1 2017-02-10 Jakub Jelinek <jakub@redhat.com>
3 PR tree-optimization/79411
4 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
5 stmt operands are SSA_NAMEs used in abnormal phis.
6 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
9 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
12 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
15 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
17 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
19 2017-02-09 Jakub Jelinek <jakub@redhat.com>
21 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
25 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
26 not arbitrary TREE_CONSTANT.
29 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30 "omp declare target link" attribute unless is_global_var.
31 * omp-offload.c (find_link_var_op): Likewise.
33 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
34 Chung-Lin Tang <cltang@codesourcery.com>
36 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38 (gimplify_adjust_omp_clauses): Don't delete TILE.
39 (gimplify_omp_for): Deal with TILE.
40 * internal-fn.c (expand_GOACC_TILE): New function.
41 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
43 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
44 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
46 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
47 avoid DIV for outermost collapse var.
48 (expand_oacc_for): Insert tile element loop as needed. Adjust.
49 Remove out of date comments, fix whitespace.
50 * omp-general.c (omp_extract_for_data): Deal with tiling.
51 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
52 adjust OLF_DIM_BASE value.
53 (struct omp_for_data): Add tiling field.
54 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
55 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
56 for auto loops. Remove default auto determining, moved to
57 oacc_loop_fixed_partitions.
58 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
59 stmts, add e_mask field.
60 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
61 (oacc_thread_numbers): Use oacc_dim_call.
62 (oacc_xform_tile): New.
63 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
64 (finish_oacc_loop): Adjust for ifns vector.
65 (oacc_loop_discover_walk): Append loop abstraction sites to list,
66 add case for GOACC_TILE fns.
67 (oacc_loop_xform_loop): Delete.
68 (oacc_loop_process): Iterate over call list directly, and add
69 handling for GOACC_TILE fns.
70 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
72 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
73 vector partitioning to outer loops. Assign 2 partitions to loops
74 when available. Add TILE handling.
75 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
76 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
77 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
78 * tree.c (omp_clause_num_ops): Adjust TILE ops.
79 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
81 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
83 * configure.ac (ACX_BUGURL): Update.
84 * configure: Regenerate.
86 2017-02-09 Richard Biener <rguenther@suse.de>
88 PR tree-optimization/69823
89 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
90 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
92 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
94 * config/arc/arc-c.def: Add __NPS400__ definition.
95 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
96 (TARGET_NPS400): Define.
98 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
100 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
102 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
104 (arc_cpu_types): Fill the arch_info field with a pointer into the
105 arc_arch_types table.
106 (arc_selected_cpu): Declare.
107 * config/arc/arc.c (arc_selected_cpu): Make global.
108 (arc_selected_arch): Delete.
109 (arc_base_cpu): Delete.
110 (arc_override_options): Remove references to deleted variables,
111 update access to arch information.
112 (ARC_OPT): Update access to arch information.
113 (ARC_OPTX): Likewise.
114 * config/arc/arc.h (arc_base_cpu): Remove declaration.
115 (TARGET_ARC600): Update access to arch information.
116 (TARGET_ARC601): Likewise.
117 (TARGET_ARC700): Likewise.
118 (TARGET_EM): Likewise.
119 (TARGET_HS): Likewise.
120 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
123 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
126 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
127 condition/operands for integer GE/LE/GEU/LEU operations.
129 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
132 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
135 2017-02-08 Martin Jambor <mjambor@suse.cz>
138 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
139 whether allocation happened.
140 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
141 nothing was allocated.
143 2017-02-08 Jakub Jelinek <jakub@redhat.com>
145 PR tree-optimization/79408
146 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
147 constant, but SSA_NAME with a known integer range, use the minimum
148 of that range instead of op1 to determine if modulo can be replaced
149 with its first operand.
151 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
153 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
155 2017-02-08 Richard Biener <rguenther@suse.de>
157 PR tree-optimization/71824
158 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
159 Check all loops contained in the merged region.
161 2017-02-07 Andrew Pinski <apinski@cavium.com>
163 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
165 2017-02-07 Andrew Pinski <apinski@cavium.com>
167 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
168 (thunderxt88): Likewise.
169 (thunderxt81): Disable LSE and change v8.1 to v8.
170 (thunderxt83): Likewise.
172 2017-02-07 Jakub Jelinek <jakub@redhat.com>
173 Richard Biener <rguenther@suse.de>
176 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
177 type from int to size_t.
178 * ira-costs.c (struct_costs_size): Change type from int to size_t.
180 2017-02-07 Jakub Jelinek <jakub@redhat.com>
182 PR rtl-optimization/79386
183 * cprop.c (bypass_conditional_jumps): Initialize
184 bypass_last_basic_block already before splitting bbs after
185 unconditional traps...
186 (bypass_conditional_jumps): ... rather than here.
189 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
190 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
191 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
192 fix -masm=intel patterns.
194 2017-02-07 Richard Biener <rguenther@suse.de>
196 PR tree-optimization/79256
198 * builtins.c (get_object_alignment_2): Use min_align_of_type
199 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
200 and ADJUST_FIELD_ALIGN.
202 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
204 * doc/tm.texi: Regenerate.
205 * stor-layout.c (layout_decl): Adjust.
206 (update_alignment_for_field): Likewise.
207 (place_field): Likewise.
208 (min_align_of_type): Likewise.
209 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
210 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
211 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
212 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
213 * config/frv/frv.c (frv_adjust_field_align): Likewise.
214 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
215 * config/i386/i386.c (x86_field_alignment): Likewise.
216 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
217 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
218 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
219 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
220 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
221 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
225 2017-01-30 Richard Biener <rguenther@suse.de>
227 PR tree-optimization/79256
228 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
229 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
232 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
234 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
235 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
236 builtins to SImode and emit a zero-extend, if necessary.
238 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
240 * docs/invoke.texi (RISC-V Options): Alphabetize.
242 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
244 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
247 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
249 * config/riscv/riscv.c: New file.
250 * gcc/common/config/riscv/riscv-common.c: Likewise.
251 * config.gcc: Likewise.
252 * config/riscv/constraints.md: Likewise.
253 * config/riscv/elf.h: Likewise.
254 * config/riscv/generic.md: Likewise.
255 * config/riscv/linux.h: Likewise.
256 * config/riscv/multilib-generator: Likewise.
257 * config/riscv/peephole.md: Likewise.
258 * config/riscv/pic.md: Likewise.
259 * config/riscv/predicates.md: Likewise.
260 * config/riscv/riscv-builtins.c: Likewise.
261 * config/riscv/riscv-c.c: Likewise.
262 * config/riscv/riscv-ftypes.def: Likewise.
263 * config/riscv/riscv-modes.def: Likewise.
264 * config/riscv/riscv-opts.h: Likewise.
265 * config/riscv/riscv-protos.h: Likewise.
266 * config/riscv/riscv.h: Likewise.
267 * config/riscv/riscv.md: Likewise.
268 * config/riscv/riscv.opt: Likewise.
269 * config/riscv/sync.md: Likewise.
270 * config/riscv/t-elf-multilib: Likewise.
271 * config/riscv/t-linux: Likewise.
272 * config/riscv/t-linux-multilib: Likewise.
273 * config/riscv/t-riscv: Likewise.
274 * configure.ac: Likewise.
275 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
276 Waterman as RISC-V maintainers.
277 * doc/install.texi: Add RISC-V entries.
278 * doc/invoke.texi: Add RISC-V options section.
279 * doc/md.texi: Add RISC-V constraints section.
280 * configure: Regenerated.
282 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
285 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
286 false values to be constant vectors with all 0 or all 1 bits set.
287 (vcondu<mode><mode>): Likewise.
288 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
290 (fpmask_comparison_operator): Update comment.
291 (vecint_comparison_operator): New predicate.
292 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
293 vector conditionals when the true and false values are constant
294 vectors with all 0 bits or all 1 bits set.
296 2017-02-06 Martin Sebor <msebor@redhat.com>
298 PR tree-optimization/79376
299 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
301 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
303 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
304 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
305 to simplify split condition.
307 2017-02-06 Jakub Jelinek <jakub@redhat.com>
309 * omp-expand.c (oxpand_omp_atomic_fetch_op,
310 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
313 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
315 PR rtl-optimization/68664
316 * target.def (can_speculate_insn): New hook.
317 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
318 * doc/tm.texi: Regenerate.
319 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
320 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
321 (rs6000_sched_can_speculate_insn): New function.
323 2017-02-06 Jakub Jelinek <jakub@redhat.com>
325 PR tree-optimization/79284
326 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
327 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
328 vectorizable_mask_load_store, vectorizable_operation,
329 vect_is_simple_cond, get_same_sized_vectype): Use it instead
330 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
331 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
332 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
333 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
334 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
335 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
336 is_gimple_assign (stmt). Replace another such test with
337 is_gimple_assign (stmt).
339 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
342 * config/avr/avr.c (rtl-iter.h): Include it.
343 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
344 (avr_legitimate_combined_insn): ...and implementation.
346 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
348 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
349 * config/s390/s390.c (s390_const_operand_ok)
350 (s390_canonicalize_comparison, s390_extract_part)
351 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
352 (s390_contiguous_bitmask_p, s390_rtx_costs)
353 (legitimize_pic_address): Likewise.
354 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
355 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
356 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
357 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
358 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
360 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
362 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
363 REGNO($0) == REGNO($1).
365 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
367 * config/s390/linux.h(SIZE_TYPE): Add comment.
369 2017-02-06 Julian Brown <julian@codesourcery.com>
370 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
371 Virendra Pathak <virendra.pathak@broadcom.com>
373 * config/aarch64/aarch64-cores.def: Change the scheduler
375 * config/aarch64/aarch64.md: Include thunderx2t99.md.
376 * config/aarch64/thunderx2t99.md: New file.
378 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
380 * doc/standards.texi (Go Language): Update link to language
383 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
385 * tree-eh.c (lower_resx): Sanitize profile.
386 (cleanup_empty_eh_move_lp): Likewise.
388 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
391 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
393 * cfgloopmanip.h (loop_version): Update prototype.
394 * modulo-sched.c (sms_schedule): Update call of loop_version.
395 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
396 * tree-parloops.c (gen_parallel_loop): Likewise.
397 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
398 * tree-ssa-loop-split.c (split_loop): Likewise.
399 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
400 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
402 2017-02-05 Martin Liska <mliska@suse.cz>
405 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
407 (print_operand_address): Initialize a struct to zero.
409 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
411 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
412 garbage collector only in textual form.
414 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
416 * doc/extend.texi (x86 specific memory model extensions for
417 transactional memory): Simplify a phrase.
419 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
422 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
423 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
424 (atomic_storedi_1): Likewise.
426 2017-02-04 Jakub Jelinek <jakub@redhat.com>
428 PR tree-optimization/79338
429 * tree-parloops.c (gather_scalar_reductions): Don't call
430 vect_analyze_loop_form for loop->inner before destroying loop's
433 2017-02-03 Martin Sebor <msebor@redhat.com>
435 PR tree-optimization/79327
436 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
437 when precision has resulted in leading zeros.
438 (format_integer): Adjust the likely counter to assume an unknown
439 argument that may be zero is non-zero.
441 2017-02-03 Jason Merrill <jason@redhat.com>
444 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
445 avoid copying non-taken branch.
447 2017-02-03 Jakub Jelinek <jakub@redhat.com>
449 PR tree-optimization/79340
450 * tree-vect-loop.c (vectorizable_reduction): Release
451 vec_defs elements after safe_splicing them into other vectors.
454 PR tree-optimization/79327
455 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
456 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
458 (format_integer): Use wide_int_to_tree instead of build_int_cst
459 + to_?hwi. If argmin is NULL, just set argmin and argmax to
460 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
461 of shortest and longest sequence.
463 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
465 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
466 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
468 2017-02-03 Walter Lee <walt@tilera.com>
471 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
472 after initial stackframe link reg save.
473 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
475 2017-02-03 Jakub Jelinek <jakub@redhat.com>
478 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
479 wu for stxssp alternative.
481 2017-02-03 Martin Sebor <msebor@redhat.com>
483 PR tree-optimization/79352
484 * gimple-fold.c (get_range_strlen): Add argument.
485 (get_range_strlen): Change return type to bool.
486 (get_maxval_strlen): Pass in a dummy argument.
487 * gimple-fold.h (get_range_strlen): Change return type to bool.
488 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
489 * tree.h (array_at_struct_end_p): Add argument.
490 * tree.c (array_at_struct_end_p): Handle it.
492 2017-02-03 Martin Liska <mliska@suse.cz>
495 * multiple_target.c (create_dispatcher_calls): Redirect edge
496 from a caller of a dispatcher.
497 (expand_target_clones): Make the clones local.
498 (ipa_target_clone): Do both target clones and resolvers.
499 (ipa_dispatcher_calls): Remove the pass.
500 (pass_dispatcher_calls::gate): Likewise.
501 (make_pass_dispatcher_calls): Likewise.
502 * passes.def (pass_target_clone): Put as very first IPA early
505 2017-02-03 Martin Liska <mliska@suse.cz>
507 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
508 in case of a function with ifunc attribute.
510 2017-02-03 Martin Liska <mliska@suse.cz>
512 * cgraph.c (cgraph_node::dump): Dump function version info.
513 * symtab.c (symtab_node::dump_base): Add missing new line.
515 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
517 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
518 (ifcombine_ifandif): Use it.
520 2017-02-03 Martin Liska <mliska@suse.cz>
522 * doc/invoke.texi: Document default value for
523 use-after-scope-direct-emission-threshold.
525 2017-02-03 Martin Liska <mliska@suse.cz>
527 PR tree-optimization/79339
528 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
529 (format_floating): Likewise.
531 2017-02-03 Martin Liska <mliska@suse.cz>
534 * ipa-prop.c (ipa_node_params_t::insert): Remove current
536 (ipa_node_params_t::remove): Likewise.
537 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
538 initialization from removed ipa_node_params_t::insert.
539 (ipa_node_params::~ipa_node_params): Move from removed
540 ipa_node_params_t::release.
541 * symbol-summary.h (symbol_summary::m_released): New member.
542 Do not release a summary twice. Do not allow to call finalizer
543 for types of a summary that live in GGC memory.
545 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
547 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
550 2017-02-02 Martin Sebor <msebor@redhat.com>
553 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
554 (format_string): Tighten up the range of output for non-constant
555 strings and correct the expected range for wide non-constant strings.
557 2017-02-02 Martin Sebor <msebor@redhat.com>
559 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
562 * doc/invoke.texi (-fdump-rtl-): Remove pass-specific options from
564 (-fdump-tree-@var): Add to index and document how to come up
565 with pass-specific option and dump file names.
566 (-fdump-passes): Clarify where to look for output.
568 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
571 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
572 statistics of the analyzed path; allow threading for speed when
573 any of BBs along the path are optimized for speed.
575 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
578 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
579 settings of the virtual registers.
582 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
584 * explow.c (get_dynamic_stack_size): Take known alignment of stack
585 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
588 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
590 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
591 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
593 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
595 * config/s390/s390.md: Add missing comments with the expanded
597 * config/s390/vector.md: Likewise.
598 * config/s390/vx-builtins.md: Likewise.
600 2017-02-02 Jakub Jelinek <jakub@redhat.com>
603 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
604 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
605 conditions on a single line.
607 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
609 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
610 __S390_VX__ to __VX__.
612 2017-02-01 Andrew Pinski <apinski@cavium.com>
614 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
615 stmt_info to record_stmt_cost.
616 (vect_get_known_peeling_cost): Pass stmt_info if known to
618 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
619 cpu_vector_cost field into
620 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
621 field into vec_int_stmt_cost and vec_fp_stmt_cost.
622 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
623 splitting of scalar_stmt_cost and vec_stmt_cost.
624 (thunderx_vector_cost): Likewise.
625 (cortexa57_vector_cost): LIkewise.
626 (exynosm1_vector_cost): Likewise.
627 (xgene1_vector_cost): Likewise.
628 (thunderx2t99_vector_cost): Improve after the splitting of the two
630 (aarch64_builtin_vectorization_cost): Update for the splitting of
631 scalar_stmt_cost and vec_stmt_cost.
633 2017-02-01 Torvald Riegel <triegel@redhat.com>
634 Richard Henderson <rth@redhat.com>
636 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
637 conditional on existance of a fast atomic load.
638 * optabs-query.c (can_atomic_load_p): New function.
639 * optabs-query.h (can_atomic_load_p): Declare it.
640 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
641 no fast atomic load is available for the particular size of access.
642 (expand_atomic_compare_and_swap): Likewise.
643 (expand_atomic_load): Likewise.
644 (expand_atomic_store): Likewise.
645 (expand_atomic_fetch_op): Likewise.
646 * testsuite/lib/target-supports.exp
647 (check_effective_target_sync_int_128): Remove x86 because it provides
649 (check_effective_target_sync_int_128_runtime): Likewise.
651 2017-02-01 Richard Biener <rguenther@suse.de>
653 * graphite.c: Include tree-vectorizer.h for find_loop_location.
654 (graphite_transform_loops): Provide opt-info for optimized nests.
655 * tree-parloop.c (parallelize_loops): Provide opt-info for
658 2017-02-01 Richard Biener <rguenther@suse.de>
661 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
664 2017-02-01 Richard Biener <rguenther@suse.de>
666 PR tree-optimization/71824
667 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
668 Verify the loops are valid in the merged SESE region.
669 (scop_detection::can_represent_loop_1): Check analyzing the
670 evolution of the number of iterations in the region succeeds.
672 2017-01-31 Ian Lance Taylor <iant@golang.org>
674 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
675 REG_ARGS_SIZE note to 32-bit push insns and call insn.
677 2017-01-31 David Malcolm <dmalcolm@redhat.com>
679 PR preprocessor/79210
680 * input.c (get_substring_ranges_for_loc): Replace line_width
681 assertion with error-handling.
683 2017-01-31 Richard Biener <rguenther@suse.de>
685 PR tree-optimization/77318
686 * graphite-sese-to-poly.c (extract_affine): Fix assert.
687 (create_pw_aff_from_tree): Take loop parameter.
688 (add_condition_to_pbb): Pass loop of the condition to
689 create_pw_aff_from_tree.
691 2017-01-31 Jakub Jelinek <jakub@redhat.com>
693 * config/s390/s390.c (s390_asan_shadow_offset): New function.
694 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
696 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
700 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
702 (convert_int_to_float128): Likewise.
703 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
704 (convert_int_to_float128): Likewise.
705 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
706 (UNSPEC_IEEE128_CONVERT): Likewise.
707 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
708 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
709 Use local variables for IBM extended format.
710 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
711 (fix_trunc<mode>si2_fprs): Likewise.
712 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
713 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
714 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
715 to know that we can now have integers of all sizes in vector
717 (fix<uns>_<mode>di2_hw): Likewise.
718 (float<uns>_<mode>si2_hw): Likewise.
719 (fix_<mode>si2_hw): Likewise.
720 (fixuns_<mode>si2_hw): Likewise.
721 (float<uns>_<mode>di2_hw): Likewise.
722 (float_<mode>di2_hw): Likewise.
723 (float_<mode>si2_hw): Likewise.
724 (floatuns_<mode>di2_hw): Likewise.
725 (floatuns_<mode>si2_hw): Likewise.
726 (xscvqp<su>wz_<mode>): Delete, no longer used.
727 (xscvqp<su>dz_<mode>): Likewise.
728 (xscv<su>dqp_<mode>): Likewise.
729 (ieee128_mfvsrd_64bit): Likewise.
730 (ieee128_mfvsrd_32bit): Likewise.
731 (ieee128_mfvsrwz): Likewise.
732 (ieee128_mtvsrw): Likewise.
733 (ieee128_mtvsrd_64bit): Likewise.
734 (ieee128_mtvsrd_32bit): Likewise.
736 2017-01-31 Martin Liska <mliska@suse.cz>
739 * ipa-prop.c (ipa_free_all_node_params): Call release method
740 instead of ~sumbol_summary to not to trigger double times
743 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
745 PR tree-optimization/71691
746 * bitmap.h (class auto_bitmap): New.
747 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
748 is_maybe_undefined instead of ssa_undefined_value_p.
750 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
752 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
753 __S390_ARCH_LEVEL__ to __ARCH__.
755 2017-01-31 Jakub Jelinek <jakub@redhat.com>
757 PR tree-optimization/79267
758 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
759 if should_remove_lhs_p is true.
761 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
764 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
765 (add_alignment_attribute): New.
766 (base_type_die): Add alignment attribute.
767 (subrange_type_die): Likewise.
768 (modified_type_die): Likewise.
769 (gen_array_type_die): Likewise.
770 (gen_descr_array_type_die: Likewise.
771 (gen_enumeration_type_die): Likewise.
772 (gen_subprogram_die): Likewise.
773 (gen_variable_die): Likewise.
774 (gen_field_die): Likewise.
775 (gen_ptr_to_mbr_type_die): Likewise.
776 (gen_struct_or_union_type_die): Likewise.
777 (gen_subroutine_type_die): Likewise.
778 (gen_typedef_die): Likewise.
779 (base_type_cmp): Compare alignment attribute.
781 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
784 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
785 (setb_unsigned) New pattern for setb with CCUNS.
786 * config/rs6000/rs6000.c (expand_block_compare): Use a different
787 subfc./subfe sequence to avoid overflow problems. Generate a
788 shorter sequence with cmpld/setb for power9.
789 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
790 for generating subfc. instruction.
791 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
792 now uses this instruction.
794 2017-01-30 Ian Lance Taylor <iant@google.com>
797 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
798 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
800 2017-01-30 Martin Sebor <msebor@redhat.com>
802 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
803 Move constant to the right of a relational operator.
804 (get_mpfr_format_length, format_character, format_string): Ditto.
805 (should_warn_p, maybe_warn): Same.
807 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
809 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
812 * asan.c (get_translation_unit_decl): Remove function.
813 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
815 2017-01-30 Martin Liska <mliska@suse.cz>
817 PR gcov-profile/79259
818 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
821 2017-01-30 Martin Liska <mliska@suse.cz>
824 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
825 Initialize variables with NULL value.
827 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
830 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
832 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
834 2017-01-30 Richard Biener <rguenther@suse.de>
836 PR tree-optimization/79276
837 * tree-vrp.c (process_assert_insertions): Properly adjust common
838 when removing a duplicate.
840 * gcc.dg/torture/pr79276.c: New testcase.
842 2017-01-30 Richard Biener <rguenther@suse.de>
844 PR tree-optimization/79256
845 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
846 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
848 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
850 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
853 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
854 ("*r<noxa>sbg_<mode>_sll_bitmask")
855 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
856 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
857 Use contiguous_bitmask_nowrap_operand.
859 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
862 * config/rs6000/altivec.h (vec_xl): Revise #define.
865 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
867 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
869 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
871 PR rtl-optimization/79194
872 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
873 traps before call to bypass_conditional_jumps.
875 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
877 PR tree-optimization/71374
878 * lra-constraints.c (check_conflict_input_operands): New.
879 (match_reload): Use it.
881 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
884 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
885 account to calculate conflict_set.
887 2017-01-27 Bin Cheng <bin.cheng@arm.com>
889 PR rtl-optimization/78559
890 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
891 other_insn in combine.
893 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
895 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
896 uint16_type_node for BT_UINT16.
898 2017-01-27 David Malcolm <dmalcolm@redhat.com>
900 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
902 (GIMPLE Tests): New node.
903 (RTL Tests): New node.
905 2017-01-27 Richard Biener <rguenther@suse.de>
907 PR tree-optimization/79245
908 * tree-loop-distribution.c (distribute_loop): Apply cost
909 modeling also to detected patterns.
911 2017-01-27 Richard Biener <rguenther@suse.de>
913 PR tree-optimization/71433
914 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
915 (compare_assert_loc): New function.
916 (process_assert_insertions): Sort and optimize assert locations
917 to remove duplicates and push down identical assertions on
918 edges to their destination block.
920 2017-01-27 Richard Biener <rguenther@suse.de>
922 PR tree-optimization/79244
923 * tree-vrp.c (remove_range_assertions): Forcefully propagate
924 out SSA names even if abnormal.
926 2017-01-27 Jakub Jelinek <jakub@redhat.com>
928 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
929 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
930 instead of MPFR_RNDN.
932 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
935 * arm.c (arm_option_override): Don't call build_target_option_node
936 until after doing all option overrides.
937 (arm_valid_target_attribute_tree): Likewise.
939 2017-01-27 Martin Liska <mliska@suse.cz>
941 * doc/invoke.texi (-fprofile-arcs): Document profiling support
942 for {cd}tors and C++ {cd}tors.
944 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
946 * config/s390/s390.md ("*setmem_long_and")
947 ("*setmem_long_and_31z"): Use zero_extend instead of and.
949 2017-01-26 Martin Sebor <msebor@redhat.com>
951 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
954 2017-01-26 Martin Sebor <msebor@redhat.com>
956 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
957 HAVE_DFmode before using XFmode or DFmode.
958 (parse_directive): Avoid using the z length modifier to avoid
959 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
962 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
963 to accept adjustment as an array.
964 (get_int_range): New function.
965 (struct directive): Make width and prec arrays.
966 (directive::set_width, directive::set_precision): Call get_int_range.
967 (format_integer, format_floating): Handle width and precision ranges.
968 (format_string, parse_directive): Same.
970 2017-01-26 Jakub Jelinek <jakub@redhat.com>
973 * dwarf2out.c (generate_skeleton_bottom_up): For children with
974 comdat_type_p set, just clone them, but keep the children in the
978 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
979 which have direct callers with -fvar-tracking-assignments enabled
981 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
982 inside of type units.
984 2017-01-26 Martin Sebor <msebor@redhat.com>
987 * gimple-ssa-sprintf.c (struct result_range): Add likely and
989 (struct format_result): Replace number_chars, number_chars_min,
990 and number_chars_max with a single member of struct result_range.
992 (format_result::operator+=): Adjust.
993 (struct fmtresult): Remove bounded. Handle likely and unlikely
995 (fmtresult::adjust_for_width_or_precision): New function.
996 (fmtresult:type_max_digits): New function.
997 (bytes_remaining): Handle likely and unlikely counters.
998 (min_bytes_remaining): Remove.
999 (format_percent): Simplify.
1000 (format_integer, format_floating): Set likely and unlikely counters.
1001 (get_string_length, format_character, format_string): Same.
1002 (format_plain, should_warn_p): New function.
1003 (maybe_warn): Call should_warn_p. Update diagnostic messages
1004 and handle those for all directives, including plain strings.
1005 (format_directive): Handle likely and unlikely counters.
1006 Remove unnecessary quoting from diagnostics. Add an informational
1008 (add_bytes): Remove.
1009 (pass_sprintf_length::compute_format_length): Simplify.
1010 (try_substitute_return_value): Handle likely and unlikely counters.
1012 2017-01-26 Carl Love <cel@us.ibm.com>
1014 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
1015 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
1017 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
1020 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
1021 endianess for subregs into account.
1022 * lra-constraints.c (lra_constraints): Do risky transformations
1023 always on the first iteration.
1024 * lra-lives.c (check_pseudos_live_through_calls): Add arg
1025 last_call_used_reg_set.
1026 (process_bb_lives): Define and use last_call_used_reg_set.
1027 * lra.c (lra): Always continue after lra_constraints on the first
1030 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
1032 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
1034 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
1036 2017-01-26 Jakub Jelinek <jakub@redhat.com>
1038 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
1039 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
1040 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
1041 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
1042 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
1043 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
1044 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
1045 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
1046 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
1048 2017-01-26 Marek Polacek <polacek@redhat.com>
1051 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
1052 for the third operand.
1054 2017-01-26 Jakub Jelinek <jakub@redhat.com>
1057 * omp-low.c (struct omp_context): Add simt_stmt field.
1058 (scan_omp_for): Return omp_context *.
1059 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
1060 context to the _simt_ SIMD stmt.
1061 (lower_omp_for): For combined SIMD with sibling _simt_
1062 SIMD, make sure to use the same decls in _looptemp_
1063 clauses as in the sibling.
1065 2017-01-26 David Sherwood <david.sherwood@arm.com>
1068 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
1071 2017-01-26 Jakub Jelinek <jakub@redhat.com>
1074 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
1075 emit fld b; fld a; if possible.
1077 * brig-builtins.def: Update copyright years.
1078 * config/arm/arm_acle_builtins.def: Update copyright years.
1080 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
1083 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
1084 constraint instead of o for the stxsd instruction.
1086 2017-01-25 Carl Love <cel@us.ibm.com>
1088 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
1089 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
1091 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
1093 * doc/invoke.texi (C++ Dialect Options): Fix typo.
1095 2017-01-25 Richard Biener <rguenther@suse.de>
1097 PR tree-optimization/69264
1098 * target.def (vector_alignment_reachable): Improve documentation.
1099 * doc/tm.texi: Regenerate.
1100 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
1102 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
1103 earlier changes with respect to TYPE_USER_ALIGN.
1104 (vector_alignment_reachable_p): Likewise. Improve dumping.
1106 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1109 * config/arm/arm.md (xordi3): Force constant operand into a register
1112 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1114 * doc/invoke.texi (-fstore-merging): Correct default optimization
1115 levels at which it is enabled.
1116 (-O): Move -fstore-merging from list to...
1119 2017-01-25 Richard Biener <rguenther@suse.de>
1122 * omp-expand.c: Include debug.h.
1123 (expand_omp_taskreg): Make sure to generate early debug before
1124 outlining anything from a function.
1125 (expand_omp_target): Likewise.
1126 (grid_expand_target_grid_body): Likewise.
1128 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
1131 * asan.c (get_translation_unit_decl): New function.
1132 (asan_add_global): Extract modules file name from globals
1133 TRANSLATION_UNIT_DECL name.
1135 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
1138 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
1139 for long calls with APCS frame and VFP.
1141 2017-01-24 David Malcolm <dmalcolm@redhat.com>
1143 * cfg.c (original_copy_tables_initialized_p): New function.
1144 * cfg.h (original_copy_tables_initialized_p): New decl.
1145 * cfgrtl.c (relink_block_chain): Guard the call to
1146 free_original_copy_tables with a call to
1147 original_copy_tables_initialized_p.
1148 * cgraph.h (symtab_node::native_rtl_p): New decl.
1149 * cgraphunit.c (symtab_node::native_rtl_p): New function.
1150 (symtab_node::needed_p): Don't assert for early assembly output
1151 for __RTL functions.
1152 (cgraph_node::finalize_function): Set "force_output" for __RTL
1154 (cgraph_node::analyze): Bail out early for __RTL functions.
1155 (analyze_functions): Update assertion to support __RTL functions.
1156 (cgraph_node::expand): Bail out early for __RTL functions.
1157 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
1159 * function.h (struct function): Update comment for field
1161 * gimple-expr.c: Include "tree-pass.h".
1162 (gimple_has_body_p): Return false for __RTL functions.
1163 * Makefile.in (OBJS): Add run-rtl-passes.o.
1164 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
1166 (gcc::pass_manager::get_clean_slate): New accessor.
1167 * passes.c: Include "insn-addr.h".
1168 (should_skip_pass_p): Add logging. Update logic for running
1169 "expand" to be compatible with both __GIMPLE and __RTL. Guard
1170 property-provider override so it is only done for gimple passes.
1172 (skip_pass): New function.
1173 (execute_one_pass): Call skip_pass when skipping passes.
1174 * read-md.c (md_reader::read_char): Support filtering
1175 the input to a subset of line numbers.
1176 (md_reader::md_reader): Initialize fields
1177 m_first_line and m_last_line.
1178 (md_reader::read_file_fragment): New function.
1179 * read-md.h (md_reader::read_file_fragment): New decl.
1180 (md_reader::m_first_line): New field.
1181 (md_reader::m_last_line): New field.
1182 * read-rtl-function.c (function_reader::create_function): Only
1183 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
1184 curr_properties. Set DECL_INITIAL to a dummy block.
1185 (read_rtl_function_body_from_file_range): New function.
1186 * read-rtl-function.h (read_rtl_function_body_from_file_range):
1188 * run-rtl-passes.c: New file.
1189 * run-rtl-passes.h: New file.
1191 2017-01-24 Jeff Law <law@redhat.com>
1193 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
1196 2017-01-24 Bin Cheng <bin.cheng@arm.com>
1198 PR tree-optimization/79159
1199 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
1200 (record_nonwrapping_iv): Improve boundary using above function if no
1201 value range information.
1203 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
1204 Martin Jambor <mjambor@suse.cz>
1206 * brig-builtins.def: New file.
1207 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
1208 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
1209 (DEF_HSAIL_SAT_BUILTIN): Likewise.
1210 (DEF_HSAIL_INTR_BUILTIN): Likewise.
1211 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
1212 * builtin-types.def (BT_INT8): New.
1213 (BT_INT16): Likewise.
1214 (BT_UINT8): Likewise.
1215 (BT_UINT16): Likewise.
1216 (BT_FN_ULONG): Likewise.
1217 (BT_FN_UINT_INT): Likewise.
1218 (BT_FN_UINT_ULONG): Likewise.
1219 (BT_FN_UINT_LONG): Likewise.
1220 (BT_FN_UINT_PTR): Likewise.
1221 (BT_FN_ULONG_PTR): Likewise.
1222 (BT_FN_INT8_FLOAT): Likewise.
1223 (BT_FN_INT16_FLOAT): Likewise.
1224 (BT_FN_UINT32_FLOAT): Likewise.
1225 (BT_FN_UINT16_FLOAT): Likewise.
1226 (BT_FN_UINT8_FLOAT): Likewise.
1227 (BT_FN_UINT64_FLOAT): Likewise.
1228 (BT_FN_UINT16_UINT32): Likewise.
1229 (BT_FN_UINT32_UINT16): Likewise.
1230 (BT_FN_UINT16_UINT16_UINT16): Likewise.
1231 (BT_FN_INT_PTR_INT): Likewise.
1232 (BT_FN_UINT_PTR_UINT): Likewise.
1233 (BT_FN_LONG_PTR_LONG): Likewise.
1234 (BT_FN_ULONG_PTR_ULONG): Likewise.
1235 (BT_FN_VOID_UINT64_UINT64): Likewise.
1236 (BT_FN_UINT8_UINT8_UINT8): Likewise.
1237 (BT_FN_INT8_INT8_INT8): Likewise.
1238 (BT_FN_INT16_INT16_INT16): Likewise.
1239 (BT_FN_INT_INT_INT): Likewise.
1240 (BT_FN_UINT_FLOAT_UINT): Likewise.
1241 (BT_FN_FLOAT_UINT_UINT): Likewise.
1242 (BT_FN_ULONG_UINT_UINT): Likewise.
1243 (BT_FN_ULONG_UINT_PTR): Likewise.
1244 (BT_FN_ULONG_ULONG_ULONG): Likewise.
1245 (BT_FN_UINT_UINT_UINT): Likewise.
1246 (BT_FN_VOID_UINT_PTR): Likewise.
1247 (BT_FN_UINT_UINT_PTR: Likewise.
1248 (BT_FN_UINT32_UINT64_PTR): Likewise.
1249 (BT_FN_INT_INT_UINT_UINT): Likewise.
1250 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
1251 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
1252 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
1253 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
1254 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
1255 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
1256 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
1257 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
1258 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
1259 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
1260 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
1261 * doc/frontends.texi: List BRIG FE.
1262 * doc/install.texi (Testing): Add BRIG tesring requirements.
1263 * doc/invoke.texi (Overall Options): Mention BRIG.
1264 * doc/standards.texi (Standards): Doucment BRIG HSA version.
1266 2017-01-24 Richard Biener <rguenther@suse.de>
1268 PR translation/79208
1269 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
1271 2017-01-24 Martin Jambor <mjambor@suse.cz>
1274 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
1275 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
1278 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
1281 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
1282 casts from signed to unsigned really don't have a range.
1284 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
1286 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
1287 GMP_RNDx for compatiblity.
1289 2017-01-24 Martin Liska <mliska@suse.cz>
1292 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
1293 that would prevent us to call alloca with -1 as argument.
1295 2017-01-24 Jakub Jelinek <jakub@redhat.com>
1297 * dwarf2out.c (output_compilation_unit_header, output_file_names):
1298 Avoid -Wformat-security warning.
1300 2017-01-23 Andrew Pinski <apinski@cavium.com>
1302 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
1305 2017-01-23 Martin Sebor <msebor@redhat.com>
1308 * gimple-ssa-sprintf.c (warn_level): New global.
1309 (format_integer): Use it here and throughout the rest of the file.
1310 Use the same switch to compute sign as base.
1311 (maybe_warn): New function.
1312 (format_directive): Factor out warnings into maybe_warn.
1313 Add debugging output. Use warn_level.
1314 (add_bytes): Use warn_level.
1315 (pass_sprintf_length::compute_format_length): Add debugging output.
1316 (try_substitute_return_value): Same.
1317 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
1320 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
1321 (struct fmtresult, format_integer, format_floating): Adjust.
1322 (fmtresult::fmtresult): Set max correctly in two argument ctor.
1323 (get_string_length, format_string,format_directive): Same.
1324 (pass_sprintf_length::compute_format_length): Same.
1325 (try_substitute_return_value): Simplify slightly.
1328 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
1329 (fmtresult::operator+=): Outlined.
1330 (struct fmtresult): Add ctors.
1331 (struct conversion_spec): Rename...
1332 (struct directive): ...to this. Add and remove data members.
1333 (directive::set_width, directive::set_precision): New functions.
1334 (format_percent): Use fmtresult ctor.
1335 (get_width_and_precision): Remove.
1336 (format_integer): Make naming changes. Avoid computing width and
1338 (format_floating): Same. Adjust indentation.
1339 (format_character, format_none): New functions.
1340 (format_string): Moved character handling to format_character.
1341 (format_directive): Remove arguments, change return type.
1342 (parse_directive): New function.
1343 (pass_sprintf_length::compute_format_length): Move directive
1344 parsing to parse_directive.
1346 2017-01-23 Jakub Jelinek <jakub@redhat.com>
1348 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
1349 (assign_assembler_name_if_needed): ... this.
1350 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
1351 (assign_assembler_name_if_needed): ... this.
1352 (free_lang_data_in_cgraph): Adjust callers.
1353 * cgraphunit.c (cgraph_node::analyze): Likewise.
1354 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
1357 2017-01-23 Richard Biener <rguenther@suse.de>
1359 PR tree-optimization/79088
1360 PR tree-optimization/79188
1361 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
1362 resetting loop bounds after last path deletion. Reset loop
1363 bounds of the target loop, make code match the comments.
1364 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
1365 Make sure loops need no fixups.
1367 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
1369 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
1370 exponent support with double type for first argument.
1371 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
1372 type returned by __builtin_vec_extract_sig,
1373 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
1374 functions from "vector int" to "vector unsigned int" or from
1375 "vector long long int" to "vector unsigned long long int".
1376 Changed type returned by __builtin_vec_extract_exp,
1377 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
1378 functions from "vector int" to "vector unsigned int" or from
1379 "vector long long int" to "vector unsigned long long int".
1380 Changed return type of __builtin_vec_test_data_class,
1381 __builtin_vec_test_data_class_sp, and
1382 __builtin_vec_test_data_class_dp from "vector int" to
1383 "vector bool int" or from "vector long long int" to "vector bool
1384 long long int" and changed second argument type from "unsigned
1385 int" to "int". Added new overloaded function forms "vector float
1386 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
1387 "vector float __builtin_vec_insert_exp_sp (vector float, vector
1388 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
1389 double, vector unsigned long long int)" and "vector double
1390 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
1391 long int)". Changed return type of
1392 __builtin_scalar_test_data_class and
1393 __builtin_scalar_test_data_class_sp and
1394 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
1395 int" and changed second argument from "unsigned int" to "int".
1396 Changed type returned by __builtin_scalar_test_neg,
1397 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
1398 from "int" to "bool int". Added new overloaded function form
1399 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
1400 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
1401 exponent double-precision with floating point first argument.
1402 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
1403 documentation of scalar_test_data_class, scalar_test_neg,
1404 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
1405 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
1406 vec_test_data_class built-in functions to reflect refinements in
1407 their type signatures.
1409 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
1411 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
1413 (aarch64_elf_asm_destructor): Likewise.
1415 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
1417 PR rtl-optimization/78634
1418 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
1419 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
1420 * ifcvt.c (noce_try_cmove): Add missing cost check.
1422 PR rtl-optimization/71724
1423 * combine.c (if_then_else_cond): Look for situations where it is
1424 beneficial to undo the work of one of the recursive calls.
1426 2017-01-23 Bin Cheng <bin.cheng@arm.com>
1428 PR tree-optimization/70754
1429 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
1430 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
1431 combined stmt before it if not NULL.
1432 (combine_chains): Process refs reversely and compute dominance point
1435 2017-01-23 Martin Liska <mliska@suse.cz>
1437 PR tree-optimization/79196
1438 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
1439 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
1441 (strlen_optimize_stmt): Call the renamed function.
1443 2017-01-23 Michael Matz <matz@suse.de>
1445 PR tree-optimization/78384
1446 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
1448 2017-01-23 Richard Biener <rguenther@suse.de>
1450 PR tree-optimization/79186
1451 * tree-vrp.c (register_new_assert_for): Make sure we've seen
1452 both incoming edges before moving an assert.
1454 2017-01-23 Martin Jambor <mjambor@suse.cz>
1456 * ipa-prop.c (load_from_param_1): Removed.
1457 (load_from_unmodified_param): Bits from load_from_param_1 put back
1459 (load_from_param): Removed.
1460 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
1461 with stmt. Reverted back to use of load_from_unmodified_param.
1463 2017-01-23 Martin Jambor <mjambor@suse.cz>
1466 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
1467 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
1468 field a pointer to garbage collected vector, mark lattices and
1469 ipcp_orig_node with GTY((skip)).
1470 (ipa_get_param_count): Adjust to descriptors being a pointer.
1471 (ipa_get_param): Likewise.
1472 (ipa_get_type): Likewise.
1473 (ipa_get_param_move_cost): Likewise.
1474 (ipa_set_param_used): Likewise.
1475 (ipa_get_controlled_uses): Likewise.
1476 (ipa_set_controlled_uses): Likewise.
1477 (ipa_is_param_used): Likewise.
1478 (ipa_node_params_t): Move into garbage collector. New methods insert
1480 (ipa_node_params_sum): Annotate wth GTY(()).
1481 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
1483 (ipa_load_from_parm_agg): Adjust declaration.
1484 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
1485 * ipa-profile.c (ipa_profile): Likewise.
1486 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
1487 (ipa_populate_param_decls): Make descriptors parameter garbage
1489 (ipa_dump_param): Adjust to descriptors being a pointer.
1490 (ipa_alloc_node_params): Likewise.
1491 (ipa_initialize_node_params): Likewise.
1492 (load_from_param_1): Make descriptors parameter garbage collected.
1493 (load_from_unmodified_param): Likewise.
1494 (load_from_param): Likewise.
1495 (ipa_load_from_parm_agg): Likewise.
1496 (ipa_node_params::~ipa_node_params): Removed.
1497 (ipa_free_all_node_params): Remove call to delete operator.
1498 (ipa_node_params_t::insert): New.
1499 (ipa_node_params_t::remove): Likewise.
1500 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
1501 copy known_csts and known_contexts vectors.
1502 (ipa_read_node_info): Adjust to descriptors being a pointer.
1503 (ipcp_modif_dom_walker): Make m_descriptors field garbage
1505 (ipcp_transform_function): Make descriptors variable garbage
1508 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
1510 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
1511 * config/i386/avx512dqintrin.h: Ditto.
1512 * config/i386/avx512fintrin.h: Ditto.
1513 * gcc/config/i386/i386.c: Handle new builtins.
1514 * config/i386/i386-builtin.def: Add new builtins.
1515 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
1516 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
1518 2017-01-23 Jakub Jelinek <jakub@redhat.com>
1519 Martin Liska <mliska@suse.cz>
1521 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
1522 * asan.c (asan_expand_poison_ifn): Support stores and use
1523 appropriate ASAN report function.
1524 * internal-fn.c (expand_ASAN_POISON_USE): New function.
1525 * internal-fn.def (ASAN_POISON_USE): Declare.
1526 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
1527 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
1528 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
1529 ASAN_POISON calls w/o LHS.
1530 * tree-ssa.c (execute_update_addresses_taken): Create clobber
1531 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
1532 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
1533 * gimplify.c (asan_poison_variables): Add attribute
1534 use_after_scope_memory to variables that really needs to live
1536 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
1537 having the attribute.
1539 2017-01-23 Martin Liska <mliska@suse.cz>
1541 * asan.c (create_asan_shadow_var): New function.
1542 (asan_expand_poison_ifn): Likewise.
1543 * asan.h (asan_expand_poison_ifn): New declaration.
1544 * internal-fn.c (expand_ASAN_POISON): Likewise.
1545 * internal-fn.def (ASAN_POISON): New builtin.
1546 * sanopt.c (pass_sanopt::execute): Expand
1547 asan_expand_poison_ifn.
1548 * tree-inline.c (copy_decl_for_dup_finish): Make function
1550 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
1551 * tree-ssa.c (is_asan_mark_p): New function.
1552 (execute_update_addresses_taken): Rewrite local variables
1553 (identified just by use-after-scope as addressable) into SSA.
1555 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
1557 * doc/install.texi (Specific): opensource.apple.com uses https
1558 now. Remove trailing slash.
1560 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
1562 * README.Portability: Remove note on an Irix compatibility issue.
1564 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
1566 * gcov.c (INCLUDE_ALGORITHM): Define.
1567 (INCLUDE_VECTOR): Define.
1568 No longer include <vector> and <algorithm> directly.
1570 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
1572 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
1574 * doc/invoke.texi (Code Gen Options): Ditto.
1576 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
1579 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
1581 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
1583 rtl-optimization/79125
1584 * cprop.c (local_cprop_pass): Handle cases where we make an
1587 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
1591 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
1592 read from, for big endian.
1594 2017-01-20 Jiong Wang <jiong.wang@arm.com>
1596 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
1597 register pauth builtins for LP64 only.
1599 2017-01-20 Marek Polacek <polacek@redhat.com>
1602 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
1605 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
1607 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
1609 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
1610 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
1611 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
1613 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1616 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
1617 in big-endian mode when they are not a single duplicated value.
1619 2017-01-20 Richard Biener <rguenther@suse.de>
1621 * BASE-VER: Bump to 7.0.1.
1623 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
1625 * omp-low.c (omplow_simd_context): New struct. Use it...
1626 (lower_rec_simd_input_clauses): ...here and...
1627 (lower_rec_input_clauses): ...here to hold common data. Adjust all
1628 references to idx, lane, max_vf, is_simt.
1630 2017-01-20 Graham Markall <graham.markall@embecosm.com>
1632 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
1635 2017-01-20 Martin Jambor <mjambor@suse.cz>
1637 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
1638 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
1640 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
1641 (GTFILES): Rename hsa.c to hsa-common.c.
1642 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
1643 * hsa-dump.c: Likewise.
1644 * hsa-gen.c: Likewise.
1645 * hsa-regalloc.c: Likewise.
1646 * ipa-hsa.c: Likewise.
1647 * omp-expand.c: Likewise.
1648 * omp-low.c: Likewise.
1649 * toplev.c: Likewise.
1651 2017-01-20 Marek Polacek <polacek@redhat.com>
1654 * doc/invoke.texi: Document -Wduplicated-branches.
1655 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
1656 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
1657 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
1658 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
1659 return 0 only when not OEP_LEXICOGRAPHIC.
1660 (fold_build_cleanup_point_expr): Use the expression
1661 location when building CLEANUP_POINT_EXPR.
1662 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
1663 * tree.c (add_expr): Handle error_mark_node.
1665 2017-01-20 Martin Liska <mliska@suse.cz>
1668 * tree-profile.c (init_ic_make_global_vars): Do not call
1670 (gimple_init_gcov_profiler): Likewise.
1672 2017-01-20 Martin Liska <mliska@suse.cz>
1675 * cgraph.h (maybe_create_reference): Remove argument and
1677 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
1679 * ipa-cp.c (create_specialized_node): Likewise.
1680 * symtab.c (symtab_node::maybe_create_reference): Handle
1681 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
1683 2017-01-20 Martin Liska <mliska@suse.cz>
1685 * read-rtl-function.c (function_reader::create_function): Use
1686 build_decl instread of build_decl_stat.
1688 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
1690 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
1691 * config/i386/avx512dqintrin.h: Ditto.
1692 * config/i386/avx512fintrin.h: Ditto.
1693 * config/i386/i386-builtin-types.def: Add new types.
1694 * gcc/config/i386/i386.c: Handle new types.
1695 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
1696 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
1697 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
1698 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
1699 (__builtin_ia32_kshiftridi): New.
1700 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
1702 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
1706 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
1707 define to rs6000_init_stack_protect_guard.
1708 (rs6000_init_stack_protect_guard): New function.
1710 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
1711 Yunqiang Su <yunqiang.su@imgtec.com>
1713 * config.gcc (supported_defaults): Add madd4.
1714 (with_madd4): Add validation.
1715 (all_defaults): Add madd4.
1716 * config/mips/mips.opt (mmadd4): New option.
1717 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
1719 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
1721 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
1722 (ISA_HAS_FUSED_MADD4): Likewise.
1723 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
1724 * gcc/doc/install.texi (--with-madd4): Document the new option.
1726 2017-01-19 Jiong Wang <jiong.wang@arm.com>
1728 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
1729 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
1730 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
1731 (aarch64_init_pauth_hint_builtins): New.
1732 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
1733 (aarch64_expand_builtin): Expand new builtins.
1735 2017-01-19 Jiong Wang <jiong.wang@arm.com>
1737 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
1738 * combine-stack-adj.c (no_unhandled_cfa): Handle
1739 REG_CFA_TOGGLE_RA_MANGLE.
1740 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
1741 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
1742 info for return address signing.
1743 (aarch64_expand_epilogue): Likewise.
1745 2017-01-19 Jiong Wang <jiong.wang@arm.com>
1747 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
1748 * config/aarch64/aarch64-protos.h
1749 (aarch64_return_address_signing_enabled): New declaration.
1750 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
1752 (aarch64_expand_prologue): Sign return address before it's pushed onto
1754 (aarch64_expand_epilogue): Authenticate return address fetched from
1756 (aarch64_override_options): Sanity check for ILP32 and ISA level.
1757 (aarch64_attributes): New function attributes for "sign-return-address".
1758 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
1759 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
1760 ("*do_return"): Generate combined instructions according to key index.
1761 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
1762 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
1764 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
1765 * config/aarch64/aarch64.opt (msign-return-address=): New.
1766 * doc/extend.texi (AArch64 Function Attributes): Documents
1767 "sign-return-address=".
1768 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
1770 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
1772 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
1773 overall option summary.
1775 2017-01-19 Jiong Wang <jiong.wang@arm.com>
1777 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
1778 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
1779 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
1780 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
1782 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
1784 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
1785 -mpower9-minmax by default for -mcpu=power9.
1786 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
1787 128-bit floating point.
1789 2017-01-20 Alan Modra <amodra@gmail.com>
1791 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
1792 optimizing for size.
1794 2017-01-20 Alan Modra <amodra@gmail.com>
1797 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
1798 for strcmp and strncmp from corresponding builtin decl.
1800 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
1802 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
1803 instead of i386/rtems-64.h.
1804 * config/i386/rtems-64.h: Remove.
1806 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
1810 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
1812 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
1814 2017-01-19 Tamar Christina <tamar.christina@arm.com>
1816 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1817 Change int to HOST_WIDE_INT.
1818 * config/aarch64/aarch64-protos.h
1819 (aarch64_simd_gen_const_vector_dup): Likewise.
1820 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
1822 2017-01-19 David Malcolm <dmalcolm@redhat.com>
1824 * langhooks-def.h (lhd_type_for_size): New decl.
1825 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
1826 * langhooks.c (lhd_type_for_size): New function, taken from
1829 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
1831 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
1832 define_bypass for CR latency.
1833 (power9-cracked-alu): Update bypass latency and remove power9-branch.
1834 (power9-alu2): Add define_bypass for CR latency.
1836 (power9-mul): Update insn latency.
1837 (power9-mul-compare): Update insn latency, bypass latency and remove
1840 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1842 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
1844 * config/aarch64/aarch64.md
1845 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
1846 aarch64_nopcrelative_literal_loads.
1847 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
1849 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
1851 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
1853 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
1855 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
1858 * config.gcc (supported_defaults): Add lxc1-sxc1.
1859 (with_lxc1_sxc1): Add validation.
1860 (all_defaults): Add lxc1-sxc1.
1861 * config/mips/mips.opt (mlxc1-sxc1): New option.
1862 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
1864 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
1865 __mips_no_lxc1_sxc1.
1866 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
1867 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
1868 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
1870 2017-01-19 Richard Biener <rguenther@suse.de>
1872 PR tree-optimization/72488
1873 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
1874 sure to restore SSA info.
1875 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
1877 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
1879 PR rtl-optimization/79121
1880 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
1881 of the inner type when shifting an extended value.
1883 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
1886 * symtab.c (symtab_node::equal_address_to): Fix comparing of
1887 interposable aliases.
1889 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
1892 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
1893 Use the evmergelohi instruction.
1894 (mov_si<mode>_e500_subreg4_2_le): Likewise.
1895 (mov_sitf_e500_subreg8_2_be): Likewise.
1896 (mov_sitf_e500_subreg12_2_le): Likewise.
1897 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
1898 (mov_si<mode>_e500_subreg4_2_be): Likewise.
1899 (mov_sitf_e500_subreg8_2_le): Likewise.
1900 (mov_sitf_e500_subreg12_2_be): Likewise.
1902 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1904 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
1905 attribute from vecsimple to vecperm.
1906 (altivec_vbpermq2): Likewise.
1908 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1911 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
1913 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1914 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
1915 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
1916 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
1917 case where N arg is SIZE_MAX.
1918 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
1919 (cmpstrsi): Add pattern.
1921 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1923 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1924 __builtin_vec_revb builtins.
1925 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
1926 built-in functions to support generation of the ISA 3.0 XXBR<x>
1927 vector byte reverse instructions.
1928 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
1929 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
1930 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
1931 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
1932 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
1933 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
1934 (P9V_BUILTIN_VEC_REVB): Likewise.
1935 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
1936 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
1937 (p9_xxbrq_v16qi): Likewise.
1938 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
1939 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
1940 (p9_xxbrh_v8hi): Likewise.
1941 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
1942 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
1943 vec_revb built-in functions.
1945 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
1947 PR rtl-optimization/78952
1948 * config/i386/i386.md (any_extract): New code iterator.
1949 (*insvqi_2): Use any_extract for source operand.
1950 (*insvqi_3): Use any_shiftrt for source operand.
1952 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
1954 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
1956 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
1958 2017-01-18 Matthias Klose <doko@ubuntu.com>
1960 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
1962 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1964 * config/rs6000/altivec.h (vec_bperm): Change #define.
1965 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
1966 (altivec_vbpermq2): New define_insn.
1967 (altivec_vbpermd): Likewise.
1968 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
1970 (VBPERMD): Likewise.
1971 (VBPERM): New polymorphic function interface.
1972 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
1973 Add entries for P9V_BUILTIN_VEC_VBPERM.
1974 * doc/extend.texi: Add interfaces for vec_bperm.
1976 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1978 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
1979 first letter of error messages.
1980 (s390_resolve_overloaded_builtin): Likewise.
1981 * config/s390/s390.c (s390_expand_builtin): Likewise.
1982 (s390_invalid_arg_for_unprototyped_fn): Likewise.
1983 (s390_valid_target_attribute_inner_p): Likewise.
1984 * config/s390/s390.md ("tabort"): Likewise.
1986 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
1988 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
1989 (ISA_AVOID_DIV_HILO): New macro.
1990 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
1991 (ISA_HAS_DDIV): Likewise.
1993 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
1995 * doc/invoke.texi (fabi-version): Correct number of occurrences.
1997 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
1999 * doc/invoke.texi (fabi-version): Spelling fix.
2001 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
2004 * doc/invoke.texi (fabi-version): Mention mangling fix for
2007 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
2010 * doc/invoke.texi (fabi-version): Document discriminator mangling.
2012 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
2015 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
2016 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
2018 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
2020 (stack_protect_test): Ditto.
2021 * config/rs6000/rs6000.opt (mstack-protector-guard=,
2022 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
2024 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
2025 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
2026 -mstack-protector-guard-offset=.
2027 (RS/6000 and PowerPC Options): Ditto.
2029 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
2031 * config/i386/i386.h (MASK_CLASS_P): New define.
2032 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
2033 there are no registers from different register sets also when
2034 mask registers are used. Update function comment.
2035 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
2036 to (*k/*r) and (*k/*km) alternatives.
2038 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
2040 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
2041 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
2042 (EH_RETURN_HANDLER_RTX): New define.
2043 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
2044 Force frame pointer in EH return functions.
2045 (aarch64_expand_epilogue): Add barrier for eh_return.
2046 (aarch64_final_eh_return_addr): Remove.
2047 (aarch64_eh_return_handler_rtx): New function.
2048 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
2050 (aarch64_eh_return_handler_rtx): New prototype.
2052 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2054 * config/rs6000/altivec.h (vec_rlmi): New #define.
2055 (vec_vrlnm): Likewise.
2056 (vec_rlnm): Likewise.
2057 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
2058 (UNSPEC_VRLNM): Likewise.
2059 (VIlong): New mode iterator.
2060 (altivec_vrl<VI_char>mi): New define_insn.
2061 (altivec_vrl<VI_char>nm): Likewise.
2062 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
2065 (RLNM): New polymorphic function entry.
2066 (VRLWMI): New monomorphic function entry.
2068 (RLMI): New polymorphic function entry.
2069 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
2070 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
2071 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
2074 2017-01-17 Jakub Jelinek <jakub@redhat.com>
2077 * dwarf2out.c (field_byte_offset): Restore the
2078 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
2079 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
2080 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
2083 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
2086 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
2088 2017-01-17 Jakub Jelinek <jakub@redhat.com>
2091 * dwarf2out.c (add_data_member_location_attribute): For constant
2092 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
2093 instead of DW_AT_data_member_location, DW_AT_bit_offset and
2094 DW_AT_byte_size attributes.
2096 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
2098 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
2099 after forcing to constant memory when the code model is medium.
2101 2017-01-17 Julia Koval <julia.koval@intel.com>
2104 * config/i386/avx512fintrin.h
2105 (_mm512_i32gather_ps): Change __addr type to void const*.
2106 (_mm512_mask_i32gather_ps): Ditto.
2107 (_mm512_i32gather_pd): Ditto.
2108 (_mm512_mask_i32gather_pd): Ditto.
2109 (_mm512_i64gather_ps): Ditto.
2110 (_mm512_mask_i64gather_ps): Ditto.
2111 (_mm512_i64gather_pd): Ditto.
2112 (_mm512_mask_i64gather_pd): Ditto.
2113 (_mm512_i32gather_epi32): Ditto.
2114 (_mm512_mask_i32gather_epi32): Ditto.
2115 (_mm512_i32gather_epi64): Ditto.
2116 (_mm512_mask_i32gather_epi64): Ditto.
2117 (_mm512_i64gather_epi32): Ditto.
2118 (_mm512_mask_i64gather_epi32): Ditto.
2119 (_mm512_i64gather_epi64): Ditto.
2120 (_mm512_mask_i64gather_epi64): Ditto.
2121 (_mm512_i32scatter_ps): Change __addr type to void*.
2122 (_mm512_mask_i32scatter_ps): Ditto.
2123 (_mm512_i32scatter_pd): Ditto.
2124 (_mm512_mask_i32scatter_pd): Ditto.
2125 (_mm512_i64scatter_ps): Ditto.
2126 (_mm512_mask_i64scatter_ps): Ditto.
2127 (_mm512_i64scatter_pd): Ditto.
2128 (_mm512_mask_i64scatter_pd): Ditto.
2129 (_mm512_i32scatter_epi32): Ditto.
2130 (_mm512_mask_i32scatter_epi32): Ditto.
2131 (_mm512_i32scatter_epi64): Ditto.
2132 (_mm512_mask_i32scatter_epi64): Ditto.
2133 (_mm512_i64scatter_epi32): Ditto.
2134 (_mm512_mask_i64scatter_epi32): Ditto.
2135 (_mm512_i64scatter_epi64): Ditto.
2136 (_mm512_mask_i64scatter_epi64): Ditto.
2137 * config/i386/avx512pfintrin.h
2138 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
2139 (_mm512_mask_prefetch_i32gather_ps): Ditto.
2140 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2141 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2142 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
2143 (_mm512_prefetch_i32scatter_ps): Ditto.
2144 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2145 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2146 (_mm512_prefetch_i64scatter_pd): Ditto.
2147 (_mm512_prefetch_i64scatter_ps): Ditto.
2148 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2149 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2150 * config/i386/avx512vlintrin.h
2151 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
2152 (_mm_mmask_i32gather_ps): Ditto.
2153 (_mm256_mmask_i32gather_pd): Ditto.
2154 (_mm_mmask_i32gather_pd): Ditto.
2155 (_mm256_mmask_i64gather_ps): Ditto.
2156 (_mm_mmask_i64gather_ps): Ditto.
2157 (_mm256_mmask_i64gather_pd): Ditto.
2158 (_mm_mmask_i64gather_pd): Ditto.
2159 (_mm256_mmask_i32gather_epi32): Ditto.
2160 (_mm_mmask_i32gather_epi32): Ditto.
2161 (_mm256_mmask_i32gather_epi64): Ditto.
2162 (_mm_mmask_i32gather_epi64): Ditto.
2163 (_mm256_mmask_i64gather_epi32): Ditto.
2164 (_mm_mmask_i64gather_epi32): Ditto.
2165 (_mm256_mmask_i64gather_epi64): Ditto.
2166 (_mm_mmask_i64gather_epi64): Ditto.
2167 (_mm256_i32scatter_ps): Change __addr type to void*.
2168 (_mm256_mask_i32scatter_ps): Ditto.
2169 (_mm_i32scatter_ps): Ditto.
2170 (_mm_mask_i32scatter_ps): Ditto.
2171 (_mm256_i32scatter_pd): Ditto.
2172 (_mm256_mask_i32scatter_pd): Ditto.
2173 (_mm_i32scatter_pd): Ditto.
2174 (_mm_mask_i32scatter_pd): Ditto.
2175 (_mm256_i64scatter_ps): Ditto.
2176 (_mm256_mask_i64scatter_ps): Ditto.
2177 (_mm_i64scatter_ps): Ditto.
2178 (_mm_mask_i64scatter_ps): Ditto.
2179 (_mm256_i64scatter_pd): Ditto.
2180 (_mm256_mask_i64scatter_pd): Ditto.
2181 (_mm_i64scatter_pd): Ditto.
2182 (_mm_mask_i64scatter_pd): Ditto.
2183 (_mm256_i32scatter_epi32): Ditto.
2184 (_mm256_mask_i32scatter_epi32): Ditto.
2185 (_mm_i32scatter_epi32): Ditto.
2186 (_mm_mask_i32scatter_epi32): Ditto.
2187 (_mm256_i32scatter_epi64): Ditto.
2188 (_mm256_mask_i32scatter_epi64): Ditto.
2189 (_mm_i32scatter_epi64): Ditto.
2190 (_mm_mask_i32scatter_epi64): Ditto.
2191 (_mm256_i64scatter_epi32): Ditto.
2192 (_mm256_mask_i64scatter_epi32): Ditto.
2193 (_mm_i64scatter_epi32): Ditto.
2194 (_mm_mask_i64scatter_epi32): Ditto.
2195 (_mm256_i64scatter_epi64): Ditto.
2196 (_mm256_mask_i64scatter_epi64): Ditto.
2197 (_mm_i64scatter_epi64): Ditto.
2198 (_mm_mask_i64scatter_epi64): Ditto.
2199 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
2200 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
2201 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
2202 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
2203 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
2204 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
2205 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
2206 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
2207 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
2208 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
2209 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
2210 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
2211 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
2212 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
2213 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
2214 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
2215 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
2216 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
2217 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
2218 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
2219 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
2220 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
2221 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
2222 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
2223 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
2224 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
2225 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
2226 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
2227 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
2228 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
2229 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
2230 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
2231 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
2232 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
2233 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
2234 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
2235 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
2236 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
2237 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
2238 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
2239 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
2240 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
2241 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
2242 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
2243 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
2244 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
2245 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
2246 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
2247 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
2248 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
2249 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
2250 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
2251 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
2252 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
2253 definitions accordingly.
2255 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
2256 Kuan-Lin Chen <kuanlinchentw@gmail.com>
2259 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
2262 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
2265 * ira-conflicts.c (ira_build_conflicts): Update total conflict
2266 hard regs for inner regno.
2268 2017-01-17 Martin Liska <mliska@suse.cz>
2271 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
2272 assumption and add comment.
2274 2017-01-17 Nathan Sidwell <nathan@acm.org>
2276 * ipa-visibility.c (localize_node): New function, broken out of ...
2277 (function_and_variable_visibility): ... here. Call it.
2279 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
2282 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
2283 correctly set frequency of oudgoing edge.
2284 (duplicate_thread_path): Fix profile updating.
2286 2017-01-17 Jakub Jelinek <jakub@redhat.com>
2289 * configure.ac: Add GCC_BASE_VER.
2290 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
2291 version from BASE-VER file.
2292 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
2293 (gcc.o): Depend on $(BASEVER).
2294 * common.opt (dumpfullversion): New option.
2295 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
2296 * doc/invoke.texi: Document -dumpfullversion.
2297 * doc/install.texi: Document --with-gcc-major-version-only.
2298 * configure: Regenerated.
2300 2017-01-17 Richard Biener <rguenther@suse.de>
2302 PR tree-optimization/71433
2303 * tree-vrp.c (register_new_assert_for): Merge same asserts
2304 on all incoming edges.
2305 (process_assert_insertions_for): Handle insertions at the
2308 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
2310 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
2311 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
2313 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
2316 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
2319 2017-01-17 Alan Modra <amodra@gmail.com>
2322 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
2323 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
2324 symbolic stack limit when pic.
2326 2017-01-16 Martin Sebor <msebor@redhat.com>
2328 PR tree-optimization/78608
2329 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
2331 2017-01-16 Jeff Law <law@redhat.com>
2334 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
2335 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
2336 for several include directories that may be relative to sysroot.
2337 * config/i386/x-mingw32 (gplus_includedir): Define.
2338 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
2339 (native_system_includedir): Likewise.
2340 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
2341 override if TARGET_SYSTEM_ROOT is defined.
2342 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
2344 PR tree-optimization/79090
2345 PR tree-optimization/33562
2346 PR tree-optimization/61912
2347 PR tree-optimization/77485
2348 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
2349 and computed trims into the dump file.
2351 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
2353 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
2355 2017-01-16 Jakub Jelinek <jakub@redhat.com>
2358 * gimplify.c (gimplify_init_constructor): If want_value and
2359 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
2363 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
2364 sequence. Formatting fixes.
2365 (doloop_optimize): Formatting fixes.
2368 * gcc.c (debug_level_greater_than_spec_func): New function.
2369 (static_spec_functions): Add debug-level-gt spec function.
2370 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
2372 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
2373 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
2374 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
2375 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
2376 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
2377 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
2379 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
2381 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
2382 QImode fixups to general and mask registers only.
2384 2017-01-16 Carl Love <cel@us.ibm.com>
2386 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
2387 for built-in functions
2388 vector signed char vec_nabs (vector signed char)
2389 vector signed short vec_nabs (vector signed short)
2390 vector signed int vec_nabs (vector signed int)
2391 vector signed long long vec_nabs (vector signed long long)
2392 vector float vec_nabs (vector float)
2393 vector double vec_nabs (vector double)
2394 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
2396 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
2397 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
2398 * doc/extend.texi: Update the documentation file for the new built-in
2401 2017-01-16 Martin Sebor <msebor@redhat.com>
2403 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
2406 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2408 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
2409 UNSPEC_VSX__XXSPLTD to require special splat handling.
2411 2017-01-16 David Malcolm <dmalcolm@redhat.com>
2414 * system.h: Poison strndup.
2416 2017-01-16 Alan Modra <amodra@gmail.com>
2419 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
2422 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
2424 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
2426 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
2428 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
2429 call recog here. Assert that INSN_CODE (insn) is non-negative.
2431 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
2434 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
2436 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
2437 in the currently scheduled RTL fragment.
2439 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
2441 PR rtl-optimization/78751
2442 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
2445 2017-01-14 Jeff Law <law@redhat.com>
2447 PR tree-optimization/79090
2448 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
2449 variable length stores.
2450 (compute_trims): Delete dead assignment to *trim_tail.
2451 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
2454 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
2456 PR rtl-optimization/78626
2457 PR rtl-optimization/78727
2458 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
2459 of a block, and split such blocks after everything else is finished.
2461 2017-01-14 Alan Modra <amodra@gmail.com>
2464 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
2465 target legitimate_combined_insn.
2466 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
2467 (rs6000_legitimate_combined_insn): New function.
2468 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
2470 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
2471 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
2472 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
2474 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
2476 * doc/frontends.texi (G++ and GCC): Remove references to Java.
2478 2017-01-13 Jeff Law <law@redhat.com>
2480 PR tree-optimization/33562
2481 PR tree-optimization/61912
2482 PR tree-optimization/77485
2483 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
2485 (delete_dead_assignment): Likewise.
2486 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
2487 statement to delete_dead_call and delete_dead_assignment.
2489 2017-01-13 David Malcolm <dmalcolm@redhat.com>
2492 * substring-locations.c (format_warning_va): Strengthen case 1 so
2493 that both endpoints of the substring must be within the format
2494 range for just the substring to be printed.
2496 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
2498 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
2499 * config/i386/i386.c (ix86_target_string): Add missing options
2500 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
2501 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
2502 flags_other and ix86_target_other to flags2_other. Display unknown
2504 (ix86_valid_target_attribute_inner_p): Add missing options and
2505 reorder options by implied ISAs, as in ix86_target_string.
2507 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
2509 * hash-table.h (hash_table::too_empty_p): New function.
2510 (hash_table::expand): Use it.
2511 (hash_table::traverse): Likewise.
2512 (hash_table::empty_slot): Use sizeof (value_type) instead of
2513 sizeof (PTR) to convert bytes to elements. Shrink the table
2514 if the current size is excessive for the current number of
2517 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
2519 * ira-costs.c (record_reg_classes): Break from the inner loop
2520 early once alt_fail is known to be true. Update outer loop
2521 handling accordingly.
2523 2017-01-13 Jeff Law <law@redhat.com>
2525 * tree-ssa-dse.c (decrement_count): New function.
2526 (increment_start_addr, maybe_trim_memstar_call): Likewise.
2527 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
2528 when we know the partially dead statement is a mem* function.
2530 PR tree-optimization/61912
2531 PR tree-optimization/77485
2532 * tree-ssa-dse.c: Include expr.h.
2533 (maybe_trim_constructor_store): New function.
2534 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
2536 PR tree-optimization/33562
2537 PR tree-optimization/61912
2538 PR tree-optimization/77485
2539 * doc/invoke.texi: Document new dse-max-object-size param.
2540 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
2541 * tree-ssa-dse.c: Include params.h.
2542 (dse_store_status): New enum.
2543 (initialize_ao_ref_for_dse): New, partially extracted from
2545 (valid_ao_ref_for_dse, normalize_ref): New.
2546 (setup_live_bytes_from_ref, compute_trims): Likewise.
2547 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
2548 (maybe_trim_partially_dead_store): Likewise.
2549 (maybe_trim_complex_store): Likewise.
2550 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
2551 Track what bytes live from the original store. Return tri-state
2552 for dead, partially dead or live.
2553 (dse_dom_walker): Add constructor, destructor and new private members.
2554 (delete_dead_call, delete_dead_assignment): New extracted from
2556 (dse_optimize_stmt): Make a member of dse_dom_walker.
2557 Use initialize_ao_ref_for_dse.
2559 PR tree-optimization/33562
2560 PR tree-optimization/61912
2561 PR tree-optimization/77485
2562 * sbitmap.h (bitmap_count_bits): Prototype.
2563 (bitmap_clear_range, bitmap_set_range): Likewise.
2564 * sbitmap.c (bitmap_clear_range): New function.
2565 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
2567 2017-01-13 Martin Liska <mliska@suse.cz>
2570 * function.c (set_cfun): Add new argument force.
2571 * function.h (set_cfun): Likewise.
2572 * ipa-inline-transform.c (inline_call): Use the function when
2573 strict alising from is dropped for function we inline to.
2575 2017-01-13 Richard Biener <rguenther@suse.de>
2577 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
2578 for dumping GIMPLE INTEGER_CSTs.
2580 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2582 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
2583 to 201112L since C++17.
2585 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
2588 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
2589 if -fsanitize=kernel-address is present.
2591 2017-01-13 Richard Biener <rguenther@suse.de>
2593 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
2594 as _Literal ( type ) number in case usual suffixes do not
2595 preserve all information.
2597 2017-01-13 Richard Biener <rguenther@suse.de>
2599 PR tree-optimization/77283
2600 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
2601 and ssa-iterators.h.
2602 (is_feasible_trace): Implement a cost model based on joiner
2605 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
2608 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
2609 char or short to __float128/_Float128 directly.
2611 2017-01-12 Martin Sebor <msebor@redhat.com>
2613 to -Wformat-overflow.
2614 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
2615 (min_bytes_remaining): Same.
2616 (get_string_length): Same.
2617 (format_string): Same.
2618 (format_directive): Same.
2620 (pass_sprintf_length::handle_gimple_call): Same.
2622 2017-01-12 Jakub Jelinek <jakub@redhat.com>
2624 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
2625 info.nowrite calls with no lhs that can't throw. Return bool
2626 whether gsi_remove has been called or not.
2627 (pass_sprintf_length::handle_gimple_call): Return bool whether
2628 try_substitute_return_value called gsi_remove. Formatting fix.
2629 (pass_sprintf_length::execute): Don't use gsi_remove if
2630 handle_gimple_call returned true.
2633 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
2634 be removed due to side-effects, don't remove following barrier nor
2635 turn the successor edge into fallthru edge.
2637 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2640 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
2641 element-reversing loads and stores as not swappable.
2643 2017-01-12 Nathan Sidwell <nathan@acm.org>
2644 Nicolai Stange <nicstange@gmail.com>
2646 * combine.c (try_combine): Don't ignore result of overlap checking
2647 loop. Combine overlap & asm check into single loop.
2649 2017-01-12 Richard Biener <rguenther@suse.de>
2651 * tree-pretty-print.c (dump_generic_node): Provide -gimple
2652 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
2654 2017-01-12 Richard Biener <rguenther@suse.de>
2656 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
2657 and TS_TARGET_OPTION directly derive from TS_BASE.
2658 * tree-core.h (tree_optimization_option): Derive from tree_base.
2659 (tree_target_option): Likewise.
2661 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
2663 * config/i386/i386.c (memory_address_length): Increase len
2664 only when rip_relative_addr_p returns false.
2666 2017-01-11 Julia Koval <julia.koval@intel.com>
2668 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
2669 (OPTION_MASK_ISA_SGX_SET): New.
2670 (ix86_handle_option): Handle OPT_msgx.
2671 * config.gcc: Added sgxintrin.h.
2672 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
2673 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
2674 * config/i386/i386.c (ix86_target_string): Add -msgx.
2676 (ix86_option_override_internal): Handle new options.
2677 (ix86_valid_target_attribute_inner_p): Add sgx.
2678 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
2679 * config/i386/i386.opt: Add msgx.
2680 * config/i386/sgxintrin.h: New file.
2681 * config/i386/x86intrin.h: Add sgxintrin.h.
2683 2017-01-11 Jakub Jelinek <jakub@redhat.com>
2686 * fold-const.c (maybe_nonzero_address): Return 1 for function
2688 (tree_single_nonzero_warnv_p): Don't handle function local objects
2692 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
2695 2017-01-11 David Malcolm <dmalcolm@redhat.com>
2698 * opts.c: Include "spellcheck.h"
2699 (struct string_fragment): New struct.
2700 (struct edit_distance_traits<const string_fragment &>): New
2702 (get_closest_sanitizer_option): New function.
2703 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
2705 2017-01-11 Jakub Jelinek <jakub@redhat.com>
2707 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
2709 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
2710 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
2711 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
2712 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
2713 for initial die_offset if dwarf_split_debug_info.
2714 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
2715 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
2717 (output_skeleton_debug_sections): Formatting fix. Use
2718 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
2719 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
2721 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
2723 * config/arm/cortex-a53.md: Add bypasses for
2725 (cortex_a53_r2f): Only use for transfers.
2726 (cortex_a53_f2r): Likewise.
2727 (cortex_a53_r2f_cvt): Add reservation for conversions.
2728 (cortex_a53_f2r_cvt): Likewise.
2730 2017-01-11 Tamar Christina <tamar.christina@arm.com>
2732 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
2733 to all inlined functions, change static to extern.
2735 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
2738 * config/arm/arm.c (legitimize_pic_address): Handle reference to
2740 (arm_assemble_integer): Likewise.
2742 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
2744 * config.gcc: Use new awk script to check CPU, FPU and architecture
2745 parameters for --with-... options.
2746 * config/arm/parsecpu.awk: New file
2747 * config/arm/arm-cpus.in: New file.
2748 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
2750 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
2752 * config/arm/t-arm: Update dependency rules.
2753 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
2754 of processing .def files.
2755 * config/arm/genopt.sh: Deleted.
2756 * config/arm/gentune.sh: Deleted.
2757 * config/arm/arm-cores.def: Deleted.
2758 * config/arm/arm-arches.def: Deleted.
2759 * config/arm/arm-fpus.def: Deleted.
2760 * config/arm/arm-tune.md: Regenerated.
2761 * config/arm/arm-tables.opt: Regenerated.
2762 * config/arm/arm-cpu.h: New generated file.
2763 * config/arm/arm-cpu-data.h: New generated file.
2764 * config/arm/arm-cpu-cdata.h: New generated file.
2766 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
2769 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
2771 (input_varpool_node): Unpack dynamically_initialized bit.
2773 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
2775 PR rtl-optimization/79032
2776 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
2777 the alignment of the adjusted memory reference against that of MODE,
2778 instead of the alignment of the original memory reference.
2780 2017-01-11 Martin Jambor <mjambor@suse.cz>
2782 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
2784 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
2785 decorated functions.
2787 2017-01-11 Richard Biener <rguenther@suse.de>
2789 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
2790 set range/nonnull info for PHI results. Do not set it on
2791 stmts marked for removal.
2793 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
2795 * expr.c (store_field): In the bitfield case, fetch the return value
2796 from the registers before applying a single big-endian adjustment.
2797 Always do a final load for a BLKmode value not larger than a word.
2799 2017-01-10 David Malcolm <dmalcolm@redhat.com>
2802 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
2803 that we correctly handle column numbers greater than
2804 LINE_MAP_MAX_COLUMN_NUMBER.
2806 2017-01-10 Martin Sebor <msebor@redhat.com>
2809 * gimple-ssa-sprintf.c (get_destination_size): Call
2810 {init,fini}object_sizes.
2811 * tree-object-size.c (addr_object_size): Adjust.
2812 (pass_through_call): Adjust.
2813 (pass_object_sizes::execute): Adjust.
2814 * tree-object-size.h (fini_object_sizes): Declare.
2816 2017-01-10 Martin Sebor <msebor@redhat.com>
2818 PR tree-optimization/78775
2819 * builtins.c (get_size_range): Move...
2820 * calls.c: ...to here.
2821 (alloc_max_size): Accept zero argument.
2822 (operand_signed_p): Remove.
2823 (maybe_warn_alloc_args_overflow): Call get_size_range.
2824 * calls.h (get_size_range): Declare.
2826 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
2828 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
2829 from TI's devices.csv file as of September 2016.
2830 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
2832 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
2834 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
2835 * doc/invoke.texi: Likewise.
2836 * doc/md.texi: Likewise.
2837 * doc/objc.texi: Likewise.
2839 2017-01-10 Joshua Conner <joshconner@google.com>
2841 * config/arm/fuchsia-elf.h: New file.
2842 * config/fuchsia.h: New file.
2843 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
2844 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
2846 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
2848 2016-01-10 Richard Biener <rguenther@suse.de>
2850 PR tree-optimization/79034
2851 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
2852 Propagate out degenerate PHIs in the joiner.
2854 2017-01-10 Martin Liska <mliska@suse.cz>
2856 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
2857 (sort_congruence_classes_by_decl_uid): Likewise.
2858 (sort_congruence_class_groups_by_decl_uid): Likewise.
2859 (sem_item_optimizer::merge_classes): Sort class, groups in these
2860 classes and members in the groups by DECL_UID of declarations.
2861 This would make merge operations stable.
2863 2017-01-10 Martin Liska <mliska@suse.cz>
2865 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
2866 usage of m_classes_vec.
2867 (sem_item_optimizer::~sem_item_optimizer): Likewise.
2868 (sem_item_optimizer::get_group_by_hash): Likewise.
2869 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
2870 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
2871 (sem_item_optimizer::verify_classes): Likewise.
2872 (sem_item_optimizer::process_cong_reduction): Likewise.
2873 (sem_item_optimizer::dump_cong_classes): Likewise.
2874 (sem_item_optimizer::merge_classes): Likewise.
2875 * ipa-icf.h (congruence_class_hash): Rename from
2876 congruence_class_group_hash. Remove declaration of m_classes_vec.
2878 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
2880 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
2881 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
2882 * config.gcc: Add avx512vpopcntdqintrin.h.
2883 * config/i386/avx512vpopcntdqintrin.h: New.
2884 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
2885 * config/i386/i386-builtin-types.def: Add new types.
2886 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
2887 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
2888 __builtin_ia32_vpopcountq_v8di_mask): New.
2889 * config/i386/i386-c.c (ix86_target_macros_internal): Define
2890 __AVX512VPOPCNTDQ__.
2891 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
2892 (PTA_AVX512VPOPCNTDQ): Define.
2893 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
2894 TARGET_AVX512VPOPCNTDQ_P): Define.
2895 * config/i386/i386.opt: Add mavx512vpopcntdq.
2896 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
2897 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
2899 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
2902 * predict.def (PRED_CALL): Set to 67.
2904 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
2906 * expr.c (store_field): In the bitfield case, if the value comes from
2907 a function call and is of an aggregate type returned in registers, do
2908 not modify the field mode; extract the value in all cases if the mode
2909 is BLKmode and the size is not larger than a word.
2911 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
2914 * config/i386/cpuid.h: Fix undefined behavior.
2916 2017-01-04 Jeff Law <law@redhat.com>
2918 PR tree-optimization/79007
2919 PR tree-optimization/67955
2920 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
2921 conservative for pt.null when flag_non_call_exceptions is on.
2923 2017-01-09 Jakub Jelinek <jakub@redhat.com>
2925 PR translation/79019
2926 PR translation/79020
2927 * params.def (PARAM_INLINE_MIN_SPEEDUP,
2928 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
2929 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
2931 * config/avr/avr.opt (maccumulate-args): Likewise.
2932 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
2933 * common.opt (freport-bug): Likewise.
2934 * cif-code.def (CIF_FINAL_ERROR): Likewise.
2935 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
2936 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
2937 translatable string.
2938 * config/i386/i386.c (function_value_32): Likewise.
2939 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
2940 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
2942 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
2943 * common/config/msp430/msp430-common.c (msp430_handle_option):
2945 * symtab.c (symtab_node::verify_base): Likewise.
2946 * opts.c (set_debug_level): Likewise.
2947 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
2948 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
2949 missing whitespace to translatable strings.
2950 * config/avr/avr.md (bswapsi2): Fix typo in comment.
2951 * config/sh/superh.h: Likewise.
2952 * config/i386/xopintrin.h: Likewise.
2953 * config/i386/znver1.md: Likewise.
2954 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
2955 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
2956 * double-int.h (struct double_int): Likewise.
2957 * double-int.c (div_and_round_double): Likewise.
2958 * wide-int.cc: Likewise.
2959 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
2960 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
2961 * cfgcleanup.c (crossjumps_occured): Renamed to ...
2962 (crossjumps_occurred): ... this.
2963 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
2966 PR tree-optimization/78899
2967 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
2968 returning bool return struct loop *, NULL for failure and the new
2970 (versionable_outer_loop_p): Don't version outer loop if it has
2971 dont_vectorized bit set.
2972 (tree_if_conversion): When versioning outer loop, ensure
2973 tree_if_conversion is performed also on the inner loop of the
2974 non-vectorizable outer loop copy.
2975 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
2976 LOOP_VECTORIZED in inner loop of the scalar outer loop and
2977 prevent vectorization of it.
2978 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
2979 the outer loop vectorization of the non-scalar version is attempted
2980 before vectorization of the inner loop in scalar version. If
2981 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
2982 vectorization of its inner loop.
2983 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
2984 has 2 inner loops, rename also on edges from bb whose single pred
2985 is outer_loop->header. Fix typo in function comment.
2987 2017-01-09 Martin Sebor <msebor@redhat.com>
2990 * asan.c (asan_emit_stack_protection): Increase local buffer size
2991 to avoid snprintf truncation warning.
2993 2017-01-09 Andrew Pinski <apinski@cavium.com>
2995 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
2996 to reference thunderx2t99 for the tuning structure
2997 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
2999 (thunderx2t99_extra_costs): This.
3000 * config/aarch64/aarch64-tune.md: Regenerate.
3001 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
3002 (vulcan_addrcost_table): This.
3003 (vulcan_regmove_cost): Rename to ...
3004 (thunderx2t99_regmove_cost): This.
3005 (vulcan_vector_cost): Rename to ...
3006 (thunderx2t99_vector_cost): this.
3007 (vulcan_branch_cost): Rename to ...
3008 (thunderx2t99_branch_cost): This.
3009 (vulcan_tunings): Rename to ...
3010 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
3011 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
3013 2017-01-09 Martin Jambor <mjambor@suse.cz>
3017 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
3018 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
3019 (propagate_vr_accross_jump_function): Use the above function for all
3020 value range computations for pass-through jump functions and type
3021 converasion from explicit value range values.
3022 (ipcp_propagate_stage): Do not attempt to deduce types of formal
3023 parameters from TYPE_ARG_TYPES.
3024 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
3025 (ipa_write_node_info): Stream type of the actual argument.
3026 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
3028 2017-01-09 Martin Liska <mliska@suse.cz>
3031 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
3032 (lookup_compiler): Do not show error message with have_E.
3034 2017-01-09 Jakub Jelinek <jakub@redhat.com>
3036 PR tree-optimization/78938
3037 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
3038 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
3039 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
3040 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
3043 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3045 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
3048 2017-01-09 Richard Biener <rguenther@suse.de>
3050 PR tree-optimization/78997
3051 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
3052 name condition properly.
3054 2017-01-09 Richard Biener <rguenther@suse.de>
3057 * dwarf2out.c (is_cxx): New overload with context.
3058 (is_naming_typedef_decl): Use it.
3060 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
3062 * invoke.texi (Option Summary): Correct spacing in option lists
3063 and add line breaks to fix over-long lines.
3065 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
3069 * extend.texi (Common Variable Attributes): Add xref to GCC
3070 Internals manual to explain mode attribute keywords.
3072 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
3075 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
3076 and Preprocessor Options.
3077 (Options for Linking): Document -pthread here....
3078 (RS/6000 and PowerPC Options): ...not here.
3079 (Solaris 2 Options): ...or here.
3080 * doc/cppopts.texi: Document -pthread.
3082 2017-01-08 Martin Sebor <msebor@redhat.com>
3085 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
3086 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
3087 New member functions.
3088 (format_directive): Used them.
3090 (pass_sprintf_length::handle_gimple_call): Same.
3091 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
3092 to avoid truncation for any argument.
3093 (extract_affine_mul): Same.
3094 * tree.c (get_file_function_name): Same.
3096 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
3099 * predict.def (PRED_INDIR_CALL): Set to 86.
3101 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
3103 PR preprocessor/54124
3104 * doc/cppopts.texi: Reformat -d subtable to list the full name
3105 of the options. Add cross-reference to the docs for the general
3106 compiler -d options.
3107 * doc/invoke.texi (Developer Options): Add cross-reference to the
3108 preprocessor-specific -d option documentation.
3110 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
3112 PR preprocessor/13498
3113 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
3114 redudant material, and reflect new command-line options.
3115 (System Headers): Likewise.
3117 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
3119 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
3120 -isystem, and -idirafter. Copy-edit.
3121 * doc/cppopts.texi: Copy-edit. Remove contradiction about
3122 default for -ftrack-macro-expansion. Delete obsolete and
3123 badly-formatted implementation details about -fdebug-cpp output.
3124 * doc/cppwarnopts.texi: Copy-edit.
3126 2017-01-07 David Malcolm <dmalcolm@redhat.com>
3129 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
3130 that the transition from a max line width >= 1<<10 to narrower
3131 lines works correctly.
3133 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
3135 * doc/options.texi (PerFunction): New.
3136 * opt-functions.awk (switch_flags): Map both Optimization and
3137 PerFunction to CL_OPTIMIZATION.
3138 * opth-gen.awk: Test for PerFunction flag along with
3140 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
3141 it only when the latter is present. Skip those that don't in
3142 the hash function generator.
3143 * common.opt (fvar-tracking): Mark as PerFunction instead of
3145 (fvar-tracking-assignments): Likewise.
3146 (fvar-tracking-assignments-toggle): Likewise.
3147 (fvar-tracking-uninit): Likewise.
3149 2017-01-07 Jakub Jelinek <jakub@redhat.com>
3151 PR translation/79018
3152 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
3155 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
3158 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
3159 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
3160 TARGET_LONG_JUMP_TABLE_OFFSETS.
3161 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
3162 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
3163 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
3164 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
3165 * config/m68k/m68k.md (tablejump expander): Likewise.
3166 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
3167 TARGET_LONG_JUMP_TABLE_OFFSETS.
3168 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
3169 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
3171 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
3172 David Holsgrove <david.holsgrove@xilinx.com>
3174 * common/config/microblaze/microblaze-common.c
3175 (TARGET_EXCEPT_UNWIND_INFO): Remove.
3176 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
3178 * config/microblaze/microblaze.c (microblaze_must_save_register)
3179 (microblaze_expand_epilogue, microblaze_return_addr): Handle
3181 (microblaze_eh_return): New function.
3182 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
3183 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
3184 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
3185 * config/microblaze/microblaze.md (eh_return): New pattern.
3187 2017-01-06 Jakub Jelinek <jakub@redhat.com>
3189 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
3190 GCC_DIAGNOSTIC_STRINGIFY): Define.
3192 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
3194 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
3196 * config/arm/arm.md (<mcrr>): New.
3198 * config/arm/arm.c (arm_arch5te): New.
3199 (arm_option_override): Set arm_arch5te.
3200 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
3202 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
3203 (arm_mcrr_qualifiers): ... this. New.
3204 (MRRC_QUALIFIERS): Define to...
3205 (arm_mrrc_qualifiers): ... this. New.
3206 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
3208 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
3209 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
3210 (MRRCI, mrrc, MRRC): New.
3211 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
3212 VUNSPEC_MRRC2): New.
3214 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
3216 * config/arm/arm.md (<mcr>): New.
3218 * config/arm/arm.c (arm_coproc_builtin_available): Add
3219 support for mcr, mrc, mcr2 and mrc2.
3220 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
3221 (arm_mcr_qualifiers): ... this. New.
3222 (MRC_QUALIFIERS): Define to ...
3223 (arm_mrc_qualifiers): ... this. New.
3224 (MCR_QUALIFIERS): Define to ...
3225 (arm_mcr_qualifiers): ... this. New.
3226 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
3228 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
3229 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
3230 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
3233 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
3235 * config/arm/arm.md (*ldc): New.
3239 * config/arm/arm.c (arm_coproc_builtin_available): Add
3240 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
3241 (arm_coproc_ldc_stc_legitimate_address): New.
3242 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
3243 'qualifier_const_pointer'.
3244 (LDC_QUALIFIERS): Define to...
3245 (arm_ldc_qualifiers): ... this. New.
3246 (STC_QUALIFIERS): Define to...
3247 (arm_stc_qualifiers): ... this. New.
3248 * config/arm/arm-protos.h
3249 (arm_coproc_ldc_stc_legitimate_address): New.
3250 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
3251 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
3252 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
3253 stc2, stcl, stc2l): New.
3254 * config/arm/constraints.md (Uz): New.
3255 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
3256 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
3257 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
3258 VUNSPEC_STC2L): New.
3260 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
3262 * config/arm/arm.md (<cdp>): New.
3263 * config/arm/arm.c (neon_const_bounds): Rename this ...
3264 (arm_const_bounds): ... this.
3265 (arm_coproc_builtin_available): New.
3266 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
3267 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
3268 (CDP_QUALIFIERS): Define to...
3269 (arm_cdp_qualifiers): ... this. New.
3271 (arm_expand_builtin_args): Add case for 6 arguments.
3272 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
3273 (arm_const_bounds): ... this.
3274 (arm_coproc_builtin_available): New.
3275 * config/arm/arm_acle.h (__arm_cdp): New.
3277 * config/arm/arm_acle_builtins.def (cdp): New.
3279 * config/arm/iterators.md (CDPI,CDP,cdp): New.
3280 * config/arm/neon.md: Rename all 'neon_const_bounds' to
3282 * config/arm/types.md (coproc): New.
3283 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
3284 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
3285 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
3286 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
3288 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
3290 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
3291 (UBINOP_QUALIFIERS): New.
3293 (acle_builtin_data): New. Change comment.
3294 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
3295 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
3296 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
3297 arm_acle_builtins.def.
3298 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
3299 (arm_init_acle_builtins): New.
3300 (CRC32_BUILTIN): Remove.
3301 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
3302 crc32cb, crc32ch and crc32cw.
3303 (arm_init_crc32_builtins): Remove.
3304 (arm_init_builtins): Use arm_init_acle_builtins rather
3305 than arm_init_crc32_builtins.
3306 (arm_expand_acle_builtin): New.
3307 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
3308 * config/arm/arm_acle_builtins.def: New.
3310 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
3312 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
3313 (arm_builtin_datum): ... this.
3314 (arm_init_neon_builtin): Rename to ...
3315 (arm_init_builtin): ... this. Add a new parameters PREFIX
3316 and USE_SIG_IN_NAME.
3317 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
3318 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
3319 'arm_builtin_datum'.
3320 (arm_init_vfp_builtins): Likewise.
3321 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
3322 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
3323 (arm_expand_neon_args): Rename to ...
3324 (arm_expand_builtin_args): ... this. Rename builtin_arg
3325 enum values and differentiate between ARG_BUILTIN_MEMORY
3326 and ARG_BUILTIN_NEON_MEMORY.
3327 (arm_expand_neon_builtin_1): Rename to ...
3328 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
3329 values, arm_expand_builtin_args and add bool parameter NEON.
3330 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
3331 (arm_expand_vfp_builtin): Likewise.
3332 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
3334 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
3337 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
3338 * predict.c (tree_estimate_probability_bb): Reverse direction of
3339 polymorphic call predictor.
3341 2017-01-06 David Malcolm <dmalcolm@redhat.com>
3343 * passes.c (execute_one_pass): Split out pass-skipping logic into...
3344 (determine_pass_name_match): ...this new function and...
3345 (should_skip_pass_p): ...this new function.
3347 2017-01-06 Nathan Sidwell <nathan@acm.org>
3349 * ipa-visibility.c (function_and_variable_visibility): Reformat
3350 comments and long lines. Remove extrneous if.
3351 * symtab.c (symtab_node::make_decl_local): Fix code format.
3352 (symtab_node::set_section_for_node): Fix comment typo.
3354 2017-01-06 Martin Liska <mliska@suse.cz>
3357 * lra-constraints.c: Rename invariant to lra_invariant.
3358 * predict.c (set_even_probabilities): Initialize e to NULL.
3360 2017-01-05 Martin Sebor <msebor@redhat.com>
3362 PR tree-optimization/78910
3363 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
3364 (format_integer): Correct off-by-one error in the handling
3365 of precision with negative numbers in signed conversions..
3367 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
3369 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
3371 2017-01-05 Jakub Jelinek <jakub@redhat.com>
3373 PR tree-optimization/71016
3374 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
3375 factor_out_conditional_conversion. Formatting fix.
3376 (factor_out_conditional_conversion): Add cond_stmt argument.
3377 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
3378 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
3381 2017-01-05 David Malcolm <dmalcolm@redhat.com>
3383 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
3384 read-rtl-function.o, and selftest-rtl.o.
3385 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
3386 (selftest::aarch64_test_loading_full_dump): New function.
3387 (selftest::aarch64_run_selftests): New function.
3388 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
3389 selftest::aarch64_run_selftests.
3390 * config/i386/i386.c
3391 (selftest::ix86_test_loading_dump_fragment_1): New function.
3392 (selftest::ix86_test_loading_call_insn): New function.
3393 (selftest::ix86_test_loading_full_dump): New function.
3394 (selftest::ix86_test_loading_unspec): New function.
3395 (selftest::ix86_run_selftests): Call the new functions.
3396 * emit-rtl.c (maybe_set_max_label_num): New function.
3397 * emit-rtl.h (maybe_set_max_label_num): New decl.
3398 * function.c (instantiate_decls): Guard call to
3399 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
3400 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
3402 * gensupport.c (gen_reader::gen_reader): Pass "false"
3403 for new "compact" param of rtx_reader.
3404 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
3405 rather than an empty string for NULL strings.
3406 * read-md.c: Potentially include config.h rather than bconfig.h.
3407 Wrap include of errors.h with #ifdef GENERATOR_FILE.
3408 (have_error): New global, copied from errors.c.
3409 (md_reader::read_name): Rename to...
3410 (md_reader::read_name_1): ...this, adding "out_loc" param,
3411 and converting "missing name or number" to returning false, rather
3413 (md_reader::read_name): Reimplement in terms of read_name_1.
3414 (md_reader::read_name_or_nil): New function.
3415 (md_reader::read_string): Handle "(nil)" by returning NULL.
3416 (md_reader::md_reader): Add new param "compact".
3417 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
3418 (md_reader::read_file): New method.
3419 * read-md.h (md_reader::md_reader): Add new param "compact".
3420 (md_reader::read_file): New method.
3421 (md_reader::is_compact): New accessor.
3422 (md_reader::read_name): Convert return type from void to file_location.
3423 (md_reader::read_name_or_nil): New decl.
3424 (md_reader::read_name_1): New decl.
3425 (md_reader::m_compact): New field.
3426 (noop_reader::noop_reader): Pass "false" for new "compact" param
3428 (rtx_reader::rtx_reader): Add new "compact" param.
3429 (rtx_reader::read_rtx_operand): Make virtual and convert return
3430 type from void to rtx.
3431 (rtx_reader::read_until): New decl.
3432 (rtx_reader::handle_any_trailing_information): New virtual function.
3433 (rtx_reader::postprocess): New virtual function.
3434 (rtx_reader::finalize_string): New virtual function.
3435 (rtx_reader::m_in_call_function_usage): New field.
3436 (rtx_reader::m_reuse_rtx_by_id): New field.
3437 * read-rtl-function.c: New file.
3438 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
3439 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
3440 (selftest::verify_three_block_rtl_cfg): New decl.
3441 * read-rtl-function.h: New file.
3442 * read-rtl.c: Potentially include config.h rather than bconfig.h.
3443 For host, include function.h, memmodel.h, and emit-rtl.h.
3444 (one_time_initialization): New function.
3445 (struct compact_insn_name): New struct.
3446 (compact_insn_names): New array.
3447 (find_code): Handle insn codes in compact dumps.
3448 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
3449 (bind_subst_iter_and_attr): Likewise.
3450 (add_condition_to_string): Likewise.
3451 (add_condition_to_rtx): Likewise.
3452 (apply_attribute_uses): Likewise.
3453 (add_current_iterators): Likewise.
3454 (apply_iterators): Likewise.
3455 (initialize_iterators): Guard usage of apply_subst_iterator with
3456 #ifdef GENERATOR_FILE.
3457 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
3458 (md_reader::read_mapping): Likewise.
3459 (add_define_attr_for_define_subst): Likewise.
3460 (add_define_subst_attr): Likewise.
3461 (read_subst_mapping): Likewise.
3462 (check_code_iterator): Likewise.
3463 (rtx_reader::read_rtx): Likewise. Move one-time initialization
3465 (one_time_initialization): New function.
3466 (rtx_reader::read_until): New method.
3467 (read_flags): New function.
3468 (parse_reg_note_name): New function.
3469 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
3470 Handle reuse_rtx ids.
3471 Wrap iterator lookup within #ifdef GENERATOR_FILE.
3472 Add parsing support for RTL dumps, mirroring the special-cases in
3473 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
3474 values, and calling handle_any_trailing_information.
3475 (rtx_reader::read_rtx_operand): Convert return type from void
3476 to rtx, returning return_rtx. Handle case 'e'. Call
3477 finalize_string on XSTR and XTMPL fields.
3478 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
3479 "(nil)" values were omitted. Call the postprocess vfunc on the
3481 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
3482 class ctor. Initialize m_in_call_function_usage. Call
3483 one_time_initialization.
3484 * rtl-tests.c (selftest::test_uncond_jump): Call
3485 set_new_first_and_last_insn.
3486 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
3487 * selftest-rtl.c: New file.
3488 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
3489 (selftest::get_insn_by_uid): New decl.
3490 * selftest-run-tests.c (selftest::run_tests): Call
3491 read_rtl_function_c_tests.
3492 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
3493 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
3496 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
3498 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
3499 operands in a special way. Assert that pos+len <= mode precision.
3501 2017-01-05 Jakub Jelinek <jakub@redhat.com>
3503 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
3504 3 argument Alias with unlimited for the negative form.
3505 (fno-vect-cost-model): Removed.
3507 2017-01-05 Martin Liska <mliska@suse.cz>
3509 * hsa-gen.c (gen_hsa_divmod): New function.
3510 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
3512 2017-01-05 Martin Liska <mliska@suse.cz>
3515 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
3518 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3520 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
3521 small constant length operands.
3523 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3525 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
3526 between loop iterations.
3528 2017-01-05 Martin Liska <mliska@suse.cz>
3531 * gimplify.c (gimplify_decl_expr): Compare to
3532 asan_poisoned_variables instread of checking flags.
3533 (gimplify_target_expr): Likewise.
3534 (gimplify_expr): Likewise.
3535 (gimplify_function_tree): Conditionally initialize
3536 asan_poisoned_variables.
3538 2017-01-04 Jeff Law <law@redhat.com>
3540 PR tree-optimizatin/78812
3541 * rtl.h (contains_mem_rtx_p): Prototype.
3542 * ifcvt.c (containts_mem_rtx_p): Move from here to...
3543 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
3544 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
3545 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
3546 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
3548 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
3550 * input.c (assert_char_at_range): Default-initialize actual_range.
3552 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
3554 * df-scan.c (df_ref_create_structure): Make regno unsigned,
3555 to match the caller.
3557 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
3559 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
3560 insns after final jump in test to emit dummy move.
3562 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
3564 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
3565 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
3567 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
3569 * multiple_target.c (create_dispatcher_calls): Init e_next.
3570 * tree-ssa-loop-split.c (split_loop): Init border.
3571 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
3574 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
3579 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
3580 (altivec_register_operand): Do not return true if the operand
3581 contains a SUBREG mixing SImode and SFmode.
3582 (vsx_register_operand): Likewise.
3583 (vsx_reg_sfsubreg_ok): New predicate.
3584 (vfloat_operand): Do not return true if the operand contains a
3585 SUBREG mixing SImode and SFmode.
3586 (vint_operand): Likewise.
3587 (vlogical_operand): Likewise.
3588 (gpc_reg_operand): Likewise.
3589 (int_reg_operand): Likewise.
3590 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
3591 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
3592 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
3594 (rs6000_emit_move_si_sf_subreg): New helper function.
3595 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
3596 fixup SUBREGs involving SImode and SFmode.
3597 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
3598 numbers for the new peephole2 optimization.
3599 (peephole2 for SFmode unions): New peephole2 to optimize cases in
3600 the GLIBC math library that do AND/IOR/XOR operations on single
3601 precision floating point.
3602 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
3603 target macros to say whether we need to avoid SUBREGs mixing
3605 (TARGET_ALLOW_SF_SUBREG): Likewise.
3606 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
3607 (UNSPEC_SI_FROM_SF): Likewise.
3608 (iorxor): Change spacing.
3609 (and_ior_xor): New iterator for AND, IOR, and XOR.
3610 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
3611 (movdi_from_sf_zero_ext): Likewise.
3612 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
3613 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
3614 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
3615 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
3616 (fms<mode>4): Likewise.
3617 (fnma<mode>4): Likewise.
3618 (fnms<mode>4): Likewise.
3619 (nfma<mode>4): Likewise.
3620 (nfms<mode>4): Likewise.
3622 2017-01-04 Marek Polacek <polacek@redhat.com>
3625 * doc/invoke.texi: Document -Wpointer-compare.
3627 2017-01-04 Jakub Jelinek <jakub@redhat.com>
3629 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
3632 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
3633 descriptions for -gdwarf-5 and emit them as uleb128 instead of
3636 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
3639 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
3640 documentation of the powerpc_popcntb_ok attribute.
3641 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3642 code to issue warning messages if a requested CPU configuration is
3643 not supported by the binary (assembler and loader) toolchain.
3644 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
3645 made to define a built-in function that has been disabled.
3646 (paired_init_builtins): Add assertion to prevent ICE if attempt is
3647 made to define a built-in function that has been disabled.
3648 (altivec_init_builtins): Add comment explaining why definition
3649 of the DST built-in functions is not preceded by an assertion
3650 check. Add assertions to prevent ICE if attempts are made to
3651 define an altivec predicate or an abs* built-in function that has
3653 (htm_init_builtins): Add comment explaining why definition of the
3654 htm built-in functions is not preceded by an assertion check.
3656 2017-01-04 Jeff Law <law@redhat.com>
3658 PR tree-optimizatin/67955
3659 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
3660 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
3661 the points-to solution does not include pt_null. Use DECL_PT_UID
3664 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
3666 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
3667 Use gen_int_mode instead of gen_lopwart for const_int operands.
3669 2017-01-04 Jakub Jelinek <jakub@redhat.com>
3671 PR tree-optimization/71563
3672 * match.pd: Simplify X << Y into X if Y is known to be 0 or
3673 out of range value - has low bits known to be zero.
3675 2017-01-04 Alan Modra <amodra@gmail.com>
3677 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
3678 * configure: Regenerate.
3679 * config.in: Regenerate.
3681 2017-01-04 Jakub Jelinek <jakub@redhat.com>
3684 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
3685 a substring of the message, but strcmp with the whole message. Ifdef
3686 ENABLE_NLS, translate the message first using dgettext.
3688 2017-01-03 Jeff Law <law@redhat.com>
3690 PR tree-optimizatin/78856
3691 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
3692 (mark_threaded_blocks): Remove code to truncate thread paths that
3693 cross multiple loop headers. Instead invalidate the cached loop
3694 iteration information and handle case of a thread path walking
3695 into an irreducible region.
3697 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
3700 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
3701 assertions. Add support for doing the signbit if the IEEE 128-bit
3702 floating point value is in a GPR.
3703 * config/rs6000/rs6000.md (Fsignbit): Delete.
3704 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
3705 Update the length attribute if the value is in a GPR.
3706 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
3707 the sign or zero extension instruction, since the value is always 0/1.
3708 (signbit<mode>2_dm2): Delete using <Fsignbit>.
3711 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
3712 extracting SImode to a GPR register so that we can generate a
3713 store, limit the vector to be in a traditional Altivec register
3714 for the vextuwrx instruction.
3716 2017-01-03 Ian Lance Taylor <iant@google.com>
3718 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
3720 2017-01-03 Martin Sebor <msebor@redhat.com>
3722 PR tree-optimization/78696
3723 * gimple-ssa-sprintf.c (format_floating): Correct handling of
3724 precision. Use MPFR for %f for greater fidelity. Correct handling
3726 (pass_sprintf_length::compute_format_length): Set width and precision
3727 specified by asrerisk to void_node for vararg functions.
3728 (try_substitute_return_value): Adjust dump output.
3730 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
3732 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
3734 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
3736 * doc/invoke.texi (SPARC options): Document -mlra as the default.
3737 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
3738 -mlra/-mno-lra was passed to the compiler.
3740 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
3742 PR rtl-optimization/65618
3743 * emit-rtl.c (try_split): Move initialization of "before" and
3744 "after" to just before the call to emit_insn_after_setloc.
3746 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
3748 * doc/md.texi (Standard Names): Remove reference to Java frontend.
3750 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
3752 * dwarf2out.c (gen_enumeration_type_die): When
3753 -gno-strict-dwarf, add a DW_AT_encoding attribute.
3755 2017-01-03 Jakub Jelinek <jakub@redhat.com>
3757 PR tree-optimization/78965
3758 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
3759 Change first argument from const call_info & to call_info &. For %n
3760 set info.nowrite to false.
3763 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
3764 possibly throwing calls.
3766 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
3767 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
3768 and fns handling, rather than in a separate case SSA_NAME.
3770 2017-01-02 Jeff Law <law@redhat.com>
3772 * config/darwin-driver.c (darwin_driver_init): Const-correctness
3773 fixes for first_period and second_period variables.
3775 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
3778 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
3779 (*insvqi_1): New insn pattern.
3780 (*insvqi_1_mem_rex64): Ditto.
3782 (*insvqi_3): Rename from *insvqi.
3784 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
3786 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
3788 * doc/cfg.texi (Edges): Remove reference to Java.
3789 (Maintaining the CFG): Ditto.
3791 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
3794 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
3795 transparent aliases.
3797 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
3800 * predict.def (PRED_CALL): Update hitrate.
3801 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
3802 * predict.c (tree_estimate_probability_bb): Split CALL predictor
3803 into direct/indirect/polymorphic variants.
3805 2017-01-01 Jakub Jelinek <jakub@redhat.com>
3807 Update copyright years.
3809 * gcc.c (process_command): Update copyright notice dates.
3810 * gcov-dump.c (print_version): Ditto.
3811 * gcov.c (print_version): Ditto.
3812 * gcov-tool.c (print_version): Ditto.
3813 * gengtype.c (create_file): Ditto.
3814 * doc/cpp.texi: Bump @copying's copyright year.
3815 * doc/cppinternals.texi: Ditto.
3816 * doc/gcc.texi: Ditto.
3817 * doc/gccint.texi: Ditto.
3818 * doc/gcov.texi: Ditto.
3819 * doc/install.texi: Ditto.
3820 * doc/invoke.texi: Ditto.
3822 Copyright (C) 2017 Free Software Foundation, Inc.
3824 Copying and distribution of this file, with or without modification,
3825 are permitted in any medium without royalty provided the copyright
3826 notice and this notice are preserved.