 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
1435 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
1436 (aarch64_init_pauth_hint_builtins): New.
1437 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
1438 (aarch64_expand_builtin): Expand new builtins.
1440 2017-01-19 Jiong Wang <jiong.wang@arm.com>
1442 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
1443 * combine-stack-adj.c (no_unhandled_cfa): Handle
1444 REG_CFA_TOGGLE_RA_MANGLE.
1445 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
1446 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
1447 info for return address signing.
1448 (aarch64_expand_epilogue): Likewise.
1450 2017-01-19 Jiong Wang <jiong.wang@arm.com>
1452 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
1453 * config/aarch64/aarch64-protos.h
1454 (aarch64_return_address_signing_enabled): New declaration.
1455 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
1457 (aarch64_expand_prologue): Sign return address before it's pushed onto
1459 (aarch64_expand_epilogue): Authenticate return address fetched from
1461 (aarch64_override_options): Sanity check for ILP32 and ISA level.
1462 (aarch64_attributes): New function attributes for "sign-return-address".
1463 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
1464 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
1465 ("*do_return"): Generate combined instructions according to key index.
1466 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
1467 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
1469 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
1470 * config/aarch64/aarch64.opt (msign-return-address=): New.
1471 * doc/extend.texi (AArch64 Function Attributes): Documents
1472 "sign-return-address=".
1473 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
1475 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
1477 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
1478 overall option summary.
1480 2017-01-19 Jiong Wang <jiong.wang@arm.com>
1482 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
1483 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
1484 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
1485 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
1487 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
1489 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
1490 -mpower9-minmax by default for -mcpu=power9.
1491 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
1492 128-bit floating point.
1494 2017-01-20 Alan Modra <amodra@gmail.com>
1496 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
1497 optimizing for size.
1499 2017-01-20 Alan Modra <amodra@gmail.com>
1502 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
1503 for strcmp and strncmp from corresponding builtin decl.
1505 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
1507 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
1508 instead of i386/rtems-64.h.
1509 * config/i386/rtems-64.h: Remove.
1511 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
1515 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
1517 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
1519 2017-01-19 Tamar Christina <tamar.christina@arm.com>
1521 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1522 Change int to HOST_WIDE_INT.
1523 * config/aarch64/aarch64-protos.h
1524 (aarch64_simd_gen_const_vector_dup): Likewise.
1525 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
1527 2017-01-19 David Malcolm <dmalcolm@redhat.com>
1529 * langhooks-def.h (lhd_type_for_size): New decl.
1530 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
1531 * langhooks.c (lhd_type_for_size): New function, taken from
1534 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
1536 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
1537 define_bypass for CR latency.
1538 (power9-cracked-alu): Update bypass latency and remove power9-branch.
1539 (power9-alu2): Add define_bypass for CR latency.
1541 (power9-mul): Update insn latency.
1542 (power9-mul-compare): Update insn latency, bypass latency and remove
1545 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1547 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
1549 * config/aarch64/aarch64.md
1550 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
1551 aarch64_nopcrelative_literal_loads.
1552 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
1554 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
1556 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
1558 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
1560 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
1563 * config.gcc (supported_defaults): Add lxc1-sxc1.
1564 (with_lxc1_sxc1): Add validation.
1565 (all_defaults): Add lxc1-sxc1.
1566 * config/mips/mips.opt (mlxc1-sxc1): New option.
1567 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
1569 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
1570 __mips_no_lxc1_sxc1.
1571 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
1572 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
1573 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
1575 2017-01-19 Richard Biener <rguenther@suse.de>
1577 PR tree-optimization/72488
1578 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
1579 sure to restore SSA info.
1580 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
1582 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
1584 PR rtl-optimization/79121
1585 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
1586 of the inner type when shifting an extended value.
1588 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
1591 * symtab.c (symtab_node::equal_address_to): Fix comparing of
1592 interposable aliases.
1594 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
1597 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
1598 Use the evmergelohi instruction.
1599 (mov_si<mode>_e500_subreg4_2_le): Likewise.
1600 (mov_sitf_e500_subreg8_2_be): Likewise.
1601 (mov_sitf_e500_subreg12_2_le): Likewise.
1602 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
1603 (mov_si<mode>_e500_subreg4_2_be): Likewise.
1604 (mov_sitf_e500_subreg8_2_le): Likewise.
1605 (mov_sitf_e500_subreg12_2_be): Likewise.
1607 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1609 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
1610 attribute from vecsimple to vecperm.
1611 (altivec_vbpermq2): Likewise.
1613 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1616 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
1618 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1619 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
1620 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
1621 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
1622 case where N arg is SIZE_MAX.
1623 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
1624 (cmpstrsi): Add pattern.
1626 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1628 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1629 __builtin_vec_revb builtins.
1630 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
1631 built-in functions to support generation of the ISA 3.0 XXBR<x>
1632 vector byte reverse instructions.
1633 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
1634 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
1635 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
1636 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
1637 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
1638 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
1639 (P9V_BUILTIN_VEC_REVB): Likewise.
1640 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
1641 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
1642 (p9_xxbrq_v16qi): Likewise.
1643 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
1644 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
1645 (p9_xxbrh_v8hi): Likewise.
1646 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
1647 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
1648 vec_revb built-in functions.
1650 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
1652 PR rtl-optimization/78952
1653 * config/i386/i386.md (any_extract): New code iterator.
1654 (*insvqi_2): Use any_extract for source operand.
1655 (*insvqi_3): Use any_shiftrt for source operand.
1657 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
1659 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
1661 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
1663 2017-01-18 Matthias Klose <doko@ubuntu.com>
1665 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
1667 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1669 * config/rs6000/altivec.h (vec_bperm): Change #define.
1670 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
1671 (altivec_vbpermq2): New define_insn.
1672 (altivec_vbpermd): Likewise.
1673 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
1675 (VBPERMD): Likewise.
1676 (VBPERM): New polymorphic function interface.
1677 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
1678 Add entries for P9V_BUILTIN_VEC_VBPERM.
1679 * doc/extend.texi: Add interfaces for vec_bperm.
1681 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1683 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
1684 first letter of error messages.
1685 (s390_resolve_overloaded_builtin): Likewise.
1686 * config/s390/s390.c (s390_expand_builtin): Likewise.
1687 (s390_invalid_arg_for_unprototyped_fn): Likewise.
1688 (s390_valid_target_attribute_inner_p): Likewise.
1689 * config/s390/s390.md ("tabort"): Likewise.
1691 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
1693 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
1694 (ISA_AVOID_DIV_HILO): New macro.
1695 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
1696 (ISA_HAS_DDIV): Likewise.
1698 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
1700 * doc/invoke.texi (fabi-version): Correct number of occurrences.
1702 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
1704 * doc/invoke.texi (fabi-version): Spelling fix.
1706 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
1709 * doc/invoke.texi (fabi-version): Mention mangling fix for
1712 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
1715 * doc/invoke.texi (fabi-version): Document discriminator mangling.
1717 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
1720 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
1721 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
1723 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
1725 (stack_protect_test): Ditto.
1726 * config/rs6000/rs6000.opt (mstack-protector-guard=,
1727 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
1729 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
1730 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
1731 -mstack-protector-guard-offset=.
1732 (RS/6000 and PowerPC Options): Ditto.
1734 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
1736 * config/i386/i386.h (MASK_CLASS_P): New define.
1737 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
1738 there are no registers from different register sets also when
1739 mask registers are used. Update function comment.
1740 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
1741 to (*k/*r) and (*k/*km) alternatives.
1743 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
1745 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
1746 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
1747 (EH_RETURN_HANDLER_RTX): New define.
1748 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
1749 Force frame pointer in EH return functions.
1750 (aarch64_expand_epilogue): Add barrier for eh_return.
1751 (aarch64_final_eh_return_addr): Remove.
1752 (aarch64_eh_return_handler_rtx): New function.
1753 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
1755 (aarch64_eh_return_handler_rtx): New prototype.
1757 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1759 * config/rs6000/altivec.h (vec_rlmi): New #define.
1760 (vec_vrlnm): Likewise.
1761 (vec_rlnm): Likewise.
1762 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
1763 (UNSPEC_VRLNM): Likewise.
1764 (VIlong): New mode iterator.
1765 (altivec_vrl<VI_char>mi): New define_insn.
1766 (altivec_vrl<VI_char>nm): Likewise.
1767 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
1770 (RLNM): New polymorphic function entry.
1771 (VRLWMI): New monomorphic function entry.
1773 (RLMI): New polymorphic function entry.
1774 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
1775 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
1776 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
1779 2017-01-17 Jakub Jelinek <jakub@redhat.com>
1782 * dwarf2out.c (field_byte_offset): Restore the
1783 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
1784 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
1785 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
1788 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
1791 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
1793 2017-01-17 Jakub Jelinek <jakub@redhat.com>
1796 * dwarf2out.c (add_data_member_location_attribute): For constant
1797 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
1798 instead of DW_AT_data_member_location, DW_AT_bit_offset and
1799 DW_AT_byte_size attributes.
1801 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
1803 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
1804 after forcing to constant memory when the code model is medium.
1806 2017-01-17 Julia Koval <julia.koval@intel.com>
1809 * config/i386/avx512fintrin.h
1810 (_mm512_i32gather_ps): Change __addr type to void const*.
1811 (_mm512_mask_i32gather_ps): Ditto.
1812 (_mm512_i32gather_pd): Ditto.
1813 (_mm512_mask_i32gather_pd): Ditto.
1814 (_mm512_i64gather_ps): Ditto.
1815 (_mm512_mask_i64gather_ps): Ditto.
1816 (_mm512_i64gather_pd): Ditto.
1817 (_mm512_mask_i64gather_pd): Ditto.
1818 (_mm512_i32gather_epi32): Ditto.
1819 (_mm512_mask_i32gather_epi32): Ditto.
1820 (_mm512_i32gather_epi64): Ditto.
1821 (_mm512_mask_i32gather_epi64): Ditto.
1822 (_mm512_i64gather_epi32): Ditto.
1823 (_mm512_mask_i64gather_epi32): Ditto.
1824 (_mm512_i64gather_epi64): Ditto.
1825 (_mm512_mask_i64gather_epi64): Ditto.
1826 (_mm512_i32scatter_ps): Change __addr type to void*.
1827 (_mm512_mask_i32scatter_ps): Ditto.
1828 (_mm512_i32scatter_pd): Ditto.
1829 (_mm512_mask_i32scatter_pd): Ditto.
1830 (_mm512_i64scatter_ps): Ditto.
1831 (_mm512_mask_i64scatter_ps): Ditto.
1832 (_mm512_i64scatter_pd): Ditto.
1833 (_mm512_mask_i64scatter_pd): Ditto.
1834 (_mm512_i32scatter_epi32): Ditto.
1835 (_mm512_mask_i32scatter_epi32): Ditto.
1836 (_mm512_i32scatter_epi64): Ditto.
1837 (_mm512_mask_i32scatter_epi64): Ditto.
1838 (_mm512_i64scatter_epi32): Ditto.
1839 (_mm512_mask_i64scatter_epi32): Ditto.
1840 (_mm512_i64scatter_epi64): Ditto.
1841 (_mm512_mask_i64scatter_epi64): Ditto.
1842 * config/i386/avx512pfintrin.h
1843 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
1844 (_mm512_mask_prefetch_i32gather_ps): Ditto.
1845 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1846 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1847 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
1848 (_mm512_prefetch_i32scatter_ps): Ditto.
1849 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1850 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1851 (_mm512_prefetch_i64scatter_pd): Ditto.
1852 (_mm512_prefetch_i64scatter_ps): Ditto.
1853 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1854 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1855 * config/i386/avx512vlintrin.h
1856 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
1857 (_mm_mmask_i32gather_ps): Ditto.
1858 (_mm256_mmask_i32gather_pd): Ditto.
1859 (_mm_mmask_i32gather_pd): Ditto.
1860 (_mm256_mmask_i64gather_ps): Ditto.
1861 (_mm_mmask_i64gather_ps): Ditto.
1862 (_mm256_mmask_i64gather_pd): Ditto.
1863 (_mm_mmask_i64gather_pd): Ditto.
1864 (_mm256_mmask_i32gather_epi32): Ditto.
1865 (_mm_mmask_i32gather_epi32): Ditto.
1866 (_mm256_mmask_i32gather_epi64): Ditto.
1867 (_mm_mmask_i32gather_epi64): Ditto.
1868 (_mm256_mmask_i64gather_epi32): Ditto.
1869 (_mm_mmask_i64gather_epi32): Ditto.
1870 (_mm256_mmask_i64gather_epi64): Ditto.
1871 (_mm_mmask_i64gather_epi64): Ditto.
1872 (_mm256_i32scatter_ps): Change __addr type to void*.
1873 (_mm256_mask_i32scatter_ps): Ditto.
1874 (_mm_i32scatter_ps): Ditto.
1875 (_mm_mask_i32scatter_ps): Ditto.
1876 (_mm256_i32scatter_pd): Ditto.
1877 (_mm256_mask_i32scatter_pd): Ditto.
1878 (_mm_i32scatter_pd): Ditto.
1879 (_mm_mask_i32scatter_pd): Ditto.
1880 (_mm256_i64scatter_ps): Ditto.
1881 (_mm256_mask_i64scatter_ps): Ditto.
1882 (_mm_i64scatter_ps): Ditto.
1883 (_mm_mask_i64scatter_ps): Ditto.
1884 (_mm256_i64scatter_pd): Ditto.
1885 (_mm256_mask_i64scatter_pd): Ditto.
1886 (_mm_i64scatter_pd): Ditto.
1887 (_mm_mask_i64scatter_pd): Ditto.
1888 (_mm256_i32scatter_epi32): Ditto.
1889 (_mm256_mask_i32scatter_epi32): Ditto.
1890 (_mm_i32scatter_epi32): Ditto.
1891 (_mm_mask_i32scatter_epi32): Ditto.
1892 (_mm256_i32scatter_epi64): Ditto.
1893 (_mm256_mask_i32scatter_epi64): Ditto.
1894 (_mm_i32scatter_epi64): Ditto.
1895 (_mm_mask_i32scatter_epi64): Ditto.
1896 (_mm256_i64scatter_epi32): Ditto.
1897 (_mm256_mask_i64scatter_epi32): Ditto.
1898 (_mm_i64scatter_epi32): Ditto.
1899 (_mm_mask_i64scatter_epi32): Ditto.
1900 (_mm256_i64scatter_epi64): Ditto.
1901 (_mm256_mask_i64scatter_epi64): Ditto.
1902 (_mm_i64scatter_epi64): Ditto.
1903 (_mm_mask_i64scatter_epi64): Ditto.
1904 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
1905 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
1906 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
1907 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
1908 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
1909 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
1910 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
1911 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
1912 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
1913 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
1914 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
1915 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
1916 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
1917 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
1918 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
1919 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
1920 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
1921 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
1922 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
1923 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
1924 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
1925 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
1926 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
1927 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
1928 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
1929 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
1930 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
1931 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
1932 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
1933 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
1934 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
1935 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
1936 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
1937 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
1938 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
1939 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
1940 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
1941 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
1942 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
1943 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
1944 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
1945 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
1946 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
1947 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
1948 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
1949 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
1950 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
1951 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
1952 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
1953 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
1954 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
1955 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
1956 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
1957 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
1958 definitions accordingly.
1960 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
1961 Kuan-Lin Chen <kuanlinchentw@gmail.com>
1964 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
1967 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
1970 * ira-conflicts.c (ira_build_conflicts): Update total conflict
1971 hard regs for inner regno.
1973 2017-01-17 Martin Liska <mliska@suse.cz>
1976 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
1977 assumption and add comment.
1979 2017-01-17 Nathan Sidwell <nathan@acm.org>
1981 * ipa-visibility.c (localize_node): New function, broken out of ...
1982 (function_and_variable_visibility): ... here. Call it.
1984 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
1987 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
1988 correctly set frequency of oudgoing edge.
1989 (duplicate_thread_path): Fix profile updating.
1991 2017-01-17 Jakub Jelinek <jakub@redhat.com>
1994 * configure.ac: Add GCC_BASE_VER.
1995 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
1996 version from BASE-VER file.
1997 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
1998 (gcc.o): Depend on $(BASEVER).
1999 * common.opt (dumpfullversion): New option.
2000 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
2001 * doc/invoke.texi: Document -dumpfullversion.
2002 * doc/install.texi: Document --with-gcc-major-version-only.
2003 * configure: Regenerated.
2005 2017-01-17 Richard Biener <rguenther@suse.de>
2007 PR tree-optimization/71433
2008 * tree-vrp.c (register_new_assert_for): Merge same asserts
2009 on all incoming edges.
2010 (process_assert_insertions_for): Handle insertions at the
2013 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
2015 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
2016 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
2018 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
2021 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
2024 2017-01-17 Alan Modra <amodra@gmail.com>
2027 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
2028 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
2029 symbolic stack limit when pic.
2031 2017-01-16 Martin Sebor <msebor@redhat.com>
2033 PR tree-optimization/78608
2034 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
2036 2017-01-16 Jeff Law <law@redhat.com>
2039 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
2040 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
2041 for several include directories that may be relative to sysroot.
2042 * config/i386/x-mingw32 (gplus_includedir): Define.
2043 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
2044 (native_system_includedir): Likewise.
2045 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
2046 override if TARGET_SYSTEM_ROOT is defined.
2047 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
2049 PR tree-optimization/79090
2050 PR tree-optimization/33562
2051 PR tree-optimization/61912
2052 PR tree-optimization/77485
2053 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
2054 and computed trims into the dump file.
2056 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
2058 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
2060 2017-01-16 Jakub Jelinek <jakub@redhat.com>
2063 * gimplify.c (gimplify_init_constructor): If want_value and
2064 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
2068 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
2069 sequence. Formatting fixes.
2070 (doloop_optimize): Formatting fixes.
2073 * gcc.c (debug_level_greater_than_spec_func): New function.
2074 (static_spec_functions): Add debug-level-gt spec function.
2075 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
2077 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
2078 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
2079 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
2080 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
2081 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
2082 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
2084 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
2086 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
2087 QImode fixups to general and mask registers only.
2089 2017-01-16 Carl Love <cel@us.ibm.com>
2091 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
2092 for built-in functions
2093 vector signed char vec_nabs (vector signed char)
2094 vector signed short vec_nabs (vector signed short)
2095 vector signed int vec_nabs (vector signed int)
2096 vector signed long long vec_nabs (vector signed long long)
2097 vector float vec_nabs (vector float)
2098 vector double vec_nabs (vector double)
2099 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
2101 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
2102 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
2103 * doc/extend.texi: Update the documentation file for the new built-in
2106 2017-01-16 Martin Sebor <msebor@redhat.com>
2108 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
2111 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2113 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
2114 UNSPEC_VSX__XXSPLTD to require special splat handling.
2116 2017-01-16 David Malcolm <dmalcolm@redhat.com>
2119 * system.h: Poison strndup.
2121 2017-01-16 Alan Modra <amodra@gmail.com>
2124 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
2127 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
2129 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
2131 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
2133 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
2134 call recog here. Assert that INSN_CODE (insn) is non-negative.
2136 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
2139 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
2141 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
2142 in the currently scheduled RTL fragment.
2144 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
2146 PR rtl-optimization/78751
2147 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
2150 2017-01-14 Jeff Law <law@redhat.com>
2152 PR tree-optimization/79090
2153 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
2154 variable length stores.
2155 (compute_trims): Delete dead assignment to *trim_tail.
2156 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
2159 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
2161 PR rtl-optimization/78626
2162 PR rtl-optimization/78727
2163 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
2164 of a block, and split such blocks after everything else is finished.
2166 2017-01-14 Alan Modra <amodra@gmail.com>
2169 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
2170 target legitimate_combined_insn.
2171 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
2172 (rs6000_legitimate_combined_insn): New function.
2173 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
2175 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
2176 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
2177 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
2179 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
2181 * doc/frontends.texi (G++ and GCC): Remove references to Java.
2183 2017-01-13 Jeff Law <law@redhat.com>
2185 PR tree-optimization/33562
2186 PR tree-optimization/61912
2187 PR tree-optimization/77485
2188 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
2190 (delete_dead_assignment): Likewise.
2191 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
2192 statement to delete_dead_call and delete_dead_assignment.
2194 2017-01-13 David Malcolm <dmalcolm@redhat.com>
2197 * substring-locations.c (format_warning_va): Strengthen case 1 so
2198 that both endpoints of the substring must be within the format
2199 range for just the substring to be printed.
2201 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
2203 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
2204 * config/i386/i386.c (ix86_target_string): Add missing options
2205 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
2206 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
2207 flags_other and ix86_target_other to flags2_other. Display unknown
2209 (ix86_valid_target_attribute_inner_p): Add missing options and
2210 reorder options by implied ISAs, as in ix86_target_string.
2212 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
2214 * hash-table.h (hash_table::too_empty_p): New function.
2215 (hash_table::expand): Use it.
2216 (hash_table::traverse): Likewise.
2217 (hash_table::empty_slot): Use sizeof (value_type) instead of
2218 sizeof (PTR) to convert bytes to elements. Shrink the table
2219 if the current size is excessive for the current number of
2222 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
2224 * ira-costs.c (record_reg_classes): Break from the inner loop
2225 early once alt_fail is known to be true. Update outer loop
2226 handling accordingly.
2228 2017-01-13 Jeff Law <law@redhat.com>
2230 * tree-ssa-dse.c (decrement_count): New function.
2231 (increment_start_addr, maybe_trim_memstar_call): Likewise.
2232 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
2233 when we know the partially dead statement is a mem* function.
2235 PR tree-optimization/61912
2236 PR tree-optimization/77485
2237 * tree-ssa-dse.c: Include expr.h.
2238 (maybe_trim_constructor_store): New function.
2239 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
2241 PR tree-optimization/33562
2242 PR tree-optimization/61912
2243 PR tree-optimization/77485
2244 * doc/invoke.texi: Document new dse-max-object-size param.
2245 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
2246 * tree-ssa-dse.c: Include params.h.
2247 (dse_store_status): New enum.
2248 (initialize_ao_ref_for_dse): New, partially extracted from
2250 (valid_ao_ref_for_dse, normalize_ref): New.
2251 (setup_live_bytes_from_ref, compute_trims): Likewise.
2252 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
2253 (maybe_trim_partially_dead_store): Likewise.
2254 (maybe_trim_complex_store): Likewise.
2255 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
2256 Track what bytes live from the original store. Return tri-state
2257 for dead, partially dead or live.
2258 (dse_dom_walker): Add constructor, destructor and new private members.
2259 (delete_dead_call, delete_dead_assignment): New extracted from
2261 (dse_optimize_stmt): Make a member of dse_dom_walker.
2262 Use initialize_ao_ref_for_dse.
2264 PR tree-optimization/33562
2265 PR tree-optimization/61912
2266 PR tree-optimization/77485
2267 * sbitmap.h (bitmap_count_bits): Prototype.
2268 (bitmap_clear_range, bitmap_set_range): Likewise.
2269 * sbitmap.c (bitmap_clear_range): New function.
2270 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
2272 2017-01-13 Martin Liska <mliska@suse.cz>
2275 * function.c (set_cfun): Add new argument force.
2276 * function.h (set_cfun): Likewise.
2277 * ipa-inline-transform.c (inline_call): Use the function when
2278 strict alising from is dropped for function we inline to.
2280 2017-01-13 Richard Biener <rguenther@suse.de>
2282 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
2283 for dumping GIMPLE INTEGER_CSTs.
2285 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2287 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
2288 to 201112L since C++17.
2290 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
2293 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
2294 if -fsanitize=kernel-address is present.
2296 2017-01-13 Richard Biener <rguenther@suse.de>
2298 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
2299 as _Literal ( type ) number in case usual suffixes do not
2300 preserve all information.
2302 2017-01-13 Richard Biener <rguenther@suse.de>
2304 PR tree-optimization/77283
2305 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
2306 and ssa-iterators.h.
2307 (is_feasible_trace): Implement a cost model based on joiner
2310 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
2313 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
2314 char or short to __float128/_Float128 directly.
2316 2017-01-12 Martin Sebor <msebor@redhat.com>
2318 to -Wformat-overflow.
2319 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
2320 (min_bytes_remaining): Same.
2321 (get_string_length): Same.
2322 (format_string): Same.
2323 (format_directive): Same.
2325 (pass_sprintf_length::handle_gimple_call): Same.
2327 2017-01-12 Jakub Jelinek <jakub@redhat.com>
2329 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
2330 info.nowrite calls with no lhs that can't throw. Return bool
2331 whether gsi_remove has been called or not.
2332 (pass_sprintf_length::handle_gimple_call): Return bool whether
2333 try_substitute_return_value called gsi_remove. Formatting fix.
2334 (pass_sprintf_length::execute): Don't use gsi_remove if
2335 handle_gimple_call returned true.
2338 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
2339 be removed due to side-effects, don't remove following barrier nor
2340 turn the successor edge into fallthru edge.
2342 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2345 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
2346 element-reversing loads and stores as not swappable.
2348 2017-01-12 Nathan Sidwell <nathan@acm.org>
2349 Nicolai Stange <nicstange@gmail.com>
2351 * combine.c (try_combine): Don't ignore result of overlap checking
2352 loop. Combine overlap & asm check into single loop.
2354 2017-01-12 Richard Biener <rguenther@suse.de>
2356 * tree-pretty-print.c (dump_generic_node): Provide -gimple
2357 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
2359 2017-01-12 Richard Biener <rguenther@suse.de>
2361 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
2362 and TS_TARGET_OPTION directly derive from TS_BASE.
2363 * tree-core.h (tree_optimization_option): Derive from tree_base.
2364 (tree_target_option): Likewise.
2366 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
2368 * config/i386/i386.c (memory_address_length): Increase len
2369 only when rip_relative_addr_p returns false.
2371 2017-01-11 Julia Koval <julia.koval@intel.com>
2373 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
2374 (OPTION_MASK_ISA_SGX_SET): New.
2375 (ix86_handle_option): Handle OPT_msgx.
2376 * config.gcc: Added sgxintrin.h.
2377 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
2378 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
2379 * config/i386/i386.c (ix86_target_string): Add -msgx.
2381 (ix86_option_override_internal): Handle new options.
2382 (ix86_valid_target_attribute_inner_p): Add sgx.
2383 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
2384 * config/i386/i386.opt: Add msgx.
2385 * config/i386/sgxintrin.h: New file.
2386 * config/i386/x86intrin.h: Add sgxintrin.h.
2388 2017-01-11 Jakub Jelinek <jakub@redhat.com>
2391 * fold-const.c (maybe_nonzero_address): Return 1 for function
2393 (tree_single_nonzero_warnv_p): Don't handle function local objects
2397 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
2400 2017-01-11 David Malcolm <dmalcolm@redhat.com>
2403 * opts.c: Include "spellcheck.h"
2404 (struct string_fragment): New struct.
2405 (struct edit_distance_traits<const string_fragment &>): New
2407 (get_closest_sanitizer_option): New function.
2408 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
2410 2017-01-11 Jakub Jelinek <jakub@redhat.com>
2412 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
2414 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
2415 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
2416 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
2417 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
2418 for initial die_offset if dwarf_split_debug_info.
2419 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
2420 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
2422 (output_skeleton_debug_sections): Formatting fix. Use
2423 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
2424 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
2426 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
2428 * config/arm/cortex-a53.md: Add bypasses for
2430 (cortex_a53_r2f): Only use for transfers.
2431 (cortex_a53_f2r): Likewise.
2432 (cortex_a53_r2f_cvt): Add reservation for conversions.
2433 (cortex_a53_f2r_cvt): Likewise.
2435 2017-01-11 Tamar Christina <tamar.christina@arm.com>
2437 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
2438 to all inlined functions, change static to extern.
2440 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
2443 * config/arm/arm.c (legitimize_pic_address): Handle reference to
2445 (arm_assemble_integer): Likewise.
2447 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
2449 * config.gcc: Use new awk script to check CPU, FPU and architecture
2450 parameters for --with-... options.
2451 * config/arm/parsecpu.awk: New file
2452 * config/arm/arm-cpus.in: New file.
2453 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
2455 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
2457 * config/arm/t-arm: Update dependency rules.
2458 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
2459 of processing .def files.
2460 * config/arm/genopt.sh: Deleted.
2461 * config/arm/gentune.sh: Deleted.
2462 * config/arm/arm-cores.def: Deleted.
2463 * config/arm/arm-arches.def: Deleted.
2464 * config/arm/arm-fpus.def: Deleted.
2465 * config/arm/arm-tune.md: Regenerated.
2466 * config/arm/arm-tables.opt: Regenerated.
2467 * config/arm/arm-cpu.h: New generated file.
2468 * config/arm/arm-cpu-data.h: New generated file.
2469 * config/arm/arm-cpu-cdata.h: New generated file.
2471 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
2474 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
2476 (input_varpool_node): Unpack dynamically_initialized bit.
2478 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
2480 PR rtl-optimization/79032
2481 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
2482 the alignment of the adjusted memory reference against that of MODE,
2483 instead of the alignment of the original memory reference.
2485 2017-01-11 Martin Jambor <mjambor@suse.cz>
2487 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
2489 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
2490 decorated functions.
2492 2017-01-11 Richard Biener <rguenther@suse.de>
2494 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
2495 set range/nonnull info for PHI results. Do not set it on
2496 stmts marked for removal.
2498 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
2500 * expr.c (store_field): In the bitfield case, fetch the return value
2501 from the registers before applying a single big-endian adjustment.
2502 Always do a final load for a BLKmode value not larger than a word.
2504 2017-01-10 David Malcolm <dmalcolm@redhat.com>
2507 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
2508 that we correctly handle column numbers greater than
2509 LINE_MAP_MAX_COLUMN_NUMBER.
2511 2017-01-10 Martin Sebor <msebor@redhat.com>
2514 * gimple-ssa-sprintf.c (get_destination_size): Call
2515 {init,fini}object_sizes.
2516 * tree-object-size.c (addr_object_size): Adjust.
2517 (pass_through_call): Adjust.
2518 (pass_object_sizes::execute): Adjust.
2519 * tree-object-size.h (fini_object_sizes): Declare.
2521 2017-01-10 Martin Sebor <msebor@redhat.com>
2523 PR tree-optimization/78775
2524 * builtins.c (get_size_range): Move...
2525 * calls.c: ...to here.
2526 (alloc_max_size): Accept zero argument.
2527 (operand_signed_p): Remove.
2528 (maybe_warn_alloc_args_overflow): Call get_size_range.
2529 * calls.h (get_size_range): Declare.
2531 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
2533 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
2534 from TI's devices.csv file as of September 2016.
2535 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
2537 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
2539 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
2540 * doc/invoke.texi: Likewise.
2541 * doc/md.texi: Likewise.
2542 * doc/objc.texi: Likewise.
2544 2017-01-10 Joshua Conner <joshconner@google.com>
2546 * config/arm/fuchsia-elf.h: New file.
2547 * config/fuchsia.h: New file.
2548 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
2549 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
2551 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
2553 2016-01-10 Richard Biener <rguenther@suse.de>
2555 PR tree-optimization/79034
2556 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
2557 Propagate out degenerate PHIs in the joiner.
2559 2017-01-10 Martin Liska <mliska@suse.cz>
2561 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
2562 (sort_congruence_classes_by_decl_uid): Likewise.
2563 (sort_congruence_class_groups_by_decl_uid): Likewise.
2564 (sem_item_optimizer::merge_classes): Sort class, groups in these
2565 classes and members in the groups by DECL_UID of declarations.
2566 This would make merge operations stable.
2568 2017-01-10 Martin Liska <mliska@suse.cz>
2570 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
2571 usage of m_classes_vec.
2572 (sem_item_optimizer::~sem_item_optimizer): Likewise.
2573 (sem_item_optimizer::get_group_by_hash): Likewise.
2574 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
2575 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
2576 (sem_item_optimizer::verify_classes): Likewise.
2577 (sem_item_optimizer::process_cong_reduction): Likewise.
2578 (sem_item_optimizer::dump_cong_classes): Likewise.
2579 (sem_item_optimizer::merge_classes): Likewise.
2580 * ipa-icf.h (congruence_class_hash): Rename from
2581 congruence_class_group_hash. Remove declaration of m_classes_vec.
2583 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
2585 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
2586 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
2587 * config.gcc: Add avx512vpopcntdqintrin.h.
2588 * config/i386/avx512vpopcntdqintrin.h: New.
2589 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
2590 * config/i386/i386-builtin-types.def: Add new types.
2591 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
2592 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
2593 __builtin_ia32_vpopcountq_v8di_mask): New.
2594 * config/i386/i386-c.c (ix86_target_macros_internal): Define
2595 __AVX512VPOPCNTDQ__.
2596 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
2597 (PTA_AVX512VPOPCNTDQ): Define.
2598 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
2599 TARGET_AVX512VPOPCNTDQ_P): Define.
2600 * config/i386/i386.opt: Add mavx512vpopcntdq.
2601 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
2602 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
2604 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
2607 * predict.def (PRED_CALL): Set to 67.
2609 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
2611 * expr.c (store_field): In the bitfield case, if the value comes from
2612 a function call and is of an aggregate type returned in registers, do
2613 not modify the field mode; extract the value in all cases if the mode
2614 is BLKmode and the size is not larger than a word.
2616 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
2619 * config/i386/cpuid.h: Fix undefined behavior.
2621 2017-01-04 Jeff Law <law@redhat.com>
2623 PR tree-optimization/79007
2624 PR tree-optimization/67955
2625 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
2626 conservative for pt.null when flag_non_call_exceptions is on.
2628 2017-01-09 Jakub Jelinek <jakub@redhat.com>
2630 PR translation/79019
2631 PR translation/79020
2632 * params.def (PARAM_INLINE_MIN_SPEEDUP,
2633 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
2634 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
2636 * config/avr/avr.opt (maccumulate-args): Likewise.
2637 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
2638 * common.opt (freport-bug): Likewise.
2639 * cif-code.def (CIF_FINAL_ERROR): Likewise.
2640 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
2641 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
2642 translatable string.
2643 * config/i386/i386.c (function_value_32): Likewise.
2644 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
2645 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
2647 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
2648 * common/config/msp430/msp430-common.c (msp430_handle_option):
2650 * symtab.c (symtab_node::verify_base): Likewise.
2651 * opts.c (set_debug_level): Likewise.
2652 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
2653 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
2654 missing whitespace to translatable strings.
2655 * config/avr/avr.md (bswapsi2): Fix typo in comment.
2656 * config/sh/superh.h: Likewise.
2657 * config/i386/xopintrin.h: Likewise.
2658 * config/i386/znver1.md: Likewise.
2659 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
2660 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
2661 * double-int.h (struct double_int): Likewise.
2662 * double-int.c (div_and_round_double): Likewise.
2663 * wide-int.cc: Likewise.
2664 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
2665 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
2666 * cfgcleanup.c (crossjumps_occured): Renamed to ...
2667 (crossjumps_occurred): ... this.
2668 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
2671 PR tree-optimization/78899
2672 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
2673 returning bool return struct loop *, NULL for failure and the new
2675 (versionable_outer_loop_p): Don't version outer loop if it has
2676 dont_vectorized bit set.
2677 (tree_if_conversion): When versioning outer loop, ensure
2678 tree_if_conversion is performed also on the inner loop of the
2679 non-vectorizable outer loop copy.
2680 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
2681 LOOP_VECTORIZED in inner loop of the scalar outer loop and
2682 prevent vectorization of it.
2683 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
2684 the outer loop vectorization of the non-scalar version is attempted
2685 before vectorization of the inner loop in scalar version. If
2686 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
2687 vectorization of its inner loop.
2688 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
2689 has 2 inner loops, rename also on edges from bb whose single pred
2690 is outer_loop->header. Fix typo in function comment.
2692 2017-01-09 Martin Sebor <msebor@redhat.com>
2695 * asan.c (asan_emit_stack_protection): Increase local buffer size
2696 to avoid snprintf truncation warning.
2698 2017-01-09 Andrew Pinski <apinski@cavium.com>
2700 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
2701 to reference thunderx2t99 for the tuning structure
2702 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
2704 (thunderx2t99_extra_costs): This.
2705 * config/aarch64/aarch64-tune.md: Regenerate.
2706 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
2707 (vulcan_addrcost_table): This.
2708 (vulcan_regmove_cost): Rename to ...
2709 (thunderx2t99_regmove_cost): This.
2710 (vulcan_vector_cost): Rename to ...
2711 (thunderx2t99_vector_cost): this.
2712 (vulcan_branch_cost): Rename to ...
2713 (thunderx2t99_branch_cost): This.
2714 (vulcan_tunings): Rename to ...
2715 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
2716 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
2718 2017-01-09 Martin Jambor <mjambor@suse.cz>
2722 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
2723 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
2724 (propagate_vr_accross_jump_function): Use the above function for all
2725 value range computations for pass-through jump functions and type
2726 converasion from explicit value range values.
2727 (ipcp_propagate_stage): Do not attempt to deduce types of formal
2728 parameters from TYPE_ARG_TYPES.
2729 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
2730 (ipa_write_node_info): Stream type of the actual argument.
2731 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
2733 2017-01-09 Martin Liska <mliska@suse.cz>
2736 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
2737 (lookup_compiler): Do not show error message with have_E.
2739 2017-01-09 Jakub Jelinek <jakub@redhat.com>
2741 PR tree-optimization/78938
2742 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
2743 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
2744 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
2745 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
2748 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2750 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
2753 2017-01-09 Richard Biener <rguenther@suse.de>
2755 PR tree-optimization/78997
2756 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
2757 name condition properly.
2759 2017-01-09 Richard Biener <rguenther@suse.de>
2762 * dwarf2out.c (is_cxx): New overload with context.
2763 (is_naming_typedef_decl): Use it.
2765 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
2767 * invoke.texi (Option Summary): Correct spacing in option lists
2768 and add line breaks to fix over-long lines.
2770 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
2774 * extend.texi (Common Variable Attributes): Add xref to GCC
2775 Internals manual to explain mode attribute keywords.
2777 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
2780 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
2781 and Preprocessor Options.
2782 (Options for Linking): Document -pthread here....
2783 (RS/6000 and PowerPC Options): ...not here.
2784 (Solaris 2 Options): ...or here.
2785 * doc/cppopts.texi: Document -pthread.
2787 2017-01-08 Martin Sebor <msebor@redhat.com>
2790 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
2791 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
2792 New member functions.
2793 (format_directive): Used them.
2795 (pass_sprintf_length::handle_gimple_call): Same.
2796 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
2797 to avoid truncation for any argument.
2798 (extract_affine_mul): Same.
2799 * tree.c (get_file_function_name): Same.
2801 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
2804 * predict.def (PRED_INDIR_CALL): Set to 86.
2806 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
2808 PR preprocessor/54124
2809 * doc/cppopts.texi: Reformat -d subtable to list the full name
2810 of the options. Add cross-reference to the docs for the general
2811 compiler -d options.
2812 * doc/invoke.texi (Developer Options): Add cross-reference to the
2813 preprocessor-specific -d option documentation.
2815 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
2817 PR preprocessor/13498
2818 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
2819 redudant material, and reflect new command-line options.
2820 (System Headers): Likewise.
2822 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
2824 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
2825 -isystem, and -idirafter. Copy-edit.
2826 * doc/cppopts.texi: Copy-edit. Remove contradiction about
2827 default for -ftrack-macro-expansion. Delete obsolete and
2828 badly-formatted implementation details about -fdebug-cpp output.
2829 * doc/cppwarnopts.texi: Copy-edit.
2831 2017-01-07 David Malcolm <dmalcolm@redhat.com>
2834 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
2835 that the transition from a max line width >= 1<<10 to narrower
2836 lines works correctly.
2838 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
2840 * doc/options.texi (PerFunction): New.
2841 * opt-functions.awk (switch_flags): Map both Optimization and
2842 PerFunction to CL_OPTIMIZATION.
2843 * opth-gen.awk: Test for PerFunction flag along with
2845 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
2846 it only when the latter is present. Skip those that don't in
2847 the hash function generator.
2848 * common.opt (fvar-tracking): Mark as PerFunction instead of
2850 (fvar-tracking-assignments): Likewise.
2851 (fvar-tracking-assignments-toggle): Likewise.
2852 (fvar-tracking-uninit): Likewise.
2854 2017-01-07 Jakub Jelinek <jakub@redhat.com>
2856 PR translation/79018
2857 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
2860 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
2863 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
2864 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
2865 TARGET_LONG_JUMP_TABLE_OFFSETS.
2866 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
2867 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
2868 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
2869 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
2870 * config/m68k/m68k.md (tablejump expander): Likewise.
2871 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
2872 TARGET_LONG_JUMP_TABLE_OFFSETS.
2873 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
2874 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
2876 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2877 David Holsgrove <david.holsgrove@xilinx.com>
2879 * common/config/microblaze/microblaze-common.c
2880 (TARGET_EXCEPT_UNWIND_INFO): Remove.
2881 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
2883 * config/microblaze/microblaze.c (microblaze_must_save_register)
2884 (microblaze_expand_epilogue, microblaze_return_addr): Handle
2886 (microblaze_eh_return): New function.
2887 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
2888 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
2889 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
2890 * config/microblaze/microblaze.md (eh_return): New pattern.
2892 2017-01-06 Jakub Jelinek <jakub@redhat.com>
2894 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
2895 GCC_DIAGNOSTIC_STRINGIFY): Define.
2897 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
2899 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
2901 * config/arm/arm.md (<mcrr>): New.
2903 * config/arm/arm.c (arm_arch5te): New.
2904 (arm_option_override): Set arm_arch5te.
2905 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
2907 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
2908 (arm_mcrr_qualifiers): ... this. New.
2909 (MRRC_QUALIFIERS): Define to...
2910 (arm_mrrc_qualifiers): ... this. New.
2911 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
2913 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
2914 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
2915 (MRRCI, mrrc, MRRC): New.
2916 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
2917 VUNSPEC_MRRC2): New.
2919 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
2921 * config/arm/arm.md (<mcr>): New.
2923 * config/arm/arm.c (arm_coproc_builtin_available): Add
2924 support for mcr, mrc, mcr2 and mrc2.
2925 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
2926 (arm_mcr_qualifiers): ... this. New.
2927 (MRC_QUALIFIERS): Define to ...
2928 (arm_mrc_qualifiers): ... this. New.
2929 (MCR_QUALIFIERS): Define to ...
2930 (arm_mcr_qualifiers): ... this. New.
2931 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
2933 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
2934 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
2935 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
2938 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
2940 * config/arm/arm.md (*ldc): New.
2944 * config/arm/arm.c (arm_coproc_builtin_available): Add
2945 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
2946 (arm_coproc_ldc_stc_legitimate_address): New.
2947 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
2948 'qualifier_const_pointer'.
2949 (LDC_QUALIFIERS): Define to...
2950 (arm_ldc_qualifiers): ... this. New.
2951 (STC_QUALIFIERS): Define to...
2952 (arm_stc_qualifiers): ... this. New.
2953 * config/arm/arm-protos.h
2954 (arm_coproc_ldc_stc_legitimate_address): New.
2955 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
2956 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
2957 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
2958 stc2, stcl, stc2l): New.
2959 * config/arm/constraints.md (Uz): New.
2960 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
2961 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
2962 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
2963 VUNSPEC_STC2L): New.
2965 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
2967 * config/arm/arm.md (<cdp>): New.
2968 * config/arm/arm.c (neon_const_bounds): Rename this ...
2969 (arm_const_bounds): ... this.
2970 (arm_coproc_builtin_available): New.
2971 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
2972 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
2973 (CDP_QUALIFIERS): Define to...
2974 (arm_cdp_qualifiers): ... this. New.
2976 (arm_expand_builtin_args): Add case for 6 arguments.
2977 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
2978 (arm_const_bounds): ... this.
2979 (arm_coproc_builtin_available): New.
2980 * config/arm/arm_acle.h (__arm_cdp): New.
2982 * config/arm/arm_acle_builtins.def (cdp): New.
2984 * config/arm/iterators.md (CDPI,CDP,cdp): New.
2985 * config/arm/neon.md: Rename all 'neon_const_bounds' to
2987 * config/arm/types.md (coproc): New.
2988 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
2989 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
2990 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
2991 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
2993 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
2995 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
2996 (UBINOP_QUALIFIERS): New.
2998 (acle_builtin_data): New. Change comment.
2999 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
3000 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
3001 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
3002 arm_acle_builtins.def.
3003 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
3004 (arm_init_acle_builtins): New.
3005 (CRC32_BUILTIN): Remove.
3006 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
3007 crc32cb, crc32ch and crc32cw.
3008 (arm_init_crc32_builtins): Remove.
3009 (arm_init_builtins): Use arm_init_acle_builtins rather
3010 than arm_init_crc32_builtins.
3011 (arm_expand_acle_builtin): New.
3012 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
3013 * config/arm/arm_acle_builtins.def: New.
3015 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
3017 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
3018 (arm_builtin_datum): ... this.
3019 (arm_init_neon_builtin): Rename to ...
3020 (arm_init_builtin): ... this. Add a new parameters PREFIX
3021 and USE_SIG_IN_NAME.
3022 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
3023 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
3024 'arm_builtin_datum'.
3025 (arm_init_vfp_builtins): Likewise.
3026 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
3027 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
3028 (arm_expand_neon_args): Rename to ...
3029 (arm_expand_builtin_args): ... this. Rename builtin_arg
3030 enum values and differentiate between ARG_BUILTIN_MEMORY
3031 and ARG_BUILTIN_NEON_MEMORY.
3032 (arm_expand_neon_builtin_1): Rename to ...
3033 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
3034 values, arm_expand_builtin_args and add bool parameter NEON.
3035 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
3036 (arm_expand_vfp_builtin): Likewise.
3037 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
3039 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
3042 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
3043 * predict.c (tree_estimate_probability_bb): Reverse direction of
3044 polymorphic call predictor.
3046 2017-01-06 David Malcolm <dmalcolm@redhat.com>
3048 * passes.c (execute_one_pass): Split out pass-skipping logic into...
3049 (determine_pass_name_match): ...this new function and...
3050 (should_skip_pass_p): ...this new function.
3052 2017-01-06 Nathan Sidwell <nathan@acm.org>
3054 * ipa-visibility.c (function_and_variable_visibility): Reformat
3055 comments and long lines. Remove extrneous if.
3056 * symtab.c (symtab_node::make_decl_local): Fix code format.
3057 (symtab_node::set_section_for_node): Fix comment typo.
3059 2017-01-06 Martin Liska <mliska@suse.cz>
3062 * lra-constraints.c: Rename invariant to lra_invariant.
3063 * predict.c (set_even_probabilities): Initialize e to NULL.
3065 2017-01-05 Martin Sebor <msebor@redhat.com>
3067 PR tree-optimization/78910
3068 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
3069 (format_integer): Correct off-by-one error in the handling
3070 of precision with negative numbers in signed conversions..
3072 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
3074 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
3076 2017-01-05 Jakub Jelinek <jakub@redhat.com>
3078 PR tree-optimization/71016
3079 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
3080 factor_out_conditional_conversion. Formatting fix.
3081 (factor_out_conditional_conversion): Add cond_stmt argument.
3082 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
3083 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
3086 2017-01-05 David Malcolm <dmalcolm@redhat.com>
3088 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
3089 read-rtl-function.o, and selftest-rtl.o.
3090 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
3091 (selftest::aarch64_test_loading_full_dump): New function.
3092 (selftest::aarch64_run_selftests): New function.
3093 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
3094 selftest::aarch64_run_selftests.
3095 * config/i386/i386.c
3096 (selftest::ix86_test_loading_dump_fragment_1): New function.
3097 (selftest::ix86_test_loading_call_insn): New function.
3098 (selftest::ix86_test_loading_full_dump): New function.
3099 (selftest::ix86_test_loading_unspec): New function.
3100 (selftest::ix86_run_selftests): Call the new functions.
3101 * emit-rtl.c (maybe_set_max_label_num): New function.
3102 * emit-rtl.h (maybe_set_max_label_num): New decl.
3103 * function.c (instantiate_decls): Guard call to
3104 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
3105 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
3107 * gensupport.c (gen_reader::gen_reader): Pass "false"
3108 for new "compact" param of rtx_reader.
3109 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
3110 rather than an empty string for NULL strings.
3111 * read-md.c: Potentially include config.h rather than bconfig.h.
3112 Wrap include of errors.h with #ifdef GENERATOR_FILE.
3113 (have_error): New global, copied from errors.c.
3114 (md_reader::read_name): Rename to...
3115 (md_reader::read_name_1): ...this, adding "out_loc" param,
3116 and converting "missing name or number" to returning false, rather
3118 (md_reader::read_name): Reimplement in terms of read_name_1.
3119 (md_reader::read_name_or_nil): New function.
3120 (md_reader::read_string): Handle "(nil)" by returning NULL.
3121 (md_reader::md_reader): Add new param "compact".
3122 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
3123 (md_reader::read_file): New method.
3124 * read-md.h (md_reader::md_reader): Add new param "compact".
3125 (md_reader::read_file): New method.
3126 (md_reader::is_compact): New accessor.
3127 (md_reader::read_name): Convert return type from void to file_location.
3128 (md_reader::read_name_or_nil): New decl.
3129 (md_reader::read_name_1): New decl.
3130 (md_reader::m_compact): New field.
3131 (noop_reader::noop_reader): Pass "false" for new "compact" param
3133 (rtx_reader::rtx_reader): Add new "compact" param.
3134 (rtx_reader::read_rtx_operand): Make virtual and convert return
3135 type from void to rtx.
3136 (rtx_reader::read_until): New decl.
3137 (rtx_reader::handle_any_trailing_information): New virtual function.
3138 (rtx_reader::postprocess): New virtual function.
3139 (rtx_reader::finalize_string): New virtual function.
3140 (rtx_reader::m_in_call_function_usage): New field.
3141 (rtx_reader::m_reuse_rtx_by_id): New field.
3142 * read-rtl-function.c: New file.
3143 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
3144 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
3145 (selftest::verify_three_block_rtl_cfg): New decl.
3146 * read-rtl-function.h: New file.
3147 * read-rtl.c: Potentially include config.h rather than bconfig.h.
3148 For host, include function.h, memmodel.h, and emit-rtl.h.
3149 (one_time_initialization): New function.
3150 (struct compact_insn_name): New struct.
3151 (compact_insn_names): New array.
3152 (find_code): Handle insn codes in compact dumps.
3153 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
3154 (bind_subst_iter_and_attr): Likewise.
3155 (add_condition_to_string): Likewise.
3156 (add_condition_to_rtx): Likewise.
3157 (apply_attribute_uses): Likewise.
3158 (add_current_iterators): Likewise.
3159 (apply_iterators): Likewise.
3160 (initialize_iterators): Guard usage of apply_subst_iterator with
3161 #ifdef GENERATOR_FILE.
3162 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
3163 (md_reader::read_mapping): Likewise.
3164 (add_define_attr_for_define_subst): Likewise.
3165 (add_define_subst_attr): Likewise.
3166 (read_subst_mapping): Likewise.
3167 (check_code_iterator): Likewise.
3168 (rtx_reader::read_rtx): Likewise. Move one-time initialization
3170 (one_time_initialization): New function.
3171 (rtx_reader::read_until): New method.
3172 (read_flags): New function.
3173 (parse_reg_note_name): New function.
3174 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
3175 Handle reuse_rtx ids.
3176 Wrap iterator lookup within #ifdef GENERATOR_FILE.
3177 Add parsing support for RTL dumps, mirroring the special-cases in
3178 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
3179 values, and calling handle_any_trailing_information.
3180 (rtx_reader::read_rtx_operand): Convert return type from void
3181 to rtx, returning return_rtx. Handle case 'e'. Call
3182 finalize_string on XSTR and XTMPL fields.
3183 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
3184 "(nil)" values were omitted. Call the postprocess vfunc on the
3186 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
3187 class ctor. Initialize m_in_call_function_usage. Call
3188 one_time_initialization.
3189 * rtl-tests.c (selftest::test_uncond_jump): Call
3190 set_new_first_and_last_insn.
3191 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
3192 * selftest-rtl.c: New file.
3193 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
3194 (selftest::get_insn_by_uid): New decl.
3195 * selftest-run-tests.c (selftest::run_tests): Call
3196 read_rtl_function_c_tests.
3197 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
3198 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
3201 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
3203 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
3204 operands in a special way. Assert that pos+len <= mode precision.
3206 2017-01-05 Jakub Jelinek <jakub@redhat.com>
3208 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
3209 3 argument Alias with unlimited for the negative form.
3210 (fno-vect-cost-model): Removed.
3212 2017-01-05 Martin Liska <mliska@suse.cz>
3214 * hsa-gen.c (gen_hsa_divmod): New function.
3215 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
3217 2017-01-05 Martin Liska <mliska@suse.cz>
3220 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
3223 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3225 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
3226 small constant length operands.
3228 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3230 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
3231 between loop iterations.
3233 2017-01-05 Martin Liska <mliska@suse.cz>
3236 * gimplify.c (gimplify_decl_expr): Compare to
3237 asan_poisoned_variables instread of checking flags.
3238 (gimplify_target_expr): Likewise.
3239 (gimplify_expr): Likewise.
3240 (gimplify_function_tree): Conditionally initialize
3241 asan_poisoned_variables.
3243 2017-01-04 Jeff Law <law@redhat.com>
3245 PR tree-optimizatin/78812
3246 * rtl.h (contains_mem_rtx_p): Prototype.
3247 * ifcvt.c (containts_mem_rtx_p): Move from here to...
3248 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
3249 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
3250 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
3251 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
3253 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
3255 * input.c (assert_char_at_range): Default-initialize actual_range.
3257 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
3259 * df-scan.c (df_ref_create_structure): Make regno unsigned,
3260 to match the caller.
3262 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
3264 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
3265 insns after final jump in test to emit dummy move.
3267 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
3269 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
3270 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
3272 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
3274 * multiple_target.c (create_dispatcher_calls): Init e_next.
3275 * tree-ssa-loop-split.c (split_loop): Init border.
3276 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
3279 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
3284 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
3285 (altivec_register_operand): Do not return true if the operand
3286 contains a SUBREG mixing SImode and SFmode.
3287 (vsx_register_operand): Likewise.
3288 (vsx_reg_sfsubreg_ok): New predicate.
3289 (vfloat_operand): Do not return true if the operand contains a
3290 SUBREG mixing SImode and SFmode.
3291 (vint_operand): Likewise.
3292 (vlogical_operand): Likewise.
3293 (gpc_reg_operand): Likewise.
3294 (int_reg_operand): Likewise.
3295 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
3296 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
3297 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
3299 (rs6000_emit_move_si_sf_subreg): New helper function.
3300 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
3301 fixup SUBREGs involving SImode and SFmode.
3302 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
3303 numbers for the new peephole2 optimization.
3304 (peephole2 for SFmode unions): New peephole2 to optimize cases in
3305 the GLIBC math library that do AND/IOR/XOR operations on single
3306 precision floating point.
3307 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
3308 target macros to say whether we need to avoid SUBREGs mixing
3310 (TARGET_ALLOW_SF_SUBREG): Likewise.
3311 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
3312 (UNSPEC_SI_FROM_SF): Likewise.
3313 (iorxor): Change spacing.
3314 (and_ior_xor): New iterator for AND, IOR, and XOR.
3315 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
3316 (movdi_from_sf_zero_ext): Likewise.
3317 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
3318 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
3319 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
3320 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
3321 (fms<mode>4): Likewise.
3322 (fnma<mode>4): Likewise.
3323 (fnms<mode>4): Likewise.
3324 (nfma<mode>4): Likewise.
3325 (nfms<mode>4): Likewise.
3327 2017-01-04 Marek Polacek <polacek@redhat.com>
3330 * doc/invoke.texi: Document -Wpointer-compare.
3332 2017-01-04 Jakub Jelinek <jakub@redhat.com>
3334 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
3337 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
3338 descriptions for -gdwarf-5 and emit them as uleb128 instead of
3341 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
3344 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
3345 documentation of the powerpc_popcntb_ok attribute.
3346 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3347 code to issue warning messages if a requested CPU configuration is
3348 not supported by the binary (assembler and loader) toolchain.
3349 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
3350 made to define a built-in function that has been disabled.
3351 (paired_init_builtins): Add assertion to prevent ICE if attempt is
3352 made to define a built-in function that has been disabled.
3353 (altivec_init_builtins): Add comment explaining why definition
3354 of the DST built-in functions is not preceded by an assertion
3355 check. Add assertions to prevent ICE if attempts are made to
3356 define an altivec predicate or an abs* built-in function that has
3358 (htm_init_builtins): Add comment explaining why definition of the
3359 htm built-in functions is not preceded by an assertion check.
3361 2017-01-04 Jeff Law <law@redhat.com>
3363 PR tree-optimizatin/67955
3364 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
3365 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
3366 the points-to solution does not include pt_null. Use DECL_PT_UID
3369 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
3371 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
3372 Use gen_int_mode instead of gen_lopwart for const_int operands.
3374 2017-01-04 Jakub Jelinek <jakub@redhat.com>
3376 PR tree-optimization/71563
3377 * match.pd: Simplify X << Y into X if Y is known to be 0 or
3378 out of range value - has low bits known to be zero.
3380 2017-01-04 Alan Modra <amodra@gmail.com>
3382 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
3383 * configure: Regenerate.
3384 * config.in: Regenerate.
3386 2017-01-04 Jakub Jelinek <jakub@redhat.com>
3389 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
3390 a substring of the message, but strcmp with the whole message. Ifdef
3391 ENABLE_NLS, translate the message first using dgettext.
3393 2017-01-03 Jeff Law <law@redhat.com>
3395 PR tree-optimizatin/78856
3396 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
3397 (mark_threaded_blocks): Remove code to truncate thread paths that
3398 cross multiple loop headers. Instead invalidate the cached loop
3399 iteration information and handle case of a thread path walking
3400 into an irreducible region.
3402 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
3405 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
3406 assertions. Add support for doing the signbit if the IEEE 128-bit
3407 floating point value is in a GPR.
3408 * config/rs6000/rs6000.md (Fsignbit): Delete.
3409 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
3410 Update the length attribute if the value is in a GPR.
3411 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
3412 the sign or zero extension instruction, since the value is always 0/1.
3413 (signbit<mode>2_dm2): Delete using <Fsignbit>.
3416 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
3417 extracting SImode to a GPR register so that we can generate a
3418 store, limit the vector to be in a traditional Altivec register
3419 for the vextuwrx instruction.
3421 2017-01-03 Ian Lance Taylor <iant@google.com>
3423 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
3425 2017-01-03 Martin Sebor <msebor@redhat.com>
3427 PR tree-optimization/78696
3428 * gimple-ssa-sprintf.c (format_floating): Correct handling of
3429 precision. Use MPFR for %f for greater fidelity. Correct handling
3431 (pass_sprintf_length::compute_format_length): Set width and precision
3432 specified by asrerisk to void_node for vararg functions.
3433 (try_substitute_return_value): Adjust dump output.
3435 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
3437 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
3439 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
3441 * doc/invoke.texi (SPARC options): Document -mlra as the default.
3442 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
3443 -mlra/-mno-lra was passed to the compiler.
3445 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
3447 PR rtl-optimization/65618
3448 * emit-rtl.c (try_split): Move initialization of "before" and
3449 "after" to just before the call to emit_insn_after_setloc.
3451 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
3453 * doc/md.texi (Standard Names): Remove reference to Java frontend.
3455 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
3457 * dwarf2out.c (gen_enumeration_type_die): When
3458 -gno-strict-dwarf, add a DW_AT_encoding attribute.
3460 2017-01-03 Jakub Jelinek <jakub@redhat.com>
3462 PR tree-optimization/78965
3463 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
3464 Change first argument from const call_info & to call_info &. For %n
3465 set info.nowrite to false.
3468 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
3469 possibly throwing calls.
3471 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
3472 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
3473 and fns handling, rather than in a separate case SSA_NAME.
3475 2017-01-02 Jeff Law <law@redhat.com>
3477 * config/darwin-driver.c (darwin_driver_init): Const-correctness
3478 fixes for first_period and second_period variables.
3480 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
3483 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
3484 (*insvqi_1): New insn pattern.
3485 (*insvqi_1_mem_rex64): Ditto.
3487 (*insvqi_3): Rename from *insvqi.
3489 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
3491 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
3493 * doc/cfg.texi (Edges): Remove reference to Java.
3494 (Maintaining the CFG): Ditto.
3496 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
3499 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
3500 transparent aliases.
3502 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
3505 * predict.def (PRED_CALL): Update hitrate.
3506 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
3507 * predict.c (tree_estimate_probability_bb): Split CALL predictor
3508 into direct/indirect/polymorphic variants.
3510 2017-01-01 Jakub Jelinek <jakub@redhat.com>
3512 Update copyright years.
3514 * gcc.c (process_command): Update copyright notice dates.
3515 * gcov-dump.c (print_version): Ditto.
3516 * gcov.c (print_version): Ditto.
3517 * gcov-tool.c (print_version): Ditto.
3518 * gengtype.c (create_file): Ditto.
3519 * doc/cpp.texi: Bump @copying's copyright year.
3520 * doc/cppinternals.texi: Ditto.
3521 * doc/gcc.texi: Ditto.
3522 * doc/gccint.texi: Ditto.
3523 * doc/gcov.texi: Ditto.
3524 * doc/install.texi: Ditto.
3525 * doc/invoke.texi: Ditto.
3527 Copyright (C) 2017 Free Software Foundation, Inc.
3529 Copying and distribution of this file, with or without modification,
3530 are permitted in any medium without royalty provided the copyright
3531 notice and this notice are preserved.