1 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
5 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
6 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
8 (rs6000_reassociation_width): Ditto.
9 (rs6000_emit_epilogue): Ditto.
10 (rs6000_adjust_cost): Ditto.
11 (is_microcoded_insn): Ditto.
12 (is_cracked_insn): Ditto.
13 (rs6000_adjust_priority): Ditto.
14 (rs6000_sched_reorder): Ditto.
15 (rs6000_sched_reorder2): Ditto.
16 (insn_must_be_first_in_group): Ditto.
17 (insn_must_be_last_in_group): Ditto.
18 (rs6000_register_move_cost): Ditto.
19 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
22 2017-12-07 Julia Koval <julia.koval@intel.com>
24 * config.gcc: Add vaesintrin.h.
25 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI): New type.
26 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
27 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi): New builtins.
28 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
29 * config/i386/immintrin.h: Include vaesintrin.h.
30 * config/i386/sse.md (vaesdec_<mode>): New pattern.
31 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
32 _mm_aesdec_epi128): New intrinsics.
34 2017-12-06 David Malcolm <dmalcolm@redhat.com>
36 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
37 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
38 c-family/c-spellcheck.cc.
39 (best_macro_match::best_macro_match): Likewise.
41 (struct edit_distance_traits<cpp_hashnode *>): Move to
42 c-family/c-spellcheck.h.
43 (class best_macro_match): Likewise.
45 2017-12-06 Jakub Jelinek <jakub@redhat.com>
47 PR tree-optimization/83293
48 * gimple-ssa-strength-reduction.c (insert_initializers): Use
49 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
50 might insert into empty bb.
53 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
54 simplify for plus with :c added, and pointer_plus without that.
55 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
56 with undefined overflow and the conversion is not widening,
57 perform negation in utype and only convert to type afterwards.
58 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
59 simplify for plus with :c added, and pointer_plus without that.
60 If type is integral with undefined overflow and the conversion is
61 not widening, perform minus in utype and only convert to type
62 afterwards. Move the last pointer_diff_expr simplify into the
65 2017-12-06 Martin Sebor <msebor@redhat.com>
67 PR tree-optimization/82646
68 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
71 2017-12-06 Martin Sebor <msebor@redhat.com>
73 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
76 PR tree-optimization/83075
77 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
78 strncat/strncpy don't change length of source string.
80 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
83 2017-11-29 Martin Aberg <maberg@gaisler.com>
85 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
86 to prevent b2bst errata sequence.
87 (sqrtdf2_fix): Likewise.
89 2017-12-06 Jakub Jelinek <jakub@redhat.com>
91 PR tree-optimization/81945
92 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
93 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
94 to dest_cfun has orig_loop_num set, either remap it to the new
95 loop number if the loop got moved too, or clear it.
97 2017-12-05 Steve Ellcey <sellcey@cavium.com>
99 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
101 (thunderx2t99_nothing): New insn reservation.
102 (thunderx2t99_mrs): New insn reservation.
103 (thunderx2t99_multiple): New insn reservation.
104 (thunderx2t99_alu_basi): Add bfx to reservation.
105 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
107 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
110 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
112 2017-12-05 Bin Cheng <bin.cheng@arm.com>
114 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
115 tree-ssa-pre.c::remove_dead_inserted_code.
116 * tree-ssa-dce.h: New file.
117 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
118 (remove_dead_inserted_code): Move and rename to function
119 tree-ssa-dce.c::simple_dce_from_worklist.
120 (pass_pre::execute): Update use.
122 2017-12-05 Richard Biener <rguenther@suse.de>
124 PR tree-optimization/83277
125 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
126 to code-gen liveout vars.
128 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
130 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
131 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
132 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
135 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
137 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
139 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
140 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
143 2017-12-05 Richard Biener <rguenther@suse.de>
145 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL):
147 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
148 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
149 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
151 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
153 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
155 2017-12-05 Martin Liska <mliska@suse.cz>
156 Jakub Jelinek <jakub@redhat.com>
158 * doc/invoke.texi: Document the options.
159 * flag-types.h (enum sanitize_code): Add
160 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
161 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
162 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
163 * opts.c: Define new sanitizer options.
164 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
165 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
167 2017-12-05 Julia Koval <julia.koval@intel.com>
169 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
170 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
171 (ix86_handle_option): Handle -mavx512vnni.
172 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
173 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
174 * config/i386/i386-c (__AVX512VNNI__): New.
175 * config/i386/i386.c (ix86_target_string): Handle new option.
176 (ix86_valid_target_attribute_inner_p): Handle new option.
177 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
178 * config/i386/i386.opt (mavx512vnni): New option.
180 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
183 * athlon.md: Disable for generic.
184 * haswell.md: Enable for generic.
185 * i386.c (ix86_sched_init_global): Add core hooks for generic.
186 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
188 (ix86_adjust_cost): Move generic to haswell path.
190 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
192 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
193 instead of MEM_P in a couple more places. Fix formatting issues.
195 2017-12-04 Jim Wilson <jimw@sifive.com>
197 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
198 instead of GP_REG_LAST-1.
199 (riscv_adjust_libcall_cfi_prologue): Likewise.
200 (riscv_adjust_libcall_cri_epilogue): Likewise.
201 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
204 2017-12-04 Luis Machado <luis.machado@linaro.org>
206 * ipa-pure-const.c (check_decl): Add missing newline.
207 (state_from_flags): Likewise.
209 2017-12-04 Jeff Law <law@redhat.com>
211 PR tree-optimizatin/78496
212 * gimple-ssa-evrp-analyze.h
213 (evrp_range_analyzer::get_vr_values): Simplify.
214 * gimple-ssa-evrp-analyze.c: Corresponding changes.
215 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
216 and gimple-ssa-evrp-analyze.h.
217 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
218 (simplify_stmt_for_jump_threading): Copy a blob of code from
219 tree-vrp.c to use ranges to simplify statements.
220 (dom_opt_dom_walker::before_dom_children): Call
221 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
222 (dom_opt_dom_walker::after_dom_children): Similarly for
223 evrp_range_analyzer::leave.
224 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
227 * gimple-ssa-evrp-analyze.c
228 (evrp_range_analyzer::extract_range_from_stmt): Always use
229 vr_values::update_value_range so preexisting range info is
230 medged with new range info, even if the new range is VR_VARYING.
232 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
234 * combine.c: Adjust comment.
235 (use_crosses_set_p): Delete.
236 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
237 (try_combine): Ditto.
239 2017-12-04 Richard Biener <rguenther@suse.de>
241 PR tree-optimization/83255
242 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
243 Re-add zero-iteration check.
245 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
247 PR rtl-optimization/83245
248 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
249 hard registers as earlyclobber, also if not in an asm.
251 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
255 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
258 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
259 rs6000_cpu based on cpu_index, not tune_index.
261 2017-12-04 Richard Biener <rguenther@suse.de>
263 PR tree-optimization/83238
264 * graphite-scop-detection.c (scop_detection::merge_sese): Make
265 code match comment, rejecting invalid SESE regions.
267 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
269 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
270 require base operand is a REG_POINTER prior to reload on targets
271 with non-equivalent space registers.
273 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
275 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
276 (update_specialized_profile): Fix updating of counts.
277 (perhaps_add_new_callers): Likewise.
279 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
282 * x86-tune.def: Remove obsolette FIXMEs.
283 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
284 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
285 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
287 (X86_TUNE_PAD_RETURNS): Disable for generic.
289 2017-12-02 Jakub Jelinek <jakub@redhat.com>
291 PR tree-optimization/83170
292 PR tree-optimization/83241
293 * gimple-ssa-store-merging.c
294 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
295 gimple_vuse (ins_stmt) in case it has changed.
296 (imm_store_chain_info::output_merged_store): Likewise.
298 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
302 * tree-cfg.c (pass_warn_function_return::execute): Handle
303 __builtin_ubsan_handle_missing_return like __builtin_unreachable
304 with BUILTINS_LOCATION.
308 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
309 is BLKmode for vector field with vector raw mode, use TYPE_MODE
310 instead of DECL_MODE.
312 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
313 last argument to rtx pointer.
314 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
315 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
316 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
317 depending on the chosen ISAs.
318 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
319 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
320 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
322 * config/i386/sse.md (mov<mode>_internal): Likewise.
323 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
325 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
327 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
329 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
331 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
332 parameters from prototype.
333 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
334 parameters. Don't print a comment.
335 (emit_fusion_gpr_load): Adjust.
336 (emit_fusion_load_store): Adjust.
337 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
338 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
339 comment on the second line.
341 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
344 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
345 rs6000_cpu based on cpu_index, not tune_index.
347 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
349 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
350 which_alternative instead of which_alternative + 1.
351 (output_asm_insn): Print an extra tab if the template is short.
353 2017-12-01 Jim Wilson <jimw@sifive.com>
355 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
357 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
358 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
360 * doc/tm.texi: Regenerate.
362 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
365 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
366 whether we can allocate pseudos before trying to fix an address.
367 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
368 memory address is indexed or indirect.
369 (floatuns_<mode>si2_hw2): Likewise.
371 2017-12-01 Jason Merrill <jason@redhat.com>
373 * Makefile.in (TAGS): Add c-family/*.cc.
375 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
377 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
378 (call_value_insn): Likewise.
379 (sibcall_insn): Likewise.
380 (sibcall_value_insn): Likewise.
381 (movsi_aarch64): Likewise.
382 (movdi_aarch64): Likewise.
383 (add_losym_): Likewise.
384 (ldr_got_small_): Likewise.
385 (ldr_got_small_sidi): Likewise.
386 (ldr_got_small_28k_): Likewise.
387 (ldr_got_small_28k_sidi): Likewise.
388 * config/aarch64/aarch64.c (aarch64_print_address_internal):
389 Move output_addr_const to symbolic case. Add error check.
391 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
393 * config/s390/predicates.md (plus16_Q_operand): New predicate.
394 * config/s390/s390.md: Disable MVC merging peephole if it would
395 disable operand forwarding.
396 (new peephole2): Split MVCs if it would turn them into up to 2
399 2017-12-01 Richard Biener <rguenther@suse.de>
401 PR tree-optimization/83232
402 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
403 detection of same access. Instead of breaking the group here
404 do not consider the duplicate. Add comment explaining real fix.
406 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
408 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
410 2017-12-01 Sudakshina Das <sudi.das@arm.com>
412 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
414 2017-12-01 Jakub Jelinek <jakub@redhat.com>
416 * function.h (struct function): Remove cilk_frame_decl,
417 is_cilk_function and calls_cilk_spawn fields.
418 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
420 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
421 cilk_elemental field.
422 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
423 * target.def: Adjust comment.
424 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
425 Don't test cilk_elemental.
427 PR tree-optimization/83233
428 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
429 bswap_stat name for the struct.
432 * tree.h (SWITCH_BREAK_LABEL_P): Define.
433 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
434 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
435 SWITCH_BREAK_LABEL_P set on the label.
436 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
437 added for default case if it was missing and not all cases covered.
438 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
439 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
441 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
442 to avoid -Wimplicit-fallthrough warning.
443 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
444 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
446 PR tree-optimization/83221
447 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
449 (init_reassoc): Formatting fix.
452 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
453 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
457 * tree-cfg.c: Include asan.h.
458 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
460 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
462 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
464 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
466 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
467 vpcmpeqd instruction.
469 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
471 * config/i386/i386.c (standard_sse_constant_opcode): Fix
472 registers type for 128bit mode.
474 2017-11-30 Jakub Jelinek <jakub@redhat.com>
476 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
478 * gengtype-state.c (read_state_pair): Likewise.
479 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
480 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
481 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
482 * ifcvt.c (noce_try_store_flag_constants): Likewise.
483 * tree-ssa-ccp.c (ccp_finalize): Likewise.
484 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
485 * builtins.c (fold_builtin_3): Likewise.
486 * graphite-scop-detection.c
487 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
488 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
491 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
494 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
495 parameter. Return true if ignore_decl_rtl_set_p is true and other
496 conditions are satisfied.
497 * asan.h (asan_protect_global): Add new parameter.
498 * varasm.c (categorize_decl_for_section): Pass true as second parameter
499 to asan_protect_global calls.
501 2017-11-30 Jim Wilson <jimw@sifive.com>
503 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
504 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
505 -mstrict-align, add info on default value. Delete redundant lines for
506 -mabi. Add missing -mexplicit-relocs docs.
508 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
510 * config/arc/arc.md (trap): New pattern.
512 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
514 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
515 instruction to end into a delay slot.
516 * config/arc/arc.md (cond_delay_insn): Check if the instruction
517 can be placed into a delay slot against reg_note.
519 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
521 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
522 labels number of usages.
524 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
526 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
528 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
530 2017-11-30 Jakub Jelinek <jakub@redhat.com>
533 * internal-fn.c (expand_mul_overflow): Optimize unsigned
534 multiplication by power of 2 constant into two shifts + comparison.
536 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
539 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
541 2017-11-30 Richard Biener <rguenther@suse.de>
543 PR tree-optimization/83202
544 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
545 allow_peel argument and guard peeling.
546 (canonicalize_loop_induction_variables): Likewise.
547 (canonicalize_induction_variables): Pass false.
548 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
549 peeling from cunrolli.
551 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
553 * combine.c (try_combine): Print a message to dump file whenever
554 I0, I1, or I2 cannot be combined into I3.
556 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
558 PR rtl-optimization/83156
559 PR rtl-optimization/82621
560 * combine.c (try_combine): Don't split an I2 if one of the dests is
561 set again before I3. Allow unused dests.
563 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
565 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
567 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
569 PR rtl-optimization/80818
570 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
571 recursively. Use insn code for clobber.
572 (lra_set_insn_recog_data): Pass the new arg to
573 collect_non_operand_hard_regs.
574 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
576 (lra_update_insn_regno_info): Pass insn to
577 add_regs_to_insn_regno_info.
579 2017-11-29 Jim Wilson <jimw@sifive.com>
580 Andrew Waterman <andrew@sifive.com>
582 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
583 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
584 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
585 use SINGLE_SHIFT_COST.
586 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
587 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
589 2017-11-29 Julia Koval <julia.koval@intel.com>
591 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
592 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
593 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
594 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
595 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
596 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
597 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
598 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
599 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
600 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
601 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
602 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
603 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
604 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
605 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
606 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
607 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
608 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
609 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
610 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
611 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
612 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
613 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
614 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
615 __builtin_ia32_vpshldv_v2di_mask,
616 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
617 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
618 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
620 2017-11-29 Julia Koval <julia.koval@intel.com>
622 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
623 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
624 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
625 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
626 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
627 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
628 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
629 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
630 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
631 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
632 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
633 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
634 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
635 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
636 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
637 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
638 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
639 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
640 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
641 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
642 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
644 2017-11-29 Daniel Cederman <cederman@gaisler.com>
646 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
647 movsi_pic_gotdata_op instruction as a load for the UT699 errata
650 2017-11-29 Martin Aberg <maberg@gaisler.com>
652 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
653 to prevent b2bst errata sequence.
654 (sqrtdf2_fix): Likewise.
656 2017-11-29 Daniel Cederman <cederman@gaisler.com>
658 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
659 (div_sqrt_insn_p): New function.
660 (sparc_do_work_around_errata): Insert NOP instructions to
661 prevent sequences that could trigger the TN-0013 errata for
662 certain LEON3 processors.
663 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
664 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
665 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
666 (in_branch_delay): Prevent div and sqrt in delay slot if
668 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
670 2017-11-29 Daniel Cederman <cederman@gaisler.com>
672 * config/sparc/sparc.c (atomic_insn_p): New function.
673 (sparc_do_work_around_errata): Insert NOP instructions to
674 prevent sequences that could trigger the TN-0010 errata for
676 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
677 instruction referable in atomic_insns_p.
679 2017-11-29 Daniel Cederman <cederman@gaisler.com>
681 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
682 (atomic_compare_and_swap_leon3_1): Likewise.
685 2017-11-29 Daniel Cederman <cederman@gaisler.com>
687 * config/sparc/sparc.c (fpop_insn_p): New function.
688 (sparc_do_work_around_errata): Insert NOP instructions to
689 prevent sequences that could trigger the TN-0012 errata for
691 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
692 * config/sparc/sparc.md (fix_gr712rc): New attribute.
693 (in_branch_annul_delay): Prevent floating-point instructions
694 in delay slot of annulled integer branch.
696 2017-11-29 Richard Biener <rguenther@suse.de>
698 PR tree-optimization/83202
699 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
701 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
702 nodes vectorized to the same stmts multiple times.
703 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
704 (vect_analyze_slp_instance): Adjust.
705 (scalar_stmts_to_slp_tree_map_t): New typedef.
706 (vect_schedule_slp_instance): Add a map recording the SLP node
707 representing the vectorized stmts for a set of scalar stmts.
708 Avoid code-generating redundancies.
709 (vect_schedule_slp): Allocate map and pass it down.
711 2017-11-29 Nathan Sidwell <nathan@acm.org>
714 * tree.c (build_complex_type): Fix canonicalization. Only fill in
717 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
719 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
720 cases for printing LDP/STP memory addresses.
721 (aarch64_print_address_internal): Renamed from
722 aarch64_print_operand_address, added parameter, add Pmode check.
723 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
724 (aarch64_print_operand_address): Indirect to
725 aarch64_print_address_internal.
726 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
727 'y' operand output specifier.
729 2017-11-29 Jakub Jelinek <jakub@redhat.com>
732 * tree.c (build_simple_mem_ref_loc): Handle
733 get_addr_base_and_unit_offset returning a MEM_REF.
736 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
739 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
742 2017-11-28 David Malcolm <dmalcolm@redhat.com>
744 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
746 (layout::move_to_column): Likewise.
748 2017-11-28 Jakub Jelinek <jakub@redhat.com>
751 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
752 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
754 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
755 Martin Jambor <mjambor@suse.cz>
758 * tree.h (expr_type_first_operand_type_p): Declare
759 * tree.c (expr_type_first_operand_type_p): New function.
760 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
761 (ipa_value_from_jfunc): Adjust declaration.
762 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
763 Use it as result type for arithmetics, unless it is NULL in which case
764 be more conservative.
765 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
766 ipa_get_jf_pass_through_result.
767 (propagate_vals_across_pass_through): Likewise.
768 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
769 is to propagate_vals_across_pass_through.
770 (propagate_constants_across_call): Pass PARM_TYPE to
771 propagate_scalar_across_jump_function.
772 (find_more_scalar_values_for_callers_subset): Pass parameter type to
773 ipa_value_from_jfunc.
774 (cgraph_edge_brings_all_scalars_for_node): Likewise.
775 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
776 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
777 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
778 target_type, pass it to ipa_value_from_jfunc.
779 (update_indirect_edges_after_inlining): Pass parameter type to
780 try_make_edge_direct_simple_call.
782 2017-11-28 Jeff Law <law@redhat.com>
784 * gimple-ssa-evrp-analyze.c
785 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
786 refine ranges if scev_initialized_p returns true.
787 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
789 2017-11-28 Julia Koval <julia.koval@intel.com>
791 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
792 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
793 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
794 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
795 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
796 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
797 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
798 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
799 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
800 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
801 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
802 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
803 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
804 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
805 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
806 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
807 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
808 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
809 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
810 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
811 __builtin_ia32_vpshrd_v2di_mask): New builtins.
812 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
814 2017-11-28 Julia Koval <julia.koval@intel.com>
816 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
817 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
818 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
819 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
820 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
821 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
822 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
823 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
824 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
825 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
826 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
827 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
828 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
829 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
830 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
831 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
832 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
833 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
834 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
835 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
836 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
837 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
838 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
839 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
840 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
841 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
842 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
843 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
844 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
845 __builtin_ia32_vpshld_v2di_mask): New builtins.
846 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
847 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
849 2017-11-28 Richard Biener <rguenther@suse.de>
851 PR tree-optimization/80776
852 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
854 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
856 (evrp_range_analyzer::record_ranges_from_incoming_edges):
857 If the incoming edge is an effective fallthru because the other
858 edge only reaches a __builtin_unreachable () then record ranges
859 derived from the controlling condition in SSA info.
860 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
861 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
863 2017-11-28 Olivier Hainque <hainque@adacore.com>
865 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
866 on mingw build hosts.
868 2017-11-28 Jakub Jelinek <jakub@redhat.com>
871 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
873 2017-11-28 Richard Biener <rguenther@suse.de>
876 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
877 copies generated from memcpy use a character array as reference
880 2017-11-28 Julia Koval <julia.koval@intel.com>
881 Sebastian Peryt <sebastian.peryt@intel.com>
883 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
884 c-family/c-cilkplus.o, c-family/array-notation-common.o,
885 cilk-common.o, cilk.h, cilk-common.c): Remove.
887 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
888 * builtins.c (is_builtin_name): Remove cilkplus condition.
889 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
890 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
891 cilk-builtins.def, cilkplus.def): Remove.
892 * cif-code.def (CILK_SPAWN): Remove.
893 * cilk-builtins.def: Delete.
894 * cilk-common.c: Ditto.
896 * cilkplus.def: Ditto.
897 * config/darwin.h (fcilkplus): Delete.
898 * cppbuiltin.c: Ditto.
899 * doc/extend.texi: Remove cilkplus doc.
900 * doc/generic.texi: Ditto.
901 * doc/invoke.texi: Ditto.
902 * doc/passes.texi: Ditto.
903 * gcc.c (fcilkplus): Remove.
904 * gengtype.c (cilk.h): Remove.
905 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
907 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
909 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
911 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
912 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
914 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
915 inline_read_section): Ditto.
916 * ipa-inline-analysis.c (cilk.h): Remove.
917 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
918 * lto-wrapper.c (merge_and_complain, append_compiler_options,
919 append_linker_options): Remove condition for fcilkplus.
920 * lto/lto-lang.c (cilk.h): Remove.
921 (lto_init): Remove condition for fcilkplus.
922 * omp-expand.c (expand_cilk_for_call): Delete.
923 (expand_omp_taskreg, expand_omp_for_static_chunk,
924 expand_omp_for): Remove cilkplus
926 (expand_cilk_for): Delete.
927 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
928 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
929 execute_lower_omp, diagnose_sb_0): Ditto.
930 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
931 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
932 * tree-nested.c: Ditto.
933 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
934 (dump_generic_node): Ditto.
935 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
936 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
938 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
940 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
942 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
943 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
944 for vector mode and !TARGET_SIMD.
946 2017-11-28 Jakub Jelinek <jakub@redhat.com>
948 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
950 * tree.h (SWITCH_LABELS): Remove.
951 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
952 assert SWITCH_BODY is non-NULL.
953 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
955 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
957 PR tree-optimization/80788
958 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
959 has TREE_OVERFLOW set, call drop_tree_overflow.
961 2017-11-28 Richard Biener <rguenther@suse.de>
963 PR tree-optimization/83158
964 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more
967 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
970 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
971 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
973 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
975 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
976 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
978 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
981 * varasm.c (compare_constant): Take the mode of the constants into
982 account when comparing floating point constants.
984 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
986 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
987 from explicit instantiation of debug_helper.
988 * vec.h (DEFINE_DEBUG_VEC): Ditto.
990 2017-11-27 Richard Biener <rguenther@suse.de>
992 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
995 2017-11-27 Richard Biener <rguenther@suse.de>
997 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
998 (type_hash_canon): Also clear int_cst_hash_table entry for
1000 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
1002 2017-11-27 Tamar Christina <tamar.christina@arm.com>
1004 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1005 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1006 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1008 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1010 * hash-map.h (gt_cleare_cache): Avoid UB.
1012 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
1014 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
1015 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
1016 (decide_unroll_runtime_iterations): Likewise.
1017 (decide_unroll_stupid): Likewise.
1019 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1022 * config/i386/i386.md: Add a loop with incssp.
1024 2017-11-27 Martin Jambor <mjambor@suse.cz>
1026 PR tree-optimization/81248
1027 * tree-sra.c (splice_param_accesses): Remove size check.
1028 (decide_one_param_reduction): Fix size check.
1029 * gimple-pretty-print.c (dump_profile): Silence warning.
1030 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
1032 2017-11-27 Jakub Jelinek <jakub@redhat.com>
1035 * dbxout.c (lastlineno): New variable.
1036 (dbx_debug_hooks): Use dbxout_switch_text_section as
1037 switch_text_section debug hook.
1038 (dbxout_function_end): Switch to current_function_section
1039 rather than function_section. If crtl->has_bb_partition,
1040 output just one N_FUN, depending on in_cold_section_p.
1041 (dbxout_source_line): Remember last lineno in lastlineno.
1042 (dbxout_switch_text_section): New function.
1043 (dbxout_function_decl): Adjust dbxout_block caller.
1044 (dbx_block_with_cold_children): New function.
1045 (dbxout_block): Return true if any LBRAC/RBRAC have been
1046 emitted. Use dbx_block_with_cold_children at depth == 0
1047 in second partition. Add PARENT_BLOCKNUM argument, pass
1048 it optionally adjusted to children. Output LBRAC/RBRAC
1049 around recursive call only if the block is in the current
1050 partition, if not and anything was output, emit empty
1052 * final.c (final_scan_insn): Compute cold_function_name
1053 before calling switch_text_section debug hook. Call
1054 that hook even if dwarf2out_do_frame if not emitting
1058 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
1059 TREE_READONLY decls.
1061 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1063 PR rtl-optimization/82488
1064 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
1066 2017-11-26 Julia Koval <julia.koval@intel.com>
1068 * config/i386/i386.c (processor_target_table): Add skylake_cost for
1070 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
1073 2017-11-26 Julia Koval <julia.koval@intel.com>
1075 * config/i386/driver-i386.c (host_detect_local_cpu):
1076 Detect skylake-avx512.
1078 2017-11-26 Julia Koval <julia.koval@intel.com>
1080 * config.gcc: Add -march=cannonlake.
1081 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
1082 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
1083 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
1084 (PTA_CANNONLAKE): New.
1085 (processor_target_table): Add cannonlake.
1086 (ix86_option_override_internal): Ditto.
1087 (fold_builtin_cpu): Ditto.
1088 (get_builtin_code_for_version): Handle cannonlake.
1089 (M_INTEL_COREI7_CANNONLAKE): New.
1090 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
1091 * doc/invoke.texi: Add -march=cannonlake.
1093 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
1095 * plugin.c (add_new_plugin): Use platform-specific library extensions.
1096 (try_init_one_plugin): Alternative implementation for MinGW.
1097 * Makefile.in (plugin_implib): New.
1098 (gengtype-lex.c): Fix broken AIX workaround.
1099 * configure: Regenerate.
1100 * doc/plugins.texi: Document support for MinGW.
1102 2017-11-25 Jakub Jelinek <jakub@redhat.com>
1104 PR rtl-optimization/81553
1105 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
1106 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
1107 is a shift where C1 has different mode than the whole shift, use C1's
1108 mode for MULT rather than the shift's mode.
1111 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
1112 builtins not enabled in the currently selected ISA.
1114 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
1116 PR tree-optimization/71026
1117 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
1118 (insert_reciprocals): Change to insert reciprocals before a division
1119 by a square and to insert the square of a reciprocal.
1120 (execute_cse_reciprocals_1): Change to consider division by a square.
1121 (register_division_in): Add importance parameter.
1123 2017-11-24 Richard Biener <rguenther@suse.de>
1125 PR tree-optimization/82402
1126 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
1127 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
1129 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
1131 * match.pd (0-ptr): New transformation.
1133 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
1136 * ipa-inline.c (inline_small_functions): Set current badnes correctly
1137 when skipping checking.
1139 2017-11-24 Richard Biener <rguenther@suse.de>
1141 PR tree-optimization/83128
1142 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
1143 (vn_reference_lookup_3): Likewise.
1145 2017-11-24 Jakub Jelinek <jakub@redhat.com>
1148 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
1149 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
1152 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
1153 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
1154 check for the latter with gimple_call_builtin_p. Do not handle
1155 BUILT_IN_STPNCPY_CHK which is not a pass through call.
1157 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
1159 * config/arm/arm_neon.h: Fix pragma GCC push_options before
1162 2017-11-23 Julia Koval <julia.koval@intel.com>
1164 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
1165 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
1166 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
1167 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
1168 _mm512_maskz_expandloadu_epi16): New intrinsics.
1169 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
1170 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
1171 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
1172 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
1173 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
1174 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
1175 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
1176 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
1177 _mm256_maskz_expandloadu_epi8): New intrinsics.
1178 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
1179 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
1180 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
1181 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
1182 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
1183 * config/i386/sse.md (VI248_VLBW): New iterator.
1184 (expand<mode>_mask, expand<mode>_maskz): New patterns.
1186 2017-11-23 Julia Koval <julia.koval@intel.com>
1188 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
1189 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
1190 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
1191 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
1192 _mm512_mask_compressstoreu_epi16): New.
1193 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
1194 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
1195 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
1196 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
1197 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
1198 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
1199 _mm256_mask_compressstoreu_epi8): New.
1200 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
1201 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
1202 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
1203 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
1204 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
1205 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
1206 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
1207 __builtin_ia32_compresshi128_mask,
1208 __builtin_ia32_compressstoreuqi512_mask,
1209 __builtin_ia32_compressstoreuhi512_mask,
1210 __builtin_ia32_compressstoreuqi256_mask,
1211 __builtin_ia32_compressstoreuqi128_mask,
1212 __builtin_ia32_compressstoreuhi256_mask,
1213 __builtin_ia32_compressstoreuhi128_mask): New builtins.
1214 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
1216 (ix86_expand_special_args_builtin): Handle new types.
1217 (s4fma_expand): Handle new builtin array.
1218 * config/i386/immintrin.h: Include new headers.
1219 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
1220 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
1222 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1224 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
1225 fixes. Declare temp and g variables at the top in order to avoid
1226 {} in most of the cases.
1228 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
1230 * match.pd (ptr-0): New transformation.
1232 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
1234 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
1235 (struct addr_mode_cost_table): New.
1236 (struct tune_params): Add field addr_mode_costs.
1237 * config/arm/arm.c (generic_addr_mode_costs): New.
1238 (arm_slowmul_tune): Initialise addr_mode_costs field.
1239 (arm_fastmul_tune): Likewise.
1240 (arm_strongarm_tune): Likewise.
1241 (arm_xscale_tune): Likewise.
1242 (arm_9e_tune): Likewise.
1243 (arm_marvell_pj4_tune): Likewise.
1244 (arm_v6t2_tune): Likewise.
1245 (arm_cortex_tune): Likewise.
1246 (arm_cortex_a8_tune): Likewise.
1247 (arm_cortex_a7_tune): Likewise.
1248 (arm_cortex_a15_tune): Likewise.
1249 (arm_cortex_a35_tune): Likewise.
1250 (arm_cortex_a53_tune): Likewise.
1251 (arm_cortex_a57_tune): Likewise.
1252 (arm_exynosm1_tune): Likewise.
1253 (arm_xgene1_tune): Likewise.
1254 (arm_cortex_a5_tune): Likewise.
1255 (arm_cortex_a9_tune): Likewise.
1256 (arm_cortex_a12_tune): Likewise.
1257 (arm_cortex_a73_tune): Likewise.
1258 (arm_v7m_tune): Likewise.
1259 (arm_cortex_m7_tune): Likewise.
1260 (arm_v6m_tune): Likewise.
1261 (arm_fa726te_tune): Likewise.
1262 (arm_mem_costs): Use table lookup to calculate cost of addressing
1265 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
1267 * config/arm/arm.c (arm_mem_costs): New function.
1268 (arm_rtx_costs_internal): Use arm_mem_costs.
1270 2017-11-23 Mark Wielaard <mark@klomp.org>
1272 * dwarf2out.c (init_sections_and_labels): Use generation to create
1273 unique ranges_section_label and ranges_base_label. Return generation.
1274 (output_rnglists): Add generation argument. Use generation to create
1275 unique ranges labels.
1276 (dwarf2out_finish): Get generation from init_sections_and_labels
1277 and pass generation to output_rnglists.
1279 2017-11-23 Mike Stump <mikestump@comcast.net>
1280 Eric Botcazou <ebotcazou@adacore.com>
1282 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
1283 * cfgloop.h (struct loop): Add unroll field.
1284 * function.h (struct function): Add has_unroll bitfield.
1285 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
1286 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
1287 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
1288 (pass_rtl_unroll_loops::gate): Likewise.
1289 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
1290 for which loop->unroll==1.
1291 (decide_unroll_constant_iterations): Use note for consistency and
1292 take loop->unroll into account. Return early if loop->unroll is set.
1293 Fix thinko in existing test.
1294 (decide_unroll_runtime_iterations): Use note for consistency and
1295 take loop->unroll into account.
1296 (decide_unroll_stupid): Likewise.
1297 * lto-streamer-in.c (input_cfg): Read loop->unroll.
1298 * lto-streamer-out.c (output_cfg): Write loop->unroll.
1299 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
1301 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
1302 (print_loop): Print loop->unroll if set.
1303 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
1304 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
1305 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
1307 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
1308 loop->unroll is set and smaller than the trip count. Otherwise bypass
1309 entirely the heuristics if loop->unroll is set. Remove dead note.
1310 Fix off-by-one bug in other note.
1311 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
1312 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
1314 (tree_unroll_loops_completely): Make static.
1315 (pass_complete_unroll::execute): Use correct type for variable.
1316 (pass_complete_unrolli::execute): Fix formatting.
1317 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
1319 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
1321 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
1322 enable when TARGET_PREFER_AVX128 is set.
1324 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1326 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
1328 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
1329 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
1331 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1333 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
1334 frequencies. Fix estimation of aggregate parameters.
1336 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1338 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
1341 2017-11-23 Tom de Vries <tom@codesourcery.com>
1343 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
1345 2017-11-23 Tom de Vries <tom@codesourcery.com>
1347 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
1350 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
1353 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
1354 sibcall_value_pcrel_fdpic): Use local variable instead of
1356 (calli_tbr_rel): Add missing operand 2.
1357 (call_valuei_tbr_rel): Add missing operand 3.
1359 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1362 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
1363 bitpos/bitsize covering the whole destination, use store_expr only if
1364 the complex mode is the same. Otherwise, use expand_normal and if
1365 it returns CONCAT, subreg each part separately instead of trying to
1366 subreg the whole result.
1368 2017-11-23 Richard Biener <rguenther@suse.de>
1370 PR tree-optimization/23094
1371 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
1372 come here from walking over backedges in the first iteration.
1373 (vn_reference_lookup_3): Skip clobbers that store the same value.
1375 2017-11-23 Richard Biener <rguenther@suse.de>
1377 PR tree-optimization/81403
1378 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
1379 a block we need a leader relative to.
1380 (phi_translate_1): For nary processing require a leader from
1381 get_representative_for given we run expression simplification
1382 using match-and-simplify. Remove previous fix.
1384 2017-11-22 Jeff Law <law@redhat.com>
1386 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
1387 Use new method allocate_value_range rather than accessing the
1388 vrp_value_range_pool data member directly.
1389 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
1390 to use extract_range_from_stmt method to avoid need for
1391 extract_range_from_assignment method.
1392 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
1393 method rather than setting values_propgated data member directly.
1394 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
1395 and values propagated data members and extract_range_from_assignment
1396 method. Reorder private data members to conform to standards.
1397 Add new methods set_lattice_propagation_complete and
1398 allocate_value_range.
1400 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
1402 PR rtl-optimization/83030
1403 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
1404 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
1405 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
1406 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
1407 CROSSING_JUMP_P flag is already set before setting it.
1408 * cfgrtl.c (fixup_partition_crossing): Likewise.
1409 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
1412 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1414 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
1415 Handle the case where both arguments are using gen_const_vec_series.
1417 2017-11-22 David Malcolm <dmalcolm@redhat.com>
1420 * pretty-print.c (pp_format): Move quoting implementation to
1421 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
1422 to pass address of "quote" local.
1423 (pp_begin_quote): New function.
1424 (pp_end_quote): New function.
1425 * pretty-print.h (printer_fn): Convert penultimate param from bool
1427 (pp_begin_quote): New decl.
1428 (pp_end_quote): New decl.
1429 * tree-diagnostic.c (default_tree_printer): Convert penultimate
1430 param from bool to bool *.
1431 * tree-diagnostic.h (default_tree_printer): Likewise.
1433 2017-11-22 Jeff Law <law@redhat.com>
1435 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
1436 Initialize vr_values.
1437 (evrp_range_analyzer::try_find_new_range): Call methods attached to
1438 vr_values via vr_values class instance rather than delegators.
1439 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
1440 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1441 (evrp_range_analyzer::push_value_range): Likewise.
1442 (evrp_range_analyzer::pop_value_range): Likewise.
1443 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
1444 most delegators. Those remaining are exposed as public interfaces.
1445 Make vr_values a pointer and private.
1446 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
1448 (evrp_range_analyzer::get_vr_value): New method.
1449 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
1450 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
1451 (class evrp_dom_walker): Attach evrp_folder class, initialize
1452 it in the ctor. Remove temporary delegators.
1453 (evrp_dom_walker::before_dom_children): Call methods in attached
1454 evrp_range_analyzer class via class instance pointer. Use
1455 free value_range_constant_singleton to remove need for
1456 op_with_constant_singleton_value delegator method. Do not
1457 create a vrp_prop class instance for every call! Narrow
1458 scope of a couple variables.
1459 (evrp_dom_walker::cleanup): Call methods in attached
1460 evrp_range_analyzer class via class instance pointer.
1461 * vr-values.h (class vr_values): Privatize many methods and
1464 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1466 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
1467 * cfgexpand.c (expand_debug_expr): Remove handling for them.
1468 * expr.c (expand_expr_real_2): Likewise.
1469 * fold-const.c (const_unop): Likewise.
1470 * optabs-tree.c (optab_for_tree_code): Likewise.
1471 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1472 * tree-inline.c (estimate_operator_cost): Likewise.
1473 * tree-pretty-print.c (dump_generic_node): Likewise.
1474 (op_code_prio): Likewise.
1475 (op_symbol_code): Likewise.
1476 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
1477 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
1478 * internal-fn.c (direct_internal_fn_optab): New function.
1479 (direct_internal_fn_array, direct_internal_fn_supported_p
1480 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
1481 * fold-const-call.c (fold_const_reduction): New function.
1482 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
1484 * tree-vect-loop.c: Include internal-fn.h.
1485 (reduction_code_for_scalar_code): Rename to...
1486 (reduction_fn_for_scalar_code): ...this and return an internal
1488 (vect_model_reduction_cost): Take an internal_fn rather than
1490 (vect_create_epilog_for_reduction): Likewise. Build calls rather
1492 (vectorizable_reduction): Use internal functions rather than tree
1493 codes for the reduction operation. Update calls to the functions
1495 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
1496 Use calls to internal functions rather than REDUC tree codes.
1497 * config/aarch64/aarch64-simd.md: Update comment accordingly.
1499 2017-11-22 Olivier Hainque <hainque@adacore.com>
1501 * config/vxworks.c (vxworks_override_options): Pick default
1502 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
1503 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
1504 DWARF_GNAT_ENCODINGS_DEFAULT.
1505 * config/vxworksae.h: Likewise.
1507 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
1509 PR tree-optimization/83104
1510 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
1513 2017-11-22 Marek Polacek <polacek@redhat.com>
1514 H.J. Lu <hongjiu.lu@intel.com>
1515 Jason Merrill <jason@redhat.com>
1520 * calls.c (initialize_argument_information): Call
1521 warn_parameter_passing_abi target hook.
1522 (store_one_arg): Use 0 for empty record size. Don't push 0 size
1523 argument onto stack.
1524 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
1525 * common.opt: Update -fabi-version description.
1526 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
1527 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
1529 (ix86_is_empty_record): New function.
1530 (ix86_warn_parameter_passing_abi): New function.
1531 (TARGET_EMPTY_RECORD_P): Redefine.
1532 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
1533 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
1534 * doc/tm.texi: Regenerated.
1535 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
1536 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
1537 * dwarf2out.c (get_ultimate_context): Move to tree.c.
1538 * explow.c (hard_function_value): Call arg_int_size_in_bytes
1539 instead of int_size_in_bytes.
1540 * expr.c (copy_blkmode_to_reg): Likewise.
1541 * function.c (aggregate_value_p): Return 0 for empty types.
1542 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi target hook.
1543 (locate_and_pad_parm): Call arg size_in_bytes instead
1545 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
1546 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
1547 * target.def (empty_record_p, warn_parameter_passing_abi): New target
1549 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
1550 (std_gimplify_va_arg_expr): Skip empty records. Call
1551 arg_size_in_bytes instead size_in_bytes.
1552 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
1553 * tree-core.h (tree_type_common): Add empty_flag.
1554 (tree_decl_common): Update comments.
1555 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
1557 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1558 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
1560 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1561 * tree.c (default_is_empty_type): New function.
1562 (default_is_empty_record): New function.
1563 (arg_int_size_in_bytes): New function.
1564 (arg_size_in_bytes): New function.
1565 (get_ultimate_context): New function.
1566 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
1567 TRANSLATION_UNIT_WARN_EMPTY_P.
1568 (default_is_empty_record, arg_int_size_in_bytes,
1569 arg_size_in_bytes, get_ultimate_context): Declare.
1571 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1573 * config/arm/arm.c (cmse_clear_registers): New function.
1574 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
1575 code by call to cmse_clear_registers.
1576 (cmse_nonsecure_entry_clear_before_return): Likewise.
1578 2017-11-22 Tamar Christina <tamar.christina@arm.com>
1580 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
1581 (vdot_s32, vdotq_s32): New.
1582 (vdot_lane_u32, vdotq_lane_u32): New.
1583 (vdot_lane_s32, vdotq_lane_s32): New.
1586 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1589 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
1590 for unsigned values with fewer HWIs than the precision.
1591 (test_overflow): New function.
1592 (wide_int_cc_tests): Call it.
1594 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1596 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
1597 has the same mode class as Pmode.
1599 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1601 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
1602 padding_bits_to_clear_ptr.
1603 (cmse_nonsecure_entry_clear_before_return): Likewise.
1605 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1607 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
1608 auto_sbitap instead of integer bitfield to control register needing
1611 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1613 PR tree-optimization/83044
1614 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
1615 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
1616 that need the upper bound. Subtract offset from
1617 get_addr_base_and_unit_offset only if positive and subtract it
1618 before division by eltsize rather than after it.
1621 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
1622 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
1623 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
1627 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
1628 expand_binop with *mul_widen_optab, make sure at least one of the
1629 operands doesn't have VOIDmode.
1632 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
1634 PR rtl-optimization/82044
1635 PR tree-optimization/82042
1636 * dse.c (record_store): Check for overflow.
1637 (check_mem_read_rtx): Properly check for overflow if width == -1, call
1638 add_wild_read instead of clear_rhs_from_active_local_stores on
1639 overflow and log it into dump_file.
1641 2017-11-22 Richard Biener <rguenther@suse.de>
1643 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
1644 fake edges to exit when looking for a place to insert.
1645 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
1647 (insert_into_preds_of_block): Commit edge insertion immediately,
1648 assert that doesn't require new BBs.
1649 (fini_pre): Release expressions.
1650 (pass_pre::execute): Shuffle things around a bit, if the fn
1651 is too large do not compute AVAIL either as this is really the
1654 2017-11-22 Richard Biener <rguenther@suse.de>
1656 PR tree-optimization/83089
1657 * tree-if-conv.c (pass_if_conversion::execute): If anything
1658 changed reset SCEV and free the number of iteration estimates.
1660 2017-11-21 Martin Sebor <msebor@redhat.com>
1662 PR tree-optimization/82945
1663 * calls.h (warn_nonstring_bound): Remove unused function.
1665 2017-11-21 Martin Sebor <msebor@redhat.com>
1667 PR tree-optimization/82945
1668 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
1669 * calls.h (maybe_warn_nonstring_arg): Declare new function.
1670 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
1672 (initialize_argument_information): Call maybe_warn_nonstring_arg.
1673 * calls.h (get_attr_nonstring_decl): Declare new function.
1674 * doc/extend.texi (attribute nonstring): Update.
1675 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
1676 get_attr_nonstring_decl and handle it.
1677 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
1678 detection of nul-termination.
1679 (strlen_to_stridx): Change to a pointer.
1680 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
1681 (pass_strlen::execute): Same.
1683 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
1685 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
1686 for the new option -mprefer-vector-width=[none|128|256|512].
1687 * config/i386/i386.c (ix86_target_string): remove old style options
1688 -mprefer-avx256 and make -mprefer-avx128 as alias.
1689 (ix86_option_override_internal): Apply defaults for the
1690 -mprefer-vector-width=[128|256] option.
1691 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
1692 Implement macros to work with -mprefer-vector-width=.
1693 * config/i386/i386.opt: Implemented option
1694 -mprefer-vector-width=[none|128|256|512].
1695 * doc/invoke.texi: Documentation for
1696 -mprefer-vector-width=[none|128|256|512].
1698 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
1700 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
1702 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
1704 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
1706 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
1708 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
1710 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
1711 * expr.c (expand_expr_real_2): Likewise.
1712 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
1713 fold_binary_loc): Likewise.
1714 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
1715 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
1716 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
1717 MINUS_EXPR transformations.
1718 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
1719 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
1720 * tree-inline.c (estimate_operator_cost): Likewise.
1721 * tree-pretty-print.c (dump_generic_node, op_code_prio,
1722 op_symbol_code): Likewise.
1723 * tree-vect-stmts.c (vectorizable_operation): Likewise.
1724 * vr-values.c (extract_range_from_binary_expr): Likewise.
1725 * varasm.c (initializer_constant_valid_p_1): Likewise.
1726 * tree.def: New tree code POINTER_DIFF_EXPR.
1728 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
1730 * config/i386/i386.md (*bswap<mode>2_movbe): Add
1731 integer suffix to movbe mnemonic.
1732 (*bswaphi2_movbe): Ditto.
1733 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
1735 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1738 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
1739 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
1740 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
1742 2017-11-21 Martin Liska <mliska@suse.cz>
1744 * tree-inline.c (expand_call_inline): Remove not needed
1747 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
1748 Jakub Jelinek <jakub@redhat.com>
1751 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
1752 Remove static keyword from f variable.
1754 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1756 PR tree-optimization/83086
1757 * gimple-ssa-store-merging.c
1758 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
1759 rather than n.base_addr.
1761 2017-11-21 Martin Liska <mliska@suse.cz>
1763 PR rtl-optimization/82044
1764 PR tree-optimization/82042
1765 * dse.c (check_mem_read_rtx): Check for overflow.
1767 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1769 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
1772 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
1774 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
1775 and its corresponding call arg location note.
1777 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
1778 Andrew Burgess <andrew.burgess@embecosm.com>
1780 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
1782 (arc_return_slot_offset): Likewise.
1783 (arc_eh_return_address_location): New declaration.
1784 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
1785 (MUST_SAVE_REGISTER): Add exception handler case.
1786 (MUST_SAVE_RETURN_ADDR): Likewise.
1787 (arc_frame_pointer_required): Likewise.
1788 (arc_frame_pointer_needed): New function.
1789 (arc_compute_frame_size): Changed.
1790 (arc_expand_prologue): Likewise.
1791 (arc_expand_epilogue): Likewise.
1792 (arc_initial_elimination_offset): Likewise.
1793 (arc_return_slot_offset): Delete.
1794 (arc_eh_return_address_location): New function.
1795 (arc_builtin_setjmp_frame_value): Likewise.
1796 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
1797 (EH_RETURN_STACKADJ_RTX): Define.
1798 (EH_RETURN_HANDLER_RTX): Likewise.
1799 * config/arc/arc.md (eh_return): Delete.
1801 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1803 * print-tree.h (debug_vec_tree): Remove prototype.
1804 * gdbinit.in (pvt): Remove macro.
1806 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1808 PR tree-optimization/83047
1809 * gimple-ssa-store-merging.c
1810 (imm_store_chain_info::output_merged_store): If the loads with the
1811 same vuse are in different basic blocks, for load_gsi pick a load
1812 location that is dominated by the other loads.
1815 * config/i386/i386.c (ix86_memmodel_check): Start
1816 -Winvalid-memory-model diagnostics with lowercase letter.
1819 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
1820 set in_cold_section_p to the partition containing loc_list->first.
1821 When seeing loc_list->last_before_switch node, update secname and
1822 perform range_across_switch second partition handling only after that.
1825 * run-rtl-passes.c: Include debug.h.
1826 (run_rtl_passes): Call debug_hooks->assembly_start.
1827 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
1831 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
1832 OPTAB_DIRECT in calls to expand_simple_binop.
1834 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1837 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
1839 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1842 * spellcheck.h (best_match::blithely_get_best_candidate): New
1845 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1847 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
1848 with lower case letter.
1850 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
1852 * config/i386/i386.md (bswaphi2): New expander.
1853 (*bswaphi2_movbe): New insn pattern.
1854 (bswaphi -> rorhi pepehole2): New peephole pattern.
1856 2017-11-20 Jeff Law <law@redhat.com>
1858 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
1859 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
1860 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
1861 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
1862 gimple-ssa-evrp-analyze.h.
1864 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
1865 set BB_VISITED here.
1866 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
1868 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1870 * tree-predcom.c: Add general comment on Store-Store chains.
1871 (split_data_refs_to_components): Postpone clearing eliminate_store_p
1873 (get_chain_last_ref_at): Rename into...
1874 (get_chain_last_write_at): ...this.
1875 (get_chain_last_write_before_load): New function.
1876 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
1877 CT_STORE_STORE when write reference is added.
1878 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
1879 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
1880 (initialize_root_vars_store_elim_1): Ditto.
1881 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
1882 definition is created.
1883 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
1884 chain by replacing it with dominant stored value.
1886 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1888 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
1890 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
1892 * vr-values.c (extract_range_from_binary_expr): Use a full range
1895 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
1897 * config/arm/arm.md (R4_REGNUM): Define constant.
1898 (nonsecure_call_internal): Remove r4 clobber.
1899 (nonsecure_call_value_internal): Likewise.
1900 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
1901 clobber and resequence match_operands.
1902 (nonsecure_call_value_reg_thumb1_v5): Likewise.
1903 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
1904 (nonsecure_call_value_reg_thumb2): Likewise.
1906 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1908 PR tree-optimization/78821
1909 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
1910 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
1911 to the address of the base rather than the base itself.
1912 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
1913 (find_bswap_or_nop_finalize): New function.
1914 (find_bswap_or_nop): Use it.
1915 (bswap_replace): Return a tree rather than bool, change first
1916 argument from gimple * to gimple_stmt_iterator, allow inserting
1917 into an empty sequence, allow ins_stmt to be NULL - then emit
1918 all stmts into gsi. Fix up MEM_REF address gimplification.
1919 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
1920 (struct store_immediate_info): Add N and INS_STMT non-static
1922 (store_immediate_info::store_immediate_info): Initialize them
1923 from newly added ctor args.
1924 (merged_store_group::apply_stores): Formatting fixes. Sort by
1926 (stmts_may_clobber_ref_p): For stores call also
1927 refs_anti_dependent_p.
1928 (gather_bswap_load_refs): New function.
1929 (imm_store_chain_info::try_coalesce_bswap): New method.
1930 (imm_store_chain_info::coalesce_immediate_stores): Use it.
1931 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
1932 (imm_store_chain_info::output_merged_store): Fail if number of
1933 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
1934 and NOP_EXPR rhs_code.
1935 (pass_store_merging::process_store): Compute n and ins_stmt, if
1936 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
1937 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
1939 (pass_store_merging::execute): Calculate dominators.
1941 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
1942 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
1943 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
1944 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
1945 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
1946 class pass_optimize_bswap, bswap_replace,
1947 pass_optimize_bswap::execute): Moved to ...
1948 * gimple-ssa-store-merging.c: ... this file.
1949 Include optabs-tree.h.
1950 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
1951 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
1952 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
1953 anonymous namespace, remove static keywords.
1954 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
1955 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
1957 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
1960 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
1962 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
1964 * vec.h (debug_helper): New function.
1965 (DEFINE_DEBUG_VEC): New macro.
1966 * hash-set.h (debug_helper): New function.
1967 (DEFINE_DEBUG_HASH_SET): New macro.
1968 * cfg.c (debug_slim (edge)): New function.
1969 Call DEFINE_DEBUG_VEC for edges.
1970 Call DEFINE_DEBUG_HASH_SET for edges.
1971 * cfghooks.c (debug_slim (basic_block)): New function.
1972 Call DEFINE_DEBUG_VEC for basic blocks.
1973 Call DEFINE_DEBUG_HASH_SET for basic blocks.
1974 * print-tree.c (debug_slim): New function to handle trees.
1975 Call DEFINE_DEBUG_VEC for trees.
1976 Call DEFINE_DEBUG_HASH_SET for trees.
1977 (debug (vec<tree, va_gc>) &): Remove.
1978 (debug (<vec<tree, va_gc>) *): Remove.
1979 * print-rtl.c (debug_slim): New function to handle const_rtx.
1980 Call DEFINE_DEBUG_VEC for rtx_def.
1981 Call DEFINE_DEBUG_VEC for rtx_insn.
1982 Call DEFINE_DEBUG_HASH_SET for rtx_def.
1983 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
1984 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
1985 (debug (vec<rtx_insn *> *ptr): Remove.
1986 (debug_insn_vector): Remove.
1987 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
1989 2017-11-20 Tom de Vries <tom@codesourcery.com>
1991 PR rtl-optimization/82020
1992 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
1993 IF_THEN_ELSE condition.
1995 2017-11-19 Jeff Law <law@redhat.com>
1997 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1998 of degenerates resulting from ignoring an edge.
2000 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2003 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
2005 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2008 * profile-count.c (profile_count::to_sreal_scale): Fix return value
2009 for uninitialied counts.
2011 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2014 * tree-inline.c (estimate_num_insns): Set to 1 at least.
2016 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2019 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
2021 2017-11-19 Tom de Vries <tom@codesourcery.com>
2023 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
2026 2017-11-19 Tom de Vries <tom@codesourcery.com>
2028 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
2031 2017-11-19 Tom de Vries <tom@codesourcery.com>
2033 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
2036 2017-11-19 Tom de Vries <tom@codesourcery.com>
2038 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
2040 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
2042 2017-11-19 Tom de Vries <tom@codesourcery.com>
2044 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
2046 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
2047 semicolon after MCORE_EXPORT_NAME call.
2049 2017-11-19 Tom de Vries <tom@codesourcery.com>
2052 * vmsdbgout.c (vmsdbgout_early_finish): New function.
2053 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
2055 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
2057 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
2058 (cgraph_edge::clone): Cleanup updating of profile.
2059 * ipa-cp.c (update_profiling_info): Likewise.
2060 * ipa-inline-transform.c (inline_transform): Likewise.
2061 * ipa-inline.c (inline_small_functions): Add missing space to dump.
2062 * ipa-split.c (execute_split_functions): Do not split when function
2064 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
2065 * profile-count.c (profile_count::dump): Add global0.
2066 (profile_count::to_cgraph_frequency): Do not ICE when entry is
2068 (profile_count::to_sreal_scale): Likewise.
2069 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
2070 (profile_count::combine_with_ipa_count): New function.
2071 * profile-count.h (profile_guessed_global0adjusted): New.
2072 (profile_count::adjusted_zero): New.
2073 (profile_count::global0adjusted): New.
2074 (profile_count::combine_with_ipa_count): New.
2075 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
2076 correct profile of return block of split functions.
2077 (copy_cfg_body): Remove unused profile_count.
2078 (copy_body): Likewise.
2079 (expand_call_inline): Update.
2080 (tree_function_versioning): Update.
2082 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
2084 * hash-set.h (hash_set::empty): New.
2085 * tree-ssa-threadbackward.h: Delete.
2086 * tree-ssa-threadbackward.c (class thread_jumps): New.
2087 Move max_threaded_paths into class.
2088 (fsm_find_thread_path): Remove arguments that are now in class.
2089 (profitable_jump_thread_path): Rename to...
2090 (thread_jumps::profitable_jump_thread_path): ...this.
2091 (convert_and_register_jump_thread_path): Rename to...
2092 (thread_jumps::convert_and_register_current_path): ...this.
2093 (check_subpath_and_update_thread_path): Rename to...
2094 (thread_jumps::check_subpath_and_update_thread_path): ...this.
2095 (register_jump_thread_path_if_profitable): Rename to...
2096 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
2097 (handle_phi): Rename to...
2098 (thread_jumps::handle_phi): ...this.
2099 (handle_assignment): Rename to...
2100 (thread_jumps::handle_assignment): ...this.
2101 (fsm_find_control_statement_thread_paths): Rename to...
2102 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
2103 (find_jump_threads_backwards): Rename to...
2104 (thread_jumps::find_jump_threads_backwards): ...this.
2105 Initialize path local data.
2106 (pass_thread_jumps::execute): Call find_jump_threads_backwards
2107 from within thread_jumps class.
2108 (pass_early_thread_jumps::execute): Same.
2110 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2112 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
2114 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2116 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
2117 to canonical location.
2119 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2122 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
2124 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
2126 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
2128 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
2129 rcrt1.o%s/grcrt1.o%s for -static-pie.
2131 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2133 * i386.c (ix86_multiplication_cost, ix86_division_cost,
2134 ix86_shift_rotate_cost): Break out from ...
2135 (ix86_rtx_costs): ... here.
2136 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
2139 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2141 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
2142 when entry block was promoted unlikely.
2143 (estimate_bb_frequencies): Increase frequency scale.
2144 * profile-count.h (profile_count): Export precision info.
2146 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2148 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
2149 disturbin profile of entry block.
2151 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2153 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
2156 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2158 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
2159 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
2161 2017-11-17 Jeff Law <law@redhat.com>
2163 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
2164 from evrp_dom_walker class. Various methods moved into new class.
2165 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
2166 (evrp_range_analyzer::enter): New method.
2167 (evrp_range_analyzer::leave): New method.
2168 (evrp_dom_walker): Remove delegators no longer needed by this class.
2169 Replace vr_values data member with evrp_range_analyzer
2171 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
2172 method extracted from evrp_dom_walker::before_dom_children.
2173 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
2174 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
2176 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
2177 Add private copy constructor and move assignment operators.
2178 Privatize methods and class data where trivially possible.
2179 (evrp_dom_walker::cleanup): New function, extracted from
2180 execute_early_vrp. Simplify access to class data.
2182 * vr-values.h (get_output_for_vrp): Prototype.
2183 * vr-values.c (get_output_for_vrp): New function extracted from
2184 vrp_visit_assignment_or_call and extract_range_from_stmt.
2185 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
2187 2017-11-17 Luis Machado <luis.machado@linaro.org>
2189 * config/aarch64/aarch64.c
2190 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
2191 (qdf24xx_tunings) <autoprefetcher_model>: Set to
2192 tune_params::AUTOPREFETCHER_WEAK.
2194 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2197 * config/arm/arm.c (arm_valid_target_attribute_rec):
2198 Parse "arch=" and "+<ext>".
2199 (arm_valid_target_attribute_tree): Re-init global options.
2200 (arm_option_override): Make non-static.
2201 (arm_options_perform_arch_sanity_checks): Make errors fatal.
2202 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
2203 (__ARM_FEATURE_CRC32): Support undef.
2204 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
2205 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
2207 2017-11-17 David Malcolm <dmalcolm@redhat.com>
2209 * gdbinit.in (break-on-diagnostic): New command.
2211 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2213 * config/i386/i386.c (ix86_expand_epilogue): Change simple
2214 return to indirect jump for EH return if control-flow
2215 protection is enabled. Change explicit 'false' argument in
2216 pro_epilogue_adjust_stack with a value of
2218 * config/i386/i386.md (simple_return_indirect_internal):
2219 Remove SImode restriction to support 64-bit.
2221 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2223 * combine.c (added_notes_insn): New.
2224 (try_combine): Handle added_notes_insn like added_links_insn.
2225 Rewrite return value code.
2226 (distribute_notes): Set added_notes_insn to the earliest insn we added
2229 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2231 PR rtl-optimization/82621
2232 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
2233 dest of one of those SETs is unused.
2235 2017-11-17 Richard Biener <rguenther@suse.de>
2238 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
2239 * tree-pretty-print.c (dump_generic_node): Handle
2240 annot_expr_parallel_kind.
2241 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
2242 * gimplify.c (gimple_boolify): Likewise.
2244 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2246 * config.gcc (extra_headers): Add cet.h for x86 targets.
2247 * config/i386/cet.h: New file.
2248 * doc/install.texi: Add --enable-cet/--disable-cet.
2250 2017-11-17 Richard Biener <rguenther@suse.de>
2252 PR tree-optimization/83017
2253 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
2254 (gen_parallel_loop): Properly count iterations.
2255 (parallelize_loops): Handle loop->can_be_parallel independent
2256 of flag_loop_parallelize_all. Make static profitability test match
2258 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
2259 * invoke.texi (parloops-min-per-thread): Document.
2261 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
2263 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
2264 upstreaming review comments.
2266 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2268 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
2269 with fast unaligned access.
2270 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
2272 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
2274 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
2275 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
2277 2017-11-17 Richard Biener <rguenther@suse.de>
2279 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
2280 folding of references.
2282 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
2285 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
2286 of replacing call to strncmp with corresponding call to strcmp when
2289 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
2291 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
2292 option prefer-avx256 for skylake-avx512 configuration.
2293 * config/i386/i386.c (ix86_option_override_internal): Ditto.
2294 (get_builtin_code_for_version): Ditto.
2296 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
2297 Monk Chiang <sh.chiang04@gmail.com>
2299 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
2300 (FIXED_REGISTERS): Reserve more register numbers.
2301 (CALL_USED_REGISTERS): Likewise.
2302 (REG_ALLOC_ORDER): Likewise.
2303 (REG_CLASS_CONTENTS): Likewise.
2304 (REGISTER_NAMES): Likewise.
2306 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
2307 Kito Cheng <kito.cheng@gmail.com>
2309 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
2311 * config/nds32/iterators.md: Add vector mode iterators and attributes.
2313 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
2315 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
2316 parameter list for vec_splats.
2318 2017-11-16 Joseph Myers <joseph@codesourcery.com>
2320 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
2322 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
2324 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
2327 * doc/install.texi: Document incompatibility of Perl >=5.6.26
2328 with the required version of automake 1.11.6.
2330 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
2332 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
2334 (DU_C2_3_power9): Correct reservation combinations.
2335 (FP_DIV_power9, VEC_DIV_power9): New.
2336 (power9-alu): Split out rotate/shift...
2337 (power9-rot): ...to here, correct dispatch resource.
2338 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
2340 (power9-fp): Correct latency.
2341 (power9-sdiv): Add div/sqrt resource.
2342 (power9-ddiv): Correct latency, add div/sqrt resource.
2343 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
2344 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
2346 (power9-qpdiv, power9-qpmul): Adjust resource usage.
2348 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2350 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
2351 switch statement mapping KF built-ins to TF built-ins if we don't
2352 have the proper ISA 3.0 assembler support.
2354 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2356 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
2357 (lower_emutls_function_body): Do not compute it.
2359 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2361 * ipa-split.c (split_bb_info): Turn time to sreal.
2362 (split_point): Likewise.
2363 (dump_split_point): Likewise.
2364 (fine_split_points): Likewise.
2365 (execute_split_functions): Only zero split_bbs; turn time to sreals.
2367 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2369 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
2372 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2374 * predict.c (combine_predictions_for_bb): Preserve zero predicted
2376 (expensive_function_p): Remove useless assert.
2377 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
2379 2017-11-16 Martin Sebor <msebor@redhat.com>
2381 PR tree-optimization/82588
2382 PR tree-optimization/82583
2383 * tree-vrp.c (check_array_ref): Handle flexible array members,
2384 string literals, and inner indices.
2385 (search_for_addr_array): Add detail to diagnostics.
2387 2017-11-16 Nathan Sidwell <nathan@acm.org>
2391 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
2392 (SET_DECL_ASSEMBLER_NAME): Forward to
2393 overwrite_decl_assembler_name.
2394 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
2395 (overwrite_decl_assembler_name): Declare.
2396 * tree.c (overwrite_decl_assembler_name): New.
2397 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
2398 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
2399 (LANG_HOOKS_INITIALIZER): Add it.
2400 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
2401 * langhooks.c (lhd_set_decl_assembler_name): Use
2402 SET_DECL_ASSEMBLER_NAME.
2403 (lhd_overwrite_decl_assembler_name): Default implementation.
2405 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2406 Jackson Woodruff <jackson.woodruff@arm.com>
2408 PR tree-optimization/71026
2409 * match.pd: Canonicalize constant multiplies in division.
2411 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2413 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
2414 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
2415 Remove OPT_fomit_frame_pointer entry.
2416 * common/config/arc/arc-common.c: Likewise.
2417 * common/config/arm/arm-common.c: Likewise.
2418 * common/config/avr/avr-common.c: Likewise.
2419 * common/config/c6x/c6x-common.c: Likewise.
2420 * common/config/cr16/cr16-common.c: Likewise.
2421 * common/config/cris/cris-common.c: Likewise.
2422 * common/config/epiphany/epiphany-common.c: Likewise.
2423 * common/config/fr30/fr30-common.c: Likewise.
2424 * common/config/frv/frv-common.c: Likewise.
2425 * common/config/ia64/ia64-common.c: Likewise.
2426 * common/config/iq2000/iq2000-common.c: Likewise.
2427 * common/config/lm32/lm32-common.c: Likewise.
2428 * common/config/m32r/m32r-common.c: Likewise.
2429 * common/config/mcore/mcore-common.c: Likewise.
2430 * common/config/microblaze/microblaze-common.c: Likewise.
2431 * common/config/mips/mips-common.c: Likewise.
2432 * common/config/mmix/mmix-common.c: Likewise.
2433 * common/config/mn10300/mn10300-common.c: Likewise.
2434 * common/config/nios2/nios2-common.c: Likewise.
2435 * common/config/pa/pa-common.c: Likewise.
2436 * common/config/pdp11/pdp11-common.c: Likewise.
2437 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
2438 * common/config/riscv/riscv-common.c: Likewise.
2439 * common/config/rs6000/rs6000-common.c: Likewise.
2440 * common/config/rx/rx-common.c: Likewise.
2441 * common/config/s390/s390-common.c: Likewise.
2442 * common/config/sh/sh-common.c: Likewise.
2443 * common/config/sparc/sparc-common.c: Likewise.
2444 * common/config/tilegx/tilegx-common.c: Likewise.
2445 * common/config/tilepro/tilepro-common.c: Likewise.
2446 * common/config/v850/v850-common.c: Likewise.
2447 * common/config/visium/visium-common.c: Likewise.
2448 * common/config/xstormy16/xstormy16-common.c: Likewise.
2449 * common/config/xtensa/xtensa-common.c: Likewise.
2450 * invoke.texi (-fomit-frame-pointer): Update documentation.
2452 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2454 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
2456 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2458 * cfg.c (scale_bbs_frequencies_int,
2459 cale_bbs_frequencies_gcov_type): Remove.
2460 * cfg.h (scale_bbs_frequencies_int,
2461 cale_bbs_frequencies_gcov_type): Remove.
2463 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2465 * tree-ssa-loop-manip.c
2466 (scale_dominated_blocks_in_loop): Update to profile counts.
2467 (tree_transform_and_unroll_loop): Likewise.
2469 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2471 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
2472 scale_bbs_frequencies_int.
2474 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2476 * final.c (compute_alignments): Use counts rather than frequencies.
2478 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2480 * cfgloopanal.c: Include sreal.h
2481 (average_num_loop_insns): Use counts and sreal for accounting.
2483 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2485 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
2488 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2489 Kito Cheng <kito.cheng@gmail.com>
2491 * config/nds32/constraints.md: Provide more constraints.
2492 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
2493 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
2494 support constraints usage.
2496 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2498 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
2500 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2501 Kito Cheng <kito.cheng@gmail.com>
2503 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
2504 * config/nds32/nds32.opt: Refine the layout.
2505 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
2506 TARGET_EXT_STRING): Support new options.
2507 * config/nds32/nds32.h: Likewise.
2508 * config/nds32/nds32.md: Likewise.
2509 * config/nds32/nds32-predicates.c: Likewise.
2510 * config/nds32/constraints.md: Likewise.
2511 * common/config/nds32/nds32-common.c: Likewise.
2513 2017-11-16 Julia Koval <julia.koval@intel.com>
2516 * config/i386/gfniintrin.h: Add sse check.
2517 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
2519 2017-11-16 Julia Koval <julia.koval@intel.com>
2521 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
2522 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
2523 (ix86_handle_option): Handle -mavx512vbmi2.
2524 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
2525 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
2526 * config/i386/i386-c.c (__AVX512VBMI2__): New.
2527 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
2528 (ix86_valid_target_attribute_inner_p): Ditto.
2529 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
2530 * config/i386/i386.opt (mavx512vbmi2): New option.
2531 * doc/invoke.texi: Add new option.
2533 2017-11-16 Julia Koval <julia.koval@intel.com>
2535 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
2536 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
2537 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
2538 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
2539 _mm512_gf2p8mul_epi8): New intrinsics.
2540 * config/i386/i386-builtin-types.def
2541 (V64QI_FTYPE_V64QI_V64QI): New type.
2542 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
2543 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
2544 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
2545 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
2546 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
2547 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2549 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
2551 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
2552 explicitly as a stream of bytes.
2554 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2556 * config/rs6000/altivec.h (vec_xst_be): New #define.
2557 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
2558 and externalize from *altivec_vperm_<mode>_internal.
2559 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
2561 (XL_BE_V8HI): Likewise.
2562 (XL_BE_V4SI): Likewise.
2563 (XL_BE_V4SI): Likewise.
2564 (XL_BE_V2DI): Likewise.
2565 (XL_BE_V4SF): Likewise.
2566 (XL_BE_V2DF): Likewise.
2567 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
2568 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
2569 all array entries with these keys: VSX_BUILTIN_VEC_XL,
2570 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
2571 VSX_BUILTIN_VEC_XST_BE.
2572 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
2573 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
2575 (altivec_init_builtins): Replace conditional calls to def_builtin
2576 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
2577 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
2578 with unconditional calls. Remove calls to def_builtin for
2579 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
2580 __builtin_vec_xst_be.
2581 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
2582 to define_expand, and add alternate RTL generation for P8.
2583 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
2584 vsx_ld_elemrev_v8hi.
2585 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
2586 add alternate RTL generation for P8.
2587 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
2588 vsx_ld_elemrev_v16qi.
2589 (vsx_st_elemrev_v8hi): Convert define_insn
2590 to define_expand, and add alternate RTL generation for P8.
2591 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
2592 vsx_st_elemrev_v8hi.
2593 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
2594 add alternate RTL generation for P8.
2595 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
2596 vsx_st_elemrev_v16qi.
2598 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
2601 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
2602 TARGET_AVX512ER check.
2603 (ix86_option_override_internal): Set MASK_VZEROUPPER if
2604 neither -mzeroupper nor -mno-zeroupper is used and
2605 TARGET_EMIT_VZEROUPPER is set.
2606 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
2607 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
2609 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
2611 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2612 folding of vector compares.
2613 (fold_build_vec_cmp): New helper function.
2614 (fold_compare_helper): New helper function.
2615 (builtin_function_type): Add compare builtins to the list of functions
2616 having unsigned arguments. Cosmetic updates to comment indentation.
2617 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
2618 the not+eq combination.
2620 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2622 PR tree-optimization/82726
2623 PR tree-optimization/70754
2624 * tree-predcom.c (order_drefs_by_pos): New function.
2625 (combine_chains): Move code setting has_max_use_after to...
2626 (try_combine_chains): ...here. New parameter. Sort combined chains
2627 according to position information.
2628 (tree_predictive_commoning_loop): Update call to above function.
2629 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
2631 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2633 PR tree-optimization/82726
2635 2017-01-23 Bin Cheng <bin.cheng@arm.com>
2637 PR tree-optimization/70754
2638 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
2639 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
2640 combined stmt before it if not NULL.
2641 (combine_chains): Process refs reversely and compute dominance point
2645 2017-02-23 Bin Cheng <bin.cheng@arm.com>
2647 PR tree-optimization/79663
2648 * tree-predcom.c (combine_chains): Process refs in reverse order
2649 only for ZERO length chains, and add explaining comment.
2651 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2653 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
2654 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
2655 * doc/invoke.texi (ARM Options): Add armv8.3-a.
2657 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2659 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
2661 2017-11-15 Martin Liska <mliska@suse.cz>
2663 * tree-cfg.c (pass_warn_function_return::execute):
2664 Compare warn_return_type for greater than zero.
2666 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
2670 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
2671 to return true on Xeon and not on Xeon Phi.
2672 (ix86_check_avx256_register): Changed to ...
2673 (ix86_check_avx_upper_register): ... this. Add extra check for
2674 VALID_AVX512F_REG_OR_XI_MODE.
2675 (ix86_avx_u128_mode_needed): Changed
2676 ix86_check_avx256_register to ix86_check_avx_upper_register.
2677 (ix86_check_avx256_stores): Changed to ...
2678 (ix86_check_avx_upper_stores): ... this. Changed
2679 ix86_check_avx256_register to ix86_check_avx_upper_register.
2680 (ix86_avx_u128_mode_after): Changed
2681 avx_reg256_found to avx_upper_reg_found. Changed
2682 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
2683 (ix86_avx_u128_mode_entry): Changed
2684 ix86_check_avx256_register to ix86_check_avx_upper_register.
2685 (ix86_avx_u128_mode_exit): Ditto.
2686 * config/i386/i386.h: (host_detect_local_cpu): New define.
2688 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2690 * config/arm/xgene1.md (xgene1): Split into automatons
2691 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
2692 (xgene1_f_load): Adjust reservations and/or types.
2693 (xgene1_f_store): Likewise.
2694 (xgene1_load_pair): Likewise.
2695 (xgene1_store_pair): Likewise.
2696 (xgene1_fp_load1): Likewise.
2697 (xgene1_load1): Likewise.
2698 (xgene1_store1): Likewise.
2699 (xgene1_move): Likewise.
2700 (xgene1_alu): Likewise.
2701 (xgene1_simd): Likewise.
2702 (xgene1_bfm): Likewise.
2703 (xgene1_neon_load1): Likewise.
2704 (xgene1_neon_store1): Likewise.
2705 (xgene1_neon_logic): Likewise.
2706 (xgene1_neon_st1): Likewise.
2707 (xgene1_neon_ld1r): Likewise.
2708 (xgene1_alu_cond): Added.
2709 (xgene1_shift_reg): Likwise.
2710 (xgene1_bfx): Likewise.
2711 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
2713 2017-11-15 Jakub Jelinek <jakub@redhat.com>
2716 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
2718 (can_widen_mult_without_libcall): New function.
2719 (expand_mul_overflow): If only checking unsigned mul overflow,
2720 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
2721 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
2722 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
2723 (expand_DIVMOD): Formatting fix.
2724 * expmed.h (expand_mult): Add NO_LIBCALL argument.
2725 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
2726 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
2728 PR tree-optimization/82977
2729 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
2730 constructed temporary to strlen_to_stridx.put.
2732 2017-11-15 Martin Liska <mliska@suse.cz>
2734 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
2735 * configure: Regenerate.
2737 2017-11-15 Martin Liska <mliska@suse.cz>
2740 * config/sh/sh-mem.cc: Use proper probability for
2743 2017-11-14 Jeff Law <law@redhat.com>
2745 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
2746 the red zone for stack_clash_protection_final_dynamic_probe targets
2747 when the total dynamic stack size is zero bytes.
2749 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
2750 blocks is post order.
2752 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
2754 * dumpfile.h (TDF_COMPARE_DEBUG): New.
2755 * final.c (rest_of_clean_state): Set it for the
2756 -fcompare-debug dump.
2757 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
2758 class when TDF_COMPARE_DEBUG is set.
2760 * dwarf2out.c (gen_producer_string): Discard
2763 2017-11-15 Joseph Myers <joseph@codesourcery.com>
2766 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
2767 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
2768 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
2769 (__TGMATH_REAL_2_3): Remove macros.
2770 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
2771 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
2773 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
2775 (remquo): Define using __TGMATH_REAL_3.
2777 2017-11-14 Jeff Law <law@redhat.com>
2779 * vr-values.c: New file with contents extracted from tree-vrp.c.
2780 * Makefile.in (OBJS): Add vr-values.o
2781 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
2782 (set_value_range, set_and_canonicalize_value_range): Likewise.
2783 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
2784 (value_range_constant_singleton, symbolic_range_p): Likewise.
2785 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
2786 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
2787 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
2788 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
2789 (find_case_label_range, find_case_label_index): Likewise.
2790 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
2791 (range_int_cst_singleton_p, value_inside_range): Likewise.
2792 (get_single_symbol): Likewise.
2793 (switch_update): Move structure definition here.
2794 (to_remove_edges, to_update_switch_stmts): Provide externs.
2795 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
2796 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
2797 (vrp_val_is_min, set_value_range): Likewise.
2798 (set_and_canonicalize_value_range, copy_value_range): Likewise.
2799 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
2800 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
2801 (range_is_nonnull, range_int_cst_p): Likewwise.
2802 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
2803 (get_single_symbol, operand_less_p): Likewise
2804 (compare_values_warnv, compare_values): Likewise.
2805 (value_inside_range, value_range_constant_singleton): Likewise.
2806 (zero_nonzero_bitgs_from_vr): Likewise.
2807 (extract_range_from_binary_expr_1): Likewise.
2808 (overflow_comparison_p): Likewise.
2809 (to_remove_edges, to_update_switch_stmts): Likewise.
2810 (find_case_label-index, find_case_label_range): Likewise.
2811 (switch_update, set_value_range_to_nonnegative): Remove.
2812 (set_value_range_to_truthvalue): Likewise.
2813 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
2814 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
2815 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
2816 (find_case_label_ranges, test_for_singularity): Likewise.
2817 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
2818 (x_vr_values): Move to its remaining use site.
2820 2017-11-10 Jeff Law <law@redhat.com>
2822 * vr-values.h (VR_INITIALIZER): Move #define here.
2823 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
2824 * Makefile.in (OBJS): Add tree-evrp.o
2825 * tree-vrp.h (assert_info): Move structure definition here.
2826 (set_value_range_to_varying): Prototype.
2827 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2828 (infer_value_range, register_edge_assert_for): Likewise.
2829 (stmt_interesting_for_vrp): Likewise.
2830 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
2831 (set_value_range_to_varying): No longer static.
2832 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2833 (infer_value_range, register_edge_assert_for): Likewise.
2835 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2837 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
2838 generate the XXBRD instruction.
2840 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
2841 (rs6000_builtin_type_compatible): Treat _Float128 and long double
2842 as being compatible if -mabi=ieeelongdouble.
2843 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
2844 to setup float128 built-ins with hardware support.
2845 (BU_FLOAT128_HW_2): Likewise.
2846 (BU_FLOAT128_HW_3): Likewise.
2847 (BU_FLOAT128_HW_VSX_1): Likewise.
2848 (BU_FLOAT128_HW_VSX_2): Likewise.
2849 (scalar_extract_expq): Change float128 built-in functions to
2850 accommodate having both KFmode and TFmode functions. Use the
2851 KFmode variant as the default.
2852 (scalar_extract_sigq): Likewise.
2853 (scalar_test_neg_qp): Likewise.
2854 (scalar_insert_exp_q): Likewise.
2855 (scalar_insert_exp_qp): Likewise.
2856 (scalar_test_data_class_qp): Likewise.
2857 (sqrtf128_round_to_odd): Delete processing the round to odd
2858 built-in functions as special built-in functions, and define them
2859 as float128 built-ins. Use the KFmode variant as the default.
2860 (truncf128_round_to_odd): Likewise.
2861 (addf128_round_to_odd): Likewise.
2862 (subf128_round_to_odd): Likewise.
2863 (mulf128_round_to_odd): Likewise.
2864 (divf128_round_to_odd): Likewise.
2865 (fmaf128_round_to_odd): Likewise.
2866 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
2867 support for KFmode and TFmode xststdcqp calls.
2868 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
2869 point, switch the built-in handlers for the get/set float128
2870 exponent, get float128 mantissa, float128 test built-ins, and the
2871 float128 round to odd built-in functions. Eliminate creating the
2872 float128 round to odd built-in functions as special built-ins.
2873 (rs6000_init_builtins): Eliminate special creation of the float128
2874 round to odd built-in functions.
2875 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
2876 function insns to support both TFmode and KFmode variants.
2877 (xsxsigqp_<mode>): Likewise.
2878 (xsiexpqpf_<mode>): Likewise.
2879 (xsiexpqp_<mode>): Likewise.
2880 (xststdcqp_<mode>): Likewise.
2881 (xststdcnegqp_<mode>): Likewise.
2882 (xststdcqp_<mode>): Likewise.
2884 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2886 * tree-ssa-threadupdate.c (compute_path_counts): Remove
2887 unused path_in_freq_ptr parameter.
2888 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
2890 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2892 * ipa-inline.c (edge_badness): Dump sreal frequency.
2893 (compute_inlined_call_time): Match natural implementaiton ...
2894 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
2895 forgotten division by CGRAPH_FREQ_BASE.
2897 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2899 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
2900 Solaris 11. Update comment.
2901 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
2903 * config/sol2.h (STARTFILE_SPEC): Likewise.
2904 * configure: Regenerate.
2906 2017-11-14 Carl Love <cel@us.ibm.com>
2908 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
2909 le_ and be_ prefixes to swap* variables. Remove
2910 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
2913 2017-11-14 Jason Merrill <jason@redhat.com>
2915 Support GTY((cache)) on hash_map.
2916 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
2917 (ggc_cache_remove): Override it instead of ggc_mx.
2918 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
2919 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
2920 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
2921 (simple_cache_map_traits): Override maybe_mx.
2922 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
2923 (hash_map): Friend gt_cleare_cache.
2924 (gt_cleare_cache): New.
2925 * tree.h (tree_cache_traits): New hash_map traits class.
2926 (tree_cache_map): New typedef.
2928 2017-11-14 Richard Biener <rguenther@suse.de>
2930 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
2931 paramter and handling.
2932 (cleanup_control_flow_bb): Likewise.
2933 (cleanup_control_flow_pre): New helper performing a DFS walk
2934 to call cleanup_control_flow_bb in PRE order.
2935 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
2936 via cleanup_control_flow_pre.
2938 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
2940 * config/aarch64/aarch64-simd.md
2941 (aarch64_simd_bsl<mode>_internal): Remove DImode.
2942 (*aarch64_simd_bsl<mode>_alt): Likewise.
2943 (aarch64_simd_bsldi_internal): New.
2944 (aarch64_simd_bsldi_alt): Likewise.
2946 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2948 * tracer.c (better_p): Do not compare frequencies.
2949 * reg-stack.c (better_edge): Likewise.
2950 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
2953 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2955 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
2956 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
2957 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
2958 * lto-streamer-in.c (input_function): Use update_max_bb_count.
2959 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
2960 * predict.c (maybe_hot_frequency_p): Inline to ...
2961 (maybe_hot_count_p): ... here; rewrite to counts.
2962 (counts_to_freqs): Rename to ...
2963 (update_max_bb_count): ... this one.
2964 (expensive_function_p): Use counts.
2965 (estimate_bb_frequencies): Update.
2966 (rebuild_frequencies): Update.
2967 * predict.h (counts_to_freqs): Rename to ...
2968 (update_max_bb_count): ... this one.
2969 * profile.c (compute_branch_probabilities): Add debug info
2970 * tree-inline.c (expand_call_inline): Update debug info.
2971 (optimize_inline_calls): Use update_max_bb_count..
2972 (tree_function_versioning): Use update_max_bb_count..
2973 * value-prof.c (gimple_value_profile_transformations):
2974 Do not use update_max_bb_count.
2976 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2978 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
2979 always use frequencies.
2981 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2983 * bb-reorder.c: Remove frequencies from comments.
2984 (better_edge_p): Use profile counts.
2985 (find_traces): Dump profile counts.
2986 (rotate_loop): Use profile counts.
2987 (find_traces_1_round): Likewise.
2988 (connect_better_edge_p): Use counts instead of probabilities for
2990 (copy_bb_p): Drop early check for non-0 frequency.
2991 (sanitize_hot_paths): Update comments.
2993 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2995 * ipa-split.c (struct split_point): Add count.
2996 (consider_split): Do not compute incoming frequency; compute incoming
2997 count and store it to split_point.
2998 (split_function): Set count of the call to split part correctly.
3000 2017-11-13 Carl Love <cel@us.ibm.com>
3002 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
3004 2017-11-13 Tom Tromey <tom@tromey.com>
3006 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
3008 2017-11-13 Carl Love <cel@us.ibm.com>
3010 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
3011 Add support for builtins:
3012 unsigned int vec_first_{,miss}_match_{,or_eos}index,
3013 vector {un,}signed {char,int,short},
3014 vector {un,}signed {char,int,short}) arguments.
3015 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
3016 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
3017 Add BU_P9V_AV_2 expansions for the builtins.
3018 * config/rs6000/altivec.h (vec_first_match_index,
3019 vec_first_mismatch_index, vec_first_match_or_eos_index,
3020 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
3021 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
3022 new extern declaration.
3023 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
3024 * config/rs6000/vsx.md (first_match_index_<mode>,
3025 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
3026 first_mismatch_or_eos_index_<mode>): Add define expand.
3027 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
3028 * doc/extend.texi: Update the built-in documenation file for the new
3031 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
3033 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
3034 into the min/max operations for _Float<N> and _Float<N>X types.
3036 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
3039 * dwarf2out.c (do_eh_frame): New static variable.
3040 (dwarf2out_begin_prologue): Set it.
3041 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
3043 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3045 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
3047 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
3049 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
3052 * debug.h (dwarf2out_do_eh_frame): Declare.
3053 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
3054 (dwarf2out_do_frame): Use it.
3055 (dwarf2out_do_cfi_asm): Likewise.
3056 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
3057 (dwarf2out_assembly_start): Likewise.
3058 (dwarf2out_begin_prologue): Fix comment.
3059 * toplev.c (compile_file): Always call dwarf2out_frame_finish
3060 if the target needs either debug or unwind DWARF2 info.
3061 * lto-opts.c (lto_write_options): Do not save -fexceptions,
3062 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
3063 -ftrapping-math, -ftrapv and -fwrapv.
3065 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3067 * cgraph.c (cgraph_edge::sreal_frequency): New function.
3068 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
3069 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
3070 (estimate_edge_size_and_time): Likewise.
3071 (ipa_merge_fn_summary_after_inlining): Likewise.
3072 * ipa-inline.c (cgraph_freq_base_rec): Remove.
3073 (compute_uninlined_call_time): Use sreal_frequency.
3074 (compute_inlined_call_time): Likewise.
3075 (ipa_inline): Do not initialize cgraph_freq_base_rec.
3076 * profile-count.c: Include sreal.h.
3077 (profile_count::to_sreal_scale): New.
3078 * profile-count.h: Forward declare sreal.
3079 (profile_count::to_sreal_scale): Declare.
3081 2017-11-13 Nathan Sidwell <nathan@acm.org>
3083 * diagnostic.c (maybe_line_and_column): New.
3084 (diagnostic_get_location_text): Use it.
3085 (diagnostic_report_current_module): Likewise.
3086 (test_diagnostic_get_location_text): Add tests.
3088 2017-11-13 Luis Machado <luis.machado@linaro.org>
3090 * doc/md.texi (Specifying processor pipeline description): Fix
3091 incorrect latency for the div instruction example.
3093 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3095 PR tree-optimization/78821
3096 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
3097 that bit_not_p is the same.
3098 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
3099 (split_group): Count precisely bit_not_p bits in each statement.
3100 (invert_op): New function.
3101 (imm_store_chain_info::output_merged_store): Use invert_op to
3102 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
3103 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
3105 2017-11-13 Martin Liska <mliska@suse.cz>
3107 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
3108 (struct source_info): Likewise.
3109 (add_branch_counts): Likewise.
3110 (add_line_counts): Likewise.
3111 (function_summary): Likewise.
3112 (output_intermediate_line): Likewise.
3113 (generate_results): Likewise.
3115 2017-11-13 Martin Liska <mliska@suse.cz>
3117 * gcov.c (struct block_info): Remove typedef for block_t.
3118 (struct line_info): Likewise.
3119 (line_info::has_block): Likewise.
3120 (EXIT_BLOCK): Likewise.
3121 (unblock): Likewise.
3122 (circuit): Likewise.
3123 (get_cycles_count): Likewise.
3124 (process_file): Likewise.
3125 (read_graph_file): Likewise.
3126 (solve_flow_graph): Likewise.
3127 (find_exception_blocks): Likewise.
3128 (add_line_counts): Likewise.
3129 (accumulate_line_info): Likewise.
3130 (output_line_details): Likewise.
3132 2017-11-13 Martin Liska <mliska@suse.cz>
3134 * gcov.c (struct arc_info): Remove typedef for arc_t.
3135 (struct line_info): Likewise.
3136 (add_branch_counts): Likewise.
3137 (output_branch_count): Likewise.
3138 (function_info::~function_info): Likewise.
3139 (circuit): Likewise.
3140 (output_intermediate_line): Likewise.
3141 (read_graph_file): Likewise.
3142 (solve_flow_graph): Likewise.
3143 (find_exception_blocks): Likewise.
3144 (add_line_counts): Likewise.
3145 (accumulate_line_info): Likewise.
3146 (output_line_details): Likewise.
3147 (output_function_details): Likewise.
3149 2017-11-13 Martin Liska <mliska@suse.cz>
3151 * gcov.c (struct function_info): Remove typedef for function_t.
3152 (struct source_info): Likewise.
3153 (source_info::get_functions_at_location): Likewise.
3154 (solve_flow_graph): Likewise.
3155 (find_exception_blocks): Likewise.
3156 (add_line_counts): Likewise.
3157 (output_intermediate_file): Likewise.
3158 (process_file): Likewise.
3159 (generate_results): Likewise.
3160 (release_structures): Likewise.
3161 (read_graph_file): Likewise.
3162 (read_count_file): Likewise.
3163 (accumulate_line_counts): Likewise.
3164 (output_lines): Likewise.
3166 2017-11-13 Martin Liska <mliska@suse.cz>
3168 * gcov.c (function_info::function_info): Remove num_counts
3169 and add vector<gcov_type>.
3170 (function_info::~function_info): Use the vector.
3171 (process_file): Likewise.
3172 (read_graph_file): Likewise.
3173 (read_count_file): Likewise.
3174 (solve_flow_graph): Likewise.
3176 2017-11-13 Martin Liska <mliska@suse.cz>
3178 * gcov.c (function_info::is_artificial): New function.
3179 (process_file): Erase all artificial early.
3180 (generate_results): Skip as all artificial are already
3183 2017-11-13 Martin Liska <mliska@suse.cz>
3185 * gcov.c (read_graph_file): Store to global vector of functions.
3186 (read_count_file): Iterate the vector.
3187 (process_file): Likewise.
3188 (generate_results): Likewise.
3189 (release_structures): Likewise.
3191 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3193 PR tree-optimization/82954
3194 * gimple-ssa-store-merging.c
3195 (imm_store_chain_info::coalesce_immediate_stores): If
3196 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
3198 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
3200 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
3201 Upddate call to ENDIAN_LANE_N.
3202 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
3203 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
3204 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
3205 and use aarch64_endian_lane_rtx.
3206 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
3208 2017-11-12 Tom de Vries <tom@codesourcery.com>
3210 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
3212 2017-11-12 Tom de Vries <tom@codesourcery.com>
3214 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
3216 2017-11-12 Tom de Vries <tom@codesourcery.com>
3218 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
3220 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3221 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3222 * defaults.h (ASM_OUTPUT_LABELREF): Same.
3224 2017-11-11 Martin Sebor <msebor@redhat.com>
3227 * doc/extend.texi (attribute nonstring): Remove spurious argument.
3230 * prefic.c (translate_name): Replace strncpy with memcpy to
3231 avoid -Wstringop-truncation.
3233 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3235 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
3237 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3239 * predict.c (maybe_hot_frequency_p): Do not use cfun.
3241 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3243 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
3246 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
3248 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
3249 enable generating XXBRH if the value is in a vector register.
3250 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
3251 value is in a vector register.
3252 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
3253 register to register bswap64's instead of doing the GPR sequence
3254 used on previous machines.
3255 (bswapdi2_xxbrd): New insn.
3256 (bswapdi2_reg): Disallow on ISA 3.0.
3257 (register to register bswap64 splitter): Do not split the insn on
3258 ISA 3.0 systems that use XXBRD.
3260 2017-11-10 Martin Sebor <msebor@redhat.com>
3263 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
3265 (find_subframework_file): Same.
3267 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3269 * auto-profile.c (afdo_indirect_call): Drop frequency.
3270 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
3271 (cgraph_node::create_edge): Drop frequency argument.
3272 (cgraph_node::create_indirect_edge): Drop frequency argument.
3273 (cgraph_edge::make_speculative): Drop frequency arguments.
3274 (cgraph_edge::resolve_speculation): Do not update frequencies
3275 (cgraph_edge::dump_edge_flags): Do not dump frequency.
3276 (cgraph_node::dump): Check consistency in IPA mode.
3277 (cgraph_edge::maybe_hot_p): Use IPA counter.
3278 (cgraph_edge::verify_count_and_frequency): Rename to ...
3279 (cgraph_edge::verify_count): ... this one; drop frequency checking.
3280 (cgraph_node::verify_node): Update.
3281 * cgraph.h (struct cgraph_edge): Drop frequency.
3282 (cgraph_edge::frequency): New function.
3283 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
3285 (cgraph_edge::rebuild_edges): Likewise.
3286 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
3287 (duplicate_thunk_for_node): Do not pass frequency.
3288 (cgraph_node::create_clone): Scale only counts.
3289 (cgraph_node::create_virtual_clone): Do not pass frequency.
3290 (cgraph_node::create_edge_including_clones): Do not pass frequency.
3291 (cgraph_node::create_version_clone): Do not pass frequency.
3292 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
3293 (cgraph_node::expand_thunk): Do not pass frequency.
3294 (cgraph_node::create_wrapper): Do not pass frequency.
3295 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
3297 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
3298 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
3299 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
3300 (gather_caller_stats): Use frequency function.
3301 (ipcp_cloning_candidate_p): Use frequency function.
3302 (ipcp_propagate_stage): Use frequency function.
3303 (get_info_about_necessary_edges): Use frequency function.
3304 (update_profiling_info): Update only IPA profile.
3305 (update_specialized_profile): Use frequency functoin.
3306 (perhaps_add_new_callers): Update only IPA profile.
3307 * ipa-devirt.c (ipa_devirt): Use IPA profile.
3308 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
3309 (dump_ipa_call_summary): Use frequency function.
3310 (estimate_edge_size_and_time): Use frequency function.
3311 (ipa_merge_fn_summary_after_inlining): Use frequency function.
3312 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
3313 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
3314 (update_noncloned_counts): ... ths one; scale counts only.
3315 (clone_inlined_nodes): Do not scale frequency.
3316 (inline_call): Do not pass frequency.
3317 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
3318 (compute_inlined_call_time): Use IPA profile.
3319 (want_inline_small_function_p): Use IPA profile.
3320 (want_inline_self_recursive_call_p): Use IPA profile.
3321 (edge_badness): Use IPA profile.
3322 (lookup_recursive_calls): Use IPA profile.
3323 (recursive_inlining): Do not pass frequency.
3324 (resolve_noninline_speculation): Do not update frequency.
3325 (inline_small_functions): Collect max of IPA profile.
3326 (dump_overall_stats): Dump IPA porfile.
3327 (dump_inline_stats): Dump IPA porfile.
3328 (ipa_inline): Collect IPA stats.
3329 * ipa-inline.h (clone_inlined_nodes): Update prototype.
3330 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
3331 (ipa_propagate_frequency): Use frequency function.
3332 (ipa_profile): Cleanup.
3333 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
3334 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
3335 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
3336 (input_node): Do not stream frequency.
3337 (input_edge): Do not stream frequency.
3338 (merge_profile_summaries): Scale only IPA profiles.
3339 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
3340 * predict.c (drop_profile): Do not recompute frequency.
3341 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
3342 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
3343 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
3344 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
3345 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
3346 * tree-inline.c (copy_bb): Do not scale frequency.
3347 (expand_call_inline): Do not scale frequency.
3348 (tree_function_versioning): Do not scale frequency.
3349 * ubsan.c (ubsan_create_edge): Do not pass frequency.
3351 2017-11-10 Julia Koval <julia.koval@intel.com>
3353 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
3354 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
3355 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
3356 (_mm256_maskz_gf2p8affine_epi64_epi8)
3357 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
3358 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
3359 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
3360 (__builtin_ia32_vgf2p8affineqb_v32qi)
3361 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
3362 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
3364 2017-11-10 Tamar Christina <tamar.christina@arm.com>
3368 (arm_option_override): Refactor.
3369 (arm_option_reconfigure_globals): New.
3370 (arm_options_perform_arch_sanity_checks): New.
3371 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
3373 (arm_options_perform_arch_sanity_checks): Likewise
3375 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
3377 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
3378 (power9-qpmul): New.
3379 * rs6000/rs6000.md ("type" attr): Add qmul.
3380 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
3381 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
3382 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
3384 2017-11-10 Martin Sebor <msebor@redhat.com>
3387 * builtins.c (compute_objsize): Handle arrays that
3388 compute_builtin_object_size likes to fail for. Make extern.
3389 * builtins.h (compute_objsize): Declare.
3390 (check_strncpy_sizes): New function.
3391 (expand_builtin_strncpy): Call check_strncpy_sizes.
3392 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
3393 -Wstringop-truncation.
3394 (gimple_fold_builtin_strncat): Same.
3395 * gimple.c (gimple_build_call_from_tree): Set call location.
3396 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
3397 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
3398 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
3399 (handle_builtin_strlen): Use strlen_to_stridx.
3400 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
3402 Use strlen_to_stridx.
3403 (pass_strlen::execute): Release strlen_to_stridx.
3404 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
3405 (-Wstringop-truncation): Document new option.
3407 2017-11-10 Martin Liska <mliska@suse.cz>
3409 PR gcov-profile/82702
3410 * gcov.c (main): Handle intermediate files in a different
3412 (get_gcov_intermediate_filename): New function.
3413 (output_gcov_file): Remove support of intermediate files.
3414 (generate_results): Allocate intermediate file.
3415 (release_structures): Clean-up properly fn_end.
3416 (output_intermediate_file): Start iterating with line 1.
3418 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3420 PR tree-optimization/82929
3421 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3422 ops_swapped_p non-static data member.
3423 (store_immediate_info::store_immediate_info): Clear it.
3424 (imm_store_chain_info::coalesce_immediate_stores): If swapping
3425 ops set ops_swapped_p.
3426 (count_multiple_uses): Handle ops_swapped_p.
3428 2017-11-10 Martin Liska <mliska@suse.cz>
3430 * coverage.c (coverage_init): Stream information about
3431 support of has_unexecuted_blocks.
3432 * doc/gcov.texi: Document that.
3433 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
3434 * gcov.c (read_graph_file): Likewise.
3435 (output_line_beginning): Fix a small issue with
3438 2017-11-10 Bin Cheng <bin.cheng@arm.com>
3440 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
3441 reference of trivial component.
3443 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3446 * gimple-ssa-store-merging.c
3447 (pass_store_merging::terminate_all_aliasing_chains): For
3448 gimple_store_p stmts also call refs_output_dependent_p.
3450 PR rtl-optimization/82913
3451 * compare-elim.c (try_merge_compare): Punt if def_insn is not
3454 2017-11-09 Jeff Law <law@redhat.com>
3456 * vr-values.h: New file with vr_values class.
3457 * tree-vrp.c: Include vr-values.h
3458 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
3459 data objects into the vr_values class.
3460 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
3461 (get_value_range): Make it a member function within vr_values class.
3462 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
3463 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
3464 (op_with_constant_singleton_value_range): Likewise.
3465 (extract_range_for_var_from_comparison_expr): Likewise.
3466 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
3467 (extract_range_from_binary_expr): Likewise.
3468 (extract_range_from_unary_expr): Likewise.
3469 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
3470 (check_for_binary_op_overflow, extract_range_basic): Likewise.
3471 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
3472 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
3473 (compare_name_with_value, compare_names): Likewise.
3474 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
3475 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
3476 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
3477 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
3478 (extract_range_from_phi_node): Likewise.
3479 (simplify_truth_ops_using_ranges): Likewise.
3480 (simplify_div_or_mod_using_ranges): Likewise.
3481 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
3482 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
3483 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
3484 (simplify_float_conversion_using_ranges): Likewise.
3485 (simplify_internal_call_using_ranges): Likewise.
3486 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
3487 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
3488 poitner via x_vr_values for calls into gimple folder.
3489 (vrp_initialize_lattice): Make this the vr_values ctor.
3490 (vrp_free_lattice): Make this the vr_values dtor.
3491 (set_vr_value): New function.
3492 (class vrp_prop): Add vr_values data member. Add various member
3493 functions as well as member functions that delegate to vr_values.
3494 (check_array_ref): Make a member function within vrp_prop class.
3495 (search_for_addr_array, vrp_initialize): Likewise.
3496 (vrp_finalize): Likewise. Revamp to avoid direct access to
3497 vr_value, values_propagated, etc.
3498 (check_array_bounds): Extract vrp_prop class instance pointer from
3499 walk info structure. Use it to call member functions.
3500 (check_all_array_refs): Make a member function within vrp_prop class.
3501 Smuggle class instance pointer via walk info structure.
3502 (x_vr_values): New local static.
3503 (vrp_valueize): Use x_vr_values to get class instance.
3504 (vr_valueize_1): Likewise.
3505 (class vrp_folder): Add vr_values data member. Add various member
3506 functions as well as member functions that delegate to vr_values.
3507 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
3508 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
3509 class instance from vr_values. Use it to call member functions.
3510 (vrp_dom_walker): Add vr_values data member.
3511 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
3512 instance via x_vr_values.
3513 (identify_jump_threads): Accept vr_values as argument. Store
3514 it into the walker structure.
3515 (evrp_dom_walker): Add vr_values class data member. Add various
3517 (evrp_dom_walker::try_find_new_range): Use vr_values data
3518 member to access the memory allocator.
3519 (evrp_dom_walker::before_dom_children): Store vr_values class
3520 instance into the vrp_folder class.
3521 (evrp_dom_walker::push_value_range): Rework to avoid direct
3522 access to num_vr_values and vr_value.
3523 (evrp_dom_walker::pop_value_range): Likewise.
3524 (execute_early_vrp): Remove call to vrp_initialize_lattice.
3525 Use vr_values to get to dump_all_value_ranges member function.
3526 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
3527 and simplify_cond_using_ranges_2 via vrp_prop class instance.
3528 Pass vr_values class instance down to identify_jump_threads.
3529 Remove call to vrp_free_lattice.
3530 (debug_all_value_ranges): Remove.
3532 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
3533 (vrp_folder): Likewise.
3535 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
3536 Get it from the existing bitmap instead.
3537 (vrp_intersect_ranges_1): Likewise.
3539 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3541 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3543 (store_immediate_info::store_immediate_info): Add bitnotp argument,
3544 set bit_not_p to it.
3545 (imm_store_chain_info::coalesce_immediate_stores): Break group
3546 if bit_not_p is different.
3547 (count_multiple_uses, split_group,
3548 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
3549 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
3550 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
3551 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
3552 to store_immediate_info ctor.
3554 2017-11-09 Jim Wilson <jimw@sifive.com>
3556 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
3557 (scan_prog_file): Likewise.
3559 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3561 * bb-reorder.c (max_entry_frequency): Remove.
3562 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
3563 connect_traces, push_to_next_round_p): Remove prototypes.
3564 (find_traces_1_round): Use counts only.
3565 (push_to_next_round_p): Likewise.
3566 (find_traces): Likewise.
3567 (rotate_loop): Likewise.
3568 (find_traces_1_round): Likewise.
3569 (connect_traces): Likewise.
3570 (edge_order): Likewise.
3572 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
3574 * config/arm/arm.c (output_return_instruction): Add comments to
3575 indicate requirement for cmse_nonsecure_entry return to account
3576 for the size of clearing instruction output here.
3577 (thumb_exit): Likewise.
3578 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
3579 return in hardfloat mode.
3581 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
3583 * config/rs6000/rs6000.c (machine_function): Add a bool,
3584 "toc_is_wrapped_separately".
3585 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
3586 if it wasn't explicitly set or unset, we are optimizing for speed, and
3587 doing separate shrink-wrapping.
3588 (rs6000_get_separate_components): Enable the TOC component if
3589 saving the TOC register in the prologue.
3590 (rs6000_components_for_bb): Handle the TOC component.
3591 (rs6000_emit_prologue_components): Store the TOC register where needed.
3592 (rs6000_set_handled_components): Mark TOC as handled, if handled.
3593 (rs6000_emit_prologue): Don't save the TOC if that is already done.
3595 2017-11-09 Martin Jambor <mjambor@suse.cz>
3597 * ipa-param-manipulation.c: New file.
3598 * ipa-param-manipulation.h: Likewise.
3599 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
3600 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
3601 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
3602 (ipa_parm_adjustment): Likewise.
3603 (ipa_parm_adjustment_vec): Likewise.
3604 (ipa_get_vector_of_formal_parms): Moved declaration to
3605 ipa-param-manipulation.h.
3606 (ipa_get_vector_of_formal_parm_types): Likewise.
3607 (ipa_modify_formal_parameters): Likewise.
3608 (ipa_modify_call_arguments): Likewise.
3609 (ipa_combine_adjustments): Likewise.
3610 (ipa_dump_param_adjustments): Likewise.
3611 (ipa_modify_expr): Likewise.
3612 (ipa_get_adjustment_candidate): Likewise.
3613 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
3614 ipa-param-manipulation.c.
3615 (ipa_get_vector_of_formal_parm_types): Likewise.
3616 (ipa_modify_formal_parameters): Likewise.
3617 (ipa_modify_call_arguments): Likewise.
3618 (ipa_modify_expr): Likewise.
3619 (get_ssa_base_param): Likewise.
3620 (ipa_get_adjustment_candidate): Likewise.
3621 (index_in_adjustments_multiple_times_p): Likewise.
3622 (ipa_combine_adjustments): Likewise.
3623 (ipa_dump_param_adjustments): Likewise.
3624 * tree-sra.c: Also include ipa-param-manipulation.h
3625 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
3628 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3629 Alan Hayward <alan.hayward@arm.com>
3630 David Sherwood <david.sherwood@arm.com>
3632 * doc/sourcebuild.texi (vect_masked_store): Document.
3634 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3635 Alan Hayward <alan.hayward@arm.com>
3636 David Sherwood <david.sherwood@arm.com>
3638 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
3640 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3641 Alan Hayward <alan.hayward@arm.com>
3642 David Sherwood <david.sherwood@arm.com>
3644 * doc/sourcebuild.texi (vect_variable_length): Document.
3646 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3647 Alan Hayward <alan.hayward@arm.com>
3648 David Sherwood <david.sherwood@arm.com>
3650 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
3652 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3653 Alan Hayward <alan.hayward@arm.com>
3654 David Sherwood <david.sherwood@arm.com>
3656 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
3658 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3659 Alan Hayward <alan.hayward@arm.com>
3660 David Sherwood <david.sherwood@arm.com>
3662 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
3663 previously undocumented selectors.
3664 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
3666 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3668 * doc/rtl.texi (const_vector): Say that elements can be
3669 const_wide_ints too.
3670 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
3671 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
3672 (gen_vec_duplicate): Use it instead of CONSTANT_P.
3673 * optabs.c (expand_vector_broadcast): Likewise.
3675 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3676 Alan Hayward <alan.hayward@arm.com>
3677 David Sherwood <david.sherwood@arm.com>
3679 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
3680 scaled index even if the unscaled address was invalid.
3681 Don't increase the complexity of using a scale in that case.
3683 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3684 Alan Hayward <alan.hayward@arm.com>
3685 David Sherwood <david.sherwood@arm.com>
3687 * doc/rtl.texi: Rewrite the subreg rules so that they partition
3688 the inner register into REGMODE_NATURAL_SIZE bytes rather than
3689 UNITS_PER_WORD bytes.
3690 * emit-rtl.c (validate_subreg): Divide subregs into blocks
3691 based on REGMODE_NATURAL_SIZE of the inner mode.
3692 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
3693 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
3694 * expmed.c (lowpart_bit_field_p): Divide the value up into
3695 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
3696 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
3697 whether something is likely to occupy more than one register.
3699 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3702 * ipa-inline-transform.c (update_noncloned_frequencies): Use
3703 profile_count::adjust_for_ipa_scaling.
3704 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
3705 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
3707 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
3709 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3711 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
3712 (split_group): Add total_orig and total_new arguments, estimate the
3713 number of statements related to the store group without store merging
3714 and with store merging.
3715 (imm_store_chain_info::output_merged_store): Adjust split_group
3716 callers, punt if estimated number of statements with store merging
3717 is not smaller than estimated number of statements without it.
3719 (handled_load): Remove has_single_use checks.
3720 (pass_store_merging::process_store): Likewise.
3722 2017-11-09 Richard Biener <rguenther@suse.de>
3724 PR tree-optimization/82902
3725 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
3727 2017-11-09 Martin Liska <mliska@suse.cz>
3730 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
3733 2017-11-09 Martin Liska <mliska@suse.cz>
3735 PR tree-optimization/82669
3736 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
3738 2017-11-09 Martin Liska <mliska@suse.cz>
3740 PR gcov-profile/48463
3741 * coverage.c (coverage_begin_function): Output also end locus
3742 of a function and information whether the function is
3744 * gcov-dump.c (tag_function): Parse and print the information.
3745 * gcov.c (INCLUDE_MAP): Add include.
3746 (INCLUDE_SET): Likewise.
3747 (struct line_info): Move earlier in the source file because
3748 of vector<line_info> in function_info structure.
3749 (line_info::line_info): Likewise.
3750 (line_info::has_block): Likewise.
3751 (struct source_info): Add new member index.
3752 (source_info::get_functions_at_location): New function.
3753 (function_info::group_line_p): New function.
3754 (output_intermediate_line): New function.
3755 (output_intermediate_file): Use the mentioned function.
3756 (struct function_start): New.
3757 (struct function_start_pair_hash): Likewise.
3758 (process_file): Add code that identifies group functions.
3759 Assign lines either to global or function scope.
3760 (generate_results): Skip artificial functions.
3761 (find_source): Assign index for each source file.
3762 (read_graph_file): Read new flag artificial and end_line.
3763 (add_line_counts): Assign it either to global of function scope.
3764 (accumulate_line_counts): Isolate core of the function to
3765 accumulate_line_info and call it for both function and global
3767 (accumulate_line_info): New function.
3768 (output_line_beginning): Fix GNU coding style.
3769 (print_source_line): New function.
3770 (output_line_details): Likewise.
3771 (output_function_details): Likewise.
3772 (output_lines): Iterate both source (global) scope and function
3774 (struct function_line_start_cmp): New class.
3775 * doc/gcov.texi: Reflect changes in documentation.
3777 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3780 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
3781 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
3782 and similarly for not instead of neg.
3784 2017-11-08 Andi Kleen <ak@linux.intel.com>
3786 * config/i386/i386.opt: Add -mforce-indirect-call.
3787 * config/i386/predicates.md: Check for flag_force_indirect_call.
3788 * doc/invoke.texi: Document -mforce-indirect-call
3790 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
3792 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
3794 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
3795 config/riscv/riscv.c (predict.h): New include.
3796 (riscv_slow_unaligned_access_p): No longer static.
3797 (riscv_block_move_straight): Add require.
3798 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3800 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3803 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
3804 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
3805 nonimmediate_operand predicate for operand 1 instead of
3808 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3810 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
3812 * config/aarch64/constraints.md (Uml): New constraint.
3813 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
3816 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3818 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
3819 of two vec_duplicates into a vec_concat.
3821 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3823 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3824 Simplify vec_merge of vec_duplicate and vec_concat.
3825 * config/aarch64/constraints.md (Utq): New constraint.
3826 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
3829 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3831 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3832 Simplify vec_merge of vec_duplicate and const_vector.
3833 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
3835 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
3836 mode iterator. Update predicate on operand 1 to
3837 handle non-const_vec constants. Delete constraints.
3838 (*aarch64_combinez_be<mode>): Likewise for operand 2.
3840 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3842 PR tree-optimization/78821
3843 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
3845 (store_operand_info::store_operand_info): Initialize it to false.
3846 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
3847 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
3848 store in the group, and if chain_info is non-NULL, to ignore altogether
3850 (compatible_load_p): Fail if bit_not_p does not match.
3851 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
3852 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
3853 (pass_store_merging::process_store): Adjust
3854 terminate_all_aliasing_chains calls to pass NULL in all current spots,
3855 call terminate_all_aliasing_chains newly when adding a store into
3856 a chain with non-NULL chain_info.
3858 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3860 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
3862 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3864 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3866 (aarch64_layout_frame): Initialise emit_frame_chain.
3867 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
3868 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
3870 2017-11-08 Martin Liska <mliska@suse.cz>
3872 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
3873 of gimple_call_internal_p.
3875 2017-11-07 Tom de Vries <tom@codesourcery.com>
3877 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
3879 2017-11-07 Tom de Vries <tom@codesourcery.com>
3881 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3884 2017-11-08 Martin Liska <mliska@suse.cz>
3887 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
3889 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3891 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
3893 2017-11-07 Jakub Jelinek <jakub@redhat.com>
3896 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
3897 (*cmp<mode>_ccz_1): New insn with $k alternative.
3900 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
3901 RTX_COMM_COMPARE as commutative as well.
3902 (ix86_binary_operator_ok): Formatting fix.
3903 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
3904 *<code><mode>3<mask_name><round_saeonly_name>,
3905 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
3906 *<s>mul<mode>3_highpart<mask_name>,
3907 *vec_widen_umult_even_v16si<mask_name>,
3908 *vec_widen_umult_even_v8si<mask_name>,
3909 *vec_widen_umult_even_v4si<mask_name>,
3910 *vec_widen_smult_even_v16si<mask_name>,
3911 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
3912 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
3913 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
3914 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
3915 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
3916 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
3917 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
3918 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
3919 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
3920 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
3921 ix86_binary_operator_ok. Formatting fixes.
3922 (*<plusminus_insn><mode>3<mask_name><round_name>,
3923 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
3926 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
3928 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
3929 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
3930 eq and ne if TARGET_ISEL.
3931 (cmp): New code_iterator.
3932 (UNS, UNSU_, UNSIK): New code_attrs.
3933 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
3934 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
3936 ("eq<mode>3"): ... this.
3937 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
3939 ("ne<mode>3"): ... this.
3941 2017-11-07 Julia Koval <julia.koval@intel.com>
3944 * common/config/i386/i386-common.c
3945 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
3946 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
3947 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
3948 * config/i386/i386.opt: Ditto.
3949 * config/i386/i386.c (ix86_target_string): Ditto.
3950 (ix86_option_override_internal): Ditto.
3951 (ix86_init_mpx_builtins): Move MPX to args2.
3952 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
3953 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3954 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3955 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3956 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3957 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3958 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
3960 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
3963 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
3964 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
3965 (zero-extendsidi peephole2): Remove peephole.
3967 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3970 * stor-layout.c: Include attribs.h.
3971 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
3972 explicit lookup of "aligned" attribute.
3974 2017-11-07 Andrew Waterman <andrew@sifive.com>
3976 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
3977 (riscv_expand_block_move): Likewise.
3978 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
3980 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
3981 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
3982 * config/riscv/riscv.c (riscv_block_move_straight): New function.
3983 (riscv_adjust_block_mem): Likewise.
3984 (riscv_block_move_loop): Likewise.
3985 (riscv_expand_block_move): Likewise.
3986 * config/riscv/riscv.md (movmemsi): New pattern.
3988 2017-11-07 Michael Clark <michaeljclark@mac.com>
3990 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
3991 (MUSL_DYNAMIC_LINKER): Likewise.
3993 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
3995 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
3996 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
3999 2017-11-07 Richard Biener <rguenther@suse.de>
4001 * match.pd: Fix build.
4003 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
4004 Jackson Woodruff <jackson.woodruff@arm.com>
4006 PR tree-optimization/71026
4007 * match.pd: Canonicalize negate in division.
4009 2017-11-07 Sudakshina Das <sudi.das@arm.com>
4012 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
4014 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4016 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
4017 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
4019 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4021 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
4022 non-scalar integral types.
4023 * match.pd (negate_expr_p): Handle MINUS_EXPR.
4024 (-(A-B), -(~A)): New transformations.
4026 2017-11-07 Tom de Vries <tom@codesourcery.com>
4028 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
4029 semicolon after "do {} while (0)".
4030 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4031 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4032 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4033 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4034 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4036 2017-11-07 Tom de Vries <tom@codesourcery.com>
4038 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
4039 after "do {} while (0)".
4040 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4041 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4042 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4043 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4044 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4046 2017-11-07 Tom de Vries <tom@codesourcery.com>
4049 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
4051 (arm_rtx_costs_internal): Add missing semicolon after
4052 HANDLE_NARROW_SHIFT_ARITH call.
4054 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
4056 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4057 disable isel if it was not set explicitly.
4059 2017-11-06 James Bowman <james.bowman@ftdichip.com>
4061 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
4062 (add_type_attribute) likewise.
4064 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
4066 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
4067 of struct ix86_frame.
4068 (ix86_initial_elimination_offset): Likewise.
4069 (ix86_expand_split_stack_prologue): Likewise.
4071 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
4073 * tree-vrp.h (enum value_range_type): Update stale comment.
4075 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4076 Alan Hayward <alan.hayward@arm.com>
4077 David Sherwood <david.sherwood@arm.com>
4079 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
4080 (aarch64_expand_vec_perm_const): Take the number of units too.
4081 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
4082 (aarch64_expand_vec_perm_const): Likewise.
4083 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
4084 (vec_perm<mode>): Update accordingly.
4086 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4087 Alan Hayward <alan.hayward@arm.com>
4088 David Sherwood <david.sherwood@arm.com>
4090 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
4091 Take the number of units too.
4092 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
4093 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
4094 but check for a vector mode before rather than after the call.
4095 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
4096 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
4097 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
4098 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
4099 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
4100 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
4101 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
4102 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
4103 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
4104 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
4105 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
4106 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
4107 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
4108 (aarch64_sqdmull2_n<mode>): Update accordingly.
4110 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4111 Alan Hayward <alan.hayward@arm.com>
4112 David Sherwood <david.sherwood@arm.com>
4114 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
4115 the number of units too.
4116 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
4117 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
4118 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
4119 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
4120 (vec_store_lanesxi<mode>): Update accordingly.
4122 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4123 Alan Hayward <alan.hayward@arm.com>
4124 David Sherwood <david.sherwood@arm.com>
4126 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
4127 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
4128 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
4129 of units rather than the mode.
4130 * config/aarch64/iterators.md (nunits): New mode attribute.
4131 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4132 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
4133 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
4134 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
4135 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
4136 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
4137 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
4138 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
4139 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
4140 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
4141 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4142 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
4143 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
4144 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
4145 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
4146 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
4147 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
4148 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
4149 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
4150 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
4151 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
4152 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
4153 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
4154 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4155 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
4156 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
4157 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4158 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
4159 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
4160 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
4161 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
4162 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
4163 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
4164 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
4165 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
4166 (aarch64_simd_vec_setv2di): Likewise.
4168 2017-11-06 Carl Love <cel@us.ibm.com>
4170 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
4172 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
4173 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
4175 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
4177 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
4178 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
4179 Add power 8 macro expansions.
4180 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
4181 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
4182 power 8 instructions. (VSX_XXBR): Add iterator.
4184 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
4186 * config/arm/arm.md (predicable_short_it): Change default to "no",
4187 improve documentation, remove uses that are identical to the default.
4188 (enabled_for_depr_it): Rename to enabled_for_short_it.
4189 * gcc/config/arm/arm-fixed.md (predicable_short_it):
4190 Remove default uses.
4191 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
4192 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
4193 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
4194 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
4196 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
4199 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
4200 float128 helper macros, which are no longer used after deleting
4201 the old 'q' built-in functions, and moving the round to odd
4202 built-in functions to being special built-in functions.
4203 (BU_FLOAT128_2): Likewise.
4204 (BU_FLOAT128_1_HW): Likewise.
4205 (BU_FLOAT128_2_HW): Likewise.
4206 (BU_FLOAT128_3_HW): Likewise.
4207 (FABSQ): Delete old 'q' built-in functions.
4208 (COPYSIGNQ): Likewise.
4209 (SQRTF128_ODD): Move round to odd built-in functions to be
4210 special built-in functions, so that we can handle
4211 -mabi=ieeelongdouble.
4212 (TRUNCF128_ODD): Likewise.
4213 (ADDF128_ODD): Likewise.
4214 (SUBF128_ODD): Likewise.
4215 (MULF128_ODD): Likewise.
4216 (DIVF128_ODD): Likewise.
4217 (FMAF128_ODD): Likewise.
4218 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
4219 built-in names to 'f128'.
4220 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
4221 old 'q' built-in functions, as the machine independent code for
4222 'f128' built-in functions handles this.
4223 (rs6000_expand_builtin): Add expansion for float128 round to odd
4224 functions, keying off on -mabi=ieeelongdouble of whether to use
4225 the KFmode or TFmode variant.
4226 (rs6000_init_builtins): Initialize the _Float128 round to odd
4228 * doc/extend.texi (PowerPC Built-in Functions): Document the old
4229 _Float128 'q' built-in functions are now mapped into the new
4230 'f128' built-in functions.
4232 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
4234 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
4235 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
4237 2017-11-06 David Malcolm <dmalcolm@redhat.com>
4240 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
4241 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
4242 * toplev.c: Include "ipa-fnsummary.h".
4243 (toplev::finalize): Call ipa_fnsummary_c_finalize.
4245 2017-11-06 Jakub Jelinek <jakub@redhat.com>
4247 PR tree-optimization/82838
4248 * gimple-ssa-store-merging.c
4249 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
4250 on a separate gimple_seq which is then appended to seq.
4252 2017-11-06 Jeff Law <law@redhat.com>
4255 * config/i386/i386.c (PROBE_INTERVAL): Remove.
4256 (get_probe_interval): New functions.
4257 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
4258 (ix86_adjust_stack_and_probe): Likewise.
4259 (output_adjust_stack_and_probe): Likewise.
4260 (ix86_emit_probe_stack_range): Likewise.
4261 (ix86_expand_prologue): Likewise.
4263 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4265 PR tree-optimization/82816
4266 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
4267 if the modes of the two types are the same.
4268 (convert_plusminus_to_widen): Likewise.
4270 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4272 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
4274 (usadv16qi): New define_expand.
4275 (usadv8hi): New define_expand.
4277 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
4280 * ipa-inline-transform.c (update_noncloned_frequencies): Always
4282 (inline_transform): Likewise.
4283 * predict.c (counts_to_freqs): Remove useless conditional.
4284 * profile-count.h (profile_count::apply_scale): Move sanity check.
4285 * tree-inline.c (copy_bb): Always scale.
4286 (copy_cfg_body): Likewise.
4288 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
4291 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
4293 (*cmp_ite0): Add enabled_for_depr_it attribute.
4294 (*cmp_ite1): Likewise.
4296 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
4298 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
4301 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4303 * tree-vrp.c (vrp_int_const_binop): Return true on success and
4304 return the value by pointer.
4305 (extract_range_from_multiplicative_op_1): Update accordingly.
4306 Return as soon as an operation fails.
4308 2017-11-05 Tom de Vries <tom@codesourcery.com>
4311 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
4312 (DEF_SANITIZER_BUILTIN): ... here.
4313 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
4314 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
4316 2017-11-05 Tom de Vries <tom@codesourcery.com>
4319 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
4321 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
4322 ASM_OUTPUT_BEFORE_CASE_LABEL call.
4323 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
4325 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4326 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4327 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4329 2017-11-05 Tom de Vries <tom@codesourcery.com>
4332 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
4335 2017-11-04 Michael Clark <michaeljclark@mac.com>
4337 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
4338 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
4340 (*addsi3_extended): Likewise.
4341 (*addsi3_extended2): Likewise.
4342 (<optab>si3): Likewise.
4343 (<optab>di3): Likewise.
4344 (<optab><mode>3): Likewise.
4345 (<*optabe>si3_internal): Likewise.
4346 (zero_extendqi<SUPERQI:mode>2): Likewise.
4347 (*add<mode>hi3): Likewise.
4348 (*xor<mode>hi3): Likewise.
4349 (<optab>di3): Likewise.
4350 (*<optab>si3_extend): Likewise.
4351 (*sge<u>_<X:mode><GPR:mode>): Likewise.
4352 (*slt<u>_<X:mode><GPR:mode>): Likewise.
4353 (*sle<u>_<X:mode><GPR:mode>): Likewise.
4355 2017-11-04 Andrew Waterman <andrew@sifive.com>
4357 * config/riscv/riscv.c (riscv_option_override): Conditionally set
4358 TARGET_STRICT_ALIGN based upon -mtune argument.
4360 2017-11-04 Andrew Waterman <andrew@sifive.com>
4362 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
4364 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
4366 * config/i386/i386.c (choose_basereg): Use optional scratch
4367 register and add assertion.
4368 (x86_emit_outlined_ms2sysv_save): Use scratch register when
4369 needed, and don't allocate stack.
4370 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
4371 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
4372 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
4374 2017-11-03 Jeff Law <law@redhat.com>
4376 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
4377 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
4378 to probe at the start of a noreturn function.
4380 2017-11-03 Jakub Jelinek <jakub@redhat.com>
4382 PR tree-optimization/78821
4383 * gimple-ssa-store-merging.c: Update the file comment.
4384 (MAX_STORE_ALIAS_CHECKS): Define.
4385 (struct store_operand_info): New type.
4386 (store_operand_info::store_operand_info): New constructor.
4387 (struct store_immediate_info): Add rhs_code and ops data members.
4388 (store_immediate_info::store_immediate_info): Add rhscode, op0r
4389 and op1r arguments to the ctor, initialize corresponding data members.
4390 (struct merged_store_group): Add load_align_base and load_align
4392 (merged_store_group::merged_store_group): Initialize them.
4393 (merged_store_group::do_merge): Update them.
4394 (merged_store_group::apply_stores): Pick the constant for
4395 encode_tree_to_bitpos from one of the two operands, or skip
4396 encode_tree_to_bitpos if neither operand is a constant.
4397 (class pass_store_merging): Add process_store method decl. Remove
4398 bool argument from terminate_all_aliasing_chains method decl.
4399 (pass_store_merging::terminate_all_aliasing_chains): Remove
4400 var_offset_p argument and corresponding handling.
4401 (stmts_may_clobber_ref_p): New function.
4402 (compatible_load_p): New function.
4403 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
4404 if there is overlap and rhs_code is not INTEGER_CST. For
4405 non-overlapping stores terminate group if rhs is not mergeable.
4406 (get_alias_type_for_stmts): Change first argument from
4407 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
4408 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
4409 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
4411 (get_location_for_stmts): Change first argument from
4412 auto_vec<gimple *> & to vec<gimple *> &.
4413 (struct split_store): Remove orig_stmts data member, add orig_stores.
4414 (split_store::split_store): Create orig_stores rather than orig_stmts.
4415 (find_constituent_stmts): Renamed to ...
4416 (find_constituent_stores): ... this. Change second argument from
4417 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
4418 to info structures rather than the statements.
4419 (split_group): Rename ALLOW_UNALIGNED argument to
4420 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
4421 it. Adjust find_constituent_stores caller.
4422 (imm_store_chain_info::output_merged_store): Handle rhs_code other
4423 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
4424 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
4425 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
4426 (mem_valid_for_store_merging): New function.
4427 (handled_load): New function.
4428 (pass_store_merging::process_store): New method.
4429 (pass_store_merging::execute): Use process_store method. Adjust
4430 terminate_all_aliasing_chains caller.
4432 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4434 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4435 Return true for more constants, symbols and label references.
4436 (aarch64_valid_floating_const): Remove unused function.
4438 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4441 * config/aarch64/aarch64.c (aarch64_layout_frame):
4442 Undo forcing of LR at bottom of frame.
4444 2017-11-03 Jeff Law <law@redhat.com>
4447 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
4448 for int_registers_saved.
4450 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
4451 extracted from tree-ssa-dom.c.
4452 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
4453 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
4454 (record_equivalences_from_incoming_edge): Add additional argument
4455 to single_pred_edge_ignoring_loop_edges call.
4456 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
4457 (uncprop_dom_walker::before_dom_children): Add additional argument
4458 to single_pred_edge_ignoring_loop_edges call.
4459 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
4460 single_pred_edge_ignoring_loop_edges rather than open coding.
4461 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
4463 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
4465 * match.pd (-(-A)): Rewrite.
4467 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4469 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
4470 (rs6000_emit_int_cmove): New declaration.
4471 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
4472 (rs6000_emit_sISEL): Delete.
4473 (rs6000_emit_int_cmove): Make non-static.
4474 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
4475 instead of rs6000_emit_sISEL.
4477 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
4479 * asan.c (create_cond_insert_point): Maintain profile.
4480 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
4482 * basic-block.h (struct basic_block_def): Remove frequency.
4483 (EDGE_FREQUENCY): Use to_frequency
4484 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
4486 (find_traces): Update to use to_frequency.
4487 (find_traces_1_round): Likewise; use only IPA counts.
4488 (bb_to_key): Likewise.
4489 (connect_traces): Use IPA counts only.
4490 (copy_bb_p): Update to use to_frequency.
4491 (fix_up_crossing_landing_pad): Likewise.
4492 (sanitize_hot_paths): Likewise.
4493 * bt-load.c (basic_block_freq): Likewise.
4494 * cfg.c (init_flow): Set count_max to uninitialized.
4495 (check_bb_profile): Remove frequencies; check counts.
4496 (dump_bb_info): Do not dump frequencies.
4497 (update_bb_profile_for_threading): Update counts only.
4498 (scale_bbs_frequencies_int): Likewise.
4499 (MAX_SAFE_MULTIPLIER): Remove.
4500 (scale_bbs_frequencies_gcov_type): Update counts only.
4501 (scale_bbs_frequencies_profile_count): Update counts only.
4502 (scale_bbs_frequencies): Update counts only.
4503 * cfg.h (struct control_flow_graph): Add count-max.
4504 (update_bb_profile_for_threading): Update prototype.
4505 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
4506 (find_many_sub_basic_blocks): Likewise.
4507 * cfgcleanup.c (try_forward_edges): Likewise.
4508 (try_crossjump_to_edge): Likewise.
4509 * cfgexpand.c (expand_gimple_cond): Likewise.
4510 (expand_gimple_tailcall): Likewise.
4511 (construct_init_block): Likewise.
4512 (construct_exit_block): Likewise.
4513 * cfghooks.c (verify_flow_info): Check consistency of counts.
4514 (dump_bb_for_graph): Do not dump frequencies.
4515 (split_block_1): Do not update frequencies.
4516 (split_edge): Do not update frequencies.
4517 (make_forwarder_block): Do not update frequencies.
4518 (duplicate_block): Do not update frequencies.
4519 (account_profile_record): Do not update frequencies.
4520 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
4521 for global heuristics.
4522 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
4523 (expected_loop_iterations_unbounded): Use counts only.
4524 * cfgloopmanip.c (scale_loop_profile): Simplify.
4525 (create_empty_loop_on_edge): Simplify
4527 (duplicate_loop_to_header_edge): Simplify
4528 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
4529 (update_br_prob_note): Take care of removing note when profile
4531 (relink_block_chain): Do not dump frequency.
4532 (rtl_account_profile_record): Use to_frequency.
4533 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
4534 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
4535 (cgraph_update_edges_for_call_stmt_node): Likewise.
4536 (cgraph_edge::verify_count_and_frequency): Update.
4537 (cgraph_node::verify_node): Temporarily disable frequency verification.
4538 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
4539 to_cgraph_frequency.
4540 (cgraph_edge::rebuild_edges): Convert to ipa counts.
4541 * cgraphunit.c (init_lowered_empty_function): Do not initialize
4543 (cgraph_node::expand_thunk): Update profile.
4544 * except.c (dw2_build_landing_pads): Do not update frequency.
4545 * final.c (compute_alignments): Use to_frequency.
4546 (dump_basic_block_info): Do not dump frequency.
4547 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
4548 (dump_gimple_bb_header): Do not dump frequency.
4549 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
4551 * gimple-streamer-in.c (input_bb): Do not stream frequency.
4552 * gimple-streamer-out.c (output_bb): Do not stream frequency.
4553 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
4554 (init_before_recovery): Do not update frequency.
4555 (sched_create_recovery_edges): Do not update frequency.
4556 * hsa-gen.c (convert_switch_statements): Do not update frequency.
4557 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
4558 (ipa_cp_c_finalize): Set max_count to uninitialized.
4559 * ipa-fnsummary.c (get_minimal_bb): Use counts.
4560 (param_change_prob): Use counts.
4561 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
4563 * ipa-split.c (consider_split): Use to_frequency.
4564 (split_function): Use to_frequency.
4565 * ira-build.c (loop_compare_func): Likewise.
4566 (mark_loops_for_removal): Likewise.
4567 (mark_all_loops_for_removal): Likewise.
4568 * loop-doloop.c (doloop_modify): Do not update frequency.
4569 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
4571 * lto-streamer-in.c (input_function): Update count_max.
4572 * omp-expand.c (expand_omp_taskreg): Update count_max.
4573 * omp-simd-clone.c (simd_clone_adjust): Update profile.
4574 * predict.c (maybe_hot_frequency_p): Use to_frequency.
4575 (maybe_hot_count_p): Use ipa counts only.
4576 (maybe_hot_bb_p): Simplify.
4577 (maybe_hot_edge_p): Simplify.
4578 (probably_never_executed): Do not take frequency argument.
4579 (probably_never_executed_bb_p): Do not pass frequency.
4580 (probably_never_executed_edge_p): Likewise.
4581 (combine_predictions_for_bb): Check that profile is nonzero.
4582 (propagate_freq): Do not set frequency.
4583 (drop_profile): Simplify.
4584 (counts_to_freqs): Simplify.
4585 (expensive_function_p): Use to_frequency.
4586 (propagate_unlikely_bbs_forward): Simplify.
4587 (determine_unlikely_bbs): Simplify.
4588 (estimate_bb_frequencies): Add hack to silence graphite issues.
4589 (compute_function_frequency): Use ipa counts.
4590 (pass_profile::execute): Update.
4591 (rebuild_frequencies): Use counts only.
4592 (force_edge_cold): Use counts only.
4593 * profile-count.c (profile_count::dump): Dump new count types.
4594 (profile_count::differs_from_p): Check compatiblity.
4595 (profile_count::to_frequency): New function.
4596 (profile_count::to_cgraph_frequency): New function.
4597 * profile-count.h (struct function): Declare.
4598 (enum profile_quality): Add profile_guessed_local and
4599 profile_guessed_global0.
4600 (class profile_proability): Decrease number of bits to 29;
4601 update from_reg_br_prob_note and to_reg_br_prob_note.
4602 (class profile_count: Update comment; decrease number of bits
4603 to 61. Check compatibility.
4604 (profile_count::compatible_p): New private member function.
4605 (profile_count::ipa_p): New member function.
4606 (profile_count::operator<): Handle global zero correctly.
4607 (profile_count::operator>): Handle global zero correctly.
4608 (profile_count::operator<=): Handle global zero correctly.
4609 (profile_count::operator>=): Handle global zero correctly.
4610 (profile_count::nonzero_p): New member function.
4611 (profile_count::force_nonzero): New member function.
4612 (profile_count::max): New member function.
4613 (profile_count::apply_scale): Handle IPA scalling.
4614 (profile_count::guessed_local): New member function.
4615 (profile_count::global0): New member function.
4616 (profile_count::ipa): New member function.
4617 (profile_count::to_frequency): Declare.
4618 (profile_count::to_cgraph_frequency): Declare.
4619 * profile.c (OVERLAP_BASE): Delete.
4620 (compute_frequency_overlap): Delete.
4621 (compute_branch_probabilities): Do not use compute_frequency_overlap.
4622 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
4623 * sched-ebb.c (rank): Use counts only.
4624 * shrink-wrap.c (handle_simple_exit): Use counts only.
4625 (try_shrink_wrapping): Use counts only.
4626 (place_prologue_for_one_component): Use counts only.
4627 * tracer.c (find_best_predecessor): Use to_frequency.
4628 (find_trace): Use to_frequency.
4629 (tail_duplicate): Use to_frequency.
4630 * trans-mem.c (expand_transaction): Do not update frequency.
4631 * tree-call-cdce.c: Do not update frequency.
4632 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
4633 (gimple_merge_blocks): Likewise.
4634 (gimple_split_edge): Likewise.
4635 (gimple_duplicate_sese_region): Likewise.
4636 (gimple_duplicate_sese_tail): Likewise.
4637 (move_sese_region_to_fn): Likewise.
4638 (gimple_account_profile_record): Likewise.
4639 (insert_cond_bb): Likewise.
4640 * tree-complex.c (expand_complex_div_wide): Likewise.
4641 * tree-eh.c (lower_resx): Update profile.
4642 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
4644 (initialize_cfun): Do not initialize frequencies
4645 (freqs_to_counts): Delete.
4646 (copy_cfg_body): Ignore count parameter.
4647 (copy_body): Update.
4648 (expand_call_inline): Update count_max.
4649 (optimize_inline_calls): Update count_max.
4650 (tree_function_versioning): Update count_max.
4651 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
4652 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
4654 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
4655 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
4656 (try_peel_loop): Likewise.
4657 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
4659 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
4660 (tree_transform_and_unroll_loop): Do not use frequencies
4661 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
4662 Use reliable prediction only.
4663 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
4664 * tree-ssa-sink.c (select_best_block): Use to_frequency.
4665 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
4666 probability scaling.
4667 * tree-ssa-threadupdate.c (create_block_for_threading): Do
4668 not update frequency
4669 (any_remaining_duplicated_blocks): Likewise.
4670 (update_profile): Likewise.
4671 (estimated_freqs_path): Delete.
4672 (freqs_to_counts_path): Delete.
4673 (clear_counts_path): Delete.
4674 (ssa_fix_duplicate_block_edges): Likewise.
4675 (duplicate_thread_path): Likewise.
4676 * tree-switch-conversion.c (gen_inbound_check): Use counts.
4677 * tree-tailcall.c (decrease_profile): Do not update frequency.
4678 (eliminate_tail_call): Likewise.
4679 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
4680 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
4681 (optimize_mask_stores): Likewise.
4682 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4683 * ubsan.c (ubsan_expand_null_ifn): Update profile.
4684 (ubsan_expand_ptr_ifn): Update profile.
4685 * value-prof.c (gimple_ic): Simplify.
4686 * value-prof.h (gimple_ic): Update prototype.
4687 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
4688 * ipa-inline.c (compute_uninlined_call_time): Be sure that
4690 (want_inline_self_recursive_call_p): Likewise.
4691 (resolve_noninline_speculation): Only cummulate defined counts.
4692 (inline_small_functions): Use nonzero_p.
4693 (ipa_inline): Do not access freed node.
4695 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4697 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4698 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
4700 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
4702 * config/riscv/riscv.c (riscv_legitimize_move): Handle
4703 non-legitimate address.
4705 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4707 * config/rs6000/rs6000.md (*lt0_disi): Delete.
4708 (*lt0_<mode>di, *lt0_<mode>si): New.
4710 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4712 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
4713 TARGET_PAIRED_FLOAT.
4715 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
4716 Jim Wilson <jim.wilson@linaro.org>
4718 * config/aarch64/aarch64-cores.def (saphira): New CPU.
4719 * config/aarch64/aarch64-tune.md: Regenerated.
4720 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
4721 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
4723 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
4725 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
4726 (arc_expand_prologue): Restore blink for millicode.
4727 * config/arc/linux.h (LINK_EH_SPEC): Defined.
4729 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
4732 * config/i386/i386.c (ix86_vector_duplicate_value): Use
4733 gen_vec_duplicate after forcing the scalar into a register.
4735 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
4737 * combine (try_combine): Print the insns input to try_combine to the
4740 2017-11-02 Steve Ellcey <sellcey@cavium.com>
4743 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
4744 Remove second argument from aarch64_process_target_attr call.
4745 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
4747 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
4749 (aarch64_handle_attr_arch): Remove second argument.
4750 (aarch64_handle_attr_cpu): Ditto.
4751 (aarch64_handle_attr_tune): Ditto.
4752 (aarch64_handle_attr_isa_flags): Ditto.
4753 (aarch64_process_one_target_attr): Ditto.
4754 (aarch64_process_target_attr): Ditto.
4755 (aarch64_option_valid_attribute_p): Remove second argument.
4756 on aarch64_process_target_attr call.
4758 2017-11-02 David Malcolm <dmalcolm@redhat.com>
4760 * diagnostic.c: Include "selftest-diagnostic.h".
4761 (selftest::assert_location_text): New function.
4762 (selftest::test_diagnostic_get_location_text): New function.
4763 (selftest::diagnostic_c_tests): Call it.
4765 2017-11-02 David Malcolm <dmalcolm@redhat.com>
4767 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
4768 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
4769 (class selftest::test_diagnostic_context): Move to...
4770 * selftest-diagnostic.c: New file.
4771 * selftest-diagnostic.h: New file.
4773 2017-11-02 James Bowman <james.bowman@ftdichip.com>
4775 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
4776 offset range for FT32B.
4777 * config/ft32/ft32.h: option "mcompress" enables relaxation.
4778 * config/ft32/ft32.md: Add TARGET_NOPM.
4779 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
4780 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
4782 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
4784 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
4786 2017-11-02 Jeff Law <law@redhat.com>
4788 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
4789 virtual keyword on FINAL OVERRIDE members.
4791 * tree-ssa-propagate.h (ssa_propagation_engine): Group
4792 virtuals together. Add virtual destructor.
4793 (substitute_and_fold_engine): Similarly.
4795 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
4797 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
4799 2017-11-02 Richard Biener <rguenther@suse.de>
4801 PR tree-optimization/82795
4802 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
4804 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4806 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
4808 (gcc_GAS_CHECK_FEATURE): Remove.
4809 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
4810 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
4811 * configure: Regenerate.
4813 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
4815 * config/arc/arc.c (hwloop_optimize): Account for empty
4818 2017-11-02 Richard Biener <rguenther@suse.de>
4821 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
4822 Truncate ARRAY_REF index and element size.
4824 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4826 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
4828 2017-11-01 Jeff Law <law@redhat.com>
4830 * tree-ssa-ccp.c (ccp_folder): New class derived from
4831 substitute_and_fold_engine.
4832 (ccp_folder::get_value): New member function.
4833 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
4834 (ccp_fold_stmt): Remove prototype.
4835 (ccp_finalize): Call substitute_and_fold from the ccp_class.
4836 * tree-ssa-copy.c (copy_folder): New class derived from
4837 substitute_and_fold_engine.
4838 (copy_folder::get_value): Renamed from get_value.
4839 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
4840 * tree-vrp.c (vrp_folder): New class derived from
4841 substitute_and_fold_engine.
4842 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
4843 (vrp_folder::get_value): New member function.
4844 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
4845 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
4846 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
4847 provide a class interface to folder/substitute routines.
4848 (ssa_prop_fold_stmt_fn): Remove typedef.
4849 (ssa_prop_get_value_fn): Likewise.
4850 (subsitute_and_fold): Remove prototype.
4851 (replace_uses_in): Likewise.
4852 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
4853 Renamed from replace_uses_in. Call the virtual member function
4854 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
4855 (substitute_and_fold_dom_walker): Remove initialization of
4856 data member entries for calbacks. Add substitute_and_fold_engine
4857 member and initialize it.
4858 (substitute_and_fold_dom_walker::before_dom_children0: Use the
4859 member functions for get_value, replace_phi_args_in c
4860 replace_uses_in, and fold_stmt calls.
4861 (substitute_and_fold_engine::substitute_and_fold): Renamed from
4862 substitute_and_fold. Remove assert. Update ctor call.
4864 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
4865 (ssa_prop_visit_phi_fn): Likewise.
4866 (class ssa_propagation_engine): New class to provide an interface
4868 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
4870 (ssa_prop_visit_phi): Likewise.
4871 (ssa_propagation_engine::simulate_stmt): Moved into class.
4872 Call visit_phi/visit_stmt from the class rather than via
4873 file scoped static variables.
4874 (ssa_propagation_engine::simulate_block): Moved into class.
4875 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
4876 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
4877 set file scoped statics for the visit_stmt/visit_phi callbacks.
4878 * tree-complex.c (complex_propagate): New class derived from
4879 ssa_propagation_engine.
4880 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
4881 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
4882 (tree_lower_complex): Call ssa_propagate via the complex_propagate
4884 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
4885 ssa_propagation_engine.
4886 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
4887 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
4888 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
4889 * tree-ssa-copy.c (copy_prop): New class derived from
4890 ssa_propagation_engine.
4891 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
4892 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
4893 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
4894 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
4895 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
4896 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
4897 (execute_vrp): Call ssa_propagate from the vrp_prop class.
4899 2017-11-01 Jakub Jelinek <jakub@redhat.com>
4901 PR rtl-optimization/82778
4902 PR rtl-optimization/82597
4903 * compare-elim.c (struct comparison): Add in_a_setter field.
4904 (find_comparison_dom_walker::before_dom_children): Remove killed
4905 bitmap and df_simulate_find_defs call, instead walk the defs.
4906 Compute last_setter and initialize in_a_setter. Merge definitions
4907 with first initialization for a few variables.
4908 (try_validate_parallel): Use insn_invalid_p instead of
4909 recog_memoized. Return insn rather than just the pattern.
4910 (try_merge_compare): Fix up comment. Don't uselessly test if
4911 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
4913 (execute_compare_elim_after_reload): Remove df_chain_add_problem
4916 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4917 Alan Hayward <alan.hayward@arm.com>
4918 David Sherwood <david.sherwood@arm.com>
4920 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
4921 aarch64_hard_regno_nregs to get the number of registers
4924 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4925 Alan Hayward <alan.hayward@arm.com>
4926 David Sherwood <david.sherwood@arm.com>
4928 * config/aarch64/constraints.md (Upl): Rename to...
4930 * config/aarch64/aarch64.md
4931 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
4934 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4935 Alan Hayward <alan.hayward@arm.com>
4936 David Sherwood <david.sherwood@arm.com>
4938 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
4939 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
4942 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4943 Alan Hayward <alan.hayward@arm.com>
4944 David Sherwood <david.sherwood@arm.com>
4946 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
4947 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
4948 (aarch64_evpc_dup): Generate rtl direcly, rather than using
4950 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
4951 of a single element.
4952 * config/aarch64/iterators.md: Add a comment above the permute
4953 unspecs to say that they are generated directly by
4954 aarch64_expand_vec_perm_const.
4955 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
4957 2017-11-01 Nathan Sidwell <nathan@acm.org>
4959 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
4961 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4963 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
4964 and medany code models, and describe what they do.
4966 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4968 Revert accidental duplicate:
4970 * combine.c (can_change_dest_mode): Reject changes in
4971 REGMODE_NATURAL_SIZE.
4973 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
4975 PR rtl-optimization/64682
4976 PR rtl-optimization/69567
4977 PR rtl-optimization/69737
4978 PR rtl-optimization/82683
4979 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
4980 register mentioned in the note, drop the note, unless it came from I3,
4981 in which case it should go to I3 again.
4983 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4985 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
4986 and return false if not.
4987 (clear_bytes_written_by, live_bytes_read): Update accordingly.
4989 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4991 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
4992 range is known to be empty.
4994 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4995 Alan Hayward <alan.hayward@arm.com>
4996 David Sherwood <david.sherwood@arm.com>
4998 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
4999 and CONST_VECTOR_NUNITS instead of computing the number of units from
5000 the byte sizes of the vector and element.
5001 (simplify_binary_operation_1): Likewise.
5002 (simplify_const_binary_operation): Likewise.
5003 (simplify_ternary_operation): Likewise.
5005 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5006 Alan Hayward <alan.hayward@arm.com>
5007 David Sherwood <david.sherwood@arm.com>
5009 * var-tracking.c (INT_MEM_OFFSET): Replace with...
5010 (int_mem_offset): ...this new function.
5011 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
5012 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
5013 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
5016 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5017 Alan Hayward <alan.hayward@arm.com>
5018 David Sherwood <david.sherwood@arm.com>
5020 * lower-subreg.c (interesting_mode_p): New function.
5021 (compute_costs, find_decomposable_subregs, decompose_register)
5022 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
5023 (resolve_clobber, dump_choices): Use it.
5025 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5026 Alan Hayward <alan.hayward@arm.com>
5027 David Sherwood <david.sherwood@arm.com>
5029 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
5031 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5032 Alan Hayward <alan.hayward@arm.com>
5033 David Sherwood <david.sherwood@arm.com>
5035 * alias.c (find_base_value, find_base_term): Only process integer
5036 truncations. Check the precision rather than the size.
5038 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5039 Alan Hayward <alan.hayward@arm.com>
5040 David Sherwood <david.sherwood@arm.com>
5042 * machmode.h (is_narrower_int_mode): New function
5043 * optabs.c (expand_float, expand_fix): Use it.
5044 * dwarf2out.c (rotate_loc_descriptor): Likewise.
5046 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5047 Alan Hayward <alan.hayward@arm.com>
5048 David Sherwood <david.sherwood@arm.com>
5050 * rtl.h (narrower_subreg_mode): New function.
5051 * ira-color.c (update_costs_from_allocno): Use it.
5053 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5054 Alan Hayward <alan.hayward@arm.com>
5055 David Sherwood <david.sherwood@arm.com>
5057 * optabs-query.h (convert_optab_p): New function, split out from...
5058 (convert_optab_handler): ...here.
5059 (widening_optab_handler): Delete.
5060 (find_widening_optab_handler): Remove permit_non_widening parameter.
5061 (find_widening_optab_handler_and_mode): Likewise. Provide an
5062 override that operates on mode class wrappers.
5063 * optabs-query.c (widening_optab_handler): Delete.
5064 (find_widening_optab_handler_and_mode): Remove permit_non_widening
5065 parameter. Assert that the two modes are the same class and that
5066 the "from" mode is narrower than the "to" mode. Use
5067 convert_optab_handler instead of widening_optab_handler.
5068 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
5069 instead of widening_optab_handler.
5070 * expr.c (expand_expr_real_2): Update calls to
5071 find_widening_optab_handler.
5072 * optabs.c (expand_widen_pattern_expr): Likewise.
5073 (expand_binop_directly): Take the insn_code as a parameter.
5074 (expand_binop): Only call find_widening_optab_handler for
5075 conversion optabs; use optab_handler otherwise. Update calls
5076 to find_widening_optab_handler and expand_binop_directly.
5077 Use convert_optab_handler instead of widening_optab_handler.
5078 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
5079 find_widening_optab_handler and use scalar_mode rather than
5081 (convert_plusminus_to_widen): Likewise.
5083 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5084 Alan Hayward <alan.hayward@arm.com>
5085 David Sherwood <david.sherwood@arm.com>
5087 * machmode.h (fixed_size_mode): New class.
5088 * rtl.h (get_pool_mode): Return fixed_size_mode.
5089 * gengtype.c (main): Add fixed_size_mode.
5090 * target.def (get_raw_result_mode): Return a fixed_size_mode.
5091 (get_raw_arg_mode): Likewise.
5092 * doc/tm.texi: Regenerate.
5093 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
5094 * targhooks.c (default_get_reg_raw_mode): Likewise.
5095 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
5096 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
5097 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
5098 (msp430_get_raw_result_mode): Likewise.
5099 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
5100 * dbxout.c (dbxout_parms): Require fixed-size modes.
5101 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
5102 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5103 * omp-low.c (lower_oacc_reductions): Likewise.
5104 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
5105 (simplify_subreg): Update accordingly.
5106 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
5107 (force_const_mem): Update accordingly. Return NULL_RTX for modes
5108 that aren't fixed-size.
5109 (get_pool_mode): Return a fixed_size_mode.
5110 (output_constant_pool_2): Take a fixed_size_mode.
5112 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5113 Alan Hayward <alan.hayward@arm.com>
5114 David Sherwood <david.sherwood@arm.com>
5116 * doc/rtl.texi (vec_series): Document.
5117 (const): Say that the operand can be a vec_series.
5118 * rtl.def (VEC_SERIES): New rtx code.
5119 * rtl.h (const_vec_series_p_1): Declare.
5120 (const_vec_series_p): New function.
5121 * emit-rtl.h (gen_const_vec_series): Declare.
5122 (gen_vec_series): Likewise.
5123 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
5124 (gen_vec_series): Likewise.
5125 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
5126 * simplify-rtx.c (simplify_unary_operation): Handle negations
5128 (simplify_binary_operation_series): New function.
5129 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
5130 (test_vector_ops_series): New function.
5131 (test_vector_ops): Call it.
5132 * config/powerpcspe/altivec.md (altivec_lvsl): Use
5133 gen_const_vec_series.
5134 (altivec_lvsr): Likewise.
5135 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
5137 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5138 Alan Hayward <alan.hayward@arm.com>
5139 David Sherwood <david.sherwood@arm.com>
5141 * doc/rtl.texi (const): Update description of address constants.
5142 Say that vector constants are allowed too.
5143 * common.md (E, F): Use CONSTANT_P instead of checking for
5145 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
5146 checking for CONST_VECTOR.
5147 * expmed.c (make_tree): Use build_vector_from_val for a CONST
5149 * expr.c (expand_expr_real_2): Check for vector modes instead
5150 of checking for CONST_VECTOR.
5151 * rtl.h (const_vec_p): New function.
5152 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
5153 (unwrap_const_vec_duplicate): Handle them here too.
5155 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5156 David Malcolm <dmalcolm@redhat.com>
5157 Alan Hayward <alan.hayward@arm.com>
5158 David Sherwood <david.sherwood@arm.com>
5160 * rtl.h (vec_duplicate_p): New function.
5161 * selftest-rtl.c (assert_rtx_eq_at): New function.
5162 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
5163 (assert_rtx_eq_at): Declare.
5164 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
5165 * selftest-run-tests.c (selftest::run_tests): Call it.
5166 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
5167 (simplify_unary_operation_1): Recursively handle vector duplicates.
5168 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
5170 (simplify_subreg): Handle subregs of vector duplicates.
5171 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
5172 (selftest::simplify_rtx_c_tests): New functions.
5174 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5175 Alan Hayward <alan.hayward@arm.com>
5176 David Sherwood <david.sherwood@arm.com>
5178 * emit-rtl.h (gen_const_vec_duplicate): Declare.
5179 (gen_vec_duplicate): Likewise.
5180 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
5182 (gen_const_vector): ...here.
5183 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
5184 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
5185 whose elements are all equal.
5186 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
5187 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
5188 (simplify_relational_operation): Likewise.
5189 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
5191 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
5192 (aarch64_expand_vector_init): Likewise.
5193 * config/arm/arm.c (neon_vdup_constant): Likewise.
5194 (neon_expand_vector_init): Likewise.
5195 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
5196 (arm_block_set_unaligned_vect): Likewise.
5197 (arm_block_set_aligned_vect): Likewise.
5198 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
5199 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
5200 (expand_vec_perm_even_odd_pack): Likewise.
5201 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
5202 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
5203 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
5204 gen_const_vec_duplicate.
5205 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
5206 * config/mips/mips.c (mips_gen_const_int_vector): Use
5207 gen_const_vec_duplicate.
5208 (mips_expand_vector_init): Use CONST0_RTX.
5209 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
5210 (define_split): Use gen_const_vec_duplicate.
5211 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
5212 (define_split): Use gen_const_vec_duplicate.
5213 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
5214 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
5215 * config/spu/spu.c (spu_const): Likewise.
5217 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5218 Alan Hayward <alan.hayward@arm.com>
5219 David Sherwood <david.sherwood@arm.com>
5221 * combine.c (can_change_dest_mode): Reject changes in
5222 REGMODE_NATURAL_SIZE.
5224 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
5226 * configure.ac (--enable-libssp): New.
5227 (gcc_cv_libc_provides_ssp): Check for explicit setting before
5228 trying to determine target-specific default. Adjust indentation.
5229 * configure: Regenerated.
5230 * doc/install.texi (Configuration): Expand --disable-libssp
5233 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
5235 config/i386/i386.c (ix86_expand_epilogue): Correct stack
5238 2017-10-31 Martin Jambor <mjambor@suse.cz>
5241 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
5243 2017-10-31 David Malcolm <dmalcolm@redhat.com>
5245 * auto-profile.c (autofdo_source_profile::read): Use
5246 UNKNOWN_LOCATION rather than 0.
5247 * diagnostic-core.h (warning_at_rich_loc): Rename to...
5248 (warning_at): ...this overload.
5249 (warning_at_rich_loc_n): Rename to...
5250 (warning_n): ...this overload.
5251 (error_at_rich_loc): Rename to...
5252 (error_at): ...this overload.
5253 (pedwarn_at_rich_loc): Rename to...
5254 (pedwarn): ...this overload.
5255 (permerror_at_rich_loc): Rename to...
5256 (permerror): ...this overload.
5257 (inform_at_rich_loc): Rename to...
5258 (inform): ...this overload.
5259 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
5260 (diagnostic_n_impl_richloc): Rename to...
5261 (diagnostic_n_impl): ...this rich_location *-based decl.
5262 (inform_at_rich_loc): Rename to...
5263 (inform): ...this, and add an assertion.
5264 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
5265 (warning_at_rich_loc): Rename to...
5266 (warning_at): ...this, and add an assertion.
5267 (warning_at_rich_loc_n): Rename to...
5268 (warning_n): ...this, and add an assertion.
5269 (warning_n): Update location_t-based implementation for removal of
5270 location_t-based diagnostic_n_impl.
5271 (pedwarn_at_rich_loc): Rename to...
5272 (pedwarn): ...this, and add an assertion.
5273 (permerror_at_rich_loc): Rename to...
5274 (permerror): ...this, and add an assertion.
5275 (error_n): Update for removal of location_t-based diagnostic_n_impl.
5276 (error_at_rich_loc): Rename to...
5277 (error_at): ...this, and add an assertion.
5278 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
5279 (driver::do_spec_on_infiles): Likewise.
5280 * substring-locations.c (format_warning_va): Update for renaming
5281 of inform_at_rich_loc.
5283 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
5285 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
5286 _Float<N>X built-in functions so that the variant without the
5287 "__builtin_" prefix is only enabled for the GNU C and Objective C
5288 languages when they are in non-strict ANSI/ISO mode.
5289 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
5290 * target.def (floatn_builtin_p): Add a target hook to control
5291 whether _Float<N> and _Float<N>X built-in functions without the
5292 "__builtin_" prefix are enabled, and return true for C and
5293 Objective C in the default hook. Include langhooks.h in
5295 * targhooks.h (default_floatn_builtin_p): Likewise.
5296 * targhooks.c (default_floatn_builtin_p): Likewise.
5297 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
5298 floatn_builtin_p target hook.
5299 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
5301 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
5302 Eric Botcazou <ebotcazou@adacore.com>
5304 PR rtl-optimization/81803
5305 * lra-constraints.c (curr_insn_transform): Also reload the whole
5306 register for a strict subreg no wider than a word if this is for
5307 a WORD_REGISTER_OPERATIONS target.
5309 2017-10-31 Jason Merrill <jason@redhat.com>
5311 * gdbinit.in: Skip over inlines from timevar.h.
5313 2017-10-31 Martin Liska <mliska@suse.cz>
5315 * doc/gcov.texi: Document new option.
5316 * gcov.c (print_usage): Likewise print it.
5317 (process_args): Support the argument.
5318 (format_count): New function.
5319 (format_gcov): Use the function.
5321 2017-10-31 Martin Liska <mliska@suse.cz>
5323 * gcov.c (struct name_map): do not use typedef.
5324 Define operator== and operator<.
5325 (name_search): Remove.
5326 (name_sort): Remove.
5327 (main): Do not allocate names.
5328 (process_file): Add vertical space.
5329 (generate_results): Use std::find.
5330 (release_structures): Do not release memory.
5331 (find_source): Use std::find.
5333 2017-10-31 Martin Liska <mliska@suse.cz>
5335 * gcov.c (struct line_info): Remove it's typedef.
5336 (line_info::line_info): Add proper ctor.
5337 (line_info::has_block): Do not use a typedef.
5338 (struct source_info): Do not use typedef.
5339 (circuit): Likewise.
5340 (get_cycles_count): Likewise.
5341 (output_intermediate_file): Iterate via vector iterator.
5342 (add_line_counts): Use std::vector methods.
5343 (accumulate_line_counts): Likewise.
5344 (output_lines): Likewise.
5346 2017-10-31 Martin Liska <mliska@suse.cz>
5348 * gcov.c (struct source_info): Remove typedef.
5349 (source_info::source_info): Add proper ctor.
5350 (accumulate_line_counts): Use struct, not it's typedef.
5351 (output_gcov_file): Likewise.
5352 (output_lines): Likewise.
5353 (main): Do not allocate an array.
5354 (output_intermediate_file): Use size of vector container.
5355 (process_file): Resize the vector.
5356 (generate_results): Do not preallocate, use newly added vector
5358 (release_structures): Do not release sources.
5359 (find_source): Use vector methods.
5360 (add_line_counts): Do not use typedef.
5362 2017-10-31 Martin Liska <mliska@suse.cz>
5364 * doc/gcov.texi: Document that.
5365 * gcov.c (add_line_counts): Mark lines with a non-executed
5367 (output_line_beginning): Handle such lines.
5368 (output_lines): Pass new argument.
5369 (output_intermediate_file): Print it in intermediate format.
5371 2017-10-31 Martin Liska <mliska@suse.cz>
5373 * color-macros.h: New file.
5374 * diagnostic-color.c: Factor out color related to macros to
5376 * doc/gcov.texi: Document -k option.
5377 * gcov.c (INCLUDE_STRING): Include string.h.
5378 (print_usage): Add -k option.
5379 (process_args): Parse it.
5380 (pad_count_string): New function.
5381 (output_line_beginning): Likewise.
5382 (DEFAULT_LINE_START): New macro.
5383 (output_lines): Support color output.
5385 2017-10-31 Martin Liska <mliska@suse.cz>
5387 PR gcov-profile/82633
5388 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
5389 their interaction with GCOV infrastructure.
5390 * configure.ac: Add -fkeep-{inline,static}-functions to
5392 * configure: Regenerate.
5394 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
5397 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
5399 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
5402 * config/rs6000/rs6000.md (allocate_stack): Force update interval
5403 into a register if it does not fit into an immediate offset field.
5405 2017-10-31 Olivier Hainque <hainque@adacore.com>
5407 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
5409 2017-10-31 Julia Koval <julia.koval@intel.com>
5411 * config.gcc: Add gfniintrin.h.
5412 * config/i386/gfniintrin.h: New.
5413 * config/i386/i386-builtin-types.def
5414 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5415 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5416 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5417 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5418 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5419 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
5420 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
5421 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
5422 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
5423 V64QI_FTYPE_V64QI_V64QI_INT): New types.
5424 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5425 * config/i386/immintrin.h: Include gfniintrin.h.
5426 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
5428 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
5430 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
5432 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5434 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
5435 (arm_ashldi3_1bit): Remove pattern.
5436 (ashrdi3): Remove shift by 1 expansion.
5437 (arm_ashrdi3_1bit): Remove pattern.
5438 (lshrdi3): Remove shift by 1 expansion.
5439 (arm_lshrdi3_1bit): Remove pattern.
5440 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
5441 cost of ashldi3 by 1.
5442 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
5443 (<shift>di3_neon): Likewise.
5445 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5447 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
5448 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
5449 and (*aarch64_simd_mov<VQ:mode>).
5450 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
5451 pattern alternative.
5452 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
5453 attributes to match pattern alternative.
5455 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
5457 * config.gcc (powerpc*-*-*): Add emmintrin.h.
5458 * config/rs6000/emmintrin.h: New file.
5459 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
5461 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5463 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
5464 * (movdi_vfp_cortexa8): Remove pattern.
5466 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5468 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
5470 (Specific, alpha*-dec-osf5.1): Remove.
5471 (Specific, mips-sgi-irix5): Remove.
5472 (Specific, mips-sgi-irix6): Remove.
5474 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5477 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
5478 arguments to clear_bit_region_be.
5480 2017-10-30 Jim Wilson <wilson@tuliptree.org>
5482 * gimplify.c: Include memmodel.h.
5484 2017-10-30 Martin Jambor <mjambor@suse.cz>
5486 * omp-grid.c (grid_attempt_target_gridification): Also insert a
5487 condition whether loop should be executed at all.
5489 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
5491 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5492 gimple folding of vec_madd() intrinsics.
5493 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
5494 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
5495 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
5497 2017-10-30 Richard Biener <rguenther@suse.de>
5499 PR tree-optimization/82762
5501 2017-10-23 Richard Biener <rguenther@suse.de>
5503 PR tree-optimization/82129
5505 2017-08-01 Richard Biener <rguenther@suse.de>
5507 PR tree-optimization/81181
5508 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5509 (compute_antic): ... end of iteration here.
5511 2017-10-30 Joseph Myers <joseph@codesourcery.com>
5513 * doc/invoke.texi (C Dialect Options): Document -std=c17,
5514 -std=iso9899:2017 and -std=gnu17.
5515 * doc/standards.texi (C Language): Document C17 support.
5516 * doc/cpp.texi (Overview): Mention -std=c17.
5517 (Standard Predefined Macros): Document C11 and C17 values of
5518 __STDC_VERSION__. Do not refer to C99 support as incomplete.
5519 * doc/extend.texi (Inline): Do not list individual options for
5520 standards newer than C99.
5521 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
5523 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
5526 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
5528 * asan.c (asan_finish_file): Align asan globals array by shadow
5531 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5534 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
5535 (struct store_immediate_info): Add bitregion_start and bitregion_end
5537 (store_immediate_info::store_immediate_info): Add brs and bre
5538 arguments and initialize bitregion_{start,end} from those.
5539 (struct merged_store_group): Add bitregion_start, bitregion_end,
5540 align_base and mask fields. Drop unnecessary struct keyword from
5541 struct store_immediate_info. Add do_merge method.
5542 (clear_bit_region_be): Use memset instead of loop storing zeros.
5543 (merged_store_group::do_merge): New method.
5544 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
5545 stores as long as the surrounding bitregions have no gaps.
5546 (merged_store_group::merge_overlapping): Use do_merge.
5547 (merged_store_group::apply_stores): Test that bitregion_{start,end}
5548 is byte aligned, rather than requiring that start and width are
5549 byte aligned. Drop unnecessary struct keyword from
5550 struct store_immediate_info. Allocate and populate also mask array.
5551 Make start of the arrays relative to bitregion_start rather than
5552 start and size them according to bitregion_{end,start} difference.
5553 (struct imm_store_chain_info): Drop unnecessary struct keyword from
5554 struct store_immediate_info.
5555 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
5556 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
5557 struct keyword from struct store_immediate_info.
5558 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
5559 between stores as long as the surrounding bitregions have no gaps.
5561 (struct split_store): Add orig non-static data member.
5562 (split_store::split_store): Initialize orig to false.
5563 (find_constituent_stmts): Return store_immediate_info *, non-NULL
5564 if there is exactly a single original stmt. Change stmts argument
5565 to pointer from reference, if NULL, don't push anything to it. Add
5566 first argument, use it to optimize skipping over orig stmts that
5567 are known to be before bitpos already. Simplify.
5568 (split_group): Return unsigned int count how many stores are or
5569 would be needed rather than a bool. Add allow_unaligned argument.
5570 Change split_stores argument from reference to pointer, if NULL,
5571 only do a dry run computing how many stores would be produced.
5572 Rewritten algorithm to use both alignment and misalign if
5573 !allow_unaligned and handle bitfield stores with gaps.
5574 (imm_store_chain_info::output_merged_store): Set start_byte_pos
5575 from bitregion_start instead of start. Compute allow_unaligned
5576 here, if true, do 2 split_group dry runs to compute which one
5577 produces fewer stores and prefer aligned if equal. Punt if
5578 new count is bigger or equal than original before emitting any
5579 statements, rather than during that. Remove no longer needed
5580 new_ssa_names tracking. Replace num_stmts with
5581 split_stores.length (). Use 32-bit stack allocated entries
5582 in split_stores auto_vec. Try to reuse original store lhs/rhs1
5583 if possible. Handle bitfields with gaps.
5584 (pass_store_merging::execute): Ignore bitsize == 0 stores.
5585 Compute bitregion_{start,end} for the stores and construct
5586 store_immediate_info with that. Formatting fixes.
5588 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
5591 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
5592 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
5594 2017-10-29 Jim Wilson <wilson@tuliptree.org>
5596 * gimplify.c: Include tm_p.h.
5598 * common.opt (gcoff): Re-add as ignored option.
5599 (gcoff1, gcoff2, gcoff3): Likewise.
5601 * Makefile.in (OBJS): Delete sdbout.o.
5602 (GTFILES): Delete $(srcdir)/sdbout.c.
5603 * debug.h: Delete sdb_debug_hooks.
5604 * final.c: Delete sdbout.h include.
5605 (final_scan_insn): Delete SDB_DEBUG check.
5606 (rest_of_clean_state): Likewise.
5607 * output.h: Delete sdb_begin_function_line.
5610 * toplev.c: Delete sdbout.h include.
5611 (process_options): Delete SDB_DEBUG check.
5612 * tree-core.h (tree_type_common): Delete pointer field of
5614 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
5615 TYPE_SYMTAB_POINTER.
5616 * tree.h (TYPE_SYMTAB_POINTER): Delete.
5617 (TYPE_SYMTAB_IS_POINTER): Delete.
5618 (TYPE_SYMTAB_IS_DIE): Renumber.
5619 * xcoffout.c: Refer to former sdbout.c file.
5620 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
5622 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
5623 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
5624 (Debugging Options): Delete -gcoff.
5625 (-gstabs): Delete SDB reference.
5627 (-gcoff@var{level}): Delete.
5628 * doc/passes.texi (Debugging information output): Delete SDB and
5629 sdbout.c references.
5630 * doc/tm.texi: Regenerate.
5631 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
5632 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
5634 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
5635 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
5636 (SDB_DEBUGGING_INFO): Delete.
5637 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
5638 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
5639 * target.def (output_source_filename): Delete COFF reference.
5641 * common.opt (gcoff): Delete.
5642 (gxcoff+): Update Negative chain.
5643 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
5645 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
5646 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
5647 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
5649 (expand_function_start): Change sdb reference to past tense.
5650 (expand_function_end): Change sdb reference to past tense.
5651 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
5652 * opts.c (debug_type_names): Delete coff entry.
5653 (common_handle_option): Delete OPT_gcoff case.
5654 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
5656 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
5657 * config/cris/cris.h: Delete SDB reference in comment.
5658 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
5659 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
5660 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
5661 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
5663 (ix86_expand_prologue): Likewise.
5664 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
5665 * config/ia64/ia64.h: Likewise.
5666 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
5667 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
5669 * config/mmix/mmix.h: Likewise.
5670 * config/nds32/nds32.c: Likewise.
5671 * config/stormy/storym16.h: Likewise.
5672 * config/visium/visium.h: Likewise.
5673 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
5675 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
5677 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
5678 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
5679 FRAME_POINTER_REGNUM point at high end of local var area.
5681 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
5683 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
5684 Move comment around. Do not reset best_edge for a copiable
5685 destination if the copy would cause a partition change.
5686 (better_edge_p): Remove redundant check.
5688 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
5690 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
5692 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5694 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
5695 for math functions that have _Float<N> and _Float<N>X variants.
5696 (mathfn_built_in_2): Add support for math functions that have
5697 _Float<N> and _Float<N>X variants.
5698 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
5699 (expand_builtin_mathfn_ternary): Add support for fma with
5700 _Float<N> and _Float<N>X variants.
5701 (expand_builtin): Likewise.
5702 (fold_builtin_3): Likewise.
5703 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
5704 create math function _Float<N> and _Float<N>X variants as external
5706 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
5707 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
5708 the __builtin_ prefix and if not strict ansi, without the prefix.
5709 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
5710 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
5711 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
5712 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
5713 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
5714 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
5715 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
5716 function signatures for fma _Float<N> and _Float<N>X variants.
5717 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
5718 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
5719 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
5720 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
5721 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
5722 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
5723 * gencfn-macros.c (print_case_cfn): Add support for math functions
5724 that have _Float<N> and _Float<N>X variants.
5725 (print_define_operator_list): Likewise.
5726 (fltfn_suffixes): Likewise.
5728 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
5729 for math functions that have _Float<N> and _Float<N>X variants.
5730 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
5731 and _Float<N>X variants.
5732 (COPYSIGN): Likewise.
5735 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
5736 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
5738 (integer_valued_read_call_p): Likewise.
5739 * fold-const-call.c (fold_const_call_ss): Likewise.
5740 (fold_const_call_sss): Add support for copysign, fmin, and fmax
5741 _Float<N> and _Float<N>X variants.
5742 (fold_const_call_ssss): Add support for fma _Float<N> and
5743 _Float<N>X variants.
5744 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
5745 support for copysign and fma _Float<N> and _Float<N>X variants.
5746 (backprop::process_builtin_call_use): Likewise.
5747 * tree-call-cdce.c (can_test_argument_range); Add support for
5748 sqrt _Float<N> and _Float<N>X variants.
5749 (edom_only_function): Likewise.
5750 (get_no_error_domain): Likewise.
5751 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
5752 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
5753 copysign _Float<N> and _Float<N>X variants.
5754 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
5755 handled by machine independent code.
5756 (FMAF128): Likewise.
5757 * doc/cpp.texi (Common Predefined Macros): Document defining
5758 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
5759 fma _Float<N> and _Float<N>X variants.
5761 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
5764 * config/i386/i386-modes.def (CCFPU): Remove definition.
5765 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
5766 (ix86_cc_modes_compatible): Ditto.
5767 (ix86_expand_carry_flag_compare): Ditto.
5768 (ix86_expand_int_movcc): Ditto.
5769 (ix86_expand_int_addcc): Ditto.
5770 (ix86_reverse_condition): Ditto.
5771 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
5772 Return true/false for unordered/ordered fp comparisons.
5773 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
5774 (ix86_prepare_fp_compare_args): Update for rename.
5775 (ix86_expand_fp_compare): Update for rename. Generate unordered
5776 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
5777 (ix86_expand_sse_compare_and_jump): Ditto.
5778 * config/i386/predicates.md (fcmov_comparison_operator):
5779 Remove CCFPU mode handling.
5780 (ix86_comparison_operator): Ditto.
5781 (ix86_carry_flag_operator): Ditto.
5782 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
5783 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
5784 (*cmpu<mode>_cc_i387): Ditto.
5785 (FPCMP): Remove mode iterator.
5786 (unord): Remove mode attribute.
5787 (unord_subst): New define_subst transformation
5788 (unord): New define_subst attribute.
5790 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
5791 (*cmpi<unord>xf_i387): Ditto.
5792 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
5793 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
5794 using unord_subst transformation.
5795 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
5796 (round_saeonly): Also handle CCFP mode.
5797 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
5798 Remove UNSPEC_SAHF unspec handling.
5800 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5802 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
5804 2017-10-27 Jeff Law <law@redhat.com>
5806 * gimple-ssa-sprintf.c: Include domwalk.h.
5807 (class sprintf_dom_walker): New class, derived from dom_walker.
5808 (sprintf_dom_walker::before_dom_children): New function.
5809 (struct call_info): Moved into sprintf_dom_walker class
5810 (compute_formath_length, handle_gimple_call): Likewise.
5811 (sprintf_length::execute): Call the dominator walker rather
5812 than walking the statements.
5814 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
5815 gimple statement locations.
5816 (check_array_bounds): Corresponding changes. Get the statement's
5817 location directly from wi->stmt.
5819 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
5822 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
5824 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5826 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
5827 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
5829 2017-10-27 Jakub Jelinek <jakub@redhat.com>
5832 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
5833 * config/i386/i386.c (maybe_get_pool_constant): Removed.
5834 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
5835 maybe_get_pool_constant.
5836 * config/i386/predicates.md (zero_extended_scalar_load_operand):
5839 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5841 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
5842 2.26 caveat. Update gas and gld versions.
5843 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
5846 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5848 * cgraph.h (set_malloc_flag): Declare.
5849 * cgraph.c (set_malloc_flag_1): New function.
5850 (set_malloc_flag): Likewise.
5851 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
5852 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
5854 (read_ipa_call_summary): Add support for reading is_return_callee.
5855 (write_ipa_call_summary): Stream is_return_callee.
5856 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
5857 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
5858 ipa-prop.h, ipa-fnsummary.h.
5859 (pure_const_names): Change to static.
5860 (malloc_state_e): Define.
5861 (malloc_state_names): Define.
5862 (funct_state_d): Add field malloc_state.
5863 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
5864 (check_retval_uses): New function.
5865 (malloc_candidate_p): Likewise.
5866 (analyze_function): Add support for malloc attribute.
5867 (pure_const_write_summary): Stream malloc_state.
5868 (pure_const_read_summary): Add support for reading malloc_state.
5869 (dump_malloc_lattice): New function.
5870 (propagate_malloc): New function.
5871 (warn_function_malloc): New function.
5872 (ipa_pure_const::execute): Call propagate_malloc and
5873 ipa_free_fn_summary.
5874 (pass_local_pure_const::execute): Add support for malloc attribute.
5875 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
5876 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
5878 2017-10-27 Martin Liska <mliska@suse.cz>
5880 PR gcov-profile/82457
5881 * doc/invoke.texi: Document that one needs a non-strict ISO mode
5882 for fork-like functions to be properly instrumented.
5884 2017-10-27 Richard Biener <rguenther@suse.de>
5887 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
5888 info when we redirected EH.
5890 2017-10-26 Michael Collison <michael.collison@arm.com>
5892 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
5894 (<optab>_trunchf<GPI:mode>2: New pattern.
5895 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
5896 * config/aarch64/iterators.md (wv): New mode attribute.
5897 (vf, VF): New mode attributes.
5898 (vgp, VGP): New mode attributes.
5899 (s): Update attribute with SImode and DImode prefixes.
5901 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
5903 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
5904 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
5905 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
5906 (nios2_option_overide): Initialize it. Don't allow R0-relative
5907 addressing with PIC.
5908 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
5909 (nios2_symbolic_constant_p): Likewise.
5910 (nios2_legitimate_address_p): Likewise.
5911 (nios2_r0rel_section_name_p): New.
5912 (nios2_symbol_ref_in_r0rel_data_p): New.
5913 (nios2_emit_move_sequence): Handle r0rel_constant_p.
5914 (r0rel_constant_p): New.
5915 (nios2_print_operand_address): Handle r0rel_constant_p.
5916 (nios2_cdx_narrow_form_p): Likewise.
5917 * config/nios2/nios2.opt (mr0rel-sec=): New option.
5918 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
5919 (Nios II Options): Document -mr0rel-sec.
5921 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
5923 * config/nios2/nios2.c: Include xregex.h.
5924 (nios2_gprel_sec_regex): New.
5925 (nios2_option_overide): Initialize it. Don't allow GP-relative
5926 addressing with PIC.
5927 (nios2_small_section_name_p): Check for regex match.
5928 * config/nios2/nios2.opt (mgprel-sec=): New option.
5929 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
5930 (Nios II Options): Document -mgprel-sec.
5932 2017-10-26 Jim Wilson <wilson@tuliptree.org>
5934 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
5936 2017-10-26 Tom de Vries <tom@codesourcery.com>
5938 PR tree-optimization/82707
5939 * gimple.c (gimple_copy): Fix unsharing of
5940 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
5942 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
5944 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
5945 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
5946 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
5947 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
5948 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
5949 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
5950 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
5951 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
5952 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
5953 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
5954 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
5955 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
5956 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
5957 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
5958 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
5959 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
5960 _mm512_mask_cmpunord_ps_mask): New intrinsics.
5962 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5964 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
5966 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
5967 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
5968 warning to rs6000.c. Remove the Undocumented flag, since it has
5970 (-mabi=ibmlongdouble): Likewise.
5971 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
5972 already set, set the default format for long double.
5973 (rs6000_debug_reg_global): Print whether long double is IBM or
5975 (rs6000_option_override_internal): Rework setting long double
5976 format. Only warn if the user is changing the long double default
5977 and they did not use -Wno-psabi.
5978 * doc/invoke.texi (PowerPC options): Update the documentation for
5979 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
5981 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5982 Alan Hayward <alan.hayward@arm.com>
5983 David Sherwood <david.sherwood@arm.com>
5985 * rtl.h (wider_subreg_mode): New function.
5986 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
5987 rather than an unsigned int *.
5988 * ira-color.c (regno_max_ref_width): Replace with...
5989 (regno_max_ref_mode): ...this new variable.
5990 (coalesced_pseudo_reg_slot_compare): Update accordingly.
5991 Use wider_subreg_mode.
5992 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
5993 rather than an unsigned int *.
5994 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
5995 (process_alt_operands): Likewise.
5996 (invariant_p): Likewise.
5997 * lra-spills.c (assign_mem_slot): Likewise.
5998 (add_pseudo_to_slot): Likewise.
5999 * lra.c (collect_non_operand_hard_regs): Likewise.
6000 (add_regs_to_insn_regno_info): Likewise.
6001 * reload1.c (regno_max_ref_width): Replace with...
6002 (regno_max_ref_mode): ...this new variable.
6003 (reload): Update accordingly. Update call to
6004 ira_sort_regnos_for_alter_reg.
6005 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
6006 (init_eliminable_invariants): Update to use regno_max_ref_mode.
6007 (scan_paradoxical_subregs): Likewise.
6009 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6011 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
6012 (aarch64_frame): Add emit_frame_chain boolean.
6013 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6014 Move eh_return case to aarch64_layout_frame.
6015 (aarch64_layout_frame): Initialize emit_frame_chain.
6016 (aarch64_expand_prologue): Use emit_frame_chain.
6018 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6020 * config/aarch64/aarch64.c (aarch64_layout_frame):
6021 Ensure LR is always stored at the bottom of the callee-saves.
6022 Remove rarely used frame layout which saves callee-saves at top of
6023 frame, so the store of LR can be used as a valid probe in all cases.
6025 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6027 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
6028 Improve unaligned TImode/TFmode base/offset split.
6030 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6031 Alan Hayward <alan.hayward@arm.com>
6032 David Sherwood <david.sherwood@arm.com>
6034 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
6035 * combine.c (find_single_use_1): Likewise.
6036 (expand_field_assignment): Likewise.
6037 (move_deaths): Likewise.
6038 * lra-constraints.c (simplify_operand_subreg): Likewise.
6039 (curr_insn_transform): Likewise.
6040 * lra.c (collect_non_operand_hard_regs): Likewise.
6041 (add_regs_to_insn_regno_info): Likewise.
6042 * rtlanal.c (reg_referenced_p): Likewise.
6043 (covers_regno_no_parallel_p): Likewise.
6045 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6047 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
6048 Don't print any bits outside the precision of the value.
6049 * wide-int.cc (test_printing): Add some new tests.
6051 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6053 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
6054 supports -xbrace_comment option.
6055 * configure: Regenerate.
6056 * config.in: Regenerate.
6057 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
6058 (ASM_CPU_SPEC): Use it.
6060 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6062 * target.def (static_rtx_alignment): New hook.
6063 * targhooks.h (default_static_rtx_alignment): Declare.
6064 * targhooks.c (default_static_rtx_alignment): New function.
6065 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
6066 * doc/tm.texi: Regenerate.
6067 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
6068 instead of targetm.constant_alignment. Remove call to
6070 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6071 (cris_preferred_mininum_alignment): New function, split out from...
6072 (cris_constant_alignment): ...here.
6073 (cris_static_rtx_alignment): New function.
6074 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
6076 (ix86_constant_alignment): ...here.
6077 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6078 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6079 (mmix_static_rtx_alignment): New function.
6080 * config/spu/spu.c (spu_static_rtx_alignment): New function.
6081 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6083 2017-10-26 Tamar Christina <tamar.christina@arm.com>
6086 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
6087 Add flag_trapping_math and flag_fp_int_builtin_inexact.
6089 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
6091 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
6092 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
6093 mark as a sign-extending load.
6094 (local_pic_load_u): Define.
6096 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
6099 * fold-const.c (operand_equal_for_comparison_p): Also return true
6100 if ARG0 is a simple variant of ARG1 with narrower precision.
6101 (fold_ternary_loc): Always pass unstripped operands to the predicate.
6103 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
6105 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
6107 * i386.h (processor_costs): Add gather_static, gather_per_elt,
6108 scatter_static, scatter_per_elt.
6109 * x86-tune-costs.h: Add new cost entries.
6111 2017-10-25 Richard Biener <rguenther@suse.de>
6113 * tree-ssa-sccvn.h (vn_eliminate): Declare.
6114 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
6115 class pass_fre): Move to ...
6116 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
6117 class pass_fre): ... here and adjust for statistics.
6119 2017-10-25 Jakub Jelinek <jakub@redhat.com>
6122 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
6123 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
6124 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
6125 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
6128 2017-10-25 Richard Biener <rguenther@suse.de>
6130 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
6131 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
6132 eliminate_push_avail, eliminate_insert): Move inside...
6133 (class eliminate_dom_walker): ... this class in preparation
6135 (fini_eliminate): Remove by merging with ...
6136 (eliminate): ... this function. Adjust for class changes.
6137 (pass_pre::execute): Remove fini_eliminate call.
6138 (pass_fre::execute): Likewise.
6140 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6143 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
6144 (VPERMI2, VPERMI2I): New mode iterators.
6145 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
6146 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
6148 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
6149 mode iterator. Remove 3 old define_insn patterns.
6150 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
6151 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
6152 VPERMI2 mode iterator, remove the other two expanders.
6153 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
6154 to use VPERMI2 mode iterator, add another alternative for vpermi2*
6155 instructions, remove the other two patterns.
6156 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
6157 mode iterator, remove the other two patterns.
6158 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
6159 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
6160 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
6161 and adjust argument order accordingly.
6162 (ix86_expand_vec_perm): Adjust caller.
6163 (expand_vec_perm_1): Likewise.
6164 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
6165 (expand_vec_perm_vpermt2_vpshub2): ... this.
6166 (ix86_expand_vec_perm_const_1): Adjust caller.
6167 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
6170 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
6171 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
6172 (vec_shl_<mode>): Remove unused expander.
6173 (avx512bw_<shift_insn><mode>3): New define_insn.
6174 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
6175 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
6177 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
6180 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
6183 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6185 PR rtl-optimization/82396
6186 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
6187 (autopref_multipass_init): Simplify initialization.
6188 (autopref_rank_data): Simplify sort order.
6189 * gcc/sched-int.h (autopref_multipass_data_): Remove
6190 multi_mem_insn_p, min_offset and max_offset.
6192 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6195 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6196 Check special value of flag_omit_frame_pointer.
6197 (aarch64_can_eliminate): Likewise.
6198 (aarch64_override_options_after_change_1): Simplify handling of
6199 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
6201 2017-10-24 Richard Biener <rguenther@suse.de>
6203 PR tree-optimization/82697
6204 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
6205 zero for conditional load and unconditional store.
6207 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6209 * doc/install.texi: Document bootstrap-cet.
6211 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6214 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
6215 ENDBR instruction at function entrance if function is only
6218 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6221 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
6222 patterns to better describe from which operation the CF is computed.
6223 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
6224 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
6225 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
6226 is 0, use _0 suffixed expanders instead of emitting a comparison
6229 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
6231 * config/i386/i386.md(*movsf_internal, *movdf_internal):
6232 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
6234 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
6237 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
6238 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
6239 * loop-iv.c (iv_get_reaching_def): Likewise.
6240 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
6241 variable is promoted and the partition contains undefined values.
6243 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6245 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
6247 (nios2_address_cost): Define.
6248 (nios2_legitimize_address): Recognize (exp + constant) directly.
6249 (TARGET_ADDRESS_COST): Define.
6251 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6253 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
6254 (nios2_symbolic_memory_operand_p): Declare.
6255 (nios2_split_large_constant): Declare.
6256 (nios2_split_symbolic_memory_operand): Declare.
6257 * config/nios2/nios2.c: Adjust includes.
6258 (nios2_symbolic_constant_allowed): New.
6259 (nios2_symbolic_constant_p): New.
6260 (nios2_plus_symbolic_constant_p): New.
6261 (nios2_valid_addr_expr_p): Recognize addresses involving
6263 (nios2_legitimate_address_p): Likewise, also LO_SUM.
6264 (nios2_symbolic_memory_operand_p): New.
6265 (nios2_large_constant_p): New.
6266 (nios2_split_large_constant): New.
6267 (nios2_split_plus_large_constant): New.
6268 (nios2_split_symbolic_memory_operand): New.
6269 (nios2_legitimize_address): Code refactoring. Handle addresses
6270 involving symbolic constants.
6271 (nios2_emit_move_sequence): Likewise.
6272 (nios2_print_operand): Improve error output.
6273 (nios2_print_operand_address): Handle LO_SUM.
6274 (nios2_cdx_narrow_form_p): Likewise.
6275 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
6276 operands involving symbolic constants.
6277 (movhi_internal, movsi_internal): Likewise.
6278 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
6279 (extendhisi2, extendqi<mode>2): Likewise.
6281 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6283 * tree-pass.h (PROP_rtl_split_insns): Define.
6284 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
6286 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6288 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
6290 2017-10-23 Jakub Jelinek <jakub@redhat.com>
6293 * target.def (const_not_ok_for_debug_p): Default to
6294 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
6295 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
6296 * targhooks.c (default_const_not_ok_for_debug_p): New function.
6297 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
6298 which targetm.const_not_ok_for_debug_p returned true.
6299 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
6301 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
6303 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
6304 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
6305 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
6307 (ix86_const_not_ok_for_debug_p): New function.
6308 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
6309 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
6311 2017-10-23 David Malcolm <dmalcolm@redhat.com>
6314 * system.h: Conditionally include "unique-ptr.h" if
6315 INCLUDE_UNIQUE_PTR is defined.
6316 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
6317 of defining INCLUDE_UNIQUE_PTR before including "system.h".
6319 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
6321 * config/rl78/rl78.md: New define_expand "subdi3".
6323 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
6326 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
6327 DF_REF_INSN if DF_REF_INSN_INFO is false.
6329 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
6331 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
6332 xmm_move instead of sse_move.
6333 (sse_store_index): New function.
6334 (ix86_register_move_cost): Be more sensible about mismatch stall;
6335 model AVX moves correctly; make difference between sse->integer and
6337 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
6338 moves; make difference between SSE and AVX.
6339 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
6340 and zmm_move. Increase size of sse load and store tables;
6341 add unaligned load and store tables; add ssemmx_to_integer.
6342 * x86-tune-costs.h: Update all entries according to real
6343 move latencies from Agner Fog's manual and chip documentation.
6345 2017-10-23 Jakub Jelinek <jakub@redhat.com>
6348 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
6349 * config/i386/constraints.md (Wf): New constraint.
6350 * config/i386/i386.md (UNSPEC_SBB): New unspec.
6351 (cmp<dwi>_doubleword): Removed.
6352 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
6353 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
6354 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
6355 expand with cmp<dwi>_doubleword. For LTU and GEU use
6356 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
6358 * common.opt (gcolumn-info): Enable by default.
6359 * doc/invoke.texi (gcolumn-info): Document new default.
6361 2017-10-23 Richard Biener <rguenther@suse.de>
6363 PR tree-optimization/82672
6364 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
6365 Fold the stmt if we propagated into it.
6367 2017-10-23 Richard Biener <rguenther@suse.de>
6369 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
6370 (bitmap_remove_expr_from_set): ... this. All callers call this
6371 for non-constant values.
6372 (bitmap_set_subtract): Rename to...
6373 (bitmap_set_subtract_expressions): ... this. Adjust and
6375 (bitmap_set_contains_value): Remove superfluous check.
6376 (bitmap_set_replace_value): Inline into single caller ...
6377 (bitmap_value_replace_in_set): ... here and simplify.
6378 (dependent_clean): Merge into ...
6379 (clean): ... this using an overload. Adjust.
6380 (prune_clobbered_mems): Adjust.
6381 (compute_antic_aux): Likewise.
6382 (compute_partial_antic_aux): Likewise.
6384 2017-10-23 Richard Biener <rguenther@suse.de>
6386 PR tree-optimization/82129
6388 2017-08-01 Richard Biener <rguenther@suse.de>
6390 PR tree-optimization/81181
6391 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6392 (compute_antic): ... end of iteration here.
6394 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6396 * target.def (starting_frame_offset): New hook.
6397 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
6398 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
6399 * doc/tm.texi.in: Regenerate.
6400 * hooks.h (hook_hwi_void_0): Declare.
6401 * hooks.c (hook_hwi_void_0): New function.
6402 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
6403 STARTING_FRAME_OFFSET.
6404 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
6405 * reload1.c (reload): Likewise.
6406 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
6407 instead of STARTING_FRAME_OFFSET.
6408 * function.c (try_fit_stack_local): Likewise.
6409 (assign_stack_local_1): Likewise
6410 (instantiate_virtual_regs): Likewise.
6411 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
6412 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
6413 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
6414 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
6415 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
6416 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
6417 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
6418 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
6419 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
6420 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
6421 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
6422 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
6423 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
6424 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
6425 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
6426 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
6427 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
6428 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
6429 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
6430 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
6431 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
6432 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
6433 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
6434 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
6435 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
6436 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
6437 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
6438 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
6439 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
6440 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
6441 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
6442 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
6443 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
6444 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
6445 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
6446 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
6447 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
6448 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
6449 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
6450 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
6451 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
6452 * config/avr/avr.c (avr_starting_frame_offset): Make static and
6453 return a HOST_WIDE_INT.
6454 (avr_builtin_setjmp_frame_value): Use it instead of
6455 STARTING_FRAME_OFFSET.
6456 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6457 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
6458 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
6460 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6461 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
6462 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
6463 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6464 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
6465 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
6466 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6467 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
6468 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
6469 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6470 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
6471 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
6473 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6474 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
6475 * config/mips/mips.c (mips_compute_frame_info): Refer to
6476 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6477 (mips_starting_frame_offset): New function.
6478 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6479 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
6480 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
6481 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
6482 and return a HOST_WIDE_INT.
6483 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6484 (mmix_initial_elimination_offset): Refer to
6485 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6486 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
6487 * config/pa/pa.c (pa_starting_frame_offset): New function.
6488 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
6489 (pa_expand_prologue): Likewise.
6490 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6491 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
6492 !FRAME_GROWS_DOWNWARD handling to...
6493 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6494 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
6495 !FRAME_GROWS_DOWNWARD handling to...
6496 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6497 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
6498 !FRAME_GROWS_DOWNWARD handling to...
6499 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6500 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
6502 (rs6000_starting_frame_offset): New function.
6503 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
6504 !FRAME_GROWS_DOWNWARD handling to...
6505 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6506 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
6507 !FRAME_GROWS_DOWNWARD handling to...
6508 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6509 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
6510 !FRAME_GROWS_DOWNWARD handling to...
6511 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6512 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
6513 (rs6000_starting_frame_offset): New function.
6514 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
6515 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
6516 * config/vax/vax.c (vax_starting_frame_offset): New function.
6517 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
6518 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6519 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
6520 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
6521 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6522 * system.h (STARTING_FRAME_OFFSET): Poison.
6524 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6526 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
6527 SCALAR_TYPE_MODE instead of TYPE_MODE.
6529 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6530 Alan Hayward <alan.hayward@arm.com>
6531 David Sherwood <david.sherwood@arm.com>
6533 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
6535 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6536 Alan Hayward <alan.hayward@arm.com>
6537 David Sherwood <david.sherwood@arm.com>
6539 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
6541 2017-10-23 Richard Biener <rguenther@suse.de>
6543 PR tree-optimization/82129
6544 * tree-ssa-pre.c (bitmap_set_and): Remove.
6545 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
6546 canonicalizing expressions in the set to those with lowest
6547 ID rather than taking that from the first edge.
6549 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6551 * combine.c (rtx_equal_for_field_assignment_p): Use
6552 byte_lowpart_offset.
6554 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6555 Alan Hayward <alan.hayward@arm.com>
6556 David Sherwood <david.sherwood@arm.com>
6558 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
6559 to a SUBREG_PROMOTED_VAR.
6561 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6562 Alan Hayward <alan.hayward@arm.com>
6563 David Sherwood <david.sherwood@arm.com>
6565 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
6566 (expand_debug_source_expr): Likewise.
6567 * combine.c (combine_simplify_rtx): Likewise.
6568 * cse.c (fold_rtx): Likewise.
6569 * optabs.c (expand_float): Likewise.
6570 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6571 (simplify_binary_operation_1): Likewise.
6573 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6574 Alan Hayward <alan.hayward@arm.com>
6575 David Sherwood <david.sherwood@arm.com>
6577 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
6578 (record_promoted_value): Likewise.
6579 * expr.c (expand_expr_real_2): Likewise.
6580 * ree.c (update_reg_equal_equiv_notes): Likewise.
6581 (combine_set_extension): Likewise.
6582 * rtlanal.c (low_bitmask_len): Likewise.
6583 * simplify-rtx.c (neg_const_int): Likewise.
6584 (simplify_binary_operation_1): Likewise.
6586 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6587 Alan Hayward <alan.hayward@arm.com>
6588 David Sherwood <david.sherwood@arm.com>
6590 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
6591 * regcprop.c (maybe_mode_change): Likewise.
6592 * reload1.c (alter_reg): Likewise.
6594 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6596 * inchash.h (inchash::hash::add_wide_int): New function.
6597 * lto-streamer-out.c (hash_tree): Use it.
6599 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6601 * inchash.h (inchash::hash::add_wide_int): Rename to...
6602 (inchash::hash::add_hwi): ...this.
6603 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
6604 (polymorphic_call_target_hasher::hash): Likewise.
6605 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
6606 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
6607 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
6608 * lto-streamer-out.c (hash_tree): Likewise.
6609 * optc-save-gen.awk: Likewise.
6610 * tree.c (add_expr): Likewise.
6612 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6615 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
6616 for ordered inequality comparisons even with TARGET_IEEE_FP.
6618 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6621 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
6622 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
6623 Expand with cmp<dwi>_doubleword.
6625 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6627 * extend.texi: Add x86 specific to 'nocf_check' attribute.
6628 List CET intrinsics.
6629 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
6630 specific to -fcf-protection option.
6632 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6634 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
6635 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
6636 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
6637 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
6638 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
6639 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
6640 (extra_objs): Add cet.o for Linux/x86 targets.
6641 (tmake_file): Add i386/t-cet for Linux/x86 targets.
6642 * config/i386/cet.c: New file.
6643 * config/i386/cetintrin.h: Likewise.
6644 * config/i386/t-cet: Likewise.
6645 * config/i386/cpuid.h (bit_SHSTK): New.
6646 (bit_IBT): Likewise.
6647 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
6648 pass IBT and SHSTK bits.
6649 * config/i386/i386-builtin-types.def
6650 (VOID_FTYPE_UNSIGNED_PVOID): New.
6651 (VOID_FTYPE_UINT64_PVOID): Likewise.
6652 * config/i386/i386-builtin.def: Add CET intrinsics.
6653 * config/i386/i386-c.c (ix86_target_macros_internal): Add
6654 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
6655 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
6656 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
6658 * config/i386/i386.c (rest_of_insert_endbranch): New.
6659 (pass_data_insert_endbranch): Likewise.
6660 (pass_insert_endbranch): Likewise.
6661 (make_pass_insert_endbranch): Likewise.
6662 (ix86_notrack_prefixed_insn_p): Likewise.
6663 (ix86_target_string): Add -mibt, -mshstk flags.
6664 (ix86_option_override_internal): Add flag_cf_protection
6666 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
6667 (ix86_print_operand): Add 'notrack' prefix output.
6668 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
6669 (ix86_expand_builtin): Expand CET intrinsics.
6670 (x86_output_mi_thunk): Add 'endbranch' instruction.
6671 * config/i386/i386.h (TARGET_IBT): New.
6672 (TARGET_IBT_P): Likewise.
6673 (TARGET_SHSTK): Likewise.
6674 (TARGET_SHSTK_P): Likewise.
6675 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
6676 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
6677 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
6678 (builtin_setjmp_setup): New pattern.
6679 (builtin_longjmp): Likewise.
6680 (rdssp<mode>): Likewise.
6681 (incssp<mode>): Likewise.
6682 (saveprevssp): Likewise.
6683 (rstorssp): Likewise.
6684 (wrss<mode>): Likewise.
6685 (wruss<mode>): Likewise.
6686 (setssbsy): Likewise.
6687 (clrssbsy): Likewise.
6688 (nop_endbr): Likewise.
6689 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
6691 * config/i386/immintrin.h: Include <cetintrin.h>.
6692 * config/i386/linux-common.h
6693 (file_end_indicate_exec_stack_and_cet): New prototype.
6694 (TARGET_ASM_FILE_END): New.
6696 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
6698 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
6699 latencies instead of having separate table; make difference between
6700 integer and float costs.
6701 * i386.h (processor_costs): Remove scalar_stmt_cost,
6702 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
6703 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
6705 * x86-tune-costs.h: Remove entries which has been removed in
6706 procesor_costs from all tables; make cond_taken_branch_cost
6707 and cond_not_taken_branch_cost COST_N_INSNS based.
6709 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
6711 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
6713 2017-10-20 Jakub Jelinek <jakub@redhat.com>
6715 * config/i386/i386.md (isa): Remove fma_avx512f.
6716 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
6717 <avx512>_fmadd_<mode>_mask3<round_name>,
6718 <avx512>_fmsub_<mode>_mask<round_name>,
6719 <avx512>_fmsub_<mode>_mask3<round_name>,
6720 <avx512>_fnmadd_<mode>_mask<round_name>,
6721 <avx512>_fnmadd_<mode>_mask3<round_name>,
6722 <avx512>_fnmsub_<mode>_mask<round_name>,
6723 <avx512>_fnmsub_<mode>_mask3<round_name>,
6724 <avx512>_fmaddsub_<mode>_mask<round_name>,
6725 <avx512>_fmaddsub_<mode>_mask3<round_name>,
6726 <avx512>_fmsubadd_<mode>_mask<round_name>,
6727 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
6728 (*vec_widen_umult_even_v16si<mask_name>,
6729 *vec_widen_smult_even_v16si<mask_name>): Likewise.
6730 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
6732 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6734 * extend.texi: Add 'nocf_check' documentation.
6735 * gimple.texi: Add second parameter to
6736 gimple_build_call_from_tree.
6737 * invoke.texi: Add -fcf-protection documentation.
6738 * rtl.texi: Add REG_CALL_NOTRACK documenation.
6740 2017-10-20 Richard Biener <rguenther@suse.de>
6742 PR tree-optimization/82473
6743 * tree-vect-loop.c (vectorizable_reduction): Properly get at
6744 the largest input type.
6746 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6748 * c-attribs.c (handle_nocf_check_attribute): New function.
6749 (c_common_attribute_table): Add 'nocf_check' handling.
6750 * gimple-parser.c: Add second argument NULL to
6751 gimple_build_call_from_tree.
6752 * attrib.c (comp_type_attributes): Check nocf_check attribute.
6753 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
6755 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
6756 * common.opt: Add fcf-protection flag.
6757 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
6758 * flag-types.h: Add enum cf_protection_level.
6759 * gimple.c (gimple_build_call_from_tree): Add second parameter.
6760 Add 'nocf_check' attribute propagation to gimple call.
6761 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
6762 (gimple_build_call_from_tree): Update prototype.
6763 (gimple_call_nocf_check_p): New function.
6764 (gimple_call_set_nocf_check): Likewise.
6765 * gimplify.c: Add second argument to gimple_build_call_from_tree.
6766 * ipa-icf.c: Add nocf_check attribute in statement hash.
6767 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
6768 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
6769 * toplev.c (process_options): Add flag_cf_protection handling.
6771 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6773 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
6775 2017-10-20 Richard Biener <rguenther@suse.de>
6777 PR tree-optimization/82603
6778 * tree-if-conv.c (predicate_mem_writes): Make sure to only
6779 remove false predicated stores.
6781 2017-10-20 Richard Biener <rguenther@suse.de>
6783 * graphite-isl-ast-to-gimple.c
6784 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
6785 Remove return value and simplify, dump copied stmt after lhs
6787 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
6788 Reduce dump verbosity.
6789 (gsi_insert_earliest): Likewise.
6790 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
6791 * graphite.c (print_global_statistics): Adjust dumping.
6792 (print_graphite_scop_statistics): Likewise.
6793 (print_graphite_statistics): Do not dump loops here.
6794 (graphite_transform_loops): But here.
6796 2017-10-20 Nicolas Roche <roche@adacore.com>
6798 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
6799 * configure: Regenerate.
6801 2017-10-20 Jakub Jelinek <jakub@redhat.com>
6804 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
6805 functions when optimizing replace GIMPLE_RETURN stmts with
6806 calls to __builtin_unreachable ().
6809 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
6810 for -fsanitize=thread link of executables.
6811 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
6812 link of executables.
6815 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
6817 (<shift_insn><mode>3<mask_name>): Change the last of the 3
6818 define_insns for logical vector shifts to use VI248_AVX512BW
6819 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
6820 condition, useless isa and prefix attributes. Change the first
6821 2 of these define_insns to ...
6822 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
6823 define_insn for avx512vl.
6824 (<shift_insn><mode>3): ... and this, new define_insn without
6825 masking for non-avx512vl.
6828 * config/i386/sse.md (*andnot<mode>3,
6829 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
6830 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
6831 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
6832 not applied use empty suffix even for TARGET_AVX512VL.
6833 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
6834 is applied, supply evex,evex or evex,evex,evex instead of just
6837 2017-10-20 Julia Koval <julia.koval@intel.com>
6839 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
6840 (OPTION_MASK_ISA_GFNI_UNSET): New.
6841 (ix86_handle_option): Handle OPT_mgfni.
6842 * config/i386/cpuid.h (bit_GFNI): New.
6843 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
6844 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
6845 * config/i386/i386.c (ix86_target_string): Add -mgfni.
6846 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
6847 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
6848 * config/i386/i386.opt: Add mgfni.
6850 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
6852 * config/msp430/msp430.c (msp430_option_override): Disable
6853 -fdelete-null-pointer-checks.
6854 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
6856 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6858 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
6859 of x87 and SSE instructions.
6861 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6863 * asan.c (create_cond_insert_point): Do not update edge count.
6864 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
6865 (afdo_propagate_circuit): Likewise.
6866 (afdo_calculate_branch_prob): Likewise.
6867 (afdo_annotate_cfg): Likewise.
6868 * basic-block.h (struct edge_def): Remove count.
6869 (edge_def::count): New accessor.
6870 * bb-reorder.c (rotate_loop): Update.
6871 (find_traces_1_round): Update.
6872 (connect_traces): Update.
6873 (sanitize_hot_paths): Update.
6874 * cfg.c (unchecked_make_edge): Update.
6875 (make_single_succ_edge): Update.
6876 (check_bb_profile): Update.
6877 (dump_edge_info): Update.
6878 (update_bb_profile_for_threading): Update.
6879 (scale_bbs_frequencies_int): Update.
6880 (scale_bbs_frequencies_gcov_type): Update.
6881 (scale_bbs_frequencies_profile_count): Update.
6882 (scale_bbs_frequencies): Update.
6883 * cfganal.c (connect_infinite_loops_to_exit): Update.
6884 * cfgbuild.c (compute_outgoing_frequencies): Update.
6885 (find_many_sub_basic_blocks): Update.
6886 * cfgcleanup.c (try_forward_edges): Update.
6887 (try_crossjump_to_edge): Update
6888 * cfgexpand.c (expand_gimple_cond): Update
6889 (expand_gimple_tailcall): Update
6890 (construct_exit_block): Update
6891 * cfghooks.c (verify_flow_info): Update
6892 (redirect_edge_succ_nodup): Update
6893 (split_edge): Update
6894 (make_forwarder_block): Update
6895 (duplicate_block): Update
6896 (account_profile_record): Update
6897 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6898 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6899 * cfgloopmanip.c (scale_loop_profile): Update.
6901 (lv_adjust_loop_entry_edge): Update.
6902 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
6903 (force_nonfallthru_and_redirect): Update.
6904 (purge_dead_edges): Update.
6905 (rtl_flow_call_edges_add): Update.
6906 * cgraphunit.c (init_lowered_empty_function): Update.
6907 (cgraph_node::expand_thunk): Update.
6908 * gimple-pretty-print.c (dump_probability): Update.
6909 (dump_edge_probability): Update.
6910 * gimple-ssa-isolate-paths.c (isolate_path): Update.
6911 * haifa-sched.c (sched_create_recovery_edges): Update.
6912 * hsa-gen.c (convert_switch_statements): Update.
6913 * ifcvt.c (dead_or_predicable): Update.
6914 * ipa-inline-transform.c (inline_transform): Update.
6915 * ipa-split.c (split_function): Update.
6916 * ipa-utils.c (ipa_merge_profiles): Update.
6917 * loop-doloop.c (add_test): Update.
6918 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
6919 * lto-streamer-in.c (input_cfg): Update.
6920 (input_function): Update.
6921 * lto-streamer-out.c (output_cfg): Update.
6922 * modulo-sched.c (sms_schedule): Update.
6923 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6924 * predict.c (maybe_hot_edge_p): Update.
6925 (unlikely_executed_edge_p): Update.
6926 (probably_never_executed_edge_p): Update.
6927 (dump_prediction): Update.
6928 (drop_profile): Update.
6929 (propagate_unlikely_bbs_forward): Update.
6930 (determine_unlikely_bbs): Update.
6931 (force_edge_cold): Update.
6932 * profile.c (compute_branch_probabilities): Update.
6933 * reg-stack.c (better_edge): Update.
6934 * shrink-wrap.c (handle_simple_exit): Update.
6935 * tracer.c (better_p): Update.
6936 * trans-mem.c (expand_transaction): Update.
6937 (split_bb_make_tm_edge): Update.
6938 * tree-call-cdce.c: Update.
6939 * tree-cfg.c (gimple_find_sub_bbs): Update.
6940 (gimple_split_edge): Update.
6941 (gimple_duplicate_sese_region): Update.
6942 (gimple_duplicate_sese_tail): Update.
6943 (gimple_flow_call_edges_add): Update.
6944 (insert_cond_bb): Update.
6945 (execute_fixup_cfg): Update.
6946 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
6947 * tree-complex.c (expand_complex_div_wide): Update.
6948 * tree-eh.c (lower_resx): Update.
6949 (unsplit_eh): Update.
6950 (cleanup_empty_eh_move_lp): Update.
6951 * tree-inline.c (copy_edges_for_bb): Update.
6952 (freqs_to_counts): Update.
6953 (copy_cfg_body): Update.
6954 * tree-ssa-dce.c (remove_dead_stmt): Update.
6955 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6956 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
6957 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
6958 (unloop_loops): Update.
6959 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6960 * tree-ssa-loop-split.c (connect_loops): Update.
6961 (split_loop): Update.
6962 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6963 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
6964 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
6965 * tree-ssa-reassoc.c (branch_fixup): Update.
6966 * tree-ssa-tail-merge.c (replace_block_by): Update.
6967 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
6968 (compute_path_counts): Update.
6969 (update_profile): Update.
6970 (recompute_probabilities): Update.
6971 (update_joiner_offpath_counts): Update.
6972 (estimated_freqs_path): Update.
6973 (freqs_to_counts_path): Update.
6974 (clear_counts_path): Update.
6975 (ssa_fix_duplicate_block_edges): Update.
6976 (duplicate_thread_path): Update.
6977 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
6978 (case_bit_test_cmp): Update.
6979 (collect_switch_conv_info): Update.
6980 (gen_inbound_check): Update.
6981 (do_jump_if_equal): Update.
6982 (emit_cmp_and_jump_insns): Update.
6983 * tree-tailcall.c (decrease_profile): Update.
6984 (eliminate_tail_call): Update.
6985 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6986 (vect_do_peeling): Update.
6987 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6988 * ubsan.c (ubsan_expand_null_ifn): Update.
6989 (ubsan_expand_ptr_ifn): Update.
6990 * value-prof.c (gimple_divmod_fixed_value): Update.
6991 (gimple_mod_pow2): Update.
6992 (gimple_mod_subtract): Update.
6993 (gimple_ic): Update.
6994 (gimple_stringop_fixed_value): Update.
6996 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6999 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
7001 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
7003 PR rtl-optimization/82395
7004 * ira-color.c (allocno_priority_compare_func): Fix comparison step
7005 based on non_spilled_static_chain_regno_p.
7007 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7009 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
7010 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
7011 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
7013 2017-10-19 Martin Sebor <msebor@redhat.com>
7015 PR tree-optimization/82596
7016 * tree.c (array_at_struct_end_p): Handle STRING_CST.
7018 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7020 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
7021 (get_mem_refs_of_builtin_call): Likewise.
7022 * builtins.c (expand_builtin_apply): Adjust call to
7023 allocate_dynamic_stack_space.
7024 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
7025 the third argument to allocate_dynamic_stack_space, otherwise -1.
7026 (expand_builtin): Deal with all alloca variants.
7027 (is_inexpensive_builtin): Likewise.
7028 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
7029 * calls.c (special_function_p): Deal with all alloca variants.
7030 (initialize_argument_information): Adjust call to
7031 allocate_dynamic_stack_space.
7032 (expand_call): Likewise.
7033 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
7034 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
7035 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
7036 use it for the stack usage computation.
7037 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
7038 * function.c (gimplify_parameters): Call build_alloca_call_expr.
7039 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
7040 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
7041 (in_loop_p): Remove first argument and useless check.
7042 (pass_walloca::execute): Remove useless test and adjust call to above.
7043 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
7044 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
7045 (gimplify_call_expr): Deal with all alloca variants.
7046 * hsa-gen.c (gen_hsa_alloca): Likewise.
7047 (gen_hsa_insns_for_call): Likewise.
7048 * ipa-pure-const.c (special_builtin_state): Likewise.
7049 * tree-chkp.c (chkp_build_returned_bound): Likewise.
7050 * tree-object-size.c (alloc_object_size): Likewise.
7051 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
7052 (call_may_clobber_ref_p_1): Likewise.
7053 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
7054 (ccp_fold_stmt): Likewise.
7055 (optimize_stack_restore): Likewise.
7056 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
7057 (mark_all_reaching_defs_necessary_1): Likewise.
7058 (propagate_necessity): Likewise.
7059 (eliminate_unnecessary_stmts): Likewise.
7060 * tree.c (build_common_builtin_nodes): Build
7061 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
7062 (build_alloca_call_expr): New function.
7063 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
7064 (CASE_BUILT_IN_ALLOCA): Likewise.
7065 (build_alloca_call_expr): Declare.
7066 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
7068 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7071 * dwarf2out.c (new_die_raw): New static inline function.
7072 (new_die): Use it to create the DIE.
7073 (add_AT_external_die_ref): Likewise.
7074 (clone_die): Likewise.
7075 (clone_as_declaration): Likewise.
7076 (dwarf2out_vms_debug_main_pointer): Likewise.
7077 (base_type_die): Likewise. Remove early return for corner cases.
7078 Do not call add_pubtype on the DIE here.
7079 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
7080 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
7081 typedefs for base types with DW_AT_endianity. Make sure a DIE with
7082 native order exists for base types, attach the DIE manually and call
7083 add_pubtype on it. Do not equate a reverse order DIE to the type.
7085 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
7087 * config/arm/arm.c (align_ok_ldrd_strd): New function.
7088 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
7090 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
7092 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7094 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
7095 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
7096 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
7097 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
7098 * opts.c (sanitizer_opts): Add builtin.
7099 * ubsan.c (instrument_builtin): New function.
7100 (pass_ubsan::execute): Call it.
7101 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
7102 * doc/invoke.texi: Document -fsanitize=builtin.
7104 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
7105 builtins, store max (log2 (align), 0) into uchar field instead of
7106 align into uptr field.
7107 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
7108 store uchar 0 field instead of uptr 0 field.
7109 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
7110 instead of passing one address of struct with 2 locations pass
7111 two addresses of structs with 1 location each.
7112 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
7113 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
7114 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
7115 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
7116 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
7117 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
7118 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
7119 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
7121 2017-10-19 Martin Liska <mliska@suse.cz>
7124 * file-find.c (remove_prefix): Remove.
7125 * file-find.h (remove_prefix): Likewise.
7126 * gcc-ar.c: Remove smartness of lookup.
7128 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
7130 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
7131 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
7132 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
7134 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7137 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
7138 (setcc + and to xor + setcc): New peephole2.
7140 2017-10-19 Tom de Vries <tom@codesourcery.com>
7142 * doc/sourcebuild.texi (Test Directives, Variants of
7143 dg-require-support): Add dg-require-stack-size.
7145 2017-10-19 Martin Liska <mliska@suse.cz>
7148 * gimplify.c (gimplify_decl_expr): Do not instrument variables
7149 that have a large alignment.
7150 (gimplify_target_expr): Likewise.
7152 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
7154 PR rtl-optimization/82602
7155 * ira.c (rtx_moveable_p): Return false for volatile asm.
7157 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
7160 * config/i386/i386-modes.def (CCGZ): New CC mode.
7161 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
7162 * config/i386/predicates.md (ix86_comparison_operator):
7164 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7165 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
7166 with double-word subtraction.
7167 (put_condition_code): Handle CCGZmode.
7169 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
7171 * wide-int.cc (debug (const wide_int &)): New.
7172 (debug (const wide_int *)): New.
7173 (debug (const widest_int &)): New.
7174 (debug (const widest_int *)): New.
7176 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
7179 * lra-constraints.c (curr_insn_transform): Use non-input operand
7180 instead of output one for matched reload.
7182 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7184 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
7185 (tree-ssa-loop-ivopts.h): New header file.
7186 (struct builtin_info): New fields.
7187 (classify_builtin_1): Compute and record base and offset parts for
7188 memset builtin partition by calling strip_offset.
7189 (offset_cmp, fuse_memset_builtins): New functions.
7190 (finalize_partitions): Fuse adjacent memset partitions by calling
7192 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
7193 Expose the interface.
7194 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
7196 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7198 PR tree-optimization/82574
7199 * tree-loop-distribution.c (find_single_drs): New parameter. Check
7200 that data reference must be executed exactly once per iteration
7201 against the outermost loop in nest.
7202 (classify_partition): Update call to above function.
7204 2017-10-18 Richard Biener <rguenther@suse.de>
7206 PR tree-optimization/82591
7207 * graphite.c (graphite_transform_loops): Move code gen message
7209 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7210 Here. Handle scop_to_isl_ast failing.
7211 (scop_to_isl_ast): Limit the number of ISL operations.
7213 2017-10-18 Richard Biener <rguenther@suse.de>
7215 * graphite-isl-ast-to-gimple.c
7216 (translate_isl_ast_to_gimple::set_rename): Simplify.
7217 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
7218 (graphite_copy_stmts_from_block): ... here.
7219 (copy_bb_and_scalar_dependences): Simplify.
7220 (add_parameters_to_ivs_params): Canonicalize.
7221 (generate_entry_out_of_ssa_copies): Simplify.
7222 * graphite-sese-to-poly.c (extract_affine_name): Simplify
7223 by passing in ISL dimension.
7224 (parameter_index_in_region_1): Rename to ...
7225 (parameter_index_in_region): ... this.
7226 (extract_affine): Adjust assert, pass down parameter index.
7227 (add_param_constraints): Use range-info when available.
7228 (build_scop_context): Adjust.
7229 * sese.c (new_sese_info): Adjust.
7230 (free_sese_info): Likewise.
7231 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
7232 Remove unused typedefs.
7233 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
7235 2017-10-18 Martin Liska <mliska@suse.cz>
7237 * combine.c (simplify_compare_const): Add gcc_fallthrough.
7239 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7241 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
7242 (s390_sched_init): Do not reset s390_sched_state if we entered the
7243 current basic block via a fallthru edge and all others are unlikely.
7245 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7247 * config/s390/s390.c (NUM_SIDES): New variable.
7248 (LONGRUNNING_THRESHOLD): New variable.
7249 (LATENCY_FACTOR): New variable.
7250 (s390_sched_score): Decrease score for long-running instructions on
7252 (s390_sched_variable_issue): Perform bookkeeping for long-running
7255 2017-10-18 Richard Biener <rguenther@suse.de>
7257 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
7258 Simplify with removal of the parameter rename map.
7259 (set_rename): Likewise.
7260 (should_copy_to_new_region): Likewise.
7261 (graphite_copy_stmts_from_block): Likewise.
7262 (copy_bb_and_scalar_dependences): Remove initialization of
7263 unused copied_bb_map.
7265 (copy_internal_parameters): Likewise.
7266 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
7267 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
7268 Use INTEGRAL_TYPE_P.
7269 (parameter_index_in_region_1): Rename to ...
7270 (assign_parameter_index_in_region): ... this. Assert we have
7271 a parameter we handle.
7272 (scan_tree_for_params): Adjust.
7273 * sese.h (parameter_rename_map_t): Remove.
7274 (struct sese_info_t): Remove unused parameter_rename_map and
7275 copied_bb_map members.
7276 * sese.c (new_sese_info): Adjust.
7277 (free_sese_info): Likewise.
7279 2017-10-18 Martin Liska <mliska@suse.cz>
7282 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
7283 on an abnormal edge.
7285 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
7287 * doc/invoke.texi (ffunction-sections and fdata-sections):
7290 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
7292 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
7293 the use statement can throw internally.
7295 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
7297 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
7298 any RTX present on the RHS of a SET.
7299 * compare-elim.c (try_eliminate_compare): Restore comment.
7301 2017-10-17 Jakub Jelinek <jakub@redhat.com>
7303 * langhooks.h (struct lang_hooks): Document that tree_size langhook
7304 may be also called on tcc_type nodes.
7305 * langhooks.c (lhd_tree_size): Likewise.
7307 2017-10-17 David Malcolm <dmalcolm@redhat.com>
7309 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
7310 format_warning_at_substring.
7311 (maybe_warn): Convert source_range * param to a location_t. Pass
7312 UNKNOWN_LOCATION rather than NULL to fmtwarn.
7313 (format_directive): Remove code to extract source_ranges and
7314 source_range * in favor of just a location_t.
7315 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
7317 * substring-locations.c (format_warning_va): Convert
7318 source_range * param to a location_t.
7319 (format_warning_at_substring): Likewise.
7320 * substring-locations.h (format_warning_va): Likewise.
7321 (format_warning_at_substring): Likewise.
7323 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
7325 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
7327 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
7328 and scatter/gather ops.
7330 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
7331 vec_gather_load and vec_scatter_store.
7332 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
7333 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
7334 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
7335 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
7336 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
7337 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
7339 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
7341 * reg-stack.c (compare_for_stack_reg): Add bool argument.
7342 Detect FTST instruction and handle its register pops. Only pop
7343 second operand if can_pop_second_op is true.
7344 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
7345 set can_pop_second_op to false in the compare_for_stack_reg call.
7347 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
7348 output_fp_compare for stack register operands.
7349 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
7350 instructions here. Do not emit stack register pops here. Assert
7351 that FCOMPP pops next to top stack register. Rewrite function.
7353 2017-10-17 Nathan Sidwell <nathan@acm.org>
7356 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
7357 use DECL_ASSEMBLER_NAME_RAW.
7360 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
7363 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
7364 Wilco Dijkstra <wilco.dijkstra@arm.com>
7366 * builtins.c (expand_builtin_update_setjmp_buf): Add a
7367 converstion to Pmode from the buf_addr.
7369 2017-10-17 Richard Biener <rguenther@suse.de>
7371 * graphite-dependences.c (scop_get_reads_and_writes): Change
7372 output parameters to references.
7374 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
7376 PR 71026/tree-optimization
7377 * fold-const.c (distribute_real_division): Removed.
7378 (fold_binary_loc): Remove calls to distribute_real_divison.
7380 2017-10-17 Richard Biener <rguenther@suse.de>
7382 * graphite-scop-detection.c
7383 (scop_detection::stmt_has_simple_data_refs_p): Always use
7384 the full nest as region.
7385 (try_generate_gimple_bb): Likewise.
7386 * sese.c (scalar_evolution_in_region): Simplify now that
7387 SCEV can handle instantiation in regions.
7388 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
7389 in the non-loop part of a function if requested.
7391 2017-10-17 Richard Biener <rguenther@suse.de>
7393 PR tree-optimization/82563
7394 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
7396 (graphite_regenerate_ast_isl): Call it.
7397 * graphite-scop-detection.c (build_scops): Remove entry edge split.
7399 2017-10-17 Jakub Jelinek <jakub@redhat.com>
7401 PR tree-optimization/82549
7402 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
7403 Formatting fixes. Instead of calling make_bit_field_ref with negative
7406 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
7408 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
7409 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
7410 _mm_maskz_reduce_ss): New.
7411 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
7412 __builtin_ia32_reducess_mask): Ditto..
7413 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
7414 * config/i386/sse.md (reduces<mode>): Renamed to ...
7415 (reduces<mode><mask_scalar_name>): ... this.
7416 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
7418 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
7419 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
7421 2017-10-16 David Malcolm <dmalcolm@redhat.com>
7423 * Makefile.in (OBJS): Add unique-ptr-tests.o.
7424 * selftest-run-tests.c (selftest::run_tests): Call
7425 selftest::unique_ptr_tests_cc_tests.
7426 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
7427 * unique-ptr-tests.cc: New file.
7429 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
7432 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7434 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7435 (make_hard_regno_born, make_hard_regno_dead): Update
7436 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
7438 2017-10-16 Jeff Law <law@redhat.com>
7440 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7442 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
7444 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
7446 2017-10-16 Olivier Hainque <hainque@adacore.com>
7448 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
7449 with_cpu if we were configured for an e500v2 target cpu name.
7451 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
7453 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
7454 * doc/invoke.texi: Document +nodsp as a valid extension for
7457 2017-10-16 Martin Liska <mliska@suse.cz>
7459 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
7460 (test_set_range): Likewise.
7461 (test_range_functions): Rename to ...
7462 (test_bit_in_range): ... this.
7463 (sbitmap_c_tests): Add new test.
7465 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7467 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
7469 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
7470 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
7472 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7474 * config/aarch64/aarch64-builtins.c
7475 (aarch64_types_quadopu_lane_qualifiers): New.
7476 (TYPES_QUADOPU_LANE): New.
7477 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
7478 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
7479 (aarch64_<sur>dot_laneq<vsi2qi>): New.
7480 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
7481 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
7482 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
7483 (Vdottype, DOTPROD): New.
7484 (sur): Add SDOT and UDOT.
7486 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7488 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
7489 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
7490 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
7492 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
7493 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
7494 Enable TARGET_DOTPROD.
7495 (cortex-a75.cortex-a55): Likewise.
7496 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
7498 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7500 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
7501 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
7503 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
7505 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
7506 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
7507 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
7508 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
7509 * config/arm/types.md (neon_dot, neon_dot_q): New.
7510 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
7512 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7514 * config/arm/arm.h (TARGET_DOTPROD): New.
7515 * config/arm/arm.c (arm_arch_dotprod): New.
7516 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
7517 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
7518 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
7519 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
7520 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
7521 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
7522 * doc/invoke.texi (armv8.2-a): Document dotprod
7524 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7526 * i386.c (ix86_vec_cost): New function.
7527 (ix86_rtx_costs): Handle vector operations better.
7528 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
7529 * x86-tune-costs.h: Add new costs to all tables.
7531 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7533 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
7535 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
7536 divsd, sqrtss and sqrtsd
7537 * x86-tune-costs.h: Add new entries to all costs.
7538 (znver1_cost): Fix to match real instruction latencies.
7540 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7541 Michael Collison <michael.collison@arm.com>
7543 * compare-elim.c: Include emit-rtl.h.
7544 (can_merge_compare_into_arith): New function.
7545 (try_validate_parallel): Likewise.
7546 (try_merge_compare): Likewise.
7547 (try_eliminate_compare): Call the above when no previous clobber
7549 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
7552 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7556 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
7557 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
7561 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
7562 to be any operand_equal_p operands. For & (B - 1) require
7563 B to be power of 2. Recognize
7564 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
7566 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7569 * optabs.c (expand_memory_blockage): Fix call of
7570 targetm.have_memory_blockage.
7572 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7575 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
7576 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
7577 objects to extra_objs instead of overwriting it.
7579 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7581 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
7582 Use any_fp_register_operand as operand[3] predicate. Simplify
7583 equality test for operands[2] and operands[4] memory location.
7584 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
7585 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
7586 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
7587 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
7588 any_fp_register_operand as operand[1] predicate. Simplify
7589 equality test for operands[0] and operands[3] memory location.
7590 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
7591 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
7592 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
7594 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7596 * target-insns.def: Add memory_blockage.
7597 * optabs.c (expand_memory_blockage): New function.
7598 (expand_asm_memory_barrier): Rename ...
7599 (expand_asm_memory_blockage): ... to this.
7600 (expand_mem_thread_fence): Call expand_memory_blockage
7601 instead of expand_asm_memory_barrier.
7602 (expand_mem_singnal_fence): Ditto.
7603 (expand_atomic_load): Ditto.
7604 (expand_atomic_store): Ditto.
7605 * doc/md.texi (Standard Pattern Names For Generation):
7606 Document memory_blockage instruction pattern.
7608 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
7610 * config/rl78/rl78.c (rl78_emit_libcall): New function.
7611 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
7612 * config/rl78/rl78.md: New define_expand "adddi3".
7614 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7616 * cfghooks.c (verify_flow_info): Disable check that all probabilities
7619 2017-10-13 Jeff Law <law@redhat.com>
7621 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
7623 2017-10-13 Jakub Jelinek <jakub@redhat.com>
7626 * internal-fn.c (expand_mul_overflow): If both operands have
7627 the same highpart of -1 or 0 and the topmost bit of lowpart
7628 is different, overflow is if res <= 0 rather than res < 0.
7630 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
7632 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
7633 TARGET_P9_VECTOR code for unaligned_load case.
7635 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7637 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
7639 2017-10-13 Nathan Sidwell <nathan@acm.org>
7641 * tree-core.h (tree_contains_struct): Make bool.
7642 * tree.c (tree_contains_struct): Likewise.
7643 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
7644 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
7645 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
7646 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
7647 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
7649 2017-10-13 Richard Biener <rguenther@suse.de>
7651 * graphite-isl-ast-to-gimple.c
7652 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
7653 parameters and dominance check.
7654 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
7655 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
7656 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
7657 Do not update SSA form here or do intermediate IL verification.
7658 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
7659 (graphite_initialize): Remove check on the number of loops in
7660 the function and inline into graphite_transform_loops.
7661 (graphite_finalize): Inline into graphite_transform_loops.
7662 (graphite_transform_loops): Perform SSA update and IL verification
7664 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
7666 2017-10-13 Richard Biener <rguenther@suse.de>
7668 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
7669 graphite_expression_type_precision): Avoid global constructor
7671 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
7672 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
7673 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
7674 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
7675 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
7676 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
7678 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
7681 * config/i386/i386.h (ix86_red_zone_size): New.
7682 * config/i386/i386.md (push peephole2s): Replace
7683 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
7685 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7686 Alan Hayward <alan.hayward@arm.com>
7687 David Sherwood <david.sherwood@arm.com>
7689 * combine.c (can_change_dest_mode): Reject changes in
7690 REGMODE_NATURAL_SIZE.
7692 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7693 Alan Hayward <alan.hayward@arm.com>
7694 David Sherwood <david.sherwood@arm.com>
7696 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
7697 (expand_debug_source_expr): Likewise.
7698 * combine.c (combine_simplify_rtx): Likewise.
7699 * cse.c (fold_rtx): Likewise.
7700 * fwprop.c (canonicalize_address): Likewise.
7701 * targhooks.c (default_shift_truncation_mask): Likewise.
7703 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7704 Alan Hayward <alan.hayward@arm.com>
7705 David Sherwood <david.sherwood@arm.com>
7707 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
7708 (widened_mode): Likewise.
7709 (expand_unop): Likewise.
7710 * ree.c (transform_ifelse): Likewise.
7711 (merge_def_and_ext): Likewise.
7712 (combine_reaching_defs): Likewise.
7713 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7715 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7716 Alan Hayward <alan.hayward@arm.com>
7717 David Sherwood <david.sherwood@arm.com>
7719 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
7720 * combine.c (gen_lowpart_for_combine): Likewise.
7721 * dwarf2out.c (rtl_for_decl_location): Likewise.
7722 * final.c (alter_subreg): Likewise.
7723 * rtlhooks.c (gen_lowpart_general): Likewise.
7724 (gen_lowpart_if_possible): Likewise.
7726 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7727 Alan Hayward <alan.hayward@arm.com>
7728 David Sherwood <david.sherwood@arm.com>
7730 * calls.c (expand_call): Use subreg_lowpart_offset.
7731 * cse.c (cse_insn): Likewise.
7732 * regcprop.c (copy_value): Likewise.
7733 (copyprop_hardreg_forward_1): Likewise.
7735 2017-10-13 Jakub Jelinek <jakub@redhat.com>
7738 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
7739 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
7740 =Q constraints to +Q and into insn condition add check
7741 that operands[0] and operands[1] are equal.
7742 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
7743 =Q constraints to +Q and into insn condition add check
7744 that operands[0] is equal to either operands[1] or operands[2].
7747 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
7748 instead of handling MINUS_EXPR twice (once for each argument),
7749 canonicalize operand order and handle just once, use rtype where
7750 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
7753 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
7754 any values of __C while still being pattern recognizable as a simple
7757 2017-10-13 Richard Biener <rguenther@suse.de>
7759 PR tree-optimization/82451
7761 2017-10-02 Richard Biener <rguenther@suse.de>
7763 PR tree-optimization/82355
7764 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7765 a mapping for the enclosing loop but avoid generating one for
7767 (copy_bb_and_scalar_dependences): Remove premature codegen
7768 error on PHIs in blocks duplicated into multiple places.
7769 * graphite-scop-detection.c
7770 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7771 in the region use it as loop and nest to analyze the DR in.
7772 (try_generate_gimple_bb): Likewise.
7773 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7774 (add_loop_constraints): For blocks in a loop not in the region
7775 create a dimension with a single iteration.
7776 * sese.h (gbb_loop_at_index): Remove assert.
7778 * cfgloop.c (loop_preheader_edge): For the loop tree root
7779 return the single successor of the entry block.
7780 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7781 Reset the SCEV hashtable and niters.
7782 * graphite-scop-detection.c
7783 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
7784 assert that we only have POLYNOMIAL_CHREC that vary in loops
7785 contained in the region.
7786 (scop_detection::graphite_can_represent_expr): Adjust.
7787 (scop_detection::stmt_has_simple_data_refs_p): For loops
7788 not in the region set loop to NULL. The nest is now the
7789 entry edge to the region.
7790 (try_generate_gimple_bb): Likewise.
7791 * sese.c (scalar_evolution_in_region): Adjust for
7792 instantiate_scev change.
7793 * tree-data-ref.h (graphite_find_data_references_in_stmt):
7794 Make nest parameter the edge into the region.
7795 (create_data_ref): Likewise.
7796 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
7797 entry edge into a region and adjust instantiate_scev calls.
7798 (create_data_ref): Likewise.
7799 (graphite_find_data_references_in_stmt): Likewise.
7800 (find_data_references_in_stmt): Pass the loop preheader edge
7801 from the nest argument.
7802 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
7803 parameter the edge into the region.
7804 (instantiate_parameters): Use the loop preheader edge as entry.
7805 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
7807 (get_instantiated_value_entry): Make instantiate_below parameter
7808 the edge into the region.
7809 (instantiate_scev_name): Likewise. Adjust dominance checks,
7810 when we cannot use loop-based instantiation instantiate by
7811 walking use-def chains.
7812 (instantiate_scev_poly): Adjust.
7813 (instantiate_scev_binary): Likewise.
7814 (instantiate_scev_convert): Likewise.
7815 (instantiate_scev_not): Likewise.
7816 (instantiate_array_ref): Remove.
7817 (instantiate_scev_3): Likewise.
7818 (instantiate_scev_2): Likewise.
7819 (instantiate_scev_1): Likewise.
7820 (instantiate_scev_r): Do not blindly handle N-operand trees.
7821 Do not instantiate array-refs. Handle all constants and invariants.
7822 (instantiate_scev): Make instantiate_below parameter
7823 the edge into the region.
7824 (resolve_mixers): Use the loop preheader edge for the region
7825 parameter to instantiate_scev_r.
7826 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
7828 2017-10-13 Richard Biener <rguenther@suse.de>
7830 PR tree-optimization/82525
7831 * graphite-isl-ast-to-gimple.c
7832 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
7834 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
7835 Fail code generation when we cannot represent the isl integer.
7836 (binary_op_to_tree): Elide modulo operations that are no-ops
7837 in the type we code generate. Remove now superfluous code
7840 2017-10-13 Richard Biener <rguenther@suse.de>
7842 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
7843 (scop_detection::harmful_loop_in_region): Remove premature
7844 IV type restriction.
7845 (scop_detection::graphite_can_represent_scev): We can handle
7846 pointer IVs just fine.
7848 2017-10-13 Alan Modra <amodra@gmail.com>
7850 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
7851 "Clobbers and Scratch Registers". Add paragraph on
7852 alternative to clobbers for scratch registers and OpenBLAS
7855 2017-10-13 Alan Modra <amodra@gmail.com>
7857 * doc/extend.texi (Clobbers): Correct vax example. Delete old
7858 example of a memory input for a string of known length. Move
7859 commentary out of table. Add a number of new examples
7860 covering array memory inputs.
7862 2017-10-12 Martin Liska <mliska@suse.cz>
7864 PR tree-optimization/82493
7865 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
7866 (test_range_functions): New function.
7867 (sbitmap_c_tests): Likewise.
7868 * selftest-run-tests.c (selftest::run_tests): Run new tests.
7869 * selftest.h (sbitmap_c_tests): New function.
7871 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7873 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7875 * config/rs6000/amo.h: Fix spacing issue.
7877 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7880 * config/i386/i386.md (*ashl<mode>3_mask_1,
7881 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
7882 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
7885 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
7887 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
7888 (profile_probability): Set max_probability
7889 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
7891 * profile-count.c (profile_probability::differs_from_p): Do not
7892 rely on max_probaiblity == 10000
7894 2017-10-12 Jeff Law <law@redhat.com>
7896 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
7899 2017-10-12 Martin Sebor <msebor@redhat.com>
7903 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
7904 (handle_alias_pairs): Call it.
7905 * common.opt (-Wattribute-alias): New option.
7906 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
7907 * doc/invoke.texi (-Wattribute-alias): Document.
7909 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
7912 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
7914 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7916 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7917 (make_hard_regno_born, make_hard_regno_dead): Update
7918 bb_killed_pseudos and bb_gen_pseudos.
7920 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
7922 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
7924 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
7926 * config/alpha/alpha.c (alpha_split_conditional_move):
7927 Use std::swap instead of manually swapping.
7928 (alpha_stdarg_optimize_hook): Ditto.
7929 (alpha_canonicalize_comparison): Ditto.
7931 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7933 * tree-loop-distribution.c (struct builtin_info): New struct.
7934 (struct partition): Refactor fields into struct builtin_info.
7935 (partition_free): Free struct builtin_info.
7936 (build_size_arg_loc, build_addr_arg_loc): Delete.
7937 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
7938 information from struct builtin_info.
7939 (find_single_drs): New function refactored from classify_partition.
7940 Also moved builtin validity checks to this function.
7941 (compute_access_range, alloc_builtin): New functions.
7942 (classify_builtin_st, classify_builtin_ldst): New functions.
7943 (classify_partition): Refactor code into functions find_single_drs,
7944 classify_builtin_st and classify_builtin_ldst.
7945 (distribute_loop): Don't do runtime alias check when distributing
7947 (find_seed_stmts_for_distribution): New function.
7948 (pass_loop_distribution::execute): Refactor code finding seed
7949 stmts into above function. Support distribution for the innermost
7950 two-level loop nest. Adjust dump information.
7952 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7954 * tree-loop-distribution.c: Adjust the general comment.
7955 (NUM_PARTITION_THRESHOLD): New macro.
7956 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
7957 (classify_partition): Skip builtin pattern of loop nest's inner loop.
7958 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
7959 in call to build_partition_graph.
7960 (finalize_partitions): New parameter. Make loop distribution more
7961 conservative by fusing more partitions.
7962 (distribute_loop): Don't do runtime alias check in case of loop nest
7964 (find_seed_stmts_for_distribution): New function.
7965 (prepare_perfect_loop_nest): New function.
7966 (pass_loop_distribution::execute): Refactor code finding seed stmts
7967 and loop nest into above functions. Support loop nest distribution.
7968 Adjust dump information accordingly.
7970 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7972 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
7973 and set PTYPE_SEQUENTIAL for merged partition.
7975 2017-10-12 Richard Biener <rguenther@suse.de>
7977 PR tree-optimization/69728
7979 2017-09-19 Richard Biener <rguenther@suse.de>
7981 PR tree-optimization/69728
7982 * graphite-sese-to-poly.c (schedule_error): New global.
7983 (add_loop_schedule): Handle empty domain by failing the
7985 (build_original_schedule): Handle schedule_error.
7987 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
7988 domain by returning an unchanged schedule.
7990 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7992 * genrecog.c (validate_pattern): For VEC_SELECT verify that
7993 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
7995 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
7997 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
8000 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8003 * expr.c (store_field): Don't optimize away bitsize == 0 store
8004 from CALL_EXPR with addressable return type.
8006 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8008 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
8009 * config/rs6000/rs6000.md (sel): Delete mode attribute.
8010 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
8011 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
8012 TARGET_ISEL instead of TARGET_ISEL<sel>.
8014 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
8016 * config/rs6000/rs6000.c
8017 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
8019 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8021 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
8023 (reg_or_cint_operand): Fix comment.
8024 (reg_or_zero_operand): New predicate.
8025 * config/rs6000/rs6000-protos.h (output_isel): Delete.
8026 * config/rs6000/rs6000.c (output_isel): Delete.
8027 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
8028 instead of reg_or_cint_operand. Output instruction directly (not via
8030 (isel_unsigned_<mode>): Ditto.
8031 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
8032 gpc_reg_operand. Add an instruction alternative for this. Output
8033 instruction directly.
8034 (*isel_reversed_unsigned_<mode>): Ditto.
8036 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8038 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
8039 (TARGET_CANONICALIZE_COMPARISON): Define.
8041 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
8044 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
8045 Check whether the dest is REG before adding REG_EQUIV note.
8047 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8050 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8052 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8053 (make_hard_regno_born, make_hard_regno_dead): Update
8054 bb_killed_pseudos and bb_gen_pseudos.
8056 2017-10-11 Nathan Sidwell <nathan@acm.org>
8058 * incpath.h (enum incpath_kind): Name enum, prefix values.
8059 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
8060 * incpath.c (heads, tails): Use INC_MAX.
8061 (add_env_var_paths, add_standard_paths): Use incpath_kind.
8062 (merge_include_chains, split_quote_chain,
8063 register_include_chains): Update incpath_kind names.
8064 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
8065 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
8067 (add_framework_path, darwin_register_objc_includes): Likewise.
8068 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
8070 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8072 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
8073 Do not use float_operator operator predicate.
8074 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
8075 * config/i386/predicates.md (float_operator): Remove predicate.
8077 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8079 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
8080 (*jccxf_i387): Ditto.
8081 (*jcc<mode>_i387): Ditto.
8082 (*jccu<mode>_i387): Ditto.
8083 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
8084 (*jcc_*_i387 splitters): Remove.
8085 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
8086 * config/i386/i386.c (ix86_split_fp_branch): Remove.
8087 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
8090 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8092 * profile-count.h (slow_safe_scale_64bit): New function.
8093 (safe_scale_64bit): New inline.
8094 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
8095 * profile-count.c: Include wide-int.h
8096 (slow_safe_scale_64bit): New.
8098 2017-10-11 Nathan Sidwell <nathan@acm.org>
8100 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
8101 HAS_DECL_ASSEMBLER_NAME_P.
8102 * gimple-expr.c (gimple_decl_printable_name: Check
8103 HAS_DECL_ASSEMBLER_NAME_P too.
8104 * ipa-utils.h (type_in_anonymous_namespace_p): Check
8105 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
8106 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
8107 * passes.c (rest_of_decl_compilation): Check
8108 HAS_DECL_ASSEMBLER_NAME_P too.
8109 * recog.c (verify_changes): Likewise.
8110 * tree-pretty-print.c (dump_decl_name): Likewise.
8111 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
8113 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
8114 (SET_DECL_ASSEMBLER_NAME): Use it.
8115 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
8116 (COPY_DECL_ASSEMBLER_NAME): Likewise.
8117 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
8119 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8121 * config.gcc (i386, x86_64): Add extra objects.
8122 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
8123 (ix86_min_insn_size): Declare.
8124 (ix86_issue_rate): Declare.
8125 (ix86_adjust_cost): Declare.
8126 (ia32_multipass_dfa_lookahead): Declare.
8127 (ix86_macro_fusion_p): Declare.
8128 (ix86_macro_fusion_pair_p): Declare.
8129 (ix86_bd_has_dispatch): Declare.
8130 (ix86_bd_do_dispatch): Declare.
8131 (ix86_core2i7_init_hooks): Declare.
8132 (ix86_atom_sched_reorder): Declare.
8133 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
8134 (COSTS_N_BYTES): Move to x86-tune-costs.h.
8135 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
8136 (rip_relative_addr_p): Rename to ...
8137 (ix86_rip_relative_addr_p): ... this one; export.
8138 (memory_address_length): Update.
8139 (ix86_issue_rate): Move to x86-tune-sched.c.
8140 (ix86_flags_dependent): Move to x86-tune-sched.c.
8141 (ix86_agi_dependent): Move to x86-tune-sched.c.
8142 (exact_dependency_1): Move to x86-tune-sched.c.
8143 (exact_store_load_dependency): Move to x86-tune-sched.c.
8144 (ix86_adjust_cost): Move to x86-tune-sched.c.
8145 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
8146 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
8147 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
8148 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
8149 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
8150 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
8151 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
8152 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
8153 (min_insn_size): Rename to ...
8154 (ix86_min_insn_size): ... this one; export.
8155 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
8156 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
8157 (core2i7_first_cycle_multipass_backtrack): Move to
8158 x86-tune-sched-core.c.
8159 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
8160 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
8161 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
8162 (ix86_avoid_jump_mispredicts): Update.
8163 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
8164 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
8165 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
8166 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
8167 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
8168 (MAX_INSN): Move to ix86-tune-sched-bd.c.
8169 (MAX_IMM): Move to ix86-tune-sched-bd.c.
8170 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
8171 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
8172 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
8173 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
8174 (MAX_STORE): Move to ix86-tune-sched-bd.c.
8175 (BIG): Move to ix86-tune-sched-bd.c.
8176 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
8177 (enum insn_path): Move to ix86-tune-sched-bd.c.
8178 (get_mem_group): Move to ix86-tune-sched-bd.c.
8179 (is_cmp): Move to ix86-tune-sched-bd.c.
8180 (dispatch_violation): Move to ix86-tune-sched-bd.c.
8181 (is_branch): Move to ix86-tune-sched-bd.c.
8182 (is_prefetch): Move to ix86-tune-sched-bd.c.
8183 (init_window): Move to ix86-tune-sched-bd.c.
8184 (allocate_window): Move to ix86-tune-sched-bd.c.
8185 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
8186 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
8187 (process_end_window): Move to ix86-tune-sched-bd.c.
8188 (allocate_next_window): Move to ix86-tune-sched-bd.c.
8189 (find_constant): Move to ix86-tune-sched-bd.c.
8190 (get_num_immediates): Move to ix86-tune-sched-bd.c.
8191 (has_immediate): Move to ix86-tune-sched-bd.c.
8192 (get_insn_path): Move to ix86-tune-sched-bd.c.
8193 (get_insn_group): Move to ix86-tune-sched-bd.c.
8194 (count_num_restricted): Move to ix86-tune-sched-bd.c.
8195 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
8196 (add_insn_window): Move to ix86-tune-sched-bd.c.
8197 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
8198 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
8199 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
8200 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
8201 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
8202 (do_dispatch): Move to ix86-tune-sched-bd.c.
8203 (has_dispatch): Move to ix86-tune-sched-bd.c.
8204 * config/i386/t-i386: Add new object files.
8205 * config/i386/x86-tune-costs.h: New file.
8206 * config/i386/x86-tune-sched-atom.c: New file.
8207 * config/i386/x86-tune-sched-bd.c: New file.
8208 * config/i386/x86-tune-sched-core.c: New file.
8209 * config/i386/x86-tune-sched.c: New file.
8211 2017-10-11 Liu Hao <lh_mouse@126.com>
8213 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
8214 the generic version below instead.
8215 (should_colorize): Recognize Windows consoles as terminals
8217 * pretty-print.c [__MINGW32__] (write_all): New function.
8218 [__MINGW32__] (find_esc_head): Likewise.
8219 [__MINGW32__] (find_esc_terminator): Likewise.
8220 [__MINGW32__] (eat_esc_sequence): Likewise.
8221 [__MINGW32__] (mingw_ansi_fputs): New function that handles
8223 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
8226 2017-10-11 Richard Biener <rguenther@suse.de>
8228 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
8229 Properly call analyze_scalar_evolution with the loop of the stmt.
8231 2017-10-11 Richard Biener <rguenther@suse.de>
8233 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
8234 * tree-core.h (tree_base): Add chrec_var union member.
8235 * tree.h (CHREC_VAR): Remove.
8236 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
8237 * tree-chrec.h (build_polynomial_chrec): Adjust.
8238 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
8239 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
8241 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
8243 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
8244 * match.pd: ... here.
8245 ((T) X == (T) Y): Relax condition.
8247 2017-10-11 Bin Cheng <bin.cheng@arm.com>
8249 PR tree-optimization/82472
8250 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
8252 (break_alias_scc_partitions): Update postorder number.
8254 2017-10-11 Martin Liska <mliska@suse.cz>
8257 * opts.c (parse_no_sanitize_attribute): Do not use error_value
8259 * opts.h (parse_no_sanitize_attribute): Remove last argument.
8261 2017-10-11 Martin Liska <mliska@suse.cz>
8263 * print-rtl.c (print_insn): Move declaration of idbuf
8264 to same scope as name.
8266 2017-10-11 Martin Liska <mliska@suse.cz>
8271 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8274 2017-10-11 Martin Liska <mliska@suse.cz>
8277 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8280 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
8282 * config.gcc (default_gnu_indirect_function): Default to yes for
8283 arm*-*-linux* with glibc.
8285 2017-10-11 Richard Biener <rguenther@suse.de>
8287 * tree-scalar-evolution.c (get_scalar_evolution): Handle
8288 default-defs and types we do not want to analyze.
8289 (interpret_loop_phi): Replace unreachable code with an assert.
8290 (compute_scalar_evolution_in_loop): Remove and inline ...
8291 (analyze_scalar_evolution_1): ... here, replacing condition with
8292 what makes the intent clearer. Remove handling of cases
8293 get_scalar_evolution now handles.
8295 2017-10-10 Jim Wilson <wilson@tuliptree.org>
8297 PR rtl-optimization/81434
8298 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
8299 comment for main loop. In sched_group_found if, also add checks for
8300 pass and min_cost_group.
8302 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
8304 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
8305 (rs6000_insn_cost): New function.
8306 * config/rs6000/rs6000.md (cost): New attribute.
8308 2017-10-10 Jakub Jelinek <jakub@redhat.com>
8309 H.J. Lu <hongjiu.lu@intel.com>
8313 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
8314 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
8315 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
8316 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
8317 to OPTION_MASK_ISA_AVX512VL - builtins that have both
8318 OPTION_MASK_ISA_MMX and some other bit set require both
8319 mmx and the ISAs without the mmx bit.
8320 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
8321 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
8322 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
8323 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
8324 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
8325 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
8326 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
8327 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
8328 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
8329 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
8330 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
8331 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
8332 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
8333 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
8334 Add OPTION_MASK_ISA_MMX.
8336 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
8338 * config.gcc (armv7*-*-freebsd*): New target.
8339 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
8341 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
8343 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
8344 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
8347 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
8349 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
8350 a template parameter.
8351 (WIDE_INT_REF_FOR): Update accordingly.
8352 * tree.h (wi::int_traits <const_tree>): Delete.
8353 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
8354 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
8355 (wi::tree_to_wide_ref): New typedef.
8356 (wi::to_wide): New function.
8357 * calls.c (get_size_range): Use wi::to_wide when operating on
8359 * cgraph.c (cgraph_node::create_thunk): Likewise.
8360 * config/i386/i386.c (ix86_data_alignment): Likewise.
8361 (ix86_local_alignment): Likewise.
8362 * dbxout.c (stabstr_O): Likewise.
8363 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
8364 * expr.c (const_vector_from_tree): Likewise.
8365 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
8366 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
8367 (fold_negate_expr_1, int_const_binop_1, const_binop)
8368 (fold_convert_const_int_from_real, optimize_bit_field_compare)
8369 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
8370 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
8371 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
8372 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
8373 (fold_not_const, round_up_loc): Likewise.
8374 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
8375 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
8376 (alloca_call_type): Likewise.
8377 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
8378 * godump.c (go_output_typedef): Likewise.
8379 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
8380 * internal-fn.c (get_min_precision): Likewise.
8381 * ipa-cp.c (ipcp_store_vr_results): Likewise.
8382 * ipa-polymorphic-call.c
8383 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
8384 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
8385 (ipa_modify_call_arguments): Likewise.
8386 * match.pd: Likewise.
8387 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
8388 * print-tree.c (print_node_brief, print_node): Likewise.
8389 * stmt.c (expand_case): Likewise.
8390 * stor-layout.c (layout_type): Likewise.
8391 * tree-affine.c (tree_to_aff_combination): Likewise.
8392 * tree-cfg.c (group_case_labels_stmt): Likewise.
8393 * tree-data-ref.c (dr_analyze_indices): Likewise.
8394 (prune_runtime_alias_test_list): Likewise.
8395 * tree-dump.c (dequeue_and_dump): Likewise.
8396 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
8397 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
8398 * tree-pretty-print.c (dump_generic_node): Likewise.
8399 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
8400 (simple_iv_with_niters): Likewise.
8401 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
8402 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
8403 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
8404 * tree-ssa-loop-niter.c (split_to_var_and_offset)
8405 (refine_value_range_using_guard, number_of_iterations_ne_max)
8406 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
8407 (get_cst_init_from_scev, record_nonwrapping_iv)
8408 (scev_var_range_cant_overflow): Likewise.
8409 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
8410 * tree-ssa-pre.c (compute_avail): Likewise.
8411 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
8412 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
8413 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
8414 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
8415 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
8416 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
8417 (dump_case_nodes, try_switch_expansion): Likewise.
8418 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
8419 (vect_do_peeling): Likewise.
8420 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
8421 * tree-vect-stmts.c (vectorizable_load): Likewise.
8422 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
8423 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
8424 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
8425 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
8426 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
8427 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
8428 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
8429 (evrp_dom_walker::before_dom_children): Likewise.
8430 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
8431 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
8432 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
8433 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
8434 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
8435 (get_range_pos_neg): Likewise.
8436 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
8437 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
8438 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
8439 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
8440 * config/avr/avr.c (avr_fold_builtin): Likewise.
8441 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
8442 * config/msp430/msp430.c (msp430_attr): Likewise.
8443 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
8444 * config/powerpcspe/powerpcspe-c.c
8445 (altivec_resolve_overloaded_builtin): Likewise.
8446 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
8447 (rs6000_expand_ternop_builtin): Likewise.
8448 * config/rs6000/rs6000-c.c
8449 (altivec_resolve_overloaded_builtin): Likewise.
8450 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
8451 (rs6000_expand_ternop_builtin): Likewise.
8452 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
8454 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8456 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
8457 when copying loop nest with only one inner loop.
8459 2017-10-10 Richard Biener <rguenther@suse.de>
8461 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
8462 blocks if SCEV is active.
8463 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
8465 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
8466 (scev_initialize): Assert we are not yet initialized.
8468 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8470 * tree-loop-distribution.c (generate_loops_for_partition): Remove
8471 inner loop's exit stmt by making it always exit the loop, otherwise
8472 we would generate an infinite empty loop.
8474 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8476 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
8477 renaming variables in new preheader if it's deleted.
8479 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8481 * tree-loop-distribution.c (struct partition): Remove unused field
8482 loops of the structure.
8483 (partition_alloc, partition_free): Ditto.
8484 (build_rdg_partition_for_vertex): Ditto.
8486 2017-10-09 Jeff Law <law@redhat.com>
8488 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
8489 return type to match prototype and documentation.
8491 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8493 * config/rs6000/rs6000.c (processor_costs): Move to ...
8494 * config/rs6000/rs6000.h: ... here.
8495 (rs6000_cost): Declare.
8497 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
8499 * except.c (setjmp_fn): New global variable.
8500 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
8501 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
8502 if DONT_USE_BUILTIN_SETJMP is defined.
8504 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8506 * target.def (insn_cost): New hook.
8507 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
8508 * doc/tm.texi: Regenerate.
8509 * rtlanal.c (insn_cost): Use the new hook.
8511 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8513 * combine.c (combine_validate_cost): Compute the new insn_cost,
8514 not just pattern_cost.
8515 (try_combine): Adjust comment.
8517 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8519 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
8521 * combine.c (uid_insn_cost): Adjust comment.
8522 (combine_validate_cost): Adjust comment. Use pattern_cost instead
8524 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
8525 * dse.c (find_shift_sequence): Ditto.
8526 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
8527 (bb_valid_for_noce_process_p): Use pattern_cost.
8528 * rtl.h (insn_rtx_cost): Delete.
8529 (pattern_cost): New prototype.
8530 (insn_cost): New prototype.
8531 * rtlanal.c (insn_rtx_cost): Rename to...
8532 (pattern_cost): ... this.
8535 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
8537 * config/i386/i386.md (*jcc_2): Remove insn pattern.
8538 (*jcc<mode>_0_r_i387): Ditto.
8539 (*jccxf_r_i387): Ditto.
8540 (*jcc<mode>_r_i387): Ditto.
8541 (*jccu<mode>_r_i387): Ditto.
8542 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
8543 (*jcc): Rename from *jcc_1.
8545 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8547 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
8548 deferred rescans after the lvx/stvx recombination pre-pass.
8550 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
8552 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
8553 memory operation instruction support.
8554 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
8555 (rs6000-ibm-aix[789]*): Likewise.
8556 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
8557 Document new functions.
8559 2017-10-09 Richard Biener <rguenther@suse.de>
8561 PR tree-optimization/82397
8562 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
8563 equality only for semantically equal trees.
8565 2017-10-09 Richard Biener <rguenther@suse.de>
8567 PR tree-optimization/82449
8568 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
8569 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
8570 allow constant addresses.
8571 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
8574 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8576 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
8579 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8582 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
8585 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8588 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
8590 2017-10-09 Jakub Jelinek <jakub@redhat.com>
8593 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
8594 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
8596 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
8598 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
8599 (WI_BINARY_PREDICATE_RESULT): Likewise.
8600 (wi::binary_traits::operator_result): New type.
8601 (wi::binary_traits::predicate_result): Likewise.
8602 (generic_wide_int::operator~, unary generic_wide_int::operator-)
8603 (generic_wide_int::operator==, generic_wide_int::operator!=)
8604 (generic_wide_int::operator&, generic_wide_int::and_not)
8605 (generic_wide_int::operator|, generic_wide_int::or_not)
8606 (generic_wide_int::operator^, generic_wide_int::operator+
8607 (binary generic_wide_int::operator-, generic_wide_int::operator*):
8609 (operator~, unary operator-, operator==, operator!=, operator&)
8610 (operator|, operator^, operator+, binary operator-, operator*): New
8612 * expr.c (get_inner_reference): Use wi::bit_and_not.
8613 * fold-const.c (fold_binary_loc): Likewise.
8614 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
8615 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
8616 (bit_value_binop): Likewise.
8617 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
8618 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
8619 (extract_range_from_binary_expr_1): Likewise.
8620 (masked_increment): Likewise.
8621 (simplify_bit_ops_using_ranges): Likewise.
8623 2017-10-09 Martin Jambor <mjambor@suse.cz>
8626 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
8627 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
8628 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
8630 (hsa_fixup_mov_insn_type): New function.
8631 (hsa_op_with_type::get_in_type): Use it.
8632 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
8633 immediates in an assert.
8634 (hsa_op_with_type::extend_int_to_32bit): New method.
8635 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
8636 types. Convert to dest type if necessary.
8637 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
8638 (reg_for_gimple_ssa): Pass false as min32int to
8639 hsa_type_for_scalar_tree_type.
8640 (gen_hsa_addr): Fixup type when creating addresable temporary.
8641 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
8642 (gen_hsa_unary_operation): Extend operands and convert to dest type if
8643 necessary. Call hsa_fixup_mov_insn_type.
8644 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
8645 extend operands and convert to dest type if necessary.
8646 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
8647 to dest type if necessary.
8648 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
8649 if conversion nt necessary and size matches.
8650 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
8651 to dest type if necessary.
8652 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
8653 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
8655 (gen_hsa_clrsb): Likewise.
8656 (gen_hsa_ffs): Likewise.
8657 (gen_hsa_divmod): Extend operands and convert to dest type if
8659 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
8661 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8663 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
8664 Remove empty default arguments. Use a brace block as output
8666 (conditional return): Ditto.
8668 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
8669 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
8671 (group_ending_nop): Ditto.
8672 (doloop_end): Ditto.
8673 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
8674 (splitters for those): Ditto.
8676 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8678 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
8679 a conditional jump (and the compare for it) so that pc_rtx is the
8681 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
8682 for the deleted and renamed ctr<mode>_internal[234] patterns.
8683 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
8684 Delete second conditional return pattern.
8685 (ctr<mode>_internal2): Delete this second bdnz pattern.
8686 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
8687 (ctr<mode>_internal4): Delete this second bdz pattern.
8689 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
8691 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
8692 (always_initialized_rtx_for_ssa_name_p): New predicate.
8693 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
8694 (finish_out_of_ssa): Free new field of SA.
8695 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
8696 * tree-ssa-coalesce.c: Include tree-ssa.h.
8697 (get_parm_default_def_partitions): Remove extern keyword.
8698 (get_undefined_value_partitions): New function.
8699 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
8700 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
8702 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
8704 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
8706 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
8708 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
8710 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
8711 for targets that preffer 128bit.
8713 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
8715 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
8717 2017-10-08 Olivier Hainque <hainque@adacore.com>
8719 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
8720 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
8722 (thumb_set_return_address): Likewise.
8724 2017-10-08 Olivier Hainque <hainque@adacore.com>
8726 * common/config/arm/arm-common.c (arm_except_unwind_info):
8727 Handle DWARF2_UNWIND_INFO.
8729 2017-10-07 Michael Collison <michael.collison@arm.com>
8731 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
8734 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
8736 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
8737 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
8738 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
8739 defined, force the creation of a new block for a dispatch label.
8741 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
8743 * invoke.texi (Wsuggest-attribute=cold): Document.
8744 * common.opt (Wsuggest-attribute=cold): New
8745 * ipa-pure-const.c (warn_function_cold): New function.
8746 * predict.c (compute_function_frequency): Use it.
8747 * predict.h (warn_function_cold): Declare.
8749 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
8751 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
8754 2017-10-06 Martin Liska <mliska@suse.cz>
8756 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
8757 keyword for member functions.
8758 (struct sanopt_tree_couple): New struct.
8759 (struct sanopt_tree_couple_hash): New function.
8760 (struct sanopt_ctx): Add new hash_map.
8761 (has_dominating_ubsan_ptr_check): New function.
8762 (record_ubsan_ptr_check_stmt): Likewise.
8763 (maybe_optimize_ubsan_ptr_ifn): Likewise.
8764 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
8765 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
8767 2017-10-06 Sudakshina Das <sudi.das@arm.com>
8770 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
8771 aarch64_simd_valid_immediate on CONST_VECTORs.
8772 (aarch64_reg_or_bic_imm): Likewise.
8774 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
8776 PR rtl-optimization/82396
8777 * haifa-sched.c (ready_sort_real): Disable qsort checking.
8779 2017-10-06 Sebastian Pop <sebpop@gmail.com>
8781 * graphite-dependences.c (scop_get_reads): Move code to...
8782 (scop_get_must_writes): Move code to...
8783 (scop_get_may_writes): Move code to...
8784 (scop_get_reads_and_writes): ... here.
8785 (scop_get_dependences): Call scop_get_reads_and_writes.
8787 2017-10-06 Jakub Jelinek <jakub@redhat.com>
8789 PR tree-optimization/82434
8790 * fold-const.h (can_native_encode_type_p,
8791 can_native_encode_string_p): Remove.
8792 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
8793 don't encode anything, just return what would be otherwise returned.
8794 (native_encode_fixed, native_encode_complex, native_encode_vector):
8796 (native_encode_string): Likewise. Inline by hand
8797 can_native_encode_string_p.
8798 (can_native_encode_type_p): Remove.
8799 (can_native_encode_string_p): Remove.
8800 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
8801 STRING_CSTs using can_native_encode_string_p, test all
8802 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
8803 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
8804 argument from native_encode_expr.
8805 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
8806 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
8809 2017-10-06 Richard Biener <rguenther@suse.de>
8811 PR tree-optimization/82397
8812 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
8813 operand_equal_p but rely on data_ref_compare_tree for detecting
8815 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
8816 to match up with dr_group_sort_cmp.
8818 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8821 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
8823 * config/s390/s390-builtin-types.def: Regenerate.
8825 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8828 * config/s390/s390-builtin-types.def: Regenerate.
8829 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
8830 Change flag from B_VXE to B_VX.
8831 (s390_vec_min_dbl): Remove B_VXE flag.
8833 2017-10-06 Richard Biener <rguenther@suse.de>
8835 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
8836 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
8837 translate_isl_ast_to_gimple::is_valid_rename,
8838 translate_isl_ast_to_gimple::get_rename,
8839 translate_isl_ast_to_gimple::get_def_bb_for_const,
8840 translate_isl_ast_to_gimple::get_new_name,
8841 translate_isl_ast_to_gimple::collect_all_ssa_names,
8842 translate_isl_ast_to_gimple::copy_loop_phi_args,
8843 translate_isl_ast_to_gimple::collect_all_ssa_names,
8844 translate_isl_ast_to_gimple::copy_loop_phi_args,
8845 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
8846 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
8847 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
8848 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
8849 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
8850 translate_isl_ast_to_gimple::copy_cond_phi_args,
8851 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
8852 translate_isl_ast_to_gimple::edge_for_new_close_phis,
8853 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
8854 translate_isl_ast_to_gimple::rename_uses,
8855 translate_isl_ast_to_gimple::rename_all_uses): Remove.
8856 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
8857 (set_rename_for_each_def): Likewise.
8858 (graphite_copy_stmts_from_block): Handle debug stmt resetting
8859 here. Handle rewriting SCEV analyzable uses here.
8860 (copy_bb_and_scalar_dependences): Generate code for PHI
8862 (graphite_regenerate_ast_isl): Adjust.
8863 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
8864 (add_write, add_read): New functions.
8865 (build_cross_bb_scalars_def): Use it and simplify.
8866 (build_cross_bb_scalars_use): Likewise.
8867 (graphite_find_cross_bb_scalar_vars): Inline into...
8868 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
8869 simulating out-of-SSA. Compute liveout and add dependencies.
8870 (build_scops): Force an empty entry block.
8871 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
8873 (sese_build_liveouts): Declare.
8874 (sese_trivially_empty_bb_p): Likewise.
8875 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
8876 compute liveout and debug_liveout.
8877 (sese_bad_liveouts_use): Remove.
8878 (sese_reset_debug_liveouts_bb): Likewise.
8879 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
8880 (sese_build_liveouts): Build liveout and debug_liveout and store
8882 (new_sese_info): Adjust.
8883 (free_sese_info): Likewise.
8884 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
8885 do not build liveout here.
8886 (move_sese_in_condition): Adjust region entry.
8887 (scev_analyzable_p): Match up with chrec_apply requirements.
8888 (sese_trivially_empty_bb_p): New.
8889 * tree-into-ssa.c (get_reaching_def): Properly support generating
8890 default-defs for incremental rewrite of anonymous names.
8892 2017-10-06 Richard Biener <rguenther@suse.de>
8894 * graphite-sese-to-poly.c (extract_affine): For casts increasing
8895 precision do not perform modulo reduction.
8897 2017-10-06 Richard Biener <rguenther@suse.de>
8899 PR tree-optimization/82436
8900 * tree-vect-slp.c (vect_supported_load_permutation_p): More
8901 conservatively choose the vectorization factor when checking
8902 whether we can perform the required load permutation.
8903 (vect_transform_slp_perm_load): Assert when we may not fail.
8905 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
8907 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
8908 message for incompatible -msdata=* and -mcall-* options.
8910 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8912 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
8913 rate for post-reload scheduling.
8915 2017-10-05 Tamar Christina <tamar.christina@arm.com>
8917 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
8919 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8921 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
8922 to improve monte carlo in scimark.
8924 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8926 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
8927 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
8928 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
8929 pentium4_cost, nocona_cost): Set reassociation width to 1.
8930 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
8931 width to 2 for fp operations and 1 otherwise.
8932 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
8934 (atom_cost): Set reassociation width to 2.
8935 (slm_cost, generic_cost): Set fp reassociation width
8936 to 2 and 1 otherwise.
8937 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
8938 (core_cost): Set fp reassociation width to 4 and vector to 2.
8939 (ix86_reassociation_width): Rewrite using cost table; special case
8940 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
8941 and TARGET_AVX128_OPTIMAL.
8942 * config/i386/i386.h (processor_costs): Add
8943 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
8944 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
8945 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
8946 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
8947 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
8948 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
8950 2017-10-05 Nathan Sidwell <nathan@acm.org>
8952 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
8954 2017-10-05 Tamar Christina <tamar.christina@arm.com>
8956 * config/arm/arm.c (arm_test_fpu_data): New.
8957 (arm_run_selftests): Call arm_test_fpu_data.
8959 2017-10-04 Nathan Sidwell <nathan@acm.org>
8961 * toplev.c (toplev::main): Remove excess parens on pretty_printer
8963 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
8965 2017-10-04 Sudakshina Das <sudi.das@arm.com>
8967 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
8968 check type for aarch64_simd_valid_immediate.
8969 (aarch64_output_simd_mov_immediate): Update prototype.
8970 (aarch64_simd_valid_immediate): Update prototype.
8971 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
8972 support for ORR-immediate.
8973 (and<mode>3): modified pattern to add support for BIC-immediate.
8974 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
8975 now checks for valid immediate for BIC and ORR based on new enum
8977 (aarch64_output_simd_mov_immediate): Function now used to output
8978 BIC/ORR imm as well based on new enum argument.
8979 * config/aarch64/constraints.md (Do): New vector immediate constraint.
8981 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
8982 (aarch64_reg_or_bic_imm): Likewise.
8984 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8986 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
8987 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
8989 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
8993 PR rtl-optimization/82396
8994 * haifa-sched.c (autopref_multipass_init): Simplify
8996 (autopref_rank_data): Simplify sort order.
8997 * sched-int.h (autopref_multipass_data_): Remove
8998 multi_mem_insn_p, min_offset and max_offset.
9000 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9002 * doc/sourcebuild.texi: Document vect_peeling_profitable.
9004 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9006 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
9009 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9011 * doc/sourcebuild.texi: Document vect_long_mult.
9013 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
9015 PR tree-optimization/82413
9016 * fold-const.c (build_range_check): Use widest_int when comparing
9017 the maximum ETYPE value with HIGH.
9019 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9021 PR rtl-optimization/82396
9022 * haifa-sched.c (autopref_multipass_init): Simplify
9024 (autopref_rank_data): Simplify sort order.
9025 * sched-int.h (autopref_multipass_data_): Remove
9026 multi_mem_insn_p, min_offset and max_offset.
9028 2017-10-04 Jakub Jelinek <jakub@redhat.com>
9030 PR tree-optimization/82381
9031 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
9032 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
9035 PR tree-optimization/82374
9036 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
9037 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
9038 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
9039 current_function_decl to the new decl.
9041 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
9043 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
9044 helper macro for IEEE float128 hardware built-in functions.
9045 (SQRTF128_ODD): Add built-in functions with the round-to-odd
9047 (TRUNCF128_ODD): Likewise.
9048 (ADDF128_ODD): Likewise.
9049 (SUBF128_ODD): Likewise.
9050 (MULF128_ODD): Likewise.
9051 (DIVF128_ODD): Likewise.
9052 (FMAF128_ODD): Likewise.
9053 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
9054 UNSPEC_TRUNC_ROUND_TO_ODD.
9055 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
9056 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
9057 floating point round to odd instructions.
9058 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
9059 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
9060 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
9061 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
9062 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
9063 (trunc<mode>sf2_hw): Change the truncate with round to odd
9064 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
9065 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
9066 to odd hardware instructions.
9067 (sub<mode>3_odd): Likewise.
9068 (mul<mode>3_odd): Likewise.
9069 (div<mode>3_odd): Likewise.
9070 (sqrt<mode>2_odd): Likewise.
9071 (fma<mode>4_odd): Likewise.
9072 (fms<mode>4_odd): Likewise.
9073 (nfma<mode>4_odd): Likewise.
9074 (nfms<mode>4_odd): Likewise.
9075 (trunc<mode>df2_odd): Change the truncate with round to odd
9076 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
9078 * doc/extend.texi (PowerPC built-in functions): Update documentation
9079 for existing IEEE float128-bit built-in functions. Add built-in
9080 functions that generate the IEEE 128-bit floating point round to
9083 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
9085 PR rtl-optimization/77729
9086 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
9087 to (X&(C1&~C2))|C2 transformations.
9089 2017-10-03 Martin Jambor <mjambor@suse.cz>
9091 PR tree-optimization/82363
9092 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
9093 mismatch, mark lacc written regardless of racc.
9095 2017-10-03 Jakub Jelinek <jakub@redhat.com>
9097 PR tree-optimization/82381
9098 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
9099 stmt_to_insert nor wheather SSA_NAMEs are default defs.
9100 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
9101 fallthrough into reassoc_stmt_dominates_stmt_p.
9104 * combine.c (combine_instructions): Don't combine in unreachable
9107 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9110 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
9111 function to not use the have_cpu variable. Do not set cpu_index,
9112 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
9114 (rs6000_valid_attribute_p): Remove duplicate initializations of
9115 old_optimize and func_optimize.
9116 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
9117 (rs6000_activate_target_options): Make global.
9118 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
9121 2017-10-02 Jakub Jelinek <jakub@redhat.com>
9123 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
9124 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
9125 Set *poffset to 0 and *psize and *pmax_size to -1 if
9126 *poffset + *psize overflows in HOST_WIDE_INT.
9128 PR tree-optimization/82387
9129 PR tree-optimization/82388
9130 PR tree-optimization/82389
9131 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
9132 instead of live_bytes non-NULL.
9134 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
9137 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
9140 2017-10-02 Richard Biener <rguenther@suse.de>
9142 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
9143 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
9144 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
9146 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
9148 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
9149 requested precision matches the type's.
9150 * calls.c (alloc_max_size): Calculate the new candidate size as
9151 a widest_int and use wi::to_widest when comparing it with the
9152 current candidate size.
9153 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
9154 zero rather than integer_zero_node.
9155 * match.pd: Check for a no-op conversion before using wi::add
9156 rather than after. Use tree_to_uhwi when summing small shift
9157 counts into an unsigned int.
9159 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
9160 Alan Hayward <alan.hayward@arm.com>
9161 David Sherwood <david.sherwood@arm.com>
9164 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
9165 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
9166 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
9167 POINTER_AND_FP_REGS.
9169 2017-10-02 Richard Biener <rguenther@suse.de>
9171 PR tree-optimization/82355
9172 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
9173 a mapping for the enclosing loop but avoid generating one for
9175 (copy_bb_and_scalar_dependences): Remove premature codegen
9176 error on PHIs in blocks duplicated into multiple places.
9177 * graphite-scop-detection.c
9178 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
9179 in the region use it as loop and nest to analyze the DR in.
9180 (try_generate_gimple_bb): Likewise.
9181 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
9182 (add_loop_constraints): For blocks in a loop not in the region
9183 create a dimension with a single iteration.
9184 * sese.h (gbb_loop_at_index): Remove assert.
9186 2017-10-01 Kevin Buettner <kevinb@redhat.com>
9188 * omp-expand.c (adjust_context_scope): New function.
9189 (expand_parallel_call): Call adjust_context_scope.
9191 2017-10-01 Jeff Law <law@redhat.com>
9193 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
9194 dom_opt_dom_walker class with direct access to private members.
9195 Add comments. Call test_for_singularity.
9196 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
9197 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
9198 m_dummy_cond anymore.
9199 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
9201 (pass_dominator:execute): Build the dummy_cond here and pass it
9202 to the dom_opt_dom_walker ctor.
9203 (test_for_singularity): New function.
9205 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
9206 Maya Rashish <coypu@sdf.org>
9208 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
9209 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
9210 (alpha*-*-netbsd*) Use nbsd_tm_file.
9211 (arm*-*-netbsdelf*) Likewise.
9212 (i[34567]86-*-netbsdelf*) Likewise.
9213 (x86_64-*-netbsd*) Likewise.
9214 (mips*-*-netbsd*) Likewise.
9215 (powerpc-*-netbsd*) Likewise.
9216 (sh*-*-netbsd*) Likewise.
9217 (sparc-*-netbsdelf*) Likewise.
9218 (sparc64-*-netbsd*) Likewise.
9219 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
9221 (vax-*-netbsdelf*) Likewise.
9222 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
9223 (UINT_FAST8_TYPE) Likewise.
9224 (INT_FAST16_TYPE) Check CHAR_FAST16.
9225 (UINT_FAST16_TYPE) Likewise.
9227 2017-09-30 Jakub Jelinek <jakub@redhat.com>
9230 * config/i386/i386.md
9231 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
9232 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
9233 *divmodsi4_zext_2): New define_insn_and_split.
9234 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
9235 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
9236 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
9237 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
9238 New define_insn_and_split.
9239 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
9240 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
9241 operands[1] having DImode when mode is SImode.
9243 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
9244 always SImode for DIV and MOD in REG_EQUAL notes.
9246 2017-09-29 Yury Gribov <tetra2005@gmail.com>
9249 * match.pd: Fix handling of NaNs in pattern.
9251 2017-09-29 Jeff Law <law@redhat.com>
9253 * sbitmap.c (bitmap_bit_in_range_p): New function.
9254 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
9255 * tree-ssa-dse.c (live_bytes_read): New function.
9256 (dse_classify_store): Ignore reads of dead bytes.
9258 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
9259 typos and whitespace errors.
9260 * config/i386/predicates.md (address_no_seg_operand): Likewise.
9261 * config/s390/s390.c (s390_emit_prologue): Likewise.
9263 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
9266 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
9267 with a symbol for LRA.
9269 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
9271 PR rtl-optimization/82338
9272 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
9274 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
9276 * genmodes.c (calc_wider_mode): Suppress qsort macro.
9277 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
9278 (qsort_chk): Declare.
9279 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
9280 (qsort_chk): New function.
9282 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9284 PR tree-optimization/82337
9285 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
9286 phi definition if the PHI result appears in an abnormal PHI.
9287 (find_basis_for_base_expr): Don't record a basis if the LHS of the
9288 basis appears in an abnormal PHI.
9290 2017-09-29 Richard Biener <rguenther@suse.de>
9292 * graphite-isl-ast-to-gimple.c
9293 (translate_isl_ast_to_gimple::set_codegen_error): New function.
9294 (binary_op_to_tree): Use it.
9295 (get_rename_from_scev): Likewise.
9296 (copy_loop_phi_nodes): Likewise.
9297 (copy_bb_and_scalar_dependences): Likewise.
9298 (translate_pending_phi_nodes): Likewise.
9300 2017-09-29 Jakub Jelinek <jakub@redhat.com>
9303 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
9304 for movabsq $(i32 << shift), r64.
9306 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
9308 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
9309 index when encoding %esp as %rsp to avoid 0x67 prefix.
9311 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
9313 * config/i386/i386.md (*movsf_internal, *movdf_internal):
9314 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
9316 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
9318 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
9319 Extensions with more than 16 double VFP registers.
9320 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
9321 to_clear_mask and all code related to it. Replace the remaining
9322 entry by a sbitmap and adapt code accordingly.
9324 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
9326 * brig-builtins.def: Change pure attributes to const.
9328 2017-09-28 Joseph Myers <joseph@codesourcery.com>
9330 * config.gcc (default_gnu_indirect_function): Default to yes for
9331 sparc*-*-linux* with glibc.
9333 2017-09-28 Joseph Myers <joseph@codesourcery.com>
9335 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
9336 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
9337 when creating .init_array and .fini_array sections with priority
9340 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
9343 * config/aarch64/aarch64.c
9344 (aarch64_builtin_support_vector_misalignment): Always return false
9345 when misalignment is unknown.
9347 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
9349 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
9350 this function to return false if the definition used by the swap
9351 instruction is artificial, or if the memory address from which the
9352 constant value is loaded is not represented by a base address held
9353 in a register or if the base address register is a frame or stack
9354 pointer. Additionally, return false if the base address of the
9355 loaded constant is a SYMBOL_REF but is not considered to be a
9357 (replace_swapped_load_constant): New function.
9358 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
9359 loaded constant vector with a load of a swapped constant vector.
9361 2017-09-27 Carl Love <cel@us.ibm.com>
9363 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
9364 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
9365 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
9368 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
9370 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
9371 first, always call autopref_rank_data otherwise.
9373 2017-09-27 Richard Biener <rguenther@suse.de>
9375 * graphite-scop-detection.c (find_scop_parameters): Move
9376 loop bound handling ...
9377 (gather_bbs::before_dom_children): ... here, avoiding the need
9378 to build scop_info->loop_nest.
9379 (record_loop_in_sese): Remove.
9380 * sese.h (sese_info_t::loop_nest): Remove.
9381 * sese.c (new_sese_info): Do not allocate loop_nest.
9382 (free_sese_info): Do not free loop_nest.
9384 2017-09-27 Jakub Jelinek <jakub@redhat.com>
9387 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
9388 lhs from calls if the lhs has addressable type.
9390 2017-09-27 Richard Biener <rguenther@suse.de>
9392 * graphite.h (scop::max_alias_set): New member.
9393 * graphite-scop-detection.c: Remove references to non-existing
9394 --param in comments.
9395 (build_alias_sets): Record the maximum alias set used for drs.
9396 (build_scops): Support zero as unlimited for
9397 --param graphite-max-arrays-per-scop.
9398 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
9400 (build_poly_sr_1): ... here. Compute alias set based on the
9401 maximum alias set used for drs rather than
9402 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
9404 2017-09-27 Richard Biener <rguenther@suse.de>
9406 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
9407 --param loop-block-tile-size=0 to disable tiling.
9409 2017-09-27 Richard Biener <rguenther@suse.de>
9411 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
9412 (graphite-max-nb-scop-params): Document special value zero.
9413 * domwalk.h (dom_walker::STOP): New symbolical constant.
9414 (dom_walker::dom_walker): Add optional parameter for bb to
9416 (dom_walker::~dom_walker): Declare.
9417 (dom_walker::before_dom_children): Document STOP return value.
9418 (dom_walker::m_user_bb_to_rpo): New member.
9419 (dom_walker::m_bb_to_rpo): Likewise.
9420 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
9421 mapping here if not provided by the user.
9422 (dom_walker::~dom_walker): Free bb to RPO mapping if not
9423 provided by the user.
9424 (dom_walker::STOP): Define.
9425 (dom_walker::walk): Do not compute bb to RPO mapping here.
9426 Support STOP return value from before_dom_children to stop
9428 * graphite-optimize-isl.c (optimize_isl): If the schedule
9429 is the same still generate code if -fgraphite-identity
9430 or -floop-parallelize-all are given.
9431 * graphite-scop-detection.c: Include cfganal.h.
9432 (gather_bbs::gather_bbs): Get and pass through bb to RPO
9434 (gather_bbs::before_dom_children): Return STOP for BBs
9436 (build_scops): Compute bb to RPO mapping and pass it to
9437 the domwalk. Treat --param graphite-max-nb-scop-params=0
9438 as not limiting the number of params.
9439 * graphite.c (graphite_initialize): Remove limit on the
9440 number of basic-blocks in a function.
9441 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
9442 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
9443 default value of 10.
9445 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9447 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
9448 Adjust code to eliminate needing to do the shift right 32-bits
9449 operation after XSCVDPSPN.
9451 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9453 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
9454 ((X / Y) != 0 -> X >= Y): Likewise.
9456 2017-09-26 Carl Love <cel@us.ibm.com>
9458 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
9459 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
9460 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
9461 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
9462 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
9463 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
9464 definitions and overloading.
9465 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
9466 statement for P9V_BUILTIN_XST_LEN_R.
9467 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
9468 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
9469 define_expand and define_insn for the instructions and builtins.
9470 * doc/extend.texi: Update the built-in documentation file for the new
9472 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
9473 define_insn for the instructions
9475 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
9478 * gcc/config/netbsd-protos.h: New file.
9479 * gcc/config/netbsd.c: New file.
9480 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
9481 * gcc/config/t-netbsd: New file.
9482 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
9483 (tmake_file) Add t-netbsd.
9484 (extra_objs) Add netbsd.o.
9486 2017-09-26 Janus Weil <janus@gcc.gnu.org>
9490 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
9492 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9494 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
9495 sign extension from a vector register to a GPR by doing a 32-bit
9496 direct move and then an EXTSW.
9497 (extendsi<mode>2 splitter): Likewise.
9498 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
9499 right or vector extract after doing XSCVDPSPN. Use
9500 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
9502 (movdi_from_sf_zero_ext): Likewise.
9503 (reload_gpr_from_vsxsf): Likewise.
9504 (p8_mfvsrd_4_disf): Delete, no longer used.
9505 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
9506 then needing to move the SFmode to a GPR to use the XSCVDPSP
9507 instruction instead of FRSP and XSCVDPSPN.
9508 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
9509 it is adjacent to the other XSCVSPDP insns.
9510 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
9511 SFmode to be in traditional Altivec registers.
9512 (vsx_xscvdpspn): Eliminate useless alternative constraint.
9513 (vsx_xscvspdpn): Likewise.
9514 (vsx_xscvspdpn_scalar): Likewise.
9516 2017-09-26 Martin Jambor <mjambor@suse.cz>
9518 * tree-sra.c (compare_access_positions): Put integral types first,
9519 stabilize sorting of integral types, remove conditions putting
9520 non-full-precision integers last.
9521 (sort_and_splice_var_accesses): Disable scalarization if a
9522 non-integert would be represented by a non-full-precision integer.
9524 2017-09-26 Joseph Myers <joseph@codesourcery.com>
9526 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
9527 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
9528 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
9529 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
9530 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
9531 conditionals inside the function instead of around it. Call
9532 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
9533 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
9535 2017-09-26 Richard Biener <rguenther@suse.de>
9537 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
9539 (scop_detection::build_scop_breadth): ... this. Removed.
9540 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
9541 (scop_detection::harmful_stmt_in_bb): Likewise.
9542 (scop_detection::graphite_can_represent_stmt): Likewise.
9543 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
9544 (scop_detection::can_represent_loop): Remove recursion, fold in ...
9545 (scop_detection::can_represent_loop_1): ... this. Removed.
9546 (scop_detection::harmful_loop_in_region): Simplify after inlining
9547 the above and remove more quadraticness.
9548 (build_scops): Adjust.
9549 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
9552 2017-09-26 Jakub Jelinek <jakub@redhat.com>
9555 * config/i386/i386.c (ix86_print_operand_address_as): Only test
9556 REGNO (base) == SP_REG if base is a REG.
9559 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
9560 if it is different SSA_NAME.
9561 (optimize_range_tests_cmp_bitwise): New function.
9562 (optimize_range_tests): Call it.
9564 2017-09-26 Richard Biener <rguenther@suse.de>
9566 PR tree-optimization/82321
9567 * graphite.c (canonicalize_loop_closed_ssa): Properly check
9568 for the def being inside the loop.
9570 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9572 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
9574 * config/s390/s390-builtins.def: Fix constraint on op4.
9576 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9578 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
9579 independent expanders.
9580 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
9581 ("vec_ordered", "vec_unordered"): New expanders.
9583 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9585 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
9588 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9590 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
9591 vec_unpacks_lo_v16qi.
9592 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
9594 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9596 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
9597 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
9598 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
9600 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9602 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
9604 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
9606 ("*vec_slb<mode>"): New insn pattern.
9607 ("vec_shr_<mode>"): New expander.
9608 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
9609 and force the shift count operand to V16QImode.
9610 ("vec_srb<mode>"): Set shift count mode to V16QI.
9612 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9614 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
9615 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
9616 ("vec_widen_smult_hi_<mode>"): New expander definitions.
9618 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
9621 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
9623 2017-09-26 Richard Biener <rguenther@suse.de>
9625 PR tree-optimization/82320
9626 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
9629 2017-09-25 Jeff Law <law@redhat.com>
9631 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
9632 prototype for new argument.
9633 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
9634 mostly extracted from rs6000_emit_allocate_stack.
9635 (rs6000_emit_probe_stack_range_stack_clash): New function.
9636 (rs6000_emit_allocate_stack): Call
9637 rs6000_emit_probe_stack_range_stack_clash as needed.
9638 (rs6000_emit_probe_stack_range): Add additional argument
9639 to call to gen_probe_stack_range{si,di}.
9640 (output_probe_stack_range): New.
9641 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
9642 (output_probe_stack_range_stack_clash): New.
9643 (rs6000_emit_prologue): Emit notes into dump file as requested.
9644 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
9645 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
9646 Add additional operand and pass it to output_probe_stack_range.
9648 2017-09-25 Bin Cheng <bin.cheng@arm.com>
9650 PR tree-optimization/82163
9651 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
9652 (checking_verify_loop_closed_ssa): New parameter.
9653 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
9654 (check_loop_closed_ssa_stmt): Delete.
9655 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
9656 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
9657 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
9660 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
9662 * brig-builtins.def: Treat HSAIL barrier builtins as
9663 setjmp/longjump style functions.
9665 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9667 * target.def (constant_alignment): New hook.
9668 * defaults.h (CONSTANT_ALIGNMENT): Delete.
9669 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
9670 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
9671 * doc/tm.texi: Regenerate.
9672 * targhooks.h (default_constant_alignment): Declare.
9673 (constant_alignment_word_strings): Likewise.
9674 * targhooks.c (default_constant_alignment): New function.
9675 (constant_alignment_word_strings): Likewise.
9676 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
9677 instead of CONSTANT_ALIGNMENT.
9678 * varasm.c (align_variable, get_variable_align, build_constant_desc)
9679 (force_const_mem): Likewise.
9680 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
9681 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
9682 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
9683 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9684 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
9686 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
9687 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9688 constant_alignment_word_strings.
9689 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
9690 (CONSTANT_ALIGNMENT): Likewise.
9691 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9692 (arm_constant_alignment): New function.
9693 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
9694 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9695 constant_alignment_word_strings.
9696 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
9697 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9698 constant_alignment_word_strings.
9699 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
9700 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9701 (cris_constant_alignment): New function.
9702 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
9703 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9704 (epiphany_constant_alignment): New function.
9705 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
9706 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9707 constant_alignment_word_strings.
9708 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
9709 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9710 constant_alignment_word_strings.
9711 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
9712 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9713 constant_alignment_word_strings.
9714 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
9715 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
9716 * config/i386/i386.c (ix86_constant_alignment): Make static.
9717 Use the same interface as the target hook.
9718 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9719 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
9720 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9721 constant_alignment_word_strings.
9722 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
9723 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
9724 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9725 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
9726 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9727 constant_alignment_word_strings.
9728 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
9729 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9730 constant_alignment_word_strings.
9731 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
9732 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9733 constant_alignment_word_strings.
9734 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
9735 * config/microblaze/microblaze.c (microblaze_constant_alignment):
9737 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9738 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
9739 * config/mips/mips.c (mips_constant_alignment): New function.
9740 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9741 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
9742 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
9743 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9744 (mmix_constant_alignment): Make static. Use the same interface
9746 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
9747 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9748 constant_alignment_word_strings.
9749 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
9750 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9751 constant_alignment_word_strings.
9752 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
9753 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9754 constant_alignment_word_strings.
9755 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
9756 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9757 (rs6000_constant_alignment): New function.
9758 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
9759 * config/riscv/riscv.c (riscv_constant_alignment): New function.
9760 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9761 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
9762 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9763 (rs6000_constant_alignment): New function.
9764 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
9765 * config/s390/s390.c (s390_constant_alignment): New function.
9766 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9767 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
9768 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9769 constant_alignment_word_strings.
9770 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
9771 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9772 (sparc_constant_alignment): New function.
9773 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
9774 * config/spu/spu.c (spu_constant_alignment): New function.
9775 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9776 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
9777 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9778 constant_alignment_word_strings.
9779 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
9780 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9781 constant_alignment_word_strings.
9782 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
9783 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9784 constant_alignment_word_strings.
9785 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
9786 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9787 (visium_constant_alignment): New function.
9788 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
9789 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9790 (xtensa_constant_alignment): New function.
9791 * system.h (CONSTANT_ALIGNMENT): Poison.
9793 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
9795 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9796 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
9797 (rs6000_builtin_valid_without_lhs): New helper function.
9798 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9799 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
9801 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9803 * target.h (vec_perm_indices): Use unsigned short rather than
9805 (auto_vec_perm_indices): Likewise.
9806 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
9807 Use unsigned int rather than unsigned char.
9808 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
9810 2017-09-25 Richard Biener <rguenther@suse.de>
9812 * cfgloop.h (sort_sibling_loops): Declare.
9813 * cfgloop.c (sort_sibling_loops_cmp): New helper.
9814 (sort_sibling_loops): New function sorting the sibling loop list
9816 * graphite.c (graphite_transform_loops): Sort sibling loops.
9818 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
9820 * target.def (vec_perm_const_ok): Change sel parameter to
9822 * optabs-query.c (can_vec_perm_p): Update accordingly.
9823 * doc/tm.texi: Regenerate.
9824 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
9825 auto_vec_perm_indices and remove separate nelt field.
9826 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
9827 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
9828 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
9829 (aarch64_expand_vec_perm_const): Update accordingly.
9830 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
9831 to vec_perm_indices.
9832 * config/arm/arm.c (expand_vec_perm_d): Change perm to
9833 auto_vec_perm_indices and remove separate nelt field.
9834 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
9835 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
9836 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
9838 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
9839 to vec_perm_indices.
9840 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
9841 sel to vec_perm_indices.
9842 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
9843 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
9844 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
9846 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
9849 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
9852 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
9853 on the FUNCTION_DECL function context if it has a DIE that is a
9856 2017-09-25 Richard Biener <rguenther@suse.de>
9858 PR tree-optimization/82285
9859 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
9862 2017-09-25 Tom de Vries <tom@codesourcery.com>
9866 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
9869 2017-09-25 Richard Biener <rguenther@suse.de>
9871 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
9872 ISL errors other than isl_error_quota happen. Dump if the
9873 schedule is the same.
9874 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
9875 errors instead of aborting inside ISL.
9877 2017-09-25 Iain Sandoe <iain@codesourcery.com>
9880 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
9881 of libgcc_eh for m64.
9882 * config/i386/darwin64.h: Likewise.
9884 2017-09-25 Richard Biener <rguenther@suse.de>
9887 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
9888 attribute for incomplete types nor twice for complete ones.
9890 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
9893 * config/i386/i386.c (ix86_print_operand_address_as): Encode
9894 %esp as %rsp to avoid 0x67 prefix if there is no index or base
9897 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
9900 * config/i386/i386.opt (mprefer-avx256): Use
9901 ix86_target_flags variable.
9902 * config/i386/i386.c (ix86_target_string): Move
9903 -mprefer-avx256 to flag2_opts.
9905 2017-09-22 Jakub Jelinek <jakub@redhat.com>
9908 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
9909 and x != -1 | y != -1 into (x & y) != -1.
9911 2017-09-22 Steve Ellcey <sellcey@cavium.com>
9913 * config.gcc: Add new case statement to set
9914 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
9915 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
9916 s390x-*-linux* case statements. Added aarch64 to the list of
9917 supported architectures.
9919 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9921 PR tree-optimization/82289
9922 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
9923 STMT_VINFO_RELEVANT_P.
9925 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9926 Alan Hayward <alan.hayward@arm.com>
9927 David Sherwood <david.sherwood@arm.com>
9929 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
9930 for VR_RANGE only; don't allow VR_ANTI_RANGE.
9931 (extract_range_from_binary_expr_1): Don't call
9932 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
9934 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9935 Alan Hayward <alan.hayward@arm.com>
9936 David Sherwood <david.sherwood@arm.com>
9938 * target.def (preferred_vector_alignment): New hook.
9939 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
9941 * doc/tm.texi: Regenerate.
9942 * targhooks.h (default_preferred_vector_alignment): Declare.
9943 * targhooks.c (default_preferred_vector_alignment): New function.
9944 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
9946 (DR_TARGET_ALIGNMENT): New macro.
9947 (aligned_access_p): Update commentary.
9948 (vect_known_alignment_in_bytes): New function.
9949 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
9951 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
9952 Calculate the misalignment based on the target alignment rather than
9954 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
9955 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
9956 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
9957 the target alignment, rather than masking the element misalignment
9958 with the number of elements in a vector. Also use the target
9959 alignment when calculating the maximum number of peels.
9960 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
9961 instead of TYPE_ALIGN_UNIT.
9962 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
9963 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
9964 (vect_create_addr_base_for_vector_ref): Update call accordingly.
9965 (vect_create_data_ref_ptr): Likewise.
9966 (vect_setup_realignment): Realign by ANDing with
9967 -DR_TARGET_MISALIGNMENT.
9968 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
9969 the number of peels based on DR_TARGET_ALIGNMENT.
9970 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
9971 with the guaranteed alignment boundary when deciding whether
9973 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
9974 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
9975 (ensure_base_align): Remove stmt_info parameter. Get the
9976 target base alignment from DR_TARGET_ALIGNMENT.
9977 (vectorizable_store): Update call accordingly. Interpret
9978 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
9980 (vectorizable_load): Likewise.
9982 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9983 Alan Hayward <alan.hayward@arm.com>
9984 David Sherwood <david.sherwood@arm.com>
9986 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
9987 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
9988 (vect_enhance_data_refs_alignment): Likewise.
9990 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9992 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9993 error. Only quit immediately if parsing is complete.
9994 (BEGIN): Initialize fatal_err and parse_done.
9995 (begin fpu, end fpu): Check number of arguments.
9996 (begin arch, end arch): Likewise.
9997 (begin cpu, end cpu): Likewise.
9998 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9999 (optalias): Likewise.
10001 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10003 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10004 * config/arm/arm-isa.h: Delete. Move definitions to ...
10005 * arm-cpus.in: ... here. Use new feature and fgroup values.
10006 * config/arm/arm.c (arm_option_override): Use lower case for feature
10008 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10009 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10010 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10012 (print_isa_bits_for): New function.
10013 (gen_isa): New function.
10014 (gen_comm_data): Use print_isa_bits_for.
10015 (define feature): New keyword.
10016 (define fgroup): New keyword.
10017 * config/arm/t-arm (TM_H): Remove.
10018 (GTM_H): Add arm-isa.h.
10019 (arm-isa.h): Add rule to generate file.
10020 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10021 case for feature bit names.
10023 2017-09-22 Richard Biener <rguenther@suse.de>
10025 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
10027 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
10028 print of no dependency loops ...
10029 * graphite.c (graphite_transform_loops): ... here.
10030 (canonicalize_loop_closed_ssa_form): Work from inner to outer
10032 (same_close_phi_node, remove_duplicate_close_phi,
10033 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
10034 (canonicalize_loop_closed_ssa): ... here and simplify.
10035 * graphite-optimize-isl.c: Include tree-vectorizer.h.
10036 (optimize_isl): Use dump_printf_loc to tell when we stopped
10037 optimizing because of an ISL timeout.
10039 2017-09-22 Richard Biener <rguenther@suse.de>
10041 PR tree-optimization/82291
10042 * tree-if-conv.c (predicate_mem_writes): Make sure to
10043 remove writes in blocks predicated with false.
10045 2017-09-22 Richard Biener <rguenther@suse.de>
10047 * sese.c: Include cfganal.h.
10048 (if_region_set_false_region): Remove.
10049 (create_if_region_on_edge): Likewise.
10050 (move_sese_in_condition): Re-implement without destroying
10053 2017-09-22 Richard Biener <rguenther@suse.de>
10055 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
10056 Verify both BBs contain loop PHI nodes before dispatching to
10057 copy_loop_phi_args.
10058 (graphite_regenerate_ast_isl): Do not recompute dominators,
10059 do not verify three times. Restructure for clarity.
10060 * graphite-scop-detection.c (same_close_phi_node,
10061 remove_duplicate_close_phi, make_close_phi_nodes_unique,
10062 defined_in_loop_p, canonicalize_loop_closed_ssa,
10063 canonicalize_loop_closed_ssa_form): Simplify, remove excess
10064 checking and SSA rewrite, move to ...
10065 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
10066 (graphite_initialize): Do not pass in ctx, do not reset the
10067 SCEV cache, compute only dominators.
10068 (graphite_transform_loops): Allocate ISL ctx after
10069 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
10070 Maintain post-dominators only around build_scops.
10071 * sese.c (if_region_set_false_region): Make static. Free
10072 and recompute dominators.
10073 (move_sese_in_condition): Assert we don't get called with
10074 post-dominators computed.
10075 * sese.h (if_region_set_false_region): Remove.
10077 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
10079 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
10080 mode attribute for TARGET_AVX512VL.
10082 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
10084 * config/i386/i386.opt (mprefer-avx256): New option.
10085 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
10087 (ix86_preferred_simd_mode): Return 256-bit AVX modes
10088 for TARGET_PREFER_AVX256.
10089 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
10091 2017-09-21 Jeff Law <law@redhat.com>
10093 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
10094 Fix dump output if the only stack space is for pushed registers.
10096 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10098 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
10101 2017-09-21 Martin Sebor <msebor@redhat.com>
10104 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
10105 code (in C++) or code that triggers warnings.
10107 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
10109 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
10111 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
10113 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
10114 * sched-rgn.c: Ditto.
10115 * sel-sched-ir.c: Ditto.
10117 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
10119 * toplev.h (set_random_seed): Adjust return type.
10120 * toplev.c (init_local_tick): Move eager initialization of random_seed
10121 to get_random_seed. Adjust comment.
10122 (init_random_seed): Inline to get_random_seed, delete.
10123 (get_random_seed): Initialize random_seed lazily.
10124 (set_random_seed): Do not return previous value.
10125 (print_switch_value): Do not call get_random_seed.
10127 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
10129 * cgraph.c (delete_function_version): New, broken out from...
10130 (cgraph_node::delete_function_version): ...here. Rename to
10131 cgraph_node::delete_function_version_by_decl. Update all uses.
10132 (cgraph_node::remove): Call delete_function_version.
10134 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10137 * tree-inline.c (expand_call_inline): Emit clobber stmts for
10138 VAR_DECLs to which addressable non-volatile parameters are mapped
10139 and for id->retvar after the return value assignment. Clear
10140 id->retval and id->retbnd after inlining.
10142 2017-09-21 Richard Biener <rguenther@suse.de>
10144 PR tree-optimization/82276
10145 PR tree-optimization/82244
10146 * tree-vrp.c (build_assert_expr_for): Set
10147 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
10149 (remove_range_assertions): Revert earlier change.
10151 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
10154 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
10156 2017-09-21 Richard Biener <rguenther@suse.de>
10158 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10159 Restore valid IL after code generation errors.
10160 * graphite.c (graphite_transform_loops): Diagnose code
10161 generation issues as MSG_MISSED_OPTIMIZATION and continue
10162 with processing SCOPs.
10164 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10165 Alan Hayward <alan.hayward@arm.com>
10166 David Sherwood <david.sherwood@arm.com>
10168 * calls.c (compute_argument_addresses): Use simplify_gen_binary
10169 rather than choosing between plus_constant and gen_rtx_<CODE>.
10170 * expr.c (emit_push_insn): Likewise.
10171 (expand_expr_real_2): Likewise.
10173 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10174 Alan Hayward <alan.hayward@arm.com>
10175 David Sherwood <david.sherwood@arm.com>
10177 * loop-unroll.c (split_iv): Call copy_rtx on the step.
10179 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10180 Alan Hayward <alan.hayward@arm.com>
10181 David Sherwood <david.sherwood@arm.com>
10183 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
10184 calling tree_to_uhwi.
10186 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10187 Alan Hayward <alan.hayward@arm.com>
10188 David Sherwood <david.sherwood@arm.com>
10190 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
10191 INTEGER_CST rather than a negative test for ADDR_EXPR.
10193 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10194 Alan Hayward <alan.hayward@arm.com>
10195 David Sherwood <david.sherwood@arm.com>
10197 * tree-vrp.c (extract_range_from_binary_expr_1): Check
10198 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
10200 2017-09-21 Richard Biener <rguenther@suse.de>
10202 PR tree-optimization/71351
10203 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
10204 graphite_create_new_loop_guard): Remove, fold remaining parts
10206 (translate_isl_ast_node_for): ... here and simplify.
10208 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10211 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
10212 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
10213 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
10214 alternative always use QI mode, for -Os imov (=R,R) alternative
10215 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
10218 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10219 Jeff Law <law@redhat.com>
10221 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
10222 (allocate_stack_space): New function, partially extracted from
10223 s390_emit_prologue.
10224 (s390_emit_prologue): Track offset to most recent stack probe.
10225 Code to allocate space moved into allocate_stack_space.
10226 Dump actions when no stack is allocated.
10227 (s390_prologue_plus_offset): New function.
10228 (s390_emit_stack_probe): Likewise.
10230 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
10232 * common.opt (Wa, Wl, Wp, g, gz=): Add
10234 (gno-column-info): Remove.
10235 (gcolumn-info): Drop RejectNegative.
10236 (gno-): New prefix.
10237 (gno-record-gcc-switches): Remove.
10238 (grecord-gcc-switches): Drop RejectNegative.
10239 (gno-split-dwarf): Remove.
10240 (gsplit-dwarf): Drop RejectNegative.
10241 (gno-strict-dwarf): Remove.
10242 (gstrict-dwarf): Drop RejectNegative.
10243 * config/darwin.opt (gfull, gused): Add RejectNegative.
10244 * dwarf2out.c (gen_producer_string): Drop
10245 gno-record-gcc-switches handler.
10246 * optc-gen.awk: Add g to prefixes with negative forms.
10247 * opts-common.c (remapping_prefix_p): New.
10248 (find_opt): Check it.
10249 (generate_canonical_option): Test g prefix.
10250 (option_map): Add -gno- mapping.
10251 (add_misspelling_candidates): Check remapping_prefix_p.
10253 2017-09-20 Jeff Law <law@redhat.com>
10255 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
10256 thinko in stack clash protection support.
10258 * explow.c (compute_stack_clash_protection_loop_data): Use
10259 CONST_INT_P instead of explicit test. Verify object is a
10260 CONST_INT_P before looking at INTVAL.
10261 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
10262 instead of explicit test.
10264 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
10267 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
10268 address instead of to r1 and r11.
10270 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
10272 * config.gcc: Support "knm".
10273 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
10274 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
10276 * config/i386/i386.c (m_KNM): Define.
10277 (processor_target_table): Add "knm".
10279 (ix86_option_override_internal): Add "knm".
10280 (ix86_issue_rate): Add PROCESSOR_KNM.
10281 (ix86_adjust_cost): Ditto.
10282 (ia32_multipass_dfa_lookahead): Ditto.
10283 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
10284 (fold_builtin_cpu): Add M_INTEL_KNM.
10285 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
10286 (processor_type): Add PROCESSOR_KNM.
10287 * config/i386/x86-tune.def: Add m_KNM.
10288 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
10290 2017-09-20 Richard Biener <rguenther@suse.de>
10292 PR tree-optimization/80213
10293 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
10294 are allowed in empty BBs as well.
10295 (canonicalize_loop_closed_ssa): Also look for other complex
10297 (scop_detection::get_sese): Include the loop-closed PHI block
10299 (scop_detection::merge_sese): Remove code adding extra blocks.
10300 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
10301 (build_scops): Assert the final returned scop is invalid.
10303 2017-09-20 Richard Biener <rguenther@suse.de>
10305 PR tree-optimization/82264
10306 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
10308 (vn_phi_lookup): Likewise.
10309 (vn_phi_insert): Likewise.
10311 2017-09-20 Jakub Jelinek <jakub@redhat.com>
10313 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
10314 that fits into uhwi or shwi, add DW_AT_const_value regardless
10315 of early_dwarf without going through RTL, using add_AT_unsigned
10318 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
10319 (DEBUG_LTO_ABBREV_SECTION): Likewise.
10320 (DEBUG_LTO_MACINFO_SECTION): Likewise.
10321 (DEBUG_MACRO_SECTION): Likewise.
10322 (DEBUG_LTO_MACRO_SECTION): Likewise.
10323 (DEBUG_STR_DWO_SECTION): Likewise.
10324 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
10325 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
10326 (DEBUG_LTO_DWO_LINE_SECTION): Define.
10327 (DEBUG_LTO_LINE_STR_SECTION): Define.
10328 (init_sections_and_labels): Initialize debug_line_str_section
10329 variable. Initialize debug_loc_section for -gdwarf-5 to
10330 DEBUG_LOCLISTS_SECTION. Formatting fixes.
10332 2017-09-20 Richard Biener <rguenther@suse.de>
10334 * graphite-sese-to-poly.c (extract_affine): Properly handle
10335 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
10337 2017-09-20 Richard Biener <rguenther@suse.de>
10339 PR tree-optimization/81373
10340 * graphite-scop-detection.c (build_cross_bb_scalars_def):
10341 Force SESE live-out defs to be handled even if they are
10344 2017-09-19 Jeff Law <law@redhat.com>
10346 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
10347 nothing for stack adjustments with REG_STACK_CHECK.
10348 * sched-deps.c (parse_add_or_inc): Reject insns with
10349 REG_STACK_CHECK from dependency breaking.
10350 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
10351 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
10352 * reg-notes.def (STACK_CHECK): New note.
10354 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
10355 (ix86_expand_prologue): Dump stack clash info as needed.
10356 Call ix86_adjust_stack_and_probe_stack_clash as needed.
10358 * function.c (dump_stack_clash_frame_info): New function.
10359 * function.h (dump_stack_clash_frame_info): Prototype.
10360 (enum stack_clash_probes): New enum.
10362 * config/alpha/alpha.c (alpha_expand_prologue): Also check
10363 flag_stack_clash_protection.
10364 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
10365 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
10366 (arm_frame_pointer_required): Likewise.
10367 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
10368 (ia64_expand_prologue): Likewise.
10369 * config/mips/mips.c (mips_expand_prologue): Likewise.
10370 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
10371 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
10372 (sparc_flat_expand_prologue): Likewise.
10373 * config/spu/spu.c (spu_expand_prologue): Likewise.
10375 * explow.c: Include "params.h".
10376 (anti_adjust_stack_and_probe_stack_clash): New function.
10377 (get_stack_check_protect): Likewise.
10378 (compute_stack_clash_protection_loop_data): Likewise.
10379 (emit_stack_clash_protection_loop_start): Likewise.
10380 (emit_stack_clash_protection_loop_end): Likewise.
10381 (allocate_dynamic_stack_space): Use get_stack_check_protect.
10382 Use anti_adjust_stack_and_probe_stack_clash.
10383 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
10384 (emit_stack_clash_protection_loop_start): Likewise.
10385 (emit_stack_clash_protection_loop_end): Likewise.
10386 * rtl.h (get_stack_check_protect): Prototype.
10387 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
10388 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
10389 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
10391 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
10393 * doc/tm.texi: Rebuilt.
10394 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
10395 get_stack_check_protect.
10396 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
10397 * config/arm/arm.c (arm_expand_prologue): Likewise.
10398 (arm_frame_pointer_required): Likewise.
10399 * config/i386/i386.c (ix86_expand_prologue): Likewise.
10400 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
10401 * config/mips/mips.c (mips_expand_prologue): Likewise.
10402 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
10403 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
10404 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
10405 (sparc_flat_expand_prologue): Likewise.
10407 * common.opt (-fstack-clash-protection): New option.
10408 * flag-types.h (enum stack_check_type): Note difference between
10409 -fstack-check= and -fstack-clash-protection.
10410 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
10411 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
10412 * toplev.c (process_options): Issue warnings/errors for cases
10413 not handled with -fstack-clash-protection.
10414 * doc/invoke.texi (-fstack-clash-protection): Document new option.
10415 (-fstack-check): Note additional problem with -fstack-check=generic.
10416 Note that -fstack-check is primarily for Ada and refer users
10417 to -fstack-clash-protection for stack-clash-protection.
10418 Document new params for stack clash protection.
10420 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
10422 * config/i386/i386.c (ix86_split_long_move): Do not handle
10423 address used for LEA in a special way.
10425 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
10427 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
10429 2017-09-19 Martin Sebor <msebor@redhat.com>
10432 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
10433 of incompatible types.
10435 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
10437 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10438 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
10439 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10440 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
10442 2017-09-19 Richard Biener <rguenther@suse.de>
10444 PR tree-optimization/82244
10445 * tree-vrp.c (remove_range_assertions): Do not propagate
10446 a constant to abnormals but replace the assert with a copy.
10448 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
10450 PR rtl-optimization/57878
10451 PR rtl-optimization/68988
10452 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
10453 avoidance test involving non_reload_pseudos. Move frequency test
10454 below the general fragmentation avoidance test.
10456 2017-09-19 Richard Biener <rguenther@suse.de>
10458 PR tree-optimization/69728
10459 * graphite-sese-to-poly.c (schedule_error): New global.
10460 (add_loop_schedule): Handle empty domain by failing the
10462 (build_original_schedule): Handle schedule_error.
10464 2017-09-19 Richard Biener <rguenther@suse.de>
10466 * graphite-scop-detection.c (scop_detection::can_represent_loop):
10467 Do not iterate to sibling loops but only to siblings of inner
10470 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
10473 * config/m68k/m68k.md (moveq feeding equality comparison): Check
10474 that the registers are different.
10476 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
10478 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
10479 to processor_model and "amdfam17h" to arch_names_table.
10480 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
10482 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10485 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
10487 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10488 Alan Hayward <alan.hayward@arm.com>
10489 David Sherwood <david.sherwood@arm.com>
10491 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
10493 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
10495 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
10496 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
10497 vect_schedule_slp_instance.
10498 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
10499 Update call to vect_slp_analyze_node_operations. Simplify return
10501 (vect_slp_analyze_bb_1): Update call accordingly.
10502 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
10503 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
10504 (vect_schedule_slp): Update call accordingly.
10506 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10507 Alan Hayward <alan.hayward@arm.com>
10508 David Sherwood <david.sherwood@arm.com>
10510 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
10511 with types that aren't in fact scalar.
10513 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10515 * tree-vect-slp.c (vect_record_max_nunits): New function,
10517 (vect_build_slp_tree_1): ...here.
10518 (vect_build_slp_tree_2): Call it for phis too.
10520 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10522 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
10523 to vect_get_vec_def_for_operand when getting the mask operand.
10525 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10526 Alan Hayward <alan.hayward@arm.com>
10527 David Sherwood <david.sherwood@arm.com>
10529 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
10532 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10533 Alan Hayward <alan.hayward@arm.com>
10534 David Sherwood <david.sherwood@arm.com>
10536 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
10537 calculation for vector booleans.
10539 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10540 Alan Hayward <alan.hayward@arm.com>
10541 David Sherwood <david.sherwood@arm.com>
10543 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
10545 (vect_transform_stmt): ...here.
10546 (vect_analyze_stmt): Use it instead of calling
10547 vectorizable_live_operation directly.
10549 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
10551 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
10552 non-SIMT targets in acc vector loops.
10554 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
10556 * configure.ac: Add arc and check if assembler supports gdwarf2.
10557 * configure: Regenerate.
10559 2017-09-18 Richard Biener <rguenther@suse.de>
10561 PR tree-optimization/82220
10562 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
10563 epilogue niters from the min_profitable_iters compute.
10565 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10568 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
10569 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
10570 (ix86_init_pic_reg): Revert 2017-09-01 changes.
10572 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
10575 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
10576 switching to a new text section.
10578 2017-09-18 Richard Biener <rguenther@suse.de>
10580 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
10582 (build_alias_set): Reject aliases with no access function.
10584 2017-09-18 Richard Biener <rguenther@suse.de>
10586 PR tree-optimization/79622
10587 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
10589 (build_cross_bb_scalars_use): Likewise.
10591 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
10593 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
10595 2017-09-18 Alan Modra <amodra@gmail.com>
10598 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
10599 stack_pointer_rtx for count 0. Update comments. Break up
10600 large rtl expression.
10602 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
10604 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
10605 Increase to 20 bytes.
10606 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
10607 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
10608 or avx version of the stub.
10610 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
10613 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
10614 compute the minimum stack alignment. Also update preferred stack
10615 boundary for leaf functions.
10617 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
10619 PR tree-optimization/82228
10620 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
10623 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10625 * common/config/nds32/nds32-common.c
10626 (nds32_option_optimization_table): Refine formatting.
10627 (nds32_option_optimization_table): Use -fsched-pressure and
10628 -fomit-frame-pointer for specific optimization level.
10630 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10632 * config/nds32/nds32.c: Refine formatting and comments.
10633 * config/nds32/nds32.h: Likewise.
10634 * config/nds32/nds32.md: Likewise.
10635 * config/nds32/nds32-cost.c: Likewise.
10636 * config/nds32/nds32-isr.c: Likewise.
10637 * config/nds32/nds32-md-auxiliary.c: Likewise.
10638 * config/nds32/nds32-multiple.md: Likewise.
10639 * config/nds32/nds32-predicates.c: Likewise.
10641 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
10642 Jakub Jelinek <jakub@redhat.com>
10644 Add support for -std=c++2a.
10645 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
10647 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
10649 2017-09-15 Steve Ellcey <sellcey@cavium.com>
10652 * doc/extend.texi (Common Function Attributes): Add
10653 references to ARM, AArch64, and S/390 specific attributes.
10654 (Function Specific Option Pragmas): Add AArch64 and S/390
10655 to list of back ends that support the target pragma.
10657 2017-09-15 Nathan Sidwell <nathan@acm.org>
10659 * doc/standards.texi: Fix C++17 description. Update URLs for
10662 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
10664 * common.opt (Wcast-align=strict): New warning option.
10665 * doc/invoke.texi: Document -Wcast-align=strict.
10667 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
10669 * cgraph.h (cgraph_thunk_info): Add comments.
10670 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
10671 assert for VIRTUAL_* arguments stricter.
10673 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
10675 PR tree-optimization/71026
10676 * match.pd: Move RDIV patterns from fold-const.c
10677 * fold-const.c (distribute_real_division): Removed.
10678 (fold_binary_loc): Remove calls to distribute_real_divison.
10680 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10682 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
10683 c++1z and gnu++1z as deprecated. Change other references to
10684 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
10685 Change -Wc++1z-compat to -Wc++17-compat.
10686 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
10687 * dwarf2out.c (highest_c_language): Handle C++17.
10688 (gen_compile_unit_die): Likewise.
10690 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10692 PR rtl-optimization/82192
10693 * combine.c (make_extraction): Don't look through non-paradoxical
10694 SUBREGs or TRUNCATE if pos + len is or might be bigger than
10697 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10698 Alan Hayward <alan.hayward@arm.com>
10699 David Sherwood <david.sherwood@arm.com>
10701 * target.def (function_arg_offset): New hook.
10702 * targhooks.h (default_function_arg_offset): Declare.
10703 * targhooks.c (default_function_arg_offset): New function.
10704 * function.c (locate_and_pad_parm): Use
10705 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
10706 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
10707 (TARGET_FUNCTION_ARG_OFFSET): ...this.
10708 * doc/tm.texi: Regenerate.
10709 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
10710 * config/spu/spu.c (spu_function_arg_offset): New function.
10711 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
10712 * system.h (FUNCTION_ARG_OFFSET): Poison.
10714 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10715 Alan Hayard <alan.hayward@arm.com>
10716 David Sherwood <david.sherwood@arm.com>
10718 * target.def (truly_noop_truncation): New hook.
10719 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
10720 than TRULY_NOOP_TRUNCATION.
10721 * hooks.h (hook_bool_uint_uint_true): Declare.
10722 * hooks.c (hook_bool_uint_uint_true): New function.
10723 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
10724 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
10725 * doc/tm.texi: Regenerate.
10726 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
10727 rather than TRULY_NOOP_TRUNCATION in comments.
10728 (simplify_comparison): Likewise.
10729 (record_truncated_value): Likewise.
10730 * expmed.c (extract_bit_field_1): Likewise.
10731 (extract_split_bit_field): Likewise.
10732 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
10733 instead of TRULY_NOOP_TRUNCATION.
10734 * function.c (assign_parm_setup_block): Likewise.
10735 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
10736 * rtlhooks.c: Include target.h.
10737 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
10738 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
10739 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
10740 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
10741 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
10742 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
10743 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
10744 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
10745 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
10746 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
10747 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
10748 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
10749 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
10750 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
10751 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
10752 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
10753 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
10754 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
10755 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
10756 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
10757 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
10758 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
10759 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
10760 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
10761 * config/mips/mips.c (mips_truly_noop_truncation): New function.
10762 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10763 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10764 rather than TRULY_NOOP_TRUNCATION in comments.
10765 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
10766 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
10767 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
10768 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
10769 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
10770 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
10771 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
10772 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
10773 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
10774 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
10775 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
10776 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10777 rather than TRULY_NOOP_TRUNCATION in comments.
10778 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
10779 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
10780 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
10781 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
10782 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
10783 TRULY_NOOP_TRUNCATION condition.
10784 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
10785 (TRULY_NOOP_TRUNCATION): Delete.
10786 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
10787 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
10788 * config/spu/spu.c (spu_truly_noop_truncation): New function.
10789 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10790 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
10791 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
10792 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
10793 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10794 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10795 rather than TRULY_NOOP_TRUNCATION in comments.
10796 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
10797 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
10798 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
10799 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
10800 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
10801 * system.h (TRULY_NOOP_TRUNCATION): Poison.
10803 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
10806 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
10807 (*cmp_ior): Likewise.
10808 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
10809 (*ior_scc_scc_cmp): Likewise.
10810 (*and_scc_scc): Likewise.
10811 (*and_scc_scc_cmp): Likewise.
10813 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10814 Alan Hayard <alan.hayward@arm.com>
10815 David Sherwood <david.sherwood@arm.com>
10817 * target.def (can_change_mode_class): New hook.
10818 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
10819 (hard_regno_nregs): Likewise.
10820 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
10821 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
10822 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
10823 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
10824 (LOAD_EXTEND_OP): Update accordingly.
10825 * doc/tm.texi: Regenerate.
10826 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
10827 CANNOT_CHANGE_MODE_CLASS.
10828 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
10829 (REG_CAN_CHANGE_MODE_P): ...this new macro.
10830 * combine.c (simplify_set): Update accordingly.
10831 * emit-rtl.c (validate_subreg): Likewise.
10832 * recog.c (general_operand): Likewise.
10833 * regcprop.c (mode_change_ok): Likewise.
10834 * reload1.c (choose_reload_regs): Likewise.
10835 (inherit_piecemeal_p): Likewise.
10836 * rtlanal.c (simplify_subreg_regno): Likewise.
10837 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
10838 instead of CANNOT_CHANGE_MODE_CLASS.
10839 (reload_cse_simplify_operands): Likewise.
10840 * reload.c (push_reload): Use targetm.can_change_mode_class
10841 instead of CANNOT_CHANGE_MODE_CLASS.
10842 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
10843 REG_CANNOT_CHANGE_MODE_P.
10844 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10845 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
10846 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10847 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10848 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10849 (arm_can_change_mode_class): New function.
10850 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10851 than CANNOT_CHANGE_MODE_CLASS in comments.
10852 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10853 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
10854 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
10855 (ix86_can_change_mode_class): ...this new function, inverting the
10856 sense of the return value.
10857 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10858 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10859 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10860 (ia64_can_change_mode_class): New function.
10861 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10862 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
10863 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
10864 (m32c_can_change_mode_class): ...this new function, inverting the
10865 sense of the return value.
10866 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10867 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10868 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
10869 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
10870 (mips_can_change_mode_class): ...this new function, inverting the
10871 sense of the return value.
10872 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10873 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10874 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10875 (msp430_can_change_mode_class): New function.
10876 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10877 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
10878 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10879 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10880 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10881 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
10882 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10883 (pa_cannot_change_mode_class): Replace with...
10884 (pa_can_change_mode_class): ...this new function, inverting the
10885 sense of the return value.
10886 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10887 than CANNOT_CHANGE_MODE_CLASS in comments.
10888 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10889 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
10890 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10891 (pdp11_cannot_change_mode_class): Replace with...
10892 (pdp11_can_change_mode_class): ...this new function, inverting the
10893 sense of the return value.
10894 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10895 * config/powerpcspe/powerpcspe-protos.h
10896 (rs6000_cannot_change_mode_class_ptr): Delete.
10897 * config/powerpcspe/powerpcspe.c
10898 (rs6000_cannot_change_mode_class_ptr): Delete.
10899 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10900 (rs6000_option_override_internal): Assign to
10901 targetm.can_change_mode_class instead of
10902 rs6000_cannot_change_mode_class_ptr.
10903 (rs6000_cannot_change_mode_class): Replace with...
10904 (rs6000_can_change_mode_class): ...this new function, inverting the
10905 sense of the return value.
10906 (rs6000_debug_cannot_change_mode_class): Replace with...
10907 (rs6000_debug_can_change_mode_class): ...this new function.
10908 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10909 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
10910 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10911 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10912 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
10914 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
10915 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10916 (rs6000_option_override_internal): Assign to
10917 targetm.can_change_mode_class instead of
10918 rs6000_cannot_change_mode_class_ptr.
10919 (rs6000_cannot_change_mode_class): Replace with...
10920 (rs6000_can_change_mode_class): ...this new function, inverting the
10921 sense of the return value.
10922 (rs6000_debug_cannot_change_mode_class): Replace with...
10923 (rs6000_debug_can_change_mode_class): ...this new function.
10924 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10925 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
10926 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
10927 (s390_can_change_mode_class): ...this new function, inverting the
10928 sense of the return value.
10929 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10930 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10931 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
10932 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10933 (sh_cannot_change_mode_class): Replace with...
10934 (sh_can_change_mode_class): ...this new function, inverting the
10935 sense of the return value.
10936 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10937 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10938 (sparc_can_change_mode_class): New function.
10939 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10940 * config/spu/spu.c (spu_can_change_mode_class): New function.
10941 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10942 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10943 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10944 (visium_can_change_mode_class): New function.
10945 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
10947 2017-09-15 Richard Biener <rguenther@suse.de>
10949 PR tree-optimization/82217
10950 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
10951 but not undefined case.
10953 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10956 * postreload.c (reload_cse_simplify_operands): Skip
10957 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
10959 2017-09-15 Richard Biener <rguenther@suse.de>
10961 PR tree-optimization/68823
10962 * graphite-scop-detection.c (build_alias_set): If we have a
10963 possible dependence check whether we can handle them by just
10964 looking at the DRs DR_ACCESS_FNs.
10965 (build_scops): If build_alias_set fails, fail the SCOP.
10967 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
10969 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
10970 to support float128 built-in functions that require the ISA 3.0
10972 (BU_FLOAT128_3_HW): Likewise.
10973 (SQRTF128): Add support for the IEEE 128-bit square root and fma
10974 built-in functions.
10975 (FMAF128): Likewise.
10977 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
10978 support for built-in functions that need the ISA 3.0 IEEE 128-bit
10979 floating point instructions.
10980 (rs6000_invalid_builtin): Likewise.
10981 (rs6000_builtin_mask_names): Likewise.
10982 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
10983 (RS6000_BTM_FLOAT128_HW): Likewise.
10984 (RS6000_BTM_COMMON): Likewise.
10985 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
10987 * doc/extend.texi (RS/6000 built-in functions): Document the
10988 IEEE 128-bit floating point square root and fused multiply-add
10989 built-in functions.
10991 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
10993 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
10994 reg (r2) isn't in the set of registers defined in the prologue.
10996 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10997 Alan Hayward <alan.hayward@arm.com>
10998 David Sherwood <david.sherwood@arm.com>
11000 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
11001 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
11002 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
11003 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
11004 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
11006 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
11007 max_vectorization_factor.
11008 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
11010 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11011 Alan Hayward <alan.hayward@arm.com>
11012 David Sherwood <david.sherwood@arm.com>
11014 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
11015 (vect_worthwhile_without_simd_p): Declare.
11016 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
11017 (vectorizable_reduction): Use it.
11018 * tree-vect-stmts.c (vectorizable_shift): Likewise.
11019 (vectorizable_operation): Likewise.
11021 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11022 Alan Hayward <alan.hayward@arm.com>
11023 David Sherwood <david.sherwood@arm.com>
11025 * tree-vectorizer.h (vect_get_num_copies): New function.
11026 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
11027 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11028 (vectorizable_induction): Likewise.
11029 (vectorizable_live_operation): Likewise.
11030 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11031 (vectorizable_bswap): Likewise.
11032 (vectorizable_call): Likewise.
11033 (vectorizable_conversion): Likewise.
11034 (vectorizable_assignment): Likewise.
11035 (vectorizable_shift): Likewise.
11036 (vectorizable_operation): Likewise.
11037 (vectorizable_store): Likewise.
11038 (vectorizable_load): Likewise.
11039 (vectorizable_condition): Likewise.
11040 (vectorizable_comparison): Likewise.
11041 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
11043 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11044 Alan Hayward <alan.hayward@arm.com>
11045 David Sherwood <david.sherwood@arm.com>
11047 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
11048 of vect_init_vector.
11050 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11051 Alan Hayward <alan.hayward@arm.com>
11052 David Sherwood <david.sherwood@arm.com>
11054 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
11055 an inline wrapper that provides a location.
11056 (gimple_build_vector): Likewise.
11057 * gimple-fold.c (gimple_build_vector_from_val): New function.
11058 (gimple_build_vector): Likewise.
11059 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
11060 functions to build the initial value. Always return a gimple value.
11061 (get_initial_defs_for_reduction): Likewise. Only compute
11063 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
11064 vect_init_vector on the results from get_initial_def(s)_for_reduction.
11065 (vectorizable_induction): Use gimple_build_vector rather than
11068 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11069 Alan Hayward <alan.hayward@arm.com>
11070 David Sherwood <david.sherwood@arm.com>
11072 * target.h (vec_perm_indices): New typedef.
11073 (auto_vec_perm_indices): Likewise.
11074 * optabs-query.h: Include target.h
11075 (can_vec_perm_p): Take a vec_perm_indices *.
11076 * optabs-query.c (can_vec_perm_p): Likewise.
11077 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
11078 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11079 * tree-vect-generic.c (lower_vec_perm): Likewise.
11080 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
11081 (vect_grouped_load_supported): Likewise.
11082 (vect_shift_permute_load_chain): Likewise.
11083 (vect_permute_store_chain): Use auto_vec_perm_indices.
11084 (vect_permute_load_chain): Likewise.
11085 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
11086 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
11087 Update uses of can_vec_perm_p.
11088 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
11089 mode with a number of elements. Take a vec_perm_indices *.
11090 (vect_create_epilog_for_reduction): Update accordingly.
11091 Use auto_vec_perm_indices.
11092 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
11093 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
11094 (vect_transform_slp_perm_load): Likewise.
11095 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
11096 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
11097 (vect_gen_perm_mask_checked): Likewise.
11098 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
11099 (vect_gen_perm_mask_checked): Likewise.
11100 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
11101 (vectorizable_store): Likewise.
11102 (vectorizable_load): Likewise.
11103 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
11104 (vectorizable_bswap): Likewise.
11106 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11107 Alan Hayward <alan.hayward@arm.com>
11108 David Sherwood <david.sherwood@arm.com>
11110 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
11111 * tree.c (build_vector): Likewise.
11112 (build_vector_from_ctor): Update accordingly.
11113 (build_vector_from_val): Likewise.
11114 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
11115 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11116 * tree-vect-generic.c (add_rshift): Likewise.
11117 (expand_vector_divmod): Likewise.
11118 (optimize_vector_constructor): Likewise.
11119 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
11120 (vect_transform_slp_perm_load): Likewise.
11121 (vect_schedule_slp_instance): Likewise.
11122 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
11123 (vectorizable_call): Likewise.
11124 (vect_gen_perm_mask_any): Likewise. Add elements in order.
11125 * expmed.c (make_tree): Likewise.
11126 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
11127 a vector passed to build_vector.
11128 (fold_convert_const): Likewise.
11129 (exact_inverse): Likewise.
11130 (fold_ternary_loc): Likewise.
11131 (fold_relational_const): Likewise.
11132 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
11133 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
11134 (const_unop): Likewise. Store the reduction accumulator in a
11135 variable rather than an array.
11136 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
11137 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
11138 the new vector, rather than constructing it after the input arrays.
11139 (native_interpret_vector): Use auto_vec<tree> when building
11140 a vector passed to build_vector. Add elements in order.
11141 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
11142 auto_vec<tree> when building a vector passed to build_vector.
11143 (vect_create_epilog_for_reduction): Likewise.
11144 (vectorizable_induction): Likewise.
11145 (get_initial_def_for_reduction): Likewise. Fix indentation of
11147 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
11149 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
11150 passed to build_vector.
11152 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11153 Alan Hayward <alan.hayward@arm.com>
11154 David Sherwood <david.sherwood@arm.com>
11156 * tree-core.h (tree_base::u): Add an "nelts" field.
11157 (tree_vector): Use VECTOR_CST_NELTS as the length.
11158 * tree.c (tree_size): Likewise.
11159 (make_vector): Initialize VECTOR_CST_NELTS.
11160 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
11161 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
11162 TYPE_VECTOR_SUBPARTS.
11163 * expr.c (const_vector_mask_from_tree): Consistently use "units"
11164 as the number of units, setting it from VECTOR_CST_NELTS.
11165 (const_vector_from_tree): Likewise.
11166 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
11167 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
11168 (fold_negate_expr_1): Likewise.
11169 (fold_convert_const): Likewise.
11170 (const_binop): Likewise. Differentiate the number of output and
11172 (const_unop): Likewise.
11173 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
11174 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
11175 in cases that did the opposite.
11177 2017-09-14 Richard Biener <rguenther@suse.de>
11179 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
11182 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
11184 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
11186 2017-09-14 Jakub Jelinek <jakub@redhat.com>
11189 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
11190 if and where to split a bb, except for splitting before debug insn
11191 sequences followed by non-label real insn. Delete debug insns
11192 in between basic blocks.
11194 * combine.c (make_compound_operation_int): Formatting fixes.
11196 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
11197 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11198 * config/netbsd.h (LINK_EH_SPEC): Likewise.
11199 * config/sol2.h (LINK_EH_SPEC): Likewise.
11200 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11201 * config/s390/linux.h (LINK_SPEC): Likewise.
11202 * config/freebsd.h (LINK_EH_SPEC): Likewise.
11203 * config/openbsd.h (LINK_EH_SPEC): Likewise.
11204 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11205 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
11206 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
11207 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11208 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
11209 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
11211 2017-09-13 Jakub Jelinek <jakub@redhat.com>
11213 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
11215 (ENDFILE_LINUX_SPEC): Likewise.
11216 (LINK_EH_SPEC): Likewise.
11217 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
11218 (LINK_OS_LINUX_SPEC32): Likewise.
11219 (LINK_OS_LINUX_SPEC64): Likewise.
11220 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
11221 (LINK_OS_LINUX_SPEC): Likewise.
11223 2017-09-13 Martin Liska <mliska@suse.cz>
11225 PR middle-end/82154
11226 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
11227 CASE_HIGH is NULL_TREE.
11229 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
11230 Alan Hayward <alan.hayward@arm.com>
11231 David Sherwood <david.sherwood@arm.com>
11233 * target.def (secondary_memory_needed): New hook.
11234 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
11235 instead of SECONDARY_MEMORY_NEEDED.
11236 (secondary_memory_needed_mode): Likewise.
11237 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
11238 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
11239 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
11240 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
11241 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
11242 * doc/tm.texi: Regenerate.
11243 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
11244 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
11245 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11246 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
11247 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
11248 * config/i386/i386.c (inline_secondary_memory_needed): Put the
11249 mode argument first and change the reg_class arguments to reg_class_t.
11250 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
11251 Make static. Update the call to inline_secondary_memory_needed.
11252 (ix86_register_move_cost): Update the call to
11253 inline_secondary_memory_needed.
11254 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11255 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
11257 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
11258 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11260 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
11261 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
11262 * config/mips/mips.c (mips_secondary_memory_needed): Make static
11263 and match hook interface. Add comment from mips.h.
11264 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11265 * config/mmix/mmix.md (truncdfsf2): Refer to
11266 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11268 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
11269 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
11270 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11271 (pa_secondary_memory_needed): New function.
11272 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
11273 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
11274 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11275 (pdp11_secondary_memory_needed): Make static and match hook interface.
11276 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
11277 * config/powerpcspe/powerpcspe-protos.h
11278 (rs6000_secondary_memory_needed_ptr): Delete.
11279 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
11281 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11282 (rs6000_option_override_internal): Assign to
11283 targetm.secondary_memory_needed rather than
11284 rs6000_secondary_memory_needed_ptr.
11285 (rs6000_secondary_memory_needed): Match hook interface.
11286 (rs6000_debug_secondary_memory_needed): Likewise.
11287 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
11288 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
11289 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
11290 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11291 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
11292 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
11294 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
11295 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11296 (rs6000_option_override_internal): Assign to
11297 targetm.secondary_memory_needed rather than
11298 rs6000_secondary_memory_needed_ptr.
11299 (rs6000_secondary_memory_needed): Match hook interface.
11300 (rs6000_debug_secondary_memory_needed): Likewise.
11301 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
11302 * config/s390/s390.c (s390_secondary_memory_needed): New function.
11303 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11304 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
11305 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11306 (sparc_secondary_memory_needed): New function.
11307 * lra-constraints.c (check_and_process_move): Refer to
11308 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11310 (curr_insn_transform): Likewise.
11311 (process_alt_operands): Use targetm.secondary_memory_needed
11312 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11313 (check_secondary_memory_needed_p): Likewise.
11314 (choose_split_class): Likewise.
11315 * reload.c: Unconditionally include code that was previously
11316 conditional on SECONDARY_MEMORY_NEEDED.
11317 (push_secondary_reload): Use targetm.secondary_memory_needed
11318 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11319 (push_reload): Likewise.
11320 * reload1.c: Unconditionally include code that was previously
11321 conditional on SECONDARY_MEMORY_NEEDED.
11322 (choose_reload_regs): Use targetm.secondary_memory_needed
11323 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11324 (gen_reload): Likewise.
11325 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
11327 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
11328 Alan Hayward <alan.hayward@arm.com>
11329 David Sherwood <david.sherwood@arm.com>
11331 * target.def (secondary_memory_needed_mode): New hook:
11332 * targhooks.c (default_secondary_memory_needed_mode): Declare.
11333 * targhooks.h (default_secondary_memory_needed_mode): New function.
11334 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
11335 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
11336 * doc/tm.texi: Regenerate.
11337 * lra-constraints.c (check_and_process_move): Use
11338 targetm.secondary_memory_needed_mode instead of
11339 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
11340 (curr_insn_transform): Likewise.
11341 * reload.c (get_secondary_mem): Likewise.
11342 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11343 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
11345 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11346 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11347 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
11348 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11349 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
11351 * config/powerpcspe/powerpcspe-protos.h
11352 (rs6000_secondary_memory_needed_mode): Delete.
11353 * config/powerpcspe/powerpcspe.c
11354 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11355 (rs6000_secondary_memory_needed_mode): Make static.
11356 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11357 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
11359 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
11361 (rs6000_secondary_memory_needed_mode): Make static.
11362 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11363 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
11364 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11365 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11366 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
11368 (sparc_secondary_memory_needed_mode): New function.
11369 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
11371 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
11373 * config/aarch64/constraints.md (Umq): New constraint.
11374 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
11376 (mov<mode>): Update condition.
11378 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11380 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
11381 when bitposition is the same.
11383 2017-09-13 Richard Biener <rguenther@suse.de>
11385 * dwarf2out.c (output_die_symbol): Remove.
11386 (output_die): Do not output a DIEs symbol.
11388 2017-09-13 Richard Biener <rguenther@suse.de>
11390 PR middle-end/82128
11391 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
11392 default-def to avoid breaking iterator update with the weird
11393 interaction with cgraph_update_edges_for_call_stmt_node.
11395 2017-09-13 Richard Biener <rguenther@suse.de>
11397 * tree-cfg.c (verify_gimple_assign_binary): Add verification
11398 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
11399 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
11400 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
11402 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
11404 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
11405 Disable pc relative literal load irrespective of
11406 TARGET_FIX_ERR_A53_84341 for default.
11408 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
11410 * config/sparc/sparc.c (output_return): Output the source location of
11411 the insn in the delay slot, if any.
11412 (output_sibcall): Likewise.
11414 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
11417 * common.opt (-static-pie): New alias.
11418 (shared): Negate static-pie.
11419 (-no-pie): Update help text.
11421 (static-pie): New option.
11422 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
11423 -static-pie support.
11424 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
11425 (LINK_EH_SPEC): Likewise.
11426 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11427 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11428 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11429 * gcc.c (LINK_COMMAND_SPEC): Likewise.
11430 (init_gcc_specs): Likewise.
11431 (init_spec): Likewise.
11432 (display_help): Update help message for -pie.
11433 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
11436 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
11438 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
11439 (movdi_aarch64): Likewise.
11440 (movti_aarch64): Likewise.
11442 2017-09-12 Simon Wright <simon@pushface.org>
11445 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
11446 calculation of the minor version, always output as 0.
11448 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11451 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
11452 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
11453 on it early, rather than manual conversion late. For
11454 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
11455 instead of performing manual conversion.
11457 2017-09-12 Carl Love <cel@us.ibm.com>
11459 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
11460 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
11462 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11463 VMULOSW): Add definitions.
11464 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11465 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11466 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11467 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11468 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11470 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11472 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
11474 (movti_aarch64): Likewise.
11475 (movdf_aarch64): Likewise.
11476 (movtf_aarch64): Likewise.
11477 (load_pairdi): Likewise.
11478 (store_pairdi): Likewise.
11479 (load_pairdf): Likewise.
11480 (store_pairdf): Likewise.
11481 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
11482 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
11483 (ldr_got_small_<mode>): Likewise.
11484 (ldr_got_small_28k_<mode>): Likewise.
11485 (ldr_got_tiny): Likewise.
11486 * config/aarch64/iterators.md (ldst_sz): New.
11487 (ldpstp_sz): Likewise.
11488 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
11490 (thunderx_load): Split load_8 to load_16.
11491 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
11493 (thunderx2t99_storepair_basic): Split store_8 to store_16.
11494 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
11495 (xgene1_store_pair): Split store_8 to store_16.
11496 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
11497 (falkor_st_0_st_sd): Split store_8 to store_16.
11499 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11501 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
11502 and store1/2/3/4 to store_4/8/12/16.
11503 * config/aarch64/aarch64.md: Update for rename.
11504 * config/arm/arm.md: Likewise.: Likewise.
11505 * config/arm/arm.c: Likewise.
11506 * config/arm/thumb1.md: Likewise.
11507 * config/arm/thumb2.md: Likewise.
11508 * config/arm/vfp.md: Likewise.
11509 * config/arm/arm-generic.md: Likewise.
11510 * config/arm/arm1020e.md: Likewise.
11511 * config/arm/arm1026ejs.md: Likewise.
11512 * config/arm/arm1136jfs.md: Likewise.
11513 * config/arm/arm926ejs.md: Likewise.
11514 * config/arm/cortex-a15.md: Likewise.
11515 * config/arm/cortex-a17.md: Likewise.
11516 * config/arm/cortex-a5.md: Likewise.
11517 * config/arm/cortex-a53.md: Likewise.
11518 * config/arm/cortex-a57.md: Likewise.
11519 * config/arm/cortex-a7.md: Likewise.
11520 * config/arm/cortex-a8.md: Likewise.
11521 * config/arm/cortex-a9.md: Likewise.
11522 * config/arm/cortex-m4.md: Likewise.
11523 * config/arm/cortex-m7.md: Likewise.
11524 * config/arm/cortex-r4.md: Likewise.
11525 * config/arm/exynos-m1.md: Likewise.
11526 * config/arm/fa526.md: Likewise.
11527 * config/arm/fa606te.md: Likewise.
11528 * config/arm/fa626te.md: Likewise.
11529 * config/arm/fa726te.md: Likewise.
11530 * config/arm/fmp626.md: Likewise.
11531 * config/arm/iwmmxt.md: Likewise.
11532 * config/arm/ldmstm.md: Likewise.
11533 * config/arm/marvell-pj4.md: Likewise.
11534 * config/arm/xgene1.md: Likewise.
11535 * config/aarch64/thunderx.md: Likewise.
11536 * config/aarch64/thunderx2t99.md: Likewise.
11537 * config/aarch64/falkor.md: Likewise.
11539 2017-09-12 Martin Liska <mliska@suse.cz>
11541 * attribs.c (private_lookup_attribute): New function.
11542 * attribs.h (private_lookup_attribute): Declared here.
11543 (lookup_attribute): Called from this place.
11545 2017-09-12 Richard Biener <rguenther@suse.de>
11547 PR tree-optimization/82157
11548 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
11549 stmts with side-effects.
11551 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11552 Alan Hayward <alan.hayward@arm.com>
11553 David Sherwood <david.sherwood@arm.com>
11555 * target.def (hard_regno_nregs): New hook.
11556 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
11557 * targhooks.h (default_hard_regno_nregs): Declare.
11558 * targhooks.c (default_hard_regno_nregs): New function.
11559 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
11560 (TARGET_HARD_REGNO_NREGS): ...this hook.
11561 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
11562 (CLASS_MAX_NREGS): Likewise.
11563 * doc/tm.texi: Regenerate.
11564 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
11565 instead of HARD_REGNO_NREGS.
11566 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
11567 HARD_REGNO_NREGS in the comment.
11568 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
11569 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
11570 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
11571 Return an unsigned int.
11572 (TARGET_HARD_REGNO_NREGS): Redefine.
11573 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
11574 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
11575 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11576 (arc_hard_regno_nregs): New function.
11577 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
11578 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
11579 (arm_hard_regno_nregs): New function.
11580 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
11581 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
11582 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
11583 (TARGET_HARD_REGNO_NREGS): Redefine.
11584 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
11585 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
11586 (HARD_REGNO_NREGS): Delete.
11587 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
11588 (cr16_hard_regno_nregs): New function.
11589 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
11590 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
11591 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
11592 (cris_hard_regno_nregs): New function.
11593 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
11594 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
11595 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11596 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
11597 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
11598 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
11599 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
11600 (frv_hard_regno_nregs): Make static. Take and return an
11602 (frv_class_max_nregs): Remove outdated copy of documentation.
11603 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
11604 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
11605 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
11606 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
11607 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
11608 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
11609 (TARGET_HARD_REGNO_NREGS): Redefine.
11610 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
11611 (CLASS_MAX_NREGS): Update comment.
11612 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
11613 (ia64_hard_regno_nregs): New function.
11614 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
11615 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
11616 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
11617 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
11618 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
11620 (m32c_hard_regno_nregs): Likewise. Make static.
11621 (TARGET_HARD_REGNO_NREGS): Redefine.
11622 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
11623 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
11624 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
11625 (m68k_hard_regno_nregs): New function.
11626 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
11627 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
11628 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
11629 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
11630 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
11631 Take and return an unsigned int.
11632 (TARGET_HARD_REGNO_NREGS): Redefine.
11633 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
11634 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11635 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
11636 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
11637 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
11638 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
11639 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
11640 (msp430_hard_regno_nregs): Make static. Take and return an
11642 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
11643 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
11644 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
11645 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
11646 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
11647 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
11648 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
11649 (TARGET_HARD_REGNO_NREGS): Redefine.
11650 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
11651 (PA_HARD_REGNO_NREGS): ...this.
11652 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
11653 (PA_HARD_REGNO_NREGS): ...this.
11654 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
11655 (pa_hard_regno_nregs): New function.
11656 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
11657 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
11658 (pdp11_hard_regno_nregs): New function.
11659 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
11660 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
11661 (rs6000_hard_regno_nregs_hook): New function.
11662 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
11663 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
11664 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
11665 Take and return an unsigned int. Move earlier in file.
11666 (TARGET_HARD_REGNO_NREGS): Redefine.
11667 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
11668 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
11669 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
11670 (rl78_hard_regno_nregs): Make static. Take and return an
11672 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
11673 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
11674 (rs6000_hard_regno_nregs_hook): New function.
11675 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
11676 * config/rx/rx.c (rx_hard_regno_nregs): New function.
11677 (TARGET_HARD_REGNO_NREGS): Redefine.
11678 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
11679 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
11680 instead of HARD_REGNO_NREGS.
11681 (s390_hard_regno_nregs): New function.
11682 (s390_hard_regno_mode_ok): Add comment from s390.h.
11683 (TARGET_HARD_REGNO_NREGS): Redefine.
11684 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
11685 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
11686 (sh_hard_regno_nregs): New function.
11687 (sh_pass_in_reg_p): Use it.
11688 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
11689 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11690 (sparc_hard_regno_nregs): New function.
11691 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
11692 * config/spu/spu.c (spu_hard_regno_nregs): New function.
11693 (spu_function_arg_advance): Use it, supplying a valid register number.
11694 (TARGET_HARD_REGNO_NREGS): Redefine.
11695 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
11696 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
11697 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
11698 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
11699 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
11700 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
11701 (CLASS_MAX_NREGS): Remove copy of old documentation.
11702 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
11703 (visium_hard_regno_nregs): New function.
11704 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
11705 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
11706 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
11707 xtensa_hard_regno_nregs): New function.
11708 * system.h (HARD_REGNO_NREGS): Poison.
11710 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11712 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
11713 hard_regno_nregs instead of HARD_REGNO_NREGS.
11714 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
11715 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
11716 (c6x_expand_epilogue): Likewise.
11717 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
11718 (frv_read_iacc_argument): Likewise.
11719 * config/sh/sh.c: Include regs.h.
11720 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
11721 (regs_used): Likewise.
11722 (output_stack_adjust): Likewise.
11723 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
11724 * expmed.c: Include regs.h.
11725 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
11726 * ree.c: Include regs.h.
11727 (combine_reaching_defs): Use hard_regno_nregs instead of
11729 (add_removable_extension): Likewise.
11731 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11733 * regs.h (hard_regno_nregs): Turn into a function.
11734 (end_hard_regno): Update accordingly.
11735 * caller-save.c (setup_save_areas): Likewise.
11736 (save_call_clobbered_regs): Likewise.
11737 (replace_reg_with_saved_mem): Likewise.
11738 (insert_restore): Likewise.
11739 (insert_save): Likewise.
11740 * combine.c (can_change_dest_mode): Likewise.
11741 (move_deaths): Likewise.
11742 (distribute_notes): Likewise.
11743 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
11744 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
11745 (rs6000_split_multireg_move): Likewise.
11746 (rs6000_register_move_cost): Likewise.
11747 (rs6000_memory_move_cost): Likewise.
11748 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
11749 (rs6000_split_multireg_move): Likewise.
11750 (rs6000_register_move_cost): Likewise.
11751 (rs6000_memory_move_cost): Likewise.
11752 * cselib.c (cselib_reset_table): Likewise.
11753 (cselib_lookup_1): Likewise.
11754 * emit-rtl.c (set_mode_and_regno): Likewise.
11755 * function.c (aggregate_value_p): Likewise.
11756 * ira-color.c (setup_profitable_hard_regs): Likewise.
11757 (check_hard_reg_p): Likewise.
11758 (calculate_saved_nregs): Likewise.
11759 (assign_hard_reg): Likewise.
11760 (improve_allocation): Likewise.
11761 (calculate_spill_cost): Likewise.
11762 * ira-emit.c (modify_move_list): Likewise.
11763 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
11764 (ira_hard_reg_in_set_p): Likewise.
11765 * ira.c (setup_reg_mode_hard_regset): Likewise.
11766 (clarify_prohibited_class_mode_regs): Likewise.
11767 (check_allocation): Likewise.
11768 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11769 (lra_setup_reg_renumber): Likewise.
11770 (setup_try_hard_regno_pseudos): Likewise.
11771 (spill_for): Likewise.
11772 (assign_hard_regno): Likewise.
11773 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
11774 * lra-constraints.c (in_class_p): Likewise.
11775 (lra_constraint_offset): Likewise.
11776 (simplify_operand_subreg): Likewise.
11777 (lra_constraints): Likewise.
11778 (split_reg): Likewise.
11779 (split_if_necessary): Likewise.
11780 (invariant_p): Likewise.
11781 (inherit_in_ebb): Likewise.
11782 * lra-lives.c (process_bb_lives): Likewise.
11783 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
11784 (get_hard_regs): Likewise.
11785 (do_remat): Likewise.
11786 * lra-spills.c (assign_spill_hard_regs): Likewise.
11787 * mode-switching.c (create_pre_exit): Likewise.
11788 * postreload.c (reload_combine_recognize_pattern): Likewise.
11789 * recog.c (peep2_find_free_register): Likewise.
11790 * regcprop.c (kill_value_regno): Likewise.
11791 (set_value_regno): Likewise.
11792 (copy_value): Likewise.
11793 (maybe_mode_change): Likewise.
11794 (find_oldest_value_reg): Likewise.
11795 (copyprop_hardreg_forward_1): Likewise.
11796 * regrename.c (check_new_reg_p): Likewise.
11797 (regrename_do_replace): Likewise.
11798 * reload.c (push_reload): Likewise.
11799 (combine_reloads): Likewise.
11800 (find_dummy_reload): Likewise.
11801 (operands_match_p): Likewise.
11802 (find_reloads): Likewise.
11803 (find_equiv_reg): Likewise.
11804 (reload_adjust_reg_for_mode): Likewise.
11805 * reload1.c (count_pseudo): Likewise.
11806 (count_spilled_pseudo): Likewise.
11807 (find_reg): Likewise.
11808 (clear_reload_reg_in_use): Likewise.
11809 (free_for_value_p): Likewise.
11810 (allocate_reload_reg): Likewise.
11811 (choose_reload_regs): Likewise.
11812 (reload_adjust_reg_for_temp): Likewise.
11813 (emit_reload_insns): Likewise.
11814 (delete_output_reload): Likewise.
11815 * rtlanal.c (subreg_get_info): Likewise.
11816 * sched-deps.c (sched_analyze_reg): Likewise.
11817 * sel-sched.c (init_regs_for_mode): Likewise.
11818 (mark_unavailable_hard_regs): Likewise.
11819 (choose_best_reg_1): Likewise.
11820 (verify_target_availability): Likewise.
11821 * valtrack.c (dead_debug_insert_temp): Likewise.
11822 * var-tracking.c (track_loc_p): Likewise.
11823 (emit_note_insn_var_location): Likewise.
11824 * varasm.c (make_decl_rtl): Likewise.
11825 * reginfo.c (choose_hard_reg_mode): Likewise.
11826 (init_reg_modes_target): Refer directly to
11827 this_target_regs->x_hard_regno_nregs.
11829 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11831 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
11832 instead of hard_regno_nregs.
11834 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11836 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
11837 end_hard_regno instead of hard_regno_nregs.
11838 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
11839 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
11840 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
11841 * ira-color.c (improve_allocation): Likewise.
11842 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11843 * lra-lives.c (mark_regno_live): Likewise.
11844 (mark_regno_dead): Likewise.
11845 * lra-remat.c (operand_to_remat): Likewise.
11846 * lra.c (collect_non_operand_hard_regs): Likewise.
11847 * postreload.c (reload_combine_note_store): Likewise.
11848 (move2add_valid_value_p): Likewise.
11849 * reload.c (regno_clobbered_p): Likewise.
11851 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11853 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
11855 * config/v850/v850.c (v850_reorg): Likewise.
11856 * reload.c (refers_to_regno_for_reload_p): Likewise.
11857 (find_equiv_reg): Likewise.
11858 * reload1.c (reload_reg_reaches_end_p): Likewise.
11860 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11862 * caller-save.c (add_used_regs): Use REG_NREGS instead of
11864 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
11865 * config/arm/arm.c (output_move_neon): Likewise.
11866 (arm_attr_length_move_neon): Likewise.
11867 (neon_split_vcombine): Likewise.
11868 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
11869 (c6x_mark_reg_written): Likewise.
11870 (c6x_dwarf_register_span): Likewise.
11871 * config/i386/i386.c (ix86_save_reg): Likewise.
11872 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
11873 (rws_access_reg): Likewise.
11874 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
11875 * mode-switching.c (create_pre_exit): Likewise.
11876 * ree.c (combine_reaching_defs): Likewise.
11877 (add_removable_extension): Likewise.
11878 * regcprop.c (find_oldest_value_reg): Likewise.
11879 (copyprop_hardreg_forward_1): Likewise.
11880 * reload.c (reload_inner_reg_of_subreg): Likewise.
11881 (push_reload): Likewise.
11882 (combine_reloads): Likewise.
11883 (find_dummy_reload): Likewise.
11884 (reload_adjust_reg_for_mode): Likewise.
11885 * reload1.c (find_reload_regs): Likewise.
11886 (forget_old_reloads_1): Likewise.
11887 (reload_reg_free_for_value_p): Likewise.
11888 (reload_adjust_reg_for_temp): Likewise.
11889 (emit_reload_insns): Likewise.
11890 (delete_output_reload): Likewise.
11891 * sel-sched.c (choose_best_reg_1): Likewise.
11892 (choose_best_pseudo_reg): Likewise.
11894 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11895 Alan Hayward <alan.hayward@arm.com>
11896 David Sherwood <david.sherwood@arm.com>
11898 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
11899 * target.def (slow_unaligned_access): New hook.
11900 * targhooks.h (default_slow_unaligned_access): Declare.
11901 * targhooks.c (default_slow_unaligned_access): New function.
11902 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
11903 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
11904 * doc/tm.texi: Regenerate.
11905 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
11906 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
11907 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
11909 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
11910 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
11912 (rs6000_slow_unaligned_access): New function.
11913 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
11914 (expand_block_compare): Likewise.
11915 (expand_strn_compare): Likewise.
11916 (rs6000_rtx_costs): Likewise.
11917 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
11918 (riscv_slow_unaligned_access): Likewise.
11919 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
11920 (riscv_slow_unaligned_access_p): ...this and make static.
11921 (riscv_option_override): Update accordingly.
11922 (riscv_slow_unaligned_access): New function.
11923 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
11924 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
11925 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
11926 (rs6000_slow_unaligned_access): New function.
11927 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
11928 (rs6000_rtx_costs): Likewise.
11929 * config/rs6000/rs6000-string.c (expand_block_compare)
11930 (expand_strn_compare): Use targetm.slow_unaligned_access instead
11931 of SLOW_UNALIGNED_ACCESS.
11932 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
11933 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
11934 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
11935 of SLOW_UNALIGNED_ACCESS.
11936 * expmed.c (simple_mem_bitfield_p): Likewise.
11937 * expr.c (alignment_for_piecewise_move): Likewise.
11938 (emit_group_load_1): Likewise.
11939 (emit_group_store): Likewise.
11940 (copy_blkmode_from_reg): Likewise.
11941 (emit_push_insn): Likewise.
11942 (expand_assignment): Likewise.
11943 (store_field): Likewise.
11944 (expand_expr_real_1): Likewise.
11945 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
11946 * lra-constraints.c (simplify_operand_subreg): Likewise.
11947 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11948 * gimple-ssa-store-merging.c: Likewise in block comment at start
11950 * tree-ssa-strlen.c: Include target.h.
11951 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
11952 of SLOW_UNALIGNED_ACCESS.
11953 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
11955 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11957 PR rtl-optimization/82185
11958 * expmed.c (emit_store_flag_int): Only test tem if it has been
11961 2017-09-12 Richard Biener <rguenther@suse.de>
11963 PR middle-end/82149
11964 * match.pd ((FTYPE) N CMP CST): Fix typo.
11966 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
11968 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
11970 (mips_near_type_p): Add 'short_call' attribute as a synonym
11972 * doc/extend.texi (short_call): Document new function attribute.
11974 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11977 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
11978 assertion check that in the condition.
11979 (get_atomic_generic_size): Likewise. Before testing if parameter
11980 has pointer type, if it has array type, call for C++
11981 default_conversion to perform array-to-pointer conversion.
11983 2017-09-12 Richard Biener <rguenther@suse.de>
11985 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
11986 for operations we cannot scalarize.
11988 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
11990 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
11991 vectors heap vectors. Clean up comments.
11992 Make visited_bbs a reference.
11993 (profitable_jump_thread_path): Make GC
11994 vectors heap vectors. Clean up comments.
11996 (convert_and_register_jump_thread_path): Make GC vectors heap
11998 (check_subpath_and_update_thread_path): Same. Clean up comments.
11999 Make visited_bbs a reference.
12000 (handle_phi): Abstract common code to to
12001 register_jump_thread_path_if_profitable.
12002 Rename VAR_BB to DEF_BB.
12004 Make GC vectors heap vectors.
12005 Make visited_bbs a reference.
12006 (handle_assignment): Same.
12007 (register_jump_thread_path_if_profitable): New.
12008 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
12010 Make GC vectors heap vectors. Clean up comments.
12011 Make visited_bbs a reference.
12012 (find_jump_threads_backwards): Make visited_bbs live in the stack.
12013 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
12016 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
12019 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
12020 words of E_DImode object are reachable by xtensa_uimm8x4 access.
12022 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
12024 Revert r251800 and r251799.
12026 2017-09-11 Martin Jambor <mjambor@suse.cz>
12029 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
12030 arguments in advance.
12031 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
12032 use it to find predecessor edges.
12033 (naive_outof_ssa): Collect vector of predecessors.
12035 2017-09-08 Jason Merrill <jason@redhat.com>
12037 PR c++/70029 - ICE with ref-qualifier and -flto
12038 * langhooks.h (struct lang_hooks_for_types): Add
12039 copy_lang_qualifiers.
12040 * attribs.c (build_type_attribute_qual_variant): Use it.
12041 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
12043 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
12044 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
12046 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
12049 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
12050 (*mulsi3_sp64): New instruction.
12051 (mulsi3): New expander.
12053 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
12055 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
12057 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12059 * sancov.c: Include memmodel.h.
12061 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
12064 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
12067 2017-09-07 Carl Love <cel@us.ibm.com>
12069 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
12070 the sldi instruction.
12072 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
12074 * sancov.c: Include tm_p.h.
12076 2017-09-07 Jakub Jelinek <jakub@redhat.com>
12079 * output.h (switch_to_other_text_partition): New declaration.
12080 * varasm.c (switch_to_other_text_partition): New function.
12081 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
12082 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
12083 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
12084 to the other text partition before emitting LCL label and switch back
12085 after emitting the word after it.
12087 2017-09-07 Richard Biener <rguenther@suse.de>
12089 * passes.def (pass_split_crit_edges): Remove instance before PRE.
12090 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
12091 critical edges here, after loop init.
12092 (pass_data_pre): Remove PROP_no_crit_edges flags.
12093 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
12094 for valueization of call args to avoid leaking VN_TOP.
12095 (visit_use): Assert we do not visit default defs.
12096 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
12097 Use error_mark_node to more easily detect leaking VN_TOP.
12098 All default-defs are varying, not VN_TOP. Mark them visited.
12099 (run_scc_vn): Make code match comment.
12101 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12103 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
12104 OPTION_MASK_FLOAT128_KEYWORD.
12105 (POWERPC_MASKS): Likewise.
12106 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
12107 support for the -mfloat128-type option, and make -mfloat128
12108 default on PowerPC Linux systems. Define or undefine
12109 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
12110 Define __float128 to be __ieee128 if IEEE 128-bit support is
12111 enabled, or undefine it.
12112 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
12113 Delete defining __FLOAT128_TYPE__.
12114 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
12115 -mfloat128-type option and make -mfloat128 default on PowerPC
12117 (TARGET_FLOAT128_TYPE): Likewise.
12118 (-mfloat128-type): Likewise.
12119 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12120 Delete the -mfloat128-type option and make -mfloat128 default on
12121 PowerPC Linux systems. Always use __ieee128 to be the keyword for
12122 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
12123 128-bit floating point is enabled. Change tests from using
12124 -mfloat128-type to -mfloat128.
12125 (rs6000_mangle_type): Use the correct mangling for the __float128
12126 type even if normal long double is restricted to 64-bits.
12127 (floatn_mode): Enable the _Float128 type by default on VSX Linux
12129 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
12130 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
12131 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
12133 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
12134 documentation for -mfloat128.
12136 2017-09-06 Olivier Hainque <hainque@adacore.com>
12138 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
12140 2017-09-06 Wish Wu <wishwu007@gmail.com>
12141 Jakub Jelinek <jakub@redhat.com>
12143 * asan.c (initialize_sanitizer_builtins): Add
12144 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
12145 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
12146 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
12147 BT_FN_VOID_UINT64_PTR variables.
12148 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
12149 (BT_FN_VOID_UINT16_UINT16): Likewise.
12150 (BT_FN_VOID_UINT32_UINT32): Likewise.
12151 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
12152 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
12153 (BT_FN_VOID_UINT64_PTR): Likewise.
12154 * common.opt (flag_sanitize_coverage): New variable.
12155 (fsanitize-coverage=trace-pc): Remove.
12156 (fsanitize-coverage=): Add.
12157 * flag-types.h (enum sanitize_coverage_code): New enum.
12158 * fold-const.c (fold_range_test): Disable non-short-circuit
12159 optimization if flag_sanitize_coverage.
12160 (fold_truth_andor): Likewise.
12161 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
12162 * opts.c (COVERAGE_SANITIZER_OPT): Define.
12163 (coverage_sanitizer_opts): New array.
12164 (get_closest_sanitizer_option): Add OPTS argument, handle also
12165 OPT_fsanitize_coverage_.
12166 (parse_sanitizer_options): Adjusted to also handle
12167 OPT_fsanitize_coverage_.
12168 (common_handle_option): Add OPT_fsanitize_coverage_.
12169 * sancov.c (instrument_comparison, instrument_switch): New function.
12170 (sancov_pass): Add trace-cmp support.
12171 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
12172 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
12173 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
12174 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
12175 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
12176 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
12177 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
12178 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
12179 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
12180 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
12182 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12184 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12185 error. Only quit immediately if parsing is complete.
12186 (BEGIN): Initialize fatal_err and parse_done.
12187 (begin fpu, end fpu): Check number of arguments.
12188 (begin arch, end arch): Likewise.
12189 (begin cpu, end cpu): Likewise.
12190 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12191 (optalias): Likewise.
12193 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12195 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12196 * config/arm/arm-isa.h: Delete. Move definitions to ...
12197 * arm-cpus.in: ... here. Use new feature and fgroup values.
12198 * config/arm/arm.c (arm_option_override): Use lower case for feature
12200 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12201 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12202 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12204 (print_isa_bits_for): New function.
12205 (gen_isa): New function.
12206 (gen_comm_data): Use print_isa_bits_for.
12207 (define feature): New keyword.
12208 (define fgroup): New keyword.
12209 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
12210 (arm-isa.h): Add rule to generate file.
12211 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12212 case for feature bit names.
12214 2017-09-06 Richard Biener <rguenther@suse.de>
12216 * tree-ssa-pre.c (NECESSARY): Remove.
12217 (create_expression_by_pieces): Do not touch pass-local flags.
12218 (insert_into_preds_of_block): Likewise.
12219 (do_pre_regular_insertion): Likewise.
12220 (eliminate_insert): Likewise.
12221 (eliminate_dom_walker::before_dom_children): Likewise.
12222 (fini_eliminate): Do not look at inserted_exprs.
12223 (mark_operand_necessary): Remove.
12224 (remove_dead_inserted_code): Replace with simple work-list
12225 algorithm based on inserted_exprs and SSA uses.
12226 (pass_pre::execute): Re-order fini_eliminate and
12227 remove_dead_inserted_code.
12229 2017-09-06 Olivier Hainque <hainque@adacore.com>
12231 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
12232 for VxWorks 7. Adjust surrounding comments.
12234 2017-09-06 Richard Biener <rguenther@suse.de>
12236 * gimple-ssa-strength-reduction.c
12237 (find_candidates_dom_walker::before_dom_children): Also allow
12240 2017-09-06 Richard Biener <rguenther@suse.de>
12242 PR tree-optimization/82108
12243 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
12244 for gap in the non-permutation SLP case.
12246 2017-09-06 Martin Jambor <mjambor@suse.cz>
12248 PR tree-optimization/82078
12249 * tree-sra.c (sort_and_splice_var_accesses): Move call to
12250 add_access_to_work_queue...
12251 (build_accesses_from_assign): ...here.
12252 (propagate_all_subaccesses): Make sure racc is the group
12253 representative, if there is one.
12255 2017-09-06 Jakub Jelinek <jakub@redhat.com>
12257 PR middle-end/82095
12258 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
12261 2017-09-06 Richard Biener <rguenther@suse.de>
12263 * gimple-ssa-strength-reduction.c
12264 (find_candidates_dom_walker::before_doom_children): Use a
12265 type and not a mode check.
12267 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
12270 * config/arm/predicates.md (arm_general_adddi_operand): Create new
12272 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
12274 2017-09-05 Jeff Law <law@redhat.com>
12276 PR tree-optimization/64910
12277 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
12278 cases where we have 3 or more operands.
12280 2017-09-05 Jakub Jelinek <jakub@redhat.com>
12282 PR middle-end/81768
12283 * omp-low.c (lower_omp_for): Recompute tree invariant if
12284 gimple_omp_for_initial/final is ADDR_EXPR.
12286 PR middle-end/81768
12287 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
12288 into gimple val before gimplification fo the COND_EXPR.
12290 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
12292 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
12293 REGION_COPY argument.
12294 (thread_through_all_blocks): Remove unused argument to
12295 duplicate_thread_path.
12297 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12298 Alan Hayward <alan.hayward@arm.com>
12299 David Sherwood <david.sherwood@arm.com>
12301 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
12302 Take a scalar_mode rather than a machine_mode.
12303 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12304 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
12305 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12306 (aarch64_gen_adjusted_ldpstp): Likewise.
12307 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
12309 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12310 Alan Hayward <alan.hayward@arm.com>
12311 David Sherwood <david.sherwood@arm.com>
12313 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
12314 Take a scalar_int_mode instead of a machine_mode.
12315 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
12316 (aarch64_output_scalar_simd_mov_immediate): Likewise.
12317 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
12318 (aarch64_simd_attr_length_rglist): Delete.
12319 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
12320 a scalar_int_mode instead of a machine_mode.
12321 (aarch64_add_offset): Likewise.
12322 (aarch64_internal_mov_immediate): Likewise
12323 (aarch64_add_constant_internal): Likewise.
12324 (aarch64_add_constant): Likewise.
12325 (aarch64_movw_imm): Likewise.
12326 (aarch64_rtx_arith_op_extract_p): Likewise.
12327 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
12328 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
12329 Remove assert that the mode isn't a vector.
12330 (aarch64_output_scalar_simd_mov_immediate): Likewise.
12331 (aarch64_expand_mov_immediate): Update calls after above changes.
12332 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
12333 (aarch64_and_bitmask_imm): Check for scalar integer modes.
12334 (aarch64_move_imm): Likewise.
12335 (aarch64_can_const_movi_rtx_p): Likewise.
12336 (aarch64_strip_extend): Likewise.
12337 (aarch64_extr_rtx_p): Likewise.
12338 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
12339 a CONST_INT when the mode parameter is VOIDmode.
12340 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
12342 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12344 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
12345 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12346 (bitwise_type_for_mode): Update accordingly.
12348 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12350 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
12351 * stor-layout.c (mode_for_size_tree): Likewise.
12352 (mode_for_array): Update accordingly.
12353 (layout_decl): Likewise.
12354 (compute_record_mode): Likewise. Only set the mode once.
12356 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12358 * target.def (get_mask_mode): Change return type to opt_mode.
12360 * doc/tm.texi: Regenerate.
12361 * targhooks.h (default_get_mask_mode): Return an opt_mode.
12362 * targhooks.c (default_get_mask_mode): Likewise.
12363 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12364 * optabs-query.c (can_vec_mask_load_store_p): Update use of
12365 targetm.get_mask_mode.
12366 * tree.c (build_truth_vector_type): Likewise.
12368 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12370 * machmode.h (mode_for_vector): Return an opt_mode.
12371 * stor-layout.c (mode_for_vector): Likewise.
12372 (mode_for_int_vector): Update accordingly.
12373 (layout_type): Likewise.
12374 * config/i386/i386.c (emit_memmov): Likewise.
12375 (ix86_expand_set_or_movmem): Likewise.
12376 (ix86_expand_vector_init): Likewise.
12377 (ix86_get_mask_mode): Likewise.
12378 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
12380 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
12381 * expmed.c (extract_bit_field_1): Likewise.
12382 * expr.c (expand_expr_real_2): Likewise.
12383 * optabs-query.c (can_vec_perm_p): Likewise.
12384 (can_vec_mask_load_store_p): Likewise.
12385 * optabs.c (expand_vec_perm): Likewise.
12386 * targhooks.c (default_get_mask_mode): Likewise.
12387 * tree-vect-stmts.c (vectorizable_store): Likewise.
12388 (vectorizable_load): Likewise.
12389 (get_vectype_for_scalar_type_and_size): Likewise.
12391 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12393 * machmode.h (mode_for_int_vector): New function.
12394 * stor-layout.c (mode_for_int_vector): Likewise.
12395 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
12396 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
12397 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12398 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12399 (s390_expand_vcond): Likewise.
12401 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12403 * machmode.h (opt_machine_mode): New type.
12404 (opt_mode<T>): Allow construction from anything that can be
12406 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
12407 (mode_for_size): Return an opt_machine_mode.
12408 * stor-layout.c (mode_for_size): Likewise.
12409 (mode_for_size_tree): Update call accordingly.
12410 (bitwise_mode_for_mode): Likewise.
12411 (make_fract_type): Likewise.
12412 (make_accum_type): Likewise.
12413 * caller-save.c (replace_reg_with_saved_mem): Update call
12415 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12416 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12417 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12418 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12419 * expmed.c (extract_bit_field_1): Likewise.
12420 * reload.c (get_secondary_mem): Likewise.
12421 * varasm.c (assemble_integer): Likewise.
12422 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
12425 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12427 * machmode.h (decimal_float_mode_for_size): New function.
12428 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
12429 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
12430 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
12431 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
12432 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
12433 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
12435 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12437 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
12438 (get_builtin_sync_mode): Likewise.
12439 (expand_ifn_atomic_compare_exchange): Likewise.
12440 (expand_builtin_atomic_clear): Likewise.
12441 (expand_builtin_atomic_test_and_set): Likewise.
12442 (fold_builtin_atomic_always_lock_free): Likewise.
12443 * calls.c (compute_argument_addresses): Likewise.
12444 (emit_library_call_value_1): Likewise.
12445 (store_one_arg): Likewise.
12446 * combine.c (combine_instructions): Likewise.
12447 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
12448 * config/arm/arm.c (arm_function_value): Likewise.
12449 (aapcs_allocate_return_reg): Likewise.
12450 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
12451 * config/i386/i386.c (construct_container): Likewise.
12452 (ix86_gimplify_va_arg): Likewise.
12453 (ix86_expand_sse_cmp): Likewise.
12454 (emit_memmov): Likewise.
12455 (emit_memset): Likewise.
12456 (expand_small_movmem_or_setmem): Likewise.
12457 (ix86_expand_pextr): Likewise.
12458 (ix86_expand_pinsr): Likewise.
12459 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
12460 * config/microblaze/microblaze.c (microblaze_block_move_straight):
12462 * config/mips/mips.c (mips_function_value_1) Likewise.
12463 (mips_block_move_straight): Likewise.
12464 (mips_expand_ins_as_unaligned_store): Likewise.
12465 * config/powerpcspe/powerpcspe.c
12466 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12467 (rs6000_darwin64_record_arg_flush): Likewise.
12468 * config/rs6000/rs6000.c
12469 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12470 (rs6000_darwin64_record_arg_flush): Likewise.
12471 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
12472 (sparc_function_value_1): Likewise.
12473 * config/spu/spu.c (adjust_operand): Likewise.
12474 (spu_emit_branch_or_set): Likewise.
12475 (arith_immediate_p): Likewise.
12476 * emit-rtl.c (gen_lowpart_common): Likewise.
12477 * expr.c (expand_expr_real_1): Likewise.
12478 * function.c (assign_parm_setup_block): Likewise.
12479 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
12480 * reload1.c (alter_reg): Likewise.
12481 * stor-layout.c (mode_for_vector): Likewise.
12482 (layout_type): Likewise.
12484 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12486 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
12487 (spu_convert_move): Likewise.
12488 * lower-subreg.c (resolve_simple_move): Likewise.
12490 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12493 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
12494 define_insn to a define_expand.
12495 (altivec_vsum2sws_direct): New define_insn.
12496 (altivec_vsumsws): Convert from a define_insn to a define_expand.
12498 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12500 * config/arm/arm.c (arm_option_params_internal): Improve setting of
12503 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
12508 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
12509 realign stack if stack alignment needed is less than incoming
12512 2017-09-05 Marek Polacek <polacek@redhat.com>
12515 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
12518 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12520 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
12522 2017-09-05 Richard Biener <rguenther@suse.de>
12524 PR tree-optimization/82084
12525 * fold-const.c (can_native_encode_string_p): Handle wide characters.
12527 2017-09-05 Richard Biener <rguenther@suse.de>
12529 PR tree-optimization/82102
12530 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
12532 2017-09-05 Martin Liska <mliska@suse.cz>
12534 PR tree-optimization/82032
12535 * tree-cfg.c (generate_range_test): New function.
12536 * tree-cfg.h (generate_range_test): Declared here.
12537 * tree-cfgcleanup.c (convert_single_case_switch): New function.
12538 (cleanup_control_expr_graph): Use it.
12539 * tree-switch-conversion.c (try_switch_expansion): Remove
12541 (emit_case_nodes): Use generate_range_test.
12543 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12546 * config/i386/i386.md (*<btsc><mode>_mask): Add
12547 TARGET_USE_BT to insn constraint.
12548 (*btr<mode>_mask): Ditto.
12550 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
12552 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
12553 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
12555 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
12558 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
12559 TARGET_NEON and TARGET_IWMMXT.
12560 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
12561 TARGET_NEON and TARGET_IWMMXT.
12562 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
12564 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12566 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
12567 (ix86_rewrite_tls_address): Ditto.
12568 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
12569 (ix86_rewrite_tls_address_1): Ditto.
12570 (ix86_rewrite_tls_address): Ditto.
12571 * config/i386/predicates.md (tls_address_pattern): New predicate.
12572 * config/i386/i386.md (TLS address splitter): New splitter.
12574 2017-09-04 Richard Biener <rguenther@suse.de>
12576 PR tree-optimization/82084
12577 * fold-const.h (can_native_encode_string_p): Declare.
12578 * fold-const.c (can_native_encode_string_p): Factor out from ...
12579 (native_encode_string): ... here.
12580 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
12581 vectorizing stores from constants we later cannot handle.
12583 2017-09-04 Marek Polacek <polacek@redhat.com>
12586 * doc/invoke.texi: Update -Wtautological-compare documentation.
12588 2017-09-04 Jeff Law <law@redhat.com>
12590 PR tree-optimization/64910
12591 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
12592 swap the first and last operand if the last is a constant.
12594 2017-09-04 Marek Polacek <polacek@redhat.com>
12597 * convert.c (do_narrow): When sanitizing signed integer overflows,
12598 bail out for signed types.
12599 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
12601 2017-09-04 Richard Biener <rguenther@suse.de>
12603 PR tree-optimization/82060
12604 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
12605 Move devirtualization after stmt folding and before EH/AB/noreturn
12606 cleanup to get the stmt refs canonicalized. Use a bool instead
12607 of gimple_modified_p since that doesn't work for NOPs. Schedule
12608 NOPs generated by folding for removal.
12610 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12611 Alan Hayward <alan.hayward@arm.com>
12612 David Sherwood <david.sherwood@arm.com>
12614 * coretypes.h (pad_direction): New enum.
12615 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
12616 (FUNCTION_ARG_PADDING): Likewise.
12617 * target.def (function_arg_padding): New hook.
12618 * targhooks.h (default_function_arg_padding): Declare.
12619 * targhooks.c (default_function_arg_padding): New function.
12620 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
12621 (TARGET_FUNCTION_ARG_PADDING): ...this.
12622 * doc/tm.texi: Regenerate.
12623 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
12624 instead of direction.
12625 (compute_argument_addresses): Likewise.
12626 (load_register_parameters): Likewise.
12627 (emit_library_call_value_1): Likewise.
12628 (store_one_arg): Use targetm.calls.function_arg_padding instead
12629 of FUNCTION_ARG_PADDING.
12630 (must_pass_in_stack_var_size_or_pad): Likewise.
12631 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
12632 (emit_group_store): Likewise.
12633 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
12634 instead of FUNCTION_ARG_PADDING.
12635 (emit_push_insn): Likewise, and propagate enum change throughout
12637 * function.h (direction): Delete.
12638 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
12640 * function.c (assign_parm_find_stack_rtl): Likewise.
12641 (assign_parm_setup_block_p): Likewise.
12642 (assign_parm_setup_block): Likewise.
12643 (gimplify_parameters): Likewise.
12644 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
12645 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
12647 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
12648 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12649 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
12650 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
12651 (aarch64_function_arg_padding): ...this new function.
12652 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
12653 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12654 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
12655 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12656 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
12657 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12658 (arm_pad_arg_upward): Replace with...
12659 (arm_function_arg_padding): ...this new function.
12660 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
12662 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
12663 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
12664 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12665 (ia64_hpux_function_arg_padding): Replace with...
12666 (ia64_function_arg_padding): ...this new function. Use pad_direction
12667 instead of direction. Check for TARGET_HPUX.
12668 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
12669 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12670 (iq2000_function_arg_padding): New function.
12671 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
12672 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
12673 (mips_function_arg_padding): ...this new function.
12674 (mips_pad_reg_upward): Update accordingly.
12675 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12676 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
12677 targetm.calls.function_arg_padding.
12678 (FUNCTION_ARG_PADDING): Delete.
12679 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12680 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
12681 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12682 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
12683 (nios2_block_reg_padding): Return pad_direction instead of direction.
12684 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
12685 instead of direction.
12686 (nios2_function_arg_padding): Likewise. Make static.
12687 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12688 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
12689 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
12690 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
12691 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12692 (pa_function_arg_padding): Make static. Return pad_direction instead
12694 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
12695 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12696 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
12697 instead of direction. Use targetm.calls.function_arg_padding.
12698 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
12699 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
12700 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
12701 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
12702 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
12704 (function_arg_padding): Rename to...
12705 (rs6000_function_arg_padding): ...this. Make static. Return
12706 pad_direction instead of direction.
12707 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
12708 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
12709 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12710 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
12711 instead of direction. Use targetm.calls.function_arg_padding.
12712 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
12713 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
12714 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
12715 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
12716 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12717 (function_arg_padding): Rename to...
12718 (rs6000_function_arg_padding): ...this. Make static. Return
12719 pad_direction instead of direction.
12720 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
12721 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
12722 * config/s390/s390.c (s390_function_arg_padding): New function.
12723 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12724 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
12725 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
12726 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12727 (function_arg_padding): Rename to...
12728 (sparc_function_arg_padding): ...this. Make static. Return
12729 pad_direction instead of direction.
12730 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
12731 * config/spu/spu.c (spu_function_arg_padding): New function.
12732 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12733 * system.h (FUNCTION_ARG_PADDING): Poison.
12735 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12736 Alan Hayward <alan.hayward@arm.com>
12737 David Sherwood <david.sherwood@arm.com>
12739 * target.def (modes_tieable_p): New hook.
12740 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
12741 (TARGET_MODES_TIEABLE_P): ...this.
12742 * doc/tm.texi.in: Regenerate.
12743 * hooks.h (hook_bool_mode_mode_true): Declare.
12744 * hooks.c (hook_bool_mode_mode_true): New function.
12745 * combine.c (subst): Use targetm.modes_tieable_p instead of
12747 * dse.c (find_shift_sequence): Likewise.
12748 * expmed.c (extract_low_bits): Likewise.
12749 * lower-subreg.c: Include target.h.
12750 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
12752 * rtlanal.c (rtx_cost): Likewise.
12753 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
12754 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
12755 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
12756 (TARGET_MODES_TIEABLE_P): Redefine.
12757 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
12758 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
12759 (TARGET_MODES_TIEABLE_P): Redefine.
12760 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
12761 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
12762 (arc_modes_tieable_p): New function.
12763 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
12764 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
12765 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
12766 (arm_modes_tieable_p): Make static.
12767 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
12768 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
12769 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
12770 (TARGET_MODES_TIEABLE_P): Redefine.
12771 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
12772 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
12773 (TARGET_MODES_TIEABLE_P): Redefine.
12774 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
12775 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
12776 (cr16_modes_tieable_p): New function.
12777 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
12778 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
12779 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
12780 (TRULY_NOOP_TRUNCATION): Update comment.
12781 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
12782 (TRULY_NOOP_TRUNCATION): Update comment.
12783 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
12784 (frv_modes_tieable_p): New function.
12785 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
12786 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
12787 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
12788 (TARGET_MODES_TIEABLE_P): Redefine.
12789 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
12790 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
12791 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
12792 (TARGET_MODES_TIEABLE_P): Redefine.
12793 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
12794 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
12795 (ia64_modes_tieable_p): New function.
12796 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
12797 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
12798 (iq2000_modes_tieable_p): New function.
12799 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
12800 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
12801 (lm32_modes_tieable_p): New function.
12802 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
12803 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
12804 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
12805 (TARGET_MODES_TIEABLE_P): Redefine.
12806 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
12807 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
12808 (m32r_modes_tieable_p): New function.
12809 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
12810 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
12811 (m68k_modes_tieable_p): New function.
12812 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
12813 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
12814 (mcore_modes_tieable_p): New function.
12815 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
12816 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
12818 (TARGET_MODES_TIEABLE_P): Redefine.
12819 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
12820 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
12821 * config/mips/mips.c (mips_modes_tieable_p): Make static.
12822 (TARGET_MODES_TIEABLE_P): Redefine.
12823 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
12824 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
12825 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
12826 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
12827 (mn10300_modes_tieable_p): ...this and make static.
12828 (TARGET_MODES_TIEABLE_P): Redefine.
12829 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
12830 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
12831 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
12832 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
12833 (msp430_modes_tieable_p): Make static.
12834 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
12835 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
12836 (TARGET_MODES_TIEABLE_P): Redefine.
12837 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
12838 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
12839 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
12840 (TARGET_MODES_TIEABLE_P): Redefine.
12841 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
12842 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
12843 * config/pa/pa.c (pa_modes_tieable_p): Make static.
12844 (TARGET_MODES_TIEABLE_P): Redefine.
12845 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
12846 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
12847 (pdp11_modes_tieable_p): New function.
12848 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
12849 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
12850 (rs6000_modes_tieable_p): New function.
12851 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12852 * config/powerpcspe/powerpcspe.md: Update comment.
12853 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
12854 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
12855 (TARGET_MODES_TIEABLE_P): Redefine.
12856 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
12857 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
12858 (rl78_modes_tieable_p): New function.
12859 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
12860 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
12861 (rs6000_modes_tieable_p): New function.
12862 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12863 * config/rs6000/rs6000.md: Update comment.
12864 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
12865 * config/rx/rx.c (rx_modes_tieable_p): New function.
12866 (TARGET_MODES_TIEABLE_P): Redefine.
12867 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
12868 * config/s390/s390.c (s390_modes_tieable_p): New function.
12869 (TARGET_MODES_TIEABLE_P): Redefine.
12870 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
12871 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
12872 (sh_modes_tieable_p): New function.
12873 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
12874 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
12875 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
12876 (sparc_modes_tieable_p): Make static.
12877 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
12878 * config/spu/spu.c (spu_modes_tieable_p): New function.
12879 (TARGET_MODES_TIEABLE_P): Redefine.
12880 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
12881 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
12882 (TARGET_MODES_TIEABLE_P): Redefine.
12883 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
12884 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
12885 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
12886 * config/v850/v850.c (v850_modes_tieable_p): New function.
12887 (TARGET_MODES_TIEABLE_P): Redefine.
12888 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
12889 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
12890 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
12891 (visium_modes_tieable_p): New function.
12892 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
12893 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
12894 (xtensa_modes_tieable_p): New function.
12895 * system.h (MODES_TIEABLE_P): Poison.
12897 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12898 Alan Hayward <alan.hayward@arm.com>
12899 David Sherwood <david.sherwood@arm.com>
12901 * target.def (hard_regno_mode_ok): New hook.
12902 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
12903 (TARGET_HARD_REGNO_MODE_OK): ...this.
12904 * doc/tm.texi.in: Regenerate.
12905 * hooks.h (hook_bool_uint_mode_true): Declare.
12906 * hooks.c (hook_bool_uint_mode_true): New function.
12907 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
12908 HARD_REGNO_MODE_OK.
12909 * genpreds.c (write_insn_preds_c): Add an include of target.h.
12910 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
12911 instead of HARD_REGNO_MODE_OK.
12912 * caller-save.c: Include target.h.
12913 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
12914 HARD_REGNO_MODE_OK.
12915 * combine.c (can_combine_p): Likewise.
12916 (combinable_i3pat): Likewise.
12917 (can_change_dest_mode): Likewise.
12918 * expr.c (init_expr_target): Likewise.
12919 (convert_move): Likewise.
12920 (convert_modes): Likewise.
12921 * ira.c (setup_prohibited_class_mode_regs): Likewise.
12922 (setup_prohibited_mode_move_regs): Likewise.
12923 * ira.h (target_ira): Likewise.
12924 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12925 * lra-constraints.c (process_alt_operands): Likewise.
12926 (split_reg): Likewise.
12927 * recog.c (peep2_find_free_register): Likewise.
12928 * ree.c (combine_reaching_defs): Likewise.
12929 * regcprop.c (maybe_mode_change): Likewise.
12930 * reginfo.c (init_reg_sets_1): Likewise.
12931 (choose_hard_reg_mode): Likewise.
12932 (simplifiable_subregs): Likewise.
12933 * regrename.c (check_new_reg_p): Likewise.
12934 * reload.c (find_valid_class): Likewise.
12935 (find_valid_class_1): Likewise.
12936 (reload_inner_reg_of_subreg): Likewise.
12937 (push_reload): Likewise.
12938 (combine_reloads): Likewise.
12939 (find_dummy_reload): Likewise.
12940 (find_reloads): Likewise.
12941 * reload1.c (find_reg): Likewise.
12942 (set_reload_reg): Likewise.
12943 (allocate_reload_reg): Likewise.
12944 (choose_reload_regs): Likewise.
12945 (reload_adjust_reg_for_temp): Likewise.
12946 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
12947 (simplify_subreg_regno): Likewise.
12948 * sel-sched.c (init_regs_for_mode): Likewise.
12949 * varasm.c (make_decl_rtl): Likewise.
12950 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
12951 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
12952 HARD_REGNO_MODE_OK.
12953 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
12954 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
12955 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12956 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
12957 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
12958 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12959 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
12960 (arc_mode_class): Delete.
12961 (HARD_REGNO_MODE_OK): Delete.
12962 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12963 (arc_hard_regno_mode_ok): Rename old array to...
12964 (arc_hard_regno_mode_ok_modes): ...this.
12965 (arc_conditional_register_usage): Update accordingly.
12966 (arc_mode_class): Make static.
12967 (arc_hard_regno_mode_ok): New function.
12968 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
12969 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
12970 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12971 (arm_hard_regno_mode_ok): Make static.
12972 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
12973 HARD_REGNO_MODE_OK.
12974 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
12975 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
12976 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
12978 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12979 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
12980 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
12981 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
12982 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
12983 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12984 * config/bfin/predicates.md (valid_reg_operand): Use
12985 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
12986 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
12987 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
12988 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12989 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
12990 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
12991 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12992 (cr16_hard_regno_mode_ok): Make static and return a bool.
12993 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
12994 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12995 (cris_hard_regno_mode_ok): New function.
12996 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
12997 (epiphany_mode_class): Delete.
12998 (HARD_REGNO_MODE_OK): Delete.
12999 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
13000 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13001 (hard_regno_mode_ok): Rename to...
13002 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
13003 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
13004 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
13005 HARD_REGNO_MODE_OK.
13006 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
13007 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
13008 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13009 (frv_hard_regno_mode_ok): Make static and return a bool.
13010 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
13011 HARD_REGNO_MODE_OK.
13012 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
13013 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
13014 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
13015 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
13017 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13018 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
13019 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
13020 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
13022 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13023 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
13024 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13025 (ia64_hard_regno_mode_ok): New function.
13026 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
13027 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13028 (iq2000_hard_regno_mode_ok): New function.
13029 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
13030 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13031 (lm32_hard_regno_mode_ok): New function.
13032 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
13033 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
13034 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
13035 instead of HARD_REGNO_MODE_OK.
13036 (m32c_hard_regno_ok): Rename to...
13037 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
13038 (m32c_cannot_change_mode_class): Update accordingly.
13039 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13040 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
13041 (m32r_mode_class): Delete.
13042 (HARD_REGNO_MODE_OK): Delete.
13043 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13044 (m32r_hard_regno_mode_ok): Rename to...
13045 (m32r_hard_regno_modes): ...this.
13046 (m32r_mode_class): Make static.
13047 (m32r_hard_regno_mode_ok): New function.
13048 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
13049 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
13050 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13051 (m68k_hard_regno_mode_ok): Make static.
13052 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
13053 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13054 (mcore_hard_regno_mode_ok): New function.
13055 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
13056 (HARD_REGNO_MODE_OK): Delete.
13057 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
13059 (microblaze_hard_regno_mode_ok_p): ...this and make static.
13060 (microblaze_hard_regno_mode_ok): New function.
13061 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13062 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
13063 (mips_hard_regno_mode_ok): Delete.
13064 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
13065 (mips_hard_regno_mode_ok_p): ...this and make static.
13066 (mips_hard_regno_mode_ok_p): Rename to...
13067 (mips_hard_regno_mode_ok_uncached): ...this.
13068 (mips_hard_regno_mode_ok): New function.
13069 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
13070 of HARD_REGNO_MODE_OK.
13071 (mips_option_override): Update after above name changes.
13072 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13073 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
13074 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
13075 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
13076 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
13077 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13078 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
13079 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
13080 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
13081 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13082 (msp430_hard_regno_mode_ok): Make static and return a bool.
13083 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
13084 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
13085 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
13087 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13088 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
13089 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
13090 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
13091 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
13092 (PA_HARD_REGNO_MODE_OK): ...this
13093 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
13094 (PA_HARD_REGNO_MODE_OK): ...this.
13095 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13096 (pa_hard_regno_mode_ok): New function.
13097 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
13098 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13099 (pdp11_hard_regno_mode_ok): New function.
13100 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
13101 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
13103 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
13105 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13106 (rs6000_hard_regno_mode_ok): Rename to...
13107 (rs6000_hard_regno_mode_ok_uncached): ...this.
13108 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13109 (rs6000_hard_regno_mode_ok): New function.
13110 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
13111 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
13112 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
13113 (riscv_hard_regno_mode_ok): ...this and make static.
13114 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13115 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
13116 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
13117 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13118 (rl78_hard_regno_mode_ok): Make static and return bool.
13119 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
13120 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
13122 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
13123 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13124 (rs6000_hard_regno_mode_ok): Rename to...
13125 (rs6000_hard_regno_mode_ok_uncached): ...this.
13126 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13127 (rs6000_hard_regno_mode_ok): New function.
13128 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
13129 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
13130 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13131 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
13132 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
13133 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
13134 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13135 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
13136 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
13137 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13138 (sh_hard_regno_mode_ok): Make static.
13139 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
13140 instead of HARD_REGNO_MODE_OK.
13141 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
13142 (sparc_mode_class): Delete.
13143 (HARD_REGNO_MODE_OK): Delete.
13144 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13145 (hard_regno_mode_classes): Make static.
13146 (sparc_mode_class): Likewise.
13147 (sparc_hard_regno_mode_ok): New function.
13148 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
13149 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
13150 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
13152 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13153 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
13154 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
13155 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
13156 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
13157 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13158 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
13159 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
13160 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13161 (visium_hard_regno_mode_ok): New function.
13162 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
13163 instead of HARD_REGNO_MODE_OK.
13164 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
13165 (HARD_REGNO_MODE_OK): Delete.
13166 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
13167 (xtensa_hard_regno_mode_ok_p): ...this and make static.
13168 (xtensa_option_override): Update accordingly.
13169 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13170 (xtensa_hard_regno_mode_ok): New function.
13171 * system.h (HARD_REGNO_MODE_OK): Poison.
13173 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13174 Alan Hayward <alan.hayward@arm.com>
13175 David Sherwood <david.sherwood@arm.com>
13177 * target.def (hard_regno_call_part_clobbered): New hook.
13178 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
13179 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
13180 * doc/tm.texi: Regenerate.
13181 * hooks.h (hook_bool_uint_mode_false): Declare.
13182 * hooks.c (hook_bool_uint_mode_false): New function.
13183 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13184 * cselib.c (cselib_process_insn): Use
13185 targetm.hard_regno_call_part_clobbered instead of
13186 HARD_REGNO_CALL_PART_CLOBBERED.
13187 * ira-conflicts.c (ira_build_conflicts): Likewise.
13188 * ira-costs.c (ira_tune_allocno_costs): Likewise.
13189 * lra-constraints.c (need_for_call_save_p): Likewise.
13190 * lra-lives.c: Include target.h.
13191 (check_pseudos_live_through_calls): Use
13192 targetm.hard_regno_call_part_clobbered instead of
13193 HARD_REGNO_CALL_PART_CLOBBERED.
13194 * regcprop.c: Include target.h.
13195 (copyprop_hardreg_forward_1): Use
13196 targetm.hard_regno_call_part_clobbered instead of
13197 HARD_REGNO_CALL_PART_CLOBBERED.
13198 * reginfo.c (choose_hard_reg_mode): Likewise.
13199 * regrename.c (check_new_reg_p): Likewise.
13200 * reload.c (find_equiv_reg): Likewise.
13201 * reload1.c (emit_reload_insns): Likewise.
13202 * sched-deps.c (deps_analyze_insn): Likewise.
13203 * sel-sched.c (init_regs_for_mode): Likewise.
13204 (mark_unavailable_hard_regs): Likewise.
13205 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
13206 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13207 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
13209 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13210 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13211 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
13213 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
13215 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13216 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13217 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
13219 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13220 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13221 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
13223 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13224 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
13226 * config/powerpcspe/powerpcspe.c
13227 (rs6000_hard_regno_call_part_clobbered): New function.
13228 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13229 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13230 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
13232 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13233 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13234 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
13236 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13237 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13238 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
13240 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13241 Alan Hayward <alan.hayward@arm.com>
13242 David Sherwood <david.sherwood@arm.com>
13244 * rtl.h (subreg_memory_offset): Declare.
13245 * emit-rtl.c (subreg_memory_offset): New function.
13246 * expmed.c (store_bit_field_1): Use it.
13247 * expr.c (undefined_operand_subword_p): Likewise.
13248 * simplify-rtx.c (simplify_subreg): Likewise.
13250 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
13252 PR rtl-optimization/57448
13255 * optabs.c (expand_atomic_load): Place compiler memory barriers if
13256 using atomic_load pattern.
13257 (expand_atomic_store): Likewise.
13259 2017-09-04 Jakub Jelinek <jakub@redhat.com>
13262 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
13263 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
13264 handling. Use replace_call_with_value with NULL instead of
13265 gsi_replace, unlink_stmt_vdef and release_defs.
13267 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
13270 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
13272 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13275 * rtl.h (emit_library_call_value_1): Declare.
13276 (emit_library_call): Replace declaration with a series of overloads.
13277 Remove the parameter count argument.
13278 (emit_library_call_value): Likewise.
13279 * calls.c (emit_library_call_value_1): Make global. Replace varargs
13280 with an "rtx_mode_t *".
13281 (emit_library_call_value): Delete.
13282 (emit_library_call): Likewise.
13283 * asan.c (asan_emit_stack_protection): Update calls accordingly.
13284 (asan_emit_allocas_unpoison): Likewise.
13285 * builtins.c (expand_builtin_powi): Likewise.
13286 (expand_asan_emit_allocas_unpoison): Likewise.
13287 * cfgexpand.c (expand_main_function): Likewise.
13288 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
13289 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
13290 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
13291 * config/arm/arm.c (arm_trampoline_init): Likewise.
13292 (arm_call_tls_get_addr): Likewise.
13293 (arm_expand_divmod_libfunc): Likewise.
13294 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
13295 (smulsi3_highpart): Likewise.
13296 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
13297 (c6x_expand_compare): Likewise.
13298 (c6x_expand_movmem): Likewise.
13299 * config/frv/frv.c (frv_trampoline_init): Likewise.
13300 * config/i386/i386.c (ix86_trampoline_init): Likewise.
13301 (ix86_expand_divmod_libfunc): Likewise.
13302 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
13303 (ia64_expand_compare): Likewise.
13304 (ia64_profile_hook): Likewise.
13305 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
13306 (nonlocal_goto): Likewise.
13307 (restore_stack_nonlocal): Likewise.
13308 * config/m32r/m32r.c (block_move_call): Likewise.
13309 (m32r_trampoline_init): Likewise.
13310 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
13311 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
13312 (m68k_call_m68k_read_tp): Likewise.
13313 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
13314 (microblaze_expand_divide): Likewise.
13315 * config/mips/mips.h (mips_args): Likewise.
13316 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
13317 (MIPS_ICACHE_SYNC): Likewise.
13318 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
13319 (nios2_trampoline_init): Likewise.
13320 * config/pa/pa.c (hppa_tls_call): Likewise.
13321 (pa_trampoline_init): Likewise.
13322 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
13323 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
13324 (expand_strn_compare): Likewise.
13325 (rs6000_generate_compare): Likewise.
13326 (rs6000_expand_float128_convert): Likewise.
13327 (output_profile_hook): Likewise.
13328 (rs6000_trampoline_init): Likewise.
13329 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
13330 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
13331 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
13332 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
13333 (rs6000_generate_compare): Likewise.
13334 (rs6000_expand_float128_convert): Likewise.
13335 (output_profile_hook): Likewise.
13336 (rs6000_trampoline_init): Likewise.
13337 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
13338 * config/sh/sh.c (sh_trampoline_init): Likewise.
13339 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
13340 (sparc_emit_float_lib_cmp): Likewise.
13341 (sparc32_initialize_trampoline): Likewise.
13342 (sparc64_initialize_trampoline): Likewise.
13343 (sparc_profile_hook): Likewise.
13344 * config/spu/spu.c (ea_load_store): Likewise.
13345 * config/spu/spu.md (floatunssidf2): Likewise.
13346 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
13347 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
13348 * config/visium/visium.c (expand_block_move_4): Likewise.
13349 (expand_block_move_2): Likewise.
13350 (expand_block_move_1): Likewise.
13351 (expand_block_set_4): Likewise.
13352 (expand_block_set_2): Likewise.
13353 (expand_block_set_1): Likewise.
13354 (visium_trampoline_init): Likewise.
13355 (visium_profile_hook): Likewise.
13356 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
13357 (xtensa_setup_frame_addresses): Likewise.
13358 (xtensa_trampoline_init): Likewise.
13359 * except.c (sjlj_emit_function_enter): Likewise.
13360 (sjlj_emit_function_exit): Likewise.
13361 * explow.c (allocate_dynamic_stack_space): Likewise.
13362 (probe_stack_range): Likewise.
13363 * expr.c (convert_mode_scalar): Likewise.
13364 * optabs.c (expand_binop): Likewise.
13365 (expand_twoval_binop_libfunc): Likewise.
13366 (expand_unop): Likewise.
13367 (prepare_cmp_insn): Likewise.
13368 (prepare_float_lib_cmp): Likewise.
13369 (expand_float): Likewise.
13370 (expand_fix): Likewise.
13371 (expand_fixed_convert): Likewise.
13372 (maybe_emit_sync_lock_test_and_set): Likewise.
13373 (expand_atomic_compare_and_swap): Likewise.
13374 (expand_mem_thread_fence): Likewise.
13375 (expand_atomic_fetch_op): Likewise.
13377 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
13379 * doc/generic.texi (OpenACC): Adjust URL.
13380 * doc/invoke.texi (C Dialect Options): Ditto.
13382 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
13384 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
13385 predicate for operand 1. Add (m,<S>) constraint.
13386 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
13387 Prevent memory operand 1 with register operand 2.
13389 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
13391 PR rtl-optimization/82024
13392 * combine.c (try_combine): If the combination result is a PARALLEL,
13393 and we only need to retain the SET in there that would be placed
13394 at I2, check that we can place that at I3 instead, before doing so.
13396 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13399 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
13401 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
13402 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
13405 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
13406 Jeff Law <law@redhat.com>
13408 * varasm.c (bss_initializer_p): Do not put constants into .bss
13409 (categorize_decl_for_section): Handle bss_initializer_p returning
13410 false when DECL_INITIAL is NULL.
13412 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13415 * config/s390/s390.c (s390_can_inline_p): New function.
13417 2017-09-01 Jeff Law <law@redhat.com>
13419 PR tree-optimization/82052
13420 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
13421 Always initialize the returned slot after a hash table miss
13422 when INSERT is true.
13424 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
13426 * config/s390/s390.md (mem_signal_fence): Remove.
13427 * doc/md.texi (mem_signal_fence): Remove.
13428 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
13430 * target-insns.def (mem_signal_fence): Remove.
13432 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13435 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
13438 * asan.c (create_odr_indicator): Strip name encoding from assembler
13439 name before appending it after __odr_asan_.
13441 2017-09-01 Martin Liska <mliska@suse.cz>
13443 PR tree-optimization/82059
13444 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
13445 frequency only when an edge is redirected.
13447 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13449 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
13450 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
13451 (arc_conditional_register_usage): Remove ARC600 lp_count
13453 (arc_file_start): Emit Tag_ARC_CPU_variation.
13454 (arc_can_use_doloop_p): New conditions to use ZOLs.
13455 (hwloop_fail): New function.
13456 (hwloop_optimize): Likewise.
13457 (hwloop_pattern_reg): Likewise.
13458 (arc_doloop_hooks): New struct, to be used with reorg_loops.
13459 (arc_reorg_loops): New function, calls reorg_loops.
13460 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
13461 (arc600_corereg_hazard): Remove ZOL checking, case handled by
13463 (arc_loop_hazard): Remove function, functionality moved into
13465 (arc_hazard): Remove arc_loop_hazard call.
13466 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
13467 into hwloop_optimize.
13468 (arc_label_align): Remove ZOL handling.
13469 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
13470 * config/arc/arc.md (doloop_begin): Remove pattern.
13471 (doloop_begin_i): Likewise.
13472 (doloop_end_i): Likewise.
13473 (doloop_fallback): Likewise.
13474 (doloop_fallback_m): Likewise.
13475 (doloop_end): Reimplement expand.
13476 (arc_lp): New pattern for LP instruction.
13477 (loop_end): New pattern.
13478 (loop_fail): Likewise.
13479 (decrement_and_branch_until_zero): Likewise.
13480 * config/arc/arc.opt (mlpc-width): New option.
13481 * doc/invoke.texi (mlpc-width): Document option.
13483 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13485 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
13486 (arc_ccfsm_advance): Fix checking for delay slots.
13487 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
13489 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13491 * config/arc/arc.md (movqi_insn): Add stores to save constant long
13493 (movhi_insn): Update store instruction constraint which are saving
13494 6-bit short immediates.
13495 (movsi_insn): Consider also short scaled load operations.
13496 (zero_extendhisi2_i): Use Usd constraint instead of T.
13497 (extendhisi2_i): Add q constraint.
13498 (arc_clzsi2): Add type and length attributes.
13499 (arc_ctzsi2): Likewise.
13500 * config/arc/constraints.md (Usc): Update constraint, the
13501 assembler can parse two relocations for a single instruction.
13503 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13505 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
13506 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
13508 2017-08-31 Olivier Hainque <hainque@adacore.com>
13510 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
13511 match as powerpc-wrs-vxworks*.
13513 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
13515 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
13516 register constraint for by-element operand.
13517 (aarch64_mls_elt_merge<mode>): Likewise.
13519 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13521 * config/arc/arc.c (arc_can_follow_jump): Check for short
13524 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13526 * config.gcc: Use g.opt for arc.
13527 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
13528 functionality moved to ...
13529 (legitimate_scaled_address_p): New function, ...here.
13530 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
13531 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
13532 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
13534 (arc_override_options): Handle G option.
13535 (arc_output_pic_addr_const): Correct function definition.
13536 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
13537 (arc_decl_anon_ns_mem_p): Delete.
13538 (arc_in_small_data_p): Overhaul this function to take into
13539 consideration the value given via G option.
13540 (arc_rewrite_small_data_1): Renamed and corrected old
13541 arc_rewrite_small_data function.
13542 (arc_rewrite_small_data): New function.
13543 (small_data_pattern): Don't use pic_offset_table_rtx.
13544 * config/arc/arc.h (CC1_SPEC): Recognize G option.
13545 * config/arc/simdext.md (movmisalignv2hi): Use
13546 prepare_move_operands function.
13548 (movmisalign*): Likewise.
13549 * doc/invoke.texi (ARC options): Document -G option.
13551 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13553 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
13555 * config/arc/arc.c (arc_print_operand): Output scalled address for
13556 sdata whenever is possible.
13557 (arc_in_small_data_p): Allow sdata for 64bit datum when double
13558 load/stores are available.
13559 (compact_sda_memory_operand): Check for the alignment required by
13560 code density instructions.
13561 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
13563 * config/arc/constraints.md (Usd): Update constraint.
13564 (Us0): New constraint.
13565 (Usc): Update constraint.
13567 2017-08-31 Richard Biener <rguenther@suse.de>
13569 PR middle-end/82054
13570 * dwarf2out.c (dwarf2out_early_global_decl): Process each
13571 function only once.
13573 2017-08-31 Tamar Christina <tamar.christina@arm.com>
13575 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
13576 Resize type_signature.
13578 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13579 Alan Hayward <alan.hayward@arm.com>
13580 David Sherwood <david.sherwood@arm.com>
13582 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
13583 subregs whose inner modes can be stored in GPRs.
13584 (aarch64_classify_index): Likewise.
13586 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13587 Alan Hayward <alan.hayward@arm.com>
13588 David Sherwood <david.sherwood@arm.com>
13590 * config/aarch64/iterators.md (V_cmp_result): Rename to...
13591 (V_INT_EQUIV): ...this.
13592 (v_cmp_result): Rename to...
13593 (v_int_equiv): ...this.
13594 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
13595 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
13596 (copysign<mode>3): Likewise.
13597 (aarch64_simd_bsl<mode>_internal): Likewise.
13598 (aarch64_simd_bsl<mode>): Likewise.
13599 (vec_cmp<mode><mode>): Likewise.
13600 (vcond<mode><mode>): Likewise.
13601 (vcond<v_cmp_mixed><mode>): Likewise.
13602 (vcondu<mode><v_cmp_mixed>): Likewise.
13603 (aarch64_cm<optab><mode>): Likewise.
13604 (aarch64_cmtst<mode>): Likewise.
13605 (aarch64_fac<optab><mode>): Likewise.
13606 (vec_perm_const<mode>): Likewise.
13607 (vcond_mask_<mode><v_cmp_result>): Rename to...
13608 (vcond_mask_<mode><v_int_equiv>): ...this.
13609 (vec_cmp<mode><v_cmp_result>): Rename to...
13610 (vec_cmp<mode><v_int_equiv>): ...this.
13612 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13613 Alan Hayward <alan.hayward@arm.com>
13614 David Sherwood <david.sherwood@arm.com>
13616 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
13618 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
13619 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
13620 (UNSPEC_LD4_DREG): New unspecs.
13621 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
13622 (aarch64_ld2<mode>_dreg_be): Replace with...
13623 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
13625 (aarch64_ld3<mode>_dreg_le)
13626 (aarch64_ld3<mode>_dreg_be): Replace with...
13627 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
13629 (aarch64_ld4<mode>_dreg_le)
13630 (aarch64_ld4<mode>_dreg_be): Replace with...
13631 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
13634 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13636 PR tree-optimization/81987
13637 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
13638 insert an initializer in a location not dominated by the stride
13641 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13643 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
13644 on the entire header of the finally block in the fallthru case.
13646 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13648 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
13650 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
13652 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
13653 rs6000_emit_move_from_cr and call renamed function.
13654 (rs6000_emit_prologue): Call renamed functions.
13655 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
13656 movesi_from_cr, remove volatile CRs.
13658 2017-08-30 Jon Beniston <jon@beniston.com>
13659 Richard Biener <rguenther@suse.de>
13661 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
13662 of VECTOR_MODE_P check.
13663 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
13664 element vector types.
13666 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13668 * df.h (df_read_modify_subreg_p): Remove in favor of...
13669 * rtl.h (read_modify_subreg_p): ...this new function. Take a
13670 const_rtx instead of an rtx.
13671 * cprop.c (local_cprop_find_used_regs): Update accordingly.
13672 * df-problems.c (df_word_lr_mark_ref): Likewise.
13673 * ira-lives.c (mark_pseudo_reg_live): Likewise.
13674 (mark_pseudo_reg_dead): Likewise.
13675 (mark_ref_dead): Likewise.
13676 * reginfo.c (init_subregs_of_mode): Likewise.
13677 * sched-deps.c (sched_analyze_1): Likewise.
13678 * df-scan.c (df_def_record_1): Likewise.
13679 (df_uses_record): Likewise.
13680 (df_read_modify_subreg_p): Remove in favor of...
13681 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
13682 const_rtx instead of an rtx.
13684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13685 Alan Hayward <alan.hayward@arm.com>
13686 David Sherwood <david.sherwood@arm.com>
13688 * rtl.h (partial_subreg_p): New function.
13689 * caller-save.c (save_call_clobbered_regs): Use it.
13690 * calls.c (expand_call): Likewise.
13691 * combine.c (combinable_i3pat): Likewise.
13692 (simplify_set): Likewise.
13693 (make_extraction): Likewise.
13694 (make_compound_operation_int): Likewise.
13695 (gen_lowpart_or_truncate): Likewise.
13696 (force_to_mode): Likewise.
13697 (make_field_assignment): Likewise.
13698 (reg_truncated_to_mode): Likewise.
13699 (record_truncated_value): Likewise.
13700 (move_deaths): Likewise.
13701 * cse.c (record_jump_cond): Likewise.
13702 (cse_insn): Likewise.
13703 * cselib.c (cselib_lookup_1): Likewise.
13704 * expmed.c (extract_bit_field_using_extv): Likewise.
13705 * function.c (assign_parm_setup_reg): Likewise.
13706 * ifcvt.c (noce_convert_multiple_sets): Likewise.
13707 * ira-build.c (create_insn_allocnos): Likewise.
13708 * lra-coalesce.c (merge_pseudos): Likewise.
13709 * lra-constraints.c (match_reload): Likewise.
13710 (simplify_operand_subreg): Likewise.
13711 (curr_insn_transform): Likewise.
13712 * lra-lives.c (process_bb_lives): Likewise.
13713 * lra.c (new_insn_reg): Likewise.
13714 (lra_substitute_pseudo): Likewise.
13715 * regcprop.c (mode_change_ok): Likewise.
13716 (maybe_mode_change): Likewise.
13717 (copyprop_hardreg_forward_1): Likewise.
13718 * reload.c (push_reload): Likewise.
13719 (find_reloads): Likewise.
13720 (find_reloads_subreg_address): Likewise.
13721 * reload1.c (alter_reg): Likewise.
13722 (eliminate_regs_1): Likewise.
13723 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13725 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
13727 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
13728 back to if statements, including unpack.
13730 2017-08-30 Martin Liska <mliska@suse.cz>
13732 PR inline-asm/82001
13733 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
13735 (func_checker::compare_asm_inputs_outputs): ... this function.
13736 (func_checker::compare_gimple_asm): Use the function to compare
13737 also ASM constrains.
13738 * ipa-icf-gimple.h: Rename the function.
13740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13741 Alan Hayward <alan.hayward@arm.com>
13742 David Sherwood <david.sherwood@arm.com>
13744 * coretypes.h (complex_mode): New type.
13745 * gdbhooks.py (build_pretty_printer): Handle it.
13746 * machmode.h (complex_mode): New class.
13747 (complex_mode::includes_p): New function.
13748 (is_complex_int_mode): Likewise.
13749 (is_complex_float_mode): Likewise.
13750 * genmodes.c (get_mode_class): Handle complex mode classes.
13751 * function.c (expand_function_end): Use is_complex_int_mode.
13753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13754 Alan Hayward <alan.hayward@arm.com>
13755 David Sherwood <david.sherwood@arm.com>
13757 * coretypes.h (scalar_mode_pod): New typedef.
13758 * gdbhooks.py (build_pretty_printer): Handle it.
13759 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
13760 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
13761 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
13762 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
13763 as_a <scalar_mode>.
13765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13766 Alan Hayward <alan.hayward@arm.com>
13767 David Sherwood <david.sherwood@arm.com>
13769 * machmode.h (mode_for_vector): Take a scalar_mode instead
13771 * stor-layout.c (mode_for_vector): Likewise.
13772 * explow.c (promote_mode): Use as_a <scalar_mode>.
13773 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
13775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13776 Alan Hayward <alan.hayward@arm.com>
13777 David Sherwood <david.sherwood@arm.com>
13779 * target.def (preferred_simd_mode): Take a scalar_mode
13780 instead of a machine_mode.
13781 * targhooks.h (default_preferred_simd_mode): Likewise.
13782 * targhooks.c (default_preferred_simd_mode): Likewise.
13783 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
13784 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
13785 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
13786 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
13787 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
13788 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
13789 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
13790 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
13792 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
13793 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
13794 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
13795 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
13796 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
13797 * doc/tm.texi: Regenerate.
13798 * optabs-query.c (can_vec_mask_load_store_p): Return false for
13801 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13802 Alan Hayward <alan.hayward@arm.com>
13803 David Sherwood <david.sherwood@arm.com>
13805 * target.def (scalar_mode_supported_p): Take a scalar_mode
13806 instead of a machine_mode.
13807 * targhooks.h (default_scalar_mode_supported_p): Likewise.
13808 * targhooks.c (default_scalar_mode_supported_p): Likewise.
13809 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
13810 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13811 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
13812 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
13813 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
13814 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
13815 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
13816 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
13817 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
13818 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
13819 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
13820 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
13822 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
13823 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
13824 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13825 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13826 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
13828 * doc/tm.texi: Regenerate.
13830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13831 Alan Hayward <alan.hayward@arm.com>
13832 David Sherwood <david.sherwood@arm.com>
13834 * coretypes.h (opt_scalar_mode): New typedef.
13835 * gdbhooks.py (build_pretty_printers): Handle it.
13836 * machmode.h (mode_iterator::get_2xwider): Add overload for
13838 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
13840 * expr.c (convert_mode_scalar): Likewise.
13841 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13842 * optabs.c (expand_float): Likewise.
13843 (expand_fix): Likewise.
13844 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13847 Alan Hayward <alan.hayward@arm.com>
13848 David Sherwood <david.sherwood@arm.com>
13850 * optabs.c (expand_float): Explicitly check for scalars before
13851 using a branching expansion.
13852 (expand_fix): Likewise.
13854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13855 Alan Hayward <alan.hayward@arm.com>
13856 David Sherwood <david.sherwood@arm.com>
13858 * expr.c (convert_mode): Split scalar handling out into...
13859 (convert_mode_scalar): ...this new function. Treat the modes
13862 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13863 Alan Hayward <alan.hayward@arm.com>
13864 David Sherwood <david.sherwood@arm.com>
13866 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
13868 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
13870 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13871 Alan Hayward <alan.hayward@arm.com>
13872 David Sherwood <david.sherwood@arm.com>
13874 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
13875 rather than a machine_mode.
13876 (fixed_from_string): Likewise.
13877 (fixed_convert): Likewise.
13878 (fixed_convert_from_int): Likewise.
13879 (fixed_convert_from_real): Likewise.
13880 (real_convert_from_fixed): Likewise.
13881 * fixed-value.c (fixed_from_double_int): Likewise.
13882 (fixed_from_string): Likewise.
13883 (fixed_convert): Likewise.
13884 (fixed_convert_from_int): Likewise.
13885 (fixed_convert_from_real): Likewise.
13886 (real_convert_from_fixed): Likewise.
13887 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
13889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13890 Alan Hayward <alan.hayward@arm.com>
13891 David Sherwood <david.sherwood@arm.com>
13893 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
13894 of separate mode class checks. Do not allow vector modes here.
13895 (immed_wide_int_const): Use as_a <scalar_mode>.
13896 * explow.c (trunc_int_for_mode): Likewise.
13897 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
13898 (wi::shwi): Likewise.
13899 (wi::min_value): Likewise.
13900 (wi::max_value): Likewise.
13901 * dwarf2out.c (loc_descriptor): Likewise.
13902 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
13903 for CONST_WIDE_INT.
13905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13906 Alan Hayward <alan.hayward@arm.com>
13907 David Sherwood <david.sherwood@arm.com>
13909 * tree.h (SCALAR_TYPE_MODE): New macro.
13910 * expr.c (expand_expr_addr_expr_1): Use it.
13911 (expand_expr_real_2): Likewise.
13912 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
13913 (fold_convert_const_fixed_from_int): Likewise.
13914 (fold_convert_const_fixed_from_real): Likewise.
13915 (native_encode_fixed): Likewise
13916 (native_encode_complex): Likewise
13917 (native_encode_vector): Likewise.
13918 (native_interpret_fixed): Likewise.
13919 (native_interpret_real): Likewise.
13920 (native_interpret_complex): Likewise.
13921 (native_interpret_vector): Likewise.
13922 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
13923 (simd_clone_adjust_argument_types): Likewise.
13924 (simd_clone_init_simd_arrays): Likewise.
13925 (simd_clone_adjust): Likewise.
13926 * stor-layout.c (layout_type): Likewise.
13927 * tree.c (build_minus_one_cst): Likewise.
13928 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13929 * tree-inline.c (estimate_move_cost): Likewise.
13930 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
13931 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
13932 (vectorizable_reduction): Likewise.
13933 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
13934 (vect_recog_mixed_size_cond_pattern): Likewise.
13935 (check_bool_pattern): Likewise.
13936 (adjust_bool_pattern): Likewise.
13937 (search_type_for_mask_1): Likewise.
13938 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
13939 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13940 (vectorizable_load): Likewise.
13941 (vectorizable_store): Likewise.
13942 * ubsan.c (ubsan_encode_value): Likewise.
13943 * varasm.c (output_constant): Likewise.
13945 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13946 Alan Hayward <alan.hayward@arm.com>
13947 David Sherwood <david.sherwood@arm.com>
13949 * coretypes.h (scalar_mode): New class.
13950 * machmode.h (scalar_mode): Likewise.
13951 (scalar_mode::includes_p): New function.
13952 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
13953 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
13954 * genmodes.c (get_mode_class): Handle remaining scalar modes.
13955 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
13956 * expmed.c (store_bit_field_1): Likewise.
13957 (extract_bit_field_1): Likewise.
13958 * expr.c (write_complex_part): Likewise.
13959 (read_complex_part): Likewise.
13960 (emit_move_complex_push): Likewise.
13961 (expand_expr_real_2): Likewise.
13962 * function.c (assign_parm_setup_reg): Likewise.
13963 (assign_parms_unsplit_complex): Likewise.
13964 * optabs.c (expand_binop): Likewise.
13965 * rtlanal.c (subreg_get_info): Likewise.
13966 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13967 * varasm.c (output_constant_pool_2): Likewise.
13969 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13970 Alan Hayward <alan.hayward@arm.com>
13971 David Sherwood <david.sherwood@arm.com>
13973 * expmed.c (extract_high_half): Use scalar_int_mode and remove
13975 (expmed_mult_highpart_optab): Likewise.
13976 (expmed_mult_highpart): Likewise.
13978 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13979 Alan Hayward <alan.hayward@arm.com>
13980 David Sherwood <david.sherwood@arm.com>
13982 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
13983 instead of a machine_mode.
13984 (builtin_memset_read_str): Likewise.
13985 * builtins.c (c_readstr): Likewise.
13986 (builtin_memcpy_read_str): Likewise.
13987 (builtin_strncpy_read_str): Likewise.
13988 (builtin_memset_read_str): Likewise.
13989 (builtin_memset_gen_str): Likewise.
13990 (expand_builtin_signbit): Use scalar_int_mode for local variables.
13991 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
13992 instead of a machine_mode.
13993 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
13995 (make_extraction): Likewise.
13996 (try_widen_shift_mode): Take and return scalar_int_modes instead
13998 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
13999 a scalar_int_mode instead of a machine_mode.
14000 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
14001 (avr_addr_space_pointer_mode): Likewise.
14002 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
14003 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
14004 (msp430_unwind_word_mode): Likewise.
14005 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
14006 (spu_addr_space_pointer_mode): Likewise.
14007 (spu_addr_space_address_mode): Likewise.
14008 (spu_libgcc_cmp_return_mode): Likewise.
14009 (spu_libgcc_shift_count_mode): Likewise.
14010 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
14011 (rl78_addr_space_pointer_mode): Likewise.
14012 (fl78_unwind_word_mode): Likewise.
14013 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
14015 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
14016 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
14017 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
14018 (mips_valid_pointer_mode): Likewise.
14019 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
14020 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
14021 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
14023 (ft32_addr_space_address_mode): Likewise.
14024 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
14025 scalar_int_mode instead of a machine_mode.
14026 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
14028 (m32c_addr_space_address_mode): Likewise.
14029 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
14030 (rs6000_eh_return_filter_mode): Likewise.
14031 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
14032 (rs6000_eh_return_filter_mode): Likewise.
14033 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
14034 (s390_libgcc_shift_count_mode): Likewise.
14035 (s390_unwind_word_mode): Likewise.
14036 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
14038 * target.def (mode_rep_extended): Likewise.
14039 (valid_pointer_mode): Likewise.
14040 (addr_space.valid_pointer_mode): Likewise.
14041 (eh_return_filter_mode): Return a scalar_int_mode rather than
14043 (libgcc_cmp_return_mode): Likewise.
14044 (libgcc_shift_count_mode): Likewise.
14045 (unwind_word_mode): Likewise.
14046 (addr_space.pointer_mode): Likewise.
14047 (addr_space.address_mode): Likewise.
14048 * doc/tm.texi: Regenerate.
14049 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
14051 (do_jump): Use scalar_int_mode for local variables.
14052 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
14053 rather than a machine_mode.
14054 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
14055 (scompare_loc_descriptor_wide): Likewise.
14056 (scompare_loc_descriptor_narrow): Likewise.
14057 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
14059 * except.c (sjlj_emit_dispatch_table): Likewise.
14060 (expand_builtin_eh_copy_values): Likewise.
14061 * explow.c (convert_memory_address_addr_space_1): Likewise.
14062 Take a scalar_int_mode rather than a machine_mode.
14063 (convert_memory_address_addr_space): Take a scalar_int_mode rather
14064 than a machine_mode.
14065 (memory_address_addr_space): Use scalar_int_mode for local variables.
14066 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
14067 rather than a machine_mode.
14068 * expmed.c (mask_rtx): Likewise.
14069 (init_expmed_one_conv): Likewise.
14070 (expand_mult_highpart_adjust): Likewise.
14071 (extract_high_half): Likewise.
14072 (expmed_mult_highpart_optab): Likewise.
14073 (expmed_mult_highpart): Likewise.
14074 (expand_smod_pow2): Likewise.
14075 (expand_sdiv_pow2): Likewise.
14076 (emit_store_flag_int): Likewise.
14077 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
14079 (extract_low_bits): Likewise.
14080 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
14082 * expr.c (pieces_addr::adjust): Likewise.
14083 (can_store_by_pieces): Likewise.
14084 (store_by_pieces): Likewise.
14085 (clear_by_pieces_1): Likewise.
14086 (expand_expr_addr_expr_1): Likewise.
14087 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
14088 (expand_expr_real_1): Likewise.
14089 (try_casesi): Likewise.
14090 * final.c (shorten_branches): Likewise.
14091 * fold-const.c (fold_convert_const_int_from_fixed): Change the
14092 type of "mode" to machine_mode.
14093 * internal-fn.c (expand_arith_overflow_result_store): Take a
14094 scalar_int_mode rather than a machine_mode.
14095 (expand_mul_overflow): Use scalar_int_mode for local variables.
14096 * loop-doloop.c (doloop_modify): Likewise.
14097 (doloop_optimize): Likewise.
14098 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
14099 than a machine_mode.
14100 (expand_doubleword_shift_condmove): Likewise.
14101 (expand_doubleword_shift): Likewise.
14102 (expand_doubleword_clz): Likewise.
14103 (expand_doubleword_popcount): Likewise.
14104 (expand_doubleword_parity): Likewise.
14105 (expand_absneg_bit): Use scalar_int_mode for local variables.
14106 (prepare_float_lib_cmp): Likewise.
14107 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
14108 rather than a machine_mode.
14109 (convert_memory_address_addr_space): Likewise.
14110 (get_mode_bounds): Likewise.
14111 (get_address_mode): Return a scalar_int_mode rather than a
14113 * rtlanal.c (get_address_mode): Likewise.
14114 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
14115 than a machine_mode.
14116 * targhooks.c (default_mode_rep_extended): Likewise.
14117 (default_valid_pointer_mode): Likewise.
14118 (default_addr_space_valid_pointer_mode): Likewise.
14119 (default_eh_return_filter_mode): Return a scalar_int_mode rather
14120 than a machine_mode.
14121 (default_libgcc_cmp_return_mode): Likewise.
14122 (default_libgcc_shift_count_mode): Likewise.
14123 (default_unwind_word_mode): Likewise.
14124 (default_addr_space_pointer_mode): Likewise.
14125 (default_addr_space_address_mode): Likewise.
14126 * targhooks.h (default_eh_return_filter_mode): Likewise.
14127 (default_libgcc_cmp_return_mode): Likewise.
14128 (default_libgcc_shift_count_mode): Likewise.
14129 (default_unwind_word_mode): Likewise.
14130 (default_addr_space_pointer_mode): Likewise.
14131 (default_addr_space_address_mode): Likewise.
14132 (default_mode_rep_extended): Take a scalar_int_mode rather than
14134 (default_valid_pointer_mode): Likewise.
14135 (default_addr_space_valid_pointer_mode): Likewise.
14136 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
14138 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
14139 rather than a machine_mode.
14140 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
14141 for local variables.
14142 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14143 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
14144 than a machine_mode.
14146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14147 Alan Hayward <alan.hayward@arm.com>
14148 David Sherwood <david.sherwood@arm.com>
14150 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
14151 the mode argument to scalar_int_mode.
14152 (do_jump_by_parts_zero_rtx): Likewise.
14153 (do_jump_by_parts_equality_rtx): Likewise.
14154 (do_jump_by_parts_greater): Take a mode argument.
14155 (do_jump_by_parts_equality): Likewise.
14156 (do_jump_1): Update calls accordingly.
14158 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14159 Alan Hayward <alan.hayward@arm.com>
14160 David Sherwood <david.sherwood@arm.com>
14162 * is-a.h (safe_dyn_cast): New function.
14163 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
14164 (jump_table_for_label): Likewise.
14165 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
14166 instead of an rtx_insn *.
14167 (shorten_branches): Use dyn_cast instead of LABEL_P and
14168 JUMP_TABLE_DATA_P. Use jump_table_for_label and
14169 rtx_jump_table_data::get_data_mode.
14170 (final_scan_insn): Likewise.
14172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14173 Alan Hayward <alan.hayward@arm.com>
14174 David Sherwood <david.sherwood@arm.com>
14176 * combine.c (try_combine): Use is_a <scalar_int_mode> when
14177 trying to combine a full-register integer set with a subreg
14180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14181 Alan Hayward <alan.hayward@arm.com>
14182 David Sherwood <david.sherwood@arm.com>
14184 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
14185 that is always either address_mode or pointer_mode.
14187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14188 Alan Hayward <alan.hayward@arm.com>
14189 David Sherwood <david.sherwood@arm.com>
14191 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
14192 when the two are known to be equal.
14194 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14195 Alan Hayward <alan.hayward@arm.com>
14196 David Sherwood <david.sherwood@arm.com>
14198 * simplify-rtx.c (simplify_const_unary_operation): Use
14199 is_a <scalar_int_mode> instead of checking for a nonzero
14200 precision. Forcibly convert op_mode to a scalar_int_mode
14201 in that case. More clearly differentiate the operand and
14202 result modes and use the former when deciding what the value
14203 of a count-bits operation should be. Use is_int_mode instead
14204 of checking for a MODE_INT. Remove redundant check for whether
14205 this mode has a zero precision.
14207 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14208 Alan Hayward <alan.hayward@arm.com>
14209 David Sherwood <david.sherwood@arm.com>
14211 * optabs.c (widen_leading): Change the type of the mode argument
14212 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
14213 (widen_bswap): Likewise.
14214 (expand_parity): Likewise.
14215 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
14216 (expand_ffs): Likewise.
14217 (epand_unop): Check for scalar integer modes before calling the
14220 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14221 Alan Hayward <alan.hayward@arm.com>
14222 David Sherwood <david.sherwood@arm.com>
14224 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
14226 (expand_expr_real_1): Update call accordingly.
14228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14229 Alan Hayward <alan.hayward@arm.com>
14230 David Sherwood <david.sherwood@arm.com>
14232 * expmed.c (store_bit_field_using_insv): Add op0_mode and
14233 value_mode arguments. Use scalar_int_mode internally.
14234 (store_bit_field_1): Rename the new integer mode from imode
14235 to op0_mode and use it instead of GET_MODE (op0). Update calls
14236 to store_split_bit_field, store_bit_field_using_insv and
14237 store_fixed_bit_field.
14238 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
14239 Use scalar_int_mode internally. Use a bit count rather than a mode
14240 when calculating the largest bit size for get_best_mode.
14241 Update calls to store_split_bit_field and store_fixed_bit_field_1.
14242 (store_fixed_bit_field_1): Add mode and value_mode arguments.
14243 Remove assertion that OP0 has a scalar integer mode.
14244 (store_split_bit_field): Add op0_mode and value_mode arguments.
14245 Update calls to extract_fixed_bit_field.
14246 (extract_bit_field_using_extv): Add an op0_mode argument.
14247 Use scalar_int_mode internally.
14248 (extract_bit_field_1): Rename the new integer mode from imode to
14249 op0_mode and use it instead of GET_MODE (op0). Update calls to
14250 extract_split_bit_field, extract_bit_field_using_extv and
14251 extract_fixed_bit_field.
14252 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
14253 to extract_split_bit_field and extract_fixed_bit_field_1.
14254 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
14255 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
14256 on the target mode.
14257 (extract_split_bit_field): Add an op0_mode argument. Update call
14258 to extract_fixed_bit_field.
14260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14261 Alan Hayward <alan.hayward@arm.com>
14262 David Sherwood <david.sherwood@arm.com>
14264 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
14265 * explow.c (hard_function_value): Likewise.
14266 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
14267 convert_to_mode call outside the loop.
14268 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
14269 for the mode iterator. Require the mode specified by max_pieces
14271 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
14273 (copy_blkmode_to_reg): Likewise.
14274 (set_storage_via_setmem): Likewise.
14275 * optabs.c (prepare_cmp_insn): Likewise.
14276 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14277 * stor-layout.c (finish_bitfield_representative): Likewise.
14279 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14280 Alan Hayward <alan.hayward@arm.com>
14281 David Sherwood <david.sherwood@arm.com>
14283 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
14284 * expr.c (convert_move): Use them.
14285 (convert_modes): Likewise.
14286 (store_expr_with_bounds): Likewise.
14288 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14289 Alan Hayward <alan.hayward@arm.com>
14290 David Sherwood <david.sherwood@arm.com>
14292 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
14293 parameter for the mode of "x". Remove the "known_x", "known_mode"
14294 and "known_ret" arguments. Change the type of the mode argument
14295 to scalar_int_mode.
14296 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
14297 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
14298 (reg_num_sign_bit_copies_for_combine): Likewise.
14299 * rtlanal.c (nonzero_bits1): Likewise.
14300 (num_sign_bit_copies1): Likewise.
14301 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
14302 (reg_num_sign_bit_copies_general): Likewise.
14303 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
14304 (reg_nonzero_bits_general): Likewise.
14306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14307 Alan Hayward <alan.hayward@arm.com>
14308 David Sherwood <david.sherwood@arm.com>
14310 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
14311 than in subroutines. Return 1 for non-integer modes.
14312 (cached_num_sign_bit_copies): Change the type of the mode parameter
14313 to scalar_int_mode.
14314 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
14315 classes. Handle CONST_INT_P first and then check whether X also
14316 has a scalar integer mode. Check the same thing for inner registers
14317 of a SUBREG and for values that are being extended or truncated.
14319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14320 Alan Hayward <alan.hayward@arm.com>
14321 David Sherwood <david.sherwood@arm.com>
14323 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
14324 in subroutines. Return the mode mask for non-integer modes.
14325 (cached_nonzero_bits): Change the type of the mode parameter
14326 to scalar_int_mode.
14327 (nonzero_bits1): Likewise. Remove early exit for other mode
14328 classes. Handle CONST_INT_P first and then check whether X
14329 also has a scalar integer mode.
14331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14332 Alan Hayward <alan.hayward@arm.com>
14333 David Sherwood <david.sherwood@arm.com>
14335 * expr.c (widest_int_mode_for_size): Make the comment match the code.
14336 Return a scalar_int_mode and assert that the size is greater than
14338 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
14339 (op_by_pieces_d::op_by_pieces_d): Likewise.
14340 (op_by_pieces_d::run): Likewise.
14341 (can_store_by_pieces): Likewise.
14343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14344 Alan Hayward <alan.hayward@arm.com>
14345 David Sherwood <david.sherwood@arm.com>
14347 * combine.c (extract_left_shift): Add a mode argument and update
14349 (make_compound_operation_int): Change the type of the mode parameter
14350 to scalar_int_mode and update the call to extract_left_shift.
14352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14353 Alan Hayward <alan.hayward@arm.com>
14354 David Sherwood <david.sherwood@arm.com>
14356 * combine.c (simplify_and_const_int): Change the type of the mode
14357 parameter to scalar_int_mode.
14358 (simplify_and_const_int_1): Likewise. Update recursive call.
14360 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14361 Alan Hayward <alan.hayward@arm.com>
14362 David Sherwood <david.sherwood@arm.com>
14364 * combine.c (simplify_compare_const): Check that the mode is a
14365 scalar_int_mode (rather than VOIDmode) before testing its
14367 (simplify_comparison): Move COMPARISON_P handling out of the
14368 loop and restrict the latter part of the loop to scalar_int_modes.
14369 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
14370 and when considering SUBREG_REGs. Use is_int_mode instead of
14371 checking GET_MODE_CLASS against MODE_INT.
14373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14374 Alan Hayward <alan.hayward@arm.com>
14375 David Sherwood <david.sherwood@arm.com>
14377 * combine.c (try_widen_shift_mode): Move check for equal modes to...
14378 (simplify_shift_const_1): ...here. Use scalar_int_mode for
14379 shift_unit_mode and for modes involved in scalar shifts.
14381 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14382 Alan Hayward <alan.hayward@arm.com>
14383 David Sherwood <david.sherwood@arm.com>
14385 * combine.c (force_int_to_mode): New function, split out from...
14386 (force_to_mode): ...here. Keep xmode up-to-date and use it
14387 instead of GET_MODE (x).
14389 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14390 Alan Hayward <alan.hayward@arm.com>
14391 David Sherwood <david.sherwood@arm.com>
14393 * optabs-query.h (extraction_insn::struct_mode): Change type to
14394 opt_scalar_int_mode and update comment.
14395 (extraction_insn::field_mode): Change type to scalar_int_mode.
14396 (extraction_insn::pos_mode): Likewise.
14397 * combine.c (make_extraction): Update accordingly.
14398 * optabs-query.c (get_traditional_extraction_insn): Likewise.
14399 (get_optab_extraction_insn): Likewise.
14400 * recog.c (simplify_while_replacing): Likewise.
14401 * expmed.c (narrow_bit_field_mem): Change the type of the mode
14402 parameter to opt_scalar_int_mode.
14404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14405 Alan Hayward <alan.hayward@arm.com>
14406 David Sherwood <david.sherwood@arm.com>
14408 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
14409 to a scalar_int_mode instead of a machine_mode.
14410 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
14411 (get_best_mode): Return a boolean and use a pointer argument to store
14412 the selected mode. Replace the limit mode parameter with a bit limit.
14413 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
14414 for the values returned by bit_field_mode_iterator::next_mode.
14415 (store_bit_field): Update call to get_best_mode.
14416 (store_fixed_bit_field): Likewise.
14417 (extract_fixed_bit_field): Likewise.
14418 * expr.c (optimize_bitfield_assignment_op): Likewise.
14419 * fold-const.c (optimize_bit_field_compare): Likewise.
14420 (fold_truth_andor_1): Likewise.
14421 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
14422 Update for new type of m_mode.
14423 (get_best_mode): As above.
14425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14426 Alan Hayward <alan.hayward@arm.com>
14427 David Sherwood <david.sherwood@arm.com>
14429 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
14430 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
14431 (store_bit_field): Check is_a <scalar_int_mode> before calling
14432 strict_volatile_bitfield_p.
14433 (extract_bit_field): Likewise.
14435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14436 Alan Hayward <alan.hayward@arm.com>
14437 David Sherwood <david.sherwood@arm.com>
14439 * target.def (cstore_mode): Return a scalar_int_mode.
14440 * doc/tm.texi: Regenerate.
14441 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
14442 * targhooks.h (default_cstore_mode): Likewise.
14443 * targhooks.c (default_cstore_mode): Likewise, using a forced
14445 * expmed.c (emit_cstore): Expect the target of the cstore to be
14448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14449 Alan Hayward <alan.hayward@arm.com>
14450 David Sherwood <david.sherwood@arm.com>
14452 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
14454 (niter_desc): Likewise mode.
14455 (iv_analyze): Add a mode parameter.
14457 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
14458 and change its type to scalar_int_mode.
14459 * loop-iv.c: Update commentary at head of file.
14460 (iv_constant): Pass the mode paraeter before the rtx it describes
14461 and change its type to scalar_int_mode. Remove VOIDmode handling.
14462 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
14463 (iv_extend): Likewise.
14464 (shorten_into_mode): Likewise.
14465 (iv_add): Use scalar_int_mode.
14466 (iv_mult): Likewise.
14467 (iv_shift): Likewise.
14468 (canonicalize_iv_subregs): Likewise.
14469 (get_biv_step_1): Pass the outer_mode parameter before the rtx
14470 it describes and change its mode to scalar_int_mode. Also change
14471 the type of the returned inner_mode to scalar_int_mode.
14472 (get_biv_step): Likewise, turning outer_mode from a pointer
14473 into a direct parameter. Update call to get_biv_step_1.
14474 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
14475 iv_constant and get_biv_step.
14476 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
14477 and change its type to scalar_int_mode. Don't initialise iv->mode
14478 to VOIDmode and remove later checks for its still being VOIDmode.
14479 Update calls to iv_analyze_op and iv_analyze_expr. Check
14480 is_a <scalar_int_mode> when changing the mode under consideration.
14481 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
14482 Update call to iv_analyze_expr.
14483 (iv_analyze_op): Add a mode parameter. Reject subregs whose
14484 inner register is not also a scalar_int_mode. Update call to
14486 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
14487 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
14488 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
14489 separate mode class checks. Update calls to iv_analyze. Remove
14490 fix-up of VOIDmodes after iv_analyze_biv.
14491 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
14492 don't have a scalar_int_mode. Update call to biv_p.
14494 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14495 Alan Hayward <alan.hayward@arm.com>
14496 David Sherwood <david.sherwood@arm.com>
14498 * cfgexpand.c (convert_debug_memory_address): Use
14499 as_a <scalar_int_mode>.
14500 * combine.c (expand_compound_operation): Likewise.
14501 (make_extraction): Likewise.
14502 (change_zero_ext): Likewise.
14503 (simplify_comparison): Likewise.
14504 * cse.c (cse_insn): Likewise.
14505 * dwarf2out.c (minmax_loc_descriptor): Likewise.
14506 (mem_loc_descriptor): Likewise.
14507 (loc_descriptor): Likewise.
14508 * expmed.c (init_expmed_one_mode): Likewise.
14509 (synth_mult): Likewise.
14510 (emit_store_flag_1): Likewise.
14511 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
14512 of a comparison with size.
14513 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
14514 (reduce_to_bit_field_precision): Likewise.
14515 * function.c (expand_function_end): Likewise.
14516 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
14517 * loop-doloop.c (doloop_modify): Likewise.
14518 * optabs.c (expand_binop): Likewise.
14519 (expand_unop): Likewise.
14520 (expand_copysign_absneg): Likewise.
14521 (prepare_cmp_insn): Likewise.
14522 (maybe_legitimize_operand): Likewise.
14523 * recog.c (const_scalar_int_operand): Likewise.
14524 * rtlanal.c (get_address_mode): Likewise.
14525 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14526 (simplify_cond_clz_ctz): Likewise.
14527 * tree-nested.c (get_nl_goto_field): Likewise.
14528 * tree.c (build_vector_type_for_mode): Likewise.
14529 * var-tracking.c (use_narrower_mode): Likewise.
14531 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14532 Alan Hayward <alan.hayward@arm.com>
14533 David Sherwood <david.sherwood@arm.com>
14535 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
14536 * builtins.c (expand_builtin_signbit): Use it.
14537 * cfgexpand.c (expand_debug_expr): Likewise.
14538 * dojump.c (do_jump): Likewise.
14539 (do_compare_and_jump): Likewise.
14540 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
14541 * expmed.c (make_tree): Likewise.
14542 * expr.c (expand_expr_real_2): Likewise.
14543 (expand_expr_real_1): Likewise.
14544 (try_casesi): Likewise.
14545 * fold-const-call.c (fold_const_call_ss): Likewise.
14546 * fold-const.c (unextend): Likewise.
14547 (extract_muldiv_1): Likewise.
14548 (fold_single_bit_test): Likewise.
14549 (native_encode_int): Likewise.
14550 (native_encode_string): Likewise.
14551 (native_interpret_int): Likewise.
14552 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
14553 * internal-fn.c (expand_addsub_overflow): Likewise.
14554 (expand_neg_overflow): Likewise.
14555 (expand_mul_overflow): Likewise.
14556 (expand_arith_overflow): Likewise.
14557 * match.pd: Likewise.
14558 * stor-layout.c (layout_type): Likewise.
14559 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14560 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
14561 * tree-ssanames.c (get_range_info): Likewise.
14562 * tree-switch-conversion.c (array_value_type) Likewise.
14563 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
14564 (vect_recog_divmod_pattern): Likewise.
14565 (vect_recog_mixed_size_cond_pattern): Likewise.
14566 * tree-vrp.c (extract_range_basic): Likewise.
14567 (simplify_float_conversion_using_ranges): Likewise.
14568 * tree.c (int_fits_type_p): Likewise.
14569 * ubsan.c (instrument_bool_enum_load): Likewise.
14570 * varasm.c (mergeable_string_section): Likewise.
14571 (narrowing_initializer_constant_valid_p): Likewise.
14572 (output_constant): Likewise.
14574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14575 Alan Hayward <alan.hayward@arm.com>
14576 David Sherwood <david.sherwood@arm.com>
14578 * machmode.h (NARROWEST_INT_MODE): New macro.
14579 * expr.c (alignment_for_piecewise_move): Use it instead of
14580 GET_CLASS_NARROWEST_MODE (MODE_INT).
14581 (push_block): Likewise.
14582 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
14584 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14586 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14587 Alan Hayward <alan.hayward@arm.com>
14588 David Sherwood <david.sherwood@arm.com>
14590 * postreload.c (move2add_valid_value_p): Change the type of the
14591 mode parameter to scalar_int_mode.
14592 (move2add_use_add2_insn): Add a mode parameter and use it instead
14594 (move2add_use_add3_insn): Likewise.
14595 (reload_cse_move2add): Update accordingly.
14597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14598 Alan Hayward <alan.hayward@arm.com>
14599 David Sherwood <david.sherwood@arm.com>
14601 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
14603 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
14604 * optabs.c (expand_unop): Likewise.
14606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14607 Alan Hayward <alan.hayward@arm.com>
14608 David Sherwood <david.sherwood@arm.com>
14610 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
14611 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
14612 (popcount_loc_descriptor): Likewise.
14613 (bswap_loc_descriptor): Likewise.
14614 (rotate_loc_descriptor): Likewise.
14615 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
14616 calling the functions above.
14618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14619 Alan Hayward <alan.hayward@arm.com>
14620 David Sherwood <david.sherwood@arm.com>
14622 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
14624 (try_combine): Likewise.
14625 (simplify_if_then_else): Likewise.
14626 * cse.c (cse_insn): Likewise.
14627 * dwarf2out.c (mem_loc_descriptor): Likewise.
14628 * emit-rtl.c (gen_lowpart_common): Likewise.
14629 * simplify-rtx.c (simplify_truncation): Likewise.
14630 (simplify_binary_operation_1): Likewise.
14631 (simplify_const_relational_operation): Likewise.
14632 (simplify_ternary_operation): Likewise.
14633 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
14635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14636 Alan Hayward <alan.hayward@arm.com>
14637 David Sherwood <david.sherwood@arm.com>
14639 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
14640 * reload.c (push_reload): Likewise.
14641 (find_reloads): Likewise.
14643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14644 Alan Hayward <alan.hayward@arm.com>
14645 David Sherwood <david.sherwood@arm.com>
14647 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
14648 (make_compound_operation_int): Likewise.
14649 (change_zero_ext): Likewise.
14650 * expr.c (convert_move): Likewise.
14651 (convert_modes): Likewise.
14652 * fwprop.c (forward_propagate_subreg): Likewise.
14653 * loop-iv.c (get_biv_step_1): Likewise.
14654 * optabs.c (widen_operand): Likewise.
14655 * postreload.c (move2add_valid_value_p): Likewise.
14656 * recog.c (simplify_while_replacing): Likewise.
14657 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14658 (simplify_binary_operation_1): Likewise. Remove redundant
14659 mode equality check.
14661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14662 Alan Hayward <alan.hayward@arm.com>
14663 David Sherwood <david.sherwood@arm.com>
14665 * combine.c (combine_simplify_rtx): Add checks for
14666 is_a <scalar_int_mode>.
14667 (simplify_if_then_else): Likewise.
14668 (make_field_assignment): Likewise.
14669 (simplify_comparison): Likewise.
14670 * ifcvt.c (noce_try_bitop): Likewise.
14671 * loop-invariant.c (canonicalize_address_mult): Likewise.
14672 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14674 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14675 Alan Hayward <alan.hayward@arm.com>
14676 David Sherwood <david.sherwood@arm.com>
14678 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
14679 is_a <scalar_int_mode> instead of != BLKmode.
14681 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14682 Alan Hayward <alan.hayward@arm.com>
14683 David Sherwood <david.sherwood@arm.com>
14685 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
14686 instead of != VOIDmode.
14687 * combine.c (if_then_else_cond): Likewise.
14688 (change_zero_ext): Likewise.
14689 * dwarf2out.c (mem_loc_descriptor): Likewise.
14690 (loc_descriptor): Likewise.
14691 * rtlanal.c (canonicalize_condition): Likewise.
14692 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
14694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14695 Alan Hayward <alan.hayward@arm.com>
14696 David Sherwood <david.sherwood@arm.com>
14698 * simplify-rtx.c (simplify_binary_operation_1): Use
14699 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
14701 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14702 Alan Hayward <alan.hayward@arm.com>
14703 David Sherwood <david.sherwood@arm.com>
14705 * wide-int.h (int_traits<unsigned char>) New class.
14706 (int_traits<unsigned short>) Likewise.
14707 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
14708 Use GET_MODE_UNIT_PRECISION and remove redundant test for
14710 * combine.c (set_nonzero_bits_and_sign_copies): Use
14711 is_a <scalar_int_mode>.
14712 (find_split_point): Likewise.
14713 (combine_simplify_rtx): Likewise.
14714 (simplify_logical): Likewise.
14715 (expand_compound_operation): Likewise.
14716 (expand_field_assignment): Likewise.
14717 (make_compound_operation): Likewise.
14718 (extended_count): Likewise.
14719 (change_zero_ext): Likewise.
14720 (simplify_comparison): Likewise.
14721 * dwarf2out.c (scompare_loc_descriptor): Likewise.
14722 (ucompare_loc_descriptor): Likewise.
14723 (minmax_loc_descriptor): Likewise.
14724 (mem_loc_descriptor): Likewise.
14725 (loc_descriptor): Likewise.
14726 * expmed.c (init_expmed_one_mode): Likewise.
14727 * lra-constraints.c (lra_constraint_offset): Likewise.
14728 * optabs.c (prepare_libcall_arg): Likewise.
14729 * postreload.c (move2add_note_store): Likewise.
14730 * reload.c (operands_match_p): Likewise.
14731 * rtl.h (load_extend_op): Likewise.
14732 * rtlhooks.c (gen_lowpart_general): Likewise.
14733 * simplify-rtx.c (simplify_truncation): Likewise.
14734 (simplify_unary_operation_1): Likewise.
14735 (simplify_binary_operation_1): Likewise.
14736 (simplify_const_binary_operation): Likewise.
14737 (simplify_const_relational_operation): Likewise.
14738 (simplify_subreg): Likewise.
14739 * stor-layout.c (bitwise_mode_for_mode): Likewise.
14740 * var-tracking.c (adjust_mems): Likewise.
14741 (prepare_call_arguments): Likewise.
14743 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14744 Alan Hayward <alan.hayward@arm.com>
14745 David Sherwood <david.sherwood@arm.com>
14747 * machmode.h (is_int_mode): New fuction.
14748 * combine.c (find_split_point): Use it.
14749 (combine_simplify_rtx): Likewise.
14750 (simplify_if_then_else): Likewise.
14751 (simplify_set): Likewise.
14752 (simplify_shift_const_1): Likewise.
14753 (simplify_comparison): Likewise.
14754 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
14755 * cse.c (notreg_cost): Likewise.
14756 (cse_insn): Likewise.
14757 * cselib.c (cselib_lookup_1): Likewise.
14758 * dojump.c (do_jump_1): Likewise.
14759 (do_compare_rtx_and_jump): Likewise.
14760 * dse.c (get_call_args): Likewise.
14761 * dwarf2out.c (rtl_for_decl_init): Likewise.
14762 (native_encode_initializer): Likewise.
14763 * expmed.c (emit_store_flag_1): Likewise.
14764 (emit_store_flag): Likewise.
14765 * expr.c (convert_modes): Likewise.
14766 (store_field): Likewise.
14767 (expand_expr_real_1): Likewise.
14768 * fold-const.c (fold_read_from_constant_string): Likewise.
14769 * gimple-ssa-sprintf.c (get_format_string): Likewise.
14770 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
14771 * optabs.c (expand_binop): Likewise.
14772 (expand_unop): Likewise.
14773 (expand_abs_nojump): Likewise.
14774 (expand_one_cmpl_abs_nojump): Likewise.
14775 * simplify-rtx.c (mode_signbit_p): Likewise.
14776 (val_signbit_p): Likewise.
14777 (val_signbit_known_set_p): Likewise.
14778 (val_signbit_known_clear_p): Likewise.
14779 (simplify_relational_operation_1): Likewise.
14780 * tree.c (vector_type_mode): Likewise.
14782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14783 Alan Hayward <alan.hayward@arm.com>
14784 David Sherwood <david.sherwood@arm.com>
14786 * machmode.h (smallest_mode_for_size): Fix formatting.
14787 (smallest_int_mode_for_size): New function.
14788 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
14789 instead of smallest_mode_for_size.
14790 * combine.c (make_extraction): Likewise.
14791 * config/arc/arc.c (arc_expand_movmem): Likewise.
14792 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
14793 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14794 * config/s390/s390.c (s390_expand_insv): Likewise.
14795 * config/sparc/sparc.c (assign_int_registers): Likewise.
14796 * config/spu/spu.c (spu_function_value): Likewise.
14797 (spu_function_arg): Likewise.
14798 * coverage.c (get_gcov_type): Likewise.
14799 (get_gcov_unsigned_t): Likewise.
14800 * dse.c (find_shift_sequence): Likewise.
14801 * expmed.c (store_bit_field_1): Likewise.
14802 * expr.c (convert_move): Likewise.
14803 (store_field): Likewise.
14804 * internal-fn.c (expand_arith_overflow): Likewise.
14805 * optabs-query.c (get_best_extraction_insn): Likewise.
14806 * optabs.c (expand_twoval_binop_libfunc): Likewise.
14807 * stor-layout.c (layout_type): Likewise.
14808 (initialize_sizetypes): Likewise.
14809 * targhooks.c (default_get_mask_mode): Likewise.
14810 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
14812 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14813 Alan Hayward <alan.hayward@arm.com>
14814 David Sherwood <david.sherwood@arm.com>
14816 * machmode.h (opt_mode::else_blk): New function.
14817 (int_mode_for_mode): Declare.
14818 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
14819 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
14821 * cfgexpand.c (expand_debug_expr): Likewise.
14822 * combine.c (gen_lowpart_or_truncate): Likewise.
14823 (gen_lowpart_for_combine): Likewise.
14824 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
14825 * config/avr/avr.c (avr_to_int_mode): Likewise.
14826 (avr_out_plus_1): Likewise.
14827 (avr_out_plus): Likewise.
14828 (avr_out_round): Likewise.
14829 * config/i386/i386.c (ix86_split_to_parts): Likewise.
14830 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14831 (s390_expand_vcond): Likewise.
14832 * config/spu/spu.c (spu_split_immediate): Likewise.
14833 (spu_expand_mov): Likewise.
14834 * dse.c (get_stored_val): Likewise.
14835 * expmed.c (store_bit_field_1): Likewise.
14836 (convert_extracted_bit_field): Use int_mode_for_mode instead of
14838 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
14839 (extract_low_bits): Likewise.
14840 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
14841 handling rather than repeating the check.
14842 (emit_group_store): Likewise.
14843 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
14844 * optabs.c (expand_absneg_bit): Likewise.
14845 (expand_copysign_absneg): Likewise.
14846 (expand_copysign_bit): Likewise.
14847 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
14848 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
14849 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
14850 * var-tracking.c (prepare_call_arguments): Likewise.
14851 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
14852 int_mode_for_mode instead of mode_for_size.
14853 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14856 Alan Hayward <alan.hayward@arm.com>
14857 David Sherwood <david.sherwood@arm.com>
14859 * machmode.h (int_mode_for_size): New function.
14860 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
14861 instead of mode_for_size.
14862 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
14864 * combine.c (expand_field_assignment): Use int_mode_for_size
14865 instead of mode_for_size.
14866 (make_extraction): Likewise.
14867 (simplify_shift_const_1): Likewise.
14868 (simplify_comparison): Likewise.
14869 * dojump.c (do_jump): Likewise.
14870 * dwarf2out.c (mem_loc_descriptor): Likewise.
14871 * emit-rtl.c (init_derived_machine_modes): Likewise.
14872 * expmed.c (flip_storage_order): Likewise.
14873 (convert_extracted_bit_field): Likewise.
14874 * expr.c (copy_blkmode_from_reg): Likewise.
14875 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
14876 * internal-fn.c (expand_mul_overflow): Likewise.
14877 * lower-subreg.c (simple_move): Likewise.
14878 * optabs-libfuncs.c (init_optabs): Likewise.
14879 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14880 * tree.c (vector_type_mode): Likewise.
14881 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
14882 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
14883 * tree-vect-generic.c (expand_vector_parallel): Likewise.
14884 * tree-vect-stmts.c (vectorizable_load): Likewise.
14885 (vectorizable_store): Likewise.
14887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14888 Alan Hayward <alan.hayward@arm.com>
14889 David Sherwood <david.sherwood@arm.com>
14891 * coretypes.h (pod_mode): New type.
14892 (scalar_int_mode_pod): New typedef.
14893 * machmode.h (pod_mode): New class.
14894 (int_n_data_t::m): Change type to scalar_int_mode_pod.
14895 * genmodes.c (emit_mode_int_n): Update accordingly.
14896 * lower-subreg.h (target_lower_subreg): Change type to
14897 scalar_int_mode_pod.
14898 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
14899 scalar_int_mode_pod.
14901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14902 Alan Hayward <alan.hayward@arm.com>
14903 David Sherwood <david.sherwood@arm.com>
14905 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
14906 machine_mode to scalar_int_mode.
14907 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
14908 (rs6000_option_override_internal): Remove cast to int.
14909 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
14910 machine_mode to scalar_int_mode.
14911 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
14912 (rs6000_option_override_internal): Remove cast to int.
14913 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
14914 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
14916 * config/s390/s390.c (s390_expand_builtin): Likewise.
14917 * coretypes.h (scalar_int_mode): New type.
14918 (opt_scalar_int_mode): New typedef.
14919 * machmode.h (scalar_int_mode): New class.
14920 (scalar_int_mode::includes_p): New function.
14921 (byte_mode): Change type to scalar_int_mode.
14922 (word_mode): Likewise.
14923 (ptr_mode): Likewise.
14924 * emit-rtl.c (byte_mode): Likewise.
14925 (word_mode): Likewise.
14926 (ptr_mode): Likewise.
14927 (init_derived_machine_modes): Update accordingly.
14928 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
14929 and MODE_PARTIAL_INT.
14930 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
14931 opt_scalar_int_mode.
14933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14934 Alan Hayward <alan.hayward@arm.com>
14935 David Sherwood <david.sherwood@arm.com>
14937 * target.def (libgcc_floating_mode_supported_p): Take a
14939 * doc/tm.texi: Regenerate.
14940 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
14942 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
14943 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
14946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14947 Alan Hayward <alan.hayward@arm.com>
14948 David Sherwood <david.sherwood@arm.com>
14950 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
14951 * doc/tm.texi: Regenerate.
14952 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
14953 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
14954 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
14955 * targhooks.h (default_floatn_mode): Likewise.
14956 * targhooks.c (default_floatn_mode): Likewise.
14957 * tree.c (build_common_tree_nodes): Update accordingly.
14959 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14960 Alan Hayward <alan.hayward@arm.com>
14961 David Sherwood <david.sherwood@arm.com>
14963 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
14964 (mode_iterator::iterate_p): Likewise.
14965 (mode_iterator::get_wider): Likewise.
14966 * expr.c (init_expr_target): Use opt_scalar_float_mode.
14968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14969 Alan Hayward <alan.hayward@arm.com>
14970 David Sherwood <david.sherwood@arm.com>
14972 * coretypes.h (opt_scalar_float_mode): New typedef.
14973 * machmode.h (float_mode_for_size): New function.
14974 * emit-rtl.c (double_mode): Delete.
14975 (init_emit_once): Use float_mode_for_size.
14976 * stor-layout.c (layout_type): Likewise.
14977 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
14979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14980 Alan Hayward <alan.hayward@arm.com>
14981 David Sherwood <david.sherwood@arm.com>
14983 * output.h (assemble_real): Take a scalar_float_mode.
14984 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
14985 * config/arm/arm.md (consttable_4): Likewise.
14986 (consttable_8): Likewise.
14987 (consttable_16): Likewise.
14988 * config/mips/mips.md (consttable_float): Likewise.
14989 * config/s390/s390.c (s390_output_pool_entry): Likewise.
14990 * varasm.c (assemble_real): Take a scalar_float_mode.
14991 (output_constant_pool_2): Update accordingly.
14992 (output_constant): Likewise.
14994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14995 Alan Hayward <alan.hayward@arm.com>
14996 David Sherwood <david.sherwood@arm.com>
14998 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
14999 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
15000 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
15001 (native_encode_real): Likewise.
15002 (native_interpret_real): Likewise.
15003 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
15004 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15007 Alan Hayward <alan.hayward@arm.com>
15008 David Sherwood <david.sherwood@arm.com>
15010 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
15011 <scalar_float_mode>. Simplify.
15012 (gen_extend_conv_libfunc): Likewise.
15014 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15015 Alan Hayward <alan.hayward@arm.com>
15016 David Sherwood <david.sherwood@arm.com>
15018 * coretypes.h (scalar_float_mode): New type.
15019 * machmode.h (mode_traits::from_int): Use machine_mode if
15020 USE_ENUM_MODES is defined.
15021 (is_a): New function.
15023 (dyn_cast): Likewise.
15024 (scalar_float_mode): New class.
15025 (scalar_float_mode::includes_p): New function.
15026 (is_float_mode): Likewise.
15027 * gdbhooks.py (MachineModePrinter): New class.
15028 (build_pretty_printer): Use it for scalar_float_mode.
15029 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
15030 (format_helper::format_helper): Turn into a template.
15031 * genmodes.c (get_mode_class): New function.
15032 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
15033 or machine_mode if none.
15034 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
15035 as_a <scalar_float_mode>.
15036 * dwarf2out.c (mem_loc_descriptor): Likewise.
15037 (insert_float): Likewise.
15038 (add_const_value_attribute): Likewise.
15039 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15040 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
15041 (expand_unop): Update accordingly.
15042 (expand_abs_nojump): Likewise.
15043 (expand_copysign_absneg): Take a scalar_float_mode.
15044 (expand_copysign_bit): Likewise.
15045 (expand_copysign): Update accordingly.
15047 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15048 Alan Hayward <alan.hayward@arm.com>
15049 David Sherwood <david.sherwood@arm.com>
15051 * coretypes.h (opt_mode): New class.
15052 * machmode.h (opt_mode): Likewise.
15053 (opt_mode::else_void): New function.
15054 (opt_mode::require): Likewise.
15055 (opt_mode::exists): Likewise.
15056 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
15057 (GET_MODE_2XWIDER_MODE): Likewise.
15058 (mode_iterator::get_wider): Update accordingly.
15059 (mode_iterator::get_2xwider): Likewise.
15060 (mode_iterator::get_known_wider): Likewise, turning into a template.
15061 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
15062 forcing a wider mode to exist.
15063 * config/cr16/cr16.h (LONG_REG_P): Likewise.
15064 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15065 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
15066 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
15067 * lower-subreg.c (init_lower_subreg): Likewise.
15068 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
15069 on the final iteration.
15070 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
15071 a wider mode exists before asking for a move pattern.
15072 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
15073 forcing a wider mode to exist.
15074 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
15075 returning false if no such mode exists.
15076 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
15077 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
15078 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
15079 Avoid checking for a MODE_INT if we already know the mode is not a
15081 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
15082 forcing a wider mode to exist.
15083 (expmed_mult_highpart_optab): Likewise.
15084 (expmed_mult_highpart): Likewise.
15085 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
15087 * lto-streamer-in.c (lto_input_mode_table): Likewise.
15088 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
15089 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15090 * internal-fn.c (expand_mul_overflow): Update use of
15091 GET_MODE_2XWIDER_MODE.
15092 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15093 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
15094 GET_MODE_WIDER_MODE.
15095 (convert_plusminus_to_widen): Likewise.
15096 * tree-switch-conversion.c (array_value_type): Likewise.
15097 * var-tracking.c (emit_note_insn_var_location): Likewise.
15098 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15099 Return false inside rather than outside the loop if no wider mode
15101 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
15102 and GET_MODE_2XWIDER_MODE
15103 (can_compare_p): Use else_void.
15104 * gdbhooks.py (OptMachineModePrinter): New class.
15105 (build_pretty_printer): Use it for opt_mode.
15107 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15108 Alan Hayward <alan.hayward@arm.com>
15109 David Sherwood <david.sherwood@arm.com>
15111 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
15112 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
15114 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15115 Alan Hayward <alan.hayward@arm.com>
15116 David Sherwood <david.sherwood@arm.com>
15118 * machmode.h (mode_traits): New structure.
15119 (get_narrowest_mode): New function.
15120 (mode_iterator::start): Likewise.
15121 (mode_iterator::iterate_p): Likewise.
15122 (mode_iterator::get_wider): Likewise.
15123 (mode_iterator::get_known_wider): Likewise.
15124 (mode_iterator::get_2xwider): Likewise.
15125 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
15126 (FOR_EACH_MODE): Likewise.
15127 (FOR_EACH_MODE_FROM): Likewise.
15128 (FOR_EACH_MODE_UNTIL): Likewise.
15129 (FOR_EACH_WIDER_MODE): Likewise.
15130 (FOR_EACH_2XWIDER_MODE): Likewise.
15131 * builtins.c (expand_builtin_strlen): Use new mode iterators.
15132 * combine.c (simplify_comparison): Likewise
15133 * config/i386/i386.c (type_natural_mode): Likewise.
15134 * cse.c (cse_insn): Likewise.
15135 * dse.c (find_shift_sequence): Likewise.
15136 * emit-rtl.c (init_derived_machine_modes): Likewise.
15137 (init_emit_once): Likewise.
15138 * explow.c (hard_function_value): Likewise.
15139 * expmed.c (extract_fixed_bit_field_1): Likewise.
15140 (extract_bit_field_1): Likewise.
15141 (expand_divmod): Likewise.
15142 (emit_store_flag_1): Likewise.
15143 * expr.c (init_expr_target): Likewise.
15144 (convert_move): Likewise.
15145 (alignment_for_piecewise_move): Likewise.
15146 (widest_int_mode_for_size): Likewise.
15147 (emit_block_move_via_movmem): Likewise.
15148 (copy_blkmode_to_reg): Likewise.
15149 (set_storage_via_setmem): Likewise.
15150 (compress_float_constant): Likewise.
15151 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15152 * optabs-query.c (get_best_extraction_insn): Likewise.
15153 * optabs.c (expand_binop): Likewise.
15154 (expand_twoval_unop): Likewise.
15155 (expand_twoval_binop): Likewise.
15156 (widen_leading): Likewise.
15157 (widen_bswap): Likewise.
15158 (expand_parity): Likewise.
15159 (expand_unop): Likewise.
15160 (prepare_cmp_insn): Likewise.
15161 (prepare_float_lib_cmp): Likewise.
15162 (expand_float): Likewise.
15163 (expand_fix): Likewise.
15164 (expand_sfix_optab): Likewise.
15165 * postreload.c (move2add_use_add2_insn): Likewise.
15166 * reg-stack.c (reg_to_stack): Likewise.
15167 * reginfo.c (choose_hard_reg_mode): Likewise.
15168 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15169 * stor-layout.c (mode_for_size): Likewise.
15170 (smallest_mode_for_size): Likewise.
15171 (mode_for_vector): Likewise.
15172 (finish_bitfield_representative): Likewise.
15173 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
15174 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
15175 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15176 * var-tracking.c (prepare_call_arguments): Likewise.
15178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15179 Alan Hayward <alan.hayward@arm.com>
15180 David Sherwood <david.sherwood@arm.com>
15182 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
15183 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
15184 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
15185 * machmode.h (mode_size): Move earlier in file.
15186 (mode_precision): Likewise.
15187 (mode_inner): Likewise.
15188 (mode_nunits): Likewise.
15189 (mode_unit_size): Likewise.
15190 (unit_unit_precision): Likewise.
15191 (mode_wider): Likewise.
15192 (mode_2xwider): Likewise.
15193 (machine_mode): New class.
15194 (mode_to_bytes): New function.
15195 (mode_to_bits): Likewise.
15196 (mode_to_precision): Likewise.
15197 (mode_to_inner): Likewise.
15198 (mode_to_unit_size): Likewise.
15199 (mode_to_unit_precision): Likewise.
15200 (mode_to_nunits): Likewise.
15201 (GET_MODE_SIZE): Use mode_to_bytes.
15202 (GET_MODE_BITSIZE): Use mode_to_bits.
15203 (GET_MODE_PRECISION): Use mode_to_precision.
15204 (GET_MODE_INNER): Use mode_to_inner.
15205 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
15206 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
15207 (GET_MODE_NUNITS): Use mode_to_nunits.
15208 * system.h (ALWAYS_INLINE): New macro.
15209 * config/powerpcspe/powerpcspe-c.c
15210 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
15211 int for arg1_mode and arg2_mode.
15213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15214 Alan Hayward <alan.hayward@arm.com>
15215 David Sherwood <david.sherwood@arm.com>
15217 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
15218 Prefix mode names with E_ in case statements.
15219 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15220 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
15221 (aarch64_split_simd_move): Likewise.
15222 (aarch64_gen_storewb_pair): Likewise.
15223 (aarch64_gen_loadwb_pair): Likewise.
15224 (aarch64_gen_store_pair): Likewise.
15225 (aarch64_gen_load_pair): Likewise.
15226 (aarch64_get_condition_code_1): Likewise.
15227 (aarch64_constant_pool_reload_icode): Likewise.
15228 (get_rsqrte_type): Likewise.
15229 (get_rsqrts_type): Likewise.
15230 (get_recpe_type): Likewise.
15231 (get_recps_type): Likewise.
15232 (aarch64_gimplify_va_arg_expr): Likewise.
15233 (aarch64_simd_container_mode): Likewise.
15234 (aarch64_emit_load_exclusive): Likewise.
15235 (aarch64_emit_store_exclusive): Likewise.
15236 (aarch64_expand_compare_and_swap): Likewise.
15237 (aarch64_gen_atomic_cas): Likewise.
15238 (aarch64_emit_bic): Likewise.
15239 (aarch64_emit_atomic_swap): Likewise.
15240 (aarch64_emit_atomic_load_op): Likewise.
15241 (aarch64_evpc_trn): Likewise.
15242 (aarch64_evpc_uzp): Likewise.
15243 (aarch64_evpc_zip): Likewise.
15244 (aarch64_evpc_ext): Likewise.
15245 (aarch64_evpc_rev): Likewise.
15246 (aarch64_evpc_dup): Likewise.
15247 (aarch64_gen_ccmp_first): Likewise.
15248 (aarch64_gen_ccmp_next): Likewise.
15249 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15250 (alpha_emit_xfloating_libcall): Likewise.
15251 (emit_insxl): Likewise.
15252 (alpha_arg_type): Likewise.
15253 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
15254 (arc_preferred_simd_mode): Likewise.
15255 (arc_secondary_reload): Likewise.
15256 (get_arc_condition_code): Likewise.
15257 (arc_print_operand): Likewise.
15258 (arc_legitimate_constant_p): Likewise.
15259 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15260 * config/arc/arc.md (casesi_load): Likewise.
15261 (casesi_compact_jump): Likewise.
15262 * config/arc/predicates.md (proper_comparison_operator): Likewise.
15263 (cc_use_register): Likewise.
15264 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15265 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
15266 (arm_init_iwmmxt_builtins): Likewise.
15267 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
15268 (neon_expand_vector_init): Likewise.
15269 (arm_attr_length_move_neon): Likewise.
15270 (maybe_get_arm_condition_code): Likewise.
15271 (arm_emit_vector_const): Likewise.
15272 (arm_preferred_simd_mode): Likewise.
15273 (arm_output_iwmmxt_tinsr): Likewise.
15274 (thumb1_output_casesi): Likewise.
15275 (thumb2_output_casesi): Likewise.
15276 (arm_emit_load_exclusive): Likewise.
15277 (arm_emit_store_exclusive): Likewise.
15278 (arm_expand_compare_and_swap): Likewise.
15279 (arm_evpc_neon_vuzp): Likewise.
15280 (arm_evpc_neon_vzip): Likewise.
15281 (arm_evpc_neon_vrev): Likewise.
15282 (arm_evpc_neon_vtrn): Likewise.
15283 (arm_evpc_neon_vext): Likewise.
15284 (arm_validize_comparison): Likewise.
15285 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
15286 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
15287 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
15288 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
15289 (c6x_preferred_simd_mode): Likewise.
15290 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
15291 (epiphany_rtx_costs): Likewise.
15292 * config/epiphany/predicates.md (proper_comparison_operator):
15294 * config/frv/frv.c (condexec_memory_operand): Likewise.
15295 (frv_emit_move): Likewise.
15296 (output_move_single): Likewise.
15297 (output_condmove_single): Likewise.
15298 (frv_hard_regno_mode_ok): Likewise.
15299 (frv_matching_accg_mode): Likewise.
15300 * config/h8300/h8300.c (split_adds_subs): Likewise.
15301 (h8300_rtx_costs): Likewise.
15302 (h8300_print_operand): Likewise.
15303 (compute_mov_length): Likewise.
15304 (output_logical_op): Likewise.
15305 (compute_logical_op_length): Likewise.
15306 (compute_logical_op_cc): Likewise.
15307 (h8300_shift_needs_scratch_p): Likewise.
15308 (output_a_shift): Likewise.
15309 (compute_a_shift_length): Likewise.
15310 (compute_a_shift_cc): Likewise.
15311 (expand_a_rotate): Likewise.
15312 (output_a_rotate): Likewise.
15313 * config/i386/i386.c (classify_argument): Likewise.
15314 (function_arg_advance_32): Likewise.
15315 (function_arg_32): Likewise.
15316 (function_arg_64): Likewise.
15317 (function_value_64): Likewise.
15318 (ix86_gimplify_va_arg): Likewise.
15319 (ix86_legitimate_constant_p): Likewise.
15320 (put_condition_code): Likewise.
15321 (split_double_mode): Likewise.
15322 (ix86_avx256_split_vector_move_misalign): Likewise.
15323 (ix86_expand_vector_logical_operator): Likewise.
15324 (ix86_split_idivmod): Likewise.
15325 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
15326 (ix86_build_const_vector): Likewise.
15327 (ix86_build_signbit_mask): Likewise.
15328 (ix86_match_ccmode): Likewise.
15329 (ix86_cc_modes_compatible): Likewise.
15330 (ix86_expand_branch): Likewise.
15331 (ix86_expand_sse_cmp): Likewise.
15332 (ix86_expand_sse_movcc): Likewise.
15333 (ix86_expand_int_sse_cmp): Likewise.
15334 (ix86_expand_vec_perm_vpermi2): Likewise.
15335 (ix86_expand_vec_perm): Likewise.
15336 (ix86_expand_sse_unpack): Likewise.
15337 (ix86_expand_int_addcc): Likewise.
15338 (ix86_split_to_parts): Likewise.
15339 (ix86_vectorize_builtin_gather): Likewise.
15340 (ix86_vectorize_builtin_scatter): Likewise.
15341 (avx_vpermilp_parallel): Likewise.
15342 (inline_memory_move_cost): Likewise.
15343 (ix86_tieable_integer_mode_p): Likewise.
15344 (x86_maybe_negate_const_int): Likewise.
15345 (ix86_expand_vector_init_duplicate): Likewise.
15346 (ix86_expand_vector_init_one_nonzero): Likewise.
15347 (ix86_expand_vector_init_one_var): Likewise.
15348 (ix86_expand_vector_init_concat): Likewise.
15349 (ix86_expand_vector_init_interleave): Likewise.
15350 (ix86_expand_vector_init_general): Likewise.
15351 (ix86_expand_vector_set): Likewise.
15352 (ix86_expand_vector_extract): Likewise.
15353 (emit_reduc_half): Likewise.
15354 (ix86_emit_i387_round): Likewise.
15355 (ix86_mangle_type): Likewise.
15356 (ix86_expand_round_sse4): Likewise.
15357 (expand_vec_perm_blend): Likewise.
15358 (canonicalize_vector_int_perm): Likewise.
15359 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
15360 (expand_vec_perm_1): Likewise.
15361 (expand_vec_perm_interleave3): Likewise.
15362 (expand_vec_perm_even_odd_pack): Likewise.
15363 (expand_vec_perm_even_odd_1): Likewise.
15364 (expand_vec_perm_broadcast_1): Likewise.
15365 (ix86_vectorize_vec_perm_const_ok): Likewise.
15366 (ix86_expand_vecop_qihi): Likewise.
15367 (ix86_expand_mul_widen_hilo): Likewise.
15368 (ix86_expand_sse2_abs): Likewise.
15369 (ix86_expand_pextr): Likewise.
15370 (ix86_expand_pinsr): Likewise.
15371 (ix86_preferred_simd_mode): Likewise.
15372 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
15373 * config/i386/sse.md (*andnot<mode>3): Likewise.
15374 (<mask_codefor><code><mode>3<mask_name>): Likewise.
15375 (*<code><mode>3): Likewise.
15376 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
15377 (ia64_expand_atomic_op): Likewise.
15378 (ia64_arg_type): Likewise.
15379 (ia64_mode_to_int): Likewise.
15380 (ia64_scalar_mode_supported_p): Likewise.
15381 (ia64_vector_mode_supported_p): Likewise.
15382 (expand_vec_perm_broadcast): Likewise.
15383 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
15384 (iq2000_function_arg_advance): Likewise.
15385 (iq2000_function_arg): Likewise.
15386 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
15387 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
15388 (m68k_libcall_value): Likewise.
15389 (m68k_function_value): Likewise.
15390 (sched_attr_op_type): Likewise.
15391 * config/mcore/mcore.c (mcore_output_move): Likewise.
15392 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
15394 (microblaze_function_arg): Likewise.
15395 * config/mips/mips.c (mips16_build_call_stub): Likewise.
15396 (mips_print_operand): Likewise.
15397 (mips_mode_ok_for_mov_fmt_p): Likewise.
15398 (mips_vector_mode_supported_p): Likewise.
15399 (mips_preferred_simd_mode): Likewise.
15400 (mips_expand_vpc_loongson_even_odd): Likewise.
15401 (mips_expand_vec_unpack): Likewise.
15402 (mips_expand_vi_broadcast): Likewise.
15403 (mips_expand_vector_init): Likewise.
15404 (mips_expand_vec_reduc): Likewise.
15405 (mips_expand_msa_cmp): Likewise.
15406 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
15407 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
15408 (cc_flags_for_mode): Likewise.
15409 * config/msp430/msp430.c (msp430_print_operand): Likewise.
15410 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
15411 (nds32_output_casesi_pc_relative): Likewise.
15412 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15413 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
15414 (nvptx_gen_unpack): Likewise.
15415 (nvptx_gen_pack): Likewise.
15416 (nvptx_gen_shuffle): Likewise.
15417 (nvptx_gen_wcast): Likewise.
15418 (nvptx_preferred_simd_mode): Likewise.
15419 * config/pa/pa.c (pa_secondary_reload): Likewise.
15420 * config/pa/predicates.md (base14_operand): Likewise.
15421 * config/powerpcspe/powerpcspe-c.c
15422 (altivec_resolve_overloaded_builtin): Likewise.
15423 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
15425 (rs6000_preferred_simd_mode): Likewise.
15426 (output_vec_const_move): Likewise.
15427 (rs6000_expand_vector_extract): Likewise.
15428 (rs6000_split_vec_extract_var): Likewise.
15429 (reg_offset_addressing_ok_p): Likewise.
15430 (rs6000_legitimate_offset_address_p): Likewise.
15431 (rs6000_legitimize_address): Likewise.
15432 (rs6000_emit_set_const): Likewise.
15433 (rs6000_const_vec): Likewise.
15434 (rs6000_emit_move): Likewise.
15435 (spe_build_register_parallel): Likewise.
15436 (rs6000_darwin64_record_arg_recurse): Likewise.
15437 (swap_selector_for_mode): Likewise.
15438 (spe_init_builtins): Likewise.
15439 (paired_init_builtins): Likewise.
15440 (altivec_init_builtins): Likewise.
15441 (do_load_for_compare): Likewise.
15442 (rs6000_generate_compare): Likewise.
15443 (rs6000_expand_float128_convert): Likewise.
15444 (emit_load_locked): Likewise.
15445 (emit_store_conditional): Likewise.
15446 (rs6000_output_function_epilogue): Likewise.
15447 (rs6000_handle_altivec_attribute): Likewise.
15448 (rs6000_function_value): Likewise.
15449 (emit_fusion_gpr_load): Likewise.
15450 (emit_fusion_p9_load): Likewise.
15451 (emit_fusion_p9_store): Likewise.
15452 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
15453 (fusion_gpr_mem_load): Likewise.
15454 (fusion_addis_mem_combo_load): Likewise.
15455 (fusion_addis_mem_combo_store): Likewise.
15456 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
15457 (fusion_gpr_mem_load): Likewise.
15458 (fusion_addis_mem_combo_load): Likewise.
15459 (fusion_addis_mem_combo_store): Likewise.
15460 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15462 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
15463 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15464 (rs6000_preferred_simd_mode): Likewise.
15465 (output_vec_const_move): Likewise.
15466 (rs6000_expand_vector_extract): Likewise.
15467 (rs6000_split_vec_extract_var): Likewise.
15468 (reg_offset_addressing_ok_p): Likewise.
15469 (rs6000_legitimate_offset_address_p): Likewise.
15470 (rs6000_legitimize_address): Likewise.
15471 (rs6000_emit_set_const): Likewise.
15472 (rs6000_const_vec): Likewise.
15473 (rs6000_emit_move): Likewise.
15474 (rs6000_darwin64_record_arg_recurse): Likewise.
15475 (swap_selector_for_mode): Likewise.
15476 (paired_init_builtins): Likewise.
15477 (altivec_init_builtins): Likewise.
15478 (rs6000_expand_float128_convert): Likewise.
15479 (emit_load_locked): Likewise.
15480 (emit_store_conditional): Likewise.
15481 (rs6000_output_function_epilogue): Likewise.
15482 (rs6000_handle_altivec_attribute): Likewise.
15483 (rs6000_function_value): Likewise.
15484 (emit_fusion_gpr_load): Likewise.
15485 (emit_fusion_p9_load): Likewise.
15486 (emit_fusion_p9_store): Likewise.
15487 * config/rx/rx.c (rx_gen_move_template): Likewise.
15488 (flags_from_mode): Likewise.
15489 * config/s390/predicates.md (s390_alc_comparison): Likewise.
15490 (s390_slb_comparison): Likewise.
15491 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
15492 (s390_vector_mode_supported_p): Likewise.
15493 (s390_cc_modes_compatible): Likewise.
15494 (s390_match_ccmode_set): Likewise.
15495 (s390_canonicalize_comparison): Likewise.
15496 (s390_emit_compare_and_swap): Likewise.
15497 (s390_branch_condition_mask): Likewise.
15498 (s390_rtx_costs): Likewise.
15499 (s390_secondary_reload): Likewise.
15500 (__SECONDARY_RELOAD_CASE): Likewise.
15501 (s390_expand_cs): Likewise.
15502 (s390_preferred_simd_mode): Likewise.
15503 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
15504 * config/sh/sh.c (sh_print_operand): Likewise.
15505 (dump_table): Likewise.
15506 (sh_secondary_reload): Likewise.
15507 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15508 * config/sh/sh.md (casesi_worker_1): Likewise.
15509 (casesi_worker_2): Likewise.
15510 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
15511 (fcc_comparison_operator): Likewise.
15512 * config/sparc/sparc.c (sparc_expand_move): Likewise.
15513 (emit_soft_tfmode_cvt): Likewise.
15514 (sparc_preferred_simd_mode): Likewise.
15515 (output_cbranch): Likewise.
15516 (sparc_print_operand): Likewise.
15517 (sparc_expand_vec_perm_bmask): Likewise.
15518 (vector_init_bshuffle): Likewise.
15519 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15520 (spu_vector_mode_supported_p): Likewise.
15521 (spu_expand_insv): Likewise.
15522 (spu_emit_branch_or_set): Likewise.
15523 (spu_handle_vector_attribute): Likewise.
15524 (spu_builtin_splats): Likewise.
15525 (spu_builtin_extract): Likewise.
15526 (spu_builtin_promote): Likewise.
15527 (spu_expand_sign_extend): Likewise.
15528 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15529 (tilegx_simd_int): Likewise.
15530 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
15531 (tilepro_simd_int): Likewise.
15532 * config/v850/v850.c (const_double_split): Likewise.
15533 (v850_print_operand): Likewise.
15534 (ep_memory_offset): Likewise.
15535 * config/vax/vax.c (vax_rtx_costs): Likewise.
15536 (vax_output_int_move): Likewise.
15537 (vax_output_int_add): Likewise.
15538 (vax_output_int_subtract): Likewise.
15539 * config/visium/predicates.md (visium_branch_operator): Likewise.
15540 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
15541 (visium_print_operand_address): Likewise.
15542 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15543 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
15544 (xtensa_expand_conditional_branch): Likewise.
15545 (xtensa_copy_incoming_a7): Likewise.
15546 (xtensa_output_literal): Likewise.
15547 * dfp.c (decimal_real_maxval): Likewise.
15548 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15550 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15551 Alan Hayward <alan.hayward@arm.com>
15552 David Sherwood <david.sherwood@arm.com>
15554 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
15555 (mode_nunits_inline): Likewise.
15556 (mode_inner_inline): Likewise.
15557 (mode_unit_size_inline): Likewise.
15558 (mode_unit_precision_inline): Likewise.
15559 (emit_insn_modes_h): Likewise. Also emit a #define of the
15561 (emit_mode_wider): Add an E_ prefix to mode names.
15562 (emit_mode_complex): Likewise.
15563 (emit_mode_inner): Likewise.
15564 (emit_mode_adjustments): Likewise.
15565 (emit_mode_int_n): Likewise.
15566 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
15567 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
15568 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
15569 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
15570 (CRC32_BUILTIN, ENTRY): Likewise.
15571 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
15572 (aarch64_pop_regs): Likewise.
15573 (aarch64_process_components): Likewise.
15574 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
15575 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
15576 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
15577 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
15578 * config/arm/arm.c (arm_init_libfuncs): Likewise.
15579 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
15581 * config/i386/i386-builtin.def (pcmpestr): Likewise.
15582 (pcmpistr): Likewise.
15583 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
15584 * config/mmix/mmix.c (mmix_output_condition): Likewise.
15585 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
15587 * config/rl78/rl78.c (mduc_regs): Likewise.
15588 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15589 (htm_expand_builtin): Likewise.
15590 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
15591 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
15592 * config/xtensa/xtensa.c (print_operand): Likewise.
15593 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
15594 (NUM_MODE_VECTOR_INT): Likewise.
15595 * genoutput.c (null_operand): Likewise.
15596 (output_operand_data): Likewise.
15597 * genrecog.c (print_parameter_value): Likewise.
15598 * lra.c (debug_operand_data): Likewise.
15600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15602 * dwarf2out.c (scompare_loc_descriptor_wide)
15603 (scompare_loc_descriptor_narrow): New functions, split out from...
15604 (scompare_loc_descriptor): ...here.
15605 * expmed.c (emit_store_flag_int): New function, split out from...
15606 (emit_store_flag): ...here.
15608 2017-08-30 Richard Biener <rguenther@suse.de>
15610 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
15611 (dwarf2out_early_finish): Move setting of AT_pubnames from
15612 early debug output to early finish.
15614 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15616 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
15617 and -mdata-region to the assembler.
15619 2017-08-30 Richard Biener <rguenther@suse.de>
15621 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
15623 (gen_subprogram_die): Add DW_AT_object_pointer only early.
15624 (dwarf2out_early_global_decl): Only generate a DIE for the
15625 abstract origin if it doesn't already exist or is a declaration DIE.
15626 (resolve_addr): Do not add the linkage name twice when
15627 generating a stub DIE for the DW_TAG_GNU_call_site target.
15629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15631 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15632 Use machine_mode rather than int for arg1_mode.
15634 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
15637 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
15638 that the second argument of the built-in functions to unpack
15639 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
15640 switch statement instead a lot of if statements.
15641 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
15642 Allow 64-bit values to be in Altivec registers as well as
15643 traditional floating point registers.
15644 (pack<mode>, FMOVE128_VSX iterator): Likewise.
15646 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
15648 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
15649 MAX_REGS_PER_ADDRESS == 1.
15651 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
15653 * config/i386/i386.opt (flag_fentry): Do not init to -1.
15654 * config/i386/i386.c (ix86_option_override_internal): Simplify
15655 setting of opts->x_flag_entry.
15657 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15658 Jakub Jelinek <jakub@redhat.com>
15659 Richard Biener <rguenther@suse.de>
15661 PR tree-optimization/81503
15662 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
15663 folded constant fits in the target type; reorder tests for clarity.
15665 2017-08-29 Martin Liska <mliska@suse.cz>
15667 * passes.def: Include pass_lower_switch.
15668 * stmt.c (dump_case_nodes): Remove and move to
15669 tree-switch-conversion.
15670 (case_values_threshold): Likewise.
15671 (expand_switch_as_decision_tree_p): Likewise.
15672 (emit_case_decision_tree): Likewise.
15673 (expand_case): Likewise.
15674 (balance_case_nodes): Likewise.
15675 (node_has_low_bound): Likewise.
15676 (node_has_high_bound): Likewise.
15677 (node_is_bounded): Likewise.
15678 (emit_case_nodes): Likewise.
15679 (struct simple_case_node): New struct.
15680 (add_case_node): Remove.
15681 (emit_case_dispatch_table): Use vector instead of case_list.
15682 (reset_out_edges_aux): Remove.
15683 (compute_cases_per_edge): Likewise.
15684 (expand_case): Build list of simple_case_node.
15685 (expand_sjlj_dispatch_table): Use it.
15686 * tree-switch-conversion.c (struct case_node): Moved from
15687 stmt.c and adjusted.
15688 (emit_case_nodes): Likewise.
15689 (node_has_low_bound): Likewise.
15690 (node_has_high_bound): Likewise.
15691 (node_is_bounded): Likewise.
15692 (case_values_threshold): Likewise.
15693 (reset_out_edges_aux): Likewise.
15694 (compute_cases_per_edge): Likewise.
15695 (add_case_node): Likewise.
15696 (dump_case_nodes): Likewise.
15697 (balance_case_nodes): Likewise.
15698 (expand_switch_as_decision_tree_p): Likewise.
15699 (emit_jump): Likewise.
15700 (emit_case_decision_tree): Likewise.
15701 (try_switch_expansion): Likewise.
15702 (do_jump_if_equal): Likewise.
15703 (emit_cmp_and_jump_insns): Likewise.
15704 (fix_phi_operands_for_edge): New function.
15705 (record_phi_operand_mapping): Likewise.
15706 (class pass_lower_switch): New pass.
15707 (pass_lower_switch::execute): New function.
15708 (make_pass_lower_switch): Likewise.
15709 (conditional_probability):
15710 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
15711 * tree-pass.h: Add make_pass_lower_switch.
15713 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15716 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
15719 2017-08-29 Richard Biener <rguenther@suse.de>
15721 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
15722 we do not add a DW_AT_inline attribute twice.
15723 (gen_subprogram_die): Remove code setting DW_AT_inline on
15724 DECL_ABSTRACT_P nodes.
15726 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
15728 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
15729 calls to internal functions.
15730 (gimplify_modify_expr): Likewise.
15731 * tree-call-cdce.c (use_internal_fn): Likewise.
15732 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
15733 (convert_to_divmod): Set the nothrow flag.
15734 * tree-if-conv.c (predicate_mem_writes): Likewise.
15735 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15736 (vectorizable_call): Likewise.
15737 (vectorizable_store): Likewise.
15738 (vectorizable_load): Likewise.
15739 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
15740 (vect_recog_mask_conversion_pattern): Likewise.
15742 2017-08-29 Martin Liska <mliska@suse.cz>
15745 * gcc.c (driver_handle_option): Add new argument.
15746 * opts-common.c (handle_option): Pass
15747 target_option_override_hook.
15748 * opts-global.c (lang_handle_option): Add new option.
15749 (set_default_handlers): Add new argument.
15750 (decode_options): Likewise.
15751 * opts.c (target_handle_option): Likewise.
15752 (common_handle_option): Call target_option_override_hook.
15753 * opts.h (struct cl_option_handler_func): Add hook for
15754 target option override.
15755 (struct cl_option_handlers): Likewise.
15756 (set_default_handlers): Add new argument.
15757 (decode_options): Likewise.
15758 (common_handle_option): Likewise.
15759 (target_handle_option): Likewise.
15760 * toplev.c (toplev::main): Pass targetm.target_option.override
15763 2017-08-29 Richard Biener <rguenther@suse.de>
15764 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
15766 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
15767 life to the active subtree.
15769 2017-08-28 Jeff Law <law@redhat.com>
15771 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
15772 derive_equivalences.
15773 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
15774 Code moved into....
15775 (edge_info::derive_equivalences): New private member function
15777 * tree-ssa-dom.c (class edge_info): Changed from a struct
15778 to a class. Add ctor/dtor, methods and data members.
15779 (edge_info::edge_info): Renamed from allocate_edge_info.
15780 Initialize additional members.
15781 (edge_info::~edge_info): New.
15782 (free_dom_edge_info): Delete the edge info.
15783 (record_edge_info): Use new class & associated member functions.
15784 Tighten forms for testing for edge equivalences.
15785 (record_temporary_equivalences): Iterate over the simple
15786 equivalences rather than assuming there's only one per edge.
15787 (cprop_into_successor_phis): Iterate over the simple
15788 equivalences rather than assuming there's only one per edge.
15789 (optimize_stmt): Use operand_equal_p rather than pointer
15790 equality for mini-DSE code.
15792 2017-08-28 Nathan Sidwell <nathan@acm.org>
15794 * gcc.c (execute): Fold SIGPIPE handling into switch
15795 statement. Adjust internal error message.
15797 2017-08-28 Richard Biener <rguenther@suse.de>
15800 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
15801 Do nothing for removed DIEs.
15803 2017-08-28 Richard Biener <rguenther@suse.de>
15805 PR tree-optimization/81977
15806 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
15809 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
15812 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
15813 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
15814 using targetm.gen_mem_thread_fence.
15816 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
15819 * config/i386/i386.md (*<btsc><mode>): Change operand 2
15820 predicate to register_operand. Reorder operands.
15821 (*btr<mode>): Ditto.
15822 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
15823 (*btr<mode>_mask): Ditto.
15825 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
15827 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
15828 * config/rs6000/xmmintrin.h: New file.
15829 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
15831 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15834 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
15835 parameter and_insn and return it.
15836 (recombine_lvx_pattern): Insert a copy to ensure availability of
15837 the base register of the copied masking operation at the point of
15838 the instruction replacement.
15839 (recombine_stvx_pattern): Likewise.
15841 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
15843 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
15844 undocumented switches.
15845 (-mpower9-dform-vector): Likewise.
15846 (-mpower9-dform): Likewise.
15847 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
15848 comments to delete references to -mpower9-dform* switches.
15849 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
15850 Delete reference to -mpower9-dform* switches, test for
15851 -mpower9-vector instead.
15852 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
15853 (OTHER_P9_VECTOR_MASKS): Likewise.
15854 (POWERPC_MASKS): Likewise.
15855 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
15856 tests against -mpower9-dform* to -mpower9-vector. Delete code
15857 that checked for -mpower9-dform* consistancy with other options.
15858 Add test for -mpower9-misc to enable other power9 switches.
15859 (rs6000_init_hard_regno_mode_ok): Likewise.
15860 (rs6000_option_override_internal): Likewise.
15861 (rs6000_emit_prologue): Likewise.
15862 (rs6000_emit_epilogue): Likewise.
15863 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
15864 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
15865 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
15867 (emit_fusion_p9_store): Likewise.
15868 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
15869 resetting these macros if the assembler does not support ISA 3.0
15871 (TARGET_P9_DFORM_VECTOR): Likewise.
15872 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
15873 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
15875 2017-08-25 Alan Modra <amodra@gmail.com>
15877 PR rtl-optimization/81747
15878 * cse.c (cse_extended_basic_block): Don't attempt to record
15879 equivalences for degenerate conditional branches that branch
15880 to their fall-through.
15882 2017-08-24 Martin Sebor <msebor@redhat.com>
15884 PR middle-end/81908
15885 * gimple-fold.c (size_must_be_zero_p): New function.
15886 (gimple_fold_builtin_memory_op): Call it.
15888 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
15890 * config/rs6000/mm_malloc.h: New file.
15892 2017-08-24 Bin Cheng <bin.cheng@arm.com>
15894 PR tree-optimization/81913
15895 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
15896 analysis when either IVs in condition can wrap.
15898 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
15900 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
15901 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
15903 2017-08-24 Richard Biener <rguenther@suse.de>
15906 * targhooks.c (default_target_can_inline_p): Properly
15907 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
15908 is present and always compare.
15909 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
15910 infer -mfpmath=sse from TARGET_SSE_P.
15911 (ix86_can_inline_p): Properly use target_option_default_node when
15912 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
15914 2017-08-24 Richard Biener <rguenther@suse.de>
15917 * dwarf2out.c (output_die): Handle flag_generate_offload like
15919 (output_comp_unit): Likewise.
15920 (gen_array_type_die): Likewise.
15921 (dwarf2out_early_finish): Likewise.
15922 (note_variable_value_in_expr): Likewise.
15923 (dwarf2out_finish): Likewise. Adjust assert.
15924 * cgraphunit.c (symbol_table::compile): Move setting of
15925 flag_generate_offload earlier ...
15926 (symbol_table::finalize_compilation_unit): ... here, before
15927 early debug finalization.
15929 2017-08-24 Richard Biener <rguenther@suse.de>
15931 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
15932 and ipa-fnsummary.h.
15933 (ix86_can_inline_p): When ix86_fpmath flags do not match
15934 check whether the callee uses FP math at all.
15936 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
15938 PR middle-end/81931
15939 * tree-ssanames.c (get_nonzero_bits): Use element_precision
15940 instead of TYPE_PRECISION.
15942 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
15943 Alan Hayward <alan.hayward@arm.com>
15944 David Sherwood <david.sherwood@arm.com>
15946 * combine.c (make_extraction): Use subreg_offset_from_lsb.
15948 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
15950 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
15952 (ix86_frame::stack_realign_allocate): New field.
15953 (struct machine_frame_state): Modify comments.
15954 (machine_frame_state::sp_realigned_fp_end): New field.
15955 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
15956 layout calculation.
15957 (sp_valid_at): Add assertion to assure no attempt to access invalid
15958 offset of a realigned stack.
15959 (fp_valid_at): Likewise.
15960 (choose_baseaddr): Modify comments.
15961 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
15962 ix86_expand_prologue.
15963 (ix86_expand_prologue): Modify stack realignment and allocation.
15964 (ix86_expand_epilogue): Modify comments.
15965 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
15966 avx2_runtime, avx512f, and avx512f_runtime.
15968 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
15970 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
15971 (mstackrealign): Do not init to -1.
15972 * config/i386/i386.c (ix86_option_override_internal):
15973 Check opts_set, not opts when setting default value of
15974 opts->x_ix86_force_align_arg_pointer.
15976 2017-08-23 Richard Biener <rguenther@suse.de>
15978 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
15979 lang_hooks.decl_printable_name.
15980 * print-rtl-function.c (print_rtx_function): Likewise.
15981 * tree-pretty-print.c (dump_function_header): Likewise.
15983 2017-08-23 Richard Biener <rguenther@suse.de>
15986 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
15987 -g0 at compile-time.
15989 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15991 PR middle-end/19706
15992 * doc/sourcebuild.texi (Other hardware attributes):
15995 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15997 PR middle-end/19706
15998 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
15999 Fix single-use check.
16001 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
16003 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
16005 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16007 * doc/install.texi: Modify to add more details on running selected
16010 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16012 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
16013 is combined with -mabi=ms.
16014 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
16017 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16019 PR tree-optimization/81488
16020 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
16021 and cached_basis fields.
16022 (MAX_SPREAD): New constant.
16023 (alloc_cand_and_find_basis): Initialize new fields.
16024 (clear_visited): New function.
16025 (create_phi_basis_1): Rename from create_phi_basis, set visited
16026 and cached_basis fields.
16027 (create_phi_basis): New wrapper function.
16028 (phi_add_costs_1): Rename from phi_add_costs, add spread
16029 parameter, set visited field, short-circuit when limits reached.
16030 (phi_add_costs): New wrapper function.
16031 (record_phi_increments_1): Rename from record_phi_increments, set
16033 (record_phi_increments): New wrapper function.
16034 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
16035 (phi_incr_cost): New wrapper function.
16036 (all_phi_incrs_profitable_1): Rename from
16037 all_phi_incrs_profitable, set visited field.
16038 (all_phi_incrs_profitable): New wrapper function.
16040 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16041 Alan Hayward <alan.hayward@arm.com>
16042 David Sherwood <david.sherwood@arm.com>
16044 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
16045 that takes the outer and inner modes.
16046 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
16047 comparison as the canonical test for a paradoxical subreg.
16048 * combine.c (simplify_set): Use paradoxical_subreg_p.
16049 (make_extraction): Likewise.
16050 (force_to_mode): Likewise.
16051 (rtx_equal_for_field_assignment_p): Likewise.
16052 (gen_lowpart_for_combine): Likewise.
16053 (simplify_comparison): Likewise.
16054 * cse.c (equiv_constant): Likewise.
16055 * expmed.c (store_bit_field_1): Likewise.
16056 * final.c (alter_subreg): Likewise.
16057 * fwprop.c (propagate_rtx): Likewise.
16058 (forward_propagate_subreg): Likewise.
16059 * ira-conflicts.c (ira_build_conflicts): Likewise.
16060 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
16061 * lra-constraints.c (curr_insn_transform): Likewise.
16062 (split_reg): Likewise.
16063 * lra-eliminations.c (move_plus_up): Likewise.
16064 (lra_eliminate_regs_1): Likewise.
16065 * recog.c (general_operand): Likewise.
16066 * ree.c (combine_reaching_defs): Likewise.
16067 * reload.c (push_reload): Likewise.
16068 (find_reloads): Likewise.
16069 * reload1.c (elimination_effects): Likewise.
16070 (compute_reload_subreg_offset): Likewise.
16071 (choose_reload_regs): Likewise.
16072 * rtlanal.c (subreg_lsb_1): Likewise.
16073 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16074 (simplify_subreg): Likewise.
16075 * var-tracking.c (track_loc_p): Likewise.
16076 * emit-rtl.c (byte_lowpart_offset): Likewise.
16077 (paradoxical_subreg_p): Delete out-of-line definition.
16079 2017-08-22 Jeff Law <law@redhat.com>
16081 PR tree-optimization/81741
16082 PR tree-optimization/71947
16083 * tree-ssa-dom.c: Include tree-inline.h.
16084 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
16085 equivalences if one is more expensive to compute than the other.
16086 * tree-ssa-scopedtables.h (class const_or_copies): Make
16087 record_const_or_copy_raw method private.
16088 (class avail_exprs_stack): New method simplify_binary_operation.
16089 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
16090 avail_exprs_stack::simplify_binary_operation as needed.
16091 (avail_exprs_stack::simplify_binary_operation): New function.
16093 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
16095 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
16096 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
16097 (DOT_SYMBOLS): Likewise.
16098 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
16099 (RELOCATABLE_NEEDS_FIXUP): Likewise.
16100 (RS6000_ABI_NAME): Likewise.
16101 (TARGET_CMODEL): Likewise.
16102 (TOC_SECTION_ASM_OP): Likewise.
16103 (SET_CMODEL): New macro.
16104 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
16106 2017-08-22 Richard Biener <rguenther@suse.de>
16108 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
16109 to allow for free-lang-data replacements similar to verify_type_variant.
16111 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16112 Alan Hayward <alan.hayward@arm.com>
16113 David Sherwood <david.sherwood@arm.com>
16115 * config/aarch64/aarch64.md (casesi): Use DImode rather than
16116 VOIDmode for the LABEL_REF.
16118 2017-08-22 Richard Biener <rguenther@suse.de>
16120 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
16122 2017-08-22 Richard Biener <rguenther@suse.de>
16124 * common.opt (feliminate-dwarf2-dups): Ignore.
16125 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
16126 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
16127 same_die_p_wrap, compute_section_prefix,
16128 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
16129 (comdat_symbol_id, comdat_symbol_number): Likewise.
16130 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
16132 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
16133 (output_die): Mark unreachable path unreachable.
16134 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
16135 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
16136 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
16137 (dwarf2out_early_finish): Likewise.
16139 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
16141 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
16143 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
16146 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
16147 not VAR_P. Filter attribute warnings with OPT_Wattributes.
16148 (avr_attribute_table) <io, io_low, address>: Initialize
16149 .decl_required with true.
16151 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
16153 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
16154 undocumented debugging options.
16155 (-mvsx-scalar-double): Likewise.
16156 (-mallow-df-permute): Likewise.
16157 (-mvectorize-builtins): Likewise.
16158 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16159 (rs6000_builtin_vectorized_function): Likewise.
16160 (rs6000_builtin_md_vectorized_function): Likewise.
16161 (rs6000_opt_vars): Likewise.
16163 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
16166 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
16167 (*btrq_imm): Rename from *btrq.
16168 (*btcq_imm): Rename from *btcq.
16169 (btsc): New code attribute.
16170 (*<btsc><mode>): New insn pattern.
16171 (*btr<mode>): Ditto.
16172 (*<btsc><mode>_mask): New insn_and_split pattern.
16173 (*btr<mode>_mask): Ditto.
16175 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16176 Alan Hayward <alan.hayward@arm.com>
16177 David Sherwood <david.sherwood@arm.com>
16179 * function.c (pad_below): Simplify padding calculation.
16181 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16182 Alan Hayward <alan.hayward@arm.com>
16183 David Sherwood <david.sherwood@arm.com>
16185 * target.def (function_prologue): Remove frame size argument.
16186 (function_epilogue): Likewise.
16187 * doc/tm.texi: Regenerate.
16188 * final.c (final_start_function): Update call to function_prologue.
16189 (final_end_function): Update call to function_epilogue.
16190 (default_function_pro_epilogue): Remove frame size argument.
16191 * output.h (default_function_pro_epilogue): Likewise.
16192 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
16193 (arm_output_function_prologue): Likewise.
16194 * config/frv/frv.c (frv_function_prologue): Likewise.
16195 (frv_function_epilogue): Likewise.
16196 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
16197 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
16198 (ia64_output_function_epilogue): Likewise.
16199 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
16200 (m32r_output_function_epilogue): Likewise.
16201 * config/microblaze/microblaze.c (microblaze_function_prologue)
16202 (microblaze_function_epilogue): Likewise.
16203 * config/mips/mips.c (mips_output_function_prologue): Likewise.
16204 (mips_output_function_epilogue): Likewise.
16205 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
16206 (mmix_target_asm_function_epilogue): Likewise.
16207 * config/msp430/msp430.c (msp430_start_function): Likewise.
16208 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
16209 (nds32_asm_function_epilogue): Likewise.
16210 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
16211 * config/pa/pa.c (pa_output_function_prologue): Likewise.
16212 (pa_output_function_epilogue): Likewise.
16213 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
16214 (rs6000_output_function_epilogue): Likewise.
16215 * config/rl78/rl78.c (rl78_start_function): Likewise.
16216 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
16217 (rs6000_output_function_epilogue): Likewise.
16218 * config/rx/rx.c (rx_output_function_prologue): Likewise.
16219 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
16220 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
16221 (sparc_asm_function_epilogue): Likewise.
16223 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16225 * tree.h (type_has_mode_precision_p): New function.
16226 * convert.c (convert_to_integer_1): Use it.
16227 * expr.c (expand_expr_real_2): Likewise.
16228 (expand_expr_real_1): Likewise.
16229 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
16230 * match.pd: Likewise.
16231 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
16232 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
16233 * tree-tailcall.c (process_assignment): Likewise.
16234 * tree-vect-loop.c (vectorizable_reduction): Likewise.
16235 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
16236 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
16237 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16238 (vectorizable_assignment): Likewise.
16239 (vectorizable_shift): Likewise.
16240 (vectorizable_operation): Likewise.
16241 * tree-vrp.c (register_edge_assert_for_2): Likewise.
16243 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
16245 * match.pd: Add pow (C, x) simplification.
16247 2017-08-21 Richard Biener <rguenther@suse.de>
16249 PR tree-optimization/81900
16250 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
16251 for blocks with abnormal predecessors.
16252 (compute_antic): Do not set visited flag prematurely.
16254 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
16257 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
16259 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16261 * stor-layout.h (vector_type_mode): Move to...
16262 * tree.h (vector_type_mode): ...here.
16263 * stor-layout.c (vector_type_mode): Move to...
16264 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
16266 2017-08-21 Richard Biener <rguenther@suse.de>
16268 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
16269 register_external_die hooks.
16270 (debug_false_tree_charstarstar_uhwistar): Declare.
16271 (debug_nothing_tree_charstar_uhwi): Likewise.
16272 * debug.c (do_nothing_debug_hooks): Adjust.
16273 (debug_false_tree_charstarstar_uhwistar): New do nothing.
16274 (debug_nothing_tree_charstar_uhwi): Likewise.
16275 * dbxout.c (dbx_debug_hooks): Adjust.
16276 (xcoff_debug_hooks): Likewise.
16277 * sdbout.c (sdb_debug_hooks): Likewise.
16278 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16279 * dwarf2out.c (macinfo_label_base): New global.
16280 (dwarf2out_register_external_die): New function for the
16281 register_external_die hook.
16282 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
16283 (dwarf2_debug_hooks): Use them.
16284 (dwarf2_lineno_debug_hooks): Adjust.
16285 (struct die_struct): Add with_offset flag.
16286 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
16287 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
16288 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
16289 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
16290 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
16291 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
16292 defining section names for the early LTO debug variants.
16293 (reset_indirect_string): New helper.
16294 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
16295 (print_dw_val): Add support for offsetted symbol references.
16296 (get_ultimate_context): Split out from is_cxx.
16297 (is_cxx): Use get_ultimate_context.
16298 (is_fortran): Add decl overload.
16299 (compute_comp_unit_symbol): Split out worker from
16300 compute_section_prefix.
16301 (compute_section_prefix): Call compute_comp_unit_symbol and
16302 set comdat_type_p here.
16303 (output_die): Skip DIE symbol output for the LTO added one.
16304 Handle DIE symbol references with offset.
16305 (output_comp_unit): Guard section name mangling properly.
16306 For LTO debug sections emit a symbol at the section beginning
16307 which we use to refer to its DIEs.
16308 (add_abstract_origin_attribute): For DIEs registered via
16309 dwarf2out_register_external_die directly refer to the early
16310 DIE rather than indirectly through the shadow one we created.
16311 Remove obsolete call to dwarf2out_abstract_function for
16312 non-function/block origins.
16313 (gen_array_type_die): When generating early LTO debug do
16314 not emit DW_AT_string_length.
16315 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
16316 late when in LTO. As suggested place a gcc_unreachable for
16317 the DECL_ABSTRACT_P case.
16318 (gen_subprogram_die): Avoid another specification DIE
16319 for early built declarations/definitions for the late LTO case.
16320 (gen_variable_die): Add type references for late duplicated VLA dies
16322 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
16323 we have the abstract instance already.
16324 (process_scope_var): Adjust decl DIE contexts in LTO which
16325 first puts them in limbo.
16326 (gen_decl_die): Do not generate type DIEs late apart from
16327 types for VLAs or for decls we do not yet have a DIE. Do not
16328 call dwarf2out_abstract_function late.
16329 (dwarf2out_early_global_decl): Make sure to create DIEs
16330 for abstract instances of a decl first.
16331 (dwarf2out_late_global_decl): Adjust comment.
16332 (output_macinfo_op): With multiple macro sections use
16333 macinfo_label_base to distinguish labels.
16334 (output_macinfo): Likewise. Update macinfo_label_base.
16335 Pass in the line info label.
16336 (note_variable_value_in_expr): When generating LTO resolve
16337 all variable values here by generating DIEs as needed.
16338 (init_sections_and_labels): Add early LTO debug flag parameter
16339 and generate different sections and names if set. Add generation
16340 counter for the labels so we can have multiple of them.
16341 (reset_dies): Helper to allow DIEs to be output multiple times.
16342 (dwarf2out_finish): When outputting DIEs to the fat part of an
16343 LTO object first reset DIEs.
16344 (dwarf2out_early_finish): Output early DIEs when generating LTO.
16345 (modified_type_die): Check for decl_ultimate_origin being self
16347 (gen_type_die_with_usage): Likewise.
16348 (gen_typedef_die): Allow decl_ultimate_origin being self.
16349 (set_decl_abstract_flags): Remove.
16350 (set_block_abstract_flags): Likewise.
16351 (dwarf2out_abstract_function): Treat the early generated DIEs
16352 as the abstract copy and only add DW_AT_inline and
16353 DW_AT_artificial here and call set_decl_origin_self.
16354 If the DIE has an abstract origin don't do anything.
16355 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
16356 if we have none yet (Go fails to build one, PR78628).
16357 (variably_modified_type_p): Prevent endless recursion for Ada
16358 cyclic pointer types.
16359 * lto-streamer-in.c: Include debug.h.
16360 (dref_queue): New global.
16361 (lto_read_tree_1): Stream in DIE references.
16362 (lto_input_tree): Register DIE references.
16363 (input_function): Stream DECL_DEBUG_ARGS.
16364 * lto-streamer-out.c: Include debug.h.
16365 (lto_write_tree_1): Output DIE references.
16366 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
16367 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
16368 (output_function): Stream DECL_DEBUG_ARGS.
16369 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
16370 Stream DECL_ABSTRACT_ORIGIN.
16371 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
16372 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
16373 DECL_CONTEXT for file-scope decls.
16374 * lto-streamer.h (struct dref_entry): Declare.
16375 (dref_queue): Likewise.
16376 * cfgexpand.c (pass_expand::execute): Do not call the
16377 outlining_inline_function hook here.
16378 * lto-wrapper.c (debug_obj): New global.
16379 (tool_cleanup): Unlink it if required.
16380 (debug_objcopy): New function.
16381 (run_gcc): Handle early debug sections in the IL files by
16382 extracting them to separate files, partially linkin them and
16383 feeding the result back as result to the linker.
16384 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
16385 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
16386 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
16387 sections into a separate segment.
16388 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
16390 (darwin_asm_dwarf_section): Likewise.
16391 (darwin_asm_output_dwarf_offset): Likewise.
16392 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
16394 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16395 Alan Hayward <alan.hayward@arm.com>
16396 David Sherwood <david.sherwood@arm.com>
16398 * read-md.h (md_reader::record_potential_iterator_use): Replace
16399 pointer argument with an rtx and an index.
16400 * read-rtl.c (iterator_group::apply_iterator): Likewise.
16401 (apply_mode_iterator): Likewise.
16402 (apply_code_iterator): Likewise.
16403 (apply_int_iterator): Likewise.
16404 (apply_subst_iterator): Likewise.
16405 (record_iterator_use): Likewise.
16406 (record_attribute_use): Likewise.
16407 (md_reader::record_potential_iterator_use): Likewise. Update calls
16408 to record_iterator_use and apply_iterator.
16409 (iterator_use): Replace ptr with x and index.
16410 (attribute_use): Likewise.
16411 (apply_attribute_uses): Update calls to apply_iterator.
16412 (apply_iterators): Likewise. Update initialization of iterator_use.
16413 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
16414 and record_potential_iterator_use.
16415 (rtx_reader::read_rtx_operand): Likewise.
16417 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16418 Alan Hayward <alan.hayward@arm.com>
16419 David Sherwood <david.sherwood@arm.com>
16421 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
16424 2017-08-21 Richard Biener <rguenther@suse.de>
16426 PR middle-end/81884
16427 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
16428 at struct end conservatively when comparing common bases.
16430 2017-08-21 Richard Biener <rguenther@suse.de>
16432 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
16433 (mem_ref_in_stmt): Remove.
16434 (determine_max_movement): Use ref index to get at the reference.
16435 (invariantness_dom_walker::before_dom_children): Deal with
16436 lim data already initialized.
16437 (gather_mem_refs_stmt): Initialize lim data and record ref index.
16439 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16441 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
16442 (TARGET_ISA_ROUND): Ditto.
16443 (TARGET_ROUND): Ditto.
16444 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
16445 * config/i386/i386.md: Ditto.
16446 * config/i386/sse.md: Ditto.
16447 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
16448 with OPTION_MASK_ISA_SSE4_1.
16450 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16453 * doc/extend.texi (x86 Built-in Functions): Correct the name of
16454 __builtin_ia32_lzcnt_u16.
16456 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
16459 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
16460 (rs6000_set_current_function): Rewrite function to use it.
16462 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
16465 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
16466 and TYPE_WARN_IF_NOT_ALIGN.
16467 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
16468 (handle_warn_if_not_align): New.
16469 (place_union_field): Call handle_warn_if_not_align.
16470 (place_field): Call handle_warn_if_not_align.
16471 Copy TYPE_WARN_IF_NOT_ALIGN.
16472 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
16473 (layout_type): Likewise.
16474 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
16476 (tree_decl_common): Add warn_if_not_align.
16477 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
16478 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
16479 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
16480 (DECL_WARN_IF_NOT_ALIGN): Likewise.
16481 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
16482 * doc/extend.texi: Document warn_if_not_aligned attribute.
16483 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
16485 2017-08-17 Martin Liska <mliska@suse.cz>
16488 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
16489 (get_data_dependence): Use it as pointer type.
16490 (distribute_loop): Likewise.
16492 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16494 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
16495 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
16496 (p8_vmrgow_<mode>_direct): New define_insn.
16497 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
16498 handle endianness for vmrgew and vmrgow permute patterns.
16500 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16502 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
16503 * config/rs6000/rs6000-cpus.def: Remove comment.
16504 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
16505 (POWERPC_MASKS): Likewise.
16506 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
16507 use of TARGET_VSX_TIMODE.
16508 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
16509 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
16510 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
16511 (rs6000_option_override_internal): Remove dead code.
16512 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
16513 (rs6000_legitimize_reload_address): Likewise.
16514 (rs6000_legitimate_address_p): Likewise.
16515 (rs6000_opt_masks): Delete "vsx-timode".
16516 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
16517 from function comment.
16518 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
16519 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
16520 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
16522 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
16523 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
16524 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
16525 (VSX_TI): Likewise.
16527 (define_peephole2): Likewise.
16529 2017-08-17 Martin Sebor <msebor@redhat.com>
16532 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
16533 past the end of an array.
16534 (test_pp_format): Add test cases.
16536 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
16538 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
16539 missing ECF_NOTHROW flags.
16541 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16544 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
16545 operands residing in integer registers.
16546 (*vsx_le_perm_load_<mode>): Likewise.
16547 (*vsx_le_perm_store_<mode>): Likewise.
16548 (define_peephole2): Add peepholes to optimize the above.
16550 2017-08-17 Marek Polacek <polacek@redhat.com>
16552 PR middle-end/81814
16553 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
16554 to mimic what shorten_compare did. Change the return type to bool.
16555 (fold_cond_expr_with_comparison): Update call to
16556 operand_equal_for_comparison_p.
16557 (fold_ternary_loc): Likewise.
16559 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
16561 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
16563 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
16565 2017-08-17 Richard Biener <rguenther@suse.de>
16567 * tree-ssa-structalias.c (solve_graph): When propagating
16568 to successors update the graphs succ edges and avoid duplicate work.
16570 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
16573 * config/i386/i386.c (ix86_option_override_internal): Save target
16574 specific options after ix86_stack_protector_guard_reg was changed.
16576 2017-08-17 Richard Biener <rguenther@suse.de>
16578 PR tree-optimization/81827
16579 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
16580 (new_var_info): Initialize it conservatively.
16581 (get_call_vi): Mark register vars.
16582 (new_scalar_tmp_constraint_exp): Likewise.
16583 (handle_rhs_call): Likewise.
16584 (handle_const_call): Likewise.
16585 (create_function_info_for): Likewise.
16586 (solve_constraints): Sort varinfos to separate register from
16587 non-register vars to pack points-to solution bitmaps during
16590 2017-08-17 Marek Polacek <polacek@redhat.com>
16592 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
16594 2017-08-17 Richard Biener <rguenther@suse.de>
16596 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
16597 to true when overflow is undefined and we saturated the result.
16599 2017-08-17 Alan Modra <amodra@gmail.com>
16602 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
16603 Don't use store multiple if only one reg needs saving.
16604 (interesting_frame_related_regno): New function.
16605 (rs6000_frame_related): Don't emit frame info for regs that
16607 (rs6000_emit_epilogue): Likewise.
16609 2017-08-16 Nathan Sidwell <nathan@acm.org>
16611 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
16612 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
16613 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
16614 * tree.c (free_lang_data_in_type): Use else-if chain. Always
16615 clear TYPE_LANG_1. Remove obsolete member-function stripping.
16616 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
16617 (verify_type): Adjust for TYPE_BINFO move.
16618 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
16619 process TYPE_BINFO directly.
16620 (hash_tree): Likewise.
16621 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16623 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16626 2017-08-16 David Malcolm <dmalcolm@redhat.com>
16628 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
16630 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
16633 * config/i386/i386.md (*anddi_1_btr): Change predicates of
16634 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
16635 Add ix86_binary_operator_ok to insn constraint.
16636 (*iordi_1_bts): Ditto.
16637 (*xordi_1_btc): Ditto.
16638 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
16639 Update corresponding peephole2 pattern.
16643 2017-08-16 Bin Cheng <bin.cheng@arm.com>
16645 PR tree-optimization/81832
16646 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
16647 copy loop header which has IFN_LOOP_DIST_ALIAS call.
16649 2017-08-16 Marek Polacek <polacek@redhat.com>
16652 * fold-const.c (fold_indirect_ref_1): Restore original behavior
16653 regarding size_zero_node.
16655 2017-08-16 Martin Liska <mliska@suse.cz>
16658 * config.gcc: Respect previously set extra_objs in case
16661 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
16663 PR tree-optimization/81835
16664 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
16665 the phi SSA_NAME. Check that the condition in a COND_EXPR does
16666 not depend on the phi.
16668 2017-08-16 Alan Modra <amodra@gmail.com>
16670 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
16673 2017-08-16 Alan Modra <amodra@gmail.com>
16675 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
16676 (save_reg_p): ..into this. Update all callers.
16677 (first_reg_to_save): Simplify.
16679 2017-08-16 Alan Modra <amodra@gmail.com>
16681 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
16684 2017-08-15 Joseph Myers <joseph@codesourcery.com>
16688 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
16689 constant count if that count is less than 32.
16691 2017-08-15 Nathan Sidwell <nathan@acm.org>
16693 * gcc.c (execute): Emit friendlier message if inferior is killed
16694 by an external cause.
16696 2017-08-15 Richard Biener <rguenther@suse.de>
16698 PR tree-optimization/81790
16699 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
16700 CONSTRUCTORs from simplifying and VN.
16702 2017-08-14 Martin Sebor <msebor@redhat.com>
16704 * builtin-attrs.def: Add comments.
16706 2017-08-14 Martin Sebor <msebor@redhat.com>
16709 * doc/extend.texi (attribute nonstring): Document new attribute.
16711 2017-08-14 Martin Sebor <msebor@redhat.com>
16714 * tree-diagnostic.c (default_tree_printer): Handle %G.
16715 * gimple-pretty-print.h (percent_G_format): Declare new function.
16716 * gimple-pretty-print.c (percent_G_format): Define.
16717 * tree-pretty-print.c (percent_K_format): Add argument.
16719 2017-08-14 Martin Sebor <msebor@redhat.com>
16721 PR translation/79998
16722 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
16723 Remove a stray space.
16725 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
16728 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
16729 (*iordi_1_bts): Ditto.
16730 (*xordi_1_btc): Ditto.
16732 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16735 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
16736 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16738 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
16739 quoted strings, and make more translator-friendly.
16740 (darwin_rs6000_override_options): Likewise.
16741 (rs6000_option_override_internal): Likewise.
16742 (rs6000_return_in_memory): Fix overlong line.
16743 (init_cmulative_args): Use quoted strings, and make more
16744 translator-friendly.
16745 (rs6000_pass_by_reference): Fix overlong line.
16746 (def_builtin): Use quoted strings.
16747 (altivec_expand_predicate_builtin): Use quoted strings, and make
16748 more translator-friendly.
16749 (htm_expand_builtin): Use quoted strings.
16750 (cpu_expand_builtin): Use quoted strings, and make more
16751 translator-friendly.
16752 (altivec_expand_builtin): Likewise.
16753 (paired_expand_predicate_builtin): Likewise.
16754 (rs6000_invalid_builtin): Likewise.
16755 (builtin_function_type): Use quoted strings.
16756 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
16757 more translator-friendly.
16758 (rs6000_trampoline_init): Likewise.
16759 (rs6000_handle_altivec_attribute): Likewise.
16760 (rs6000_inner_target_options): Use quoted strings.
16761 (rs6000_disable_incompatible_switches): Likewise.
16762 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
16763 strings, and make more translator-friendly.
16764 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
16766 2017-08-14 Bin Cheng <bin.cheng@arm.com>
16768 PR tree-optimization/81799
16769 * tree-loop-distribution.c (version_loop_by_alias_check): Force
16770 cond_expr to simple gimple operand.
16772 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
16774 PR middle-end/46932
16775 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
16777 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
16781 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
16782 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
16783 TARGET_GASISR_PROLOGUES.
16784 * config/avr/avr.c (avr_option_override): Same.
16785 (avr_pass_pre_proep::execute): Same.
16787 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
16790 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
16791 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
16793 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
16795 * config/i386/i386.md (*load_tp_<mode>): Redefine as
16796 define_insn_and_split. Split to a memory load from 0 in
16797 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
16798 using PTR mode iterator.
16799 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
16800 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
16801 (*add_tp_<mode>): Redefine as define_insn_and_split.
16802 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
16803 address space. Merge with *add_tp_x32 using PTR mode iterator.
16804 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
16805 Split to an add with a memory load from 0 in
16806 DEFAULT_TLS_SEG_REG address space.
16808 2017-08-12 Andrew Pinski <apinski@cavium.com>
16810 * config/aarch64/aarch64-option-extensions.def (rdma):
16811 Fix feature string to what Linux prints out in /proc/cpuinfo.
16813 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
16816 * dwarf2out.c (modified_type_die): For C typedef types that have
16817 an ultimate origin, process the ultimate origin instead of the
16819 (gen_typedef_die): Assert that input DECLs have no ultimate
16821 (gen_type_die_with_usage): For typedef variants that have an
16822 ultimate origin, just call gen_decl_die on the original DECL.
16823 (process_scope_var): Avoid creating DIEs for local typedefs and
16824 concrete static variables.
16826 2017-08-12 Alan Modra <amodra@gmail.com>
16830 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
16831 match gnu-user.h startfile.
16832 (ENDFILE_LINUX_SPEC): Similarly.
16834 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
16837 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
16839 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
16841 2017-08-11 Tamar Christina <tamar.christina@arm.com>
16842 * config/aarch64/aarch64.md (mov<mode>): Change.
16843 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16844 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
16845 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
16847 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
16849 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
16850 for storage order barriers.
16852 2017-08-11 Martin Liska <mliska@suse.cz>
16855 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
16856 variables of void type.
16858 2017-08-11 Martin Liska <mliska@suse.cz>
16860 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
16861 TARGET_SUPPORTS_ALIASES.
16862 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
16863 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
16864 (optimize_weakref): Likewise.
16865 * symtab.c (symtab_node::noninterposable_alias): Likewise.
16866 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
16867 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
16869 2017-08-11 Martin Liska <mliska@suse.cz>
16872 * config/i386/i386.c (make_resolver_func): Do complete
16873 refactoring of the function.
16875 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16878 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
16879 * config/i386/i386.c (ix86_stack_protect_guard): Use
16880 ix86_stack_protect_guard_symbol_str to generate varible declaration.
16881 * doc/invoke.texi (x86 Options): Document
16882 -mstack-protector-guard-symbol= option.
16884 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16886 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
16887 * config/i386/i386.c (ix86_split_stack_guard): New function.
16888 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
16889 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
16890 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
16891 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
16892 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
16893 (split_stack_space_check): Call ix86_split_stack_guard.
16895 2017-08-10 Martin Sebor <msebor@redhat.com>
16897 * print-tree.c (print_node): Print location using the established
16899 Replace spaces with colons.
16900 (debug_raw, debug): Ditto.
16902 2017-08-10 Martin Sebor <msebor@redhat.com>
16905 * pretty-print.c (pp_format): Correct the handling of %s precision.
16907 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
16910 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
16912 (ix86_finalize_stack_frame_flags): This. Also clear
16913 frame_pointer_needed if -fno-omit-frame-pointer is used without
16915 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
16916 with ix86_finalize_stack_frame_flags.
16917 (ix86_expand_epilogue): Likewise.
16918 (ix86_expand_split_stack_prologue): Likewise.
16919 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
16921 2017-08-10 Martin Liska <mliska@suse.cz>
16924 * c-attribs.c (handle_target_attribute):
16925 Report warning for an empty string argument of target attribute.
16927 2017-08-09 Jakub Jelinek <jakub@redhat.com>
16930 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
16932 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
16933 or DECL_NONLOCAL labels.
16934 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
16935 or DECL_NONLOCAL labels here.
16937 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
16939 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
16940 to indicate when early gimple folding has been disabled.
16941 (rs6000_gimple_fold_builtin): Add debug content.
16942 (rs6000_invalid_builtin): Fix whitespace.
16943 (rs6000_expand_builtin): Fix whitespace.
16944 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
16946 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
16949 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
16950 SAVE_MULTIPLE if not all the registers that saves, should be saved.
16952 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
16954 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
16955 (qdf24xx): Likewise.
16956 * config/aarch64/aarch64.md: Include falkor.md.
16957 * config/aarch64/falkor.md: New.
16959 2017-08-09 Marek Polacek <polacek@redhat.com>
16962 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
16963 * diagnostic.c (emit_diagnostic): Add a comment.
16964 (emit_diagnostic_valist): New function.
16966 2017-08-09 Marek Polacek <polacek@redhat.com>
16969 * input.c (make_location): New overload.
16970 * input.h (make_location): Declare.
16972 2017-08-08 Alan Modra <amodra@gmail.com>
16973 H.J. Lu <hongjiu.lu@intel.com>
16976 * gcc.c (NO_PIE_SPEC): Delete.
16977 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
16979 (LINK_PIE_SPEC): ..to here.
16980 (LINK_COMMAND_SPEC): Support -no-pie.
16981 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
16982 chain of crtbegin*.o selection, update for PIE_SPEC changes and
16984 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
16985 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
16986 (ENDFILE_CRTEND_SPEC): Similarly.
16988 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
16991 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
16992 (mstack-protector-guard-offset=): Ditto.
16993 * config/i386/i386.c (ix86_option_override): Handle
16994 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
16996 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
16997 ix86_stack_protect_guard_offset variables.
16998 (TARGET_STACK_PROTECT_GUARD): Always define.
16999 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
17000 and -mstack-protector-guard-offset= options.
17002 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17004 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
17005 boundary case for the last candidate.
17007 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17009 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
17010 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
17012 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17014 PR middle-end/19706
17015 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
17016 * config/aarch64/aarch64-builtins.c
17017 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
17018 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
17019 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
17021 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17022 Andrew Pinski <pinskia@gmail.com>
17024 PR middle-end/19706
17025 * internal-fn.def (XORSIGN): New.
17026 * optabs.def (xorsign_optab): New.
17027 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
17028 (convert_expand_mult_copysign): New.
17029 (pass_optimize_widening_mul::execute): Call
17030 convert_expand_mult_copysign.
17032 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17034 PR tree-optimization/81354
17035 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
17036 Insert on edges rather than explicitly creating landing pads.
17037 (analyze_candidates_and_replace): Commit edge inserts.
17039 2017-08-08 Richard Biener <rguenther@suse.de>
17041 PR middle-end/81719
17042 * tree-ssa-loop-niter.c: Include tree-dfa.h.
17043 (expand_simple_operations): Also look through ADDR_EXPRs with
17044 MEM_REF bases treating them as POINTER_PLUS_EXPR.
17046 2017-08-08 Richard Biener <rguenther@suse.de>
17048 PR tree-optimization/81723
17049 * tree-vect-slp.c (struct bst_traits): New hash traits.
17050 (bst_fail): New global.
17051 (vect_build_slp_tree_2): New worker, split out from ...
17052 (vect_build_slp_tree): ... this now wrapping it with using
17053 bst_fail set to cache SLP tree build fails. Properly handle
17055 (vect_analyze_slp_instance): Allocate and free bst_fail.
17057 2017-08-08 Martin Liska <mliska@suse.cz>
17060 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
17061 LABEL_DECLs that can be from a different function.
17063 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17065 PR tree-optimization/81744
17066 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
17067 loop's number of iterations.
17069 2017-08-08 Martin Liska <mliska@suse.cz>
17071 * asan.c: Include header files.
17072 * attribs.c (build_decl_attribute_variant): New function moved
17074 (build_type_attribute_qual_variant): Likewise.
17075 (cmp_attrib_identifiers): Likewise.
17076 (simple_cst_list_equal): Likewise.
17077 (omp_declare_simd_clauses_equal): Likewise.
17078 (attribute_value_equal): Likewise.
17079 (comp_type_attributes): Likewise.
17080 (build_type_attribute_variant): Likewise.
17081 (lookup_ident_attribute): Likewise.
17082 (remove_attribute): Likewise.
17083 (merge_attributes): Likewise.
17084 (merge_type_attributes): Likewise.
17085 (merge_decl_attributes): Likewise.
17086 (merge_dllimport_decl_attributes): Likewise.
17087 (handle_dll_attribute): Likewise.
17088 (attribute_list_equal): Likewise.
17089 (attribute_list_contained): Likewise.
17090 * attribs.h (lookup_attribute): New function moved from tree.[ch].
17091 (lookup_attribute_by_prefix): Likewise.
17092 * bb-reorder.c: Include header files.
17093 * builtins.c: Likewise.
17094 * calls.c: Likewise.
17095 * cfgexpand.c: Likewise.
17096 * cgraph.c: Likewise.
17097 * cgraphunit.c: Likewise.
17098 * convert.c: Likewise.
17099 * dwarf2out.c: Likewise.
17100 * final.c: Likewise.
17101 * fold-const.c: Likewise.
17102 * function.c: Likewise.
17103 * gimple-expr.c: Likewise.
17104 * gimple-fold.c: Likewise.
17105 * gimple-pretty-print.c: Likewise.
17106 * gimple.c: Likewise.
17107 * gimplify.c: Likewise.
17108 * hsa-common.c: Likewise.
17109 * hsa-gen.c: Likewise.
17110 * internal-fn.c: Likewise.
17111 * ipa-chkp.c: Likewise.
17112 * ipa-cp.c: Likewise.
17113 * ipa-devirt.c: Likewise.
17114 * ipa-fnsummary.c: Likewise.
17115 * ipa-inline.c: Likewise.
17116 * ipa-visibility.c: Likewise.
17118 * lto-cgraph.c: Likewise.
17119 * omp-expand.c: Likewise.
17120 * omp-general.c: Likewise.
17121 * omp-low.c: Likewise.
17122 * omp-offload.c: Likewise.
17123 * omp-simd-clone.c: Likewise.
17124 * opts-global.c: Likewise.
17125 * passes.c: Likewise.
17126 * predict.c: Likewise.
17127 * sancov.c: Likewise.
17128 * sanopt.c: Likewise.
17129 * symtab.c: Likewise.
17130 * toplev.c: Likewise.
17131 * trans-mem.c: Likewise.
17132 * tree-chkp.c: Likewise.
17133 * tree-eh.c: Likewise.
17134 * tree-into-ssa.c: Likewise.
17135 * tree-object-size.c: Likewise.
17136 * tree-parloops.c: Likewise.
17137 * tree-profile.c: Likewise.
17138 * tree-ssa-ccp.c: Likewise.
17139 * tree-ssa-live.c: Likewise.
17140 * tree-ssa-loop.c: Likewise.
17141 * tree-ssa-sccvn.c: Likewise.
17142 * tree-ssa-structalias.c: Likewise.
17143 * tree-ssa.c: Likewise.
17144 * tree-streamer-in.c: Likewise.
17145 * tree-vectorizer.c: Likewise.
17146 * tree-vrp.c: Likewise.
17147 * tsan.c: Likewise.
17148 * ubsan.c: Likewise.
17149 * varasm.c: Likewise.
17150 * varpool.c: Likewise.
17151 * tree.c: Remove functions moved to attribs.[ch].
17152 * tree.h: Likewise.
17153 * config/aarch64/aarch64.c: Add attrs.h header file.
17154 * config/alpha/alpha.c: Likewise.
17155 * config/arc/arc.c: Likewise.
17156 * config/arm/arm.c: Likewise.
17157 * config/avr/avr.c: Likewise.
17158 * config/bfin/bfin.c: Likewise.
17159 * config/c6x/c6x.c: Likewise.
17160 * config/cr16/cr16.c: Likewise.
17161 * config/cris/cris.c: Likewise.
17162 * config/darwin.c: Likewise.
17163 * config/epiphany/epiphany.c: Likewise.
17164 * config/fr30/fr30.c: Likewise.
17165 * config/frv/frv.c: Likewise.
17166 * config/ft32/ft32.c: Likewise.
17167 * config/h8300/h8300.c: Likewise.
17168 * config/i386/winnt.c: Likewise.
17169 * config/ia64/ia64.c: Likewise.
17170 * config/iq2000/iq2000.c: Likewise.
17171 * config/lm32/lm32.c: Likewise.
17172 * config/m32c/m32c.c: Likewise.
17173 * config/m32r/m32r.c: Likewise.
17174 * config/m68k/m68k.c: Likewise.
17175 * config/mcore/mcore.c: Likewise.
17176 * config/microblaze/microblaze.c: Likewise.
17177 * config/mips/mips.c: Likewise.
17178 * config/mmix/mmix.c: Likewise.
17179 * config/mn10300/mn10300.c: Likewise.
17180 * config/moxie/moxie.c: Likewise.
17181 * config/msp430/msp430.c: Likewise.
17182 * config/nds32/nds32-isr.c: Likewise.
17183 * config/nds32/nds32.c: Likewise.
17184 * config/nios2/nios2.c: Likewise.
17185 * config/nvptx/nvptx.c: Likewise.
17186 * config/pa/pa.c: Likewise.
17187 * config/pdp11/pdp11.c: Likewise.
17188 * config/powerpcspe/powerpcspe.c: Likewise.
17189 * config/riscv/riscv.c: Likewise.
17190 * config/rl78/rl78.c: Likewise.
17191 * config/rx/rx.c: Likewise.
17192 * config/s390/s390.c: Likewise.
17193 * config/sh/sh.c: Likewise.
17194 * config/sol2.c: Likewise.
17195 * config/sparc/sparc.c: Likewise.
17196 * config/spu/spu.c: Likewise.
17197 * config/stormy16/stormy16.c: Likewise.
17198 * config/tilegx/tilegx.c: Likewise.
17199 * config/tilepro/tilepro.c: Likewise.
17200 * config/v850/v850.c: Likewise.
17201 * config/vax/vax.c: Likewise.
17202 * config/visium/visium.c: Likewise.
17203 * config/xtensa/xtensa.c: Likewise.
17205 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17208 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
17209 constraints since the -mupper-regs-* switches have been
17211 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
17212 into a vector from a double word element that was extracted from
17213 another vector, and eliminate extra XXPERMDI instructions.
17214 (vsx_concat_<mode>_2): Likewise.
17215 (vsx_concat_<mode>_3): Likewise.
17216 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
17217 concat to allow optimizing inserts from previous extracts.
17219 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
17221 * config/i386/i386.c (ix86_stack_protect_guard): Generate
17222 memory reference to a SSP offset in TLS address space.
17223 (ix86_print_operand) <case '@'>: Remove.
17224 (ix86_print_operand_punct_valid_p): Remove '@' code.
17225 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
17226 UNSPEC_SP_TLS_TEST.
17227 (stack_tls_protect_set_<mode>): Remove.
17228 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
17229 (stack_tls_protect_test_<mode>): Remove.
17230 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
17232 2017-08-07 Olivier Hainque <hainque@adacore.com>
17235 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
17237 2017-08-07 Douglas Rupp <rupp@adacore.com>
17239 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
17240 variable was referenced as multidir in command.
17242 2017-08-07 Jakub Jelinek <jakub@redhat.com>
17245 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
17248 2017-08-07 Martin Liska <mliska@suse.cz>
17250 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
17251 * config/rl78/rl78.c: Add include of attribs.h.
17252 * config/sh/sh.c: Likewise.
17253 * config/v850/v850.c: Likewise.
17255 2017-08-07 Tom de Vries <tom@codesourcery.com>
17257 PR middle-end/78266
17258 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
17260 2017-08-07 Martin Liska <mliska@suse.cz>
17262 * config/mips/mips.c: Include attribs.h.
17264 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
17267 * doc/invoke.texi: Document change in behvaior for -Ofast for
17270 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
17272 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
17274 (aarch64_pop_regs): Likewise.
17275 (aarch64_gen_load_pair): Likewise.
17276 (aarch64_save_callee_saves): Likewise.
17277 (aarch64_restore_callee_saves): Likewise.
17279 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17281 * config/i386/i386.c: Revert the last change.
17283 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17286 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17288 (ix86_finalize_stack_frame_flags): This. Also clear
17289 frame_pointer_needed if -fno-omit-frame-pointer is used without
17291 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17292 with ix86_finalize_stack_frame_flags.
17293 (ix86_expand_epilogue): Likewise.
17294 (ix86_expand_split_stack_prologue): Likewise.
17296 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17299 * config/i386/i386.c (get_builtin_code_for_version): Set priority
17300 to P_AES for Westmere.
17302 2017-08-07 Jonathan Yong <10walls@gmail.com>
17304 * config/i386/mingw.opt (fset-stack-executable): Removed.
17305 * config/i386/cygming.opt (fset-stack-executable): Moved
17307 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
17309 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
17311 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
17313 2017-08-07 Marek Polacek <polacek@redhat.com>
17315 PR middle-end/81737
17316 * fold-const.c (fold_indirect_ref_1): Check type_domain.
17318 2017-08-07 Martin Liska <mliska@suse.cz>
17320 * attribs.h (canonicalize_attr_name): New function.
17321 (cmp_attribs): Move from c-format.c and adjusted.
17322 (is_attribute_p): Moved from tree.h.
17323 * tree-inline.c: Add new includes.
17324 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
17325 (private_is_attribute_p): Remove.
17326 (private_lookup_attribute): Likewise.
17327 (private_lookup_attribute_by_prefix): Simplify.
17328 (remove_attribute): Use is_attribute_p.
17329 * tree.h: Remove removed declarations.
17331 2017-08-07 Jakub Jelinek <jakub@redhat.com>
17333 PR middle-end/81698
17334 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
17335 instead of computing it in the function. Formatting fix.
17336 (expand_case): Don't rely on default_edge being the first edge,
17337 clear it if removing it, pass default_edge to
17338 emit_case_dispatch_table.
17339 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
17342 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
17344 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
17345 insn in the function, emit NOP after the insn.
17347 2017-08-06 Tom de Vries <tom@codesourcery.com>
17349 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
17352 2017-08-06 Tom de Vries <tom@codesourcery.com>
17354 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
17357 2017-08-04 Yury Gribov <tetra2005@gmail.com>
17359 PR tree-optimization/57371
17360 * match.pd: New pattern.
17362 2017-08-04 Marek Polacek <polacek@redhat.com>
17364 PR middle-end/81695
17365 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
17366 perform the computation in offset_int.
17368 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17370 PR tree-optimization/81136
17371 * tree-vectorizer.h: Include tree-hash-traits.h.
17372 (vec_base_alignments): New typedef.
17373 (vec_info): Add a base_alignments field.
17374 (vect_record_base_alignments): Declare.
17375 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
17377 (DR_IS_CONDITIONAL_IN_STMT): New macro.
17378 (create_data_ref): Add an is_conditional_in_stmt argument.
17379 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
17380 the is_conditional_in_stmt field.
17381 (data_ref_loc): Add an is_conditional_in_stmt field.
17382 (get_references_in_stmt): Set the is_conditional_in_stmt field.
17383 (find_data_references_in_stmt): Update call to create_data_ref.
17384 (graphite_find_data_references_in_stmt): Likewise.
17385 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
17386 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
17387 (vect_record_base_alignment): New function.
17388 (vect_record_base_alignments): Likewise.
17389 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
17390 for nested statements even if we fail to compute a misalignment.
17391 Use pooled base alignments for unconditional references.
17392 (vect_find_same_alignment_drs): Compare base addresses instead
17394 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
17395 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
17397 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17399 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
17400 Add an explicit name for the enum. Use auto_vec for slp_instances
17401 and grouped_stores.
17402 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
17404 (_bb_vec_info): Add a constructor and destructor.
17405 (vinfo_for_stmt): Return NULL for uids of -1 as well.
17406 (destroy_loop_vec_info): Delete.
17407 (vect_destroy_datarefs): Likewise.
17408 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
17409 (vec_info::vec_info): New function.
17410 (vec_info::~vec_info): Likewise.
17411 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
17412 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
17413 destroy_loop_vec_info.
17414 * tree-vect-loop.c (new_loop_vec_info): Replace with...
17415 (_loop_vec_info::_loop_vec_info): ...this.
17416 (destroy_loop_vec_info): Replace with...
17417 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
17418 the stmt_vec_infos. Leave handling of vec_info information to its
17419 destructor. Remove explicit vector releases.
17420 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
17421 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
17422 * tree-vect-slp.c (new_bb_vec_info): Replace with...
17423 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
17424 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
17425 (destroy_bb_vec_info): Replace with...
17426 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
17427 information to its destructor.
17428 (vect_slp_analyze_bb_1): Use new and delete instead of
17429 new_bb_vec_info and destroy_bb_vec_info.
17430 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
17433 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17435 * tree-data-ref.h (subscript): Add access_fn field.
17436 (data_dependence_relation): Add could_be_independent_p.
17437 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
17438 (same_access_functions): Move to tree-data-ref.c.
17439 * tree-data-ref.c (ref_contains_union_access_p): New function.
17440 (access_fn_component_p): Likewise.
17441 (access_fn_components_comparable_p): Likewise.
17442 (dr_analyze_indices): Add a reference to access_fn_component_p.
17443 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
17445 (constant_access_functions): Likewise.
17446 (add_other_self_distances): Likewise.
17447 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
17448 (initialize_data_dependence_relation): Use XCNEW and remove
17449 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
17450 of access functions that have the same type. Allow the
17451 subsequence to end with different bases in some circumstances.
17452 Record the chosen access functions in SUB_ACCESS_FN.
17453 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
17454 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
17455 (subscript_dependence_tester_1): Likewise dra and drb.
17456 (build_classic_dist_vector): Update calls accordingly.
17457 (subscript_dependence_tester): Likewise.
17458 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
17459 DDR_COULD_BE_INDEPENDENT_P.
17460 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
17461 comp_alias_ddrs instead of may_alias_ddrs.
17462 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
17464 (vect_analyze_data_ref_dependence): Use it if
17465 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
17466 distance vectors if that fails.
17467 (dependence_distance_ge_vf): New function.
17468 (vect_prune_runtime_alias_test_list): Use it. Don't clear
17469 LOOP_VINFO_MAY_ALIAS_DDRS.
17471 2017-08-04 Richard Biener <rguenther@suse.de>
17473 PR middle-end/81705
17474 * fold-const.c (fold_binary_loc): Properly restrict
17475 minus_var0 && minus_var1 case when associating undefined overflow
17478 2017-08-04 Tom de Vries <tom@codesourcery.com>
17480 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
17482 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17484 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17485 Don't start diagnostic messages with a capital letter.
17486 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17488 (rs6000_invalid_builtin): Likewise.
17489 (rs6000_trampoline_init): Likewise.
17491 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17494 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
17495 after setting changeable df flags.
17497 2017-08-03 Richard Biener <rguenther@suse.de>
17499 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
17500 up if the use is in USE - X.
17502 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
17504 * toplev.c (dumpfile.h): New include.
17505 (internal_error_reentered): New static function. Use it...
17506 (internal_error_function): ...here to handle reentered internal_error.
17508 2017-08-03 Richard Biener <rguenther@suse.de>
17510 PR middle-end/81148
17511 * fold-const.c (split_tree): Add minus_var and minus_con
17512 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
17513 here but always use minus_*.
17514 (associate_trees): Assert we never associate with MINUS_EXPR
17515 and NULL first operand. Do not recurse for PLUS_EXPR operands
17516 when associating as MINUS_EXPR either.
17517 (fold_binary_loc): Track minus_var and minus_con.
17519 2017-08-03 Tom de Vries <tom@codesourcery.com>
17522 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
17523 ACCEL_COMPILER, apply finish_options on
17524 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
17526 2017-08-03 Tom de Vries <tom@codesourcery.com>
17529 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
17530 function_entry_patch_area_size > 0.
17532 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17535 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
17536 instead of 10??LU, perform unit multiplication in wide_int,
17537 don't change alloc_object_size_limit if the limit is larger
17540 PR tree-optimization/81655
17541 PR tree-optimization/81588
17542 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
17543 the case when ranges[i].low and high are 1 for unsigned type with
17546 PR middle-end/81052
17547 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
17548 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
17550 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17552 * tree-vrp.h: Add include guard.
17554 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
17557 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
17558 (ud2): New insn pattern.
17559 * config/i386/i386.c (ix86_expand_epilogue):
17560 For naked functions, generate ud2 instead of trap insn.
17562 2017-08-02 Marek Polacek <polacek@redhat.com>
17565 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
17567 2017-08-02 Tom de Vries <tom@codesourcery.com>
17568 Cesar Philippidis <cesar@codesourcery.com>
17570 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
17571 Add missing edge probabilities.
17573 2017-08-02 Tamar Christina <tamar.christina@arm.com>
17575 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
17576 Correct endianness.
17578 2017-08-02 Jakub Jelinek <jakub@redhat.com>
17580 PR middle-end/79499
17581 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
17582 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
17583 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
17584 prologue_seq sequences - if any.
17586 2017-08-02 Richard Biener <rguenther@suse.de>
17588 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
17589 via vectors if supported, integer extracts via punning if supported
17590 or otherwise vector extracts.
17592 2017-08-02 Richard Biener <rguenther@suse.de>
17594 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
17596 (bitmap_insert_into_set): ... this.
17598 2017-08-02 Richard Biener <rguenther@suse.de>
17600 PR tree-optimization/81633
17602 2015-08-17 Alan Hayward <alan.hayward@arm.com>
17604 PR tree-optimization/71752
17605 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
17607 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
17609 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
17610 (machine_function::call_ms2sysv_pad_out): Remove field.
17611 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
17612 (ix86_compute_frame_layout): Likewise.
17614 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
17617 * config/i386/i386.c (ix86_set_func_type): Disallow naked
17618 attribute with interrupt attribute.
17620 2017-08-01 Andrew Pinski <apinski@cavium.com>
17622 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
17623 BIT_INSERT_EXPR's operand 1
17624 to see if the types precision matches.
17626 2017-08-01 Martin Liska <mliska@suse.cz>
17628 PR middle-end/70140
17629 * builtins.c (expand_builtin_memcpy_args): Remove.
17630 (expand_builtin_memcpy): Call newly added function
17631 expand_builtin_memory_copy_args.
17632 (expand_builtin_memcpy_with_bounds): Likewise.
17633 (expand_builtin_mempcpy): Remove last argument.
17634 (expand_builtin_mempcpy_with_bounds): Likewise.
17635 (expand_builtin_memory_copy_args): New function created from
17636 expand_builtin_mempcpy_args with small modifications.
17637 (expand_builtin_mempcpy_args): Remove.
17638 (expand_builtin_stpcpy): Remove unused argument.
17639 (expand_builtin): Likewise.
17640 (expand_builtin_with_bounds): Likewise.
17642 2017-08-01 Martin Liska <mliska@suse.cz>
17645 Make mempcpy more optimal (PR middle-end/70140).
17647 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17650 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
17651 __builtin_vec_cmpne verify both arguments are compatible vectors
17652 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
17653 verify arg1_type is a pointer or array type. For __builtin_vec_st,
17654 move computation of aligned to after checking the argument types.
17658 * config/rs6000/vsx.md (vextract_fp_from_shorth,
17659 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
17662 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
17663 Doug Rupp <rupp@adacore.com>
17664 Olivier Hainque <hainque@adacore.com>
17666 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
17667 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
17669 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
17670 for TARGET_OS_CPP_BUILTIN.
17671 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
17672 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
17674 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
17675 passing required abi options to the assembler for EABI configurations.
17676 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
17677 of .text.hot and .text.unlikely sections for kernel modules when
17678 using ARM style exceptions.
17679 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
17680 options. Add EXTRA_CC1_SPEC.
17681 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
17683 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
17685 (ARM_TARGET2_DWARF_FORMAT): Define.
17686 * config/arm/t-vxworks: Adjust multilib control to removal of the
17687 Diab command line options.
17689 2017-08-01 Martin Liska <mliska@suse.cz>
17691 PR gcov-profile/81561
17692 * gcov.c (unblock): Make unblocking safe as we need to preserve
17693 index correspondence of blocks and block_lists.
17695 2017-08-01 Richard Biener <rguenther@suse.de>
17697 PR tree-optimization/81181
17698 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
17699 (compute_antic): ... end of iteration here.
17701 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
17703 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
17704 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
17705 (ftree-slp-vectorize): Likewise.
17706 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
17707 can no longer be set independent of flag_tree_loop_vectorize.
17708 * omp-general.c (emp_max_vf): Likewise.
17709 * opts.c (enable_fdo_optimizations): Remove references to
17710 flag_tree_vectorize, these are now implicit.
17711 (common_handle_option): Remove handling for OPT_ftree_vectorize,
17712 and leave it for the options machinery.
17714 2017-08-01 Martin Liska <mliska@suse.cz>
17716 PR middle-end/70140
17717 * builtins.c (expand_builtin_memcpy_args): Remove.
17718 (expand_builtin_memcpy): Call newly added function
17719 expand_builtin_memory_copy_args.
17720 (expand_builtin_memcpy_with_bounds): Likewise.
17721 (expand_builtin_mempcpy): Remove last argument.
17722 (expand_builtin_mempcpy_with_bounds): Likewise.
17723 (expand_builtin_memory_copy_args): New function created from
17724 expand_builtin_mempcpy_args with small modifications.
17725 (expand_builtin_mempcpy_args): Remove.
17726 (expand_builtin_stpcpy): Remove unused argument.
17727 (expand_builtin): Likewise.
17728 (expand_builtin_with_bounds): Likewise.
17730 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
17733 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
17734 print "ds:" only for immediates in generic address space.
17736 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
17739 * config/i386/i386.c (ix86_funciton_naked): New prototype.
17740 (ix86_function_ok_for_sibcall): Return false for naked functions.
17742 2017-08-01 Richard Biener <rguenther@suse.de>
17744 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
17745 (compute_antic): Seed worklist with exit block predecessors.
17746 * cfganal.c (dfs_find_deadend): For a cycle return the source
17747 of the edge closing it.
17749 2017-08-01 Tamar Christina <tamar.christina@arm.com>
17751 * config/aarch64/aarch64.c
17752 (aarch64_can_const_movi_rtx_p): Move 0 check.
17754 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17756 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
17757 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
17759 * match.pd: Ditto in address comparison pattern.
17761 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17763 PR tree-optimization/81627
17764 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
17765 closed ssa form for store-store chain.
17767 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17769 PR tree-optimization/81620
17770 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
17771 for store-store chain.
17773 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17775 PR tree-optimization/81588
17776 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
17777 ranges[i].in_p, invert comparison code ccode. For >/>=,
17778 swap rhs1 and rhs2 and comparison code unconditionally,
17779 for </<= don't do that. Don't swap rhs1/rhs2 again if
17780 ranges[i].in_p, instead invert comparison code ccode if
17781 opcode or oe->rank is BIT_IOR_EXPR.
17784 * optabs.def (vec_extract_optab, vec_init_optab): Change from
17785 a direct optab to conversion optab.
17786 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
17787 with GET_MODE_INNER as last argument instead of optab_handler.
17788 * expmed.c (extract_bit_field_1): Likewise. Use vector from
17789 vector extraction if possible and optab is available.
17790 * expr.c (store_constructor): Use convert_optab_handler instead
17791 of optab_handler. Use vector initialization from smaller
17792 vectors if possible and optab is available.
17793 * tree-vect-stmts.c (vectorizable_load): Likewise.
17794 * doc/md.texi (vec_extract, vec_init): Document that the optabs
17795 now have two modes.
17796 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
17797 of vec_init from half-sized vectors with the same element mode.
17798 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
17799 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
17800 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
17801 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
17802 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
17803 after mode in gen_vec_extract* calls.
17804 (vec_extract<mode>): Renamed to ...
17805 (vec_extract<mode><ssescalarmodelower>): ... this.
17806 (vec_extract<mode><ssehalfvecmodelower>): New expander.
17807 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
17808 element mode after mode in gen_vec_init* calls.
17809 (VEC_INIT_HALF_MODE): New mode iterator.
17810 (vec_init<mode>): Renamed to ...
17811 (vec_init<mode><ssescalarmodelower>): ... this.
17812 (vec_init<mode><ssehalfvecmodelower>): New expander.
17813 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
17814 (vec_extractv2sfsf): ... this.
17815 (vec_initv2sf): Renamed to ...
17816 (vec_initv2sfsf): ... this.
17817 (vec_extractv2si): Renamed to ...
17818 (vec_extractv2sisi): ... this.
17819 (vec_initv2si): Renamed to ...
17820 (vec_initv2sisi): ... this.
17821 (vec_extractv4hi): Renamed to ...
17822 (vec_extractv4hihi): ... this.
17823 (vec_initv4hi): Renamed to ...
17824 (vec_initv4hihi): ... this.
17825 (vec_extractv8qi): Renamed to ...
17826 (vec_extractv8qiqi): ... this.
17827 (vec_initv8qi): Renamed to ...
17828 (vec_initv8qiqi): ... this.
17829 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
17830 (vec_init<mode>): Renamed to ...
17831 (vec_init<mode><VEC_base_l>): ... this.
17832 (vec_extract<mode>): Renamed to ...
17833 (vec_extract<mode><VEC_base_l>): ... this.
17834 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
17835 (vec_initv2sfsf): ... this.
17836 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
17837 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17838 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
17839 element mode after mode in gen_vec_init* calls.
17840 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
17841 (vec_init<mode><Vel>): ... this.
17842 (vec_extract<mode>): Renamed to ...
17843 (vec_extract<mode><Vel>): ... this.
17844 * config/aarch64/iterators.md (Vel): New mode attribute.
17845 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
17846 Add element mode after mode in gen_vec_extract* calls.
17847 * config/s390/vector.md (non_vec_l): New mode attribute.
17848 (vec_extract<mode>): Renamed to ...
17849 (vec_extract<mode><non_vec_l>): ... this.
17850 (vec_init<mode>): Renamed to ...
17851 (vec_init<mode><non_vec_l>): ... this.
17852 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
17853 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
17855 * config/arm/iterators.md (V_elem_l): New mode attribute.
17856 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
17857 (vec_extract<mode><V_elem_l>): ... this.
17858 (vec_extractv2di): Renamed to ...
17859 (vec_extractv2didi): ... this.
17860 (vec_init<mode>): Renamed to ...
17861 (vec_init<mode><V_elem_l>): ... this.
17862 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
17863 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
17864 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
17865 Add element mode after gen_vec_extract* calls.
17866 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
17867 (vec_init<mode><unitmode>): ... this.
17868 (vec_extract<mode>): Renamed to ...
17869 (vec_extract<mode><unitmode>): ... this.
17870 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
17871 (vec_init<mode><unitmode>): ... this.
17872 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
17873 (vec_initv2sfsf): ... this.
17874 (vec_extractv2sf): Renamed to ...
17875 (vec_extractv2sfsf): ... this.
17876 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
17877 Add element mode after gen_vec_extract* calls.
17878 * config/mips/mips.md (unitmode): New mode iterator.
17879 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
17880 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
17881 * config/spu/spu.md (inner_l): New mode attribute.
17882 (vec_init<mode>): Renamed to ...
17883 (vec_init<mode><inner_l>): ... this.
17884 (vec_extract<mode>): Renamed to ...
17885 (vec_extract<mode><inner_l>): ... this.
17886 * config/sparc/sparc.md (veltmode): New mode iterator.
17887 (vec_init<VMALL:mode>): Renamed to ...
17888 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
17889 * config/ia64/vect.md (vec_initv2si): Renamed to ...
17890 (vec_initv2sisi): ... this.
17891 (vec_initv2sf): Renamed to ...
17892 (vec_initv2sfsf): ... this.
17893 (vec_extractv2sf): Renamed to ...
17894 (vec_extractv2sfsf): ... this.
17895 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
17896 (vec_init<mode>): Renamed to ...
17897 (vec_init<mode><VEC_base_l>): ... this.
17898 (vec_extract<mode>): Renamed to ...
17899 (vec_extract<mode><VEC_base_l>): ... this.
17900 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
17901 (vec_initv2sfsf): ... this.
17902 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
17903 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17904 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
17905 gen_vec_init* calls.
17907 2017-08-01 Richard Biener <rguenther@suse.de>
17909 PR tree-optimization/81297
17910 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
17911 TREE_OVERFLOW from INTEGER_CSTs.
17913 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
17915 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
17917 2017-07-31 Carl Love <cel@us.ibm.com>
17919 * config/rs6000/rs6000-c: Add support for built-in functions
17920 vector signed char vec_xl_be (signed long long, signed char *);
17921 vector unsigned char vec_xl_be (signed long long, unsigned char *);
17922 vector signed int vec_xl_be (signed long long, signed int *);
17923 vector unsigned int vec_xl_be (signed long long, unsigned int *);
17924 vector signed long long vec_xl_be (signed long long, signed long long *);
17925 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
17926 vector signed short vec_xl_be (signed long long, signed short *);
17927 vector unsigned short vec_xl_be (signed long long, unsigned short *);
17928 vector double vec_xl_be (signed long long, double *);
17929 vector float vec_xl_be (signed long long, float *);
17930 * config/rs6000/altivec.h (vec_xl_be): Add #define.
17931 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
17932 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
17934 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
17935 (altivec_expand_builtin): Add switch statement to call
17936 altivec_expand_xl_be for each builtin.
17937 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
17938 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
17939 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
17940 __builtin_vsx_le_be_v16qi.
17941 * doc/extend.texi: Update the built-in documentation file for the
17942 new built-in functions.
17944 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17947 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
17949 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17951 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17953 * config.gcc: Add z14.
17954 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
17955 CPU model numbers for z13s and z14.
17956 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
17958 * config/s390/s390-opts.h (enum processor_type): Rename
17959 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
17960 * config/s390/s390.c (processor_table): Add field for CPU name to
17961 be passed to Binutils.
17962 (s390_asm_output_machine_for_arch): Use the new field in
17963 processor_table for Binutils.
17964 (s390_expand_builtin): Replace arch12 with z14.
17965 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
17966 (s390_get_sched_attrmask): Likewise.
17967 (s390_get_unit_mask): Likewise.
17968 * config/s390/s390.opt: Add z14 to processor_type enum.
17970 2017-07-31 Martin Jambor <mjambor@suse.cz>
17973 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
17974 regardless of optimization level.
17976 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
17977 Martin Liska <mliska@suse.cz>
17979 * predict.def: Remove old comment and adjust probability.
17980 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
17981 PREDICT statements.
17983 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17986 * config/i386/i386.c (ix86_function_naked): New function.
17987 (ix86_can_use_return_insn_p): Return false for naked functions.
17988 (ix86_expand_prologue): Skip prologue for naked functions.
17989 (ix86_expand_epilogue): Skip epilogue for naked functions
17990 and emit trap instruction.
17991 (ix86_warn_func_return): New function.
17992 (ix86_attribute_table): Add "naked" attribute specification.
17993 (TARGET_WARN_FUNC_RETURN): Define.
17994 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
17996 2017-07-31 Martin Liska <mliska@suse.cz>
17998 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
17999 (dump_gimple_bb_header): Always dump BB info.
18000 (pp_cfg_jump): Do not append info about BB when dumping a jump.
18002 2017-07-31 Martin Liska <mliska@suse.cz>
18005 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
18006 also with current_function_decl non-null equality.
18008 2017-07-31 Jakub Jelinek <jakub@redhat.com>
18011 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
18012 change type to the element type, instead add eltype variable and
18013 use it where we are interested in the element type.
18015 PR tree-optimization/81603
18016 * ipa-polymorphic-call.c
18017 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
18018 offset arithmetic in offset_int, bail out if the resulting bit offset
18019 doesn't fit into shwi.
18021 2017-07-31 Martin Liska <mliska@suse.cz>
18023 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
18024 (gimplify_save_expr): Fix comment.
18026 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
18029 * config/i386/i386.c (ix86_function_arg): Update arguments for
18031 (ix86_compute_frame_layout): Set the initial stack offset to
18032 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
18033 INCOMING_FRAME_SP_OFFSET.
18034 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
18035 stack before exception handler returns.
18036 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
18037 the 'ERROR_CODE' for exception handler.
18039 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
18041 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
18042 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
18043 (ASM_OUTPUT_REG_POP): Ditto.
18044 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
18045 instead of asm_fprintf to output pure string.
18047 2017-07-29 Jakub Jelinek <jakub@redhat.com>
18049 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
18050 to imported_module_or_decl hook.
18051 (debug_nothing_tree_tree_tree_bool): Remove.
18052 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
18053 * debug.c (do_nothing_debug_hooks): Use
18054 debug_nothing_tree_tree_tree_bool_bool instead of
18055 debug_nothing_tree_tree_tree_bool.
18056 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
18057 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
18058 * sdbout.c (sdb_debug_hooks): Likewise.
18059 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
18060 (gen_namespace_die): Add DW_AT_export_symbols attribute if
18062 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
18063 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
18064 attribute, don't add anything.
18066 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18068 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
18069 (fold_build2_stat_loc): Likewise.
18070 (fold_build3_stat_loc): Likewise.
18071 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
18072 (fold_build1_loc): Remove macro.
18073 (fold_build2_loc): Likewise.
18074 (fold_build3_loc): Likewise.
18076 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18078 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
18079 (gimple_build_debug_bind_source_stat): Likewise.
18080 * gimple.h (gimple_build_debug_bind): Remove macro.
18081 (gimple_build_debug_bind_source): Likewise.
18083 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18085 * bitmap.c (bitmap_alloc): Adjust.
18086 (bitmap_gc_alloc): Likewise.
18087 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
18089 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18091 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
18092 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
18093 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
18094 (bitmap_gc_alloc_stat): Likewise.
18095 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
18097 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18099 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
18100 * rtl.h (shallow_copy_rtx): Remove macro.
18102 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18104 * emit-rtl.c (gen_raw_REG): Adjust.
18105 * gengenrtl.c (gendef): Likewise.
18106 * rtl.c (rtx_alloc_stat): Remove _stat from name.
18107 * rtl.h (rtx_alloc): Remove macro.
18109 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18111 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
18112 (build_tree_list_stat): Likewise.
18113 * tree.h (build_tree_list): Remove macro.
18114 (build_tree_list_vec): Likewise.
18116 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18118 * tree.c (make_vector_stat): Remove _stat from name.
18119 (build_vector_stat): Likewise.
18120 * tree.h (make_vector_stat): Remove macro.
18121 (build_vector_stat): Likewise.
18123 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18125 * tree.h (build_var_debug_value): Remove prototype.
18127 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18129 * tree.c (tree_cons_stat): Remove _stat from name.
18130 * tree.h (tree_cons): Remove macro.
18132 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18134 * tree.c (build_vl_exp_stat): Remove _stat from name.
18135 * tree.h (build_vl_exp): Remove macro.
18137 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18139 * tree.c (build_decl_stat): Remove _stat from name.
18140 * tree.h (build_decl): Remove macro.
18142 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18144 * gimple.c (gimple_build_with_ops_stat): Adjust.
18145 (gimple_alloc_stat): Remove _stat from name.
18146 * gimple.h (gimple_alloc): Remove macro.
18148 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18150 * tree.c (make_tree_vec_stat): Remove _stat from name.
18151 (grow_tree_vec_stat): Likewise.
18152 * tree.h (make_tree_vec_stat): Adjust prototype.
18153 (grow_tree_vec_stat): Likewise.
18154 (make_tree_vec): Remove macro.
18155 (grow_tree_vec): Likewise.
18157 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18159 * fold-const.c (fold_build1_stat_loc): Adjust.
18160 (fold_build2_stat_loc): Likewise.
18161 (fold_build3_stat_loc): Likewise.
18162 * tree.c (build0_stat): Remove _stat from name.
18163 (build1_stat): Likewise.
18164 (build2_stat): Likewise.
18165 (build3_stat): Likewise.
18166 (build4_stat): Likewise.
18167 (build5_stat): Likewise.
18168 * tree.h (build1_loc): Remove macro, and rename _stat function
18170 (build2_loc): Likewise.
18171 (build3_loc): Likewise.
18172 (build4_loc): Likewise.
18173 (build5_loc): Likewise.
18175 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18177 * tree.c (make_int_cst_stat): Remove _stat from name.
18178 * tree.h (make_int_cst_stat): Adjust prototype.
18179 (make_int_cst): Remove macro.
18181 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18183 * tree.c (make_tre_binfo_stat): Remove _stat from name.
18184 * tree.h (make_tree_binfo_stat): Adjust prototype.
18185 (make_tree_binfo): Remove.
18187 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18189 * tree.c (copy_node_stat): Rename to copy_node.
18190 (build_distinct_type_copy): Adjust.
18191 * tree.h (copy_node_stat): Adjust prototype.
18192 (copy_node): Remove macro.
18194 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18196 * tree.c (make_node_stat): rename to make_node.
18197 (build_tree_list_stat): Adjust.
18198 (build0_stat): Likewise.
18199 (build2_stat): Likewise.
18200 (build3_stat): Likewise.
18201 (build4_stat): Likewise.
18202 (build5_stat): Likewise.
18203 (build_decl_stat): Likewise.
18204 * tree.h (make_node_stat): Adjust prototype.
18205 (make_node): remove macro.
18207 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
18209 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
18210 (PPC_FEATURE2_SCV): Likewise.
18211 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
18213 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18215 * config/aarch64/aarch64.c
18216 (aarch64_internal_mov_immediate): Add new special pattern.
18217 * config/aarch64/aarch64.md (*movdi_aarch64):
18218 Add reg/32bit const mov case.
18220 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18221 Richard Sandiford <richard.sandiford@linaro.org>
18223 * config/aarch64/aarch64.md (mov<mode>): Generalize.
18224 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18225 Add integer and movi cases.
18226 (movi-split-hf-df-sf split, fp16): New.
18227 (enabled): Added TARGET_FP_F16INST.
18228 * config/aarch64/iterators.md (GPF_HF): New.
18229 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
18231 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18233 * config/aarch64/aarch64.c
18234 (aarch64_simd_container_mode): Add prototype.
18235 (aarch64_expand_mov_immediate): Add HI support.
18236 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
18237 (aarch64_can_const_movi_rtx_p): New.
18238 (aarch64_preferred_reload_class):
18239 Remove restrictions of using FP registers for certain SIMD operations.
18240 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
18241 (aarch64_valid_floating_const): Add integer move validation.
18242 (aarch64_simd_imm_scalar_p): Remove.
18243 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
18244 (aarch64_legitimate_constant_p): Expand list of supported cases.
18245 * config/aarch64/aarch64-protos.h
18246 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
18247 (aarch64_reinterpret_float_as_int): New.
18248 (aarch64_simd_imm_scalar_p): Remove.
18249 * config/aarch64/constraints.md (Uvi): New.
18250 (Dd): Split into Ds and new Dd.
18251 * config/aarch64/aarch64.md (*movsi_aarch64):
18253 (*movdi_aarch64): Add SIMD mov case.
18255 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18257 * tree-predcom.c: (struct chain): Handle store-store chain in which
18258 stores for elimination only store loop invariant values.
18259 (execute_pred_commoning_chain): Ditto.
18260 (prepare_initializers_chain_store_elim): Ditto.
18261 (prepare_finalizers): Ditto.
18262 (is_inv_store_elimination_chain): New function.
18263 (initialize_root_vars_store_elim_1): New function.
18265 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18267 * tree-predcom.c: Revise general description of the pass.
18268 (enum chain_type): New enum type for store elimination.
18269 (struct chain): New field supporting store elimination.
18270 (struct component): Ditto.
18271 (dump_chain): Dump store-stores chain.
18272 (release_chain): Release resources.
18273 (split_data_refs_to_components): Compute and create component
18274 contains only stores for elimination.
18275 (get_chain_last_ref_at): New function.
18276 (make_invariant_chain): Initialization.
18277 (make_rooted_chain): Specify chain type in parameter and record it.
18278 (add_looparound_copies): Skip for store-stores chain.
18279 (determine_roots_comp): Compute type of chain and pass it to
18281 (initialize_root_vars_store_elim_2): New function.
18282 (finalize_eliminated_stores): New function.
18283 (remove_stmt): Handle store for elimination.
18284 (execute_pred_commoning_chain): Execute predictive commoning on
18285 store-store chains.
18286 (determine_unroll_factor): Skip unroll for store-stores chain.
18287 (prepare_initializers_chain_store_elim): New function.
18288 (prepare_initializers_chain): Hanlde store-store chain.
18289 (prepare_finalizers_chain, prepare_finalizers): New function.
18290 (tree_predictive_commoning_loop): Return integer value indicating
18291 if loop is unrolled or lcssa form is corrupted.
18292 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
18294 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18296 * tree-predcom.c (initialize_root): Delete.
18297 (execute_pred_commoning_chain): Initialize root vars and replace
18298 reference of non-combined chain directly, rather than call above
18301 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18303 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
18304 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
18306 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18308 * tree-predcom.c (struct chain): New field init_seq.
18309 (release_chain): Release init_seq.
18310 (prepare_initializers_chain): Record intialization stmts in above
18312 (insert_init_seqs): New function.
18313 (tree_predictive_commoning_loop): Call insert_init_seqs.
18315 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18317 * tree-predcom.c (determine_roots_comp): Skip trivial components.
18319 2017-07-28 Richard Biener <rguenther@suse.de>
18321 * match.pd: Remove superfluous :c.
18322 * genmatch.c (simplify::id): Add member.
18323 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
18325 (current_id): New global.
18326 (dt_node::parent): Move from ...
18327 (dt_operand::parent): ... here. Add for_id member.
18328 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
18329 (decision_tree::find_node): Relax order requirement when
18330 merging DT_TRUE nodes to ones inbetween the current simplify
18331 and the one we try to merge with. Add diagnostic whenever
18332 we need to enforce pattern order by not merging.
18333 (decision_tree::insert): Set current_id.
18334 (decision_tree::print_node): Dump parent node and for_id.
18335 (parser::last_id): Add member.
18336 (parser::push_simplify): Assign unique id.
18337 (parser::parser): Initialize last_id.
18339 2017-07-28 Martin Liska <mliska@suse.cz>
18342 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
18343 gimple_build_debug_bind.
18345 2017-07-28 Richard Biener <rguenther@suse.de>
18347 PR tree-optimization/81502
18348 * match.pd: Add pattern combining BIT_INSERT_EXPR with
18350 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
18352 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
18353 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
18354 for BIT_FIELD_REF args.
18355 * fold-const.c (make_bit_field_ref): Likewise.
18356 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
18358 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18361 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
18362 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
18363 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
18364 Or it into SANITIZER_UNDEFINED.
18365 * ubsan.c: Include gimple-fold.h and varasm.h.
18366 (ubsan_expand_ptr_ifn): New function.
18367 (instrument_pointer_overflow): New function.
18368 (maybe_instrument_pointer_overflow): New function.
18369 (instrument_object_size): Formatting fix.
18370 (pass_ubsan::execute): Call instrument_pointer_overflow
18371 and maybe_instrument_pointer_overflow.
18372 * internal-fn.c (expand_UBSAN_PTR): New function.
18373 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
18374 * sanitizer.def (__ubsan_handle_pointer_overflow,
18375 __ubsan_handle_pointer_overflow_abort): New builtins.
18376 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
18377 * internal-fn.def (UBSAN_PTR): New internal function.
18378 * opts.c (sanitizer_opts): Add pointer-overflow.
18379 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
18380 * fold-const.c (build_range_check): Compute pointer range check in
18381 integral type if pointer arithmetics would be needed. Formatting
18384 2017-07-28 Martin Liska <mliska@suse.cz>
18387 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
18388 parameters that are of a variable-length.
18390 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
18392 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
18394 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
18395 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
18396 (CRT_CALL_STATIC_FUNCTION): Likewise.
18397 (ASM_DEFAULT_SPEC): New define.
18398 (ASM_SPEC32): Likewise.
18399 (ASM_SPEC64): Likewise.
18400 (ASM_SPEC_COMMON): Likewise.
18401 (ASM_SPEC): Likewise.
18402 (INVALID_64BIT): Likewise.
18403 (LINK_OS_DEFAULT_SPEC): Likewise.
18404 (LINK_OS_SPEC32): Likewise.
18405 (LINK_OS_SPEC64): Likewise.
18406 (POWERPC_LINUX): Likewise.
18407 (PTRDIFF_TYPE): Likewise.
18408 (RESTORE_FP_PREFIX): Likewise.
18409 (RESTORE_FP_SUFFIX): Likewise.
18410 (SAVE_FP_PREFIX): Likewise.
18411 (SAVE_FP_SUFFIX): Likewise.
18412 (SIZE_TYPE): Likewise.
18413 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
18414 (TARGET_64BIT): Likewise.
18415 (TARGET_64BIT): Likewise.
18416 (TARGET_AIX): Likewise.
18417 (WCHAR_TYPE_SIZE): Likewise.
18418 (WCHAR_TYPE): Undefine.
18419 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
18420 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
18421 (CPP_OS_RTEMS_SPEC): Delete.
18422 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
18423 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
18425 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
18427 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18429 PR tree-optimization/81578
18430 * tree-parloops.c (build_new_reduction): Bail out if
18431 reduction_code isn't one of the standard OpenMP reductions.
18432 Move the details printing after that decision.
18434 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18436 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
18437 related to reload_in_progress.
18438 (splat_input_operand): Likewise.
18439 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
18441 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
18443 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
18444 (TARGET_INSTANTIATE_DECLS): Likewise.
18445 (legitimate_indexed_address_p): Delete reload_in_progress code.
18446 (rs6000_debug_legitimate_address_p): Likewise.
18447 (rs6000_eliminate_indexed_memrefs): Likewise.
18448 (rs6000_emit_le_vsx_store): Likewise.
18449 (rs6000_emit_move_si_sf_subreg): Likewise.
18450 (rs6000_emit_move): Likewise.
18451 (register_to_reg_type): Likewise.
18452 (rs6000_pre_atomic_barrier): Likewise.
18453 (rs6000_machopic_legitimize_pic_address): Likewise.
18454 (rs6000_allocate_stack_temp): Likewise.
18455 (rs6000_address_for_fpconvert): Likewise.
18456 (rs6000_address_for_altivec): Likewise.
18457 (rs6000_secondary_memory_needed_rtx): Delete function.
18458 (rs6000_check_sdmode): Likewise.
18459 (rs6000_alloc_sdmode_stack_slot): Likewise.
18460 (rs6000_instantiate_decls): Likewise.
18461 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
18462 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
18463 Delete reload_in_progress.
18464 (*vec_reload_and_plus_<mptrsize>): Likewise.
18465 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
18466 (vsx_div_v2di): Likewise.
18467 (vsx_udiv_v2di): Likewise.
18469 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18471 * config/rs6000/rs6000.opt (mlra): Replace with stub.
18472 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
18473 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
18474 (rs6000_debug_reg_global): Delete print of LRA status.
18475 (rs6000_option_override_internal): Delete dead LRA related code.
18476 (rs6000_lra_p): Delete function.
18477 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
18479 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
18481 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
18482 * config/riscv/rtems.h: New file.
18484 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18485 Sudakshina Das <sudi.das@arm.com>
18487 * config/aarch64/aarch64.md
18488 (define_split for and<mode>3nr_compare): Move
18489 non aarch64_logical_operand to a register.
18490 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
18491 register immediate operand to a register.
18492 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
18494 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18496 PR middle-end/81564
18497 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
18499 2017-07-27 Richard Biener <rguenther@suse.de>
18501 PR tree-optimization/81573
18502 PR tree-optimization/81494
18503 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
18504 multi defuse cycle case.
18506 2017-07-27 Richard Biener <rguenther@suse.de>
18508 PR tree-optimization/81571
18509 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
18512 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
18514 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
18515 earlier and only if MASK_FPU is set. Adjust formatting.
18517 2017-07-27 Martin Liska <mliska@suse.cz>
18519 * opt-functions.awk: Add validation of value of Init.
18520 * optc-gen.awk: Pass new argument.
18522 2017-07-27 Martin Liska <mliska@suse.cz>
18524 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
18525 Fix wrong condition.
18527 2017-07-27 Martin Liska <mliska@suse.cz>
18529 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
18530 BBs and edges seen by autoFDO.
18532 2017-07-27 Richard Biener <rguenther@suse.de>
18534 PR tree-optimization/81502
18535 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
18536 with incompatible but same sized type.
18537 (execute_update_addresses_taken): Likewise.
18539 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
18541 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
18542 flag_tree_loop_vectorize rather than flag_tree_vectorize.
18544 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18547 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
18548 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
18549 Change s_operand to memory_operand.
18551 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
18553 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
18554 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
18555 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
18556 Emit instructions rather than returning an expression. Handle TFmode
18557 and KFmode by casting to TImode.
18558 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
18559 (rs6000_emit_le_vsx_store): Likewise.
18560 * config/rs6000/vsx.md (VSX_TI): New iterator.
18561 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
18562 (*vsx_le_undo_permute_<mode>): Likewise.
18563 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
18564 emit the split sequence.
18565 (*vsx_le_perm_store_<mode>): Likewise.
18567 2017-07-27 Jakub Jelinek <jakub@redhat.com>
18569 PR tree-optimization/81555
18570 PR tree-optimization/81556
18571 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
18572 if true, force CHANGED for the recursive invocation.
18573 (reassociate_bb): Remember original length of ops array, pass
18574 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
18576 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
18577 attributes for noipa attribute. For naked attribute use
18578 lookup_attribute first before lookup_attribute_spec.
18579 * final.c (rest_of_handle_final): Disable IPA RA for functions with
18581 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
18582 for functions with noipa attribute.
18583 (cgraph_externally_visible_p): Return true for functions with noipa
18585 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
18586 for functions with noipa attribute.
18587 * doc/extend.texi: Document noipa function attribute.
18588 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
18589 also for functions with noipa attribute.
18590 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
18592 2017-07-26 Andrew Pinski <apinski@cavium.com>
18594 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
18595 vec_unalign_load_cost and vec_unalign_store_cost.
18597 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
18599 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
18600 -mvsx-small-integer option.
18601 (ISA_3_0_MASKS_IEEE): Likewise.
18602 (OTHER_VSX_VECTOR_MASKS): Likewise.
18603 (POWERPC_MASKS): Likewise.
18604 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
18605 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
18606 code, only testing for DImode being allowed in non-VSX floating
18608 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
18609 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
18611 (rs6000_option_override_internal): Delete -mvsx-small-integer.
18612 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
18613 TARGET_P8_VECTOR test.
18614 (rs6000_secondary_reload_simple_move): Likewise.
18615 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
18616 since TARGET_P9_VECTOR was already tested.
18617 (rs6000_opt_masks): Remove -mvsx-small-integer.
18618 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
18619 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18621 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
18622 test for TARGET_VEXTRACTUB was used, and that uses
18624 (p9 extract splitter): Likewise.
18625 (vsx_extract_<mode>_di_p9): Likewise.
18626 (vsx_extract_<mode>_store_p9): Likewise.
18627 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
18628 for TARGET_P9_VECTOR was used. Delete code that is now dead with
18629 the elimination of TARGET_VSX_SMALL_INTEGER.
18630 (vsx_extract_<mode>_p8): Likewise.
18631 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
18632 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
18633 (vsx_set_<mode>_p9): Likewise.
18634 (vsx_set_v4sf_p9): Likewise.
18635 (vsx_set_v4sf_p9_zero): Likewise.
18636 (vsx_insert_extract_v4sf_p9): Likewise.
18637 (vsx_insert_extract_v4sf_p9_2): Likewise.
18638 * config/rs6000/rs6000.md (sign extend splitter): Change
18639 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
18640 (floatsi<mode>2_lfiwax_mem): Likewise.
18641 (floatunssi<mode>2_lfiwzx_mem): Likewise.
18642 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
18643 since a test for TARGET_P9_VECTOR was used.
18644 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18645 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
18646 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18647 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18648 TARGET_P8_VECTOR test.
18649 (fix_trunc<mode>si2_stfiwx): Likewise.
18650 (fix_trunc<mode>si2_internal): Likewise.
18651 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
18652 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18654 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18655 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18656 TARGET_P8_VECTOR test.
18657 (fixuns_trunc<mode>si2_stfiwx): Likewise.
18658 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
18659 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18661 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18662 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
18663 since a test for TARGET_P9_VECTOR was used.
18664 (splitter for loading small constants): Likewise.
18666 2017-07-26 Andrew Pinski <apinski@cavium.com>
18668 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
18671 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
18674 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
18675 (fp_valid_at): Likewise.
18677 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18679 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
18680 (qdf24xx_addrcost_table): Likewise.
18681 (cortexa57_tunings): Update to use generic_branch_cost.
18682 (cortexa72_tunings): Likewise.
18683 (cortexa73_tunings): Likewise.
18684 (qdf24xx_tunings): Likewise.
18686 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18688 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
18689 (thunderx2t99_branch_cost): Likewise.
18690 (cortexa35_tunings): Update to use generic_branch_cost.
18691 (cortexa53_tunings): Likewise.
18692 (cortexa57_tunings): Likewise.
18693 (cortexa72_tunings): Likewise.
18694 (cortexa73_tunings): Likewise.
18695 (thunderx2t99_tunings): Likewise.
18697 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
18699 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
18700 (sparc_option_override): Honour MASK_FSMULD.
18701 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
18702 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
18703 * config/sparc/sparc.opt (mfsmuld): New option.
18704 * doc/invoke.texi (mfsmuld): Document option.
18706 2017-07-26 Marek Polacek <polacek@redhat.com>
18708 PR middle-end/70992
18709 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
18711 2017-07-26 Richard Biener <rguenther@suse.de>
18713 * gimple-match-head.c (do_valueize): Return OP if valueize
18715 (get_def): New helper to get at the def stmt of a SSA name
18716 if valueize allows.
18717 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
18718 do_valueize to get at the def stmt.
18719 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
18721 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
18723 PR middle-end/46932
18724 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
18726 2017-07-26 Martin Liska <mliska@suse.cz>
18729 * function.c (expand_function_start): Make expansion of
18730 nonlocal_goto_save_area after parm_birth_insn.
18732 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
18734 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
18735 from all CPU target flags enable members.
18737 2017-07-26 Richard Biener <rguenther@suse.de>
18739 * genmatch.c (dt_simplify::gen): Make iterator vars const.
18740 (decision_tree::gen): Make 'type' const.
18741 (write_predicate): Likewise.
18743 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
18745 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
18746 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
18747 (rs6000_option_override_internal): Likewise.
18748 (rs6000_expand_vector_set): Likewise.
18749 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
18750 (TARGET_UPPER_REGS_SF): Likewise.
18751 (TARGET_UPPER_REGS_DI): Likewise.
18752 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
18753 (TARGET_DIRECT_MOVE_64BIT): Likewise.
18754 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
18755 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18756 (Splitters for DI constants in Altivec registers): Likewise.
18757 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
18758 (vsx_set_v4sf_p9): Likewise.
18759 (vsx_set_v4sf_p9_zero): Likewise.
18760 (vsx_insert_extract_v4sf_p9): Likewise.
18761 (vsx_insert_extract_v4sf_p9_2): Likewise.
18763 2017-07-25 Carl Love <cel@us.ibm.com>
18765 * doc/extend.texi: Update the built-in documentation file for the
18766 existing built-in functions
18767 vector signed char vec_cnttz (vector signed char);
18768 vector unsigned char vec_cnttz (vector unsigned char);
18769 vector signed short vec_cnttz (vector signed short);
18770 vector unsigned short vec_cnttz (vector unsigned short);
18771 vector signed int vec_cnttz (vector signed int);
18772 vector unsigned int vec_cnttz (vector unsigned int);
18773 vector signed long long vec_cnttz (vector signed long long);
18774 vector unsigned long long vec_cnttz (vector unsigned long long);
18776 2017-07-25 Andrew Pinski <apinski@cavium.com>
18778 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
18779 accesses where the use is for the first operand of a BIT_INSERT.
18781 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18784 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18785 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18787 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18789 * config/i386/gstabs.h: Delete.
18790 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
18792 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
18794 * config/i386/i386.c (ix86_decompose_address): Do not check for
18795 register RTX when looking at index_reg or base_reg.
18796 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
18798 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
18800 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
18801 to update EH info here.
18803 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18805 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
18807 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18809 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
18811 2017-07-25 Torsten Duwe <duwe@suse.de>
18813 * common.opt: Introduce -fpatchable-function-entry
18814 command line option, and its variables function_entry_patch_area_size
18815 and function_entry_patch_area_start.
18816 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
18817 including a two-value parser.
18818 * target.def (print_patchable_function_entry): New target hook.
18819 * targhooks.h (default_print_patchable_function_entry): New function.
18820 * targhooks.c (default_print_patchable_function_entry): Likewise.
18821 * toplev.c (process_options): Switch off IPA-RA if
18822 patchable function entries are being generated.
18823 * varasm.c (assemble_start_function): Look at the
18824 patchable-function-entry command line switch and current
18825 function attributes and maybe generate NOP instructions by
18826 calling the print_patchable_function_entry hook.
18827 * doc/extend.texi: Document patchable_function_entry attribute.
18828 * doc/invoke.texi: Document -fpatchable_function_entry
18829 command line option.
18830 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
18832 * doc/tm.texi: Re-generate.
18834 2017-07-25 Jakub Jelinek <jakub@redhat.com>
18837 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
18838 TARGET_AVX512DQ rather than TARGET_AVX512BW.
18840 2017-07-25 Tamar Christina <tamar.christina@arm.com>
18842 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
18844 2017-07-25 Richard Biener <rguenther@suse.de>
18846 PR tree-optimization/81455
18847 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
18848 not walk in cycles when looking for guards.
18850 2017-07-25 Richard Biener <rguenther@suse.de>
18852 PR tree-optimization/81529
18853 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
18854 when optimizing backedge uses.
18856 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
18858 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
18860 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
18861 to dl_section_ref. On AIX, append an expression to subtract
18862 the size of the section length to dl_section_ref.
18864 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
18866 * configure.ac: If any of the config.* scripts fail, exit 1.
18867 * configure: Regenerate.
18869 2017-07-25 Richard Biener <rguenther@suse.de>
18871 PR middle-end/81546
18872 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
18873 of immediate uses, be more verbose on errors.
18875 2017-07-25 Richard Biener <rguenther@suse.de>
18877 PR tree-optimization/81510
18878 * tree-vect-loop.c (vect_is_simple_reduction): When the
18879 reduction stmt is not inside the loop bail out.
18881 2017-07-25 Richard Biener <rguenther@suse.de>
18883 PR tree-optimization/81303
18884 * tree-vect-loop-manip.c (vect_loop_versioning): Build
18885 profitability check against LOOP_VINFO_NITERSM1.
18887 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18889 * domwalk.c (cmp_bb_postorder): Simplify.
18890 (sort_bbs_postorder): New function. Use it...
18891 (dom_walker::walk): ...here to optimize common cases.
18893 2017-07-25 Martin Liska <mliska@suse.cz>
18896 * ipa-visibility.c (function_and_variable_visibility): Make the
18897 redirection just on target that supports aliasing.
18898 Fix GNU coding style.
18900 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
18903 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
18905 * config/arm/rtems.h: Likewise.
18906 * config/bfin/rtems.h: Likewise.
18907 * config/i386/rtemself.h: Likewise.
18908 * config/lm32/rtems.h: Likewise.
18909 * config/m32c/rtems.h: Likewise.
18910 * config/m68k/rtemself.h: Likewise.
18911 * config/microblaze/rtems.h: Likewise.
18912 * config/mips/rtems.h: Likewise.
18913 * config/moxie/rtems.h: Likewise.
18914 * config/nios2/rtems.h: Likewise.
18915 * config/powerpcspe/rtems.h: Likewise.
18916 * config/rs6000/rtems.h: Likewise.
18917 * config/rtems.h: Likewise.
18918 * config/sh/rtems.h: Likewise.
18919 * config/sh/rtemself.h: Likewise.
18920 * config/sparc/rtemself.h: Likewise.
18922 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
18925 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
18926 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
18927 * tree-ssa-structalias.c (alias_get_name): Same.
18929 2017-07-25 Bin Cheng <bin.cheng@arm.com>
18932 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
18933 instructions if no du chain is found.
18935 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
18937 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
18939 2017-07-25 Richard Biener <rguenther@suse.de>
18941 PR middle-end/81505
18942 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
18945 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
18947 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
18948 upper-regs options.
18949 (ISA_2_7_MASKS_SERVER): Likewise.
18950 (ISA_3_0_MASKS_IEEE): Likewise.
18951 (OTHER_P8_VECTOR_MASKS): Likewise.
18952 (OTHER_VSX_VECTOR_MASKS): Likewise.
18953 (POWERPC_MASKS): Likewise.
18954 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
18955 duplicate list of options.
18956 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
18957 explicit -mupper-regs options.
18958 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
18959 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
18960 alias for -mupper-regs-df.
18961 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18962 (rs6000_init_hard_regno_mode_ok): Likewise.
18963 (rs6000_option_override_internal): Likewise.
18964 (rs6000_opt_masks): Likewise.
18965 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
18966 options in terms of whether -mvsx or -mpower8-vector was used.
18967 (TARGET_UPPER_REGS_DI): Likewise.
18968 (TARGET_UPPER_REGS_SF): Likewise.
18969 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
18970 -mupper-regs-* options.
18972 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
18974 * passes.c (emergency_dump_function): Print some empty lines and a
18975 header before the RTL dump.
18977 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
18979 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
18981 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
18984 * config/aarch64/aarch64.c (aarch64_classify_symbol):
18985 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
18987 2017-07-24 Carl Love <cel@us.ibm.com>
18989 * config/rs6000/rs6000-c.c: Add support for built-in functions
18990 vector float vec_extract_fp32_from_shorth (vector unsigned short);
18991 vector float vec_extract_fp32_from_shortl (vector unsigned short);
18992 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
18993 vec_extract_fp_from_shortl): Add defines for the two builtins.
18994 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
18995 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
18997 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
18998 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
18999 * doc/extend.texi: Update the built-in documentation file for the
19000 new built-in function.
19002 2017-07-24 Jakub Jelinek <jakub@redhat.com>
19005 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
19007 * doc/generic.texi: Likewise.
19008 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19009 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19011 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
19013 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
19014 (aarch64_mls_elt_merge<mode>): Likewise.
19016 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
19018 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
19019 having __cxa_atexit.
19021 2017-07-23 Michael Collison <michael.collison@arm.com>
19023 * config/arm/arm.c (arm_option_override): Deprecate
19024 use of -mstructure-size-boundary.
19025 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
19026 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
19028 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19031 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19032 Reduce cost estimate for direct moves.
19034 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
19037 * config/i386/i386.c (ix86_option_override_internal): Disable
19038 BMI, BMI2 and TBM instructions for -m16.
19040 2017-07-21 Carl Love <cel@us.ibm.com>
19042 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19043 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19044 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19045 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19046 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19047 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19048 VMULOSW): New enum "unspec" values.
19049 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19050 altivec_vmulosw): New patterns.
19051 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19052 VMULOSW): Add definitions.
19054 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
19056 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
19057 (qdf24xx): Likewise.
19058 * config/aarch64/aarch64-options-extensions.def (rdma); New.
19059 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
19060 (AARCH64_FL_V8_1): Renumber.
19061 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
19062 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
19063 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
19064 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
19065 rdma to feature modifiers list.
19067 2017-07-21 Yury Gribov <tetra2005@gmail.com>
19069 PR middle-end/56727
19070 * ipa-visibility (function_and_variable_visibility): Convert
19071 recursive PLT call to direct call if appropriate.
19073 2017-07-21 Andrew Pinski <apinski@cavium.com>
19075 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
19076 operand 1 to see if the types precision matches.
19077 * fold-const.c (operand_equal_p): Likewise.
19079 2017-07-21 Richard Biener <rguenther@suse.de>
19081 PR tree-optimization/81303
19082 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
19083 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
19084 (vect_peeling_hash_get_lowest_cost): Adjust.
19085 (vect_enhance_data_refs_alignment): Likewise. Use
19086 vect_get_peeling_costs_all_drs to compute the penalty for no
19087 peeling to match up costs.
19089 2017-07-21 Richard Biener <rguenther@suse.de>
19091 PR tree-optimization/81500
19092 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
19093 we didn't identify a reduction path.
19095 2017-07-21 Tom de Vries <tom@codesourcery.com>
19096 Cesar Philippidis <cesar@codesourcery.com>
19098 PR gcov-profile/81442
19099 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
19102 2017-07-21 Tom de Vries <tom@codesourcery.com>
19105 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
19107 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
19108 nvptx_override_options_after_change.
19110 2017-07-21 Ulrich Drepper <drepper@redhat.com>
19112 * dwarf2out.c (output_file_names): Avoid double testing for
19113 dwarf_version >= 5.
19115 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
19117 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
19119 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19121 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
19123 (try_crossjump_to_edge): Do not punt on partitioned functions.
19125 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19127 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
19128 Put all BBs reachable only via paths crossing cold region to cold
19130 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
19132 2016-07-21 Richard Biener <rguenther@suse.de>
19134 PR tree-optimization/81303
19135 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
19136 into account prologue and epilogue iterations when raising
19137 min_profitable_iters to sth at least covering one vector iteration.
19139 2017-07-21 Tamar Christina <tamar.christina@arm.com>
19141 * config/arm/arm.c (arm_test_cpu_arch_dat):
19144 2017-07-20 Nathan Sidwell <nathan@acm.org>
19146 Remove TYPE_METHODS.
19147 * tree.h (TYPE_METHODS): Delete.
19148 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
19149 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
19150 (dbxout_type_methods): Scan TYPE_FIELDS.
19151 (dbxout_type): Don't check TYPE_METHODS here.
19152 * function.c (use_register_for_decl): Always ignore register for
19153 class types when not optimizing.
19154 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
19155 * tree.c (free_lang_data_in_type): Stitch out member functions and
19156 templates from TYPE_FIELDS.
19157 (build_distinct_type_copy, verify_type_variant,
19158 verify_type): Member fns are on TYPE_FIELDS.
19159 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
19160 * tree-pretty-print.c (dump_generic_node): Likewise.
19162 2017-07-20 Jakub Jelinek <jakub@redhat.com>
19165 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
19166 V2TImode and V4TImode.
19167 (ix86_expand_vector_extract): Likewise.
19168 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
19169 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
19170 (ssescalarmode): Handle V4TImode and V2TImode.
19171 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
19172 (*vec_extractv2ti, *vec_extractv4ti): New insns.
19173 (VEXTRACTI128_MODE): New mode iterator.
19174 (splitter for *vec_extractv?ti first element): New.
19175 (VEC_INIT_MODE): New mode iterator.
19176 (vec_init<mode>): Consolidate 3 expanders into one using
19177 VEC_INIT_MODE mode iterator.
19179 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
19181 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
19182 non_spilled_static_chain_regno_p.
19184 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
19186 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
19188 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
19190 * bb-reorder.c (connect_traces): Allow copying of blocks within
19193 2017-07-20 Richard Biener <rguenther@suse.de>
19195 * gimple.h (gimple_phi_result): Add gphi * overload.
19196 (gimple_phi_result_ptr): Likewise.
19197 (gimple_phi_arg): Likewise. Adjust index assert to only
19198 allow actual argument accesses rather than all slots available
19200 (gimple_phi_arg_def): Add gphi * overload.
19201 * tree-phinodes.c (make_phi_node): Initialize only actual
19203 (resize_phi_node): Clear memory not covered by old node,
19204 do not initialize excess argument slots.
19205 (reserve_phi_args_for_new_edge): Initialize new argument slot
19208 2017-07-20 Bin Cheng <bin.cheng@arm.com>
19210 PR tree-optimization/81388
19212 2016-07-21 Bin Cheng <bin.cheng@arm.com>
19214 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
19215 by removing computation of may_be_zero.
19217 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19218 Tom de Vries <tom@codesourcery.com>
19220 PR middle-end/81030
19221 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
19222 when gimple level profile disagrees with what RTL expander did.
19224 2017-07-20 Richard Biener <rguenther@suse.de>
19226 PR tree-optimization/61171
19227 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
19228 (vect_analyze_stmt): Add slp instance parameter.
19229 (vectorizable_reduction): Likewise.
19230 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
19231 (vect_is_simple_reduction): Deal with chains not detected
19232 as SLP reduction chain, specifically not properly associated
19233 chains containing a mix of plus/minus.
19234 (get_reduction_op): Remove.
19235 (get_initial_defs_for_reduction): Simplify, pass in whether
19236 this is a reduction chain, pass in the SLP node for the PHIs.
19237 (vect_create_epilog_for_reduction): Get the SLP instance as
19239 (vectorizable_reduction): Get the SLP instance as arg.
19240 During analysis remember the SLP node with the PHIs in the
19241 instance. Simplify getting at the vectorized reduction PHIs.
19242 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
19243 through SLP instance.
19244 (vect_slp_analyze_operations): Likewise.
19245 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
19246 (vect_transform_stmt): Likewise.
19248 2017-07-20 Tom de Vries <tom@codesourcery.com>
19250 PR tree-optimization/81489
19251 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
19252 read of phi arg location to before loop that modifies phi.
19254 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19256 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
19259 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19261 PR middle-end/81331
19262 * except.c (execute): Fix ordering issue.
19264 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
19266 PR rtl-optimization/81423
19267 * combine.c (make_compound_operation_int): Don't try to optimize
19268 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
19270 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
19272 PR rtl-optimization/81423
19273 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
19274 with a constant that is -1 in the truncated to mode.
19276 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19278 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
19279 (determine_unlikely_bbs): ... here.
19280 * predict.h (propagate_unlikely_bbs_forward): Declare.
19281 * cfgexpand.c (pass_expand::execute): Use it.
19282 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
19284 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
19285 propagate_unlikely_bbs_forward.
19287 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19289 PR middle-end/81331
19290 * except.c (maybe_add_nop_after_section_switch): New function.
19293 2017-07-19 Tom de Vries <tom@codesourcery.com>
19295 * gimple.h (gimple_phi_set_arg): Make assert more strict.
19297 2017-07-19 Tom de Vries <tom@codesourcery.com>
19299 * gimple.h (gimple_phi_arg): Make assert more strict.
19301 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
19303 * config.gcc (powerpc*-*-*): Add mmintrin.h.
19304 * config/rs6000/mmintrin.h: New file.
19305 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
19307 2017-07-19 Jakub Jelinek <jakub@redhat.com>
19309 PR tree-optimization/81346
19310 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
19312 2017-07-19 Tom de Vries <tom@codesourcery.com>
19314 * config/nvptx/nvptx.md (VECIM): Add V2DI.
19316 2017-07-19 Tom de Vries <tom@codesourcery.com>
19318 * config/nvptx/nvptx-modes.def: Add V2DImode.
19319 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
19320 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
19321 (nvptx_output_mov_insn): Handle lack of mov.b128.
19322 (nvptx_print_operand): Handle 'H' and 'L' codes.
19323 (nvptx_vector_mode_supported): Allow V2DImode.
19324 (nvptx_preferred_simd_mode): New function.
19325 (nvptx_data_alignment): New function.
19326 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
19327 nvptx_preferred_simd_mode.
19328 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
19330 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
19332 2017-07-19 Tom de Vries <tom@codesourcery.com>
19334 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
19335 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
19336 (nvptx_vector_mode_supported): New function. Allow V2SImode.
19337 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
19338 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
19339 (mov<VECIM>_insn): New define_insn.
19340 (define_expand "mov<VECIM>): New define_expand.
19342 2017-07-19 Tom de Vries <tom@codesourcery.com>
19344 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
19346 2017-07-19 Jakub Jelinek <jakub@redhat.com>
19348 PR tree-optimization/81346
19349 * fold-const.h (fold_div_compare, range_check_type): Declare.
19350 * fold-const.c (range_check_type): New function.
19351 (build_range_check): Use range_check_type.
19352 (fold_div_compare): No longer static, rewritten into
19353 a match.pd helper function.
19354 (fold_comparison): Don't call fold_div_compare here.
19355 * match.pd (X / C1 op C2): New optimization using fold_div_compare
19356 as helper function.
19358 2017-07-19 Nathan Sidwell <nathan@acm.org>
19360 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
19361 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
19362 * tree.c (find_decls_types_r, verify_type): Use
19363 TYPE_{MIN,MAX}_VALUE_RAW.
19364 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
19365 (hash_tree): Likewise.
19366 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19368 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19371 2017-07-18 Tom de Vries <tom@codesourcery.com>
19373 PR middle-end/81464
19374 * omp-expand.c (expand_omp_for_static_chunk): Handle
19375 equal-argument loop exit phi.
19377 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
19380 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
19381 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
19382 operand 2 predicate.
19383 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
19384 operand 2 predicate.
19385 (ror,rol -> rorx splitters): Use const_int_operand as
19386 operand 2 predicate.
19388 2017-06-18 Richard Biener <rguenther@suse.de>
19390 PR tree-optimization/81410
19391 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
19392 the gap in the ! slp_perm SLP case after each group.
19394 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19396 PR middle-end/81463
19397 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
19400 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19402 PR middle-end/81462
19403 * predict.c (set_even_probabilities): Cleanup; do not affect
19404 probabilities that are already known.
19405 (combine_predictions_for_bb): Call even when count is set.
19407 2017-07-18 Nathan Sidwell <nathan@acm.org>
19409 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
19412 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19415 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
19416 optimization for loop niter analysis.
19418 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
19421 * config/avr/avr.c (avr_optimize_casesi): Don't use
19422 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
19424 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
19426 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
19427 body_cost_vec from _vect_peel_extended_info.
19428 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
19429 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
19432 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19434 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
19436 2017-07-18 Richard Biener <rguenther@suse.de>
19438 PR tree-optimization/80620
19439 PR tree-optimization/81403
19440 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
19441 info when re-using a VN table entry.
19443 2017-07-18 Richard Biener <rguenther@suse.de>
19445 PR tree-optimization/81418
19446 * tree-vect-loop.c (vectorizable_reduction): Properly compute
19447 vectype_in. Verify that with lane-reducing reduction operations
19448 we have a single def-use cycle.
19450 2017-07-17 Carl Love <cel@us.ibm.com>
19452 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
19454 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19455 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19456 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19457 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19458 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19459 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19460 VMULOSW): New enum "unspec" values.
19461 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19462 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19463 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19464 altivec_vmulosw): New patterns.
19465 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19466 VMULOSW): Add definitions.
19467 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
19469 * config/alpha/alpha.c: Include predict.h.
19471 2017-07-17 Yury Gribov <tetra2005@gmail.com>
19473 * tree-vrp.c (compare_assert_loc): Fix comparison function
19474 to return predictable results.
19476 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19478 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
19480 (subdi3): Likewise.
19481 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
19482 * doc/invoke.texi (mexpand-adddi): Update text.
19484 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19486 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
19487 that also clobbers the CC register. The old expand code is moved
19489 (*arc_clzsi2): ... here.
19490 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
19491 the CC register. The old expand code is moved to ...
19492 (arc_ctzsi2): ... here.
19494 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19496 * config/arc/arc.opt (mindexed-loads): Use initial value
19497 TARGET_INDEXED_LOADS_DEFAULT.
19498 (mauto-modify-reg): Use initial value
19499 TARGET_AUTO_MODIFY_REG_DEFAULT.
19500 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19501 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19502 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19503 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19505 2017-07-17 Martin Liska <mliska@suse.cz>
19508 * opts.c (finish_options): Do not allow -fgnu-tm
19509 w/ -fsanitize={kernel-,}address. Say sorry.
19511 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19514 * tree-loop-distribution.c (classify_partition): Only assert on
19515 numer of iterations.
19516 (merge_dep_scc_partitions): Delete prameter. Update function call.
19517 (distribute_loop): Remove code handling loop with unknown niters.
19518 (pass_loop_distribution::execute): Skip loop with unknown niters.
19520 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19523 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
19524 function sort_partitions_by_post_order.
19526 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19528 PR tree-optimization/81374
19529 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
19530 the max index of basic blocks, rather than number of basic blocks.
19532 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19534 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
19536 (arc_legitimate_pic_operand_p): Likewise.
19537 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
19539 (arc_needs_pcl_p): Likewise.
19540 (arc_legitimate_pc_offset_p): Likewise.
19541 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
19542 function is also used in constrains.md.
19543 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
19544 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
19545 PLUS. Only return true/false in known cases, otherwise assert.
19546 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
19547 is already called in arc_legitimate_constant_p.
19548 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
19550 (LEGITIMATE_PIC_OPERAND_P): Use
19551 arc_raw_symbolic_reference_mentioned_p function.
19552 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
19557 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19558 Andrew Burgess <andrew.burgess@embecosm.com>
19560 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
19561 (arc_return_address_register): New function.
19562 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
19563 (arc_handle_fndecl_attribute): Add naked attribute.
19564 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19565 (TARGET_WARN_FUNC_RETURN): Likewise.
19566 (arc_allocate_stack_slots_for_args): New function.
19567 (arc_warn_func_return): Likewise.
19568 (machine_function): Change type fn_type.
19569 (arc_compute_function_type): Consider new naked function type,
19570 change function return type.
19571 (arc_must_save_register): Adapt to handle new
19572 arc_compute_function_type's return type.
19573 (arc_expand_prologue): Likewise.
19574 (arc_expand_epilogue): Likewise.
19575 (arc_return_address_regs): Delete.
19576 (arc_return_address_register): New function.
19577 (arc_epilogue_uses): Use above function.
19578 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
19579 (arc_function_type): Change encoding, add naked type.
19580 (ARC_INTERRUPT_P): Change to handle the new encoding.
19581 (ARC_FAST_INTERRUPT_P): Likewise.
19582 (ARC_NORMAL_P): Define.
19583 (ARC_NAKED_P): Likewise.
19584 (arc_compute_function_type): Delete prototype.
19585 * config/arc/arc.md (in_ret_delay_slot): Use
19586 arc_return_address_register function.
19587 (simple_return): Likewise.
19588 (p_return_i): Likewise.
19590 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19592 PR tree-optimization/81428
19593 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
19594 can't be built for those types.
19596 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19598 Remove stuff dead since r239246.
19600 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
19601 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
19602 (avr_inform_devices): Remove dead stuff.
19604 2017-07-17 Tamar Christina <tamar.christina@arm.com>
19606 * config/arm/arm_neon.h: Fix softp typo.
19608 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19610 PR tree-optimization/81365
19611 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
19612 aggregate moves onto bb predecessor edges, make sure there are no
19613 loads that could alias the lhs in between the start of bb and the
19616 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19619 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
19620 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
19621 [LSHIFTRT, outer_code = TRUNCATE]: Same.
19623 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19625 PR tree-optimization/81396
19626 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
19627 (init_symbolic_number): Initialize it to 1.
19628 (perform_symbolic_merge): Add n_ops from both operands into the new
19630 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
19631 without base_addr as useless if they need more than one operation.
19632 (bswap_replace): Handle !bswap case for NULL base_addr.
19634 2017-07-17 Tom de Vries <tom@codesourcery.com>
19637 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
19640 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
19642 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
19643 conditional builtin define __FIX_LEON3FT_B2BST.
19645 2017-07-17 Daniel Cederman <cederman@gaisler.com>
19647 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
19648 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
19651 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
19653 PR rtl-optimization/81424
19654 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
19655 to remove potential trapping from operands if -fnon-call-exceptions.
19657 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19659 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
19660 profile_proability for scalling.
19661 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
19663 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19665 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
19667 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19669 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
19670 fixpoint arithmetics.
19672 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19674 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
19675 fixpoint arithmetics.
19677 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19679 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
19680 fixpoint arithmetics.
19682 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19684 * profile-count.h (profile_probability::from_reg_br_prob_note,
19685 profile_probability::to_reg_br_prob_note): New functions.
19686 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
19687 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
19688 * predict.c (probability_reliable_p): Update.
19689 (edge_probability_reliable_p): Update.
19690 (br_prob_note_reliable_p): Update.
19691 (invert_br_probabilities): Update.
19692 (add_reg_br_prob_note): New function.
19693 (combine_predictions_for_insn): Update.
19694 * asan.c (asan_clear_shadow): Update.
19695 * cfgbuild.c (compute_outgoing_frequencies): Update.
19696 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
19697 (update_br_prob_note): Update.
19698 (rtl_verify_edges): Update.
19699 (purge_dead_edges): Update.
19700 (fixup_reorder_chain): Update.
19701 * emit-rtl.c (try_split): Update.
19702 * ifcvt.c (cond_exec_process_insns): Update.
19703 (cond_exec_process_if_block): Update.
19704 (dead_or_predicable): Update.
19705 * internal-fn.c (expand_addsub_overflow): Update.
19706 (expand_neg_overflow): Update.
19707 (expand_mul_overflow): Update.
19708 * loop-doloop.c (doloop_modify): Update.
19709 * loop-unroll.c (compare_and_jump_seq): Update.
19710 * optabs.c (emit_cmp_and_jump_insn_1): Update.
19711 * predict.h: Update.
19712 * reorg.c (mostly_true_jump): Update.
19714 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
19715 * config/alpha/alpha.c (emit_unlikely_jump): Update.
19716 * config/arc/arc.c: (emit_unlikely_jump): Update.
19717 * config/arm/arm.c: (emit_unlikely_jump): Update.
19718 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
19719 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
19720 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
19721 (ix86_print_operand): Update.
19722 (ix86_split_fp_branch): Update.
19723 (predict_jump): Update.
19724 * config/ia64/ia64.c (ia64_print_operand): Update.
19725 * config/mmix/mmix.c (mmix_print_operand): Update.
19726 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
19727 (rs6000_expand_split_stack_prologue): Update.
19728 * config/rs6000/rs6000.c: Update.
19729 * config/s390/s390.c (s390_expand_vec_strlen): Update.
19730 (s390_expand_vec_movstr): Update.
19731 (s390_expand_cs_tdsi): Update.
19732 (s390_expand_split_stack_prologue): Update.
19733 * config/sh/sh.c (sh_print_operand): Update.
19734 (expand_cbranchsi4): Update.
19735 (expand_cbranchdi4): Update.
19736 * config/sparc/sparc.c (output_v9branch): Update.
19737 * config/spu/spu.c (get_branch_target): Update.
19738 (ea_load_store_inline): Update.
19739 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
19740 * config/tilepro/tilepro.c: Update.
19742 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
19744 * gimplify.c (mostly_copy_tree_r): Revert latest change.
19745 (gimplify_save_expr): Likewise.
19747 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19749 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
19751 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19753 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
19755 * timevar.def (TV_IPA_FNSUMMARY): Define.
19757 2017-07-16 Daniel Cederman <cederman@gaisler.com>
19759 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
19760 to back store errata sensitive sequence from being generated.
19761 (sqrtdf2_fix): Likewise.
19763 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19765 * tree-ssa-threadupdate.c (compute_path_counts,
19766 update_joiner_offpath_counts): Use profile_probability.
19768 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
19771 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19773 * config/arm/arm-c.c (arm_cpu_builtins): Define
19774 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19776 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
19778 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19779 array entries to represent __ieee128 versions of the
19780 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
19781 scalar_extract_sig, and scalar_insert_exp built-in functions.
19782 (altivec_resolve_overloaded_builtin): Add special case handling
19783 for the __builtin_scalar_insert_exp function, as represented by
19784 the P9V_BUILTIN_VEC_VSIEDP constant.
19785 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
19786 exponent support for __ieee128 argument.
19787 (VSESQP): Add scalar extract signature support for __ieee128
19789 (VSTDCNQP): Add scalar test negative support for __ieee128
19791 (VSIEQP): Add scalar insert exponent support for __int128 argument
19792 with __ieee128 result.
19793 (VSIEQPF): Add scalar insert exponent support for __ieee128
19794 argument with __ieee128 result.
19795 (VSTDCQP): Add scalar test data class support for __ieee128
19797 (VSTDCNQP): Add overload support for scalar test negative with
19798 __ieee128 argument.
19799 (VSTDCQP): Add overload support for scalar test data class
19800 __ieee128 argument.
19801 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
19802 UNSPEC_VSX_SXSIGDP.
19803 (UNSPEC_VSX_SIEXPQP): New constant.
19804 (xsxexpqp): New insn for VSX scalar extract exponent quad
19806 (xsxsigqp): New insn for VSX scalar extract significand quad
19808 (xsiexpqpf): New insn for VSX scalar insert exponent quad
19809 precision with floating point argument.
19810 (xststdcqp): New expand for VSX scalar test data class quad
19812 (xststdcnegqp): New expand for VSX scalar test negative quad
19814 (xststdcqp): New insn to match expansions for VSX scalar test data
19815 class quad precision and VSX scalar test negative quad precision.
19816 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
19817 special case operand checking to enforce that second operand of
19818 VSX scalar test data class with quad precision argument is a 7-bit
19820 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19821 prototypes and descriptions of __ieee128 versions of
19822 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
19823 scalar_test_data_class, and scalar_test_neg built-in functions.
19825 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19827 PR tree-optimization/81162
19828 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
19829 replace a negate with an add.
19831 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
19833 * doc/invoke.texi (arm/-mcpu): Document +crypto.
19835 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19837 * config/arm/arm-c.c (arm_cpu_builtins): Define
19838 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19840 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19842 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
19843 (armv8-r): Set ARM Cortex-R52 as default CPU.
19844 * config/arm/arm-tables.opt: Regenerate.
19845 * config/arm/arm-tune.md: Regenerate.
19846 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19848 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
19849 extension for -mcpu=cortex-r52.
19851 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19853 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
19854 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
19855 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
19856 (fp-armv8): Define it as FP_ARMv8 only.
19857 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
19858 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
19860 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
19861 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
19862 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
19863 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
19864 than TARGET_FPU_ARMV8.
19865 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
19866 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
19867 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
19869 * config/arm/neon.md (neon_vrint): Likewise.
19870 (neon_vcvt): Likewise.
19871 (neon_<fmaxmin_op><mode>): Likewise.
19872 (<fmaxmin><mode>3): Likewise.
19873 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
19874 * config/arm/predicates.md (arm_cond_move_operator): Check against
19875 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
19877 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
19879 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
19880 to top of function.
19882 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19884 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
19885 loop in comment with memset.
19887 2017-07-14 Martin Liska <mliska@suse.cz>
19889 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
19890 * dwarf2out.c (is_java): Remove the function.
19891 (output_pubname): Remove usage of the function.
19892 (lower_bound_default): Remove usage of DW_LANG_Java.
19893 (gen_compile_unit_die): Likewise.
19894 * gcc.c: Remove compiler defaults for .java and .zip files.
19895 * gimple-expr.c (remove_suffix): Change as there's no longer
19896 extension than 4-letter one.
19897 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
19898 (gimplify_save_expr): Likewise.
19899 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
19900 as it's possible even for other languages than Java.
19901 * langhooks.h (struct lang_hooks): Remove Java from a comment.
19902 * lto-opts.c (lto_write_options): Remove reference to Java.
19903 * opts.c (strip_off_ending): Update file extension handling.
19904 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
19905 * tree-eh.c (lower_resx): Likewise.
19906 * tree.c (free_lang_data_in_type): Remove dead code.
19907 (find_decls_types_r): Likewise.
19908 (build_common_builtin_nodes): Remove Java from a comment.
19909 (verify_type): Remove dead code.
19910 * varasm.c (assemble_external): Remove Java from a comment.
19912 2017-07-14 Martin Liska <mliska@suse.cz>
19914 * opts.c (finish_options): Add quotes.
19915 (common_handle_option): Likewise.
19917 2017-07-14 Martin Liska <mliska@suse.cz>
19919 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
19920 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
19921 Remove N_SO_PASCAL.
19922 * dwarf2out.c (lower_bound_default): Do not handle
19924 (gen_compile_unit_die): Likewise.
19925 * gcc.c: Remove default extension binding for GNU Pascal.
19926 * stmt.c: Remove Pascal language from a comment.
19927 * xcoffout.c: Likewise.
19929 2017-07-13 David Malcolm <dmalcolm@redhat.com>
19932 * diagnostic-show-locus.c (fixit_cmp): New function.
19933 (layout::layout): Sort m_fixit_hints.
19934 (column_range::column_range): Assert that the values are valid.
19935 (struct char_span): New struct.
19936 (correction::overwrite): New method.
19937 (struct source_line): New struct.
19938 (line_corrections::add_hint): Add assertions. Reimplement memcpy
19939 calls in terms of classes source_line and char_span, and
19940 correction::overwrite.
19941 (selftest::test_overlapped_fixit_printing_2): New function.
19942 (selftest::diagnostic_show_locus_c_tests): Call it.
19944 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
19946 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
19947 early if there is no lhs.
19949 2017-07-13 Martin Liska <mliska@suse.cz>
19951 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
19952 (gen_reference_type_die): Likewise.
19953 * stor-layout.c: Remove Pascal-related comment.
19955 2017-07-13 Martin Liska <mliska@suse.cz>
19957 * opts.c (finish_options): Add quotes to error messages.
19958 (parse_sanitizer_options): Likewise.
19960 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
19962 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
19964 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
19966 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
19968 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
19970 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
19972 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
19974 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
19977 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
19978 provides the hardware capability bits, define the macro
19979 __BUILTIN_CPU_SUPPORTS__.
19980 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
19981 if GLIBC does not provide the hardware capability bits. Add a
19982 gcc_unreachable call if the built-in cpu function is neither
19983 __builtin_cpu_is nor __builtin_cpu_supports.
19984 (rs6000_get_function_versions_dispatcher): Change the warning
19985 that an old GLIBC is used which does not export the capability
19986 bits to be an error.
19987 * doc/extend.texi (target_clones attribute): Document the
19988 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
19989 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
19990 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
19991 the macros defined by GCC if the newer GLIBC is available.
19993 2017-07-12 Jeff Law <law@redhat.com>
19995 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
19996 remaining includes slightly.
19997 * config/riscv/riscv-builtins.c: Include profile-count.h.
19999 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20002 * config/avr/avr.c (avr_set_current_function): In diagnostic
20003 messages: Quote keywords and (parts of) identifiers.
20004 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
20007 2017-07-12 Carl Love <cel@us.ibm.com>
20009 * config/rs6000/rs6000-c.c: Add support for built-in functions
20010 vector bool char vec_revb (vector bool char);
20011 vector bool short vec_revb (vector short char);
20012 vector bool int vec_revb (vector bool int);
20013 vector bool long long vec_revb (vector bool long long);
20014 * doc/extend.texi: Update the built-in documentation file for the
20015 new built-in functions.
20017 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20019 * config/s390/s390.md: Remove movcc splitter.
20021 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20023 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
20024 load/store on condition.
20026 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20029 * config/avr/avr.c (avr_encode_section_info)
20030 [progmem && !TREE_READONLY]: Error if progmem object needs
20033 2017-07-11 Michael Collison <michael.collison@arm.com>
20035 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
20038 2017-07-11 Carl Love <cel@us.ibm.com>
20040 * config/rs6000/rs6000-c.c: Add support for builtins
20041 vector unsigned int vec_parity_lsbb (vector signed int);
20042 vector unsigned int vec_parity_lsbb (vector unsigned int);
20043 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
20044 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
20045 vector unsigned long long vec_parity_lsbb (vector signed long long);
20046 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
20047 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
20048 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
20049 * doc/extend.texi: Update the built-in documentation file for the
20050 new built-in functions.
20052 2017-07-11 David Malcolm <dmalcolm@redhat.com>
20054 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
20055 (layout::m_primary_loc): New field.
20056 (layout::layout): Initialize new field. Move location filtering
20057 logic from here to...
20058 (layout::maybe_add_location_range): ...this new method. Add
20059 support for filtering to just the lines already specified by other
20061 (layout::will_show_line_p): New method.
20062 (gcc_rich_location::add_location_if_nearby): New method.
20063 (selftest::test_add_location_if_nearby): New test function.
20064 (selftest::diagnostic_show_locus_c_tests): Call it.
20065 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
20068 2017-07-11 Tom de Vries <tom@codesourcery.com>
20070 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
20071 (bb_first_real_insn): New function.
20072 (nvptx_single): Add extra initialization of broadcasted condition
20075 2017-07-11 Nathan Sidwell <nathan@acm.org>
20077 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
20079 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
20081 * doc/extend.texi (AVR Function Attributes): Remove weblink to
20082 Binutils doc as TEXI will mess them up.
20083 * doc/invoke.texi (AVR Options): Same here.
20085 2017-07-11 Daniel Cederman <cederman@gaisler.com>
20087 * config/sparc/sparc.opt (mfix-ut700): New option.
20088 (mfix-gr712rc): Likewise.
20089 (sparc_fix_b2bst): New variable.
20090 * doc/invoke.texi (SPARC options): Document them.
20091 (ARM options): Fix warnings.
20092 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
20093 instructions to prevent sequences that can trigger the store-store
20094 errata for certain LEON3FT processors.
20095 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
20096 (sparc_option_override): Set sparc_fix_b2bst appropriately.
20097 * config/sparc/sparc.md (fix_b2bst): New attribute.
20098 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
20100 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
20103 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
20105 (*rsqrtsf2_sse): Ditto.
20107 (div<mode>3): Macroize insn from divdf3 and divsf3
20108 using MODEF mode iterator.
20110 2017-07-10 Martin Sebor <msebor@redhat.com>
20112 PR tree-optimization/80397
20113 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
20114 instead of testing for equality to INTEGER_TYPE.
20116 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
20118 * config.gcc: Remove uclibc from arc target spec.
20120 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
20122 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
20124 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20127 * lto-wrapper.c (remove_option): New function.
20128 (merge_and_complain): Merge PIC/PIE options more realistically.
20130 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
20132 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
20136 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
20137 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
20138 * config.in: Regenerate.
20139 * configure: Regenerate.
20140 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
20141 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
20142 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
20143 (TARGET_GASISR_PROLOGUES): ...target mask.
20144 * common/config/avr/avr-common.c
20145 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
20146 Set -mgas-isr-prologues.
20147 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
20148 INSERT_PASS_BEFORE for it.
20149 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
20150 * config/avr/avr.c (avr_option_override)
20151 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
20152 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
20153 (avr_attribute_table) <no_gccisr>: Add new function attribute.
20154 (avr_set_current_function) <is_no_gccisr>: Init machine field.
20155 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
20157 (make_avr_pass_pre_proep): New function.
20158 (emit_push_sfr) <treg>: Add argument to function and use it
20159 instead of TMP_REG.
20160 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
20161 and set machine->gasisr.yes.
20162 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
20163 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
20164 __gcc_isr.n_pushed to .L__stack_usage.
20165 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
20166 (avr_asm_final_postscan_insn): ...this new static function.
20167 * config/avr/avr.h (machine_function)
20168 <is_no_gccisr, use_L__stack_usage>: New fields.
20169 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
20170 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
20171 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
20172 (gasisr, *gasisr): New expander and insn.
20173 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20174 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
20175 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
20177 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
20179 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
20182 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
20184 Move jump-tables out of .text again.
20187 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
20188 (ASM_OUTPUT_ADDR_VEC): New function.
20189 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
20190 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
20191 INSN_ADDRESSes as asm comment.
20192 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
20193 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
20194 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
20195 * config/avr/avr.md (*tablejump): Adjust comment.
20196 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
20197 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
20199 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
20200 (avr_output_addr_vec): New proto.
20201 (avr_log_t) <insn_addresses>: New field.
20203 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
20206 * config/i386/i386.c (ix86_function_arg_advance): Set
20207 outgoing_args_on_stack to true if there are outgoing arguments
20209 (ix86_function_arg): Likewise.
20210 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
20211 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
20212 * config/i386/i386.h (machine_function): Add
20213 outgoing_args_on_stack.
20215 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
20217 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20218 supporting pthreds.
20219 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
20221 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
20223 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
20224 (REAL_H): Remove $(MACHMODE_H).
20225 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
20227 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
20228 $(MACHMODE_H) and double-int.h.
20229 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
20231 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
20234 2017-07-07 Andrew Pinski <apinski@cavium.com>
20236 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
20237 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
20239 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20241 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
20242 Add warning if GCC was not configured to link against a GLIBC that
20243 exports the hardware capability bits.
20244 (make_resolver_func): Make resolver function private and not a
20245 COMDAT function. Create the name with clone_function_name instead
20246 of make_unique_name.
20249 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
20250 correct operand in doing the split.
20252 2017-07-07 Carl Love <cel@us.ibm.com>
20254 * config/rs6000/rs6000-c: Add support for built-in function
20255 vector unsigned short vec_pack_to_short_fp32 (vector float,
20257 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
20258 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
20259 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
20260 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
20261 (convert_4f32_8i16): Add define_expand.
20262 * doc/extend.texi: Update the built-in documentation file for the
20263 new built-in function.
20265 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20267 * config/sparc/m8.md: New file.
20268 * config/sparc/sparc.md: Include m8.md.
20270 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20272 * config/sparc/sparc.opt: New option -mvis4b.
20273 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
20274 (sparc_option_override): Handle VIS4B.
20275 (enum sparc_builtins): Define
20276 SPARC_BUILTIN_DICTUNPACK{8,16,32},
20277 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
20278 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
20279 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
20280 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
20281 (check_constant_argument): New function.
20282 (sparc_vis_init_builtins): Define builtins
20283 __builtin_vis_dictunpack{8,16,32},
20284 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
20285 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
20286 __builtin_vis_fpcmpde{8,16,32}shl and
20287 __builtin_vis_fpcmpur{8,16,32}shl.
20288 (sparc_expand_builtin): Check that the constant operands to
20289 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
20290 constant and in range.
20291 * config/sparc/sparc-c.c (sparc_target_macros): Handle
20293 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
20294 (SPARC_IMM5_P): Likewise.
20295 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
20296 (enabled): Handle vis4b.
20297 (UNSPEC_DICTUNPACK): New unspec.
20298 (UNSPEC_FPCMPSHL): Likewise.
20299 (UNSPEC_FPUCMPSHL): Likewise.
20300 (UNSPEC_FPCMPDESHL): Likewise.
20301 (UNSPEC_FPCMPURSHL): Likewise.
20302 (cpu_feature): New CPU feature `vis4b'.
20303 (dictunpack{8,16,32}): New insns.
20304 (FPCSMODE): New mode iterator.
20305 (fpcscond): New code iterator.
20306 (fpcsucond): Likewise.
20307 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
20308 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
20309 (fpcmpde{8,16,32}{si,di}shl): Likewise.
20310 (fpcmpur{8,16,32}{si,di}shl): Likewise.
20311 * config/sparc/constraints.md: Define constraints `q' for unsigned
20312 2-bit integer constants and `t' for unsigned 5-bit integer
20314 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
20316 (imm5_operand_dictunpack16): Likewise.
20317 (imm5_operand_dictunpack32): Likewise.
20318 (imm2_operand): Likewise.
20319 * doc/invoke.texi (SPARC Options): Document -mvis4b.
20320 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
20321 ditunpack* and fpcmp*shl builtins.
20323 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20325 * config.gcc: Handle m8 in --with-{cpu,tune} options.
20326 * config.in: Add HAVE_AS_SPARC6 define.
20327 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
20329 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
20331 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
20332 (CPP_CPU_SPEC): Handle m8.
20333 (ASM_CPU_SPEC): Likewise.
20334 * config/sparc/sparc-opts.h (enum processor_type): Add
20336 * config/sparc/sparc.c (m8_costs): New struct.
20337 (sparc_option_override): Handle TARGET_CPU_m8.
20338 (sparc32_initialize_trampoline): Likewise.
20339 (sparc64_initialize_trampoline): Likewise.
20340 (sparc_issue_rate): Likewise.
20341 (sparc_register_move_cost): Likewise.
20342 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
20343 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
20344 (ASM_CPU64_DEFAULT_SPEC): Likewise.
20345 (CPP_CPU_SPEC): Handle M8.
20346 (ASM_CPU_SPEC): Likewise.
20347 (AS_M8_FLAG): Define.
20348 * config/sparc/sparc.md: Add m8 to the cpu attribute.
20349 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
20350 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
20352 * configure: Regenerate.
20353 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
20356 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20358 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
20360 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
20361 ("*movdi_insn_sp32"): Do not set v3pipe.
20362 ("*movsi_insn"): Likewise.
20363 ("*movdi_insn_sp64"): Likewise.
20364 ("*movsf_insn"): Likewise.
20365 ("*movdf_insn_sp32"): Likewise.
20366 ("*movdf_insn_sp64"): Likewise.
20367 ("*zero_extendsidi2_insn_sp64"): Likewise.
20368 ("*sign_extendsidi2_insn"): Likewise.
20369 ("*mov<VM32:mode>_insn"): Likewise.
20370 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20371 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20372 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
20373 ("<vlop:code><VL:mode>3"): Likewise.
20374 ("*not_<vlop:code><VL:mode>3"): Likewise.
20375 ("*nand<VL:mode>_vis"): Likewise.
20376 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
20377 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
20378 ("one_cmpl<VL:mode>2"): Likewise.
20379 ("faligndata<VM64:mode>_vis"): Likewise.
20380 ("alignaddrsi_vis"): Likewise.
20381 ("alignaddrdi_vis"): Likweise.
20382 ("alignaddrlsi_vis"): Likewise.
20383 ("alignaddrldi_vis"): Likewise.
20384 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20385 ("bmaskdi_vis"): Likewise.
20386 ("bmasksi_vis"): Likewise.
20387 ("bshuffle<VM64:mode>_vis"): Likewise.
20388 ("cmask8<P:mode>_vis"): Likewise.
20389 ("cmask16<P:mode>_vis"): Likewise.
20390 ("cmask32<P:mode>_vis"): Likewise.
20391 ("pdistn<P:mode>_vis"): Likewise.
20392 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20394 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20396 * config/sparc/sparc.md ("subtype"): New insn attribute.
20397 ("*wrgsr_sp64"): Set insn subtype.
20398 ("*rdgsr_sp64"): Likewise.
20399 ("alignaddrsi_vis"): Likewise.
20400 ("alignaddrdi_vis"): Likewise.
20401 ("alignaddrlsi_vis"): Likewise.
20402 ("alignaddrldi_vis"): Likewise.
20403 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
20404 ("fexpand_vis"): Likewise.
20405 ("fpmerge_vis"): Likewise.
20406 ("faligndata<VM64:mode>_vis"): Likewise.
20407 ("bshuffle<VM64:mode>_vis"): Likewise.
20408 ("cmask8<P:mode>_vis"): Likewise.
20409 ("cmask16<P:mode>_vis"): Likewise.
20410 ("cmask32<P:mode>_vis"): Likewise.
20411 ("fchksm16_vis"): Likewise.
20412 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
20413 ("fmean16_vis"): Likewise.
20414 ("fp<plusminus_insn>64_vis"): Likewise.
20415 ("<plusminus_insn>v8qi3"): Likewise.
20416 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20417 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
20418 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
20419 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
20420 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
20421 ("*movqi_insn"): Likewise.
20422 ("*movhi_insn"): Likewise.
20423 ("*movsi_insn"): Likewise.
20424 ("movsi_pic_gotdata_op"): Likewise.
20425 ("*movdi_insn_sp32"): Likewise.
20426 ("*movdi_insn_sp64"): Likewise.
20427 ("movdi_pic_gotdata_op"): Likewise.
20428 ("*movsf_insn"): Likewise.
20429 ("*movdf_insn_sp32"): Likewise.
20430 ("*movdf_insn_sp64"): Likewise.
20431 ("*zero_extendhisi2_insn"): Likewise.
20432 ("*zero_extendqihi2_insn"): Likewise.
20433 ("*zero_extendqisi2_insn"): Likewise.
20434 ("*zero_extendqidi2_insn"): Likewise.
20435 ("*zero_extendhidi2_insn"): Likewise.
20436 ("*zero_extendsidi2_insn_sp64"): Likewise.
20437 ("ldfsr"): Likewise.
20438 ("prefetch_64"): Likewise.
20439 ("prefetch_32"): Likewise.
20440 ("tie_ld32"): Likewise.
20441 ("tie_ld64"): Likewise.
20442 ("*tldo_ldub_sp32"): Likewise.
20443 ("*tldo_ldub1_sp32"): Likewise.
20444 ("*tldo_ldub2_sp32"): Likewise.
20445 ("*tldo_ldub_sp64"): Likewise.
20446 ("*tldo_ldub1_sp64"): Likewise.
20447 ("*tldo_ldub2_sp64"): Likewise.
20448 ("*tldo_ldub3_sp64"): Likewise.
20449 ("*tldo_lduh_sp32"): Likewise.
20450 ("*tldo_lduh1_sp32"): Likewise.
20451 ("*tldo_lduh_sp64"): Likewise.
20452 ("*tldo_lduh1_sp64"): Likewise.
20453 ("*tldo_lduh2_sp64"): Likewise.
20454 ("*tldo_lduw_sp32"): Likewise.
20455 ("*tldo_lduw_sp64"): Likewise.
20456 ("*tldo_lduw1_sp64"): Likewise.
20457 ("*tldo_ldx_sp64"): Likewise.
20458 ("*mov<VM32:mode>_insn"): Likewise.
20459 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20460 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20462 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20464 * config/sparc/sparc.md ("type"): New insn type viscmp.
20465 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
20467 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
20468 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
20469 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20470 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
20472 ("n7_vis_logical_11cycle"): Likewise.
20473 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
20474 * config/sparc/niagara2.md ("niag3_vis": Likewise.
20475 * config/sparc/niagara.md ("niag_vis"): Likewise.
20476 * config/sparc/ultra3.md ("us3_fga"): Likewise.
20477 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
20479 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20481 * config/sparc/sparc.md: New instruction type `bmask'.
20482 (bmaskdi_vis): Use the `bmask' type.
20483 (bmasksi_vis): Likewise.
20484 * config/sparc/ultra3.md (us3_array): Likewise.
20485 * config/sparc/niagara7.md (n7_array): Likewise.
20486 * config/sparc/niagara4.md (n4_array): Likewise.
20487 * config/sparc/niagara2.md (niag2_vis): Likewise.
20488 (niag3_vis): Likewise.
20489 * config/sparc/niagara.md (niag_vis): Likewise.
20491 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20493 * ipa-comdats.c: Remove optimize check from gate.
20494 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
20495 for functions not optimized.
20496 (ipa_fn_summary_read): Skip optimize check.
20497 (ipa_fn_summary_write): Likewise.
20498 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
20500 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
20502 (can_inline_edge_p): Check flag_pcc_struct_return for match.
20503 (check_callers): Give up on caller which is not optimized.
20504 (inline_small_functions): Likewise.
20505 (ipa_inline): Do not give up when not optimizing.
20506 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
20507 away unoptimizes cdtors.
20508 (whole_program_function_and_variable_visibility): Do
20509 ipa_discover_readonly_nonaddressable_vars in LTO mode.
20510 * ipa.c (process_references): Do not check optimize.
20511 (symbol_table::remove_unreachable_nodes): Update optimize check.
20512 (set_writeonly_bit): Update optimize check.
20513 (pass_ipa_cdtor_merge::gate): Do not check optimize.
20514 (pass_ipa_single_use::gate): Remove.
20516 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20518 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
20519 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
20520 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
20521 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
20522 permute_load, permute_store, adjust_extract, adjust_splat,
20523 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
20524 replace_swap_with_copy, dump_swap_insn_table,
20525 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
20526 recombine_lvx_pattern, recombine_stvx_pattern,
20527 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
20528 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
20529 to file rs6000-p8swap.c.
20530 * config/rs6000/rs6000-p8swap.c: New file.
20531 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
20532 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
20533 and rs6000*-*-* targets.
20535 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20537 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
20539 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20541 * lto-wrapper.c (merge_and_complain): Do not merge
20542 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
20543 fsigned_zeros, ftrapping_math, fwrapv.
20544 (append_compiler_options): Do not track these options.
20545 (append_linker_options): Likewie
20547 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20549 * cgraphunit.c (cgraph_node::finalize_function): When
20550 !flag_toplevel_reorde set no_reorder flag.
20551 (varpool_node::finalize_decl): Likewise.
20552 (symbol_table::compile): Drop no toplevel reorder path.
20554 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20556 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
20557 edges; zero probability is not better than uninitialized.
20559 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20561 * asan.h (asan_sanitize_allocas_p): Declare.
20562 * asan.c (asan_sanitize_allocas_p): New function.
20563 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
20564 (handle_builtin_alloca): Likewise.
20565 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
20566 if !asan_sanitize_allocas_p.
20567 * params.def (asan-instrument-allocas): Add new option.
20568 * params.h (ASAN_PROTECT_ALLOCAS): Define.
20569 * opts.c (common_handle_option): Disable allocas sanitization for
20572 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20574 * asan.c: Include gimple-fold.h.
20575 (get_last_alloca_addr): New function.
20576 (handle_builtin_stackrestore): Likewise.
20577 (handle_builtin_alloca): Likewise.
20578 (asan_emit_allocas_unpoison): Likewise.
20579 (get_mem_refs_of_builtin_call): Add new parameter, remove const
20580 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
20581 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
20582 (instrument_builtin_call): Pass gimple iterator to
20583 get_mem_refs_of_builtin_call.
20584 (last_alloca_addr): New global.
20585 * asan.h (asan_emit_allocas_unpoison): Declare.
20586 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
20587 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
20588 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
20589 if function calls alloca.
20590 * gimple-fold.c (replace_call_with_value): Remove static keyword.
20591 * gimple-fold.h (replace_call_with_value): Declare.
20592 * internal-fn.c: Include asan.h.
20593 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
20594 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
20596 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20598 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
20599 (C_SELFTEST_FLAGS): New.
20600 (CPP_SELFTEST_FLAGS): New.
20601 (SELFTEST_DEPS): New, from deps of s-selftest.
20602 (C_SELFTEST_DEPS): New, from deps of s-selftest.
20603 (CPP_SELFTEST_DEPS): New.
20604 (selftest): Add dependency on s-selftest-c++.
20605 (s-selftest): Rename to...
20606 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
20607 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
20608 than SELFTEST_FLAGS.
20609 (selftest-gdb): Rename to...
20610 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
20612 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
20613 (selftest-valgrind): Rename to...
20614 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
20616 (selftest-valgrind): Reintroduce as an alias for
20617 selftest-c-valgrind.
20618 (s-selftest-c++): New.
20619 (selftest-c++-gdb): New.
20620 (selftest-c++-valgrind): New.
20622 2017-07-06 Olivier Hainque <hainque@adacore.com>
20624 * gcc.c (process_command): When deciding if undefined variables
20625 should be ignored when processing specs, accept "gcc -v" as well.
20627 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20629 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
20630 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
20632 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20634 * config/arm/arm-cpus.in (armv8-r): Add new entry.
20635 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
20636 * config/arm/arm-tables.opt: Regenerate.
20637 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
20639 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
20641 2017-07-06 Carl Love <cel@us.ibm.com>
20643 * ChangeLog: Clean up from mid air collision
20645 2017-07-06 Carl Love <cel@us.ibm.com>
20647 * config/rs6000/rs6000-c.c: Add support for built-in functions
20648 vector signed int vec_subc (vector signed int, vector signed int);
20649 vector signed __int128 vec_subc (vector signed __int128,
20650 vector signed __int128);
20651 vector unsigned __int128 vec_subc (vector unsigned __int128,
20652 vector unsigned __int128);
20653 vector signed int vec_sube (vector signed int, vector signed int,
20654 vector signed int);
20655 vector unsigned int vec_sube (vector unsigned int,
20656 vector unsigned int,
20657 vector unsigned int);
20658 vector signed __int128 vec_sube (vector signed __int128,
20659 vector signed __int128,
20660 vector signed__int128);
20661 vector unsigned __int128 vec_sube (vector unsigned __int128,
20662 vector unsigned __int128,
20663 vector unsigned __int128);
20664 vector signed int vec_subec (vector signed int, vector signed int,
20665 vector signed int);
20666 vector unsigned int vec_subec (vector unsigned int,
20667 vector unsigned int,
20668 vector unsigned int);
20669 vector signed __int128 vec_subec (vector signed __int128,
20670 vector signed __int128,
20671 vector signed__int128);
20672 vector unsigned __int128 vec_subec (vector unsigned __int128,
20673 vector unsigned __int128,
20674 vector unsigned __int128);
20675 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
20676 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
20677 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
20678 BU_ALTIVEC_OVERLOAD_X definitions.
20679 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
20680 * doc/extend.texi: Update the built-in documentation file for the new
20681 built-in functions.
20683 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20686 * diagnostic-show-locus.c (layout::layout): Use start and finish
20687 spelling location for the start and finish of each range.
20688 * genmatch.c (linemap_client_expand_location_to_spelling_point):
20689 Add unused aspect param.
20690 * input.c (expand_location_1): Add "aspect" param, and use it
20691 to access the correct part of the location.
20692 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
20694 (expand_location_to_spelling_point): Likewise.
20695 (linemap_client_expand_location_to_spelling_point): Add "aspect"
20696 param, and pass it to expand_location_1.
20698 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
20700 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
20701 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
20702 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
20703 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
20704 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
20705 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
20706 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
20707 _mm_maskz_getmant_ss): New intrinsics.
20708 (__builtin_ia32_getexpss128_mask): Changed to ...
20709 __builtin_ia32_getexpss128_round ... this.
20710 (__builtin_ia32_getexpsd128_mask): Changed to ...
20711 __builtin_ia32_getexpsd128_round ... this.
20712 * config/i386/i386-builtin-types.def
20713 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
20714 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
20715 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
20716 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
20717 __builtin_ia32_getmantss_mask_round): New builtins.
20718 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
20719 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
20720 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
20721 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
20722 * config/i386/sse.md
20723 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
20724 avx512f_sgetexp<mode><mask_scalar_name>
20725 <round_saeonly_scalar_name> ... this.
20726 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
20727 %0, %1, %2<round_saeonly_op3>}): Changed to ...
20728 vgetexp<ssescalarmodesuffix>
20729 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20730 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
20731 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
20732 avx512f_vgetmant<mode><mask_scalar_name>
20733 <round_saeonly_scalar_name> ... this.
20734 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
20735 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
20736 vgetmant<ssescalarmodesuffix>
20737 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
20738 %0<mask_scalar_operand4>, %1, %2
20739 <round_saeonly_scalar_mask_op4>, %3} ... this.
20740 * config/i386/subst.md (mask_scalar_operand4,
20741 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
20742 round_saeonly_scalar_nimm_predicate): New subst attributes.
20744 2017-07-06 Julia Koval <julia.koval@intel.com>
20746 * config/i386/i386.c (ix86_erase_embedded_rounding):
20747 Remove code for old rounding pattern.
20749 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
20751 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
20753 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
20755 * doc/sourcebuild.texi (Test Directives, Variants of
20756 dg-require-support): Add documentation for dg-require-stack-check.
20758 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
20760 * config/i386/subst.md (mask_scalar, round_scalar,
20761 round_saeonly_scalar): New meta-templates.
20762 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
20763 round_scalar_mask_operand3, round_scalar_mask_op3,
20764 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
20765 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
20766 round_saeonly_scalar_constraint,
20767 round_saeonly_scalar_prefix): New subst attribute.
20768 * config/i386/sse.md
20769 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
20770 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
20771 <round_scalar_name> ... this.
20772 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
20773 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
20774 <round_scalar_name> ... this.
20775 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
20776 <sse>_vm<code><mode>3<mask_scalar_name>
20777 <round_saeonly_scalar_name> ... this.
20778 (v<plusminus_mnemonic><ssescalarmodesuffix>
20779 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20780 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20781 v<plusminus_mnemonic><ssescalarmodesuffix>
20782 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20783 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20784 (v<multdiv_mnemonic><ssescalarmodesuffix>
20785 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20786 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20787 v<multdiv_mnemonic><ssescalarmodesuffix>
20788 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20789 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20790 (v<maxmin_float><ssescalarmodesuffix>
20791 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
20792 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
20793 v<maxmin_float><ssescalarmodesuffix>
20794 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20795 %0<mask_scalar_operand3>, %1, %<iptr>2
20796 <round_saeonly_scalar_mask_op3>} ... this.
20798 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
20800 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
20801 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
20803 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
20804 Alan Hayward <alan.hayward@arm.com>
20805 David Sherwood <david.sherwood@arm.com>
20807 * combine.c (simplify_if_then_else): Remove "enum" before
20809 * compare-elim.c (can_eliminate_compare): Likewise.
20810 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
20812 (aarch64_lookup_simd_builtin_type): Likewise.
20813 (aarch64_simd_builtin_type): Likewise.
20814 (aarch64_init_simd_builtin_types): Likewise.
20815 (aarch64_simd_expand_args): Likewise.
20816 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
20818 (aarch64_reverse_mask): Likewise.
20819 (aarch64_simd_emit_reg_reg_move): Likewise.
20820 (aarch64_gen_adjusted_ldpstp): Likewise.
20821 (aarch64_ccmp_mode_to_code): Likewise.
20822 (aarch64_operands_ok_for_ldpstp): Likewise.
20823 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20824 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
20826 (aarch64_min_divisions_for_recip_mul): Likewise.
20827 (aarch64_reassociation_width): Likewise.
20828 (aarch64_get_condition_code_1): Likewise.
20829 (aarch64_simd_emit_reg_reg_move): Likewise.
20830 (aarch64_simd_attr_length_rglist): Likewise.
20831 (aarch64_reverse_mask): Likewise.
20832 (aarch64_operands_ok_for_ldpstp): Likewise.
20833 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20834 (aarch64_gen_adjusted_ldpstp): Likewise.
20835 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
20837 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
20838 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
20839 (arm_lookup_simd_builtin_type): Likewise.
20840 (arm_simd_builtin_type): Likewise.
20841 (arm_init_simd_builtin_types): Likewise.
20842 (arm_expand_builtin_args): Likewise.
20843 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
20844 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
20845 (ft32_setup_incoming_varargs): Likewise.
20846 (ft32_function_arg): Likewise.
20847 (ft32_function_arg_advance): Likewise.
20848 (ft32_pass_by_reference): Likewise.
20849 (ft32_arg_partial_bytes): Likewise.
20850 (ft32_valid_pointer_mode): Likewise.
20851 (ft32_addr_space_pointer_mode): Likewise.
20852 (ft32_addr_space_legitimate_address_p): Likewise.
20853 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
20855 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
20856 (ix86_emit_outlined_ms2sysv_restore): Likewise.
20857 (iamcu_alignment): Likewise.
20858 (canonicalize_vector_int_perm): Likewise.
20859 (ix86_noce_conversion_profitable_p): Likewise.
20860 (ix86_mpx_bound_mode): Likewise.
20861 (ix86_operands_ok_for_move_multiple): Likewise.
20862 * config/microblaze/microblaze-protos.h
20863 (microblaze_expand_conditional_branch_reg): Likewise.
20864 * config/microblaze/microblaze.c
20865 (microblaze_expand_conditional_branch_reg): Likewise.
20866 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20868 (rs6000_reassociation_width): Likewise.
20869 (rs6000_invalid_binary_op): Likewise.
20870 (fusion_p9_p): Likewise.
20871 (emit_fusion_p9_load): Likewise.
20872 (emit_fusion_p9_store): Likewise.
20873 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
20875 (riscv_hard_regno_mode_ok_p): Likewise.
20876 (riscv_address_insns): Likewise.
20877 (riscv_split_symbol): Likewise.
20878 (riscv_legitimize_move): Likewise.
20879 (riscv_function_value): Likewise.
20880 (riscv_hard_regno_nregs): Likewise.
20881 (riscv_expand_builtin): Likewise.
20882 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
20883 (riscv_build_integer): Likewise.
20884 (riscv_split_integer): Likewise.
20885 (riscv_legitimate_constant_p): Likewise.
20886 (riscv_cannot_force_const_mem): Likewise.
20887 (riscv_regno_mode_ok_for_base_p): Likewise.
20888 (riscv_valid_base_register_p): Likewise.
20889 (riscv_valid_offset_p): Likewise.
20890 (riscv_valid_lo_sum_p): Likewise.
20891 (riscv_classify_address): Likewise.
20892 (riscv_legitimate_address_p): Likewise.
20893 (riscv_address_insns): Likewise.
20894 (riscv_load_store_insns): Likewise.
20895 (riscv_force_binary): Likewise.
20896 (riscv_split_symbol): Likewise.
20897 (riscv_force_address): Likewise.
20898 (riscv_legitimize_address): Likewise.
20899 (riscv_move_integer): Likewise.
20900 (riscv_legitimize_const_move): Likewise.
20901 (riscv_legitimize_move): Likewise.
20902 (riscv_address_cost): Likewise.
20903 (riscv_subword): Likewise.
20904 (riscv_output_move): Likewise.
20905 (riscv_canonicalize_int_order_test): Likewise.
20906 (riscv_emit_int_order_test): Likewise.
20907 (riscv_function_arg_boundary): Likewise.
20908 (riscv_pass_mode_in_fpr_p): Likewise.
20909 (riscv_pass_fpr_single): Likewise.
20910 (riscv_pass_fpr_pair): Likewise.
20911 (riscv_get_arg_info): Likewise.
20912 (riscv_function_arg): Likewise.
20913 (riscv_function_arg_advance): Likewise.
20914 (riscv_arg_partial_bytes): Likewise.
20915 (riscv_function_value): Likewise.
20916 (riscv_pass_by_reference): Likewise.
20917 (riscv_setup_incoming_varargs): Likewise.
20918 (riscv_print_operand): Likewise.
20919 (riscv_elf_select_rtx_section): Likewise.
20920 (riscv_save_restore_reg): Likewise.
20921 (riscv_for_each_saved_reg): Likewise.
20922 (riscv_register_move_cost): Likewise.
20923 (riscv_hard_regno_mode_ok_p): Likewise.
20924 (riscv_hard_regno_nregs): Likewise.
20925 (riscv_class_max_nregs): Likewise.
20926 (riscv_memory_move_cost): Likewise.
20927 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
20928 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
20929 (rl78_addr_space_address_mode): Likewise.
20930 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20932 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20933 (rs6000_reassociation_width): Likewise.
20934 (rs6000_invalid_binary_op): Likewise.
20935 (fusion_p9_p): Likewise.
20936 (emit_fusion_p9_load): Likewise.
20937 (emit_fusion_p9_store): Likewise.
20938 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
20939 (ok_for_simple_move_operands): Likewise.
20940 (ok_for_simple_move_strict_operands): Likewise.
20941 (ok_for_simple_arith_logic_operands): Likewise.
20942 (visium_legitimize_reload_address): Likewise.
20943 (visium_select_cc_mode): Likewise.
20944 (output_cbranch): Likewise.
20945 (visium_split_double_move): Likewise.
20946 (visium_expand_copysign): Likewise.
20947 (visium_expand_int_cstore): Likewise.
20948 (visium_expand_fp_cstore): Likewise.
20949 * config/visium/visium.c (visium_pass_by_reference): Likewise.
20950 (visium_function_arg): Likewise.
20951 (visium_function_arg_advance): Likewise.
20952 (visium_libcall_value): Likewise.
20953 (visium_setup_incoming_varargs): Likewise.
20954 (visium_legitimate_constant_p): Likewise.
20955 (visium_legitimate_address_p): Likewise.
20956 (visium_legitimize_address): Likewise.
20957 (visium_secondary_reload): Likewise.
20958 (visium_register_move_cost): Likewise.
20959 (visium_memory_move_cost): Likewise.
20960 (prepare_move_operands): Likewise.
20961 (ok_for_simple_move_operands): Likewise.
20962 (ok_for_simple_move_strict_operands): Likewise.
20963 (ok_for_simple_arith_logic_operands): Likewise.
20964 (visium_function_value_1): Likewise.
20965 (rtx_ok_for_offset_p): Likewise.
20966 (visium_legitimize_reload_address): Likewise.
20967 (visium_split_double_move): Likewise.
20968 (visium_expand_copysign): Likewise.
20969 (visium_expand_int_cstore): Likewise.
20970 (visium_expand_fp_cstore): Likewise.
20971 (visium_split_cstore): Likewise.
20972 (visium_select_cc_mode): Likewise.
20973 (visium_split_cbranch): Likewise.
20974 (output_cbranch): Likewise.
20975 (visium_print_operand_address): Likewise.
20976 * expmed.c (flip_storage_order): Likewise.
20977 * expmed.h (emit_cstore): Likewise.
20978 (flip_storage_order): Likewise.
20979 * genrecog.c (validate_pattern): Likewise.
20980 * hsa-gen.c (gen_hsa_addr): Likewise.
20981 * internal-fn.c (expand_arith_overflow): Likewise.
20982 * ira-color.c (allocno_copy_cost_saving): Likewise.
20983 * lra-assigns.c (find_hard_regno_for_1): Likewise.
20984 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
20985 (process_invariant_for_inheritance): Likewise.
20986 * lra-eliminations.c (move_plus_up): Likewise.
20987 * omp-low.c (lower_oacc_reductions): Likewise.
20988 * simplify-rtx.c (simplify_subreg): Likewise.
20989 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
20990 (TARGET_CHKP_BOUND_MODE): Likewise..
20991 * targhooks.c (default_chkp_bound_mode): Likewise.
20992 (default_setup_incoming_vararg_bounds): Likewise.
20993 * targhooks.h (default_chkp_bound_mode): Likewise.
20994 (default_setup_incoming_vararg_bounds): Likewise.
20995 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
20996 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
20997 (have_whole_vector_shift): Likewise.
20998 * tree-vect-stmts.c (vectorizable_load): Likewise.
20999 * doc/tm.texi: Regenerate.
21001 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21003 Graceful degrade if Binutils PR21472 is not available.
21006 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
21007 .rodata in flash test fails.
21008 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
21009 * confgure: Regenerate.
21010 * config.in: Regenerate.
21011 * config/avr/avr.c (avr_asm_named_section)
21012 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
21013 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
21014 (avr_asm_init_sections): Same.
21016 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21018 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
21019 (fma<VH:mode>4_intrinsic): Likewise.
21020 (*fmsub<VCVTF:mode>4): Likewise.
21021 (*fmsub<VH:mode>4_intrinsic): Likewise.
21023 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21026 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
21027 Don't depend on "optimize > 0".
21028 (out_movhi_r_mr, out_movqi_mr_r): Same.
21029 (out_movhi_mr_r, out_movqi_r_mr): Same.
21030 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
21031 io_address_operand on "optimize > 0".
21033 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21035 * tree-loop-distribution.c: Add general explanantion on the pass.
21036 (generate_loops_for_partition): Mark distributed loop.
21037 (pg_add_dependence_edges): New parameter. Handle alias data
21038 dependence specially and record it in the parameter if asked.
21039 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
21040 (init_partition_graph_vertices, add_partition_graph_edge): New.
21041 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
21042 (free_partition_graph_vdata, build_partition_graph): New.
21043 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
21044 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
21045 (data_ref_segment_size, latch_dominated_by_data_ref): New.
21046 (compute_alias_check_pairs, version_loop_by_alias_check): New.
21047 (version_for_distribution_p, finalize_partitions): New.
21048 (distribute_loop): Handle alias data dependence specially. Factor
21049 out loop fusion code as functions and call these functions.
21051 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21053 * tree-loop-distribution.c (classify_partition): New parameter and
21054 better handle reduction statement.
21055 (rdg_build_partitions): Revise comment.
21056 (distribute_loop): Compute statements in all partitions and pass it
21057 to classify_partition.
21059 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21061 * tree-loop-distribution.c (enum partition_type): New.
21062 (struct partition): New field type.
21063 (partition_merge_into): Add parameter. Update partition type.
21064 (data_dep_in_cycle_p, update_type_for_merge): New functions.
21065 (build_rdg_partition_for_vertex): Compute partition type.
21066 (rdg_build_partitions): Dump partition type.
21067 (distribute_loop): Update calls to partition_merge_into.
21069 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21071 * tree-loop-distribution.c (struct ddr_hasher): New.
21072 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
21074 (classify_partition): Call get_data_dependence.
21075 (pg_add_dependence_edges): Ditto.
21076 (distribute_loop): Release data dependence hash table.
21078 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21080 * tree-loop-distribution.c (ref_base_address): Delete.
21081 (similar_memory_accesses): Rename ...
21082 (share_memory_accesses): ... to this. Check if partitions access
21083 the same memory reference.
21084 (distribute_loop): Call share_memory_accesses.
21086 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21088 * tree-loop-distribution.c (struct partition): New field recording
21089 its data reference.
21090 (partition_alloc, partition_free): Init and release data refs.
21091 (partition_merge_into): Merge data refs.
21092 (build_rdg_partition_for_vertex): Collect data refs for partition.
21093 (pg_add_dependence_edges): Change parameters from vector to bitmap.
21095 (distribute_loop): Remve data refs from vertice data of partition
21098 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21100 * tree-loop-distribution.c (params.h): Include header file.
21101 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
21102 (datarefs_vec): New global var.
21103 (create_rdg_vertices): Use datarefs_vec directly.
21104 (free_rdg): Don't free data references.
21105 (build_rdg): Update use. Don't free data references.
21106 (distribute_loop): Compute global variable for data references.
21107 Bail out if there are too many data references.
21109 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21111 * tree-loop-distribution.c (loop_nest): New global var.
21112 (build_rdg): Use loop directly, rather than loop nest.
21113 (pg_add_dependence_edges): Remove loop nest parameter. Use global
21115 (distribute_loop): Compute global variable loop nest. Update use.
21117 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21119 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
21120 (partition_merge_into): New parameter. Dump reason for fusion.
21121 (distribute_loop): Update use of partition_merge_into.
21123 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21125 * tree-loop-distribution.c (bb_top_order_index): New.
21126 (bb_top_order_index_size, bb_top_order_cmp): New.
21127 (stmts_from_loop): Use topological order.
21128 (pass_loop_distribution::execute): Compute and release topological
21129 order for basic blocks.
21131 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21133 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
21136 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21138 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
21139 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
21140 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
21141 * internal-fn.def (LOOP_DIST_ALIAS): New.
21142 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
21143 (fold_loop_internal_call): ... this.
21144 (vect_loop_dist_alias_call): New function.
21145 (set_uid_loop_bbs): Call fold_loop_internal_call.
21146 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
21149 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
21152 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
21153 Require dead FLAGS_REG at the beginning of a peephole.
21155 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
21158 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
21159 arguments in the call to __builtin_ia32_sbb_u32.
21160 (_subborrow_u64): Swap _X and _Y arguments in the call to
21161 __builtin_ia32_sbb_u64.
21163 2017-07-04 Jakub Jelinek <jakub@redhat.com>
21166 * tree-vrp.c (compare_assert_loc): Turn into a function template
21167 with stable template parameter. Only test if a->e is NULL,
21168 !a->e == !b->e has been verified already. Use e == NULL or
21169 e != NULL instead of e or ! e tests. If stable is true, don't use
21170 iterative_hash_expr, on the other side allow a or b or both NULL
21171 and sort the NULLs last.
21172 (process_assert_insertions): Sort using compare_assert_loc<false>
21173 instead of compare_assert_loc, later sort using
21174 compare_assert_loc<true> before calling process_assert_insertions_for
21175 in a loop. Use break instead of continue once seen NULL pointer.
21177 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21179 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21180 Cortex-R7 and Cortex-R8 processors.
21182 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21184 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
21185 uninitialized while src is not.
21187 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
21189 * common/config/arm/arm-common.c: Adjust include path for
21191 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
21192 (arm-cpu.h): Create in build directory. Adjust dependency rules.
21193 (arm-cpu-data.h): Likewise.
21194 (arm-cpu-cdata.h): Likewise.
21195 * config/arm/arm-cpu.h: Delete.
21196 * config/arm/arm-cpu-cdata.h: Delete.
21197 * config/arm/arm-cpu-data.h: Delete.
21199 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
21201 * config/arm/arm-cpus.in (cortex-a55): New.
21202 (cortex-a75): Likewise.
21203 (cortex-a75.cortex-a55): Likewise.
21204 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
21206 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
21207 * config/arm/arm-cpu-cdata.h: Regenerate.
21208 * config/arm/arm-cpu-data.h: Regenerate.
21209 * config/arm/arm-cpu.h: Regenerate.
21210 * config/arm/arm-tables.opt: Regenerate.
21211 * config/arm/arm-tune.md: Regenerate.
21213 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21215 * haifa-sched.c (sched_create_recovery_edges): Update profile.
21217 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21219 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
21222 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
21224 PR tree-optimization/81292
21225 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
21226 full_string_p, also call adjust_related_strinfos if the adjustment
21227 is simple, otherwise invalidate related strinfos.
21229 2017-07-04 Martin Liska <mliska@suse.cz>
21232 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
21233 newly created variable as DECL_IGNORED_P.
21235 2017-07-04 Martin Liska <mliska@suse.cz>
21238 * ipa-inline.c (inline_small_functions):
21239 Use xstrdup_for_dump.
21241 2017-07-04 Tom de Vries <tom@codesourcery.com>
21243 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
21245 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
21248 * config/darwin.c (darwin_function_switched_text_sections):
21251 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
21253 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
21255 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
21257 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
21259 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21261 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
21262 min_profitable_iters, and th as inclusive lower bounds.
21263 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
21264 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
21265 for min_profitable_iters and min_profitable_estimate.
21266 (vect_transform_loop): Treat th as an inclusive lower bound.
21267 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
21269 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
21272 * config/darwin.c (darwin_function_switched_text_sections):
21273 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
21274 in two pieces, and suppress the use of buf.
21276 2017-07-03 Nathan Sidwell <nathan@acm.org>
21278 * hash-table.h (hash_table_mod1): Fix indentation.
21280 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21282 PR middle-end/81290
21283 * predict.c (force_edge_cold): Be more careful about propagation
21285 * profile-count.h (profile_probability::guessed,
21286 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
21288 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
21290 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
21292 * doc/invoke.texi (rcpc architecture extension): Document it.
21294 2017-07-03 Richard Biener <rguenther@suse.de>
21296 PR tree-optimization/60510
21297 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
21298 the scalar reduction PHI and use it.
21299 (vectorizable_reduction): Properly guard the single_defuse_cycle
21300 path for non-SLP reduction chains where we cannot use it.
21301 Rework reduc_def/index and vector type deduction. Rework
21302 vector operand gathering during reduction op code-gen.
21303 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
21304 chains dissolve the chain and leave it to non-SLP reduction
21307 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21309 * tree-data-ref.h (dr_alignment): Declare.
21310 * tree-data-ref.c (dr_alignment): New function.
21311 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
21312 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
21314 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
21316 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21318 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
21319 and base_misalignment fields.
21320 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
21321 * tree-data-ref.c: Include builtins.h.
21322 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
21323 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
21324 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
21325 * tree-vect-data-refs.c: Include tree-cfg.h.
21326 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
21327 fields instead of calculating an alignment here.
21328 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
21329 innermost_loop_behavior fields.
21331 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21333 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
21335 (DR_STEP_ALIGNMENT): New macro.
21336 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
21337 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
21338 (create_data_ref): Print it.
21339 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
21340 to tell whether the step preserves vector (mis)alignment.
21341 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
21342 Move the check for an integer step and generalise to all INTEGER_CST.
21343 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
21344 Print the outer step alignment.
21346 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21348 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
21349 with offset_alignment.
21350 (DR_ALIGNED_TO): Delete.
21351 (DR_OFFSET_ALIGNMENT): New macro.
21352 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
21353 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
21354 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
21355 (create_data_ref): Likewise.
21356 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
21357 (vect_analyze_data_refs): Likewise.
21358 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
21359 creating dummy innermost behavior.
21361 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21363 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
21364 with a "innermost_loop_behavior *" and refeence tree.
21365 * tree-data-ref.c (dr_analyze_innermost): Likewise.
21366 (create_data_ref): Update call accordingly.
21367 * tree-predcom.c (find_looparound_phi): Likewise.
21369 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21371 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
21372 fields with dr_wrt_vec_loop.
21373 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
21374 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
21375 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
21376 (vect_dr_behavior): New function.
21377 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
21378 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
21379 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
21380 track whether the step preserves the misalignment.
21381 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
21382 Use vect_dr_behavior.
21383 (vect_setup_realignment): Update call accordingly.
21384 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
21385 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
21386 call to vect_create_addr_base_for_vector_ref.
21387 (vect_create_cond_for_align_checks): Likewise.
21388 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
21389 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
21390 (vect_recog_mask_conversion_pattern): Likewise.
21391 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
21392 (new_stmt_vec_info): Remove redundant zeroing.
21394 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
21396 * common/config/arm/arm-common.c (arm_be8_option): New function.
21397 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
21398 (ISA_ARMv6): Add isa_bit_be8.
21399 * config/arm/arm.h (arm_be8_option): Add prototype.
21400 (BE8_SPEC_FUNCTION): New define.
21401 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
21402 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
21403 (mlittle-endian): Similarly.
21404 (mbe8, mbe32): New options.
21405 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
21406 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
21408 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21410 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
21412 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21414 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
21415 (cleanup_tree_cfg_bb): Use it.
21416 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
21418 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
21420 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21423 * loop-doloop.c (add_test): Update profile.
21425 2017-07-03 Martin Liska <mliska@suse.cz>
21428 * sanopt.c (rewrite_usage_of_param): New function.
21429 (sanitize_rewrite_addressable_params): Likewise.
21430 (pass_sanopt::execute): Call rewrite_usage_of_param.
21432 2017-07-03 Richard Biener <rguenther@suse.de>
21434 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
21435 back to using VIEW_CONVERT_EXPR.
21437 2017-07-03 Martin Liska <mliska@suse.cz>
21440 * doc/extend.texi: Document when a resolver function is
21441 generated for target_clones.
21443 2017-07-03 Martin Liska <mliska@suse.cz>
21445 * asan.c (asan_emit_stack_protection): Unpoison just red zones
21446 and shadow memory of auto variables which are subject of
21447 use-after-scope sanitization.
21448 (asan_expand_mark_ifn): Add do set only when is_poison.
21450 2016-07-03 Richard Biener <rguenther@suse.de>
21452 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
21454 (vect_force_simple_reduction): Record reduction def -> phi mapping.
21455 (vectorizable_reduction): Perform reduction PHI creation when
21456 visiting a reduction PHI and adjust and simplify code generation
21457 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
21458 (vect_transform_loop): Visit reduction PHIs.
21459 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
21460 defs into the SLP tree.
21461 (vect_build_slp_tree): Reduction defs terminate the recursion.
21462 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
21464 (vect_get_vec_defs_for_stmt_copy): Export.
21465 (vect_get_vec_defs): Likewise.
21466 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
21468 (vect_get_vec_defs_for_stmt_copy): Declare.
21469 (vect_get_vec_defs): Likewise.
21471 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21473 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
21474 parameter with a "loop" parameter and use it instead of the
21475 loop containing DR_STMT. Don't check simple_iv when doing
21476 BB analysis. Describe the two analysis modes in the comment.
21478 2017-07-03 Tom de Vries <tom@codesourcery.com>
21480 PR tree-optimization/69468
21481 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
21482 (find_same_succ_bb): Handle ignore_edge_flags.
21484 2017-07-03 Tom de Vries <tom@codesourcery.com>
21486 PR tree-optimization/81192
21487 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
21489 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
21491 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
21493 2017-07-03 Tom de Vries <tom@codesourcery.com>
21495 PR tree-optimization/81192
21496 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
21497 BB_SAME_SUCC (bb) == NULL.
21499 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21501 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
21504 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21506 * dumpfile.c: Include profile-count.h
21507 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
21509 (insert_cond_bb): Update profile.
21510 * tree-cfg.h (insert_cond_bb): Update prototype.
21511 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
21512 * tree-dump.c: Do not include tree-cfg.
21514 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21516 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
21518 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21520 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
21523 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21525 * tree-complex.c (expand_complex_div_wide): update profile.
21527 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21528 Alan Hayward <alan.hayward@arm.com>
21529 David Sherwood <david.sherwood@arm.com>
21531 * Makefile.in (MACHMODE_H): Remove insn-modes.h
21532 (CORETYPES_H): New define.
21533 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
21534 (insn-modes-inline.h, s-modes-inline-h): New rules.
21535 (generated_files): Add insn-modes-inline.h.
21536 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
21537 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
21538 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
21539 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
21540 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
21541 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
21542 (build/gencodes.o, build/genconditions.o): Likewise.
21543 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
21544 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
21545 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
21546 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
21547 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
21548 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
21549 * coretypes.h: Include everything up to real.h for generators.
21550 Include insn-modes.h first. Include wide-int-print.h after
21551 wide-int.h. Include insn-modes-inline.h and then machmode.h.
21552 * machmode.h: Don't include insn-modes.h here.
21553 * function-tests.c: Remove includes of signop.h, machmode.h,
21554 double-int.h and wide-int.h.
21556 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
21558 * optc-save-gen.awk: Likewise.
21559 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
21560 * godump.c: Remove include of wide-int-print.h.
21561 * pretty-print.h: Likewise.
21562 * wide-int-print.cc: Likewise.
21563 * wide-int.cc: Likewise.
21564 * hash-map-tests.c: Remove include of signop.h.
21565 * hash-set-tests.c: Likewise.
21566 * rtl-tests.c: Likewise.
21567 * mkconfig.sh: Remove include of machmode.h.
21568 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
21570 (emit_insn_modes_inline_h): ...this new function. Emit the code
21571 into an insn-modes-inline.h header file, adding appropriate
21572 include guards and end comments.
21573 (emit_insn_modes_c_header): Remove include of machmode.h.
21574 (emit_min_insn_modes_c_header): Include coretypes.h rather than
21576 (main): Handle -i flag and call emit_insn_modes_inline_h when
21579 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21581 * tree-ssa-strlen.c (strinfo): Rename the length field to
21582 nonzero_chars. Add a full_string_p field.
21583 (compare_nonzero_chars, zero_length_string_p): New functions.
21584 (get_addr_stridx): Add an offset_out parameter.
21585 Use compare_nonzero_chars.
21586 (get_stridx): Update accordingly. Use compare_nonzero_chars.
21587 (new_strinfo): Update after above changes to strinfo.
21588 (set_endptr_and_length): Set full_string_p.
21589 (get_string_length): Update after above changes to strinfo.
21590 (unshare_strinfo): Update call to new_strinfo.
21591 (maybe_invalidate): Likewise.
21592 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
21593 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
21594 as a uhwi instead of an shwi. Update after above changes to
21595 strinfo and new_strinfo.
21596 (zero_length_string): Assert that chainsi contains full strings.
21597 Use zero_length_string_p. Update call to new_strinfo.
21598 (adjust_related_strinfos): Update after above changes to strinfo.
21599 Copy full_string_p from origsi.
21600 (adjust_last_stmt): Use zero_length_string_p.
21601 (handle_builtin_strlen): Update after above changes to strinfo and
21602 new_strinfo. Install the lhs as the string length if the previous
21603 entry didn't describe a full string.
21604 (handle_builtin_strchr): Update after above changes to strinfo
21606 (handle_builtin_strcpy): Likewise.
21607 (handle_builtin_strcat): Likewise.
21608 (handle_builtin_malloc): Likewise.
21609 (handle_pointer_plus): Likewise.
21610 (handle_builtin_memcpy): Likewise. Track nonzero characters
21611 that aren't necessarily followed by a nul terminator.
21612 (handle_char_store): Likewise.
21614 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21616 PR tree-optimization/80769
21617 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
21618 for malloc and calloc. Document the new invariant that all related
21619 strinfos have delayed lengths or none do.
21620 (verify_related_strinfos): Move earlier in file.
21621 (set_endptr_and_length): New function, split out from...
21622 (get_string_length): ...here. Also set the lengths of related
21624 (zero_length_string): Assert that chainsi has known (rather than
21626 (adjust_related_strinfos): Likewise.
21628 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21630 PR tree-optimization/81136
21631 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
21632 assert that two references with the same misalignment have the same
21633 compile-time misalignment if those compile-time misalignments
21636 2017-07-01 Andi Kleen <ak@linux.intel.com>
21638 * print-tree.c (print_node): Print all attributes.
21640 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21642 * cfg.c (scale_bbs_frequencies): New function.
21643 * cfg.h (scale_bbs_frequencies): Declare it.
21644 * cfgloopanal.c (single_likely_exit): Cleanup.
21645 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
21647 (scale_loop_profile): Likewise.
21648 (loop_version): Likewise.
21649 (create_empty_loop_on_edge): Update.
21650 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
21651 scale_loop_frequencies, scale_loop_profile, loopify,
21652 loop_version): Update prototypes.
21653 * modulo-sched.c (sms_schedule): Update.
21654 * predict.c (unlikely_executed_edge_p): Also check probability.
21655 (probably_never_executed_edge_p): Fix typo.
21656 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21657 * tree-parloops.c (gen_parallel_loop): Update.
21658 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
21659 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21660 * tree-ssa-loop-split.c (split_loop): Update.
21661 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21662 * tree-vect-loop-manip.c (vect_do_peeling): Update.
21663 (vect_loop_versioning): Update.
21664 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21666 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21668 * trans-mem.c (split_bb_make_tm_edge): Update profile.
21670 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21672 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
21673 to keep profile consistent.
21675 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21677 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
21678 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
21679 * profile-count.h (max_safe_multiplier): Make unsigned.
21680 (profile_count::guessed_zero): New.
21682 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21684 * bb-reorder.c (fix_up_crossing_landing_pad,
21685 fix_crossing_conditional_branches): Use make_single_succ_edge
21686 to keep profile consistent.
21688 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21690 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
21693 2017-07-01 Jakub Jelinek <jakub@redhat.com>
21696 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
21697 the right scopes, make sure cond_jump isn't preserved between multiple
21698 iterations. Search for fallthru edge whenever there are 3+ edges and
21699 use find_fallthru_edge for it.
21701 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21703 Patch by Alexander Monakov <amonakov@ispras.ru>
21704 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
21705 probabilities consistently.
21707 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21709 * pa.c (pa_expand_compare_and_swap_loop): Update call of
21710 emit_cmp_and_jump_insns.
21712 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21715 * tree-inline.c (expand_call_inline): Combine profile statuses.
21717 2017-06-30 Andrew Pinski <apinski@cavium.com>
21719 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
21720 fold_stmt returned true.
21722 2017-06-30 Nathan Sidwell <nathan@acm.org>
21724 * ggc.h (empty_string): Delete.
21725 * cfgexpand.c (expand_asm_stmt): Use plain "".
21726 * optabs.c (expand_asm_memory_barrier): Likewise.
21727 * stringpool.c (empty_string): Delete.
21728 (digit_vector, digit_string): Delete.
21729 (ggc_alloc_string): Use plain "", don't optimize single digit
21730 strings. Use ggc_alloc_atomic.
21732 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
21734 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
21735 comparison set and one other set, use the cost of the non-comparison
21738 2017-06-30 Nathan Sidwell <nathan@acm.org>
21740 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
21743 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
21745 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
21746 loops. Remove now unneeded calls to gimple_switch_set_label() that
21747 just set removed labels to NULL_TREE.
21749 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
21751 * tree-ssanames.c (set_range_info_raw): Abstract from ...
21752 (set_range_info): ...here. Only call set_range_info_raw if domain
21754 (set_nonzero_bits): Call set_range_info_raw.
21755 * tree-ssanames.h (set_range_info_raw): New.
21757 2017-06-30 Jakub Jelinek <jakub@redhat.com>
21760 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
21761 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
21762 of nonimmediate_operand and <store_mask_constraint> instead of m
21763 for the input operand. For V8FI iterator, always split if input
21764 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
21765 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
21766 <store_mask_predicate> instead of register_operand and
21767 <store_mask_constraint> instead of v for the input operand. Make
21768 sure both operands aren't MEMs for if not <mask_applied>.
21770 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
21772 * lto-wrapper.c (copy_file) Close both file descriptors before
21775 2017-06-30 Martin Liska <mliska@suse.cz>
21778 * multiple_target.c (create_dispatcher_calls): Make ifunc
21779 also for function that don't have calls or are not referenced.
21781 2017-06-30 Richard Biener <rguenther@suse.de>
21783 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
21784 analyze the first scalar stmt. Move vector type computation
21785 for the BB case here from ...
21786 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
21787 live operation processing in the SLP case properly.
21789 2017-06-30 Richard Biener <rguenther@suse.de>
21791 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
21793 2017-06-30 Martin Liska <mliska@suse.cz>
21796 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
21797 before BUILT_IN_UNWIND_RESUME when ASAN is used.
21799 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
21801 * doc/invoke.texi (AArch64): Add missing options and remove redundant
21804 2017-06-30 Richard Biener <rguenther@suse.de>
21806 PR tree-optimization/81249
21807 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
21808 condition reduction result to original scalar type.
21810 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21812 * profile-count.h (enum profile_quality): Fix typos and whitespace
21815 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21817 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
21818 type for branch probabilities.
21820 2017-06-29 Julian Brown <julian@codesourcery.com>
21821 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21823 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
21824 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
21825 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
21826 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
21828 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21830 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
21831 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
21832 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
21833 CC usage from generic code to here.
21834 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
21835 CC usage into the target macros.
21837 2017-06-29 Maya Rashish <coypu@sdf.org>
21839 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
21842 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21844 * arm/arm-builtins.c: Include profile-count.h
21845 * except.c (sjlj_emit_function_enter): Use
21846 profile_probability::unlikely.
21848 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21850 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
21851 and tocrel_offset be pointer args rather than implicitly using
21853 (legitimate_constant_pool_address_p, rs6000_emit_move,
21854 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
21855 tocrel_offset and use in toc_relative_expr_p call.
21856 (print_operand, print_operand_address): Use static tocrel_base_oac
21857 and tocrel_offset_oac.
21858 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
21861 2017-06-29 Maya Rashish <coypu@sdf.org>
21863 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
21865 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
21867 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
21868 objects, take into account only the alignment of 'op0' and 'mode1' if
21871 2017-06-29 Steve Ellcey <sellcey@cavium.com>
21873 * ccmp.c (ccmp_tree_comparison_p): New function.
21874 (ccmp_candidate_p): Update to use above function.
21875 (get_compare_parts): New function.
21876 (expand_ccmp_next): Update to use new functions.
21877 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
21879 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
21880 take mode as argument.
21881 * ccmp.h (expand_ccmp_expr): Add mode as argument.
21882 * expr.c (expand_expr_real_1): Pass mode as argument.
21884 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
21886 * combine.c (combine_instructions): Print insns to dump_file, together
21889 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21891 * asan.c (asan_emit_stack_protection): Update.
21892 (create_cond_insert_point): Update.
21893 * auto-profile.c (afdo_propagate_circuit): Update.
21894 * basic-block.h (struct edge_def): Turn probability to
21895 profile_probability.
21896 (EDGE_FREQUENCY): Update.
21897 * bb-reorder.c (find_traces_1_round): Update.
21898 (better_edge_p): Update.
21899 (sanitize_hot_paths): Update.
21900 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
21901 (make_single_succ_edge): Update.
21902 (check_bb_profile): Update.
21903 (dump_edge_info): Update.
21904 (update_bb_profile_for_threading): Update.
21905 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
21906 probabilitycount to 0.
21907 * cfgbuild.c (compute_outgoing_frequencies): Update.
21908 * cfgcleanup.c (try_forward_edges): Update.
21909 (outgoing_edges_match): Update.
21910 (try_crossjump_to_edge): Update.
21911 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
21912 (expand_gimple_tailcall): Update.
21913 (construct_init_block): Use make_single_succ_edge.
21914 (construct_exit_block): Use make_single_succ_edge.
21915 * cfghooks.c (verify_flow_info): Update.
21916 (redirect_edge_succ_nodup): Update.
21917 (split_edge): Update.
21918 (account_profile_record): Update.
21919 * cfgloopanal.c (single_likely_exit): Update.
21920 * cfgloopmanip.c (scale_loop_profile): Update.
21921 (set_zero_probability): Remove.
21922 (duplicate_loop_to_header_edge): Update.
21923 * cfgloopmanip.h (loop_version): Update prototype.
21924 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
21925 (force_nonfallthru_and_redirect): Update.
21926 (update_br_prob_note): Update.
21927 (rtl_verify_edges): Update.
21928 (purge_dead_edges): Update.
21929 (rtl_lv_add_condition_to_bb): Update.
21930 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
21931 * cgraphunit.c (init_lowered_empty_function): Update.
21932 (cgraph_node::expand_thunk): Update.
21933 * cilk-common.c: Include profile-count.h
21934 * dojump.c (inv): Remove.
21935 (jumpifnot): Update.
21936 (jumpifnot_1): Update.
21937 (do_jump_1): Update.
21939 (do_jump_by_parts_greater_rtx): Update.
21940 (do_compare_rtx_and_jump): Update.
21941 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
21942 do_jump_1. do_compare_rtx_and_jump): Update prototype.
21943 * dwarf2cfi.c: Include profile-count.h
21944 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
21945 (sjlj_emit_dispatch_table): Likewise.
21946 * explow.c: Include profile-count.h
21947 * expmed.c (emit_store_flag_force): Update.
21948 (do_cmp_and_jump): Update.
21949 * expr.c (compare_by_pieces_d::generate): Update.
21950 (compare_by_pieces_d::finish_mode): Update.
21951 (emit_block_move_via_loop): Update.
21952 (store_expr_with_bounds): Update.
21953 (store_constructor): Update.
21954 (expand_expr_real_2): Update.
21955 (expand_expr_real_1): Update.
21956 * expr.h (try_casesi, try_tablejump): Update prototypes.
21957 * gimple-pretty-print.c (dump_probability): Update.
21958 (dump_profile): New.
21959 (dump_gimple_label): Update.
21960 (dump_gimple_bb_header): Update.
21961 * graph.c (draw_cfg_node_succ_edges): Update.
21962 * hsa-gen.c (convert_switch_statements): Update.
21963 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
21964 (find_if_case_1): Update.
21965 (find_if_case_2): Update.
21966 * internal-fn.c (expand_arith_overflow_result_store): Update.
21967 (expand_addsub_overflow): Update.
21968 (expand_neg_overflow): Update.
21969 (expand_mul_overflow): Update.
21970 (expand_vector_ubsan_overflow): Update.
21971 * ipa-cp.c (good_cloning_opportunity_p): Update.
21972 * ipa-split.c (split_function): Use make_single_succ_edge.
21973 * ipa-utils.c (ipa_merge_profiles): Update.
21974 * loop-doloop.c (add_test): Update.
21975 (doloop_modify): Update.
21976 * loop-unroll.c (compare_and_jump_seq): Update.
21977 (unroll_loop_runtime_iterations): Update.
21978 * lra-constraints.c (lra_inheritance): Update.
21979 * lto-streamer-in.c (input_cfg): Update.
21980 * lto-streamer-out.c (output_cfg): Update.
21981 * mcf.c (adjust_cfg_counts): Update.
21982 * modulo-sched.c (sms_schedule): Update.
21983 * omp-expand.c (expand_omp_for_init_counts): Update.
21984 (extract_omp_for_update_vars): Update.
21985 (expand_omp_ordered_sink): Update.
21986 (expand_omp_for_ordered_loops): Update.
21987 (expand_omp_for_generic): Update.
21988 (expand_omp_for_static_nochunk): Update.
21989 (expand_omp_for_static_chunk): Update.
21990 (expand_cilk_for): Update.
21991 (expand_omp_simd): Update.
21992 (expand_omp_taskloop_for_outer): Update.
21993 (expand_omp_taskloop_for_inner): Update.
21994 * omp-simd-clone.c (simd_clone_adjust): Update.
21995 * optabs.c (expand_doubleword_shift): Update.
21996 (expand_abs): Update.
21997 (emit_cmp_and_jump_insn_1): Update.
21998 (expand_compare_and_swap_loop): Update.
21999 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
22000 * predict.c (predictable_edge_p): Update.
22001 (edge_probability_reliable_p): Update.
22002 (set_even_probabilities): Update.
22003 (combine_predictions_for_insn): Update.
22004 (combine_predictions_for_bb): Update.
22005 (propagate_freq): Update.
22006 (estimate_bb_frequencies): Update.
22007 (force_edge_cold): Update.
22008 * profile-count.c (profile_count::dump): Add missing space into dump.
22009 (profile_count::debug): Add newline.
22010 (profile_count::differs_from_p): Explicitly convert to unsigned.
22011 (profile_count::stream_in): Update.
22012 (profile_probability::dump): New member function.
22013 (profile_probability::debug): New member function.
22014 (profile_probability::differs_from_p): New member function.
22015 (profile_probability::differs_lot_from_p): New member function.
22016 (profile_probability::stream_in): New member function.
22017 (profile_probability::stream_out): New member function.
22018 * profile-count.h (profile_count_quality): Rename to ...
22019 (profile_quality): ... this one.
22020 (profile_probability): New.
22021 (profile_count): Update.
22022 * profile.c (compute_branch_probabilities): Update.
22023 * recog.c (peep2_attempt): Update.
22024 * sched-ebb.c (schedule_ebbs): Update.
22025 * sched-rgn.c (find_single_block_region): Update.
22026 (compute_dom_prob_ps): Update.
22027 (schedule_region): Update.
22028 * sel-sched-ir.c (compute_succs_info): Update.
22029 * stmt.c (struct case_node): Update.
22030 (do_jump_if_equal): Update.
22031 (get_outgoing_edge_probs): Update.
22032 (conditional_probability): Update.
22033 (emit_case_dispatch_table): Update.
22034 (expand_case): Update.
22035 (expand_sjlj_dispatch_table): Update.
22036 (emit_case_nodes): Update.
22037 * targhooks.c: Update.
22038 * tracer.c (better_p): Update.
22039 (find_best_successor): Update.
22040 * trans-mem.c (expand_transaction): Update.
22041 * tree-call-cdce.c: Update.
22042 * tree-cfg.c (gimple_split_edge): Upate.
22043 (move_sese_region_to_fn): Upate.
22044 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
22045 * tree-eh.c (lower_resx): Upate.
22046 (cleanup_empty_eh_move_lp): Upate.
22047 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22048 * tree-inline.c (copy_edges_for_bb): Update.
22049 (copy_cfg_body): Update.
22050 * tree-parloops.c (gen_parallel_loop): Update.
22051 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
22052 (gimple_gen_time_profiler): Update.
22053 * tree-ssa-dce.c (remove_dead_stmt): Update.
22054 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22055 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
22056 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
22057 (unloop_loops): Update.
22058 (try_peel_loop): Update.
22059 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22060 * tree-ssa-loop-split.c (connect_loops): Update.
22061 (split_loop): Update.
22062 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22063 (hoist_guard): Update.
22064 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
22065 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
22066 (value_replacement): Update.
22067 * tree-ssa-reassoc.c (branch_fixup): Update.
22068 * tree-ssa-tail-merge.c (replace_block_by): Update.
22069 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
22070 (create_edge_and_update_destination_phis): Update.
22071 (compute_path_counts): Update.
22072 (recompute_probabilities): Update.
22073 (update_joiner_offpath_counts): Update.
22074 (freqs_to_counts_path): Update.
22075 (duplicate_thread_path): Update.
22076 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
22077 (struct switch_conv_info): Update.
22078 (gen_inbound_check): Update.
22079 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22080 (vect_do_peeling): Update.
22081 (vect_loop_versioning): Update.
22082 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22083 (optimize_mask_stores): Update.
22084 * ubsan.c (ubsan_expand_null_ifn): Update.
22085 * value-prof.c (gimple_divmod_fixed_value): Update.
22086 (gimple_divmod_fixed_value_transform): Update.
22087 (gimple_mod_pow2): Update.
22088 (gimple_mod_pow2_value_transform): Update.
22089 (gimple_mod_subtract): Update.
22090 (gimple_mod_subtract_transform): Update.
22091 (gimple_ic): Update.
22092 (gimple_stringop_fixed_value): Update.
22093 (gimple_stringops_transform): Update.
22094 * value-prof.h: Update.
22096 2017-06-29 Carl Love <cel@us.ibm.com>
22098 * config/rs6000/rs6000-c.c: Add support for built-in functions
22099 vector signed int vec_signed (vector float);
22100 vector signed long long vec_signed (vector double);
22101 vector signed int vec_signed2 (vector double, vector double);
22102 vector signed int vec_signede (vector double);
22103 vector signed int vec_signedo (vector double);
22104 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
22105 instruction generator.
22106 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22107 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
22108 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
22110 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
22111 vunsignede_v2df): Add define_expands.
22112 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
22113 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
22114 VEC_UNSIGNEDO): Add definitions.
22115 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22116 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
22117 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
22118 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
22119 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
22120 * config/rs6000/altivec.h (vec_signed, vec_signed2,
22121 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
22122 vec_unsignede, vec_unsignedo): Add builtin defines.
22123 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
22125 * doc/extend.texi: Update the built-in documentation file for the
22126 new built-in functions.
22128 2017-06-29 Richard Biener <rguenther@suse.de>
22130 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
22131 reduction chains to LOOP_VINFO_REDUCTIONS.
22132 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
22133 SLP reductions after processing reduction chains.
22135 2017-06-29 Nathan Sidwell <nathan@acm.org>
22137 * builtins.c (fold_builtin_FUNCTION): Use
22138 lang_hooks.decl_printable_name.
22140 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
22142 PR middle-end/81194
22143 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
22144 with only one label.
22145 * stmt.c (expand_case): Assert NCASES is greater than one.
22147 2017-06-29 Richard Biener <rguenther@suse.de>
22149 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
22151 (group_case_labels): Likewise.
22152 (find_taken_edge): Push sanity checking on val to workers...
22153 (find_taken_edge_cond_expr): ... here
22154 (find_taken_edge_switch_expr): ... and here, handle cases
22155 with just a default label.
22156 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
22157 (group_case_labels): Likewise.
22158 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
22159 group_case_labels does anything cleanup the CFG again.
22161 2017-06-29 Bin Cheng <bin.cheng@arm.com>
22163 PR tree-optimization/81196
22164 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
22165 exit condition comparing two IVs.
22167 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
22169 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
22170 profile to the dummy entry at the end of the list of architectures.
22171 * config/arm/arm-cpu-cdata.h: Regenerated.
22173 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22174 Michael Collison <michael.collison@arm.com>
22177 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
22179 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
22180 (*aarch64_reg_<mode>3_minus_mask): New pattern.
22181 (*aarch64_<optab>_reg_di3_mask2): New pattern.
22182 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
22183 of shift when the shift amount is masked with constant equal to
22184 the size of the mode.
22185 * config/aarch64/predicates.md (subreg_lowpart_operator): New
22188 2017-06-29 Martin Liska <mliska@suse.cz>
22190 * config/i386/i386.opt: Change range from [1,5] to [0,5].
22192 2017-06-29 Yury Gribov <tetra2005@gmail.com>
22195 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
22198 (edge_growth_cache_entry::edge_growth_cache_entry): New
22200 (reset_edge_growth_cache): Update to use constructor.
22202 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22204 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
22205 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22206 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22208 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
22210 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
22211 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
22213 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
22215 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
22216 (*-linux-uclibc*): Add t-uclibc tmake_file.
22217 * config/t-musl: New.
22218 * config/t-uclibc: New.
22220 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
22222 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
22224 (gen_comm_data): Emit architectural setting of arch_prof.
22225 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
22227 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
22228 (armv8-m.base, armv8-m.main): Likewise.
22229 * arm-protos.h (arm_build_target): Add profile field.
22230 (arch_option): Likewise.
22231 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
22233 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
22234 arm_active_target.profile.
22236 2017-06-28 Richard Biener <rguenther@suse.de>
22238 PR middle-end/81227
22239 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
22240 TYPE_OVERFLOW_WRAPS.
22241 * match.pd (negate_expr_p): Likewise.
22242 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
22243 fold_build2, not fold_binary.
22245 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
22247 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
22248 Convert memory address to Pmode.
22249 (aarch64_print_operand): Assert MEM operands are always Pmode.
22251 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
22254 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
22255 Remove redundant if.
22256 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
22257 * config/arm/aarch-common-protos.h
22258 (aarch_forward_to_shift_is_not_shifted_re): Remove.
22259 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
22261 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
22264 * multiple_target.c (create_dispatcher_calls): Set the default
22265 clone to be static, not public.
22267 2017-06-28 Richard Biener <rguenther@suse.de>
22269 * tree-vect-loop.c (vectorizable_reduction): Move special
22270 cond reduction IV var creation ...
22271 (vect_create_epilog_for_reduction): ... here. Remove induction_index
22272 parameter. Use STMT_VINFO_VECTYPE.
22273 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
22276 2017-06-28 Martin Liska <mliska@suse.cz>
22279 * ipa-visibility.c (non_local_p): Handle visibility.
22281 2017-06-28 Martin Liska <mliska@suse.cz>
22284 * common.opt: Add IntegerRange to various options.
22285 * opt-functions.awk (integer_range_info): New function.
22286 * optc-gen.awk: Add integer_range_info to cl_options struct.
22287 * opts-common.c (decode_cmdline_option): Handle
22288 CL_ERR_INT_RANGE_ARG.
22289 (cmdline_handle_error): Likewise.
22290 * opts.c (print_filtered_help): Show valid interval in
22291 when --help is provided.
22292 * opts.h (struct cl_option): Add range_min and range_max fields.
22293 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
22295 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
22297 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
22298 (x * C EQ/NE y * C): New transformation.
22300 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
22302 * genmultilib (combination_space): Accept '+' in option names.
22304 2017-06-28 Martin Liska <mliska@suse.cz>
22307 * asan.c (instrument_derefs): Bail out inner references
22308 that are hard register variables.
22310 2017-06-28 Jakub Jelinek <jakub@redhat.com>
22313 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
22314 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
22316 2017-06-28 Richard Biener <rguenther@suse.de>
22318 * tree-vectorizer.h (vect_get_vec_defs): Remove.
22319 (vect_get_slp_defs): Adjust.
22320 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
22322 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
22324 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
22325 get_initial_defs_for_reduction instead of vect_get_vec_defs.
22326 (vectorizable_reduction): Adjust.
22327 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
22329 (vect_get_slp_defs): Likewise.
22330 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
22331 (vectorizable_bswap): Adjust.
22332 (vectorizable_call): Likewise.
22333 (vectorizable_conversion): Likewise.
22334 (vectorizable_assignment): Likewise.
22335 (vectorizable_shift): Likewise.
22336 (vectorizable_operation): Likewise.
22337 (vectorizable_store): Likewise.
22338 (vectorizable_condition): Likewise.
22339 (vectorizable_comparison): Likewise.
22341 2017-06-28 Michael Collison <michael.collison@arm.com>
22344 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
22346 (arm_gen_movmemqi): Removed unused variable 'i'.
22347 Convert 'for' loop into 'while' loop.
22348 (arm_expand_prologue): Remove last unnecessary set of insn.
22349 (thumb_pop): Remove unused variable 'pushed_words'.
22350 (thumb_exit): Remove last unnecessary set of regs_to_pop.
22352 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22354 * config/s390/predicates.md: Use s390_rel_address_ok_p.
22355 * config/s390/s390-protos.h: Add prototype of
22356 s390_rel_address_ok_p.
22357 * config/s390/s390.c (s390_got_symbol): New function.
22358 (s390_rel_address_ok_p): New function.
22359 (legitimize_pic_address): Use s390_rel_address_ok_p.
22360 (s390_load_got): Use s390_got_symbol.
22361 (s390_option_override): Issue error if
22362 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
22363 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
22365 * config/s390/s390.opt: New option mpic-data-is-text-relative.
22367 2017-06-27 Andrew Pinski <apinski@cavium.com>
22369 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
22370 (X * copysign (1.0, X)): New pattern.
22371 (X * copysign (1.0, -X)): New pattern.
22372 (copysign (-1.0, CST)): New pattern.
22374 2017-06-27 Joseph Myers <joseph@codesourcery.com>
22376 * genmultilib (combination_space): Remove variable.
22377 Validate reuse rules against regular expression for any sequence
22378 of multilib options in any order.
22380 2017-06-27 Michael Collison <michael.collison@arm.com>
22382 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
22383 call aarch64_split_simd_combine.
22384 * (aarch64_combine_internal<mode>): Delete pattern.
22385 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
22386 Allow register and subreg operands.
22388 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22390 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
22391 specific need, just fallback on defaults.
22392 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
22394 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22395 Olivier Hainque <hainque@adacore.com>
22397 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
22399 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
22400 targets. Pick a default if no particular attempt applied.
22401 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
22404 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22406 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
22407 (x86_64-wrs-vxworks7): Likewise.
22409 2017-06-27 Marek Polacek <polacek@redhat.com>
22412 * ubsan.c (instrument_null): Check get_base_address's result for null.
22414 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22416 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
22418 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22420 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
22421 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
22422 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
22423 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
22424 New function types.
22425 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
22426 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
22427 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
22428 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
22429 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
22430 BUILT_IN_FEUPDATEENV): New builtins.
22431 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
22432 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
22433 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
22434 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
22436 (builtin_structptr_types): Adjust size.
22437 * tree.c (builtin_structptr_types): Add four entries.
22439 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22440 Olivier Hainque <hainque@adacore.com>
22442 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
22443 (TLS_SYM): New local macro, forcing reference to __tls__ on
22444 link command lines for VxWorks 7 RTPs, triggering initialization
22446 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
22447 OS features TLS support, true for RTPs on VxWorks 7.
22448 * config/vxworks.c (vxworks_override_options): Setup emutls
22451 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22453 * predict.c (test_prediction_value_range): Use -1U instead of -1
22454 to avoid narrowing conversion warning.
22455 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
22456 to avoid narrowing conversion warning.
22457 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
22459 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
22461 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22463 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
22464 64bit configurations.
22465 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
22466 (SIZE_TYPE): Likewise.
22467 * config/vxworks.c (vxworks_emutls_var_fields): Use
22468 long_unsigned_type_node instead of unsigned_type_node as the offset
22469 field type, which is "pointer" mode in emutls.c.
22471 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22474 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
22476 PR middle-end/81207
22477 * gimple-fold.c (replace_call_with_call_and_fold): Handle
22478 gimple_vuse copying separately from gimple_vdef copying.
22480 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22482 * value-prof.c (free_hist): Remove call to memset and the enclosing if
22485 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22486 Olivier Hainque <hainque@adacore.com>
22488 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
22489 for all vxworks7 targets.
22490 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
22491 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
22492 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
22493 variations for VX6/VX7 and 32/64bits later on in ...
22494 (VXWORKS_LIB_SPEC): Leverage new macros.
22495 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
22496 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
22498 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22500 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
22501 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
22503 2017-06-26 Carl Love <cel@us.ibm.com>
22505 * config/rs6000/rs6000-c.c: Add support for built-in functions
22506 vector bool char vec_reve (vector bool char);
22507 vector signed char vec_reve (vector signed char);
22508 vector unsigned char vec_reve (vector unsigned char);
22509 vector bool int vec_reve (vector bool int);
22510 vector signed int vec_reve (vector signed int);
22511 vector unsigned int vec_reve (vector unsigned int);
22512 vector bool long long vec_reve (vector bool long long);
22513 vector signed long long vec_reve (vector signed long long);
22514 vector unsigned long long vec_reve (vector unsigned long long);
22515 vector bool short vec_reve (vector bool short);
22516 vector signed short vec_reve (vector signed short);
22517 vector double vec_reve (vector double);
22518 vector float vec_reve (vector float);
22519 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
22520 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
22521 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
22522 (altivec_vreve): New pattern.
22523 * config/rs6000/altivec.h (vec_reve): New define.
22524 * doc/extend.texi (vec_rev): Update the built-in documentation file
22525 for the new built-in functions.
22527 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22529 PR tree-optimization/71815
22530 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
22532 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
22534 (slsr_process_phi): Likewise.
22535 (replace_uncond_cands_and_profitable_phis): Don't replace a
22536 multiply candidate with a stride of 1 (copy or cast).
22537 (phi_incr_cost): Call uses_consumed_by_stmt rather than
22539 (lowest_cost_path): Likewise.
22540 (total_savings): Likewise.
22542 2017-06-26 Richard Biener <rguenther@suse.de>
22545 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
22546 Use def_builtin_pure for all gather builtins.
22548 2017-06-26 Richard Biener <rguenther@suse.de>
22550 PR tree-optimization/81203
22551 * tree-tailcall.c (find_tail_calls): Do not move stmts into
22552 non-dominating BBs.
22554 2017-06-26 Marek Polacek <polacek@redhat.com>
22557 * doc/invoke.texi: Document -Wmultistatement-macros.
22559 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
22561 * doc/sourcebuild.texi (ARM-specific attributes): Document new
22562 arm_neon_ok_no_float_abi effective target.
22564 2017-06-26 Richard Biener <rguenther@suse.de>
22566 PR tree-optimization/80928
22567 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
22568 (copy_bbs): Set BB_DUPLICATED flag early.
22569 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
22571 (execute_on_shrinking_pred): Likewise.
22572 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
22573 BB_DUPLICATED blocks.
22574 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
22575 iterate over all PHIs considering removal of *gsi.
22577 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
22579 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
22582 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22584 * config/rs6000/rs6000-string.c: (expand_block_clear,
22585 do_load_for_compare, select_block_compare_mode,
22586 compute_current_alignment, expand_block_compare,
22587 expand_strncmp_align_check, expand_strn_compare,
22588 expand_block_move, rs6000_output_load_multiple)
22589 Move functions related to string/block move/compare
22590 to a separate file.
22591 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
22592 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
22593 for this function which is now used in two files.
22594 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
22595 * config.gcc: Add rs6000-string.o to extra_objs for
22596 targets powerpc*-*-* and rs6000*-*-*.
22598 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
22601 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
22602 32-bit, since indexed is not valid for DImode.
22603 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
22604 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
22605 (define_peephole2 for Altivec d-form load): Add 32-bit support.
22606 (define_peephole2 for Altivec d-form store): Likewise.
22609 * multiple_target.c (create_dispatcher_calls): Only create the
22610 dispatcher call if the function is the default clone of a
22611 versioned function.
22613 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
22615 PR middle-end/80902
22616 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
22617 a call, force the call to not be a tail call.
22619 2017-06-23 Jeff Law <law@redhat.com>
22621 * doc/contrib.texi: Add entry for Steven Pemberton's work on
22624 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
22626 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
22627 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
22628 handling for early expansion of vector shifts (sl,sr,sra,rl).
22629 (builtin_function_type): Add vector shift right instructions
22630 to the unsigned argument list.
22632 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
22634 rtl-optimizatoin/79286
22635 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
22636 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
22637 trap. PIC register plus a const unspec without offset can never trap.
22639 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
22641 * tree.h (builtin_structptr_type): New type.
22642 (builtin_structptr_types): Declare new array.
22643 * tree.c (builtin_structptr_types): New array.
22644 (free_lang_data, build_common_tree_nodes): Use it.
22646 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
22649 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
22652 2017-06-22 Matt Turner <mattst88@gmail.com>
22654 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
22655 Lake models to skylake case. Assume skylake for unknown
22656 models with clflushopt.
22658 2017-06-22 Jeff Law <law@redhat.com>
22660 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
22661 frame sizes that do not satisfy aarch64_uimm12_shift.
22663 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
22665 * profile-count.h (apply_probability,
22666 apply_scale, probability_in): Fix checks for zero.
22668 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22670 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
22671 * doc/cppdiropts.texi (-I @var{dir}): Document it.
22673 2016-06-22 Richard Biener <rguenther@suse.de>
22675 * tree-vect-loop.c (vect_model_reduction_cost): Handle
22676 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
22677 REDUC_MAX_EXPR support.
22678 (vectorizable_reduction): Likewise.
22679 (vect_create_epilog_for_reduction): Likewise.
22681 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
22683 * match.pd (A / (1 << B) -> A >> B): New.
22684 * generic-match-head.c: Include optabs-tree.h.
22685 * gimple-match-head.c: Likewise.
22686 * optabs-tree.h (target_supports_op_p): New.
22687 * optabs-tree.c (target_supports_op_p): New.
22689 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22691 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
22692 $gcc_cv_ld --help output.
22693 (gcc_cv_ld_demangle): Likewise.
22694 (gcc_cv_ld_eh_frame_hdr): Likewise.
22695 (gcc_cv_ld_pie): Likewise.
22696 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
22697 (gcc_cv_ld_buildid): Likewise.
22698 (gcc_cv_ld_sysroot): Likewise.
22699 (ld_bndplt_support): Likewise.
22700 (ld_pushpopstate_support): Likewise.
22701 * configure: Regenerate.
22702 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
22704 2017-06-21 Jakub Jelinek <jakub@redhat.com>
22707 * config/i386/sse.md (round<mode>2): Renumber match_dup and
22708 operands indexes to avoid gap between operands and match_dups.
22710 2017-06-21 Andrew Pinski <apinski@cavium.com>
22712 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
22713 Increment Arith_shift and Arith_shift_reg by 1.
22714 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
22716 * config/aarch64/aarch64.c (thunderx_tunings): Enable
22717 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
22718 (aarch64_strip_extend): Add new argument and test for it.
22719 (aarch64_cheap_mult_shift_p): New function.
22720 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
22721 add a cost if it is true.
22722 Update calls to aarch64_strip_extend.
22723 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
22725 2017-06-21 Andrew Pinski <apinski@cavium.com>
22727 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
22729 (thunderxt88): Likewise.
22730 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
22731 (thunderx_prefetch_tune): New variable.
22732 (thunderx2t99_prefetch_tune): Update for the correct values.
22733 (thunderxt88_tunings): New variable.
22734 (thunderx_tunings): Use thunderx_prefetch_tune instead of
22735 generic_prefetch_tune.
22736 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
22738 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22740 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
22741 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
22742 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
22743 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
22744 (aarch64_atomic_cas<mode>, GPI): Likewise.
22746 2017-06-21 Martin Liska <mliska@suse.cz>
22748 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
22749 statements on cold and hot labels.
22750 * predict.c (tree_estimate_probability_bb): Remove the
22751 prediction from this place.
22753 2017-06-21 Martin Liska <mliska@suse.cz>
22755 PR tree-optimization/79489
22756 * gimplify.c (maybe_add_early_return_predict_stmt): New
22758 (gimplify_return_expr): Call the function.
22759 * predict.c (tree_estimate_probability_bb): Remove handling
22761 * predict.def: Update comment about early return predictor.
22762 * gimple-predict.h (is_gimple_predict): New function.
22763 * predict.def: Change default value of early return to 66.
22764 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
22766 * passes.def: Put pass_strip_predict_hints to the beginning of
22769 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
22771 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
22772 FUNCTION_DECL declarations.
22773 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
22775 (dwaf2out_decl): Likewise.
22776 * godump.c (go_early_global_decl): Skip call to the real debug hook
22777 for FUNCTION_DECL declarations.
22778 * passes.c (rest_of_decl_compilation): Skip call to the
22779 early_global_decl debug hook for FUNCTION_DECL declarations, unless
22780 -fdump-go-spec is passed.
22782 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22784 * config/i386/i386.c (struct builtin_isa): New field pure_p.
22785 Reorder for compactness.
22786 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
22787 (def_builtin_pure, def_builtin_pure2): New functions.
22788 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
22790 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22792 * match.pd (nop_convert): New predicate.
22793 ((A +- CST1) +- CST2): Allow some NOP conversions.
22795 2017-06-21 Jakub Jelinek <jakub@redhat.com>
22798 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
22799 with ctors/dtors if GOVD_SHARED is set.
22801 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22803 * config/aarch64/aarch64.md (movti_aarch64):
22804 Emit mov rather than orr.
22805 (movtf_aarch64): Likewise.
22806 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
22807 Emit mov rather than orr.
22809 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22811 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
22812 Swap alternatives, make integer dup more expensive.
22814 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22816 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
22817 Return true for non-tls symbols.
22819 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
22821 * config/aarch64/aarch64-cores.def (cortex-a55): New.
22822 (cortex-a75): Likewise.
22823 (cortex-a75.cortex-a55): Likewise.
22824 * config/aarch64/aarch64-tune.md: Regenerate.
22825 * doc/invoke.texi (-mtune): Document new values for -mtune.
22827 2017-06-21 Tom de Vries <tom@codesourcery.com>
22829 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
22830 stack_size feature.
22831 (Effective-Target Keywords, Other attributes): Suggest using
22832 dg-add-options stack_size feature to get stack limit in stack_size
22833 effective target documentation.
22835 2017-06-21 Julian Brown <julian@codesourcery.com>
22836 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22838 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
22839 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
22840 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
22842 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
22843 attribute type list for neon_multiply.
22844 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
22845 attribute type list for neon_multiply.
22846 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
22847 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
22848 attribute type list for neon_multiply.
22849 * config/arm/types.md (crypto_pmull): Add.
22850 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
22851 attribute type list.
22853 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
22855 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
22858 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22860 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
22861 to make sure not to dereference a NULL cost_classes_ptr pointer.
22863 2017-06-20 Carl Love <cel@us.ibm.com>
22865 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22866 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22867 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22868 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22869 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22870 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22871 VMULOSW): New enum "unspec" values.
22872 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22873 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22874 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22875 altivec_vmulosw): New patterns.
22876 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22877 VMULOSW): Add definitions.
22879 2017-06-20 Julia Koval <julia.koval@intel.com>
22881 * config/i386/i386.c: Fix rounding expand for new pattern.
22882 * config/i386/subst.md: Fix pattern (parallel -> unspec).
22884 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22886 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
22887 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
22889 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22891 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
22894 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22896 * config/aarch64/aarch64-cores.def: Rearrange to sort by
22897 architecture, then by implementer ID.
22898 * config/aarch64/aarch64-tune.md: Regenerate.
22900 2017-06-20 Richard Biener <rguenther@suse.de>
22902 PR middle-end/81097
22903 * fold-const.c (split_tree): Fold to type before negating.
22905 2017-06-20 David Malcolm <dmalcolm@redhat.com>
22907 * diagnostic-show-locus.c
22908 (selftest::test_fixit_deletion_affecting_newline): New function.
22909 (selftest::diagnostic_show_locus_c_tests): Call it.
22911 2017-06-20 Andreas Schwab <schwab@suse.de>
22914 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
22917 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22919 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
22920 __ARM_FEATURE_COPROC according to support.
22922 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22924 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22925 Rewritten to avoid overflow for > 32-bit pointers.
22928 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
22929 by removing enum keyword.
22930 (ubsan_type_descriptor): Likewise. Formatting fix.
22933 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
22934 splitter): Require TARGET_SSE2 in the condition.
22936 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
22939 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
22940 for doing vector set of SFmode on ISA 3.0.
22941 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
22942 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
22944 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
22945 SFmode value into a V4SF variable that was extracted from another
22946 V4SF variable without converting the element to double precision
22947 and back to single precision vector format.
22948 (vsx_insert_extract_v4sf_p9_2): Likewise.
22950 2017-06-19 Jakub Jelinek <jakub@redhat.com>
22952 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
22953 in UWHI to avoid undefined overflow.
22956 * ubsan.h (enum ubsan_encode_value_phase): New.
22957 (ubsan_encode_value): Change second argument to
22958 enum ubsan_encode_value_phase with default value of
22959 UBSAN_ENCODE_VALUE_GENERIC.
22960 * ubsan.c (ubsan_encode_value): Change second argument to
22961 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
22962 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
22963 create_tmp_var_raw instead of create_tmp_var and use a
22965 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
22966 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
22967 ubsan_encode_value callers.
22970 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
22971 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
22972 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
22974 2017-06-19 Richard Biener <rguenther@suse.de>
22976 PR middle-end/81118
22977 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
22978 estimates if we changed anything.
22980 2017-06-19 Richard Biener <rguenther@suse.de>
22982 PR tree-optimization/80887
22983 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
22984 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
22985 simplified lookups, then reset mprts_hook.
22986 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
22988 (try_to_simplify): Likewise.
22990 2017-06-19 Martin Liska <mliska@suse.cz>
22993 * gimplify.c (gimplify_switch_expr):
22994 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
22996 2017-06-19 Martin Liska <mliska@suse.cz>
22998 * doc/install.texi: Document that PGO runs in 4 stages.
23000 2017-06-19 Martin Liska <mliska@suse.cz>
23003 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
23004 to dispatcher function name.
23005 * multiple_target.c (replace_function_decl): New function.
23006 (create_dispatcher_calls): Redirect both edges and references.
23008 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
23010 * profile-count.c (profile_count::dump): Dump quality.
23011 (profile_count::differs_from_p): Update for unsigned val.
23012 * profile-count.h (profile_count_quality): New enum.
23013 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
23015 2017-06-19 Richard Biener <rguenther@suse.de>
23017 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
23018 struct function as arg.
23019 (estimate_numbers_of_iterations): Export overload with loop arg.
23020 (free_numbers_of_iterations_estimates_loop): Use an overload of
23021 free_numbers_of_iterations_estimates instead.
23022 * tree-cfg.c (remove_bb): Adjust.
23023 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
23024 * tree-parloops.c (gen_parallel_loop): Likewise.
23025 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
23027 (tree_unroll_loops_completely): Likewise.
23028 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
23029 Use an overload instead and export.
23030 (estimated_loop_iterations): Adjust.
23031 (max_loop_iterations): Likewise.
23032 (likely_max_loop_iterations): Likewise.
23033 (estimate_numbers_of_iterations): Take struct function as arg
23035 (loop_exits_before_overflow): Adjust.
23036 (free_numbers_of_iterations_estimates_loop): Use an overload.
23037 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
23038 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
23040 2017-06-19 Richard Biener <rguenther@suse.de>
23043 * ipa-prop.c (find_constructor_constant_at_offset): Handle
23044 RANGE_EXPR conservatively.
23046 2017-06-16 Carl Love <cel@us.ibm.com>
23048 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23049 definitions for vec_float, vec_float2, vec_floato,
23050 vec_floate built-ins.
23051 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
23052 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
23054 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
23055 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
23056 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
23057 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
23058 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
23059 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
23060 vec_floato): Add builtin defines.
23061 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
23062 Update the built-in documentation file for the new built-in
23065 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23067 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
23068 (mthumb): Mark as the negative of -marm.
23070 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23072 * doc/invoke.texi (ARM Options, -mcpu): Document supported
23074 (ARM Options, -mtune): Document that this accepts the same
23075 extension options as -mcpu.
23076 (ARM Options, -mfpu): Document addition of -mfpu=auto.
23078 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23080 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
23081 permitted extensions.
23083 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23085 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
23086 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
23087 (armv8-m.main): Add option +nodsp.
23088 * config/arm/arm-cpu-cdata.h: Regenerated.
23090 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23092 * config/arm/t-fuchsia: New file.
23093 * config.gcc (arm*-*-fuchsia*): Use it.
23095 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23097 * config/arm/t-symbian: Rewrite for new option infrastructure.
23099 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23101 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
23102 (MULTILIB_REQUIRED): Likewise.
23104 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23106 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
23107 (MULTILIB_RESUE): Likewise.
23108 (MULTILIB_MATCHES): Likewise.
23109 (MULTLIB_REQUIRED): Likewise.
23111 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23113 * config/arm/t-rtems: Rewrite for new option framework.
23115 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23117 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
23118 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
23119 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
23120 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
23121 * config/arm/t-multilib: ... here.
23122 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
23123 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
23124 armv7-a and armv8*-a when A-profile libraries have not been built.
23125 * config/arm/t-rmprofile: Rewrite.
23127 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23129 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
23130 with a backslash. Remove the backslash after substituting unescaped
23132 * doc/fragments.texi (MULTILIB_REUSE): Document it.
23134 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23136 * config.gcc: (arm*-*-*): When building a-profile libraries, force
23137 the driver to pass through the default setting of -mfloat-abi.
23138 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
23140 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
23141 (all_feat_combs): New rule.
23142 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
23144 * config/arm/t-aprofile: Rewrite.
23146 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23148 * config/arm/arm.h (FPUTYPE_AUTO): Define.
23149 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
23150 fpu is not specified by the user/command-line.
23151 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
23152 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
23153 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
23154 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
23155 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
23156 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
23158 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23160 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
23161 * config/arm/t-arm-elf: Rewritten.
23163 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23165 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
23166 have some floating-point instructions.
23167 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
23168 (TARGET_MAYBE_HARD_FLOAT): New macro.
23169 * config/arm/arm-builtins.c (arm_init_builtins): Use
23170 TARGET_MAYBE_HARD_FLOAT.
23171 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
23173 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23175 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
23176 (configargs.h): Include it.
23177 (arm_print_hint_for_fpu_option): New function.
23178 (arm_parse_fpu_option): New function.
23179 (candidate_extension): New class.
23180 (arm_canon_for_multilib): New function.
23181 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
23182 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
23183 (ARCH_CANONICAL_SPECS): New macro.
23184 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
23186 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23188 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
23189 are set after handling multilib fragments. Set target_cpu_default2
23192 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23194 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
23196 (arm*-*-*): Set target_cpu_default2 to a quoted string.
23197 * config/arm/parsecpu.awk (check_cpu): Validate any extension
23199 (check_arch): Likewise.
23200 * config/arm/arm.c (arm_configure_build_target): Handle
23201 TARGET_CPU_DEFAULT being a string constant. Scan any feature
23202 options in the default.
23204 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23206 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
23207 when an option is an alias of another.
23208 * config/arm/parsecpu.awk (optalias): New parser token.
23209 (gen_comm_data): Mark non-alias options as such. Emit entries
23210 for extension aliases.
23211 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
23212 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
23213 (armv6kz, armv6zk, armv6t2): Likewise.
23214 (armv7): Make vfpv3-d16 an alias.
23215 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
23217 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
23218 Sort in canonical order.
23219 (armv8-a): Sort in canonical order.
23220 (armv8.1-a, armv8.2-a): Likewise.
23221 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
23223 (cortex-a9): Sort in canonical order.
23224 * config/arm/arm.c (selftests.h): Include it.
23225 (arm_test_cpu_arch_data): New function.
23226 (arm_run_self_tests): New function.
23227 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
23228 (targetm): Move declaration to the end of the file.
23229 * arm-cpu-cdata.h: Regenerated.
23231 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23233 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
23234 call to target_mode_check describing the type of option passed.
23235 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
23236 (arm_target_thumb_only): Use arm_parse_arch_option_name or
23237 arm_parse_cpu_option_name to match parameters against list of
23239 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
23240 arm_arch_core_flags data structure.
23241 * config/arm/arm-cpu_cdata.h: Regenerated.
23243 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23245 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
23247 (arm_print_hint_for_cpu_option): Likewise.
23248 (arm_print_hint_for_arch_option): Likewise.
23249 (arm_parse_cpu_option_name): Likewise.
23250 (arm_parse_arch_option_name): Likewise.
23251 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
23252 of entries in the all_fpus list.
23253 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
23254 (arm_parse_cpu_option_name): Declare.
23255 (arm_parse_arch_option_name): Declare.
23256 (arm_parse_option_features): Declare.
23257 (arm_intialize_isa): Declare.
23258 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
23260 (gen_comm_data): ... here. Make definitions non-static.
23261 * config/arm/arm-cpu-data.h: Regenerated.
23262 * config/arm/arm-cpu-cdata.h: Regenerated.
23264 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23266 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
23267 (cpu_arch_extension): New structure.
23268 (cpu_arch_option, arch_option, cpu_option): New structures.
23269 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
23270 architecture types.
23271 (gen_data): Generate new format data tables.
23272 * config/arm/arm.c (cpu_tune): New structure.
23273 (cpu_option, processors): Delete.
23274 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
23275 (arm_print_hint_for_cpu_option): ... this and ...
23276 (arm_print_hint_for_arch_option): ... this.
23277 (arm_parse_arch_cpu_name): Delete. Replace with ...
23278 (arm_parse_cpu_option_name): ... this and ...
23279 (arm_parse_arch_option_name): ... this.
23280 (arm_unrecognized_feature): Change type of target parameter to
23282 (arm_parse_arch_cpu_features): Delete. Replace with ...
23283 (arm_parse_option_features): ... this.
23284 (arm_configure_build_target): Rework to use new configuration data
23286 (arm_print_tune_info): Rework for new configuration data tables.
23287 * config/arm/arm-cpu-data.h: Regenerated.
23288 * config/arm/arm-cpu.h: Regenerated.
23290 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23292 * Makefile.in (OBJS): Move sbitmap.o from here ...
23293 (OBJS-libcommon): ... to here.
23295 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23297 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
23298 (ISA_ALL_CRYPTO): New macro.
23299 (ISA_ALL_SIMD): New macro
23300 (ISA_ALL_FP): New macro.
23301 * config/arm/arm.c (fpu_bitlist): Update initializer.
23302 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
23304 (arm9e): Add fpu. Add option for nofp
23305 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
23306 (arm926ej-s, arm1026ej-s): Likewise.
23307 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
23308 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
23309 neon-fp16, neon-vfpv4, nofp and nosimd.
23310 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
23311 (cortex-a8): Add fpu. Add option for nofp.
23312 (cortex-a9): Add fpu. Add options for nosimd and nofp.
23313 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
23314 (cortex-r4f): Add fpu.
23315 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
23316 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
23318 (cortex-r8): Likewise.
23319 (cortex-m4): Add fpu. Add option for nofp.
23320 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
23321 (cortex-a17.cortex-a7): Likewise.
23322 (cortex-a32): Add fpu. Add options for crypto and nofp.
23323 (cortex-a35, cortex-a53): Likewise.
23324 (cortex-a57): Add fpu. Add option for crypto.
23325 (cortex-a72, cortex-a73): Likewise.
23326 (exynos-m1): Likewise.
23327 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
23328 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
23329 (cortex-m33): Add fpu. Add option for nofp.
23330 * config/arm/arm-cpu-cdata.h: Regenerated
23331 * config/arm/arm-cpu-data.h: Regenerated.
23333 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23335 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
23336 (armv5te, armv5tej): Likewise.
23337 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
23338 (armv7): Add options fp and vfpv3-d16.
23339 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
23340 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
23342 (armv7ve): Likewise.
23343 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
23344 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
23345 (armv8-a): Add nocrypto option.
23346 (armv8.1-a, armv8.2-a): Likewise.
23347 (armv8-m.main): add options fp, fp.dp and nofp.
23349 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23351 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
23353 (armv8-a+crc): Delete.
23354 (armv8.1-a): Add options simd, crypto and nofp.
23355 (armv8.2-a): Add options fp16, simd, crypto and nofp.
23356 (armv8.2-a+fp16): Delete.
23357 (armv8-m.main): Add option dsp.
23358 (armv8-m.main+dsp): Delete.
23359 (cortex-a8): Add fpu. Add nofp option.
23360 (cortex-a9): Add fpu. Add nofp and nosimd options.
23361 * config/arm/parsecpu.awk (gen_data): Generate option tables and
23362 link to main cpu and architecture data structures.
23363 (gen_comm_data): Only put isa attributes from the main architecture
23365 (option): New statement for architecture and CPU entries.
23366 * arm.c (struct cpu_option): New structure.
23367 (struct processors): Add entry for options.
23368 (arm_unrecognized_feature): New function.
23369 (arm_parse_arch_cpu_name): Ignore any characters after the first
23371 (arm_parse_arch_cpu_feature): New function.
23372 (arm_configure_build_target): Separate out any CPU and architecture
23373 features and parse separately. Don't error out if -mfpu=auto is
23374 used with only an architecture string.
23375 (arm_print_asm_arch_directives): New function.
23376 (arm_file_start): Call it.
23377 * config/arm/arm-cpu-cdata.h: Regenerated.
23378 * config/arm/arm-cpu-data.h: Likewise.
23379 * config/arm/arm-tables.opt: Likewise.
23381 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23383 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
23384 assembler when it is not -mfpu=auto.
23386 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23388 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
23389 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
23390 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
23391 (ASM_CPU_SPEC): Rewrite.
23392 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
23393 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
23394 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
23395 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
23396 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
23397 copied string is NUL-terminated. Also strip any characters prefixed
23399 (arm_rewrite_selected_arch): New function.
23400 (arm_rewrite_march): New function.
23402 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23404 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
23405 (x_arm_cpu_string, x_arm_tune_string): Likewise.
23406 (march, mcpu, mtune): Convert to string-based options.
23407 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
23408 (arm_parse_arch_cpu_name): New function.
23409 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
23410 identify selected architecture or CPU.
23411 (arm_option_save): New function.
23412 (TARGET_OPTION_SAVE): Redefine.
23413 (arm_option_restore): Restore string options.
23414 (arm_option_print): Print string options.
23416 2017-06-16 Martin Sebor <msebor@redhat.com>
23418 PR tree-optimization/80933
23419 PR tree-optimization/80934
23420 * builtins.c (fold_builtin_3): Do not handle bcmp here.
23421 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
23422 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
23423 (gimple_fold_builtin): Call them.
23425 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23427 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
23428 as unlikely; update profile.
23430 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23432 * predict.c (force_edge_cold): Handle declaring edges impossible
23435 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23437 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
23439 (try_unroll_loop_completely): Fix reporting.
23441 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23443 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
23445 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
23448 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
23449 if given a non-constant argument for an intrinsic which requires a
23452 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23454 * profile.c (compare_freqs): New function.
23455 (branch_prob): Sort edge list.
23456 (find_spanning_tree): Assume that the list is priority sorted.
23458 2017-06-16 Richard Biener <rguenther@suse.de>
23460 PR tree-optimization/81090
23461 * passes.def (pass_record_bounds): Remove.
23462 * tree-pass.h (make_pass_record_bounds): Likewise.
23463 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
23464 make_pass_record_bounds): Likewise.
23465 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
23466 not free niter estimates at the beginning but at the end.
23467 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
23469 2017-06-16 Richard Biener <rguenther@suse.de>
23471 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
23472 initializer to workaround ICE in host GCC 4.8.
23474 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23476 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
23478 (clone_inlined_nodes): Update.
23480 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23482 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
23483 prefetch settings, and enable prefetching by default at -O3.
23485 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23487 * config/aarch64/aarch64.c (aarch64_override_options_internal):
23488 Set flag_prefetch_loop_arrays according to tuning data.
23490 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23492 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
23493 New tune structure.
23494 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
23495 [Unrelated to main purpose of the patch] Place the pointer field last
23496 to enable type checking errors when tune structure are wrongly merged.
23497 * config/aarch64/aarch64.c (generic_prefetch_tune,)
23498 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
23499 (thunderx2t99_prefetch_tune): New tune constants.
23500 (tune_params *_tunings): Update all tunings (no functional change).
23501 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
23502 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
23503 from tunings structures.
23505 2017-06-16 Jakub Jelinek <jakub@redhat.com>
23508 * ubsan.c (instrument_null): Add T argument, use it instead
23509 of computing it based on IS_LHS.
23510 (instrument_object_size): Likewise.
23511 (pass_ubsan::execute): Adjust instrument_null and
23512 instrument_object_size callers to pass gimple_get_lhs or
23513 gimple_assign_rhs1 result to it. Use instrument_null instead of
23514 calling get_base_address and instrument_mem_ref. Handle
23515 aggregate call arguments for object-size sanitization.
23517 2017-06-16 Yury Gribov <tetra2005@gmail.com>
23519 PR tree-optimization/81089
23520 * tree-vrp.c (is_masked_range_test): Validate operands of
23523 2017-06-15 Martin Sebor <msebor@redhat.com>
23526 * dumpfile.c (dump_register): Avoid calling memset to initialize
23527 a class with a default ctor.
23528 * gcc.c (struct compiler): Remove const qualification.
23529 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
23530 * hash-table.h: Ditto.
23531 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
23533 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
23534 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
23536 * params.h (struct param_info): Make struct members non-const.
23537 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
23538 with default initialization.
23539 * vec.h (vec_copy_construct, vec_default_construct): New helper
23541 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
23542 with vec_copy_construct.
23543 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
23544 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
23545 * doc/invoke.texi (-Wclass-memaccess): Document.
23547 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23549 * emit-rtl.h (is_leaf): Update comment about local
23550 register allocator.
23552 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23555 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
23556 for a variable to have a section before checking if the section has a
23558 Set section to.persistent if persistent attribute is set.
23559 Warn if .persistent attribute is used on an automatic variable.
23561 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
23563 PR rtl-optimization/80474
23564 * reorg.c (update_block): Do not ignore instructions in a delay slot.
23566 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
23568 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
23571 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
23573 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
23574 (casesi): Emit bounds checking as RTL.
23575 (casesi_internal_mips16_<mode>): Remove bounds checking.
23577 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
23579 * config/xtensa/xtensa.c (xtensa_option_override): Append
23580 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
23581 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
23582 xtensa_doloop_hooks): Define unconditionally.
23583 (xtensa_reorg_loops): Only call reorg_loops in the presence of
23585 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
23586 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
23587 for it in xtensa_option_override.
23588 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
23589 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
23591 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
23593 * doc/cppopts.texi: Document '-' special value to -MF.
23595 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
23597 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
23598 (cortex_a53_fconst): Likewise.
23599 (cortex_a53_fpmul): Likewise.
23600 (cortex_a53_f_load_64): Likewise.
23601 (cortex_a53_f_load_many): Likewise.
23602 (cortex_a53_advsimd_alu): Likewise.
23603 (cortex_a53_advsimd_alu_q): Likewise.
23604 (cortex_a53_advsimd_mul): Likewise.
23605 (cortex_a53_advsimd_mul_q): Likewise.
23606 (fpmac bypass): Add new bypass for fpmac-fpmac case.
23607 Add missing fmul, r2f_cvt and fconst cases.
23609 2017-06-14 Richard Biener <rguenther@suse.de>
23611 PR middle-end/81088
23612 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
23614 (fold_binary_loc): When associating do not treat pre-existing
23615 TREE_OVERFLOW on literal constants as a reason to allow
23616 TREE_OVERFLOW on associated literal constants.
23618 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23620 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
23621 (MASK_FEATURES): New macro.
23622 * config/sparc/sparc.c (sparc_option_override): Remove the special
23623 handling of -mfpu and generalize it to all MASK_FEATURES switches.
23625 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23627 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
23628 a division of 0 if non-call exceptions are enabled.
23630 2017-06-14 Andrew Pinski <apinski@cavium.com>
23631 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23634 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
23635 Improve vector initialization code gen for only variable case.
23637 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23639 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
23641 2017-06-14 Richard Biener <rguenther@suse.de>
23643 PR tree-optimization/81083
23644 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
23647 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23649 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
23650 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
23651 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
23652 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
23653 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
23654 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
23656 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23658 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
23659 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
23661 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23663 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
23665 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23667 * config/rs6000/t-rtems: Don't handle SPE.
23669 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23671 * config/rs6000/t-linux: Don't handle SPE.
23673 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23675 * config/rs6000/eabispe.h: Delete file.
23677 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23679 * config/rs6000/t-spe: Delete file.
23681 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23683 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
23684 (rs6000_legitimate_offset_address_p): Return false for anything in
23685 V2SImode or V2SFmode.
23687 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23689 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
23690 except V2SF and V2SI. Rearrange the vector modes, and add comments.
23691 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
23693 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
23694 (rs6000_legitimate_offset_address_p): Ditto.
23695 (rs6000_emit_move): Ditto.
23696 (rs6000_init_builtins): Remove V4HI_type_node.
23698 2017-06-13 Martin Liska <mliska@suse.cz>
23701 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
23702 (gate_asan): Likewise.
23703 * asan.h (asan_no_sanitize_address_p): Remove the function.
23704 (sanitize_flags_p): New function.
23705 * builtins.def: Fix coding style.
23706 * common.opt: Use renamed enum value.
23707 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
23708 * doc/extend.texi: Document no_sanitize attribute.
23709 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
23710 to SANITIZE_UNDEFINED_NONDEFAULT.
23711 * gcc.c (sanitize_spec_function): Use the renamed enum value.
23712 * gimple-fold.c (optimize_atomic_compare_exchange_p):
23713 Use sanitize_flags_p.
23714 * gimplify.c (gimplify_function_tree): Likewise.
23715 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
23716 * opts.c (parse_no_sanitize_attribute): New function.
23717 (common_handle_option): Use renamed enum value.
23718 * opts.h (parse_no_sanitize_attribute): Declare.
23719 * tree.c (sanitize_flags_p): New function.
23720 * tree.h: Declared here.
23721 * tsan.c: Use sanitize_flags_p.
23722 * ubsan.c (ubsan_expand_null_ifn): Likewise.
23723 (instrument_mem_ref): Likewise.
23724 (instrument_bool_enum_load): Likewise.
23725 (do_ubsan_in_current_function): Remove the function.
23726 (pass_ubsan::execute): Use sanitize_flags_p.
23727 * ubsan.h: Remove do_ubsan_in_current_function
23728 * tree-cfg.c (print_no_sanitize_attr_value): New function.
23729 (dump_function_to_file): Use it here.
23731 2017-06-13 Martin Jambor <mjambor@suse.cz>
23733 PR tree-optimization/80803
23734 PR tree-optimization/81063
23735 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
23736 (propagate_subaccesses_across_link): Enqueue subtree whenever
23737 necessary instead of relying on the caller.
23739 2017-06-13 Martin Jambor <mjambor@suse.cz>
23741 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
23742 that have a first_link.
23743 (sort_and_splice_var_accesses): Do not check first_link before
23745 (subtree_mark_written_and_enqueue): Likewise.
23746 (propagate_all_subaccesses): Likewise and do not stop at first
23747 parent with a first_link.
23749 2017-06-13 Martin Jambor <mjambor@suse.cz>
23751 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
23754 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23756 * match.pd: New pattern.
23758 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23760 * tree-vrp.c (is_masked_range_test): New function.
23761 (register_edge_assert_for): Determine ranges for
23764 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23766 PR tree-optimization/67328
23767 * fold-const.c (maskable_range_p): New function.
23768 (build_range_check): Generate bittests if possible.
23770 2017-06-13 Martin Liska <mliska@suse.cz>
23772 * gimple-pretty-print.c (dump_probability): Add new argument.
23773 (dump_edge_probability): Dump both probability and count.
23774 (dump_gimple_label): Likewise.
23775 (dump_gimple_bb_header): Likewise.
23777 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
23780 * config/avr/avr-devices.c: Fix indentation.
23781 * config/avr/gen-avr-mmcu-specs.c: Dito.
23783 2017-06-13 Richard Biener <rguenther@suse.de>
23785 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
23786 instead get vector type from stmt_info.
23787 (vectorizable_reduction): Adjust. Remove dead code.
23789 2017-06-13 Richard Biener <rguenther@suse.de>
23791 PR middle-end/81065
23792 * fold-const.c (extract_muldiv_1): Remove bogus distribution
23793 case of C * (x * C2 + C3).
23794 (fold_addr_of_array_ref_difference): Properly fold index difference.
23796 2017-06-12 David S. Miller <davem@davemloft.net>
23799 * config/sparc/sparc.md (return expander): Emit frame blockage if
23800 function uses alloca.
23802 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
23804 * combine.c (make_field_assignment): Check len rather than the mode
23805 precision when calling force_to_mode.
23807 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
23809 Support multilibs and devices that see flash in RAM address range.
23812 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
23813 (avr_mcu_t) <flash_pm_offset>: New field.
23814 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
23815 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
23816 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
23817 (AVR_TINY_PM_OFFSET): Remove macro.
23818 * config/avr/avr.opt (-mshort-calls): New option.
23819 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23820 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
23821 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
23822 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
23823 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
23824 instead of avr_arch->have_jmp_call.
23825 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
23826 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
23827 avr_arch->flash_pm_offset to define.
23828 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
23829 new field flash_pm_offset. Add entry for avrxmega3.
23830 (avr_texinfo): Add entry for avrxmega3.
23831 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
23832 attiny212, attiny214,
23833 attiny412, attiny414, attiny416, attiny417,
23834 attiny814, attiny816, attiny817,
23835 attiny1614, attiny1616, attiny1617,
23836 attiny3214, attiny3216, attiny3217.
23837 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
23838 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
23839 (avr_print_operand_address) [AVR_TINY]: Same.
23840 (avr_asm_init_sections) <readonly_data_section>: Only patch
23841 callback if avr_arch->flash_pm_offset = 0.
23842 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
23843 for rodata if avr_arch->flash_pm_offset != 0.
23844 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
23845 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
23846 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
23847 (m_options): Append opt_rcall.
23848 (m_dirnames): Append dir_rcall.
23849 * config/avr/t-multilib: Regenerate.
23851 * configure.ac [target=avr]: Check whether avrxmega3 default
23852 linker description file works as needed.
23853 * configure: Regenerate.
23854 * doc/avr-mmcu.texi: Regenerate.
23855 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
23856 <__AVR_ARCH__>: Document avrxmega3 and 103.
23857 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
23858 <__AVR_SHORT_CALLS__>: Document it.
23859 <__AVR_PM_BASE_ADDRESS__>: Document it.
23860 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
23861 (AVR Variable Attributes) <progmem>: Document this is
23862 not needed for avrxmega3.
23863 (AVR Named Address Spaces) <__flash>: Dito.
23865 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
23867 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
23869 2017-06-12 Doug Rupp <rupp@adacore.com>
23871 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
23872 Append vxworks-stdint.h to the tm_file list.
23873 * config/vxworks-stdint.h: New file.
23875 2017-06-12 Martin Liska <mliska@suse.cz>
23877 PR tree-optimization/81041
23878 * tree-profile.c (gimple_gen_ic_func_profiler):
23879 Create an extra BB in profile-generate
23880 (gimple_gen_time_profiler): Likewise.
23882 2017-06-12 Jakub Jelinek <jakub@redhat.com>
23884 PR tree-optimization/81003
23885 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
23886 (update_range_test): Use it instead of force_gimple_operand_gsi.
23888 2017-06-12 Richard Biener <rguenther@suse.de>
23890 PR tree-optimization/81053
23891 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
23892 with backedge value not defined in loop. Simplify def stmt
23895 2017-06-11 Tom de Vries <tom@codesourcery.com>
23898 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
23900 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
23901 nvptx_cannot_force_const_mem.
23903 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23905 * opts.c (finish_options): Move test for flag_split_stack after
23906 it has been initialized.
23908 2017-06-11 Jason Merrill <jason@redhat.com>
23910 * tree.h (id_equal): New.
23911 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
23912 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
23913 instead of strcmp of IDENTIFIER_POINTER.
23915 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23917 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
23918 (mark_all_inlined_calls_cdtor): Fix formating.
23919 (inline_transform): Rescale profile before inlining.
23921 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23923 * cgraph.h (cgraph_edge::clone): Update prototype.
23924 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
23925 (cgraph_node::create_clone): Update.
23926 (cgraph_node::create_version_clone): Update.
23927 * tree-inline.c (copy_bb): Update.
23928 (expand_call_inline): Update.
23930 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
23932 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
23933 factored out from ...
23934 (rs6000_emit_prologue): ... here.
23936 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
23938 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
23939 factored out from ...
23940 (rs6000_emit_prologue): ... here.
23942 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23944 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
23946 (handle_missing_profiles): Fix computation of tp_first_run.
23947 (counts_to_freqs): Do not touch freqs when count is 0.
23949 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23951 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
23954 2017-06-10 Tom de Vries <tom@codesourcery.com>
23956 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23957 attributes): Document signal effective target.
23959 2017-06-10 Tom de Vries <tom@codesourcery.com>
23961 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23962 Document effective target stack_size.
23964 2017-06-09 David Malcolm <dmalcolm@redhat.com>
23966 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
23967 to the edit_context if they can be auto-applied.
23969 2017-06-9 Ian Lance Taylor <iant@golang.org>
23971 * opts.c (finish_options): If -fsplit-stack, disable implicit
23972 -forder-blocks-and-partition.
23973 * doc/invoke.texi (Optimize Options): Document that when using
23974 -fsplit-stack -forder-blocks-and-partition is not implicitly
23977 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23979 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
23980 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
23981 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
23982 * builtins.def (abort, trap, unreachable): Declare cold.
23983 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
23984 * tree-core.h (ECF_COLD): New.
23985 * tree.c (set_call_expr_flags): Handle ECF_COLD.
23986 (build_common_builtin_nodes): Mark unreachable and abort as cold.
23988 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23990 * predict.c (unlikely_executed_stmt_p): Cleanup.
23992 2017-06-09 Richard Biener <rguenther@suse.de>
23994 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
23995 model if the ref is always written to.
23997 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23999 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
24001 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24003 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
24006 2017-06-09 Tom de Vries <tom@codesourcery.com>
24009 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
24010 "target cannot support label values" when encountering LABEL_REF.
24012 2017-06-09 Martin Liska <mliska@suse.cz>
24014 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
24015 (gimple_gen_ic_func_profiler): Emit direct comparison
24016 of __gcov_indirect_call_callee with NULL.
24017 (gimple_gen_time_profiler): Change probability from
24018 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
24020 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24022 * profile.c (edge_gcov_counts): Turn to pointer.
24023 (compute_branch_probabilities, compute_branch_probabilities): Update.
24024 (branch_prob): Do not clear edge_gcov_count.
24025 * profile.h (edge_gcov_counts): Turn to pointer.
24026 (edge_gcov_count): Update.
24028 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24030 * gimple.h (gimple_check_failed): Mark cold.
24032 2017-06-09 Richard Biener <rguenther@suse.de>
24034 PR tree-optimization/66623
24035 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
24036 refactor check_reduction into two parts, properly computing
24037 whether we have to check reduction validity for outer loop
24040 2017-06-09 Richard Biener <rguenther@suse.de>
24042 PR tree-optimization/79483
24043 * graphite-scop-detection.c (order): New global.
24044 (get_order): Compute bb to order mapping that satisfies code
24045 generation constraints.
24046 (cmp_pbbs): New helper.
24047 (build_scops): Start domwalk at entry block, sort generated
24050 2017-06-09 Richard Biener <rguenther@suse.de>
24052 PR middle-end/81007
24053 * ipa-polymorphic-call.c
24054 (ipa_polymorphic_call_context::restrict_to_inner_class):
24055 Skip FIELD_DECLs with error_mark_node type.
24056 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
24059 2017-06-09 Martin Liska <mliska@suse.cz>
24061 * predict.c (struct branch_predictor): New struct.
24062 (test_prediction_value_range): New test.
24063 (predict_c_tests): New function.
24064 * selftest-run-tests.c (selftest::run_tests): Run the function.
24065 * selftest.h: Declare new tests.
24067 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
24070 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
24071 gen_add3_insn did not fail.
24072 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
24073 r0, construct that number in a temporary reg and add that reg to r0.
24074 If asked to put the result in r0 as well, fail.
24076 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
24078 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24079 for early expansion of vec_eqv.
24081 2017-06-08 Jakub Jelinek <jakub@redhat.com>
24083 PR middle-end/81005
24084 * ubsan.c (instrument_null): Avoid pointless code temporary.
24085 (pass_ubsan::execute): Instrument aggregate arguments of calls.
24087 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
24091 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
24094 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
24097 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24099 * predict.c (unlikely_executed_edge_p): Move ahead.
24100 (probably_never_executed_edge_p): Use it.
24102 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24104 PR middle-end/79988
24105 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
24106 gimple_call_builtin_p call.
24108 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24110 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
24111 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
24112 rtl_check_failed_type2, rtl_check_failed_code1,
24113 rtl_check_failed_code2, rtl_check_failed_code_mode,
24114 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
24115 rtvec_check_failed_bounds, rtl_check_failed_flag,
24116 _fatal_insn_not_found, _fatal_insn): Likewise.
24117 * tree.h (tree_contains_struct_check_failed,
24118 tree_check_failed, tree_not_check_failed,
24119 tree_class_check_failed, tree_range_check_failed,
24120 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
24121 tree_vec_elt_check_failed, phi_node_elt_check_failed,
24122 tree_operand_check_failed, omp_clause_check_failed,
24123 omp_clause_operand_check_failed, omp_clause_range_check_failed):
24126 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24128 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
24129 flag_branch_probabilities.
24130 * ipa-inline.c (edge_badness): Likewise.
24131 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
24132 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
24133 * predict.c (maybe_hot_frequency_p): Likewise.
24134 (probably_never_executed): Likewise.
24135 * sched-ebb.c (schedule_ebbs): Likewise.
24136 * sched-rgn.c (find_single_block_region): Likewise.
24137 * tracer.c (tail_duplicate): Likewise.
24139 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24141 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
24142 longer requires x_flag_profile_use.
24144 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24146 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
24147 instead of flag_reorder_blocks_and_partition.
24148 * dbxout.c (dbxout_function_end): Likewise.
24149 * dwarf2out.c (gen_subprogram_die): Likewise.
24150 * haifa-sched.c (sched_create_recovery_edges): Likewise.
24151 * hw-doloop.c (reorg_loops): Likewise.
24152 * varasm.c (assemble_start_function,
24153 assemble_end_function): Likewise.
24154 (decide_function_section): Do not check for
24155 flag_reorder_blocks_and_partition.
24157 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24159 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24161 (chkp_get_hard_register_fake_addr_expr): Ditto.
24162 (chkp_build_addr_expr): Add check for hard reg case.
24163 (chkp_parse_array_and_component_ref): Ditto.
24164 (chkp_find_bounds_1): Ditto.
24165 (chkp_process_stmt): Don't generate bounds store for
24168 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24170 * predict.c (maybe_hot_bb_p): Do not check profile status.
24171 (maybe_hot_edge_p): Likewise.
24172 (probably_never_executed): Check for zero counts even if profile
24174 (unlikely_executed_edge_p): New function.
24175 (unlikely_executed_stmt_p): New function.
24176 (unlikely_executed_bb_p): New function.
24177 (set_even_probabilities): Use unlikely predicates.
24178 (combine_predictions_for_bb): Likewise.
24179 (predict_paths_for_bb): Likewise.
24180 (predict_paths_leading_to_edge): Likewise.
24181 (determine_unlikely_bbs): New function.
24182 (estimate_bb_frequencies): Use it.
24183 (compute_function_frequency): Use zero counts even if profile is
24185 * profile-count.h: Fix typo.
24187 2017-08-08 Julia Koval <julia.koval@intel.com>
24189 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
24190 _mm512_mask_cvtsepi16_storeu_epi8,
24191 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
24192 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
24193 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
24194 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
24195 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
24196 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
24197 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
24198 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
24199 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
24200 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
24201 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
24202 __builtin_ia32_pmovuswb256mem_mask,
24203 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
24204 __builtin_ia32_pmovwb512mem_mask): New builtins.
24206 2017-08-08 Julia Koval <julia.koval@intel.com>
24208 PR target/73350,80862
24209 * config/i386/subst.md (round): Fix round pattern.
24210 * config/i386/i386.c (ix86_erase_embedded_rounding):
24211 Fix erasing rounding for the fixed pattern.
24213 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24215 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
24217 2017-06-08 Martin Liska <mliska@suse.cz>
24219 PR gcov-profile/80911
24220 * gcov.c (block_info::block_info): New constructor.
24222 2017-06-07 Carl Love <cel@us.ibm.com>
24224 * config/rs6000/rs6000-c: The return type of the following
24225 built-in functions was implemented as int not long long. Fix sign
24226 of return value for the unsigned version of vec_mulo and vec_mule.
24227 vector unsigned long long vec_bperm (vector unsigned long long,
24228 vector unsigned char)
24229 vector signed long long vec_mule (vector signed int,
24231 vector unsigned long long vec_mule (vector unsigned int,
24232 vector unsigned int)
24233 vector signed long long vec_mulo (vector signed int,
24235 vector unsigned long long vec_mulo (vector unsigned int,
24236 vector unsigned int)
24237 * doc/extend.texi: Fix the documentation for the built-in
24240 2017-06-07 Carl Love <cel@us.ibm.com>
24243 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
24246 2017-06-07 Carl Love <cel@us.ibm.com>
24248 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
24249 support, Generate doublehv for signed int/float for BE case only.
24251 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
24253 * doc/invoke.texi (mcx16): Rewrite.
24255 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24257 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
24258 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
24259 *mov<mode>_softfloat, and an anonymous splitter): Use
24260 nonimmediate_operand instead of rs6000_nonimmediate_operand.
24262 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24264 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
24266 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
24267 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
24268 (rs6000_debug_reg_global): Adjust.
24269 (rs6000_init_hard_regno_mode_ok): Adjust.
24270 (rs6000_dbx_register_number): Adjust.
24271 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
24272 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
24273 Remove SPE_ACC and SPEFSCR.
24274 (REG_ALLOC_ORDER): Ditto.
24275 (FRAME_POINTER_REGNUM): Change to 111.
24276 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
24277 (REG_CLASS_NAMES): Ditto.
24278 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
24279 (REGISTER_NAMES): Ditto.
24280 (ADDITIONAL_REG_NAMES): Ditto.
24281 (rs6000_reg_names): Ditto.
24282 * config/rs6000/rs6000.md: Renumber some register number
24285 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24287 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
24289 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
24290 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
24292 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
24293 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
24294 Delete the SPE high registers.
24295 (REG_ALLOC_ORDER): Ditto.
24296 (enum reg_class): Remove SPE_HIGH_REGS.
24297 (REG_CLASS_NAMES): Ditto.
24298 (REG_CLASS_CONTENTS): Delete the SPE high registers.
24299 (REGISTER_NAMES): Ditto.
24300 (rs6000_reg_names): Ditto.
24301 * doc/tm.texi.in: Remove SPE as example.
24302 * doc/tm.texi: Regenerate.
24304 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24306 * config/rs6000/8540.md (ppc8540_brinc): Delete.
24307 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
24308 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
24309 * config/rs6000/rs6000.md (type): Remove "brinc".
24311 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24313 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
24314 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
24315 * config/rs6000/linuxspe.h: Delete file.
24316 * config/rs6000/rs6000.md: Don't include spe.md.
24317 * config/rs6000/spe.h: Delete file.
24318 * config/rs6000/spe.md: Delete file.
24319 * config/rs6000/t-rs6000: Remove spe.md.
24321 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24323 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
24324 (reg_or_none500mem_operand): Delete.
24325 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
24326 instead of reg_or_none500mem_operand.
24328 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24330 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
24331 handling of SPE flags.
24332 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
24334 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24336 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
24338 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
24339 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
24340 paired_divv2sf3): Similar.
24341 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
24342 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
24343 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
24345 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
24346 Rename the paired_* instruction patterns.
24347 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
24349 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
24350 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
24351 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
24352 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
24353 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
24354 PAIRED_VECTOR_MODE.
24355 (struct machine_function): Delete field spe_insn_chain_scanned_p.
24356 (spe_func_has_64bit_regs_p): Delete.
24357 (spe_expand_predicate_builtin): Delete.
24358 (spe_expand_evsel_builtin): Delete.
24359 (TARGET_DWARF_REGISTER_SPAN): Do not define.
24360 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
24361 (invalid_e500_subreg): Delete.
24362 (rs6000_legitimize_address): Always force_reg op2 as well, for
24363 paired single memory accesses.
24364 (rs6000_member_type_forces_blk): Delete.
24365 (rs6000_spe_function_arg): Delete.
24366 (rs6000_expand_unop_builtin): Delete SPE handling.
24367 (rs6000_expand_binop_builtin): Ditto.
24368 (spe_expand_stv_builtin): Delete.
24369 (bdesc_2arg_spe): Delete.
24370 (spe_expand_builtin): Delete.
24371 (spe_expand_predicate_builtin): Delete.
24372 (spe_expand_evsel_builtin): Delete.
24373 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
24374 (spe_init_builtins): Delete.
24375 (spe_func_has_64bit_regs_p): Delete.
24376 (savres_routine_name): Delete "info" parameter. Adjust callers.
24377 (rs6000_emit_stack_reset): Ditto.
24378 (rs6000_dwarf_register_span): Delete.
24379 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
24380 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
24381 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
24383 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
24385 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
24386 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
24387 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
24388 mulv2sf3, divv2sf3): Delete expanders.
24390 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24392 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
24394 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24396 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
24397 * config/rs6000/rs6000.c: Ditto.
24399 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24401 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
24402 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
24403 comparison_operator.
24405 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24407 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
24408 * config/rs6000/rs6000.opt: Ditto.
24409 * config/rs6000/t-rtems: Ditto.
24411 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24413 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
24414 TARGET_E500_SINGLE by 0, simplify.
24415 * config/rs6000/rs6000.c: Ditto.
24416 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
24417 (spe_build_register_parallel): Delete.
24418 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
24419 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
24420 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
24421 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
24422 (E500_CONVERT): Delete.
24423 * config/rs6000/spe.md: Remove many patterns and all define_constants.
24425 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24427 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
24428 * config/rs6000/dfp.md: Ditto.
24429 (negdd2, *negdd2_fpr): Merge.
24430 (absdd2, *absdd2_fpr): Merge.
24431 (negtd2, *negtd2_fpr): Merge.
24432 (abstd2, *abstd2_fpr): Merge.
24433 * config/rs6000/e500.h: Delete file.
24434 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
24435 TARGET_FPRS by 1 and simplify.
24436 * config/rs6000/rs6000-c.c: Ditto.
24437 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
24438 TARGET_DF_SPE by 0.
24439 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
24441 * config/rs6000/rs6000.md: Ditto.
24442 (floatdidf2, *floatdidf2_fpr): Merge.
24443 (move_from_CR_gt_bit): Delete.
24444 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
24445 (E500_CR_IOR_COMPARE): Delete.
24446 (All patterns that require !TARGET_FPRS): Delete.
24447 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
24449 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24451 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
24453 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24455 * graphds.c (add_edge): Intitialize edge's attached data.
24456 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
24457 pointer parameter. Call pointed function on each edge during
24458 graph traversing. Skip traversing the edge when the function
24460 (graphds_dfs, graphds_scc): Ditto.
24461 (for_each_edge): New parameter. Pass the new parameter to callback
24463 * graphds.h (skip_edge_callback): New function pointer type.
24464 (graphds_dfs, graphds_scc): New function pointer parameter.
24465 (graphds_edge_callback, for_each_edge): New parameter.
24467 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24469 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
24470 out code checking if runtime alias check is possible to below ...
24471 Call the new function.
24472 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
24473 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
24475 2017-06-07 Marek Polacek <polacek@redhat.com>
24478 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
24479 TYPE_OVERFLOW_WRAPS check.
24481 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24483 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
24484 if versioning is required.
24485 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
24486 peeling with the check for versioning.
24488 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24490 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
24491 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
24492 Set true to new parameter if new ssa variable is defined.
24493 (vect_gen_vector_loop_niters): Refactor. Set range information
24494 for the new vector loop bound variable.
24495 (vect_do_peeling): Ditto.
24497 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24499 * tree-affine.c (ssa.h): Include header file.
24500 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
24501 has wrapping overflow behavior.
24503 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24505 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
24507 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24509 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
24510 (tree_to_aff_combination): ... here.
24512 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24514 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
24515 reg_pressure model function.
24516 (ivopts_global_cost_for_size): Delete.
24517 (determine_set_costs, iv_ca_recount_cost): Call new model function
24518 ivopts_estimate_reg_pressure.
24520 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24522 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
24523 expensive than udiv. Remove floating point cases from mod.
24525 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24527 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
24528 Increase idiv cost.
24530 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24532 * config/aarch64/aarch64.md
24533 (copysignsf3): Fix mask generation.
24535 2017-06-07 Jakub Jelinek <jakub@redhat.com>
24537 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
24539 (class dump_manager): Add register_dumps method.
24540 * dumpfile.c: Include langhooks.h.
24541 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
24542 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
24543 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
24544 (dump_manager::dump_register): Start with 512 entries instead of 32.
24545 (dump_manager::register_dumps): New method.
24546 * toplev.c (general_init): Instead of invoking register_dumps
24547 langhook, invoke register_dumps method on the dump manager.
24548 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
24551 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
24553 * doc/md.texi: Clarify the restrictions on a define_insn condition.
24554 Say that # requires an associated define_split to exist, and that
24555 the define_split must be suitable for use after register allocation.
24557 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24559 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
24560 (compute_outgoing_frequencies): Also initialize zero counts.
24561 (find_many_sub_basic_blocks): Do not produce uninitialized profile
24562 around loops; preserve more of profile when nothing changes.
24564 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
24566 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
24568 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
24569 * config/arm/arm-cpu-cdata.h: Regenerate.
24570 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
24571 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
24572 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
24573 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
24574 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
24576 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
24578 * config/arm/t-rmprofile: Likewise.
24579 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
24581 2017-06-06 David S. Miller <davem@davemloft.net>
24584 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
24585 blockage if function uses alloca.
24587 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24589 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
24590 New "uid" fields to hold pretty-print IDs of group and ref.
24591 Memory references are now identified as <group_id>:<ref_id>
24592 instead of using [random] addresses.
24593 (dump_mem_details): Simplify, no functional change.
24594 (dump_mem_ref): Simplify and make output more concise.
24595 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
24596 (find_or_create_group): Initialize group uid.
24597 (record_ref): Initialize ref uid. Improve debug output.
24598 (prune_group_by_reuse, should_issue_prefetch_p,)
24599 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
24600 (mark_nontemporal_store, determine_loop_nest_reuse):
24601 Improve debug output.
24603 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24605 * dbgcnt.def (prefetch): New debug counter.
24606 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
24607 (schedule_prefetches): Stop issueing prefetches if debug counter
24610 2017-06-06 Tom de Vries <tom@codesourcery.com>
24612 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
24613 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
24615 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24617 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
24618 Use aarch64_reg_or_zero predicate for operand 4.
24619 (aarch64_compare_and_swap<mode> define_insn_and_split):
24620 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
24621 (aarch64_store_exclusive<mode>): Likewise for operand 2.
24623 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24625 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
24626 (arm_compute_save_core_reg_mask): This.
24627 (thumb1_compute_save_reg_mask): Rename into ...
24628 (thumb1_compute_save_core_reg_mask): This.
24629 (arm_compute_save_reg0_reg12_mask): Adapt comment.
24630 (arm_compute_frame_layout): Likewise.
24632 2017-06-06 Richard Biener <rguenther@suse.de>
24634 PR tree-optimization/80974
24635 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
24636 keep or clear leaders SSA info.
24638 2017-06-06 Tom de Vries <tom@codesourcery.com>
24640 * config/nvptx/nvptx.c (split_mode_p): New function.
24641 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
24643 2017-06-06 Tom de Vries <tom@codesourcery.com>
24645 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
24647 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24650 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
24653 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24655 * shrink-wrap.c (handle_simple_exit): Update profile.
24656 (try_shrink_wrapping): Upate profile.
24658 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24660 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
24661 (tree_guess_outgoing_edge_probabilities): New.
24662 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
24663 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
24665 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24667 * ipa-split.c (split_function): Initialize return bb profile.
24669 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24671 * profile.c (compute_branch_probabilities): Also initialize
24672 EXIT_BLOCK profile.
24674 2017-06-06 Richard Biener <rguenther@suse.de>
24676 PR tree-optimization/80928
24677 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
24678 (vect_analyze_loop_operations): Properly guard analysis for
24680 (vect_transform_loop): Likewise.
24681 (vect_analyze_loop_2): Also reset SLP type on PHIs.
24682 (vect_model_induction_cost): Do not cost for pure SLP.
24683 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
24684 of induction in inner loop vectorization.
24685 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
24686 (vect_get_and_check_slp_defs): Handle vect_induction_def.
24687 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
24689 (vect_analyze_slp_cost_1): Cost induction.
24690 (vect_detect_hybrid_slp_stmts): Handle PHIs.
24691 (vect_get_slp_vect_defs): Likewise.
24692 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
24693 (vect_transform_stmt): Handle SLP reductions.
24694 * tree-vectorizer.h (vectorizable_induction): Adjust.
24696 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24698 * config/rs6000/rs6000.c (make_resolver_func): Update
24699 init_lowered_empty_function call.
24701 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
24703 * doc/invoke.texi: Document the -fprofile-abs-path option.
24704 * common.opt (fprofile-abs-path): New option.
24705 * gcov-io.h (gcov_write_filename): Declare.
24706 * gcov-io.c (gcov_write_filename): New function.
24707 * coverage.c (coverage_begin_function): Use gcov_write_filename.
24708 * profile.c (output_location): Likewise.
24710 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24712 * shring-wrap.c: Revert accidental commit.
24714 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
24716 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
24718 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24720 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
24722 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
24723 profile in callgraph edge.
24724 * profile-count.h (apply_probability): If THIS is 0, then result is 0
24725 (apply_scale): Likewise.
24726 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
24727 Also scale profile when inlining function with zero profile.
24728 (initialize_cfun): Update exit block profile even when it is zero.
24729 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
24730 when profile is read.
24732 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24734 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
24735 (CLONE_*): New constants to define the processors we can generate
24736 code for with the target_clone attribute.
24737 (rs6000_clone_map): New array to identify which clone processors
24738 the current program is running on.
24739 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
24740 target_clone attribute.
24741 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
24742 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
24743 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
24744 (cpu_expand_builtin): Add support for target_clone attribute.
24745 (rs6000_valid_attribute_p): Allow "default" attribute.
24746 (get_decl_name): New debug function to simplify printing the
24747 current function name in debugging statements.
24748 (rs6000_clone_priority): New functions to support the target_clone
24749 attribute, and be able to generate code to switch between ISA 2.05
24750 through ISA 3.0 (power6 through power9).
24751 (rs6000_compare_version_priority): Likewise.
24752 (rs6000_get_function_versions_dispatcher): Likewise.
24753 (make_resolver_func): Likewise.
24754 (add_condition_to_bb): Likewise.
24755 (dispatch_function_versions): Likewise.
24756 (rs6000_generate_version_dispatcher_body): Likewise.
24757 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
24758 (fusion_gpr_load_p): Fix a spacing issue.
24759 * doc/extend.texi (Common Function Attributes): Document that the
24760 PowerPC supports the target_clone attribute.
24762 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24764 * config/arm/arm.h: explain F symbol found in description of ARM
24765 register allocation in its legend.
24767 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24769 * config/mips/frame-header-opt.c: Include profile-count.h.
24770 * config/riscv/riscv.c: Include profile-count.h
24772 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24774 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
24776 (sm_set_flag_if_changed): Add bbs field.
24777 (execute_sm_if_changed_flag_set): Pass BBS.
24778 (execute_sm): Update.
24780 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24782 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
24785 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24787 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
24788 (peephole2): New peephole2 to emit the above.
24789 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
24791 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24793 * config/aarch64/aarch64.c (define_peephole2 above
24794 *sub_<shift>_<mode>): New peephole.
24796 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24798 * config/i386/i386.c (make_resolver_func): Update.
24799 * Makefile.in: Add profile-count.h and profile-count.o
24800 * auto-profile.c (afdo_indirect_call): Update to new API.
24801 (afdo_set_bb_count): Update.
24802 (afdo_propagate_edge): Update.
24803 (afdo_propagate_circuit): Update.
24804 (afdo_calculate_branch_prob): Update.
24805 (afdo_annotate_cfg): Update.
24806 * basic-block.h: Include profile-count.h
24807 (struct edge_def): Turn count to profile_count.
24808 (struct basic_block_def): Likewie.
24809 (REG_BR_PROB_BASE): Move to profile-count.h
24810 (RDIV): Move to profile-count.h
24811 * bb-reorder.c (max_entry_count): Turn to profile_count.
24812 (find_traces): Update.
24813 (rotate_loop):Update.
24814 (connect_traces):Update.
24815 (sanitize_hot_paths):Update.
24816 * bt-load.c (migrate_btr_defs): Update.
24817 * cfg.c (RDIV): Remove.
24818 (init_flow): Use alloc_block.
24819 (alloc_block): Uninitialize count.
24820 (unchecked_make_edge): Uninitialize count.
24821 (check_bb_profile): Update.
24822 (dump_edge_info): Update.
24823 (dump_bb_info): Update.
24824 (update_bb_profile_for_threading): Update.
24825 (scale_bbs_frequencies_int): Update.
24826 (scale_bbs_frequencies_gcov_type): Update.
24827 (scale_bbs_frequencies_profile_count): New.
24828 * cfg.h (update_bb_profile_for_threading): Update.
24829 (scale_bbs_frequencies_profile_count): Declare.
24830 * cfgbuild.c (compute_outgoing_frequencies): Update.
24831 (find_many_sub_basic_blocks): Update.
24832 * cfgcleanup.c (try_forward_edges): Update.
24833 (try_crossjump_to_edge): Update.
24834 * cfgexpand.c (expand_gimple_tailcall): Update.
24835 (construct_exit_block): Update.
24836 * cfghooks.c (verify_flow_info): Update.
24837 (dump_bb_for_graph): Update.
24838 (split_edge): Update.
24839 (make_forwarder_block): Update.
24840 (duplicate_block): Update.
24841 (account_profile_record): Update.
24842 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
24843 (get_estimated_loop_iterations): Update.
24844 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
24845 (single_likely_exit): Update.
24846 * cfgloopmanip.c (scale_loop_profile): Update.
24848 (set_zero_probability): Update.
24849 (lv_adjust_loop_entry_edge): Update.
24850 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24851 (purge_dead_edges): Update.
24852 (rtl_account_profile_record): Update.
24853 * cgraph.c (cgraph_node::create): Uninitialize count.
24854 (symbol_table::create_edge): Uninitialize count.
24855 (cgraph_update_edges_for_call_stmt_node): Update.
24856 (cgraph_edge::dump_edge_flags): Update.
24857 (cgraph_node::dump): Update.
24858 (cgraph_edge::maybe_hot_p): Update.
24859 * cgraph.h: Include profile-count.h
24860 (create_clone), create_edge, create_indirect_edge): Update.
24861 (cgraph_node): Turn count to profile_count.
24862 (cgraph_edge0: Likewise.
24863 (make_speculative, clone): Update.
24864 (create_edge): Update.
24865 (init_lowered_empty_function): Update.
24866 * cgraphclones.c (cgraph_edge::clone): Update.
24867 (duplicate_thunk_for_node): Update.
24868 (cgraph_node::create_clone): Update.
24869 * cgraphunit.c (cgraph_node::analyze): Update.
24870 (cgraph_node::expand_thunk): Update.
24871 * final.c (dump_basic_block_info): Update.
24872 * gimple-streamer-in.c (input_bb): Update.
24873 * gimple-streamer-out.c (output_bb): Update.
24874 * graphite.c (print_global_statistics): Update.
24875 (print_graphite_scop_statistics): Update.
24876 * hsa-brig.c: Include basic-block.h.
24877 * hsa-dump.c: Include basic-block.h.
24878 * hsa-gen.c (T sum_slice): Update.
24879 (convert_switch_statements):Update.
24880 * hsa-regalloc.c: Include basic-block.h.
24881 * ipa-chkp.c (chkp_produce_thunks): Update.
24882 * ipa-cp.c (struct caller_statistics): Update.
24883 (init_caller_stats): Update.
24884 (gather_caller_stats): Update.
24885 (ipcp_cloning_candidate_p): Update.
24886 (good_cloning_opportunity_p): Update.
24887 (get_info_about_necessary_edges): Update.
24888 (dump_profile_updates): Update.
24889 (update_profiling_info): Update.
24890 (update_specialized_profile): Update.
24891 (perhaps_add_new_callers): Update.
24892 (decide_about_value): Update.
24893 (ipa_cp_c_finalize): Update.
24894 * ipa-devirt.c (struct odr_type_warn_count): Update.
24895 (struct decl_warn_count): Update.
24896 (struct final_warning_record): Update.
24897 (possible_polymorphic_call_targets): Update.
24898 (ipa_devirt): Update.
24899 * ipa-fnsummary.c (redirect_to_unreachable): Update.
24900 * ipa-icf.c (sem_function::merge): Update.
24901 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
24902 * ipa-inline.c (compute_uninlined_call_time): Update.
24903 (compute_inlined_call_time): Update.
24904 (want_inline_small_function_p): Update.
24905 (want_inline_self_recursive_call_p): Update.
24906 (edge_badness): Update.
24907 (lookup_recursive_calls): Update.
24908 (recursive_inlining): Update.
24909 (inline_small_functions): Update.
24910 (dump_overall_stats): Update.
24911 (dump_inline_stats): Update.
24912 * ipa-profile.c (ipa_profile_generate_summary): Update.
24913 (ipa_propagate_frequency): Update.
24914 (ipa_profile): Update.
24915 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24916 * ipa-utils.c (ipa_merge_profiles): Update.
24917 * loop-doloop.c (doloop_modify): Update.
24918 * loop-unroll.c (report_unroll): Update.
24919 (unroll_loop_runtime_iterations): Update.
24920 * lto-cgraph.c (lto_output_edge): Update.
24921 (lto_output_node): Update.
24922 (input_node): Update.
24923 (input_edge): Update.
24924 (merge_profile_summaries): Update.
24925 * lto-streamer-in.c (input_cfg): Update.
24926 * lto-streamer-out.c (output_cfg): Update.
24927 * mcf.c (create_fixup_graph): Update.
24928 (adjust_cfg_counts): Update.
24929 (sum_edge_counts): Update.
24930 * modulo-sched.c (sms_schedule): Update.
24931 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
24932 * predict.c (maybe_hot_count_p): Update.
24933 (probably_never_executed): Update.
24934 (dump_prediction): Update.
24935 (combine_predictions_for_bb): Update.
24936 (propagate_freq): Update.
24937 (handle_missing_profiles): Update.
24938 (counts_to_freqs): Update.
24939 (rebuild_frequencies): Update.
24940 (force_edge_cold): Update.
24941 * predict.h: Include profile-count.h
24942 (maybe_hot_count_p, counts_to_freqs): UPdate.
24943 * print-rtl-function.c: Do not include cfg.h
24944 * print-rtl.c: Include basic-block.h
24945 * profile-count.c: New file.
24946 * profile-count.h: New file.
24947 * profile.c (is_edge_inconsistent): Update.
24948 (correct_negative_edge_counts): Update.
24949 (is_inconsistent): Update.
24950 (set_bb_counts): Update.
24951 (read_profile_edge_counts): Update.
24952 (compute_frequency_overlap): Update.
24953 (compute_branch_probabilities): Update; Initialize and deinitialize
24955 (branch_prob): Update.
24956 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
24957 (edge_gcov_count): New.
24958 (bb_gcov_count): New.
24959 * shrink-wrap.c (try_shrink_wrapping): Update.
24960 * tracer.c (better_p): Update.
24961 * trans-mem.c (expand_transaction): Update.
24962 (ipa_tm_insert_irr_call): Update.
24963 (ipa_tm_insert_gettmclone_call): Update.
24964 * tree-call-cdce.c: Update.
24965 * tree-cfg.c (gimple_duplicate_sese_region): Update.
24966 (gimple_duplicate_sese_tail): Update.
24967 (gimple_account_profile_record): Update.
24968 (execute_fixup_cfg): Update.
24969 * tree-inline.c (copy_bb): Update.
24970 (copy_edges_for_bb): Update.
24971 (initialize_cfun): Update.
24972 (freqs_to_counts): Update.
24973 (copy_cfg_body): Update.
24974 (expand_call_inline): Update.
24975 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24976 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
24977 (try_unroll_loop_completely): Update.
24978 (try_peel_loop): Update.
24979 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24980 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
24981 * tree-ssa-loop-split.c (connect_loops): Update.
24982 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
24983 * tree-ssa-reassoc.c (branch_fixup): Update.
24984 * tree-ssa-tail-merge.c (replace_block_by): Update.
24985 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
24986 (compute_path_counts): Update.
24987 (update_profile): Update.
24988 (recompute_probabilities): Update.
24989 (update_joiner_offpath_counts): Update.
24990 (estimated_freqs_path): Update.
24991 (freqs_to_counts_path): Update.
24992 (clear_counts_path): Update.
24993 (ssa_fix_duplicate_block_edges): Update.
24994 (duplicate_thread_path): Update.
24995 * tree-switch-conversion.c (case_bit_test_cmp): Update.
24996 (struct switch_conv_info): Update.
24997 * tree-tailcall.c (decrease_profile): Update.
24998 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24999 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25000 * value-prof.c (check_counter): Update.
25001 (gimple_divmod_fixed_value): Update.
25002 (gimple_mod_pow2): Update.
25003 (gimple_mod_subtract): Update.
25004 (gimple_ic_transform): Update.
25005 (gimple_stringop_fixed_value): Update.
25006 * value-prof.h (gimple_ic): Update.
25008 2017-06-02 Carl Love <cel@us.ibm.com>
25010 * config/rs6000/rs6000-c: Add support for built-in functions
25011 vector double vec_doublee (vector signed int);
25012 vector double vec_doublee (vector unsigned int);
25013 vector double vec_doublee (vector float);
25014 vector double vec_doubleh (vector signed int);
25015 vector double vec_doubleh (vector unsigned int);
25016 vector double vec_doubleh (vector float);
25017 vector double vec_doublel (vector signed int);
25018 vector double vec_doublel (vector unsigned int);
25019 vector double vec_doublel (vector float);
25020 vector double vec_doubleo (vector signed int);
25021 vector double vec_doubleo (vector unsigned int);
25022 vector double vec_doubleo (vector float);.
25023 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
25024 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
25026 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
25027 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
25028 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
25030 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
25031 vec_doublel, vec_doubleh.
25032 * doc/extend.texi: Update the built-in documentation file for the
25033 new built-in functions.
25035 2017-06-02 David Malcolm <dmalcolm@redhat.com>
25038 * ipa-inline-analysis.c (free_growth_caches): Set
25039 edge_removal_hook_holder to NULL after removing it.
25041 2017-06-02 Sudakshina Das <sudi.das@arm.com>
25043 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
25044 comparision with zero.
25046 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
25047 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25048 for early expansion of vec_min and vec_max builtins.
25049 (builtin_function_type): Add min/max unsigned variants to those
25050 identified as having unsigned arguments.
25052 2017-06-02 Olivier Hainque <hainque@adacore.com>
25054 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
25056 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25058 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
25059 Use VALL_F16 iterator rather than VALL.
25061 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25063 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
25064 Emit CBNZ inside loop when doing a strong exchange and comparing
25065 against zero. Generate the CC flags after the loop.
25067 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
25069 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
25070 (dl_section_ref): New.
25071 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
25072 On AIX, append an expression to subtract the size of the
25073 section length to dl_section_ref.
25075 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
25077 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25078 for early expansion of vector absolute builtins.
25080 2017-06-02 Richard Biener <rguenther@suse.de>
25082 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
25083 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
25085 2017-06-02 Richard Biener <rguenther@suse.de>
25087 PR tree-optimization/80948
25088 * tree-tailcall.c (find_tail_calls): Track stmts to move in
25089 stmt order as well.
25091 2017-06-02 Richard Biener <rguenther@suse.de>
25093 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
25095 * tree-vect-stmts.c (process_use): Do not mark backedge defs
25096 for inductions as relevant.
25098 2017-06-02 Richard Biener <rguenther@suse.de>
25100 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
25101 (vectorizable_induction): ... this. Remove dead code.
25103 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
25105 * builtins. (expand_builtin_alloca): Remove second parameter and
25106 infer its value from the first parameter instead.
25107 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
25109 2017-06-02 Jakub Jelinek <jakub@redhat.com>
25111 PR rtl-optimization/80903
25112 * loop-doloop.c (add_test): Unshare sequence.
25114 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
25116 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
25118 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
25120 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
25122 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
25123 xlogue_layout::get_instance, logue_layout::xlogue_layout,
25124 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
25125 (xlogue_layout::get_stub_rtx): Make static.
25126 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
25127 (xlogue_layout::compute_stub_managed_regs): Rename to...
25128 (xlogue_layout::count_stub_managed_regs): ...this.
25129 (xlogue_layout::is_stub_managed_reg): New function.
25130 (xlogue_layout::m_stub_names): Rename to...
25131 (xlogue_layout::s_stub_names): ...this, make static.
25132 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
25133 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
25134 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
25135 xlogue_layout::s_stub_names): Instantiate statics.
25136 (stub_managed_regs): Remove.
25137 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
25138 (disable_call_ms2sysv_xlogues): Rename to...
25139 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
25140 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
25142 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
25143 change after reload_completed.
25144 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
25146 (ix86_expand_prologue): Likewise.
25147 (ix86_expand_epilogue): Likewise.
25148 (ix86_expand_split_stack_prologue): Likewise.
25149 (ix86_compute_frame_layout): Remove frame parameter ...
25150 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
25151 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
25153 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
25154 (ix86_frame): Move from here ...
25155 * config/i386/i386.h (ix86_frame): ... to here.
25156 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
25157 complete ix86_frame data structure instead. Remove some_ld_name.
25159 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
25161 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
25162 symbols that hold a DECL_VALUE_EXPR.
25164 2017-06-01 Martin Jambor <mjambor@suse.cz>
25166 PR tree-optimization/80898
25167 * tree-sra.c (process_subtree_disqualification): Removed.
25168 (disqualify_candidate): Do not acll
25169 process_subtree_disqualification.
25170 (subtree_mark_written_and_enqueue): New function.
25171 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
25172 RHS has been disqualified and re-queue LHS if necessary. Apart
25173 from that, ignore disqualified RHS.
25175 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25177 * config/s390/s390.c (s390_emit_epilogue): Disable early return
25178 address fetch for z10 or later.
25180 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25182 * config/arc/arc.md (tst_movb): Add guard when splitting.
25184 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25186 * config/arc/arc.c (arc_can_eliminate): Test against
25187 arc_frame_pointer_needed.
25189 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25191 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
25192 to prevent store reordering.
25193 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
25194 (type): Add block type.
25195 (stack_tie): Define special instruction to be used in
25198 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25200 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
25201 constraint. It is not valid for the pattern.
25202 (noncommutative_binary_comparison): Likewise.
25204 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25206 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
25209 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25211 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
25212 be used by the reg-alloc.
25214 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25216 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
25217 reg-alloc when having mul64 or mul32x16 instructions.
25218 (mulsidi3): Likewise.
25219 (umulsidi3): Likewise.
25220 (mulsi32x16): New pattern.
25221 (mulsi64): Likewise.
25222 (mulsidi64): Likewise.
25223 (umulsidi64): Likewise.
25224 (MUL32x16_REG): Define.
25225 (mul64_600): Use MUL32x16_REG.
25226 (mac64_600): Likewise.
25227 (umul64_600): Likewise.
25228 (umac64_600): Likewise.
25230 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25232 * config/arc/arc.md (mulsi3_700): Make it commutative.
25234 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
25236 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
25238 (*sign_extendsidi2_insn): Likewise for movstosw.
25240 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
25242 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
25243 the type of the input discriminant value. Convert the
25244 discriminant value of signedness vary.
25246 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
25248 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
25249 Add to -Wall section.
25251 2017-06-01 Richard Biener <rguenther@suse.de>
25253 PR middle-end/66313
25254 * fold-const.c (fold_plusminus_mult_expr): If the factored
25255 factor may be zero use a wrapping type for the inner operation.
25256 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
25257 and handle moved defs.
25258 (process_assignment): Properly guard the unary op case. Return a
25259 tri-state indicating that moving the stmt before the call may allow
25260 to continue. Pass through to_move.
25261 (find_tail_calls): Handle moving unrelated defs before
25264 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
25267 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
25268 splitter result in the canonical way.
25270 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
25272 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
25273 also for 32bit target. Update insn attributes.
25274 (zero-extendsidi2 splitter): Allow all registers for operand 1.
25276 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
25278 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
25279 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
25280 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
25281 (_mm_maskz_min_ss): New intrinsics.
25283 2017-05-31 Martin Liska <mliska@suse.cz>
25285 * tree-vect-loop.c (vect_create_epilog_for_reduction):
25286 Change comment style to one we normally use.
25287 (vectorizable_reduction): Likewise.
25288 (vectorizable_induction): Likewise.
25289 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
25290 (vectorizable_call): Likewise.
25291 (vectorizable_simd_clone_call): Likewise.
25292 (vectorizable_conversion): Likewise.
25293 (vectorizable_assignment): Likewise.
25294 (vectorizable_shift): Likewise.
25295 (vectorizable_operation): Likewise.
25296 (vectorizable_store): Likewise.
25297 (vectorizable_load): Likewise.
25298 * tree-vectorizer.h: Likewise.
25300 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
25302 * passes.c (emergency_dump_function): New.
25303 * tree-pass.h (emergency_dump_function): Declare.
25304 * plugin.c (plugins_internal_error_function): Remove.
25305 * plugin.h (plugins_internal_error_function): Remove declaration.
25306 * toplev.c (internal_error_function): New static function. Use it...
25307 (general_init): ...here.
25309 2017-05-31 Graham Markall <graham.markall@embecosm.com>
25311 * config/arc/arc.c (arc_print_operand): Handle constant operands.
25312 (arc_rtx_costs): Add costs for new patterns.
25313 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
25314 * config/arc/predicates.md: Add _1_2_3_operand predicate.
25316 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
25318 * tree-ssa-strlen.c (get_next_strinfo): New function.
25319 (get_stridx_plus_constant): Use it.
25320 (zero_length_string): Likewise.
25321 (adjust_related_strinfos): Likewise.
25322 (adjust_last_stmt): Likewise.
25324 2017-05-31 Richard Biener <rguenther@suse.de>
25327 * config/i386/i386.c (ix86_expand_builtin): Remove assert
25328 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
25330 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
25332 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
25333 loop_vinfo argument and use of dependence distance vectors.
25334 Check instead whether the two references differ only in their
25335 initial value and assume that they have the same alignment if the
25336 difference is a multiple of the vector alignment.
25337 (vect_analyze_data_refs_alignment): Update call accordingly.
25339 2017-05-31 Martin Liska <mliska@suse.cz>
25342 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
25344 2017-05-31 Bin Cheng <bin.cheng@arm.com>
25346 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
25347 (create_intersect_range_checks): Move from ...
25348 * tree-data-ref.c (create_intersect_range_checks_index)
25349 (create_intersect_range_checks): ... to here.
25350 (create_runtime_alias_checks): New function factored from ...
25351 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
25352 here. Call above function.
25353 * tree-data-ref.h (create_runtime_alias_checks): New function.
25355 2017-05-31 Bin Cheng <bin.cheng@arm.com>
25357 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
25358 segment length for dr_b and compute it in wide_int.
25360 2017-05-31 Richard Biener <rguenther@suse.de>
25362 PR tree-optimization/80906
25363 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
25364 and pass through iv_map.
25365 (copy_bb_and_scalar_dependences): Adjust.
25366 (translate_pending_phi_nodes): Likewise.
25367 (copy_loop_close_phi_args): Handle code-generating IVs instead
25370 2017-05-30 David Malcolm <dmalcolm@redhat.com>
25372 * diagnostic-color.c (color_dict): Add "type-diff".
25373 (parse_gcc_colors): Update comment.
25374 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
25375 -fdiagnostics-show-template-tree and -fno-elide-type.
25376 (GCC_COLORS): Add type-diff to example.
25378 (-fdiagnostics-show-template-tree): New.
25379 (-fno-elide-type): New.
25380 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
25381 the pp_format_decoder callback. Call any m_format_postprocessor's
25383 (pretty_printer::pretty_printer): Initialize
25384 m_format_postprocessor.
25385 (pretty_printer::~pretty_printer): Delete any
25386 m_format_postprocessor.
25387 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
25388 (class format_postprocessor): New class.
25389 (struct pretty_printer::format_decoder): Document the new parameters.
25390 (struct pretty_printer::m_format_postprocessor): New field.
25391 * tree-diagnostic.c (default_tree_printer): Update for new
25392 bool and const char ** params.
25393 * tree-diagnostic.h (default_tree_printer): Likewise.
25395 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
25397 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
25398 (lwa_operand): Delete rs6000_gen_cell_microcode test.
25399 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25400 rs6000_gen_cell_microcode code.
25401 (rs6000_final_prescan_insn): Delete.
25402 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
25403 "warn-cell-microcode" entries.
25404 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
25405 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
25406 throughout. Change cc_reg_not_micro_cr0_operand to
25407 cc_reg_not_cr0_operand throughout.
25408 (*extendhi<mode>2_noload): Delete.
25409 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
25410 (mwarn-cell-microcode): Delete.
25411 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
25412 -mgen-cell-microcode and -mwarn-cell-microcode.
25414 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
25417 * config/i386/constraints.md (Yd): New constraint.
25419 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
25420 and (?Yd, r) alternatives. Update insn attributes.
25421 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
25422 and (?*Yd, r) alternatives. Update insn attributes.
25423 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
25425 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
25427 * gimplify.c (gimplify_modify_expr): Don't create a
25428 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
25431 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
25433 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
25435 2017-05-30 Richard Biener <rguenther@suse.de>
25437 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
25438 and reduc_def fields.
25439 (STMT_VINFO_REDUC_TYPE): New define.
25440 (STMT_VINFO_REDUC_DEF): Likewise.
25441 (vect_force_simple_reduction): Adjust prototype.
25442 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
25443 (vect_is_simple_reduction): Remove check_reduction argument.
25444 (vect_force_simple_reduction): Adjust and set
25445 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25446 (vectorizable_reduction): Do not re-do reduction analysis
25447 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25448 * tree-parloops.c (gather_scalar_reductions): Adjust.
25450 2017-05-30 Richard Biener <rguenther@suse.de>
25452 PR middle-end/80901
25453 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
25456 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
25458 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
25459 Introduce unknown_misalignment parameter and remove vf.
25460 (vect_peeling_hash_get_lowest_cost):
25461 Pass unknown_misalignment parameter.
25462 (vect_enhance_data_refs_alignment):
25463 Fix unsupportable data ref treatment.
25465 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25467 * tree-vect-data-refs.c (vect_get_data_access_cost):
25468 Workaround for SLP handling.
25469 (vect_enhance_data_refs_alignment):
25470 Compute costs for doing no peeling at all, compare to the best
25471 peeling costs so far and avoid peeling if cheaper.
25473 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25475 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
25476 Return peeling info and set costs to zero for unlimited cost
25478 (vect_enhance_data_refs_alignment): Also inspect all datarefs
25479 with unknown misalignment. Compute and costs for unknown
25480 misalignment, compare them to the costs for known misalignment
25481 and choose the cheapest for peeling.
25483 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25485 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
25486 (vect_get_peeling_costs_all_drs): Create function.
25487 (vect_peeling_hash_get_lowest_cost):
25488 Use vect_get_peeling_costs_all_drs.
25489 (vect_peeling_supportable): Create function.
25490 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
25492 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25494 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
25495 DR_HAS_NEGATIVE_STEP.
25496 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
25497 (vect_enhance_data_refs_alignment): Use.
25498 (vect_duplicate_ssa_name_ptr_info): Use.
25499 * tree-vectorizer.h (dr_misalignment): Use.
25500 (known_alignment_for_access_p): Use.
25502 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25505 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
25507 (has_section_name): New function.
25509 2017-05-30 Martin Liska <mliska@suse.cz>
25512 * auto-profile.c (get_function_decl_from_block): Fix
25515 2017-05-30 Richard Biener <rguenther@suse.de>
25517 PR middle-end/80876
25518 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
25520 2017-05-30 Martin Liska <mliska@suse.cz>
25522 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
25523 * dumpfile.h (struct dump_file_info): Remove ctors.
25525 2017-05-30 Martin Liska <mliska@suse.cz>
25527 * predict.def: Fix GNU coding style.
25529 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25531 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25532 Mark 'to' argument with ATTRIBUTE_UNUSED.
25534 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25536 * config/xtensa/xtensa.c (xtensa_emit_call): Use
25537 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
25538 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
25541 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
25543 * doc/install.texi (Options specification): Restore entry of
25544 --enable-sjlj-exceptions.
25546 2017-05-27 Michael Eager <eager@eagercon.com>
25549 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
25551 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
25553 * config/microblaze/microblaze.h
25554 (FIXED_REGISTERS): Update in macro.
25555 (CALL_USED_REGISTERS): Update in macro.
25557 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
25559 * doc/install.texi: Add links to macOS binary distributions.
25561 2017-05-27 Jakub Jelinek <jakub@redhat.com>
25565 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25567 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25569 2017-05-26 Martin Liska <mliska@suse.cz>
25571 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
25573 2017-05-26 Martin Liska <mliska@suse.cz>
25575 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
25576 always leading ';; '.
25577 (dump_bb_info): Likewise.
25578 (brief_dump_cfg): Likewise.
25579 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
25580 * dumpfile.c: Remove usage of TDF_VERBOSE.
25581 * dumpfile.h (enum dump_kind): Likewise.
25582 (dump_gimple_bb_header): Do not use TDF_COMMENT.
25583 * print-tree.c (debug_verbose): Remove.
25584 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
25585 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
25586 * tree-diagnostic.c (default_tree_printer): Replace
25587 TDF_DIAGNOSTIC with TDF_SLIM.
25589 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25591 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
25592 in parameter loop, rather than loop_vinfo.
25593 (create_intersect_range_checks): Ditto.
25594 (vect_create_cond_for_alias_checks): Update call to above functions.
25596 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25598 PR tree-optimization/80815
25599 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
25600 for merging runtime alias checks. Handle negative DR_STEPs.
25602 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25604 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
25606 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
25607 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
25608 out code pruning runtime alias checks.
25609 * tree-data-ref.c (prune_runtime_alias_test_list): New function
25610 factored out from above.
25611 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25613 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25615 (prune_runtime_alias_test_list): New decalaration.
25617 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25619 * tree-vect-data-refs.c (compare_tree): Rename and move ...
25620 * tree-data-ref.c (data_ref_compare_tree): ... to here.
25621 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
25622 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
25623 (operator==, comp_dr_with_seg_len_pair): Ditto.
25624 (vect_prune_runtime_alias_test_list): Ditto.
25626 2017-05-26 Martin Liska <mliska@suse.cz>
25629 * params.def: Bound partial-inlining-entry-probability param.
25631 2017-05-26 Marek Polacek <polacek@redhat.com>
25634 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
25637 2017-05-26 Richard Biener <rguenther@suse.de>
25639 PR tree-optimization/80842
25640 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
25643 2017-05-26 Richard Biener <rguenther@suse.de>
25645 PR tree-optimization/80844
25646 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
25648 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
25650 * doc/md.texi (Machine Constraints): Update x86 family
25651 machine constraints section to match 'config/i386/constraints.md'.
25653 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
25655 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
25657 2017-05-25 Nathan Sidwell <nathan@acm.org>
25659 * doc/invoke.texi (--enable-languages): Update documentation.
25661 2017-05-25 Martin Liska <mliska@suse.cz>
25663 * dumpfile.c: Add TDF_FOLDING.
25664 * dumpfile.h (enum dump_kind): Likewise.
25665 * genmatch.c (dt_simplify::gen_1): Use it.
25667 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25669 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
25671 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25673 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25674 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
25676 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25678 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
25679 * match.pd (X == C): Rewrite it here.
25680 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
25681 with_certain_nonzero_bits2): New predicates.
25682 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
25684 2017-05-24 Nathan Sidwell <nathan@acm.org>
25686 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
25689 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
25692 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
25694 * config/powerpcspe: New port. Files are copied from the rs6000
25695 port, with "rs6000" in filenames replaced by "powerpcspe".
25697 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
25699 PR rtl-optimization/80754
25700 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
25702 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
25704 * config/sparc/sparc.md (length): Return the correct value for -mflat
25705 sibcalls to match output_sibcall.
25707 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
25711 * config/rs6000/rs6000.c (struct machine_function): Add new field
25713 (rs6000_get_separate_components): Init that field, use it.
25714 (rs6000_components_for_bb): Use the field.
25716 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
25718 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
25720 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
25722 PR middle-end/80823
25723 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
25725 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25728 * config/s390/s390.c (s390_check_qrst_address): Check incoming
25729 address against address_operand predicate.
25730 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
25732 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
25734 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
25735 parameters passed indirectly.
25737 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
25739 * config/i386/i386.md (*movdi_internal): Remove SSE4
25740 alternative 18 (?r, *v). Update insn attributes.
25741 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
25742 Update insn attributes.
25743 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
25744 Update insn attributes.
25745 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
25746 alternative 1 (r, v). Remove isa attribute.
25747 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
25748 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
25749 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
25751 2017-05-23 Tom de Vries <tom@codesourcery.com>
25753 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
25756 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25758 * cgraphunit.c (symbol_table::process_new_functions): Update.
25759 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
25760 (inline_generate_summary): Rename to ...
25761 (ipa_fn_summary_generate): ... this one.
25762 (inline_read_summary): Rename to ...
25763 (ipa_fn_summary_read): ... this one.
25764 (inline_write_summary): Rename to ...
25765 (ipa_fn_summary_write): ... this one.
25766 (inline_free_summary): Rename to ...
25767 (ipa_free_fn_summary): ... this one.
25768 (pass_data_local_fn_summary, pass_local_fn_summary,
25769 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
25770 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
25771 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
25772 make_pass_ipa_fn_summary): New.
25773 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
25774 inline_write_summary, inline_free_summary): Remove.
25775 (ipa_free_fn_summary) : New.
25776 * ipa-inline.c (ipa_inline): Update.
25777 (pass_ipa_inline): Do not generate summaries.
25778 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
25780 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
25781 and add pass_ipa_fn_summary.
25782 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
25784 (make_pass_inline_parameters): Remove.
25786 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
25788 * omp-low.c (struct omp_context): Remove "default_kind" member.
25791 * omp-offload.c (execute_oacc_device_lower): Remove the
25792 parallelism dimensions function attributes for unparallelized
25793 OpenACC kernels constructs.
25795 2017-05-23 Martin Liska <mliska@suse.cz>
25797 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
25799 (cgraph_edge::make_speculative): Likewise.
25800 (cgraph_edge::resolve_speculation): Likewise.
25801 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25802 (cgraph_node::dump): Likewise.
25803 * cgraph.h: Likewise.
25804 * cgraphunit.c (analyze_functions): Likewise.
25805 (symbol_table::compile): Likewise.
25806 * ipa-cp.c (print_all_lattices): Likewise.
25807 (determine_versionability): Likewise.
25808 (initialize_node_lattices): Likewise.
25809 (ipcp_verify_propagated_values): Likewise.
25810 (estimate_local_effects): Likewise.
25811 (update_profiling_info): Likewise.
25812 (create_specialized_node): Likewise.
25813 (perhaps_add_new_callers): Likewise.
25814 (decide_about_value): Likewise.
25815 (decide_whether_version_node): Likewise.
25816 (identify_dead_nodes): Likewise.
25817 (ipcp_store_bits_results): Likewise.
25818 * ipa-devirt.c (dump_targets): Likewise.
25819 (ipa_devirt): Likewise.
25820 * ipa-icf.c (sem_item::dump): Likewise.
25821 (sem_function::equals): Likewise.
25822 (sem_variable::equals): Likewise.
25823 (sem_item_optimizer::read_section): Likewise.
25824 (sem_item_optimizer::execute): Likewise.
25825 (congruence_class::dump): Likewise.
25826 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
25827 (dump_inline_summary): Likewise.
25828 (estimate_node_size_and_time): Likewise.
25829 (inline_analyze_function): Likewise.
25830 * ipa-inline-transform.c (inline_call): Likewise.
25831 * ipa-inline.c (report_inline_failed_reason): Likewise.
25832 (want_early_inline_function_p): Likewise.
25833 (edge_badness): Likewise.
25834 (update_edge_key): Likewise.
25835 (inline_small_functions): Likewise.
25836 * ipa-profile.c (ipa_profile): Likewise.
25837 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
25838 (ipa_make_edge_direct_to_target): Likewise.
25839 (remove_described_reference): Likewise.
25840 (ipa_impossible_devirt_target): Likewise.
25841 (propagate_controlled_uses): Likewise.
25842 (ipa_print_node_params): Likewise.
25843 (ipcp_transform_function): Likewise.
25844 * ipa-pure-const.c (pure_const_read_summary): Likewise.
25845 (propagate_pure_const): Likewise.
25846 * ipa-reference.c (generate_summary): Likewise.
25847 (read_write_all_from_decl): Likewise.
25848 (propagate): Likewise.
25849 (ipa_reference_read_optimization_summary): Likewise.
25850 * ipa-utils.c (ipa_merge_profiles): Likewise.
25851 * ipa.c (walk_polymorphic_call_targets): Likewise.
25852 (symbol_table::remove_unreachable_nodes): Likewise.
25853 (ipa_single_use): Likewise.
25854 * passes.c (execute_todo): Likewise.
25855 * predict.c (drop_profile): Likewise.
25856 * symtab.c (symtab_node::get_dump_name): New function.
25857 (symtab_node::dump_name): Likewise.
25858 (symtab_node::dump_asm_name): Likewise.
25859 (symtab_node::dump_references): Likewise.
25860 (symtab_node::dump_referring): Likewise.
25861 (symtab_node::dump_base): Likewise.
25862 (symtab_node::debug_symtab): Likewise.
25863 * tree-sra.c (convert_callers_for_node): Likewise.
25864 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
25865 * value-prof.c (init_node_map): Likewise.
25867 2017-05-23 Martin Liska <mliska@suse.cz>
25869 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
25870 and symtab_node::debug_symtab to symbol_table::debug.
25871 * cgraphunit.c (analyze_functions): Use the renamed function.
25872 (symbol_table::compile): Likewise.
25873 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
25874 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
25875 * passes.c (execute_todo): Likewise.
25876 * symtab.c (symbol_table::dump): New function.
25877 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
25879 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25881 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
25882 that nonconst implies exec.
25884 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25886 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
25887 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
25888 (inline_edge_summary_vec): Turn into ...
25889 (ipa_call_summaries): ... this one.
25890 (redirect_to_unreachable, edge_set_predicate,
25891 evaluate_properties_for_edge, inline_summary_alloc,
25892 reset_ipa_call_summary, reset_inline_summary,
25893 inline_summary_t::duplicate): Update.
25894 (inline_edge_duplication_hook): Turn to ...
25895 (ipa_call_summary_t::duplicate): ... this one.
25896 (inline_edge_removal_hook): Turn to ...
25897 (ipa_call_summary_t::remove): ... this one.
25898 (dump_inline_edge_summary): Turn to ...
25899 (dump_ipa_call_summary): ... this one.
25900 (estimate_function_body_sizes): Update.
25901 (inline_update_callee_summaries): Update.
25902 (remap_edge_change_prob): Update.
25903 (remap_edge_summaries): Update.
25904 (inline_merge_summary): Update.
25905 (do_estimate_edge_time): Update.
25906 (inline_generate_summary): Update.
25907 (inline_read_section): Update.
25908 (inline_read_summary): Update.
25909 (inline_free_summary): Update.
25910 * ipa-inline.c (can_inline_edge_p): Update.
25911 (compute_inlined_call_time): Update.
25912 (want_inline_small_function_p): Update.
25913 (edge_badness): Update.
25914 (early_inliner): Update.
25915 * ipa-inline.h (inline_edge_summary): Turn to ...
25916 (ipa_call_summary): ... this one.
25917 (ipa_call_summary_t): New class.
25918 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
25919 (ipa_call_summaries): New.
25920 (inline_edge_summary): Remove.
25921 (estimate_edge_growth): Update.
25922 * ipa-profile.c (ipa_propagate_frequency_1): Update.
25923 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25924 * ipa-split.c (execute_split_functions): Update.
25925 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
25927 2017-05-23 Tom de Vries <tom@codesourcery.com>
25929 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
25930 attributes): Document rdrand effective target.
25932 2017-05-23 Tom de Vries <tom@codesourcery.com>
25934 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
25935 attributes): Sort alphabetically.
25937 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
25939 * config/avr/genmultilib.awk: Use gsub instead of gensub.
25941 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
25944 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
25945 V2DF/V2DI splat into two separate patterns, one that handles
25946 registers, and the other that only handles memory. Drop support
25947 for splatting from a GPR on ISA 2.07 and then splitting the
25948 splat into direct move and splat.
25949 (vsx_splat_<mode>_reg): Likewise.
25950 (vsx_splat_<mode>_mem): Likewise.
25952 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
25954 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
25956 2017-05-22 Jakub Jelinek <jakub@redhat.com>
25958 PR middle-end/80809
25959 * omp-low.c (finish_taskreg_remap): New function.
25960 (finish_taskreg_scan): If unit size of ctx->record_type
25961 is non-constant, unshare the size expression and replace
25962 decls in it with possible outer var refs.
25964 PR middle-end/80809
25965 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
25966 GOVD_SHARED rather than GOVD_PRIVATE with it.
25967 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
25968 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
25970 PR middle-end/80853
25971 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
25972 as last argument to build_outer_var_ref for pointer bases of array
25973 section reductions.
25975 2017-05-19 Martin Sebor <msebor@redhat.com>
25977 * print-tree.c (print_node): Print DECL_READ_P flag.
25979 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25981 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
25982 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
25983 * cgraph.c: Likewise.
25984 * cgraphunit.c: Likewise.
25985 * gengtype.c: Likewise.
25986 * ipa-cp.c: Likewise.
25987 * ipa-devirt.c: Likewise.
25988 * ipa-icf.c: Likewise.
25989 * ipa-predicate.c: Likewise.
25990 * ipa-profile.c: Likewise.
25991 * ipa-prop.c: Likewise.
25992 * ipa-split.c: Likewise.
25994 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
25995 edge_predicate_pool, dump_inline_hints,
25996 inline_summary::account_size_time, redirect_to_unreachable,
25997 edge_set_predicate, set_hint_predicate,
25998 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
25999 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
26000 inline_summary_t::remove, remap_hint_predicate_after_duplication,
26001 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
26002 ipa_call_summary_t::remove, initialize_growth_caches,
26003 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
26004 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
26005 mark_modified, unmodified_parm_1, unmodified_parm,
26006 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
26007 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
26008 compute_bb_predicates, will_be_nonconstant_expr_predicate,
26009 will_be_nonconstant_predicate, record_modified_bb_info,
26010 get_minimal_bb, record_modified, param_change_prob,
26011 phi_result_unknown_predicate, predicate_for_phi_result,
26012 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
26013 estimate_function_body_sizes, compute_inline_parameters,
26014 compute_inline_parameters_for_curren, pass_data_inline_parameters,
26015 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
26016 inline_update_callee_summaries, remap_edge_change_prob,
26017 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
26018 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
26019 inline_analyze_function, inline_summary_t::insert,
26020 inline_generate_summary, read_ipa_call_summary, inline_read_section,
26021 inline_read_summary, write_ipa_call_summary, inline_write_summary,
26022 inline_free_summary): Move to ipa-fnsummary.h
26023 (predicate_t): Remove.
26024 * ipa-fnsummary.c: New file.
26025 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
26026 (enum inline_hints_vals, inline_hints, agg_position_info,
26027 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
26028 inline_summaries, ipa_call_summary, ipa_call_summary_t,
26029 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
26030 dump_inline_summary, dump_inline_hints, inline_generate_summary,
26031 inline_read_summary, inline_write_summary, inline_free_summary,
26032 inline_analyze_function, initialize_inline_failed,
26033 inline_merge_summary, inline_update_overall_summary,
26034 compute_inline_parameters): Move to ipa-fnsummary.h
26035 * ipa-fnsummary.h: New file.
26036 * ipa-inline-transform.h: Include ipa-inline.h.
26037 * ipa-inline.c: LIkewise.
26039 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26041 * ipa-inline.c (edge_badness): Use inlined_time instead of
26042 inline_summaries->get.
26044 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26046 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
26048 2017-05-22 Nathan Sidwell <nathan@acm.org>
26050 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
26051 (fdump-lang): Document 'raw' option.
26052 * dumpfile.h (TDI_tu): Delete.
26053 * dumpfile.c (dump_files): Remove translation-unit.
26054 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
26056 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
26058 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
26059 command option from $(AWK) call.
26060 * config/avr/genmultilib.awk: Simplify and rewrite so that it
26061 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
26062 [FORMAT]: Remove handling of variable.
26063 * config/avr/t-multilib: Regenerate.
26065 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26067 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
26069 (dump_inline_summary): Do not print self_time.
26070 (estimate_function_body_sizes): Do not set self_time.
26071 (compute_inline_parameters): Likewise.
26072 (inline_read_section, inline_write_summary): Do not stream self_time.
26073 * ipa-inline.h (inline_summary): Drop self_time.
26075 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26077 * ipa-inline-analysis.c (account_size_time): Rename to ...
26078 (inline_summary::account_size_time): ... this one.
26079 (reset_ipa_call_summary): Turn to ...
26080 (ipa_call_summary::reset): ... this one.
26081 (reset_inline_summary): Turn to ...
26082 (inline_summary::reset): ... this one.
26083 (inline_summary_t::remove): Update.
26084 (inline_summary_t::duplicate): Update.
26085 (ipa_call_summary_t::remove): Update.
26086 (dump_inline_summary): Update.
26087 (estimate_function_body_sizes): Update.
26088 (compute_inline_parameters): Update.
26089 (estimate_node_size_and_time): Update.
26090 (inline_merge_summary): Update.
26091 (inline_update_overall_summary): Update.
26092 (inline_read_section): Update.
26093 (inline_write_summary): Update.
26094 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
26095 add account_size_time and reset member functions.
26096 (ipa_call_summary): Add reset function.
26097 * ipa-predicate.h (predicate::operator &): Constify.
26099 2017-05-22 Richard Biener <rguenther@suse.de>
26101 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
26103 2017-05-19 Jason Merrill <jason@redhat.com>
26105 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
26107 2017-05-19 Marek Polacek <polacek@redhat.com>
26110 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
26111 TYPE_OVERFLOW_WRAPS checks.
26113 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
26115 * tree-core.h (enum omp_clause_default_kind): Add
26116 "OMP_CLAUSE_DEFAULT_PRESENT".
26117 * tree-pretty-print.c (dump_omp_clause): Handle it.
26118 * gimplify.c (enum gimplify_omp_var_data): Add
26119 "GOVD_MAP_FORCE_PRESENT".
26120 (gimplify_adjust_omp_clauses_1): Map it to
26121 "GOMP_MAP_FORCE_PRESENT".
26122 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
26124 * gimplify.c (oacc_default_clause): Clarify.
26126 2017-05-19 Nathan Sidwell <nathan@acm.org>
26128 LANG_HOOK_REGISTER_DUMPS
26129 * toplev.c (general_init): Call register dump lang hook.
26130 * doc/invoke.texi: Document -fdump-lang option family.
26131 * dumpfile.c (dump_files): Remove class dump here.
26132 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
26133 * dumpfile.h (tree_dump_index): Remove TDI_class.
26134 * langhooks-def.h (lhd_register_dumps): Declare.
26135 (LANG_HOOKS_REGISTER_DUMPS): Define.
26136 (LANG_HOOKS_INITIALIZER): Add it.
26137 * langhooks.c (lhd_register_dumps): Define.
26138 * langhooks.h (struct lang_hooks): Add register_dumps.
26140 2017-05-19 Nathan Sidwell <nathan@acm.org>
26142 * context.h (context::set_passes): New.
26143 * context.c (context::context): Do not create pass manager.
26144 * toplev.c (general_init): Create pass manager here.
26146 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
26148 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
26149 use this splitter if two add or or instructions would also work for
26150 the constant we want to generate.
26152 2017-05-19 Richard Biener <rguenther@suse.de>
26155 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
26156 predicate evaluation.
26158 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
26160 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
26162 * ipa-inline.c (want_inline_small_function_p): Do not cast to
26165 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
26167 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26168 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26169 (inline_edge_summary_vec): Turn into ...
26170 (ipa_call_summaries): ... this one.
26171 (redirect_to_unreachable, edge_set_predicate,
26172 evaluate_properties_for_edge, inline_summary_alloc,
26173 reset_ipa_call_summary, reset_inline_summary,
26174 inline_summary_t::duplicate): Update.
26175 (inline_edge_duplication_hook): Turn to ...
26176 (ipa_call_summary_t::duplicate): ... this one.
26177 (inline_edge_removal_hook): Turn to ...
26178 (ipa_call_summary_t::remove): ... this one.
26179 (dump_inline_edge_summary): Turn to ...
26180 (dump_ipa_call_summary): ... this one.
26181 (estimate_function_body_sizes): Update.
26182 (inline_update_callee_summaries): Update.
26183 (remap_edge_change_prob): Update.
26184 (remap_edge_summaries): Update.
26185 (inline_merge_summary): Update.
26186 (do_estimate_edge_time): Update.
26187 (inline_generate_summary): Update.
26188 (inline_read_section): Update.
26189 (inline_read_summary): Update.
26190 (inline_free_summary): Update.
26191 * ipa-inline.c (can_inline_edge_p): Update.
26192 (compute_inlined_call_time): Update.
26193 (want_inline_small_function_p): Update.
26194 (edge_badness): Update.
26195 (early_inliner): Update.
26196 * ipa-inline.h (inline_edge_summary): Turn to ...
26197 (ipa_call_summary): ... this one.
26198 (ipa_call_summary_t): New class.
26199 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26200 (ipa_call_summaries): New.
26201 (inline_edge_summary): Remove.
26202 (estimate_edge_growth): Update.
26203 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26204 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26205 * ipa-split.c (execute_split_functions): Update.
26206 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26208 2017-05-19 Richard Biener <rguenther@suse.de>
26210 PR middle-end/80764
26211 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
26213 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
26215 * config/rs6000/rs6000.c (struct machine_function): Add field
26216 fpr_is_wrapped_separately.
26217 (rs6000_get_separate_components): Use 64 components. Handle the
26218 new FPR components.
26219 (rs6000_components_for_bb): Handle the FPR components.
26220 (rs6000_emit_prologue_components): Handle the FPR components.
26221 (rs6000_emit_epilogue_components): Handle the FPR components.
26222 (rs6000_set_handled_components): Handle the FPR components.
26223 (rs6000_emit_prologue): Don't output prologue code for those FPRs
26224 that are already separately shrink-wrapped.
26225 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
26226 that are already separately shrink-wrapped.
26228 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26231 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
26234 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
26235 (define_peephole2 for Altivec d-form load): Add peepholes to catch
26236 cases where the register allocator uses a move and an offsettable
26237 memory operation to/from a FPR register on ISA 2.06/2.07.
26238 (define_peephole2 for Altivec d-form store): Likewise.
26240 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
26243 * config/i386/mmx.md (*mov<mode>_internal): Enable
26244 alternatives 11, 12, 13 and 14 also for 32bit targets.
26245 Remove alternatives 15, 16, 17 and 18.
26246 * config/i386/sse.md (vec_concatv2di): Change
26247 alternative (!x, *y) to (x, ?!*Yn).
26249 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
26251 * dumpfile.h (enum dump_kind): Remove stray comma.
26253 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
26255 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
26256 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
26257 predicate::num_conditions
26258 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
26259 (CHANGED): turn into predicate::changed.
26260 (agg_position_info): Move to ipa-predicate.h
26261 (add_condition, predicate::add_clause, predicate::operator &=,
26262 predicate::or_with, predicate::evaluate, predicate::probability,
26263 dump_condition, dump_clause, predicate::dump,
26264 predicate::remap_after_duplication, predicate::remap_after_inlining,
26265 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
26266 (evaluate_conditions_for_known_args): Update.
26267 (set_cond_stmt_execution_predicate): Update.
26268 * ipa-inline.h: Include ipa-predicate.h
26269 (condition, inline_param_summary, conditions, agg_position_info,
26270 predicate): Move to ipa-predicate.h
26271 * ipa-predicate.c: New file.
26272 * ipa-predicate.h: New file.
26274 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
26276 * final.c (leaf_function_p): Check we are not in a sequence.
26278 2017-05-18 Martin Liska <mliska@suse.cz>
26280 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
26281 * dumpfile.c (dump_register): Use new enum dump_kind.
26282 (get_dump_file_name): Likewise.
26283 (dump_enable_all): Likewise.
26284 (dump_switch_p_1): Likewise.
26285 (enable_rtl_dump_file): Remove usage of TDF_RTL.
26286 * dumpfile.h (enum dump_kind): New enum type.
26287 (struct dump_file_info): Create constructor and
26288 format fields and comments.
26289 * passes.c (pass_manager::register_one_dump_file):
26291 * statistics.c (statistics_early_init): Likewise.
26292 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
26293 TDF_TREE with TDF_SLIM.
26294 (gather_memory_references_ref): Likewise.
26296 2017-05-18 Martin Liska <mliska@suse.cz>
26298 * vec.h (struct vnull): Use it.
26300 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
26302 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
26303 (true_predicate, false_predicate, true_predicate_p,
26304 false_predicate_p): Remove.
26305 (single_cond_predicate, not_inlined_predicate): Turn to member function
26307 (add_condition): Update.
26308 (add_clause): Turn to...
26309 (predicate::add_clause): ... this one; update; allow passing NULL
26311 (and_predicates): Turn to ...
26312 (predicate::operator &=): ... this one.
26313 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
26314 (or_predicates): Turn to ...
26315 (predicate::or_with): ... this one.
26316 (evaluate_predicate): Turn to ...
26317 (predicate::evaluate): ... this one.
26318 (predicate_probability): Turn to ...
26319 (predicate::probability): ... this one.
26320 (dump_condition): Update.
26321 (dump_predicate): Turn to ...
26322 (predicate::dump): ... this one.
26323 (account_size_time): Update.
26324 (edge_set_predicate): Update.
26325 (set_hint_predicate): UPdate.
26326 (evaluate_conditions_for_known_args): Update.
26327 (evaluate_properties_for_edge): Update.
26328 (remap_predicate_after_duplication): Turn to...
26329 (predicate::remap_after_duplication): ... this one.
26330 (remap_hint_predicate_after_duplication): Update.
26331 (inline_summary_t::duplicate): UPdate.
26332 (dump_inline_edge_summary): Update.
26333 (dump_inline_summary): Update.
26334 (set_cond_stmt_execution_predicate): Update.
26335 (set_switch_stmt_execution_predicate): Update.
26336 (compute_bb_predicates): Update.
26337 (will_be_nonconstant_expr_predicate): Update.
26338 (will_be_nonconstant_predicate): Update.
26339 (phi_result_unknown_predicate): Update.
26340 (predicate_for_phi_result): Update.
26341 (array_index_predicate): Update.
26342 (estimate_function_body_sizes): Update.
26343 (estimate_node_size_and_time): Update.
26344 (estimate_ipcp_clone_size_and_time): Update.
26345 (remap_predicate): Rename to ...
26346 (predicate::remap_after_inlining): ... this one.
26347 (remap_hint_predicate): Update.
26348 (inline_merge_summary): Update.
26349 (inline_update_overall_summary): Update.
26350 (estimate_size_after_inlining): Update.
26351 (read_predicate): Rename to ...
26352 (predicate::stream_in): ... this one.
26353 (read_inline_edge_summary): Update.
26354 (write_predicate): Rename to ...
26355 (predicate::stream_out): ... this one.
26356 (write_inline_edge_summary): Update.
26357 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
26358 (clause_t): Turn to uint32_t
26359 (predicate): Turn to class; implement constructor and operators
26361 (size_time_entry): Update.
26362 (inline_summary): Update.
26363 (inline_edge_summary): Update.
26365 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
26367 * fold-const.c (fold_binary_loc): Move transformation...
26368 * match.pd (C - X CMP X): ... here.
26370 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
26372 * config/sparc/sparc.c (sparc_option_override): Set function
26373 alignment for -mcpu=niagara7 to 64 to match the I$ line.
26374 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
26376 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
26378 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
26380 2017-05-18 Marek Polacek <polacek@redhat.com>
26383 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
26384 (pass_ubsan::execute): Call gimple_assign_single_p instead of
26385 gimple_assign_load_p.
26387 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
26389 PR middle-end/80692
26390 * real.c (do_compare): Give decimal_do_compare preference over
26391 comparing just the signs.
26393 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
26395 * doc/md.texi (Canonicalization of Instructions): Describe the
26396 canonical form of instructions that inherently set a condition
26399 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
26401 PR middle-end/80775
26402 * tree-cfg.c: Move deletion of unreachable case statements to after
26403 the merging of consecutive case labels.
26405 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
26407 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
26408 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
26409 restoring of callee-saved registers.
26411 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
26413 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
26414 * config/visium/visium.c (single_set_and_flags): Likewise.
26415 * config/visium/visium.md (Substitutions): Likewise.
26417 2017-05-17 Martin Liska <mliska@suse.cz>
26419 * cfg.c: Introduce dump_flags_t type and
26420 use it instead of int type.
26422 * cfghooks.c: Likewise.
26423 * cfghooks.h (struct cfg_hooks): Likewise.
26424 * cfgrtl.c: Likewise.
26425 * cfgrtl.h: Likewise.
26426 * cgraph.c (cgraph_node::get_body): Likewise.
26427 * coretypes.h: Likewise.
26428 * domwalk.c: Likewise.
26429 * domwalk.h: Likewise.
26430 * dumpfile.c (struct dump_option_value_info): Likewise.
26431 (dump_enable_all): Likewise.
26432 (dump_switch_p_1): Likewise.
26433 (opt_info_switch_p): Likewise.
26434 * dumpfile.h (enum tree_dump_index): Likewise.
26435 (struct dump_file_info): Likewise.
26436 * genemit.c: Likewise.
26437 * generic-match-head.c: Likewise.
26438 * gengtype.c (open_base_files): Likewise.
26439 * gimple-pretty-print.c: Likewise.
26440 * gimple-pretty-print.h: Likewise.
26441 * graph.c (print_graph_cfg): Likewise.
26442 * graphite-scop-detection.c (dot_all_sese): Likewise.
26443 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
26444 * loop-unroll.c (report_unroll): Likewise.
26445 * passes.c (pass_manager::register_one_dump_file): Likewise.
26446 * print-tree.c: Likewise.
26447 * statistics.c: Likewise.
26448 * tree-cfg.c: Likewise.
26449 * tree-cfg.h: Likewise.
26450 * tree-dfa.c: Likewise.
26451 * tree-dfa.h: Likewise.
26452 * tree-dump.c (dump_function): Likewise.
26453 * tree-dump.h (struct dump_info): Likewise.
26454 * tree-pretty-print.c: Likewise.
26455 * tree-pretty-print.h: Likewise.
26456 * tree-ssa-live.c: Likewise.
26457 * tree-ssa-live.h: Likewise.
26458 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
26459 * tree-vect-loop.c: Likewise.
26460 * tree-vect-slp.c: Likewise.
26462 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
26463 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26465 PR tree-optimization/80457
26466 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
26467 of all arguments to a statement as scalar_to_vec operations.
26468 (vectorizable_call): Adjust call to vect_model_simple_cost for
26470 (vectorizable_conversion): Likewise.
26471 (vectorizable_assignment): Likewise.
26472 (vectorizable_shift): Likewise.
26473 (vectorizable_operation): Likewise.
26474 (vectorizable_comparison): Likewise.
26475 (vect_is_simple_cond): Record the def types for operands.
26476 (vectorizable_condition): Likewise, call vect_model_simple_cost.
26477 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
26478 for statement argument count.
26480 2017-05-16 Carl Love <cel@us.ibm.com>
26482 * config/rs6000/rs6000-c: Add support for built-in functions
26483 vector unsigned long long vec_bperm (vector unsigned long long,
26484 vector unsigned char)
26485 vector signed long long vec_mule (vector signed int,
26487 vector unsigned long long vec_mule (vector unsigned int,
26488 vector unsigned int)
26489 vector signed long long vec_mulo (vector signed int,
26491 vector unsigned long long vec_mulo (vector unsigned int,
26492 vector unsigned int)
26493 vector signed char vec_sldw (vector signed char,
26494 vector signed char,
26496 vector unsigned char vec_sldw (vector unsigned char,
26497 vector unsigned char,
26499 vector signed short vec_sldw (vector signed short,
26500 vector signed short,
26502 vector unsigned short vec_sldw (vector unsigned short,
26503 vector unsigned short,
26505 vector signed int vec_sldw (vector signed int,
26508 vector unsigned int vec_sldw (vector unsigned int,
26509 vector unsigned int,
26511 vector signed long long vec_sldw (vector signed long long,
26512 vector signed long long,
26514 vector unsigned long long vec_sldw (vector unsigned long long,
26515 vector unsigned long long,
26517 * config/rs6000/rs6000-c: Add support for built-in functions
26518 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
26519 * config/rs6000/altivec.h: Add defintion for vec_sldw.
26520 * doc/extend.texi: Update the built-in documentation for the
26521 new built-in functions.
26523 2017-05-16 Marek Polacek <polacek@redhat.com>
26527 * tree.c (save_expr): Don't fold the expression.
26529 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
26531 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
26532 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
26533 and (?*y,m). Update insn attributes.
26535 2017-05-16 Martin Liska <mliska@suse.cz>
26537 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
26538 flags argument of print_gimple_stmt, print_gimple_expr,
26539 print_generic_stmt and print_generic_expr.
26540 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
26541 * coretypes.h: Likewise.
26542 * except.c (dump_eh_tree): Likewise.
26543 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
26544 * gimple-pretty-print.h: Likewise.
26545 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
26546 (backprop::push_to_worklist): Likewise.
26547 (backprop::pop_from_worklist): Likewise.
26548 (backprop::process_use): Likewise.
26549 (backprop::intersect_uses): Likewise.
26550 (note_replacement): Likewise.
26551 * gimple-ssa-store-merging.c
26552 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
26553 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
26554 (pass_store_merging::execute): Likewise.
26555 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
26556 (ssa_base_cand_dump_callback): Likewise.
26557 (dump_incr_vec): Likewise.
26558 (replace_refs): Likewise.
26559 (replace_mult_candidate): Likewise.
26560 (create_add_on_incoming_edge): Likewise.
26561 (create_phi_basis): Likewise.
26562 (insert_initializers): Likewise.
26563 (all_phi_incrs_profitable): Likewise.
26564 (introduce_cast_before_cand): Likewise.
26565 (replace_one_candidate): Likewise.
26566 * gimplify.c (gimplify_expr): Likewise.
26567 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
26568 (set_rename): Likewise.
26569 (rename_uses): Likewise.
26570 (copy_loop_phi_nodes): Likewise.
26571 (add_close_phis_to_merge_points): Likewise.
26572 (copy_loop_close_phi_args): Likewise.
26573 (copy_cond_phi_args): Likewise.
26574 (graphite_copy_stmts_from_block): Likewise.
26575 (translate_pending_phi_nodes): Likewise.
26576 * graphite-poly.c (print_pdr): Likewise.
26577 (dump_gbb_cases): Likewise.
26578 (dump_gbb_conditions): Likewise.
26579 (print_scop_params): Likewise.
26580 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
26581 (build_cross_bb_scalars_use): Likewise.
26582 (gather_bbs::before_dom_children): Likewise.
26583 * hsa-dump.c (dump_hsa_immed): Likewise.
26584 * ipa-cp.c (print_ipcp_constant_value): Likewise.
26585 (get_replacement_map): Likewise.
26586 * ipa-inline-analysis.c (dump_condition): Likewise.
26587 (estimate_function_body_sizes): Likewise.
26588 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
26589 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
26590 * ipa-prop.c (ipa_dump_param): Likewise.
26591 (ipa_print_node_jump_functions_for_edge): Likewise.
26592 (ipa_modify_call_arguments): Likewise.
26593 (ipa_modify_expr): Likewise.
26594 (ipa_dump_param_adjustments): Likewise.
26595 (ipa_dump_agg_replacement_values): Likewise.
26596 (ipcp_modif_dom_walker::before_dom_children): Likewise.
26597 * ipa-pure-const.c (check_stmt): Likewise.
26598 (pass_nothrow::execute): Likewise.
26599 * ipa-split.c (execute_split_functions): Likewise.
26600 * omp-offload.c (dump_oacc_loop_part): Likewise.
26601 (dump_oacc_loop): Likewise.
26602 * trans-mem.c (tm_log_emit): Likewise.
26603 (tm_memopt_accumulate_memops): Likewise.
26604 (dump_tm_memopt_set): Likewise.
26605 (dump_tm_memopt_transform): Likewise.
26606 * tree-cfg.c (gimple_verify_flow_info): Likewise.
26607 (print_loop): Likewise.
26608 * tree-chkp-opt.c (chkp_print_addr): Likewise.
26609 (chkp_gather_checks_info): Likewise.
26610 (chkp_get_check_result): Likewise.
26611 (chkp_remove_check_if_pass): Likewise.
26612 (chkp_use_outer_bounds_if_possible): Likewise.
26613 (chkp_reduce_bounds_lifetime): Likewise.
26614 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
26615 (chkp_mark_completed_bounds): Likewise.
26616 (chkp_register_incomplete_bounds): Likewise.
26617 (chkp_mark_invalid_bounds): Likewise.
26618 (chkp_maybe_copy_and_register_bounds): Likewise.
26619 (chkp_build_returned_bound): Likewise.
26620 (chkp_get_bound_for_parm): Likewise.
26621 (chkp_build_bndldx): Likewise.
26622 (chkp_get_bounds_by_definition): Likewise.
26623 (chkp_generate_extern_var_bounds): Likewise.
26624 (chkp_get_bounds_for_decl_addr): Likewise.
26625 * tree-chrec.c (chrec_apply): Likewise.
26626 * tree-data-ref.c (dump_data_reference): Likewise.
26627 (dump_subscript): Likewise.
26628 (dump_data_dependence_relation): Likewise.
26629 (analyze_overlapping_iterations): Likewise.
26630 * tree-inline.c (expand_call_inline): Likewise.
26631 (tree_function_versioning): Likewise.
26632 * tree-into-ssa.c (dump_defs_stack): Likewise.
26633 (dump_currdefs): Likewise.
26634 (dump_names_replaced_by): Likewise.
26635 (dump_update_ssa): Likewise.
26636 (update_ssa): Likewise.
26637 * tree-object-size.c (pass_object_sizes::execute): Likewise.
26638 * tree-parloops.c (build_new_reduction): Likewise.
26639 (try_create_reduction_list): Likewise.
26640 (ref_conflicts_with_region): Likewise.
26641 (oacc_entry_exit_ok_1): Likewise.
26642 (oacc_entry_exit_single_gang): Likewise.
26643 * tree-pretty-print.h: Likewise.
26644 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
26645 (get_scalar_evolution): Likewise.
26646 (add_to_evolution): Likewise.
26647 (get_loop_exit_condition): Likewise.
26648 (analyze_evolution_in_loop): Likewise.
26649 (analyze_initial_condition): Likewise.
26650 (analyze_scalar_evolution): Likewise.
26651 (instantiate_scev): Likewise.
26652 (number_of_latch_executions): Likewise.
26653 (gather_chrec_stats): Likewise.
26654 (final_value_replacement_loop): Likewise.
26655 (scev_const_prop): Likewise.
26656 * tree-sra.c (dump_access): Likewise.
26657 (disqualify_candidate): Likewise.
26658 (create_access): Likewise.
26659 (reject): Likewise.
26660 (maybe_add_sra_candidate): Likewise.
26661 (create_access_replacement): Likewise.
26662 (analyze_access_subtree): Likewise.
26663 (analyze_all_variable_accesses): Likewise.
26664 (sra_modify_assign): Likewise.
26665 (initialize_constant_pool_replacements): Likewise.
26666 (find_param_candidates): Likewise.
26667 (decide_one_param_reduction): Likewise.
26668 (replace_removed_params_ssa_names): Likewise.
26669 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
26670 * tree-ssa-copy.c (dump_copy_of): Likewise.
26671 (copy_prop_visit_cond_stmt): Likewise.
26672 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
26673 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
26674 (record_equivalences_from_stmt): Likewise.
26675 * tree-ssa-dse.c (compute_trims): Likewise.
26676 (delete_dead_call): Likewise.
26677 (delete_dead_assignment): Likewise.
26678 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
26679 (forward_propagate_into_cond): Likewise.
26680 (pass_forwprop::execute): Likewise.
26681 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
26682 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
26684 (move_computations_worker): Likewise.
26685 (execute_sm): Likewise.
26686 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
26687 (remove_exits_and_undefined_stmts): Likewise.
26688 (remove_redundant_iv_tests): Likewise.
26689 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
26690 (adjust_iv_update_pos): Likewise.
26691 * tree-ssa-math-opts.c (bswap_replace): Likewise.
26692 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
26693 (value_replacement): Likewise.
26694 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
26695 * tree-ssa-pre.c (print_pre_expr): Likewise.
26696 (get_representative_for): Likewise.
26697 (create_expression_by_pieces): Likewise.
26698 (insert_into_preds_of_block): Likewise.
26699 (eliminate_insert): Likewise.
26700 (eliminate_dom_walker::before_dom_children): Likewise.
26701 (eliminate): Likewise.
26702 (remove_dead_inserted_code): Likewise.
26703 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
26704 * tree-ssa-reassoc.c (get_rank): Likewise.
26705 (eliminate_duplicate_pair): Likewise.
26706 (eliminate_plus_minus_pair): Likewise.
26707 (eliminate_not_pairs): Likewise.
26708 (undistribute_ops_list): Likewise.
26709 (eliminate_redundant_comparison): Likewise.
26710 (update_range_test): Likewise.
26711 (optimize_range_tests_var_bound): Likewise.
26712 (optimize_vec_cond_expr): Likewise.
26713 (rewrite_expr_tree): Likewise.
26714 (rewrite_expr_tree_parallel): Likewise.
26715 (linearize_expr): Likewise.
26716 (break_up_subtract): Likewise.
26717 (linearize_expr_tree): Likewise.
26718 (attempt_builtin_powi): Likewise.
26719 (attempt_builtin_copysign): Likewise.
26720 (transform_stmt_to_copy): Likewise.
26721 (transform_stmt_to_multiply): Likewise.
26722 (dump_ops_vector): Likewise.
26723 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
26724 (print_scc): Likewise.
26725 (set_ssa_val_to): Likewise.
26726 (visit_reference_op_store): Likewise.
26727 (visit_use): Likewise.
26728 (sccvn_dom_walker::before_dom_children): Likewise.
26729 (run_scc_vn): Likewise.
26730 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
26732 (expr_hash_elt::print): Likewise.
26733 (const_and_copies::pop_to_marker): Likewise.
26734 (const_and_copies::record_const_or_copy_raw): Likewise.
26735 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
26736 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
26737 (dump_predicates): Likewise.
26738 (find_uninit_use): Likewise.
26739 (warn_uninitialized_phi): Likewise.
26740 (pass_late_warn_uninitialized::execute): Likewise.
26741 * tree-ssa.c (verify_vssa): Likewise.
26742 (verify_ssa): Likewise.
26743 (maybe_optimize_var): Likewise.
26744 * tree-vrp.c (dump_value_range): Likewise.
26745 (dump_all_value_ranges): Likewise.
26746 (dump_asserts_for): Likewise.
26747 (register_edge_assert_for_2): Likewise.
26748 (vrp_visit_cond_stmt): Likewise.
26749 (vrp_visit_switch_stmt): Likewise.
26750 (vrp_visit_stmt): Likewise.
26751 (vrp_visit_phi_node): Likewise.
26752 (simplify_cond_using_ranges_1): Likewise.
26753 (fold_predicate_in): Likewise.
26754 (evrp_dom_walker::before_dom_children): Likewise.
26755 (evrp_dom_walker::push_value_range): Likewise.
26756 (evrp_dom_walker::pop_value_range): Likewise.
26757 (execute_early_vrp): Likewise.
26759 2017-05-16 Richard Biener <rguenther@suse.de>
26761 * dwarf2out.c (loc_list_from_tree_1): Do not create
26762 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
26764 2017-05-16 Richard Biener <rguenther@suse.de>
26766 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
26768 (note_variable_value_in_expr): If we resolved the decl ref
26769 do not push to the stack.
26771 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
26773 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
26774 operations in fast-math mode.
26775 (vaddq_f16): Likewise.
26776 (vmul_f16): Likewise.
26777 (vmulq_f16): Likewise.
26778 (vsub_f16): Likewise.
26779 (vsubq_f16): Likewise.
26780 * config/arm/neon.md (add<mode>3): New.
26782 (fma:<VH:mode>3): New. Also remove outdated comment.
26785 2017-05-16 Martin Liska <mliska@suse.cz>
26789 * ipa-devirt.c (warn_types_mismatch): Fix typo.
26790 (odr_types_equivalent_p): Likewise.
26792 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
26794 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
26796 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
26799 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
26800 non-interunit SSE move alternatives with '?'.
26801 (zero-extendsidi peephole2): New peephole to skip intermediate
26802 general register in SSE zero-extend sequence.
26804 2017-05-15 Jeff Law <law@redhat.com>
26806 * reorg.c (relax_delay_slots): Create a new variable to hold
26807 the temporary target rather than clobbering TARGET_LABEL.
26809 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
26810 missing argument to extract_bit_field call.
26811 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
26813 2017-05-15 Martin Liska <mliska@suse.cz>
26816 * gcc.c (process_command): Do not allow empty argument of -o option.
26818 2017-05-15 Renlin Li <renlin.li@arm.com>
26820 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
26821 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
26822 * config/aarch64/constraints.md (Usf): Add long call check.
26823 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
26824 (call_value): Likewise.
26825 (sibcall): Likewise.
26826 (sibcall_value): Likewise.
26828 (call_value_insn): New.
26829 (sibcall_insn): Update rtx pattern.
26830 (sibcall_value_insn): Likewise.
26831 (call_internal): Remove.
26832 (call_value_internal): Likewise.
26833 (sibcall_internal): Likewise.
26834 (sibcall_value_internal): Likewise.
26835 (call_reg): Likewise.
26836 (call_symbol): Likewise.
26837 (call_value_reg): Likewise.
26838 (call_value_symbol): Likewise.
26840 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
26843 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
26845 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
26847 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
26848 compatible with CCGOCmode and with CCZmode.
26850 2017-05-14 Martin Sebor <msebor@redhat.com>
26852 PR middle-end/77671
26853 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
26854 (gimple_fold_builtin_snprintf): Same.
26855 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
26856 (gimple_fold_builtin_snprintf): Same.
26857 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
26858 of character types.
26859 (is_call_safe): New function.
26860 (try_substitute_return_value): Call it.
26861 (try_simplify_call): New function.
26862 (pass_sprintf_length::handle_gimple_call): Call it.
26864 2017-05-14 Martin Sebor <msebor@redhat.com>
26866 PR middle-end/80669
26867 * builtins.c (expand_builtin_stpncpy): Simplify.
26869 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
26871 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
26872 * config/i386/i386.h
26873 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
26874 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
26875 (struct machine_function): Add new members call_ms2sysv,
26876 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
26877 (struct machine_frame_state): New fields sp_realigned and
26878 sp_realigned_offset.
26879 * config/i386/i386.c
26880 (enum xlogue_stub): New enum.
26881 (enum xlogue_stub_sets): New enum.
26882 (class xlogue_layout): New class.
26883 (struct ix86_frame): New fields stack_realign_allocate_offset,
26884 stack_realign_offset and outlined_save_offset. Modify comments to
26885 detail stack layout when using out-of-line stubs.
26886 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
26887 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
26888 -mcall-ms2sysv-xlogues.
26889 (stub_managed_regs): New static variable.
26890 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
26891 registers managed by out-of-line stub.
26892 (disable_call_ms2sysv_xlogues): New function.
26893 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
26894 m->call_ms2sysv when appropriate and compute frame layout for
26896 (sp_valid_at, fp_valid_at): New inline functions.
26897 (choose_basereg): New function.
26898 (choose_baseaddr): Add align parameter, use choose_basereg and modify
26900 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
26901 Use align parameter of choose_baseaddr to generated aligned SSE movs
26903 (pro_epilogue_adjust_stack): Modify to track
26904 machine_frame_state::sp_realigned.
26905 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
26906 (ix86_nsaved_sseregs): Likewise.
26907 (ix86_emit_save_regs): Likewise.
26908 (ix86_emit_save_regs_using_mov): Likewise.
26909 (ix86_emit_save_sse_regs_using_mov): Likewise.
26910 (get_scratch_register_on_entry): Likewise.
26911 (gen_frame_set): New function.
26912 (gen_frame_load): Likewise.
26913 (gen_frame_store): Likewise.
26914 (emit_outlined_ms2sysv_save): Likewise.
26915 (emit_outlined_ms2sysv_restore): Likewise.
26916 (ix86_expand_prologue): Modify stack re-alignment code and call
26917 emit_outlined_ms2sysv_save when appropriate.
26918 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
26919 parameter rtx_insn *insn, which allows the function to be used to only
26920 generate the notes.
26921 (ix86_expand_epilogue): Modify validity checks of frame and stack
26922 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
26923 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
26924 * config/i386/predicates.md
26925 (save_multiple): New predicate.
26926 (restore_multiple): Likewise.
26927 * config/i386/sse.md
26928 (save_multiple<mode>): New pattern.
26929 (save_multiple_realign<mode>): Likewise.
26930 (restore_multiple<mode>): Likewise.
26931 (restore_multiple_and_return<mode>): Likewise.
26932 (restore_multiple_leave_return<mode>): Likewise.
26933 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
26935 2017-05-14 Julia Koval <julia.koval@intel.com>
26937 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
26938 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
26939 (__builtin_ia32_xsetbv): New builtins.
26940 * config/i386/i386.c (ix86_expand_special_args_builtin):
26942 (ix86_expand_builtin): Special expand for new intrinsics.
26943 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
26944 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
26945 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
26947 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26949 * cfganal.c (inverted_post_order_compute): Change argument type
26951 * cfganal.h (inverted_post_order_compute): Adjust prototype.
26952 * df-core.c (rest_of_handle_df_initialize): Adjust.
26953 (rest_of_handle_df_finish): Likewise.
26954 (df_analyze_1): Likewise.
26955 (df_analyze): Likewise.
26956 (loop_inverted_post_order_compute): Change argument to be a vec *.
26957 (df_analyze_loop): Adjust.
26958 (df_get_n_blocks): Likewise.
26959 (df_get_postorder): Likewise.
26960 * df.h (struct df_d): Change field to be a vec.
26961 * lcm.c (compute_laterin): Adjust.
26962 (compute_available): Likewise.
26963 * lra-lives.c (lra_create_live_ranges_1): Likewise.
26964 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
26965 * tree-ssa-pre.c (compute_antic): Likewise.
26967 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26969 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
26970 (depth_first_search::depth_first_search): Change structure init
26971 function to this constructor.
26972 (depth_first_search::add_bb): Rename function to this member.
26973 (depth_first_search::execute): Likewise.
26974 (flow_dfs_compute_reverse_finish): Adjust.
26976 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26978 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
26979 (longest_simple_path): Likewise.
26980 * shrink-wrap.c (spread_components): Likewise.
26981 (disqualify_problematic_components): Likewise.
26982 (emit_common_heads_for_components): Likewise.
26983 (emit_common_tails_for_components): Likewise.
26984 (insert_prologue_epilogue_for_components): Likewise.
26986 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26988 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
26991 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26993 * df-core.c (df_set_blocks): Start using auto_bitmap.
26994 (df_compact_blocks): Likewise.
26995 * df-problems.c (df_rd_confluence_n): Likewise.
26996 * df-scan.c (df_insn_rescan_all): Likewise.
26997 (df_process_deferred_rescans): Likewise.
26998 (df_update_entry_block_defs): Likewise.
26999 (df_update_exit_block_uses): Likewise.
27000 (df_entry_block_bitmap_verify): Likewise.
27001 (df_exit_block_bitmap_verify): Likewise.
27002 (df_scan_verify): Likewise.
27003 * lra-constraints.c (lra_constraints): Likewise.
27004 (undo_optional_reloads): Likewise.
27005 (lra_undo_inheritance): Likewise.
27006 * lra-remat.c (calculate_gen_cands): Likewise.
27007 (do_remat): Likewise.
27008 * lra-spills.c (assign_spill_hard_regs): Likewise.
27009 (spill_pseudos): Likewise.
27010 * tree-ssa-pre.c (bitmap_set_and): Likewise.
27011 (bitmap_set_subtract_values): Likewise.
27013 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27015 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
27016 management with auto_bitmap.
27017 (fix_inter_tick): Likewise.
27018 (fix_recovery_deps): Likewise.
27019 * ira.c (add_store_equivs): Likewise.
27020 (find_moveable_pseudos): Likewise.
27021 (split_live_ranges_for_shrink_wrap): Likewise.
27022 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
27023 (rtx_reuse_manager::seen_def_p): Likewise.
27024 (rtx_reuse_manager::set_seen_def): Likewise.
27025 * print-rtl.h (class rtx_reuse_manager): Likewise.
27027 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27029 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
27031 (migrate_btr_def): Likewise.
27032 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
27033 * df-core.c (loop_post_order_compute): Likewise.
27034 (loop_inverted_post_order_compute): Likewise.
27035 * hsa-common.h: Likewise.
27036 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
27037 * init-regs.c (initialize_uninitialized_regs): Likewise.
27038 * ipa-inline.c (resolve_noninline_speculation): Likewise.
27039 (inline_small_functions): Likewise.
27040 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
27041 * ira.c (combine_and_move_insns): Likewise.
27042 (build_insn_chain): Likewise.
27043 * loop-invariant.c (find_invariants): Likewise.
27044 * lower-subreg.c (propagate_pseudo_copies): Likewise.
27045 * predict.c (tree_predict_by_opcode): Likewise.
27046 (predict_paths_leading_to): Likewise.
27047 (predict_paths_leading_to_edge): Likewise.
27048 (estimate_loops_at_level): Likewise.
27049 (estimate_loops): Likewise.
27050 * shrink-wrap.c (try_shrink_wrapping): Likewise.
27051 (spread_components): Likewise.
27052 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
27053 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
27054 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
27055 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
27056 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
27057 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
27058 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
27059 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
27060 (mark_threaded_blocks): Likewise.
27061 (thread_through_all_blocks): Likewise.
27062 * tree-ssa.c (verify_ssa): Likewise.
27063 (execute_update_addresses_taken): Likewise.
27064 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
27066 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27068 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
27070 (post_order_compute): Likewise.
27071 (inverted_post_order_compute): Likewise.
27072 (pre_and_rev_post_order_compute_fn): Likewise.
27074 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27076 * genrecog.c (int_set::int_set): Explicitly construct our
27077 auto_vec base class.
27078 * vec.h (auto_vec::auto_vec): New constructor.
27080 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27082 * bitmap.h (class auto_bitmap): New constructor taking
27083 bitmap_obstack * argument.
27085 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27087 * bitmap.h (class auto_bitmap): Change type of m_bits to
27088 bitmap_head, and adjust ctor / dtor and member operators.
27090 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
27092 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
27093 when returned register mode doesn't match original mode.
27095 2017-05-12 Jeff Law <law@redhat.com>
27096 Jakub Jelinek <jakub@redhat.com>
27098 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
27099 we look for cc setter after the compare-elim changes.
27100 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
27101 within the vector to match what compare-elim now expects.
27102 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
27103 (xorsi3_flags, one_cmplsi2_flags): Likewise.
27105 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
27106 after the compare-elim changes.
27107 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
27108 the vector to match what compare-elim now expects.
27109 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
27110 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
27111 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
27112 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
27113 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
27115 * config/visium/visium.c (single_set_and_flags): Fix where
27116 we look for cc setter after the compare-elim changes.
27117 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
27118 with the vector to match what compare-elim now expects.
27119 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
27120 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
27121 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
27122 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
27123 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
27124 (neg<mode>2_insn_set_overflow): Likewise.
27126 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
27128 PR middle-end/79794
27129 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
27130 maybe_expand_insn call, set ops[0].target. If still set after call,
27131 set alt_rtl. Add extra arg to recursive calls.
27132 (extract_bit_field): Add alt_rtl argument. Pass to
27134 * expmed.h (extract_bit_field): Fix prototype.
27135 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
27136 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
27137 to extract_bit_field_calls.
27138 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
27139 Pass alt_rtl to extract_bit_field calls.
27140 * calls.c (store_unaligned_arguments_into_psuedos)
27141 load_register_parameters): Pass extra NULL to extract_bit_field calls.
27142 * optabs.c (maybe_legitimize_operand): Clear op->target when call
27144 * optabs.h (struct expand_operand): Add target bitfield.
27146 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
27148 * compare-elim.c (try_eliminate_compare): Canonicalize
27149 operation with embedded compare to
27150 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
27151 (set (reg) (operation)].
27153 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
27155 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
27158 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
27159 cost of adding a carry flag for ADC instruction.
27160 [case MINUS]: Ignore the cost of subtracting a carry flag
27161 for SBB instruction.
27163 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
27165 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
27167 * config/rs6000/bmiintrin.h: New file.
27168 * config/rs6000/bmi2intrin.h: New file.
27169 * config/rs6000/x86intrin.h: New file.
27171 2017-05-12 Jeff Law <law@redhat.com>
27173 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
27176 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
27178 PR middle-end/80707
27179 * tree-cfg.c: Remove cfg edges of unreachable case statements.
27181 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27183 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27184 early expansion of vector divide builtins.
27185 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
27186 builtins identified as having unsigned arguments.
27188 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27190 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
27191 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
27192 expansion of vector logical operations (and, andc, or, xor,
27195 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27197 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
27198 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
27200 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27202 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27203 early GIMPLE expansion of vector multiplies.
27205 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27207 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
27208 TARGET_HAVE_MOVT conditional.
27209 (movt splitter): Likewise.
27211 2017-05-12 Richard Biener <rguenther@suse.de>
27213 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
27214 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27215 Fold all stmts not inplace.
27217 2017-05-12 Richard Biener <rguenther@suse.de>
27219 PR tree-optimization/80713
27220 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
27221 inserted_exprs bit for not removed stmts.
27223 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
27225 PR middle-end/69921
27226 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
27227 parallelized" attribute for parallelized OpenACC kernels.
27228 * omp-offload.c (execute_oacc_device_lower): Use it.
27230 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
27231 Set "oacc kernels" attribute.
27232 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
27233 parameter. Adjust all users.
27234 (oacc_fn_attrib_kernels_p): Remove function.
27235 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
27236 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
27237 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
27238 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
27239 assert "oacc kernels" attribute is set.
27241 2017-05-11 Carl Love <cel@us.ibm.com>
27243 * config/rs6000/rs6000-c: Add support for built-in functions
27244 vector unsigned char vec_popcnt (vector signed char)
27245 vector unsigned char vec_popcnt (vector unsigned char)
27246 vector unsigned short vec_popcnt (vector signed short)
27247 vector unsigned short vec_popcnt (vector unsigned short)
27248 vector unsigned int vec_popcnt (vector signed int)
27249 vector unsigned int vec_popcnt (vector unsigned int)
27250 vector unsigned long long vec_popcnt (vector signed long long)
27251 vector unsigned long long vec_popcnt (vector unsigned long long)
27252 vector signed long long vec_slo (vector signed long long,
27253 vector signed char)
27254 vector signed long long vec_slo (vector signed long long,
27255 vector unsigned char)
27256 vector unsigned long long vec_slo (vector unsigned long long,
27257 vector signed char)
27258 vector unsigned long long vec_slo (vector unsigned long long,
27259 vector unsigned char)
27260 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
27261 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
27262 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
27263 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
27264 * doc/extend.texi: Update the built-in documentation file for the
27265 new built-in functions.
27267 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
27269 * attribs.h (sorted_attr_string): Move machine independent
27270 functions for target clone support from the i386 port to common
27271 code. Rename ix86_function_versions to common_function_versions.
27272 Rename make_name to make_unique_name.
27273 (common_function_versions): Likewise.
27274 (make_unique_name): Likewise.
27275 (make_dispatcher_decl): Likewise.
27276 (is_function_default_version): Likewise.
27277 * attribs.c (attr_strcmp): Likewise.
27278 (sorted_attr_string): Likewise.
27279 (common_function_versions): Likewise.
27280 (make_unique_name): Likewise.
27281 (make_dispatcher_decl): Likewise.
27282 (is_function_default_version): Likewise.
27283 * config/i386/i386.c (attr_strcmp): Likewise.
27284 (sorted_attr_string): Likewise.
27285 (ix86_function_versions): Likewise.
27286 (make_name): Likewise.
27287 (make_dispatcher_decl): Likewise.
27288 (is_function_default_version): Likewise.
27289 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
27291 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27294 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
27295 Account for direct move costs for vec_construct of integer
27298 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
27301 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
27302 (UNSPEC_STX_ATOMIC): Ditto.
27303 (loaddi_via_sse): New insn.
27304 (storedi_via_sse): Ditto.
27305 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
27306 Update corresponding peephole2 patterns.
27307 (atomic_storedi_fpu): Ditto.
27309 2017-05-11 Julia Koval <julia.koval@intel.com>
27311 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
27312 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
27314 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
27315 (__builtin_ia32_rsqrt14ss_mask): New builtins.
27316 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
27318 2017-05-11 Nathan Sidwell <nathan@acm.org>
27320 * graphite-poly.c: Include dumpfile.h.
27322 * dumpfle.h (dump_function): Declare here ...
27323 * tree-dump.h (dump_function): ... not here.
27324 * dumpfile.c: #include tree-cfg.h.
27325 (dump_function): Move here from ...
27326 * tree-dump.c (dump_function): ... here.
27327 * gimplify.c: #include splay-tree.h, not tree-dump.h.
27328 * graphite-poly.c: Don't include tree-dump.h.
27329 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
27330 * print-tree.c: Likewise.
27331 * stor-layout.c: Likewise.
27332 * tree-nested.c: Likewise.
27334 * dumpfile.c (dump_start): Use TDF_FLAGS.
27335 (dump_enable_all): Fix TDF_KIND check thinko.
27337 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
27339 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27340 array entries to represent two legal parameterizations of the
27341 overloaded __builtin_cmpb function, as represented by the
27342 P6_OV_BUILTIN_CMPB constant.
27343 (altivec_resolve_overloaded_builtin): Add special case handling
27344 for the __builtin_cmpb function, as represented by the
27345 P6_OV_BUILTIN_CMPB constant.
27346 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
27347 (BU_P6_64BIT_2): New macro.
27348 (BU_P6_OVERLOAD_2): New macro
27349 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
27350 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
27351 (CMPB): Add overload support to represent both 32-bit and 64-bit
27352 compare-bytes function.
27353 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
27354 support for TARGET_CMPB.
27355 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
27356 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27357 documentation of the __builtin_cmpb overloaded built-in function.
27359 2017-05-11 Richard Biener <rguenther@suse.de>
27361 PR tree-optimization/80705
27362 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
27363 bases are not vectorizable.
27365 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27367 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
27368 when counting register pressure.
27370 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27372 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
27373 (iv_ca_more_deps): Renamed to ...
27374 (iv_ca_compare_deps): ... this.
27375 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
27377 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27379 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
27381 (determine_group_iv_costs): ... here.
27382 (find_inv_vars_cb): Record inv var if it's not recorded before.
27384 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27386 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
27387 (get_shiftadd_cost): Ditto.
27389 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27391 * tree-ssa-address.c: Include header file.
27392 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
27394 (add_to_parts): Refactor.
27395 (addr_to_parts): New parameter. Update use of move_hint_to_base.
27396 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
27399 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27401 PR tree-optimization/53090
27402 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
27404 (extract_cond_operands): Detect condition with IV on both sides
27405 and return COMP_IV_EXPR_2.
27406 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
27407 (rewrite_use_compare): Simplify by removing call to function
27408 extract_cond_operands.
27410 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27412 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
27413 (extract_cond_operands): Detect condition comparing against non-
27414 invariant bound and return appropriate enum value.
27415 (find_interesting_uses_cond): Update use of extract_cond_operands.
27416 Handle its return value accordingly.
27417 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
27419 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27421 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
27422 nonlinear iv_use computation in loop invariant sensitive way.
27424 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27426 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
27427 (find_iv_candidates): Call relate_compare_use_with_all_cands.
27429 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27431 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
27432 (dump_cand): Support iv_cand.inv_exprs.
27433 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
27435 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
27438 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27440 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
27442 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
27443 as local function. Include necessary header files.
27444 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
27446 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27448 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
27450 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27452 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
27453 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
27454 RSHIFT_EXPR and BIT_NOT_EXPR.
27456 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27458 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
27459 (adjust_setup_cost): New parameter supporting round up adjustment.
27460 (struct address_cost_data): Delete.
27461 (force_expr_to_var_cost): Don't bound cost with spill_cost.
27462 (split_address_cost, ptr_difference_cost): Delete.
27463 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
27464 (struct ainc_cost_data): New struct.
27465 (get_address_cost_ainc): New function.
27466 (get_address_cost, get_computation_cost): Reimplement.
27467 (determine_group_iv_cost_address): Record inv_expr for all uses of
27469 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
27470 (iv_ca_has_deps): Reimplemented to ...
27471 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
27473 (iv_ca_extend): Call iv_ca_more_deps.
27475 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27477 * tree-ssa-address.c (struct mem_address): Move to header file.
27478 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
27479 * tree-ssa-address.h (struct mem_address): Move from C file.
27480 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
27482 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27484 * tree-affine.h (aff_combination_type): New interface.
27485 (aff_combination_zero_p): Remove static.
27486 (aff_combination_const_p): New interface.
27487 (aff_combination_singleton_var_p): New interfaces.
27489 2017-05-11 Richard Biener <rguenther@suse.de>
27491 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27492 Skip unreachable blocks and destinations.
27493 (eliminate): Move stmt removal and fixup ...
27494 (fini_eliminate): ... here. Skip inserted exprs.
27495 (pass_pre::execute): Move fini_pre after fini_eliminate.
27496 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
27497 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
27498 PRE to get rid of dead code that has invalid SSA form and
27499 split critical edges again.
27501 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27503 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27505 2017-05-11 Richard Biener <rguenther@suse.de>
27507 * passes.c (execute_function_todo): Verify loops if they are
27508 said to be up-to-date.
27509 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
27510 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
27512 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
27515 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
27516 handle calling assemble_external ourself.
27519 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
27520 modes with zero size. Enhance comment.
27522 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27524 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
27525 built-ins for vec_xl and vec_xst with short and char pointer
27528 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27530 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
27531 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
27532 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
27533 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
27534 (_mm_maskz_min_round_ss): New intrinsics.
27535 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
27536 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
27537 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
27538 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
27539 (__builtin_ia32_minss_mask_round): New builtins.
27540 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27541 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27542 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
27544 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
27545 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
27547 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
27550 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27552 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
27553 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
27554 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
27555 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
27556 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
27557 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
27558 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
27559 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27560 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27561 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
27562 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
27563 (__builtin_ia32_mulss_mask_round): New builtins.
27564 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27565 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27566 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
27568 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
27569 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27571 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27574 2017-05-10 Julia Koval <julia.koval@intel.com>
27576 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
27577 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
27578 (_mm256_setr_m128i): New intrinsics.
27580 2017-05-10 Julia Koval <julia.koval@intel.com>
27582 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
27583 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
27584 (_mm_maskz_rcp14_ss): New intrinsics.
27585 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
27586 (__builtin_ia32_rcp14ss_mask): New builtins.
27587 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
27589 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
27591 PR tree-optimization/51513
27592 * tree-cfg.c (gimple_seq_unreachable_p): New function.
27593 (assert_unreachable_fallthru_edge_p): Use it.
27594 (group_case_labels_stmt): Likewise.
27595 * tree-cfg.h: Prototype it.
27596 * stmt.c: Include cfghooks.h and tree-cfg.h.
27597 (emit_case_dispatch_table) <gap_label>: New local variable.
27598 Use it to fill dispatch table gaps.
27599 Test for default_label before updating probabilities.
27600 (expand_case) <default_label>: Remove unneeded initialization.
27601 Test for unreachable default case statement and remove its edge.
27602 Set default_label accordingly.
27603 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
27605 2017-05-10 Carl Love <cel@us.ibm.com>
27607 * config/rs6000/rs6000-c: Add support for built-in functions
27608 vector signed char vec_neg (vector signed char)
27609 vector signed short int vec_neg (vector short int)
27610 vector signed int vec_neg (vector signed int)
27611 vector signed long long vec_neg (vector signed long long)
27612 vector float vec_neg (vector float)
27613 vector double vec_neg (vector double)
27614 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
27616 * config/rs6000/altivec.h: Add define for vec_neg
27617 * doc/extend.texi: Update the built-in documentation for the
27618 new built-in functions.
27620 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27622 PR tree-optimization/77644
27623 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
27625 2017-05-10 Nathan Sidwell <nathan@acm.org>
27627 * dumpfile.h (TDI_lang_all): New.
27628 (TDF_KIND): New. Renumber others
27629 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
27631 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
27633 (get_dump_file_name): Adjust suffix generation.
27634 (dump_enable_all): Use TDF_KIND.
27635 * doc/invoke.texi (-fdump-lang-all): Document.
27637 * dumpfile.h: Tabify.
27639 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
27642 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
27643 Move member access before delete.
27645 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
27647 * tree-inline.c (expand_call_inline): Split block at stmt
27650 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
27653 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
27654 are now unused after splitting mov{sf,sd}_hardfloat.
27655 (f32_lr2): Likewise.
27656 (f32_lm): Likewise.
27657 (f32_lm2): Likewise.
27658 (f32_li): Likewise.
27659 (f32_li2): Likewise.
27660 (f32_lv): Likewise.
27661 (f32_sr): Likewise.
27662 (f32_sr2): Likewise.
27663 (f32_sm): Likewise.
27664 (f32_sm2): Likewise.
27665 (f32_si): Likewise.
27666 (f32_si2): Likewise.
27667 (f32_sv): Likewise.
27668 (f32_dm): Likewise.
27669 (f32_vsx): Likewise.
27670 (f32_av): Likewise.
27671 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
27672 For movsf, order stores so the VSX stores occur before the GPR
27673 store which encourages the register allocator to use a traditional
27674 FPR instead of a GPR. For movsd, order the stores so that the GPR
27675 store comes before the VSX stores to allow the power6 to work.
27676 This is due to the power6 not having a 32-bit integer store
27677 instruction from a FPR.
27678 (movsf_hardfloat): Likewise.
27679 (movsd_hardfloat): Likewise.
27681 2017-05-09 Martin Sebor <msebor@redhat.com>
27683 PR translation/80280
27684 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
27687 PR translation/80280
27688 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
27689 data member added in r247778.
27690 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
27692 2017-05-09 Nathan Sidwell <nathan@acm.org>
27694 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
27696 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
27699 2017-05-09 Marek Polacek <polacek@redhat.com>
27701 * doc/invoke.texi: Fix typo.
27703 2017-05-09 Richard Biener <rguenther@suse.de>
27705 * tree-vrp.c (vrp_val_is_max): Adjust comment.
27706 (vrp_val_is_min): Likewise.
27707 (set_value_range_to_value): Likewise.
27708 (set_value_range_to_nonnegative): Likewise.
27709 (gimple_assign_nonzero_p): Likewise.
27710 (gimple_stmt_nonzero_p): Likewise.
27711 (vrp_int_const_binop): Likewise. Remove unreachable case.
27712 (adjust_range_with_scev): Adjust comments.
27713 (compare_range_with_value): Likewise.
27714 (extract_range_from_phi_node): Likewise.
27715 (test_for_singularity): Likewise.
27717 2017-05-09 Richard Biener <rguenther@suse.de>
27719 * tree-vrp.c (get_single_symbol): Add assert that we don't
27720 get overflowed constants as invariant part.
27721 (compare_values_warnv): Add comment before the TREE_NO_WARNING
27722 checks. Use wi::cmp instead of recursing for integer constants.
27723 (compare_values): Just ignore whether we assumed undefined
27724 overflow instead of failing the compare.
27725 (extract_range_for_var_from_comparison_expr): Add comment before the
27726 TREE_NO_WARNING sets.
27727 (test_for_singularity): Likewise.
27728 (extract_range_from_comparison): Do not disable optimization
27729 when we assumed undefined overflow.
27730 (extract_range_basic): Remove init of unused var.
27732 2017-05-09 Richard Biener <rguenther@suse.de>
27734 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
27735 (extract_range_from_multiplicative_op_1): Adjust.
27736 (extract_range_from_binary_expr_1): Use int_const_binop.
27738 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
27741 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
27742 rs6000_store_data_bypass_p in seven define_bypass directives and
27743 in several comments.
27744 * config/rs6000/rs6000-protos.h: Add prototype for
27745 rs6000_store_data_bypass_p function.
27746 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
27747 function implements slightly different (rs6000-specific) semantics
27748 than store_data_bypass_p, returning false rather than aborting
27749 with assertion error when arguments do not satisfy the
27750 requirements of store data bypass.
27751 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
27752 rs6000_store_data_bypass_p.
27754 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
27756 * config/xtensa/xtensa-protos.h
27757 (xtensa_initial_elimination_offset): New declaration.
27758 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
27759 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
27760 macro definition, add case for FRAME_POINTER_REGNUM when
27761 FRAME_GROWS_DOWNWARD.
27762 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
27763 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
27764 xtensa_initial_elimination_offset.
27766 2017-05-08 Nathan Sidwell <nathan@acm.org>
27768 * doc/invoke.texi: Alphabetize -fdump options.
27770 2017-05-08 Martin Sebor <msebor@redhat.com>
27772 PR translation/80280
27773 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
27775 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
27777 * target.def (compute_frame_layout): New optional target hook.
27778 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
27779 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
27780 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
27782 * reload1.c (verify_initial_elim_offsets): Likewise.
27783 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
27784 (use_simple_return_p): Call arm_compute_frame_layout if needed.
27785 (arm_get_frame_offsets): Split up into this ...
27786 (arm_compute_frame_layout): ... and this function.
27788 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
27790 * config/aarch64/constraints.md (Usa): New constraint.
27791 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
27793 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
27795 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
27796 with_multilib_list after it has been checked.
27798 2017-05-08 Richard Biener <rguenther@suse.de>
27800 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
27801 (bitmap_set_subtract_values): Likewise.
27803 2017-05-08 Richard Biener <rguenther@suse.de>
27805 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
27806 (gimple_assign_nonzero): ... this and remove strict_overflow_p
27808 (gimple_stmt_nonzero_warnv_p): Rename to ...
27809 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
27811 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
27812 (extract_range_basic): Adjust, do not disable propagation on
27813 strict overflow sensitive simplification.
27814 (vrp_visit_cond_stmt): Likewise.
27816 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
27818 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
27819 body size unconditionally.
27821 2017-05-07 Jeff Law <law@redhat.com>
27824 2017-05-06 Jeff Law <law@redhat.com>
27825 PR tree-optimization/78496
27826 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27829 PR tree-optimization/78496
27830 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27831 (simplify_stmt_using_ranges): Call it.
27832 (vrp_dom_walker::before_dom_children): Extract equivalences
27833 from an ASSERT_EXPR with an equality comparison against a
27836 2017-05-06 Jeff Law <law@redhat.com>
27838 PR tree-optimization/78496
27839 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27842 PR tree-optimization/78496
27843 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27844 (simplify_stmt_using_ranges): Call it.
27845 (vrp_dom_walker::before_dom_children): Extract equivalences
27846 from an ASSERT_EXPR with an equality comparison against a
27849 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27851 * lra-constraints.c (lra_copy_reg_equiv): New function.
27852 (split_reg): Use it to copy equivalence information from the
27853 original register to the spill register.
27855 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27857 PR rtl-optimization/75964
27858 * simplify-rtx.c (simplify_const_relational_operation): Remove
27859 invalid handling of comparisons of integer ABS.
27861 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
27863 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
27864 initialize to zero.
27865 (init_regs): Remove declaration.
27866 (function_arg_advance_32): Initialize error_p as boolean variable.
27868 2017-05-05 Nathan Sidwell <nathan@acm.org>
27870 * store-motion.c (remove_reachable_equiv_notes): Reformat long
27871 lines. Use for (;;).
27873 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27875 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
27876 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
27877 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
27878 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
27879 VF=2 that require versioning.
27881 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27883 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
27886 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27888 * diagnostic.h (diagnostic_override_option_index): Convert from
27889 macro to inline function.
27891 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27893 * diagnostic.c (last_module_changed_p): New function.
27894 (set_last_module): New function.
27895 (diagnostic_report_current_module): Convert macro usage to
27896 the above functions.
27897 * diagnostic.h (diagnostic_context::last_module): Strengthen
27898 from const line_map * to const line_map_ordinary *.
27899 (diagnostic_last_module_changed): Delete macro.
27900 (diagnostic_set_last_module): Delete macro.
27902 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27904 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
27905 with diagnostic_report_diagnostic.
27906 (diagnostic_n_impl_richloc): Likewise.
27907 * diagnostic.h (report_diagnostic): Delete macro.
27908 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
27909 with diagnostic_report_diagnostic.
27910 * substring-locations.c (format_warning_va): Likewise.
27912 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27914 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
27915 save/restor of format_spec. Move option-printing code to...
27916 (print_option_information): ...this new function, and
27917 reimplement by simply printing to the pretty_printer,
27918 rather than appending to the format string.
27920 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27922 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
27923 handling logic into...
27924 (update_effective_level_from_pragmas): ...this new function.
27926 2017-05-04 Andrew Waterman <andrew@sifive.com>
27928 * config/riscv/riscv.opt (mstrict-align): New option.
27929 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
27930 (SLOW_UNALIGNED_ACCESS): Define.
27931 (riscv_slow_unaligned_access): Declare.
27932 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
27934 (riscv_slow_unaligned_access): New variable.
27935 (rocket_tune_info): Set slow_unaligned_access to true.
27936 (optimize_size_tune_info): Set slow_unaligned_access to false.
27937 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
27938 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
27939 (riscv_option_override): Set riscv_slow_unaligned_access.
27940 * doc/invoke.texi: Add -mstrict-align to RISC-V.
27942 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
27944 * config/riscv/riscv.md: Unify indentation.
27946 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27951 * config/rs6000/rs6000.md (u code attribute): Add FIX and
27953 (extendsi<mode>2): Add support for doing sign extension via
27954 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
27955 don't have ISA 3.0 instructions.
27956 (extendsi<mode>2 splitter): Likewise.
27957 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
27958 generate the normal insns since SImode can now go in vector
27959 registers. Disallow the special UNSPECs needed for previous
27960 machines to hide SImode being used. Add new insns
27961 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
27962 (fix_trunc<mode>si2_stfiwx): Likewise.
27963 (fix_trunc<mode>si2_internal): Likewise.
27964 (fixuns_trunc<mode>si2): Likewise.
27965 (fixuns_trunc<mode>si2_stfiwx): Likewise.
27966 (fctiw<u>z_<mode>_smallint): Likewise.
27967 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
27968 of floating point to 32-bit integer from doing a direct move to
27969 the GPR registers to do a store.
27970 (fctiwz_<mode>): Break long line.
27972 2017-05-05 Bin Cheng <bin.cheng@arm.com>
27974 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
27975 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
27976 (addr_list, addr_offset_valid_p): New.
27977 (split_address_groups): Check offset validity with above function.
27978 (gt-tree-ssa-loop-ivopts.h): Include header file.
27980 2017-05-05 Nathan Sidwell <nathan@acm.org>
27982 * config.gcc (arm*-*-*): Add missing 'fi'.
27984 2017-05-05 Steve Ellcey <sellcey@cavium.com>
27986 * doc/invoke.texi (-fopt-info): Explicitly say order of options
27987 included in -fopt-info does not matter.
27988 * doc/optinfo.texi (-fopt-info): Fix description of default
27989 behavour. Explicitly say order of options included in -fopt-info
27992 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27994 * config.gcc: Allow combinations of aprofile and rmprofile values for
27995 --with-multilib-list.
27996 * config/arm/t-multilib: New file.
27997 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
27998 variables. Remove setting of ISA and floating-point ABI in
27999 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
28000 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
28001 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
28002 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
28004 * config/arm/t-rmprofile: Likewise except for the matches changes.
28005 * doc/install.texi (--with-multilib-list): Document the combination of
28006 aprofile and rmprofile values and warn about pitfalls in doing that.
28008 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28010 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
28011 (movdi_aarch64): Likewise.
28013 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28015 PR tree-optimization/80632
28016 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
28018 (build_arrays): Initialize it for virtual phis.
28019 (fix_phi_nodes): Use it for virtual phis.
28021 PR tree-optimization/80558
28022 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
28023 [x, y] op z into [x op, y op z] for op & or | if conditions
28026 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
28027 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28030 * config/arm/arm.md (use_literal_pool): Remove.
28031 (64-bit immediate split): No longer takes cost into consideration
28032 if arm_disable_literal_pool is enabled.
28033 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
28034 used when arm_disable_literal_pool is enabled.
28035 (arm_max_const_double_inline_cost): Remove use of
28036 arm_disable_literal_pool.
28037 (push_minipool_fix): Add assert.
28038 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
28039 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
28040 (no_literal_pool_sf_immediate): New.
28042 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28044 PR tree-optimization/80613
28045 * tree-ssa-dce.c (propagate_necessity): Remove cases for
28046 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
28048 2017-05-05 Richard Biener <rguenther@suse.de>
28050 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
28052 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
28054 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
28055 of this flag from insn conditions due to removal from r247495.
28057 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28059 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
28061 (arm_early_store_addr_dep_ptr): Likewise.
28062 * config/arm/aarch-common-protos.h
28063 (arm_early_load_addr_dep_ptr): Add prototype.
28064 (arm_early_store_addr_dep_ptr): Likewise.
28065 * config/arm/cortex-a53.md: Add new bypasses.
28067 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28069 * tree.c (next_type_uid): Change type to unsigned.
28070 (type_hash_canon): Decrement back next_type_uid if
28071 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
28072 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
28075 2017-05-04 Martin Sebor <msebor@redhat.com>
28077 * builtins.c: Fix a trivial typo in a comment.
28079 PR middle-end/79234
28080 * builtins.c (check_sizes): Adjust to handle reading past the end.
28081 Avoid printing excessive upper bound of ranges. Use %E to print
28082 tree nodes instead of converting them to %wu.
28083 (expand_builtin_memchr): New function.
28084 (compute_dest_size): Rename...
28085 (compute_objsize): ...to this.
28086 (expand_builtin_memcpy): Adjust.
28087 (expand_builtin_mempcpy): Adjust.
28088 (expand_builtin_strcat): Adjust.
28089 (expand_builtin_strcpy): Adjust.
28090 (check_strncat_sizes): Adjust.
28091 (expand_builtin_strncat): Adjust.
28092 (expand_builtin_strncpy): Adjust and simplify.
28093 (expand_builtin_memset): Adjust.
28094 (expand_builtin_bzero): Adjust.
28095 (expand_builtin_memcmp): Adjust.
28096 (expand_builtin): Handle memcmp.
28097 (maybe_emit_chk_warning): Check strncat just once.
28099 2017-05-04 Martin Sebor <msebor@redhat.com>
28101 PR preprocessor/79214
28102 PR middle-end/79222
28103 PR middle-end/79223
28104 * builtins.c (check_sizes): Add inlining context and issue
28105 warnings even when -Wno-system-headers is set.
28106 (check_strncat_sizes): Same.
28107 (expand_builtin_strncat): Same.
28108 (expand_builtin_memmove): New function.
28109 (expand_builtin_stpncpy): Same.
28110 (expand_builtin): Handle memmove and stpncpy.
28112 2017-05-04 Bin Cheng <bin.cheng@arm.com>
28114 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
28115 which is not used any more.
28117 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28119 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
28121 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28123 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
28124 (cortexa53_tunings): Likewise.
28125 (cortexa57_tunings): Likewise.
28126 (cortexa72_tunings): Likewise.
28127 (cortexa73_tunings): Likewise.
28129 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28131 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
28132 Set loop alignment to 8.
28134 2017-05-04 Martin Sebor <msebor@redhat.com>
28136 PR translation/80280
28137 * builtins.c (expand_builtin_object_size): Add missing quoting to
28138 %D and like directives.
28139 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
28140 (hsa_type_for_tree_type): Same.
28141 (verify_function_arguments): Same.
28142 * symtab.c (symbol_table::change_decl_assembler_name): Same.
28143 * varasm.c (get_section): Same.
28146 2017-05-04 Martin Sebor <msebor@redhat.com>
28148 PR translation/80280
28149 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
28151 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28153 * config/aarch64/aarch64.c (generic_addrcost_table):
28154 Change HI/TI mode setting.
28156 2017-05-04 Martin Jambor <mjambor@suse.cz>
28158 PR tree-optimization/80622
28159 * tree-sra.c (comes_initialized_p): New function.
28160 (build_accesses_from_assign): Only set write lazily when
28161 comes_initialized_p is false.
28162 (analyze_access_subtree): Use comes_initialized_p.
28163 (propagate_subaccesses_across_link): Assert !comes_initialized_p
28164 instead of testing for PARM_DECL.
28166 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28168 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
28169 constraint on operand 0 to allow more general addressing modes.
28170 Adjust output template.
28171 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
28173 * config/aarch64/aarch64-protos.h
28174 (aarch64_address_valid_for_prefetch_p): Declare prototype.
28175 * config/aarch64/constraints.md (Dp): New address constraint.
28176 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
28179 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
28181 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
28182 update use of estimate_ipcp_clone_size_and_time.
28183 (estimate_local_effects): Update use of
28184 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
28185 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
28186 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
28187 Return nonspecialized time.
28189 2017-05-04 Richard Biener <rguenther@suse.de>
28191 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
28192 for the last VUSE which def dominates the PHI. Directly call
28194 (get_continuation_for_phi_1): Remove.
28196 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
28198 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
28199 to explain the use of truncating division. Cap the number of
28200 iterations to the maximum given by nb_iterations_upper_bound,
28203 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28205 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
28206 * configure: Regenerate.
28207 * config.in: Regenerate.
28208 * config/i386/driver-mingw32.c: new file.
28209 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
28210 * config.host: Link driver-mingw32.o on MinGW host.
28211 * doc/install.texi: Document new --enable-mingw-wildcard configure
28214 2017-05-04 Marek Polacek <polacek@redhat.com>
28216 PR tree-optimization/80612
28217 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
28219 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28220 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
28222 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
28223 (movt splitter): Likewise.
28224 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
28225 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
28226 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
28227 block for Thumb-1 with MOVT.
28228 (thumb2_legitimate_address_p): Move code block ...
28229 (can_avoid_literal_pool_for_label_p): ... into this new function.
28230 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
28232 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
28233 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
28234 "M-profile targets with the MOVT instruction".
28236 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28238 * config/arm/arm-builtins.c (arm_init_builtins): Rename
28239 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
28240 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
28242 2017-05-04 Martin Liska <mliska@suse.cz>
28244 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
28245 variable cond_code.
28247 2017-05-04 Richard Biener <rguenther@suse.de>
28249 * tree.c (array_at_struct_end_p): Handle arrays at struct
28250 end with flexarrays more conservatively. Refactor and treat
28251 arrays of arrays or aggregates more strict. Fix
28252 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
28253 * tree.c (array_at_struct_end_p): Adjust prototype.
28254 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
28255 * gimple-fold.c (get_range_strlen): Likewise.
28256 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
28258 2017-05-04 Richard Biener <rguenther@suse.de>
28260 PR tree-optimization/31130
28261 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
28263 (supports_overflow_infinity): Likewise.
28264 (is_negative_overflow_infinity): Likewise.
28265 (is_positive_overflow_infinity): Likewise.
28266 (is_overflow_infinity): Likewise.
28267 (stmt_overflow_infinity): Likewise.
28268 (overflow_infinity_range_p): Likewise.
28269 (usable_range_p): Remove as always returning true.
28270 (make_overflow_infinity): Remove.
28271 (negative_overflow_infinity): Likewise.
28272 (positive_overflow_infinity): Likewise.
28273 (avoid_overflow_infinity): Likewise.
28274 (set_value_range): Adjust accordingly.
28275 (set_value_range_to_nonnegative): Likewise, remove now unused
28276 overflow_infinity arg.
28277 (vrp_operand_equal_p): Adjust.
28278 (update_value_range): Likewise.
28279 (range_int_cst_singleton_p): Likewise.
28280 (operand_less_p): Likewise.
28281 (compare_values_warnv): Likewise.
28282 (extract_range_for_var_from_comparison_expr): Likewise.
28283 (vrp_int_const_binop): Likewise.
28284 (zero_nonzero_bits_from_vr): Likewise.
28285 (extract_range_from_multiplicative_op_1): Likewise.
28286 (extract_range_from_binary_expr_1): Likewise.
28287 (extract_range_from_unary_expr): Likewise.
28288 (extract_range_from_comparison): Likewise.
28289 (extract_range_basic): Likewise.
28290 (adjust_range_with_scev): Likewise.
28291 (compare_ranges): Likewise.
28292 (compare_range_with_value): Likewise.
28293 (dump_value_range): Likewise.
28294 (test_for_singularity): Likewise, remove strict_overflow_p parameter
28296 (simplify_cond_using_ranges): Adjust.
28298 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28300 * brig-builtins.def: Added a builtin for class_f64.
28301 * builtin-types.def: Added a builtin type needed by class_f64.
28303 2017-05-03 Jason Merrill <jason@redhat.com>
28305 * timevar.def: Add TV_CONSTEXPR.
28307 2017-05-03 David Malcolm <dmalcolm@redhat.com>
28309 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
28311 2017-05-03 Martin Jambor <mjambor@suse.cz>
28313 * ipa-prop.c (ipa_update_after_lto_read): Removed.
28314 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
28315 * ipa-cp.c (ipcp_propagate_stage): Do not call
28316 ipa_update_after_lto_read.
28317 * ipa-inline.c (ipa_inline): Likewise.
28319 2017-05-03 Martin Jambor <mjambor@suse.cz>
28321 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
28322 tag. Added a default constructor and a destructor.
28323 (ipa_edge_args_sum_t): New class;
28324 (ipa_edge_args_sum): Declare.
28325 (ipa_edge_args_vector): Remove declaration.
28326 (IPA_EDGE_REF): Use ipa_edge_args_sum.
28327 (ipa_free_edge_args_substructures): Remove declaration.
28328 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
28329 (ipa_edge_args_info_available_for_edge_p): Likewise.
28330 * ipa-prop.c (ipa_edge_args_vector): Removed.
28331 (edge_removal_hook_holder): Likewise.
28332 (edge_duplication_hook_holder): Likewise.
28333 (ipa_edge_args_sum): New variable.
28334 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
28335 ipa_edge_args_vector.
28336 (ipa_free_edge_args_substructures): Likewise.
28337 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
28338 ipa_edge_args_vector.
28339 (ipa_edge_removal_hook): Turned into method
28340 ipa_edge_args_sum_t::remove.
28341 (ipa_edge_duplication_hook): Turned into method
28342 ipa_edge_args_sum_t::duplicate.
28343 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
28344 registering edge hooks.
28345 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
28346 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
28347 ipa_edge_args_sum instead of ipa_edge_args_vector.
28348 * ipa-profile.c (ipa_profile): Likewise.
28350 2017-05-03 Martin Jambor <mjambor@suse.cz>
28352 * symbol-summary.h (function_summary): New method exists.
28353 (function_summary::symtab_removal): Deallocate through release.
28354 (call_summary): New class.
28355 (gt_ggc_mx): New overload.
28356 (gt_pch_nx): Likewise.
28357 (gt_pch_nx): Likewise.
28359 2017-05-03 Jeff Law <law@redhat.com>
28361 PR tree-optimization/78496
28362 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
28363 from simplify_cond_using_ranges. Split off code to walk
28364 backwards through casts into ...
28365 (simplify_cond_using_ranges_2): New function.
28366 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
28367 (execute_vrp): After identifying jump threads, call
28368 simplify_cond_using_ranges_2.
28370 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
28373 * ipa-inline.h (inline_summary): Add ctor.
28374 (create_ggc): Do not use ggc_cleared_alloc.
28376 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
28377 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28379 * gcc.c (handle_braces): Support escaping in switch matching
28381 * doc/invoke.texi (Spec Files): Document it.
28382 Remove superfluous @code markup in items.
28384 2017-05-03 David Malcolm <dmalcolm@redhat.com>
28386 * diagnostic-show-locus.c (struct column_range): New struct.
28387 (get_affected_columns): New function.
28388 (get_printed_columns): New function.
28389 (struct correction): New struct.
28390 (correction::ensure_capacity): New function.
28391 (correction::ensure_terminated): New function.
28392 (struct line_corrections): New struct.
28393 (line_corrections::~line_corrections): New dtor.
28394 (line_corrections::add_hint): New function.
28395 (layout::print_trailing_fixits): Reimplement in terms of the new
28397 (selftest::test_overlapped_fixit_printing): New function.
28398 (selftest::diagnostic_show_locus_c_tests): Call it.
28400 2017-05-03 Nathan Sidwell <nathan@acm.org>
28402 Canonicalize canonical type hashing
28403 * tree.h (type_hash_canon_hash): Declare.
28404 * tree.c (type_hash_list, attribute_hash_list): Move into
28405 type_hash_canon_hash.
28406 (build_type_attribute_qual_variant): Break out hash code calc into
28407 type_hash_canon_hash.
28408 (type_hash_canon_hash): New. Generic type hash computation.
28409 (build_range_type_1, build_array_type_1, build_function_type,
28410 build_method_type_directly, build_offset_type, build_complex_type,
28411 make_vector_type): Call it.
28413 2017-05-03 Richard Biener <rguenther@suse.de>
28415 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
28416 When all DRs have unknown misaligned do not always peel
28417 when there is a store but apply the same costing model as if
28418 there were only loads.
28420 2017-05-03 Richard Biener <rguenther@suse.de>
28423 PR tree-optimization/80492
28424 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
28425 compare_base_decls returning dont-know properly.
28427 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
28429 * config/arm/iterators.md (CCSI): New mode iterator.
28430 (arch): New mode attribute.
28431 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
28432 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
28433 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
28434 code iterator for success result mode.
28435 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
28436 the corresponding new insn generators.
28438 2017-05-03 Bin Cheng <bin.cheng@arm.com>
28441 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28442 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28444 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
28446 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
28447 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
28448 (DDR_A): Wrap DDR argument in brackets.
28449 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
28450 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
28451 (DDR_REVERSED_P): Likewise.
28453 2017-05-03 Jakub Jelinek <jakub@redhat.com>
28455 PR tree-optimization/79472
28456 * tree-switch-conversion.c (struct switch_conv_info): Add
28457 contiguous_range and default_case_nonstandard fields.
28458 (collect_switch_conv_info): Compute contiguous_range and
28459 default_case_nonstandard fields, don't clear final_bb if
28460 contiguous_range and only the default case doesn't have the required
28462 (check_all_empty_except_final): Set default_case_nonstandard instead
28463 of failing if contiguous_range and the default case doesn't have empty
28465 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
28466 and only the default case doesn't have the required constants. Skip
28468 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
28469 if default_case_nonstandard.
28470 (build_constructors): Build constant 1 just once. Assert that default
28471 values aren't inserted in between cases if contiguous_range. Skip
28473 (build_arrays): Skip virtual phis.
28474 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
28475 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
28476 Handle virtual phis.
28477 (gen_inbound_check): Handle default_case_nonstandard case.
28478 (process_switch): Adjust check_final_bb caller. Call
28479 gather_default_values with the first non-default case instead of
28480 default case if default_case_nonstandard.
28482 2017-05-02 Nathan Sidwell <nathan@acm.org>
28484 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
28485 check. Fix formatting.
28487 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
28489 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
28490 errors when comparing specialized and unspecialized times.
28492 2017-05-02 David Malcolm <dmalcolm@redhat.com>
28494 * diagnostic-show-locus.c
28495 (layout::should_print_annotation_line_p): Make private.
28496 (layout::print_annotation_line): Make private.
28497 (layout::annotation_line_showed_range_p): Make private.
28498 (layout::show_ruler): Make private.
28499 (layout::print_source_line): Make private. Pass in line and
28500 line_width, rather than calling location_get_source_line. Drop
28502 (layout::print_leading_fixits): New method.
28503 (layout::print_any_fixits): Rename to...
28504 (layout::print_trailing_fixits): ...this, and make private.
28505 Don't print newline fixits.
28506 (diagnostic_show_locus): Move logic for printing one row into...
28507 (layout::print_line): ...this new function. Move the
28508 location_get_source_line call and error-handling from
28509 print_source_line to here. Call print_leading_fixits, and rename
28510 print_any_fixits to print_trailing_fixits.
28511 (selftest::test_fixit_insert_containing_newline): Update now that
28512 newlines are partially supported.
28513 (selftest::test_fixit_insert_containing_newline_2): New test.
28514 (selftest::test_fixit_replace_containing_newline): Update comments.
28515 (selftest::diagnostic_show_locus_c_tests): Call the new test.
28516 * edit-context.c (class added_line): New class.
28517 (class edited_line): Describe newline handling in comment.
28518 (edited_line::actually_edited_p): New method.
28519 (edited_line::print_content): Delete redundant decl.
28520 (edited_line::m_predecessors): New field.
28521 (edited_file::print_content): Call edited_line::print_content.
28522 (edited_file::print_diff): Update to support newlines.
28523 (edited_file::print_diff_hunk): Likewise.
28524 (edited_file::print_run_of_changed_lines): New function.
28525 (edited_file::print_diff_line): Convert to...
28526 (print_diff_line): ...this.
28527 (edited_file::get_effective_line_count): New function.
28528 (edited_line::edited_line): Initialize new field m_predecessors.
28529 (edited_line::~edited_line): Clean up m_predecessors.
28530 (edited_line::apply_fixit): Handle newlines.
28531 (edited_line::get_effective_line_count): New function.
28532 (edited_line::print_content): New function.
28533 (edited_line::print_diff_lines): New function.
28534 (selftest::test_applying_fixits_insert_containing_newline): New
28536 (selftest::test_applying_fixits_replace_containing_newline): New
28538 (selftest::insert_line): New function.
28539 (selftest::test_applying_fixits_multiple_lines): Add example of
28541 (selftest::edit_context_c_tests): Call the new tests.
28543 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28545 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
28546 parameter cand. Update dump information.
28547 (get_computation_cost): Update uses.
28549 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28551 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
28552 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
28553 (get_computation_at, rewrite_use_address): Update use of
28554 get_computation_aff.
28556 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28558 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
28559 (get_computation): Delete.
28560 (get_computation_cost): Implement like get_computation_cost_at.
28561 Use get_computation_at.
28562 (get_computation_cost_at): Delete.
28563 (rewrite_use_nonlinear_expr): Use get_computation_at.
28564 (rewrite_use_compare, remove_unused_ivs): Ditto.
28566 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28568 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
28570 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28572 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
28573 (ivopts_global_cost_for_size): Rename parameter and update uses.
28574 (iv_ca_recount_cost): Update uses.
28575 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
28576 candidates seperately in n_invs and n_cands.
28577 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
28579 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28581 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
28582 (find_inv_vars_cb): New.
28583 (find_depends): Renamed to ...
28584 (find_inv_vars): ... this.
28585 (add_candidate_1, force_var_cost): Call find_inv_vars.
28586 (split_address_cost, determine_group_iv_cost_cond): Ditto.
28588 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28590 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
28591 inv_vars. Add inv_exprs.
28592 (struct iv_cand): Rename depends_on to inv_vars.
28593 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
28594 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
28595 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
28596 (dump_cand): Dump inv_vars.
28597 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
28598 (record_invariant, find_depends, add_candidate_1): Ditto.
28599 (set_group_iv_cost, force_var_cost): Ditto.
28600 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
28601 (get_computation_cost_at, get_computation_cost): Ditto.
28602 (determine_group_iv_cost_generic): Ditto.
28603 (determine_group_iv_cost_address): Ditto.
28604 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
28605 (determine_group_iv_costs): Ditto.
28606 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
28607 (iv_ca_set_remove_invariants): Renamed to ...
28608 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
28609 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
28610 (iv_ca_set_add_invariants): Renamed to ...
28611 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
28612 (iv_ca_set_cp): Use iv_ca_set_add_invs.
28613 (iv_ca_has_deps): Support inv_vars and inv_exprs.
28614 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
28615 (create_new_ivs): Remove useless dump.
28617 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28619 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
28621 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
28622 (iv_ca_set_no_cp, create_new_iv): Ditto.
28624 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28626 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28628 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28630 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
28631 function tree_check2.
28633 2017-05-02 Martin Liska <mliska@suse.cz>
28635 * doc/gcov.texi: Add missing preposition.
28636 * gcov.c (function_info::function_info): Properly fill up
28637 all member variables.
28639 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28641 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
28643 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28645 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
28647 2017-05-02 Martin Liska <mliska@suse.cz>
28650 * lto-streamer-in.c (lto_read_tree_1): Remove
28651 LTO_STREAMER_DEBUG.
28652 * lto-streamer.c (struct tree_hash_entry): Likewise.
28653 (struct tree_entry_hasher): Likewise.
28654 (tree_entry_hasher::hash): Likewise.
28655 (tree_entry_hasher::equal): Likewise.
28656 (lto_streamer_init): Likewise.
28657 (lto_orig_address_map): Likewise.
28658 (lto_orig_address_get): Likewise.
28659 (lto_orig_address_remove): Likewise.
28660 * lto-streamer.h: Likewise.
28661 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
28662 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
28664 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
28666 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
28667 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
28668 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
28669 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
28670 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
28671 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
28672 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
28673 (mm_maskz_sub_ss): New intrinsics.
28674 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28675 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28676 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
28677 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
28678 (__builtin_ia32_subss_mask_round): New builtins.
28679 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28680 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28681 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
28683 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
28684 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28686 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28689 2017-05-02 Martin Jambor <mjambor@suse.cz>
28691 PR tree-optimization/78687
28692 * tree-sra.c (access): New field parent.
28693 (process_subtree_disqualification): New function.
28694 (disqualify_candidate): Call it.
28695 (build_accesses_from_assign): Reset write flag if creating an
28697 (build_access_subtree): Fill in parent field and also prpagate
28698 down grp_write flag.
28699 (create_artificial_child_access): New parameter set_grp_write, set
28700 grp_write to its value.
28701 (propagate_subaccesses_across_link): Also propagate grp_write flag
28703 (propagate_all_subaccesses): Push the closest parent back to work
28704 queue if add_access_to_work_queue returned true.
28706 2017-05-02 Richard Biener <rguenther@suse.de>
28708 * common.opt (fstrict-overflow): Alias negative to fwrapv.
28709 * doc/invoke.texi (fstrict-overflow): Remove all traces of
28710 -fstrict-overflow documentation.
28711 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
28712 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
28713 flag_strict_overflow.
28714 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
28715 * lto-opts.c (lto_write_options): Do not stream it.
28716 * lto-wrapper.c (merge_and_complain): Do not handle it.
28717 * opts.c (default_options_table): Do not set -fstrict-overflow.
28718 (finish_options): Likewise do not clear it when sanitizing.
28719 * simplify-rtx.c (simplify_const_relational_operation): Do not
28720 test flag_strict_overflow.
28722 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
28724 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
28725 using enabled attribute.
28726 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
28727 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
28728 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
28729 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
28730 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
28731 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
28732 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
28733 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
28734 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
28735 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
28737 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
28739 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
28741 2017-05-02 Richard Biener <rguenther@suse.de>
28743 PR tree-optimization/80591
28745 2017-04-10 Richard Biener <rguenther@suse.de>
28747 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28750 2017-05-02 Richard Biener <rguenther@suse.de>
28752 PR tree-optimization/80549
28753 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
28754 (cleanup_tree_cfg_noloop): Create forwarders to known loop
28755 headers if they do not have a preheader.
28757 2017-05-02 Martin Liska <mliska@suse.cz>
28760 * common.opt: Fix typo.
28761 * doc/invoke.texi: Likewise.
28763 2017-05-01 Jan Beulich <jbeulich@suse.com>
28765 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
28766 swapping, add (x,x,m,x,n) alternative.
28768 2017-05-01 Nathan Sidwell <nathan@acm.org>
28770 * calls.c (combine_pending_stack_adjustment_and_call): Remove
28771 unnecessary unadjusted_alignment check.
28773 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28776 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
28778 * gimplify.c (gimplify_cilk_detach): New function.
28779 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
28780 * tree-core.h: Document EXPR_CILK_SPAWN.
28781 * tree.h (EXPR_CILK_SPAWN): Define.
28783 2017-05-01 David Malcolm <dmalcolm@redhat.com>
28785 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
28786 to use new fixit_hint representation, using the "replace" logic.
28787 (get_line_span_for_fixit_hint): Likewise.
28788 (layout::print_any_fixits): Likewise.
28789 (selftest::test_one_liner_many_fixits): Rename to...
28790 (selftest::test_one_liner_many_fixits_1): ...this, and update
28791 comment and expected output to reflect that the multiple fix-it
28792 hints are now consolidated into one insertion.
28793 (selftest::test_one_liner_many_fixits_2): New test.
28794 (selftest::test_diagnostic_show_locus_one_liner): Update for
28796 (selftest::test_fixit_consolidation): Update for fix-it API
28798 * diagnostic.c (print_parseable_fixits): Likewise.
28799 * edit-context.c (edited_line::m_line_events): Convert from
28800 auto_vec <line_event *> to auto_vec <line_event>.
28801 (class line_event): Convert from abstract base class to a concrete
28802 class, taking over the role of replace_event.
28803 (class insert_event): Delete.
28804 (class replace_event): Rename to class line_event. Convert to
28806 (edit_context::add_fixits): Reimplement.
28807 (edit_context::apply_insert): Delete.
28808 (edit_context::apply_replace): Rename to...
28809 (edit_context::apply_fixit): ...this. Convert to half-open range.
28810 (edited_file::apply_insert): Delete.
28811 (edited_file::apply_replace): Rename to...
28812 (edited_file::apply_fixit): ...this.
28813 (edited_line::~edited_line): Drop deletion of events.
28814 (edited_line::apply_insert): Delete.
28815 (edited_line::apply_replace): Rename to...
28816 (edited_line::apply_fixit): ...this. Convert to half-open range.
28817 Update for change to type of m_line_events.
28818 * edit-context.h (edit_context::apply_insert): Delete.
28819 (edit_context::apply_replace): Rename to...
28820 (edit_context::apply_fixit): ...this.
28822 2017-05-01 Martin Sebor <msebor@redhat.com>
28824 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
28827 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
28830 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
28831 __get_cpuid_max returns 0.
28832 (__get_cpuid_count): Ditto.
28834 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
28836 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
28837 replacement expression is another instance of one of its arguments.
28839 2017-05-01 Jakub Jelinek <jakub@redhat.com>
28842 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
28843 check for stack push/pop autoinc.
28844 * config/i386/i386.c (ix86_agi_dependent): Return false
28845 if the only reason why modified_in_p returned true is that
28846 addr is SP based and set_insn is a push or pop.
28848 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28850 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
28853 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28856 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
28857 (account_size_time): Use two predicates - exec_pred and
28859 (evaluate_conditions_for_known_args): Compute both clause and
28861 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
28862 (inline_summary_t::duplicate): Update.
28863 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
28865 (compute_inline_parameters): Likewise.
28866 (estimate_edge_size_and_time): Update caluclation of time.
28867 (estimate_node_size_and_time): Compute both time and nonspecialized
28869 (estimate_ipcp_clone_size_and_time): Update.
28870 (inline_merge_summary): Update.
28871 (do_estimate_edge_time): Update.
28872 (do_estimate_edge_size): Update.
28873 (do_estimate_edge_hints): Update.
28874 (inline_read_section, inline_write_summary): Stream both new predicates.
28875 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
28877 (compute_inlined_call_time): Cleanup.
28878 (big_speedup_p): Update.
28879 (edge_badness): Update.
28880 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
28881 (size_time_entry): Replace predicate by exec_predicate and
28882 nonconst_predicate.
28883 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
28884 (estimate_edge_time): Return also nonspec_time.
28885 (reset_edge_growth_cache): Update.
28887 2017-04-29 Jakub Jelinek <jakub@redhat.com>
28889 PR rtl-optimization/80491
28890 * ifcvt.c (noce_process_if_block): When looking for x setter
28891 with missing else_bb, don't check only the insn right before
28892 cond_earliest, but look for the last insn that x is modified in
28893 within the same bb.
28895 PR rtl-optimization/80491
28896 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
28898 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
28900 PR tree-optimization/80487
28901 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
28903 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28905 PR tree-optimization/79697
28906 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
28907 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
28908 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
28910 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
28911 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
28913 2017-04-28 Martin Sebor <msebor@redhat.com>
28915 PR tree-optimization/80523
28916 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
28917 (init_target_to_host_charmap, target_to_host, target_strtol10): New
28919 (maybe_warn, format_directive, parse_directive): Use new functions.
28920 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
28922 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
28924 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
28926 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
28928 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
28929 target_header_dir): Set correctly.
28930 * configure: Regenerated.
28931 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28932 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28933 instead of SYSTEM_HEADER_DIR.
28935 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
28937 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
28938 (estimate_local_effects): Likewise.
28939 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
28940 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
28941 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
28942 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
28943 do_estimate_edge_time, estimate_edge_time): Likewise.
28944 * ipa-inline-analysis.c (estimate_node_size_and_time,
28945 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
28946 (estimate_time_after_inlining): Remove.
28948 2017-04-28 Martin Liska <mliska@suse.cz>
28950 * doc/gcov.texi: Enhance documentation of gcov.
28952 2017-04-28 Martin Liska <mliska@suse.cz>
28954 * doc/gcov.texi: Sort options in alphabetic order.
28955 * doc/gcov-dump.texi: Likewise.
28956 * doc/gcov-tool.texi: Likewise.
28957 * gcov.c (print_usage): Likewise.
28958 * gcov-dump.c (print_usage): Likewise.
28959 * gcov-tool.c (print_merge_usage_message): Likewise.
28960 (print_rewrite_usage_message): Likewise.
28961 (print_overlap_usage_message): Likewise.
28963 2017-04-28 Martin Liska <mliska@suse.cz>
28965 PR gcov-profile/53915
28966 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
28968 2017-04-28 Martin Liska <mliska@suse.cz>
28970 PR gcov-profile/79891
28971 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
28972 is marked by compiler as living on a line.
28973 (get_cycles_count): Remove usage of the union.
28974 (output_intermediate_file): Likewise.
28975 (find_source): Fix GNU coding style.
28976 (accumulate_line_counts): Remove old non-all block mode.
28977 (output_lines): Remove usage of the union.
28978 * profile.c (output_location): Include all BBs, even if
28979 belonging to a same line (and file) as a previous BB.
28981 2017-04-28 Martin Liska <mliska@suse.cz>
28983 * gcov.c (process_args): Handle new argument 'w'.
28984 (read_graph_file): Assign ID to BBs.
28985 (output_branch_count): Display BB # if verbose flag is set.
28986 (output_lines): Likewise for arcs.
28987 (print_usage): Add '--verbose' option help.
28988 * doc/gcov.texi: Document --verbose (-w) option.
28990 2017-04-28 Martin Liska <mliska@suse.cz>
28992 * gcov.c (struct block_location_info): New struct.
28993 (process_file): Fill up the new structure.
28994 (read_graph_file): Replace usage of encoding by the newly added
28996 (add_line_counts): Likewise.
28997 (accumulate_line_counts): Remove usage of the union.
28998 (function_info::function_info): New function.
28999 (function_info::~function_info): Likewise.
29000 (process_file): Call delete instead of release_function.
29001 (release_function): Release the function.
29002 (release_structures): Call delete instead of release_function.
29003 (solve_flow_graph): Replace usage of num_blocks.
29004 (find_exception_blocks): Likewise.
29005 (output_lines): Fix GNU coding style.
29007 2017-04-28 Martin Liska <mliska@suse.cz>
29010 * coverage.c (coverage_remove_note_file): New function.
29011 * coverage.h: Declare the function.
29012 * toplev.c (finalize): Clean if an error has been seen.
29014 2017-04-28 Martin Liska <mliska@suse.cz>
29016 PR gcov-profile/80031
29017 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
29018 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
29019 * gcov.c (read_graph_file): Read just number of blocks.
29020 * profile.c (branch_prob): Do not stream 0 flags per a basic
29023 2017-04-28 Martin Liska <mliska@suse.cz>
29025 * gcov-dump.c (tag_*): Add new argument to declarations.
29026 (dump_gcov_file): Likewise.
29027 (tag_blocks): Add and use new argument depth.
29028 (tag_arcs): Likewise.
29029 (tag_lines): Likewise.
29030 (tag_counters): Likewise.
29031 (tag_summary): Likewise.
29032 (dump_working_sets): Use depth to do a proper indentation.
29034 2017-04-28 Jakub Jelinek <jakub@redhat.com>
29037 * cgraph.h (symtab_node::debug_symtab): No longer inline.
29038 * symtab.c (symtab_node::debug_symtab): Move definition here.
29040 2017-04-28 Richard Biener <rguenther@suse.de>
29042 * lto-streamer.h (LTO_major_version): Bump to 7.
29044 2017-04-28 Richard Biener <rguenther@suse.de>
29046 * tree-vrp.c (assert_info): New struct.
29047 (add_assert_info): New helper.
29048 (register_edge_assert_for_2): Refactor to add asserts to a vector
29050 (register_edge_assert_for_1): Likewise.
29051 (register_edge_assert_for): Likewise.
29052 (finish_register_edge_assert_for): New helper actually registering
29053 asserts where live on edge.
29054 (find_conditional_asserts): Adjust.
29055 (find_switch_asserts): Likewise.
29056 (evrp_dom_walker::try_find_new_range): Generalize.
29057 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
29059 2017-04-27 Marek Polacek <polacek@redhat.com>
29062 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
29063 arg10 and arg11 to itype.
29065 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
29067 * doc/extend.texi (Object Size Checking): Improve grammar.
29069 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
29072 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
29073 that the logic for permitting reciprocal estimates matches that
29076 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29079 * tree.c (type_cache_hasher::equal): Only compare
29080 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
29081 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
29082 non-aggregate element types.
29083 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
29084 about the flag on ARRAY_TYPEs in the comment, formatting fix.
29086 2017-04-27 Richard Biener <rguenther@suse.de>
29088 PR middle-end/80533
29089 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
29090 stripping ARRAY_REFs from MEM_EXPR make sure we're not
29091 keeping a reference to a trailing array.
29093 2017-04-27 Richard Biener <rguenther@suse.de>
29095 PR middle-end/80539
29096 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
29097 being in loop-closed SSA form conservatively.
29098 (chrec_fold_multiply_poly_poly): Likewise.
29100 2017-04-27 Tamar Christina <tamar.christina@arm.com>
29102 PR middle-end/79665
29103 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
29104 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
29106 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29109 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
29110 (aarch64_function_arg_alignment): Return unsigned int again, but still
29111 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
29112 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29113 Don't emit -Wpsabi note.
29114 (aarch64_function_arg_boundary): Likewise.
29115 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29118 2017-04-26 Nathan Sidwell <nathan@acm.org>
29120 * tree.h (crc32_unsigned_n): Declare.
29121 (crc32_unsigned, crc32_unsigned): Make inline.
29122 * tree.c (crc32_unsigned_bits): Replace with ...
29123 (crc32_unsigned_n): ... this.
29124 (crc32_unsigned, crc32_byte): Remove.
29125 (crc32_string): Remove unnecessary braces.
29127 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
29129 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
29130 * ipa-inline-analysis.c (MAX_TIME): Remove.
29131 (account_size_time): Use sreal for time.
29132 (dump_inline_summary): Update.
29133 (estimate_function_body_sizes): Update.
29134 (estimate_edge_size_and_time): Update.
29135 (estimate_calls_size_and_time): Update.
29136 (estimate_node_size_and_time): Update.
29137 (inline_merge_summary): Update.
29138 (inline_update_overall_summary): Update.
29139 (estimate_time_after_inlining): Update.
29140 (inline_read_section): Update.
29141 (inline_write_summary): Update.
29142 * ipa-inline.c (compute_uninlined_call_time): Update.
29143 (compute_inlined_call_time): Update.
29144 (recursive_inlining): Update.
29145 (inline_small_functions): Update.
29146 (dump_overall_stats): Update.
29147 * ipa-inline.h: Include sreal.h.
29148 (size_time_entry): Turn time to sreal.
29149 (inline_summary): Turn self_time nad time to sreal.
29151 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
29153 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
29155 (sreal::stream_out, sreal::stream_in): New.
29156 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
29158 2017-04-25 Jakub Jelinek <jakub@redhat.com>
29160 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
29163 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
29166 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
29168 (dimode_scalar_chain::compute_convert_gain): Ditto.
29169 (dimode_scalar_chain::make_vector_copies): Ditto.
29170 (dimode_scalar_chain::convert_reg): Ditto.
29171 (dimode_scalar_chain::convert_insn): Ditto.
29172 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
29173 (VI248_AVX512BW_1): New mode iterator.
29174 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
29175 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
29178 2017-04-25 Martin Sebor <msebor@redhat.com>
29180 PR tree-optimization/80497
29181 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
29182 constants are representable in HOST_WIDE_INT.
29183 (parse_directive): Ditto.
29185 2017-04-25 Martin Sebor <msebor@redhat.com>
29188 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
29189 (new_zero_array): Adjust signature.
29190 (dom_info::dom_init): Used unsigned rather that size_t.
29191 (dom_info::dom_info): Same.
29193 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29194 Jakub Jelinek <jakub@redhat.com>
29197 * config/arm/arm.c: Include gimple.h.
29198 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29199 returns negative, increment ncrn only if it returned positive.
29200 (arm_needs_doubleword_align): Return int instead of bool,
29201 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
29202 members, but if there is any such non-FIELD_DECL
29203 > PARM_BOUNDARY aligned decl, return -1 instead of false.
29204 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29205 returns negative, increment nregs only if it returned positive.
29206 (arm_setup_incoming_varargs): Likewise.
29207 (arm_function_arg_boundary): Emit -Wpsabi note if
29208 arm_needs_doubleword_align returns negative, return
29209 DOUBLEWORD_ALIGNMENT only if it returned positive.
29211 2017-04-25 Marek Polacek <polacek@redhat.com>
29214 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29215 first argument to type.
29217 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
29220 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
29221 type checks to test for compatibility instead of equality.
29223 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29224 Jakub Jelinek <jakub@redhat.com>
29227 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
29229 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
29230 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
29231 the alignment computation, but return their maximum in warn_alignment.
29232 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29233 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
29235 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
29236 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29239 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29241 * config/arc/simdext.md (dmpyh): Fix typo.
29243 2017-04-25 Richard Biener <rguenther@suse.de>
29245 PR tree-optimization/80492
29246 * alias.c (compare_base_decls): Handle registers with asm
29247 specification conservatively.
29248 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29249 compare_base_decls returning dont-know properly.
29251 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29253 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
29254 (legitimate_offset_address_p): New function.
29255 (arc_legitimate_address_p): Use above function.
29257 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29259 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
29261 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29263 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
29264 ACCH registers whenever they are available.
29266 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29268 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
29269 double regs fix when not used.
29271 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29273 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
29275 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
29276 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
29278 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29280 * config/arc/arc.c (arc_output_addsi): Check for h-register class
29281 when emitting short ADD instructions.
29283 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29285 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
29287 (cmpsi_cc_c_insn): Likewise.
29288 (cbranchsi4_scratch): Compute proper instruction length using
29289 compact_hreg_operand.
29290 * config/arc/predicates.md (compact_hreg_operand): New predicate.
29292 2017-04-25 Richard Biener <rguenther@suse.de>
29294 PR middle-end/80509
29295 * passes.c (pass_manager::pass_manager): Initialize
29296 m_name_to_pass_map.
29298 2017-04-25 Richard Biener <rguenther@suse.de>
29300 PR tree-optimization/79201
29301 * tree-ssa-sink.c (statement_sink_location): Handle calls.
29303 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29306 * config/s390/vector.md: Split MEM->GPR vector moves for
29307 non-s_operand addresses.
29309 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29312 * config/s390/predicates.md (reload_const_wide_int_operand): New
29314 * config/s390/s390.md ("movti"): Remove d/P alternative.
29315 ("movti_bigconst"): New pattern definition.
29317 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29320 * s390-protos.h (s390_expand_cs_hqi): Removed.
29321 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
29322 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
29323 modes as well as CCZ1mode and CCZmode.
29324 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
29325 signature of s390_emit_compare_and_swap.
29326 (s390_expand_cs_hqi): Likewise, make static.
29327 (s390_expand_cs_tdsi): Generate an explicit compare before trying
29328 compare-and-swap, in some cases.
29329 (s390_expand_cs): Wrapper function.
29330 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
29332 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
29333 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
29334 patterns for small and large integers. Forbid symref memory operands.
29335 Move expander to s390.c. Require cc register.
29336 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
29337 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
29338 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
29339 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
29340 symref memory operands. Remove CC mode and call s390_match_ccmode
29342 ("atomic_exchange<mode>"): Allow and implement all integer modes.
29344 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29346 * config/s390/s390.md (define_peephole2): New peephole to help
29347 combining the load-and-test pattern with volatile memory.
29349 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29351 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
29352 with CCZmode for TARGET_Z196.
29354 2017-04-25 Jakub Jelinek <jakub@redhat.com>
29356 PR rtl-optimization/80501
29357 * combine.c (make_compound_operation_int): Set subreg_code to SET
29358 even for AND with mask of the sign bit of mode.
29360 PR rtl-optimization/80500
29361 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
29362 sum's initial value.
29364 2017-04-25 Julian Brown <julian@codesourcery.com>
29365 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29367 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
29369 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
29371 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
29373 2017-04-25 Julian Brown <julian@codesourcery.com>
29374 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29376 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
29377 (thunderx2t99_sha): New Reservation.
29379 2017-04-25 Julian Brown <julian@codesourcery.com>
29380 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29382 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
29383 type for 1-element load.
29385 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
29387 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
29389 2017-04-24 Martin Jambor <mjambor@suse.cz>
29391 PR tree-optimization/80293
29392 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
29393 char arrays not totally scalarizable if it is false.
29394 (analyze_all_variable_accesses): Pass correct value in the new
29395 parameter. Add a statistics counter.
29397 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
29399 PR middle-end/79931
29400 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
29402 2017-04-24 Richard Biener <rguenther@suse.de>
29404 PR tree-optimization/80494
29405 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
29406 out for complex types.
29408 2017-04-24 Richard Biener <rguenther@suse.de>
29410 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
29411 * tree-ssa-sccvn.c (print_scc): Print SCC size.
29412 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
29413 (DFS): Adjust and never fail.
29414 (sccvn_dom_walker::fail): Remove.
29415 (sccvn_dom_walker::before_dom_children): Adjust.
29416 (run_scc_vn): Likewise and never fail.
29417 * tree-ssa-pre.c (pass_pre::execute): Adjust.
29418 (pass_fre::execute): Likewise.
29420 2017-04-24 Richard Biener <rguenther@suse.de>
29422 PR tree-optimization/79725
29423 * tree-ssa-sink.c (statement_sink_location): Return whether
29424 failure reason was zero uses. Move that check later.
29425 (sink_code_in_bb): Deal with zero uses by removing the stmt
29428 2017-04-24 Richard Biener <rguenther@suse.de>
29431 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
29432 pointer-based references.
29434 2017-04-24 Richard Biener <rguenther@suse.de>
29437 * pass_manager.h (pass_manager::operator new): Remove.
29438 (pass_manager::operator delete): Likewise.
29439 * passes.c (pass_manager::operator new): Remove.
29440 (pass_manager::operator delete): Likewise.
29441 (pass_manager::pass_manager): Zero individual pass members.
29443 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
29446 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
29447 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
29448 Check "XEXP (src, 1)" operand here.
29449 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
29450 Check "XEXP (src, 1)" operand here.
29451 (dimode_scalar_chain::make_vector_copies): Detect count register
29452 of a shift instruction. Zero extend count register from QImode
29453 to DImode to satisfy vector shift pattern count operand predicate.
29454 Substitute vector shift count operand with a DImode copy.
29455 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
29458 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
29460 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
29461 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29462 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
29463 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29464 (UNSPEC_NOREX_MEM): Remove definition.
29466 2017-04-21 Richard Biener <rguenther@suse.de>
29468 PR tree-optimization/79547
29469 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
29470 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
29471 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
29472 without any constraints.
29474 2017-04-21 Richard Biener <rguenther@suse.de>
29476 PR tree-optimization/78847
29477 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
29479 2017-04-21 Richard Biener <rguenther@suse.de>
29481 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
29482 (build_distinct_type_copy): Likewise.
29483 (build_variant_type_copy): Likewise.
29484 * tree.c (build_qualified_type): Pass down mem-stat info.
29485 (build_distinct_type_copy): Likewise.
29486 (build_variant_type_copy): Likewise.
29488 2017-04-21 Richard Biener <rguenther@suse.de>
29490 PR tree-optimization/80237
29491 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
29493 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
29494 for a simplified result.
29496 2016-04-21 Richard Biener <rguenther@suse.de>
29498 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
29499 sth as strict as a simple_iv but a chrec without symbols and an
29500 operand defined in the loop we are peeling (and not some subloop).
29501 (propagate_constants_for_unrolling): Propagate all constants.
29503 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29506 * config/i386/i386.c (print_reg): Remove assert for disalowed
29507 regno values, call output_operand_lossage instead.
29509 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29512 * config/i386/constraints.md (Yc): New register constraint.
29513 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
29514 Use Yc constraint for alternative 2 of operand 0. Remove
29515 preferred_for_speed attribute.
29517 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29519 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
29520 lastprivate clauses in SIMT case.
29522 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
29524 * doc/invoke.texi (-Wextra-semi): Document new warning option.
29526 2017-04-20 Richard Biener <rguenther@suse.de>
29528 PR tree-optimization/57796
29529 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
29530 as N scalar stores.
29531 (vect_model_load_cost): Cost gathers as N scalar loads.
29533 2017-04-20 Richard Biener <rguenther@suse.de>
29535 * ggc-page.c (ggc_allocated_p): Rename to ...
29536 (safe_lookup_page_table_entry): ... this and return the lookup
29538 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
29540 2017-04-20 Richard Biener <rguenther@suse.de>
29542 PR tree-optimization/80453
29543 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
29544 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
29545 from the conditions.
29546 (vn_phi_eq): Pass them down.
29547 (vn_phi_lookup): Record them.
29548 (vn_phi_insert): Likewise.
29550 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
29552 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
29553 uninitialized variable warning to avoid buffer overrun.
29555 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29558 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
29559 is suppressed for '{ 0 }' in C.
29561 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29563 * BASE-VER: Set to 8.0.0.
29565 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
29567 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
29568 priority .init_array and .fini_array section with SECTION_NOTYPE
29571 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29573 PR middle-end/80423
29574 * tree.h (build_array_type): Add typeless_storage default argument.
29575 * tree.c (type_cache_hasher::equal): Also compare
29576 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
29577 (build_array_type): Add typeless_storage argument, set
29578 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
29580 (build_nonshared_array_type): Adjust build_array_type_1 caller.
29581 (build_array_type): Likewise. Add typeless_storage argument.
29583 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29584 Jakub Jelinek <jakub@redhat.com>
29586 PR tree-optimization/80426
29587 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
29588 operation on symbolic operands, also compute the overflow for the
29589 invariant part when the operation degenerates into a negation.
29591 2017-04-19 Jakub Jelinek <jakub@redhat.com>
29594 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
29595 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
29598 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
29600 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
29603 * config/avr/avr.c (tree.h): Include it.
29604 (cgraph.h): Include it.
29605 (avr_encode_section_info): Don't warn for uninitialized progmem
29606 variable if it's just an alias.
29608 2017-04-19 Richard Biener <rguenther@suse.de>
29611 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
29612 when needed by AutoPGO.
29614 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
29617 * doc/lto.texi: Remove an extra 'that'.
29619 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
29621 PR rtl-optimization/80429
29622 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
29623 are only used in debug insns.
29625 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29626 Vladimir Makarov <vmakarov@redhat.com>
29628 * config/sparc/predicates.md (input_operand): Add comment. Return
29629 true for any memory operand when LRA is in progress.
29630 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
29632 2017-04-18 Jeff Law <law@redhat.com>
29635 * mips.md ({return,simple_return}_internal): Do not overwrite
29638 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29640 PR tree-optimization/80443
29641 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
29642 instead of adding 1, subtract -1 and similarly instead of subtracting
29645 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
29647 PR rtl-optimization/80357
29648 * haifa-sched.c (tmp_bitmap): New variable.
29649 (model_recompute): Handle duplicate use records.
29650 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
29651 (free_global_sched_pressure_data): Free it.
29653 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
29656 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29657 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29658 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29659 instead of SYSTEM_HEADER_DIR.
29661 2017-04-18 Jeff Law <law@redhat.com>
29663 PR middle-end/80422
29664 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
29665 predecessors after walking up the insn chain.
29667 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29670 * dwarf2out.c (modified_type_die): Try harder not to emit internal
29671 sizetype type into debug info.
29673 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29676 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
29677 unneeded test for TARGET_UPPER_REGS_SF.
29678 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
29680 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29683 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
29684 instead of gsi_after_labels.
29686 2017-04-18 Jeff Law <law@redhat.com>
29688 * regcprop.c (maybe_mode_change): Avoid creating copies of the
29692 2017-04-13 Jeff Law <law@redhat.com>
29693 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
29694 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
29696 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
29699 * config/avr/avr.c (intl.h): Include it.
29700 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
29702 2017-04-18 Martin Liska <mliska@suse.cz>
29704 PR gcov-profile/78783
29705 * gcov-tool.c (gcov_output_files): Validate that destination
29706 file is either removed by the tool or by a user.
29708 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
29709 Guy Benyei <guybe@mellanox.com>
29711 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
29712 block, and do not negate it, the stored id is already negative.
29714 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
29716 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
29718 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29721 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
29722 masks of options that should be turned off if the VSX vector
29723 options are turned off.
29724 (OTHER_P8_VECTOR_MASKS): Likewise.
29725 (OTHER_VSX_VECTOR_MASKS): Likewise.
29726 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
29727 rs6000_disable_incompatible_switches to validate no type switches
29729 (rs6000_incompatible_switch): New function to disallow turning on
29730 other vector options if -mno-vsx, -mno-power8-vector, or
29731 -mno-power9-vector are specified.
29733 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29735 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
29737 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29739 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
29740 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
29741 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
29742 (ARG_POINTER_CFA_OFFSET): Likewise.
29744 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29746 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
29747 conditions to take advantage of various optimizations.
29749 2017-04-13 Jeff Law <law@redhat.com>
29751 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
29752 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
29753 (zero_extendsidi2_dext): Likewise.
29755 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29758 * fold-const.c (fold_ternary_loc): Revert
29759 use op0 instead of fold_convert_loc (loc, type, arg0) part of
29762 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
29764 PR rtl-optimization/80343
29765 * lra-remat.c (update_scratch_ops): Assign original hard reg to
29766 new scratch pseudo.
29768 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
29771 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
29772 to ubsan_encode_value.
29774 2017-04-13 Jeff Law <law@redhat.com>
29776 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
29777 appearing in DEBUG_INSNs.
29779 2017-04-13 Martin Liska <mliska@suse.cz>
29781 PR gcov-profile/80413
29782 * gcov-io.c (gcov_write_string): Copy to buffer just when
29783 allocated size is greater than zero.
29785 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29788 * dwarf2out.c (decls_for_scope): Ignore declarations of
29789 current_function_decl in BLOCK_NONLOCALIZED_VARS.
29791 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
29794 * ipa-visibility.c (non_local_p): Fix typos.
29795 (localize_node): When localizing symbol in same comdat group,
29796 dissolve the group only when we know external symbols are going
29798 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
29800 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29802 PR tree-optimization/79390
29803 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
29804 order does not result in usable sequence, retry with reversed operand
29810 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
29811 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
29812 op0 instead of fold_convert_loc (loc, type, arg0).
29814 2017-04-12 Jeff Law <law@redhat.com>
29816 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
29817 has a delay slot in the generated code.
29819 * config/cris/cris.md (cris_preferred_reload_class): Return
29820 GENNONACR_REGS rather than GENERAL_REGS.
29822 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29825 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
29826 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
29827 signedness of the result type.
29829 2017-04-12 Richard Biener <rguenther@suse.de>
29830 Jeff Law <law@redhat.com>
29832 PR tree-optimization/80359
29833 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
29834 trim stores to TARGET_MEM_REFs.
29836 2017-04-12 Richard Biener <rguenther@suse.de>
29838 PR tree-optimization/79390
29839 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
29840 threading case even more.
29842 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
29845 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
29846 for quad_address_p for TImode, instead of just not indexed_address.
29848 2017-04-12 Richard Biener <rguenther@suse.de>
29849 Bernd Edlinger <bernd.edlinger@hotmail.de>
29851 PR middle-end/79671
29852 * alias.c (component_uses_parent_alias_set_from): Handle
29853 TYPE_TYPELESS_STORAGE.
29854 (get_alias_set): Likewise.
29855 * tree-core.h (tree_type_common): Add typeless_storage flag.
29856 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
29857 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
29858 for types containing members with TYPE_TYPELESS_STORAGE.
29859 (place_field): Likewise.
29860 (layout_type): Likewise for ARRAY_TYPE.
29861 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
29862 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
29863 TYPE_TYPELESS_STORAGE.
29864 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
29866 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29869 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29870 first argument to type.
29872 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29876 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
29877 CONST0_RTX (mode) rather than const0_rtx where appropriate.
29878 (rs6000_expand_binop_builtin): Likewise.
29879 (rs6000_expand_ternop_builtin): Likewise; also add missing
29880 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
29881 vshasigma built-ins.
29882 * doc/extend.texi: Document that vec_xxpermdi's third argument
29883 must be a constant.
29885 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
29887 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
29888 Use shift_const cost parameter when calculating gain of STV shifts.
29890 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
29892 PR rtl-optimization/70478
29893 * lra-constraints.c (process_alt_operands): Check memory for
29894 disfavoring memory insn operand.
29896 2017-04-11 Jakub Jelinek <jakub@redhat.com>
29898 PR middle-end/80100
29899 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
29900 left shift in unsigned HOST_WIDE_INT type.
29902 PR rtl-optimization/80385
29903 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
29904 (not (neg X)) into (plus X -1) for complex or non-integral modes.
29907 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
29908 if they have any depend clauses.
29910 2017-04-11 Martin Liska <mliska@suse.cz>
29913 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
29914 * ipa-split.c (split_function): Create a local comdat symbol
29915 if caller is in a comdat group.
29917 2017-04-11 Martin Liska <mliska@suse.cz>
29920 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
29923 2017-04-11 Martin Sebor <msebor@redhat.com>
29925 PR middle-end/80364
29926 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
29927 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
29929 (directive::set_width, directive::set_precision, format_character):
29931 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
29934 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
29937 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
29938 conflict, set target->arch_name instead of target->cpu_name.
29940 2017-04-11 Richard Biener <rguenther@suse.de>
29942 PR tree-optimization/80374
29943 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
29944 build_zero_cst, remove fold_convertible_p check again.
29946 2017-04-11 Martin Liska <mliska@suse.cz>
29949 * ubsan.c (instrument_object_size): Do not instrument register
29952 2017-04-11 Jakub Jelinek <jakub@redhat.com>
29955 * config/i386/i386-builtin-types.def
29956 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
29957 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
29958 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
29959 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
29960 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
29961 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
29962 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
29963 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
29964 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
29965 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
29966 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
29967 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
29968 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
29969 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
29970 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
29971 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
29972 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
29973 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
29974 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
29975 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
29976 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
29977 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
29978 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
29979 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
29980 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
29981 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
29982 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
29983 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
29984 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
29985 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
29986 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
29987 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
29988 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
29989 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
29990 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
29991 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
29992 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
29993 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
29994 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
29995 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
29996 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
29997 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
29998 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
29999 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
30000 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
30001 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
30003 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
30004 flag to second_arg_count, handle 4 argument function type _COUNT
30005 aliases, handle second_arg_count on second argument rather than last.
30007 2017-04-10 Jeff Law <law@redhat.com>
30009 PR tree-optimization/80374
30010 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
30011 record anything if we can not convert integer_zero_node to the
30014 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30017 * config/rs6000/rs6000.c (rs6000_option_override_internal):
30018 Enhance special handling given to the TARGET_P9_MINMAX option in
30019 relation to certain other options.
30021 2017-04-10 Bin Cheng <bin.cheng@arm.com>
30023 PR tree-optimization/80153
30024 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
30025 remove POINTER_PLUS_EXPR's base part directly, rather than through
30028 2017-04-10 Richard Biener <rguenther@suse.de>
30029 Bin Cheng <bin.cheng@arm.com>
30031 PR tree-optimization/80153
30032 * tree-affine.c (aff_combination_to_tree): Get base pointer from
30033 the first element of pointer type aff_tree. Build result expr in
30035 (add_elt_to_tree): Convert to type unconditionally. Remove other
30036 fold_convert calls.
30037 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
30038 (rewrite_use_nonlinear_expr): Check invariant using iv information.
30040 2017-04-10 Richard Biener <rguenther@suse.de>
30042 * tree-ssa-structalias.c (find_func_aliases): Properly handle
30045 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
30047 PR rtl-optimization/70478
30048 * lra-constraints.c (curr_small_class_check): New.
30049 (update_and_check_small_class_inputs): New.
30050 (process_alt_operands): Update curr_small_class_check. Disfavor
30051 alternative insn memory operands. Check available regs for small
30054 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30057 * config/mips/mips.opt (-mvirt): Update description.
30058 * doc/invoke.texi (-mvirt): Likewise.
30060 2017-04-10 Richard Biener <rguenther@suse.de>
30062 PR middle-end/80362
30063 * fold-const.c (fold_binary_loc): Look at unstripped ops when
30064 looking for NEGATE_EXPR in -A / -B to A / B folding.
30066 2017-04-10 Martin Liska <mliska@suse.cz>
30068 PR gcov-profile/80224
30069 * gcov.c (print_usage): Fix usage string.
30070 (get_gcov_intermediate_filename): Remove.
30071 (output_gcov_file): Use both for normal and intermediate format.
30072 (generate_results): Do not initialize special file for
30073 intermediate format.
30075 2017-04-10 Richard Biener <rguenther@suse.de>
30077 PR tree-optimization/80304
30078 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
30081 2017-04-10 Nathan Sidwell <nathan@acm.org>
30084 * config/rs6000/rs6000.c (rs6000_vector_type): New.
30085 (rs6000_init_builtins): Use it.
30087 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30089 * config/arm/arm.md (<mrc>): Add mode to SET source.
30090 (<mrrc>): Likewise.
30092 2017-04-10 Richard Biener <rguenther@suse.de>
30094 PR middle-end/80344
30095 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
30097 2017-04-10 Jakub Jelinek <jakub@redhat.com>
30100 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
30101 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
30102 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
30103 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
30104 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
30105 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
30106 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
30107 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
30108 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
30109 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
30110 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
30111 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
30112 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
30113 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
30114 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
30115 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
30116 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
30117 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
30118 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
30119 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
30120 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
30121 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
30122 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
30124 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
30126 PR rtl-optimization/70478
30127 * lra-constraints.c: Reverse the last patch.
30129 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
30131 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
30132 Add comment for WCHAR_T.
30134 2017-04-08 Martin Liska <mliska@suse.cz>
30137 2017-04-07 Martin Liska <mliska@suse.cz>
30140 * ipa-split.c (split_function): Add function part to a same comdat
30143 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30146 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
30148 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
30150 * rs6000/rs6000.c (vec_load_pendulum): Rename...
30151 (vec_pairing): ...to this.
30152 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
30153 (rs6000_sched_init): Adjust for name change.
30154 (struct rs6000_sched_context): Likewise.
30155 (rs6000_init_sched_context): Likewise.
30156 (rs6000_set_sched_context): Likewise.
30158 2017-04-07 Jakub Jelinek <jakub@redhat.com>
30164 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
30166 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
30167 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
30168 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
30170 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
30172 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
30174 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
30176 PR rtl-optimization/70703
30177 * ira-color.c (update_conflict_hard_regno_costs): Use
30178 int64_t instead of HOST_WIDE_INT.
30180 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
30182 PR rtl-optimization/70478
30183 * lra-constraints.c (process_alt_operands): Disfavor alternative
30184 insn memory operands.
30186 2017-04-07 Jeff Law <law@redhat.com>
30188 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
30189 CALL and NOTE_INSN_CALL_ARG_LOCATION.
30191 2017-04-07 Martin Liska <mliska@suse.cz>
30194 * config/aarch64/aarch64.c (aarch64_process_target_attr):
30195 Show error message instead of an ICE.
30197 2017-04-07 Martin Liska <mliska@suse.cz>
30200 * ipa-split.c (split_function): Add function part to a same comdat
30203 2017-04-07 Richard Biener <rguenther@suse.de>
30205 PR middle-end/80341
30206 * tree.c (get_unwidened): Also handle ! for_type case for
30208 * convert.c (do_narrow): Split out from ...
30209 (convert_to_integer_1): ... here. Do not pass final truncation
30210 type to get_unwidened for TRUNC_DIV_EXPR.
30212 2017-04-07 Richard Biener <rguenther@suse.de>
30214 * tree-affine.c (wide_int_ext_for_comb): Take type rather
30216 (aff_combination_const): Adjust.
30217 (aff_combination_scale): Likewise.
30218 (aff_combination_add_elt): Likewise.
30219 (aff_combination_add_cst): Likewise.
30220 (aff_combination_convert): Likewise.
30221 (add_elt_to_tree): Likewise. Remove unused argument.
30222 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
30224 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
30226 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
30228 * config/arm/arm.c (arm_default_short_enums): Use
30229 ARM_DEFAULT_SHORT_ENUMS.
30230 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
30232 2017-04-06 Jakub Jelinek <jakub@redhat.com>
30235 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
30236 members with redundant out-of-class redeclaration.
30238 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
30241 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
30242 * config/i386/i386.md (*zero_extendsidi2):
30243 Add (?*x,*x) and (?*v,*v) alternatives.
30245 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
30248 * config/i386/i386.c (ix86_expand_builtin)
30249 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
30250 mode from insn data. Convert operands to insn operand mode.
30251 Copy operands that don't satisfy insn predicate to a register.
30253 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
30255 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
30258 2017-04-06 Richard Biener <rguenther@suse.de>
30260 PR tree-optimization/80334
30261 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
30262 preserve alignment of accesses.
30264 2017-04-06 Richard Biener <rguenther@suse.de>
30266 PR tree-optimization/80262
30267 * tree-sra.c (build_ref_for_offset): Preserve address-space
30269 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
30270 Drop useless address-space information on MEM_REF offsets.
30272 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
30274 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
30276 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
30278 PR rtl-optimization/70703
30279 * ira-color.c (update_conflict_hard_regno_costs): Use
30280 HOST_WIDE_INT instead of long.
30282 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
30285 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
30286 not defined for x86_64 target. Add -mmmx target option when __SSE2__
30288 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
30289 for x86_64 target. Handle -m3dnowa option.
30291 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
30293 PR rtl-optimization/70703
30294 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
30295 (update_conflict_hard_regno_costs): Use long instead of unsigned
30296 arithmetic for cost calculation.
30298 2017-04-05 Jakub Jelinek <jakub@redhat.com>
30299 Bernd Edlinger <bernd.edlinger@hotmail.de>
30302 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
30305 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
30308 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
30309 ptr_mode with Pmode throughout.
30310 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
30311 into probe_stack_range and use DImode.
30313 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
30316 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
30317 call_eh_return is true.
30319 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30321 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
30322 Initialize last_match_fntype_index.
30324 2017-04-05 Jakub Jelinek <jakub@redhat.com>
30327 * tree-nvr.c: Include internal-fn.h.
30328 (pass_return_slot::execute): Ignore internal calls without
30331 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30332 Richard Biener <rguenther@suse.de>
30335 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
30336 captures used multiple times, except for the last use.
30337 * generic-match-head.c: Include gimplify.h.
30339 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30341 PR tree-optimization/79390
30342 * target.h (struct noce_if_info): Declare.
30343 * targhooks.h (default_noce_conversion_profitable_p): Declare.
30344 * target.def (noce_conversion_profitable_p): New target hook.
30345 * ifcvt.h (struct noce_if_info): New type, moved from ...
30346 * ifcvt.c (struct noce_if_info): ... here.
30347 (noce_conversion_profitable_p): Renamed to ...
30348 (default_noce_conversion_profitable_p): ... this. No longer
30350 (noce_try_store_flag_constants, noce_try_addcc,
30351 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
30352 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
30353 instead of noce_conversion_profitable_p.
30354 * config/i386/i386.c: Include ifcvt.h.
30355 (ix86_option_override_internal): Don't override
30356 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
30357 (ix86_noce_conversion_profitable_p): New function.
30358 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
30359 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
30360 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
30361 * doc/tm.texi: Regenerated.
30363 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30365 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
30368 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
30371 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
30372 instructions for small multiply cores.
30374 2017-04-04 Jeff Law <law@redhat.com>
30376 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
30378 (mips_expand_vec_perm_const): Initialize elements in orig_perm
30379 that are not set by the loop over the elements.
30381 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30384 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
30385 int mode, convert_modes it to mode as unsigned, otherwise use
30386 lowpart_subreg to mode rather than SImode.
30387 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
30388 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
30389 Use DImode instead of SImode for the shift count operand.
30390 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
30393 2017-04-04 Richard Biener <rguenther@suse.de>
30395 PR middle-end/80281
30396 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
30397 arithmetic done for the negate or the plus. Simplify.
30398 (A - (-B) -> A + B): Likewise.
30399 * fold-const.c (split_tree): Make sure to not negate pointers.
30401 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
30403 PR rtl-optimization/60818
30404 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
30405 a compare of comparisons with the thing compared if this results
30406 in a different machine mode.
30408 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
30410 * alias.c (base_alias_check): Fix typo in comment.
30411 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
30412 * cgraphunit.c (symbol_table::compile): Likewise.
30413 * collect2.c (maybe_run_lto_and_relink): Likewise.
30414 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
30415 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
30416 * config/avr/avr.c (avr_map_op_t): Likewise.
30417 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
30418 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
30419 * config/epiphany/epiphany.md (movcc): Likewise.
30420 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
30421 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
30423 * config/mips/mips.c (mips_save_restore_reg): Likewise.
30424 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
30425 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
30426 * config/sh/sh.c (sh_rtx_costs): Likewise.
30427 * fold-const.c (fold_truth_andor): Likewise.
30428 * genautomata.c (collapse_flag): Likewise.
30429 * gengtype.h (struct type::u::s): Likewise.
30430 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
30431 * input.c (FORMAT_AMOUNT): Likewise.
30432 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
30433 (known_aggs_to_agg_replacement_list): Likewise.
30434 * ipa-inline-analysis.c: Likewise.
30435 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
30436 * ipa-polymorphic-call.c
30437 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
30438 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
30439 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
30441 * modulo-sched.c (apply_reg_moves): Likewise.
30442 * omp-expand.c (build_omp_regions_1): Likewise.
30443 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
30444 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
30445 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
30446 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
30447 * value-prof.c: Likewise.
30448 * var-tracking.c (val_reset): Likewise.
30450 2017-04-03 Richard Biener <rguenther@suse.de>
30452 PR tree-optimization/80275
30453 * fold-const.c (split_address_to_core_and_offset): Handle
30456 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
30458 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
30459 descriptors is at least equal to that of functions.
30461 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30463 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
30465 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30468 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
30469 (mov<IMOD4:mode>): New expander.
30470 (*mov<IMOD4:mode>_internal): New insn and split pattern.
30472 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
30474 PR rtl-optimization/79405
30475 * fwprop.c (propagations_left): New variable.
30476 (forward_propagate_into): Decrement it.
30477 (fwprop_init): Initialize it.
30478 (fw_prop): If the variable has reached zero, stop propagating.
30479 (fwprop_addr): Ditto.
30481 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30484 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
30485 a FUNCTION_DECL, pass it as decl instead of origin to
30488 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
30490 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
30493 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
30496 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
30497 TARGET_VSX_SMALL_INTEGER.
30499 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30501 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30502 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
30504 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30506 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
30507 extraction from odd-numbered MSA register.
30509 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30511 PR middle-end/80173
30512 * expmed.c (store_bit_field_1): Don't attempt to create
30513 a word subreg out of hard registers wider than word if they
30514 have HARD_REGNO_NREGS of 1 for their mode.
30516 PR middle-end/80163
30517 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
30518 conversions to integer types wider than word and pointer.
30521 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
30522 (rtx_equal_for_cselib_p): Pass 0 to it.
30523 * cselib.c (cselib_hasher::equal): Likewise.
30524 (rtx_equal_for_cselib_1): Add depth argument. If depth
30525 is 128, don't look up VALUE locs and punt. Increment
30526 depth in recursive calls when walking VALUE locs.
30528 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
30530 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
30531 (make_gcov_file_name): Use the canonical path name for generating
30533 (read_line): Fix handling of files with ascii null bytes.
30535 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
30537 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
30538 to initialise a vector register instead
30539 of using a const_int.
30541 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30543 PR translation/80189
30544 * gimplify.c (omp_default_clause): Use %qs instead of %s in
30545 diagnostic messages.
30547 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
30550 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
30551 (dfp_diex_<mode>): Update mode of operand 1.
30552 * doc/extend.texi (dxex, dxexq): Document change to return type.
30553 (diex, diexq): Document change to argument type.
30555 2017-03-30 Martin Jambor <mjambor@suse.cz>
30558 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
30559 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
30560 it reflects the signature changes performed at the callee side.
30561 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
30562 to cgraph_build_function_type_skip_args.
30563 (build_function_decl_skip_args): Adjust call to the above function.
30565 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30568 * config/i386/sse.md
30569 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
30570 register as dest whenever it is a MEM not rtx_equal_p to the
30571 corresponding dup operand, and when forcing into reg move the
30572 reg into the memory afterwards.
30573 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
30574 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
30575 for the force_reg mode.
30576 (avx512vl_vextractf128<mode>): Use register as dest either
30577 always when a MEM, or when it is a MEM not rtx_equal_p to the
30578 corresponding dup operand, or even not when it is a CONST_VECTOR
30579 depending on the mode and lo vs. hi.
30580 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
30582 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
30583 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
30584 Likewise. Require that operands[2] is even.
30585 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
30586 Remove extraneous parens. Require that operands[2] is a multiple
30588 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
30589 operands[0] is a MEM if <mask_applied>, the predicates/constraints
30590 disallow memory then.
30592 2017-03-30 Richard Biener <rguenther@suse.de>
30594 PR tree-optimization/77498
30595 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
30596 to non-constants over backedges.
30598 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
30600 PR rtl-optimization/80233
30601 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
30602 as last_combined_insn. Do not test for BARRIER_P separately.
30604 2017-03-29 Andreas Schwab <schwab@suse.de>
30607 * calls.c (prepare_call_address): Convert funexp to Pmode before
30608 copying to temp reg.
30610 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30612 PR tree-optimization/80158
30613 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
30614 Handle possible future case of more than one alternate
30616 (replace_rhs_if_not_dup): Likewise.
30617 (replace_one_candidate): Likewise.
30619 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
30621 PR rtl-optimization/80193
30622 * ira.c (ira): Do not check allocation for LRA.
30624 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
30626 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
30627 (nvptx_output_simt_exit): Declare.
30628 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
30629 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
30630 (init_softstack_frame): Move initialization of crtl->is_leaf to...
30631 (nvptx_declare_function_name): ...here. Emit declaration of local
30632 memory space buffer for omp_simt_enter insn.
30633 (nvptx_output_unisimt_switch): New.
30634 (nvptx_output_softstack_switch): New.
30635 (nvptx_output_simt_enter): New.
30636 (nvptx_output_simt_exit): New.
30637 * config/nvptx/nvptx.h (struct machine_function): New fields
30638 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
30639 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
30640 (UNSPECV_SIMT_EXIT): Ditto.
30641 (omp_simt_enter_insn): New insn.
30642 (omp_simt_enter): New expansion.
30643 (omp_simt_exit): New insn.
30644 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
30646 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
30647 (expand_GOMP_SIMT_ENTER_ALLOC): New.
30648 (expand_GOMP_SIMT_EXIT): New.
30649 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
30650 (GOMP_SIMT_ENTER_ALLOC): Ditto.
30651 (GOMP_SIMT_EXIT): Ditto.
30652 * target-insns.def (omp_simt_enter): New insn.
30653 (omp_simt_exit): Ditto.
30654 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
30656 (lower_rec_simd_input_clauses): Implement SIMT privatization.
30657 (lower_rec_input_clauses): Likewise.
30658 (lower_lastprivate_clauses): Handle SIMT privatization.
30660 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
30661 (ompdevlow_adjust_simt_enter): New.
30662 (find_simtpriv_var_op): New.
30663 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
30664 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
30666 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
30667 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
30668 (copy_decl_for_dup_finish): Ditto.
30670 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
30672 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
30675 * config/i386/i386.c (ix86_option_override_internal): Always
30676 allow -mpreferred-stack-boundary=3 for 64-bit targets.
30678 2017-03-28 Bin Cheng <bin.cheng@arm.com>
30680 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
30682 2017-03-28 Bin Cheng <bin.cheng@arm.com>
30684 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
30685 mark new edge's irreducible flag accordign to it.
30686 (vect_do_peeling): Check loop preheader edge's irreducible flag
30687 and pass it to function slpeel_add_loop_guard.
30689 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
30691 PR tree-optimization/80218
30692 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30693 Update block frequencies and counts.
30695 2017-03-28 Richard Biener <rguenther@suse.de>
30697 PR tree-optimization/78644
30698 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
30699 of a simplification result we may not use it at all.
30701 2017-03-28 Richard Biener <rguenther@suse.de>
30704 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
30705 without arguments, generate default definition of a SSA name.
30707 2017-03-28 Richard Biener <rguenther@suse.de>
30709 PR middle-end/80222
30710 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
30711 TYPE_REF_CAN_ALIAS_ALL references.
30712 * fold-const.c (fold_indirect_ref_1): Likewise.
30714 2017-03-28 Martin Liska <mliska@suse.cz>
30717 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
30718 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
30720 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
30721 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
30723 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
30724 (EXTRA_SPECS): Define.
30725 (SUBTARGET_EXTRA_SPECS): Likewise.
30726 (SUBTARGET_CPP_SPEC): Likewise.
30727 * config/arc/elf.h (EXTRA_SPECS): Renamed to
30728 SUBTARGET_EXTRA_SPECS.
30729 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
30731 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
30733 * config/arc/simdext.md (vst64_insn): Update pattern.
30734 (vld32wh_insn): Likewise.
30735 (vld32wl_insn): Likewise.
30736 (vld64_insn): Likewise.
30737 (vld32_insn): Likewise.
30739 2017-03-28 Marek Polacek <polacek@redhat.com>
30742 * fold-const.c (fold_comparison): Use protected_set_expr_location
30743 instead of SET_EXPR_LOCATION.
30745 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
30747 * tree.c (add_expr): Avoid name lookup warning.
30749 2017-03-27 Jeff Law <law@redhat.com>
30751 PR tree-optimization/80216
30752 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
30753 function name. Limit recursion depth.
30754 (record_temporary_equivalences): Corresponding changes.
30756 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
30758 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
30761 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30764 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
30766 * cfgcleanup.c (reg_note_cfa_p): New array.
30767 (insns_have_identical_cfa_notes): New function.
30768 (old_insns_match_p): Don't cross-jump in between /f
30769 and non-/f instructions. If both i1 and i2 are frame related,
30770 verify all CFA notes, their order and content.
30772 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
30775 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
30776 HImode and SImode with zero extend to DImode to one insn.
30777 (bswap<mode>2_extenddi): Likewise.
30778 (bswapsi2_extenddi): Likewise.
30779 (bswaphi2_extendsi): Likewise.
30780 (bswaphi2): Combine bswap HImode and SImode into one insn.
30781 Separate memory insns from swapping register.
30782 (bswapsi2): Likewise.
30783 (bswap<mode>2): Likewise.
30784 (bswaphi2_internal): Delete, no longer used.
30785 (bswapsi2_internal): Likewise.
30786 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
30787 store, and gpr<-gpr swap insns.
30788 (bswap<mode>2_store): Likewise.
30789 (bswaphi2_reg): Register only splitter, combine with the splitter.
30790 (bswaphi2 splitter): Likewise.
30791 (bswapsi2_reg): Likewise.
30792 (bswapsi2 splitter): Likewise.
30793 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
30794 the insns into load, store, and register/register insns.
30795 (bswapdi2_ldbrx): Likewise.
30796 (bswapdi2_load): Likewise.
30797 (bswapdi2_store): Likewise.
30798 (bswapdi2_reg): Likewise.
30800 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
30802 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
30803 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
30805 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
30808 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
30810 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30811 special handling for target option conflicts between dform
30812 options (-mpower9-dform, -mpower9-dform-vector,
30813 -mpower9-dform-scalar) and -mno-direct-move.
30815 2017-03-27 Richard Biener <rguenther@suse.de>
30817 PR tree-optimization/80181
30818 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
30820 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30822 * config/arc/predicates.md (move_double_src_operand): Replace the
30823 call to move_double_src_operand with a call to address_operand.
30825 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30827 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
30828 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
30829 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
30831 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30833 * config/arc/predicates.md (long_immediate_loadstore_operand):
30834 Consider scaled addresses cases.
30836 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30838 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
30839 restored when in interrupt.
30840 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
30841 doesn't have delay slot.
30843 2017-03-27 Richard Biener <rguenther@suse.de>
30846 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
30847 inlined thunk clones.
30849 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30852 * asan.c (instrument_derefs): Copy over last operand from
30853 original COMPONENT_REF to the new COMPONENT_REF with
30854 DECL_BIT_FIELD_REPRESENTATIVE.
30855 * ubsan.c (instrument_object_size): Likewise.
30857 2017-03-27 Richard Biener <rguenther@suse.de>
30859 PR tree-optimization/80170
30860 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
30861 sure DR/SCEV didnt fold in constants we do not see when looking
30862 at the reference base alignment.
30864 2017-03-27 Richard Biener <rguenther@suse.de>
30866 PR middle-end/80171
30867 * gimple-fold.c (fold_ctor_reference): Properly guard against
30868 NULL return value from canonicalize_constructor_val.
30870 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
30873 * config/i386/i386.c (ix86_expand_builtin)
30874 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
30875 flags reg setting and flags reg using instructions.
30876 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
30877 clobbering instructions to zero extend op2.
30879 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
30881 * doc/install.texi (Configuration) <--with-aix-soname>:
30882 Update link to AIX ld.
30884 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
30886 PR rtl-optimization/80160
30887 PR rtl-optimization/80159
30888 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
30889 reg_alternate_class into account.
30891 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
30894 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
30895 to consider in curr_insn_transform.
30897 2017-03-24 Jakub Jelinek <jakub@redhat.com>
30899 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
30900 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
30901 and emit_mode_inner.
30903 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30905 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
30906 argument to the overloaded builtin variants. Use the new flag to
30907 deprecate certain builtin variants.
30908 * config/s390/s390-builtin-types.def: Add new builtin types.
30909 * config/s390/s390-builtins.h: Support new flags field for
30910 overloaded builtins.
30911 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
30912 (s390_macro_to_expand): Enable vector float data type.
30913 (s390_cpu_cpp_builtins_internal): Indicate support of the new
30914 builtins by incrementing the __VEC__ version number.
30915 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
30917 (s390_resolve_overloaded_builtin): Emit error messages depending
30918 on the builtin flags.
30919 * config/s390/s390.c (s390_expand_builtin): Support additional
30920 flags argument. Change error message to match the messages
30921 emitted in s390-c.c.
30922 * config/s390/s390.md: New UNSPEC_* constants.
30923 (op_type): Add new instruction types.
30924 * config/s390/vecintrin.h: Add new builtins and test data class
30926 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
30927 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
30928 (VEC_INEXACT, VEC_NOINEXACT): New constants.
30929 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
30930 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
30931 ("vec_mergel<mode>"): V_HW -> VEC_HW.
30933 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
30934 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
30935 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
30936 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
30938 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
30939 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
30940 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
30941 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
30943 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
30944 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
30945 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
30946 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
30947 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
30948 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
30949 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
30951 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
30952 ("vec_scatter_element<V_HW_4:mode>_DI")
30953 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
30954 ("vec_fpint<mode>", "vflls")
30955 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
30956 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
30957 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
30958 ("*vec_cmphe<mode>_cc"): ... these.
30960 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
30961 mode constant instead of magic value.
30963 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30965 * config/s390/s390.c (s390_expand_vec_compare): Support other
30966 vector floating point modes than just V2DF.
30967 (s390_expand_vcond): Likewise.
30968 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
30969 (s390_cannot_change_mode_class): Prevent mode changes between TF
30970 and V1TF in vector registers.
30971 * config/s390/s390.md (DF, SF): New mode attributes.
30972 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
30973 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
30974 SFmode support for VRs.
30975 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
30977 (VFT, VF_HW): New mode iterators.
30978 (vw, sdx): New mode attributes.
30979 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
30980 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
30981 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
30982 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
30983 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
30984 also the new vector floating point modes. Renaming to ...
30986 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
30987 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
30988 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
30989 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
30990 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
30991 ("vec_unordered<mode>"): ... these.
30993 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
30994 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
30995 ("*vec_extendv2df"): New insn definitions.
30997 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30999 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
31000 ("mulditi3_2", "*muldi3_sign"): New patterns.
31001 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
31002 rename the pattern definition.
31004 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31006 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
31008 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
31010 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31012 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
31013 instruction if possible.
31014 * config/s390/vector.md (vec_halfnumelts): New mode
31016 ("*vec_vllezlf<mode>"): New pattern.
31018 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31020 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
31021 ("popcountv4si2", "popcountv2di2"): Rename to ...
31022 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
31023 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
31025 ("popcount<mode>2_vxe"): New pattern.
31027 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31029 * common/config/s390/s390-common.c (processor_flags_table): Add
31031 * config.gcc: Add arch12.
31032 * config/s390/driver-native.c (s390_host_detect_local_cpu):
31033 Default to arch12 for unknown CPU model numbers.
31034 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
31035 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
31036 PROCESSOR_max sanity check.
31037 * config/s390/s390-opts.h (enum processor_type): Add
31039 * config/s390/s390.c (processor_table): Add arch12.
31040 (s390_expand_builtin): Add check for B_VXE flag.
31041 (s390_issue_rate): Add PROCESSOR_ARCH12.
31042 (s390_get_sched_attrmask): Likewise.
31043 (s390_get_unit_mask): Likewise.
31044 (s390_sched_score): Enable z13 scheduling for arch12.
31045 (s390_sched_reorder): Likewise.
31046 (s390_sched_variable_issue): Likewise.
31047 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
31049 (s390_tune_attr): Use z13 scheduling also for arch12.
31050 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
31051 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
31052 (TARGET_VXE_P): New macros.
31053 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
31054 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
31055 * config/s390/s390.opt: Add arch12 as processor_type.
31057 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31059 * config/s390/s390.md
31060 ("fixuns_truncdddi2", "fixuns_trunctddi2")
31061 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
31062 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
31064 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
31065 Rename expanders to ...
31067 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
31068 ("fixuns_truncdddi2_emu"): ... these.
31070 ("fixuns_trunc<mode>si2_emu"): New expander.
31072 ("*fixuns_truncdfdi2_z13"): Rename to ...
31073 ("*fixuns_truncdfdi2_vx"): ... this.
31075 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31077 * config/s390/2964.md: Remove the single element vector compare
31078 instructions which are no longer used.
31079 * config/s390/s390.c (s390_select_ccmode): Remove handling of
31081 (s390_canonicalize_comparison): Remove handling of DFmode
31083 (s390_expand_vec_compare_scalar): Remove function.
31084 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
31085 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
31087 ("*cmp<mode>_ccs"): Add wfcdb instruction.
31089 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31091 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
31093 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
31094 will anyway by matched by mov<mode>_64dfp.
31096 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31098 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
31099 vlef/vstef. Add missing operand to vleif.
31101 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31103 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
31104 pair for all vector types with 64 bit elements.
31105 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
31106 * config/s390/vector.md (V_HW_64): ... here.
31107 (V_128_NOSINGLE): New mode iterator.
31108 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
31109 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
31110 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
31111 ("*vec_load_pairv2di"): Change to ...
31112 ("*vec_load_pair<mode>"): ... this one.
31114 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31116 * config/s390/constraints.md: Add comments.
31117 (jKK): Reject element sizes > 8 bytes.
31118 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
31120 * config/s390/s390.md: Add the s_operand checks formerly in
31121 s390_split_ok_p to various splitters where they are still
31123 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
31124 for 128 bit vectors. Plus two splitters.
31126 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31128 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
31131 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31134 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
31135 error if the boundary argument is not constant.
31137 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31139 PR rtl-optimization/80112
31140 * loop-doloop.c (doloop_condition_get): Don't check condition
31141 if cmp isn't SET with IF_THEN_ELSE src.
31143 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31145 PR tree-optimization/80158
31146 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
31147 replacing a candidate statement, also replace it for the
31148 candidate's alternate interpretation.
31149 (replace_rhs_if_not_dup): Likewise.
31150 (replace_one_candidate): Likewise.
31152 2017-03-24 Richard Biener <rguenther@suse.de>
31154 PR tree-optimization/80167
31155 * graphite-isl-ast-to-gimple.c
31156 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
31158 (translate_isl_ast_to_gimple::get_rename): Likewise.
31160 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31162 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
31163 handling of certain combinations of target options, including the
31164 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
31165 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
31167 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31170 * config/arm/arm.md (*load_multiple): Add reload_completed to
31171 matching condition.
31173 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31174 Richard Biener <rguenth@suse.de>
31176 PR tree-optimization/79908
31177 PR tree-optimization/80136
31178 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
31179 been cast away, gimplify_and_add suffices.
31181 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
31183 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
31185 2017-03-23 Richard Biener <rguenther@suse.de>
31187 PR tree-optimization/80032
31188 * gimplify.c (gimple_push_cleanup): Forced unconditional
31189 cleanups still have to go to the conditional_cleanups
31192 2017-03-22 Jakub Jelinek <jakub@redhat.com>
31194 PR tree-optimization/80072
31195 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
31197 (next_operand_entry_id): Change type to unsigned int.
31198 (sort_by_operand_rank): Make sure to return the right return value
31199 even if unsigned fields are bigger than INT_MAX.
31200 (struct oecount): Change cnt and id type to unsigned int.
31201 (oecount_hasher::equal): Formatting fix.
31202 (oecount_cmp): Make sure to return the right return value
31203 even if unsigned fields are bigger than INT_MAX.
31204 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
31207 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
31208 TREE_READONLY on result if writing it more than once.
31211 * doc/invoke.texi (-fsanitize=thread): Document that with
31212 -fnon-call-exceptions atomics are not able to throw
31216 * tsan.c: Include tree-eh.h.
31217 (instrument_builtin_call): Call maybe_clean_eh_stmt or
31218 maybe_clean_or_replace_eh_stmt where needed.
31219 (instrument_memory_accesses): Add cfg_changed argument.
31220 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
31221 if it returned true.
31222 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
31224 PR rtl-optimization/63191
31225 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
31226 wrapper function, moved the whole old content into ...
31227 (ix86_delegitimize_address_1): ... this. New inline function.
31228 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
31229 true as last argument instead of ix86_delegitimize_address.
31231 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
31233 * config/aarch64/aarch64.c (generic_branch_cost): Copy
31234 cortexa57_branch_cost.
31236 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
31238 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
31240 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31243 * doc/md.texi (Constraints): Document wA constraint.
31244 * config/rs6000/constraints.md (wA): New.
31245 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
31246 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
31247 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
31248 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
31250 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
31253 * gimplify.c (is_oacc_declared): New function.
31254 (oacc_default_clause): Use it to set default flags for acc declared
31255 variables inside parallel regions.
31256 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
31257 declared variables.
31258 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
31259 declare attribute to any decl as necessary.
31261 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
31264 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
31265 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
31266 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
31267 (arm_arch_lpae): This.
31268 * config/arm/arm.c (arm_arch7ve): Rename into ...
31269 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
31270 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
31273 2017-03-22 Martin Liska <mliska@suse.cz>
31276 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
31277 error message instead of an ICE.
31279 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31281 * doc/extend.texi (6.11 Additional Floating Types): Revise.
31283 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
31285 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
31287 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31290 2017-03-21 Martin Sebor <msebor@redhat.com>
31292 * doc/extend.texi: Use "cannot" instead of "can't."
31293 * doc/hostconfig.texi: Same.
31294 * doc/install.texi: Same.
31295 * doc/invoke.texi: Same.
31296 * doc/loop.texi: Same.
31297 * doc/md.texi: Same.
31298 * doc/objc.texi: Same.
31299 * doc/rtl.texi: Same.
31300 * doc/tm.texi: Same.
31301 * doc/tm.texi.in: Same.
31302 * doc/trouble.texi: Same.
31304 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
31307 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
31308 (collect_checksum_attributes): Set it.
31309 (die_checksum_ordered): Use it.
31311 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31313 PR tree-optimization/79908
31314 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
31315 change: For a VA_ARG whose LHS has been cast away, use
31316 force_gimple_operand to construct the side effects.
31318 2017-03-21 David Malcolm <dmalcolm@redhat.com>
31320 PR translation/80001
31321 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
31322 more amenable to translation.
31323 (oacc_loop_auto_partitions): Likewise.
31325 2017-03-21 Marek Polacek <polacek@redhat.com>
31326 Martin Sebor <msebor@redhat.com>
31328 PR tree-optimization/80109
31329 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
31330 on INTEGRAL_TYPE_P.
31332 2017-03-21 Jakub Jelinek <jakub@redhat.com>
31333 Segher Boessenkool <segher@kernel.crashing.org>
31336 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
31337 check reg_used_between_p between insn and one of succ or succ2
31338 depending on if succ is artificial insn not inserted into insn
31341 2017-03-21 Martin Liska <mliska@suse.cz>
31343 PR gcov-profile/80081
31344 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
31345 * doc/gcc.texi: Include gcov-dump stuff.
31346 * doc/gcov-dump.texi: New file.
31348 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
31350 PR rtl-optimization/79150
31351 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
31352 conditional jump, if the jump is the last insn of the loop.
31354 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31355 Richard Biener <rguenth@suse.de>
31357 PR tree-optimization/79908
31358 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
31359 been cast away, use force_gimple_operand to construct the side
31362 2017-03-21 Martin Liska <mliska@suse.cz>
31364 PR libfortran/79956
31365 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
31368 2017-03-21 Brad Spengler <spender@grsecurity.net>
31371 * plugin.c (htab_hash_plugin): New function.
31372 (add_new_plugin): Use it and adjust.
31373 (parse_plugin_arg_opt): Adjust.
31374 (init_one_plugin): Likewise.
31376 2017-03-21 Richard Biener <rguenther@suse.de>
31378 PR tree-optimization/80032
31379 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
31380 if set force the cleanup to happen unconditionally.
31381 (gimplify_target_expr): Push inserted clobbers with force_uncond
31382 to avoid them being removed by control-dependent DCE.
31384 2017-03-21 Richard Biener <rguenther@suse.de>
31386 PR tree-optimization/80122
31387 * tree-inline.c (copy_bb): Do not expans va-arg packs or
31388 va_arg_pack_len when the inlined call stmt requires pack
31390 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
31392 2017-03-21 Jakub Jelinek <jakub@redhat.com>
31395 * tsan.c (instrument_builtin_call): If the memory model argument
31396 is not a constant, assume it is valid.
31399 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
31402 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
31404 PR rtl-optimization/79910
31405 * combine.c (can_combine_p): Do not allow combining an I0 or I1
31406 if its dest is used by an insn before I2 (other than the combined
31407 insns themselves, which are properly handled already).
31409 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
31412 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31414 * combine.c (record_used_regs): New static function.
31415 (try_combine): Handle situations where there is an additional
31416 instruction between I2 and I3 which needs to have a LOG_LINK
31420 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31422 * combine.c (try_combine): Delete redundant i1 test. Call
31423 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31425 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31428 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
31429 alternatives 13/14.
31431 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31433 PR tree-optimization/80054
31434 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
31435 the optimization if a PHI or any of its arguments is not dominated
31436 by the candidate's basis. Use gphi* rather than gimple* as
31438 (replace_profitable_candidates): Clean up a gimple* variable that
31439 should be a gphi* variable.
31441 2017-03-20 Martin Sebor <msebor@redhat.com>
31444 * doc/extend.texi (attribute constructor): Document present limitation.
31446 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
31449 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
31450 __POWER9_VECTOR__ #ifdef control, change template definition to
31451 use Power9-specific built-in function.
31452 (vec_any_eq): Likewise.
31453 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
31454 to control outcomes from this test.
31455 (vector_ae_<mode>p): For VEC_F modes, likewise.
31457 2017-03-20 Ian Lance Taylor <iant@google.com>
31459 * config/i386/i386.c (ix86_function_regparm): Save an extra
31460 register for -fsplit-stack with DECL_STATIC_CHAIN.
31462 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31465 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
31466 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
31468 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31470 * config/riscv/riscv.c (riscv_print_operand): Use "fence
31472 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
31475 2017-03-20 Marek Polacek <polacek@redhat.com>
31478 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
31480 2017-03-20 Richard Biener <rguenther@suse.de>
31482 PR tree-optimization/80113
31483 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
31484 allocate extra SSA name for PHI def.
31485 (add_close_phis_to_outer_loops): Likewise.
31486 (add_close_phis_to_merge_points): Likewise.
31487 (copy_loop_close_phi_args): Likewise.
31488 (copy_cond_phi_nodes): Likewise.
31490 2017-03-20 Martin Liska <mliska@suse.cz>
31492 PR middle-end/79753
31493 * tree-chkp.c (chkp_build_returned_bound): Do not build
31494 returned bounds for a LHS that's not a BOUNDED_P type.
31496 2017-03-20 Martin Liska <mliska@suse.cz>
31500 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
31501 COMPLEX_CST and VECTOR_CST.
31503 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31506 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
31507 target operand. A new splitter adds the clobber statement in case
31508 the target operand is dead anyway.
31510 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
31512 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
31513 to age-old versions of binutils and glibc.
31515 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
31517 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
31519 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31521 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
31523 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31525 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
31526 requirement for binutils 2.13.
31528 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31530 * combine.c (try_combine): Delete redundant i1 test. Call
31531 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31533 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
31535 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
31536 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
31538 <riscv64-*-elf>: Re-arrange section
31539 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
31540 <riscv32-*-linux>: Likewise.
31541 <riscv64-*-elf>: Likewise
31542 <riscv64-*-linux>: Likewise.
31544 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
31547 * aarch64.opt(verbose-cost-dump): Fix typo.
31549 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
31552 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
31553 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
31555 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31557 * reload.c (find_reloads): When reloading a nonoffsettable address,
31558 use RELOAD_OTHER for it and its address reloads.
31560 PR rtl-optimization/79910
31561 * combine.c (record_used_regs): New static function.
31562 (try_combine): Handle situations where there is an additional
31563 instruction between I2 and I3 which needs to have a LOG_LINK
31566 2017-03-17 Jeff Law <law@redhat.com>
31568 PR tree-optimization/71437
31569 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
31570 conditional in the hash table first.
31571 (vrp_dom_walker::before_dom_children): Extract condition from
31572 ASSERT_EXPR. Record condition, its inverion and any implied
31573 conditions as well.
31575 2017-03-17 Marek Polacek <polacek@redhat.com>
31576 Markus Trippelsdorf <markus@trippelsdorf.de>
31578 PR tree-optimization/80079
31579 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
31582 2017-03-17 Richard Biener <rguenther@suse.de>
31584 PR middle-end/80075
31585 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
31586 Properly verify the LHS before the RHS possibly claims to be
31588 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
31591 2017-03-17 Martin Jambor <mjambor@suse.cz>
31593 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
31594 (List of -O2 options): Likewise.
31595 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
31598 2017-03-17 Tom de Vries <tom@codesourcery.com>
31600 * gcov-dump.c (print_usage): Print bug_report_url.
31602 2017-03-17 Richard Biener <rguenther@suse.de>
31604 PR middle-end/80050
31605 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
31606 (parser::peek): Likewise.
31608 2017-03-17 Richard Biener <rguenther@suse.de>
31610 PR tree-optimization/80048
31611 * sese.c (free_sese_info): Properly release rename_map and
31612 copied_bb_map elements.
31614 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
31616 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
31617 Add linked-list forward and backlinks. Insert on
31618 construction, remove on destruction.
31619 (class pass_store_merging): Add m_stores_head field.
31620 (pass_store_merging::terminate_and_process_all_chains):
31621 Iterate over m_stores_head list.
31622 (pass_store_merging::terminate_all_aliasing_chains):
31624 (pass_store_merging::execute): Check for debug stmts first.
31625 Push new chains onto the m_stores_head stack.
31627 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
31630 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
31631 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
31632 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
31634 2017-03-16 Jeff Law <law@redhat.com>
31636 PR tree-optimization/71437
31637 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
31638 member function. Implementation moved into after_dom_children
31639 member function and into the threader's thread_outgoing_edges
31641 (dom_opt_dom_walker::after_dom_children): Simplify by moving
31642 some code into new thread_outgoing_edges.
31643 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
31644 definition. Simplify marker handling (do it here). Assume we always
31645 have the available expression and the const/copies tables.
31646 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
31648 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
31649 * tree-vrp.c (equiv_stack): No longer file scoped.
31650 (vrp_dom_walker): New class.
31651 (vrp_dom_walker::before_dom_children): New member function.
31652 (vrp_dom_walker::after_dom_children): Likewise.
31653 (identify_jump_threads): Setup domwalker. Use it rather than
31654 walking edges in a random order by hand. Simplify setup/finalization.
31655 (finalize_jump_threads): Remove.
31656 (vrp_finalize): Do not call identify_jump_threads here.
31657 (execute_vrp): Do it here instead and call thread_through_all_blocks
31660 PR tree-optimization/71437
31661 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
31663 (simplify_stmt_for_jump_threading): Add basic_block argument. All
31665 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
31666 (dom_opt_dom_walker::thread_across_edge): Remove
31667 handle_dominating_asserts argument. All callers changed.
31668 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
31669 changes. Remove calls to lhs_of_dominating_assert. Other
31670 uses of handle_dominating_asserts turn into unconditional code
31671 (simplify_control_stmt_condition_1): Likewise.
31672 (simplify_control_stmt_condition): Likewise.
31673 (thread_through_normal_block, thread_across_edge): Likewise.
31674 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
31675 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
31676 object if it is not an SSA_NAME.
31677 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
31678 before calling into the VRP specific simplifiers.
31679 (identify_jump_threads): Remove handle_dominating_asserts
31682 2017-03-16 Jakub Jelinek <jakub@redhat.com>
31685 * tree-diagnostic.c (default_tree_printer): No longer static.
31686 * tree-diagnostic.h (default_tree_printer): New prototype.
31688 2017-03-16 Tamar Christina <tamar.christina@arm.com>
31690 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
31691 Change ins into fmov.
31693 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31695 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
31696 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
31697 Use h_con constraint for operand 1.
31698 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
31699 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
31701 2017-03-15 Jeff Law <law@redhat.com>
31703 PR tree-optimization/71437
31704 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
31705 (record_temporary_equivalences): Use it.
31707 PR tree-optimization/71437
31708 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
31709 tree-ssa-scopedtables.
31710 (lookup_avail_expr, build_and_record_new_cond): Likewise.
31711 (record_conditions, record_cond, vuse_eq): Likewise.
31712 (record_edge_info): Adjust to API tweak of record_conditions.
31713 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
31714 (record_temporary_equivalences, optimize_stmt): Likewise.
31715 (eliminate_redundant_computations): Likewise.
31716 (record_equivalences_from_stmt): Likewise.
31717 * tree-ssa-scopedtables.c: Include options.h and params.h.
31718 (vuse_eq): New function, moved from tree-ssa-dom.c
31719 (build_and_record_new_cond): Likewise.
31720 (record_conditions): Likewise. Accept vector of conditions rather
31721 than edge_equivalence structure for first argument.
31722 for the first argument.
31723 (avail_exprs_stack::lookup_avail_expr): New member function, moved
31724 from tree-ssa-dom.c.
31725 (avail_exprs_stack::record_cond): Likewise.
31726 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
31727 from tree-ssa-dom.c.
31728 (avail_exprs_stack): Add new member functions lookup_avail_expr
31730 (record_conditions): Declare.
31732 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
31735 * lra-constraints.c (process_alt_operands): Increase reject for
31736 reloading an input/output operand.
31738 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
31741 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
31742 insns to convert from signed/unsigned char/short to IEEE 128-bit
31744 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
31746 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
31749 * config/i386/i386.c (ix86_vector_duplicate_value): Create
31750 subreg of inner mode for values already in registers.
31752 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
31754 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
31755 iteration reg is used after the loop.
31757 2017-03-14 Martin Sebor <msebor@redhat.com>
31759 PR tree-optimization/79800
31760 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
31761 precision in negative-positive range.
31762 (format_floating): Call non-const overload with adjusted precision.
31764 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
31767 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
31770 2017-03-14 Martin Sebor <msebor@redhat.com>
31772 PR middle-end/80020
31773 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
31774 * builtins.def (aligned_alloc): Use it.
31777 * Makefile.in (GTFILES): Add calls.c.
31778 * calls.c: Include "gt-calls.h".
31780 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
31782 PR rtl-optimization/79728
31783 * regs.h (struct target_regs): New field
31784 x_contains_allocatable_regs_of_mode.
31785 (contains_allocatable_regs_of_mode): New macro.
31786 * reginfo.c (init_reg_sets_1): Initialize it, and change
31787 contains_reg_of_mode so it includes global regs as well.
31788 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
31789 rather than contains_regs_of_mode.
31791 2017-03-14 Martin Liska <mliska@suse.cz>
31793 * doc/invoke.texi: Document options that can't be combined with
31794 -fcheck-pointer-bounds.
31796 2017-03-14 Martin Liska <mliska@suse.cz>
31798 PR middle-end/79831
31799 * doc/invoke.texi (-Wchkp): Document the option.
31801 2017-03-14 Martin Liska <mliska@suse.cz>
31803 * Makefile.in: Install gcov-dump.
31805 2017-03-14 Martin Liska <mliska@suse.cz>
31807 * multiple_target.c (expand_target_clones): Bail out for
31808 an invalid attribute.
31810 2017-03-14 Richard Biener <rguenther@suse.de>
31812 * alias.c (struct alias_set_entry): Pack properly.
31813 * cfgloop.h (struct loop): Likewise.
31814 * cse.c (struct set): Likewise.
31815 * ipa-utils.c (struct searchc_env): Likewise.
31816 * loop-invariant.c (struct invariant): Likewise.
31817 * lra-remat.c (struct cand): Likewise.
31818 * recog.c (struct change_t): Likewise.
31819 * rtl.h (struct address_info): Likewise.
31820 * symbol-summary.h (function_summary): Likewise.
31821 * tree-loop-distribution.c (struct partition): Likewise.
31822 * tree-object-size.c (struct object_size_info): Likewise.
31823 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
31824 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
31825 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
31826 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
31827 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
31828 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
31829 (struct _stmt_vec_info): Likewise.
31831 2017-03-14 Martin Liska <mliska@suse.cz>
31834 * multiple_target.c (create_dispatcher_calls): Check that
31835 a target can create a function dispatcher.
31837 2017-03-14 Martin Liska <mliska@suse.cz>
31840 * multiple_target.c (expand_target_clones): Drop local.local
31841 flag for default implementation.
31843 2017-03-14 Richard Biener <rguenther@suse.de>
31845 PR tree-optimization/80030
31846 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
31848 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
31850 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
31851 gcc_fallthrough() instead of __attribute__((fallthrough));
31853 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31855 * doc/gcc.texi: Remove "up" link to (DIR).
31856 * doc/gccint.texi: Ditto.
31858 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31860 * doc/install.texi (Specific) <avr>: Remove reference to
31863 2017-03-13 Jeff Law <law@redhat.com>
31865 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
31866 attribute rather than comments.
31868 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
31869 match_scratch operand is highest.
31871 2017-03-13 Martin Liska <mliska@suse.cz>
31873 PR middle-end/78339
31874 * ipa-pure-const.c (warn_function_noreturn): If the declarations
31875 is a CHKP clone, use original declaration.
31877 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31879 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
31880 (arc_conditional_register_usage): Use a different allocation order
31881 when optimizing for size.
31882 * common/config/arc/arc-common.c (arc_option_optimization_table):
31883 Section anchors default on when optimizing for size.
31885 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31887 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
31889 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31891 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
31892 * config/arc/arc.md (cpu_facility): Add cd variant.
31893 (*movqi_insn): Add code density variant.
31894 (*movhi_insn): Likewise.
31895 (*movqi_insn): Likewise.
31896 (*addsi3_mixed): Likewise.
31897 (subsi3_insn): Likewise.
31899 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31901 * config/arc/arc.md (movsi_cond_exec): Update constraint.
31903 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31905 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
31906 expressions with MINUS and UNARY ops.
31908 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31911 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
31913 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
31914 between vec_select and vector argument.
31915 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
31916 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
31917 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
31918 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
31919 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
31920 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
31922 2017-03-13 Richard Biener <rguenther@suse.de>
31925 * params.def (vect-max-peeling-for-alignment): Fix typo.
31927 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31929 * doc/install.texi (Specific) <mips-*-*>: Remove description of
31930 issue that only occurred with binutils below 2.18.
31932 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31934 * doc/install.texi (Specific) <cris-axis-elf>: No longer
31935 refer to binutils 2.11/2.12 minimum.
31937 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31939 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
31940 ftp.kernel.org and simplify binutils requirement.
31942 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
31944 * doc/invoke.texi (Warning Options): Fix spelling of link-time
31946 (Optimize Options): Ditto. Also remove redundancy.
31948 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31950 PR translation/79848
31951 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
31953 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
31954 to G_ to avoid double translation.
31956 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31958 PR translation/79923
31959 * auto-profile.c (get_combined_location): Convert leading
31960 character of diagnostics to lower case and remove trailing period.
31961 (read_profile): Likewise for various diagnostics.
31962 * config/arm/arm.c (arm_option_override): Remove trailing period
31963 from various diagnostics.
31964 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
31965 (msp430_expand_delay_cycles): Likewise.
31967 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31970 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
31971 full command-line argument, rather than just "str".
31972 (aarch64_validate_march): Likewise.
31973 (aarch64_validate_mtune): Likewise.
31975 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
31977 PR rtl-optimization/78911
31978 * lra-assigns.c (must_not_spill_p): New function.
31979 (spill_for): Use it.
31981 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31983 PR tree-optimization/79981
31984 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
31985 ATOMIC_COMPARE_EXCHANGE ifn result.
31986 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
31987 IFN_ATOMIC_COMPARE_EXCHANGE.
31989 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31992 * opts.c (parse_sanitizer_options): Add missing question mark to
31993 "did you mean" message.
31995 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31997 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
31999 (VMULEUH_UNS): Likewise.
32000 (VMULOUB_UNS): Likewise.
32001 (VMULOUH_UNS): Likewise.
32002 * config/rs6000/rs6000.c (builtin_function_type): Remove
32003 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
32005 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32008 * read-rtl-function.c (function_reader::read_rtx_operand): Update
32009 x with result of extra_parsing_for_operand_code_0.
32010 (function_reader::extra_parsing_for_operand_code_0): Convert
32011 return type from void to rtx, returning x. When reading
32012 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
32013 larger size containing struct block_symbol.
32015 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
32017 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
32018 -mfloat128-hardware without -m64.
32020 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
32023 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
32024 entries to the case statement that marks unsigned arguments to
32025 overloaded functions.
32027 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32029 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
32030 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
32032 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
32035 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
32036 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
32038 2017-03-10 Martin Liska <mliska@suse.cz>
32042 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
32043 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
32046 2017-03-10 Olivier Hainque <hainque@adacore.com>
32048 * tree-switch-conversion (array_value_type): Start by resetting
32049 candidate type to it's main variant.
32051 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32053 PR rtl-optimization/79909
32054 * combine.c (try_combine): Use simplify_replace_rtx on individual
32055 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
32056 of the whole CALL_INSN_FUNCTION_USAGE.
32058 PR tree-optimization/79972
32059 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
32060 get_range_info on SSA_NAMEs. Formatting fixes.
32062 2017-03-10 Richard Biener <rguenther@suse.de>
32063 Jakub Jelinek <jakub@redhat.com>
32065 PR tree-optimization/77975
32066 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
32067 edge to be constant.
32068 (get_val_for): For constant x return it. Formatting fix.
32069 (loop_niter_by_eval): Avoid pointless looping if the next iteration
32070 would use the same bases as the current one.
32072 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32074 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
32075 instead of vec_select for V1TImode.
32076 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
32078 (VSX_LE_128): Add V1TI to this mode iterator.
32079 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
32080 (*vsx_le_perm_store_<mode>): Likewise.
32081 (pre-reload splitter for VSX stores): Likewise.
32082 (post-reload splitter for VSX stores): Likewise.
32083 (*vsx_xxpermdi2_le_<mode>): Likewise.
32084 (*vsx_lxvd2x2_le_<mode>): Likewise.
32085 (*vsx_stxvd2x2_le_<mode>): Likewise.
32087 2017-03-09 Michael Eager <eager@eagercon.com>
32089 Correct failures with --enable-checking=yes,rtl.
32091 * config/microblaze/microblaze.c (microblaze_expand_shift):
32092 Replace GET_CODE test with CONST_INT_P and INTVAL test with
32093 test for const0_rtx.
32094 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
32095 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
32097 2017-03-09 Richard Biener <rguenther@suse.de>
32099 PR tree-optimization/79977
32100 * graphite-scop-detection.c (scop_detection::merge_sese):
32101 Handle the case of extra exits to blocks dominating the entry.
32103 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
32105 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
32108 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
32110 PR rtl-optimization/79949
32111 * lra-constraints.c (process_alt_operands): Check memory when
32112 trying to predict a cycle. Print about the overall increase.
32114 2017-03-09 Richard Biener <rguenther@suse.de>
32116 PR middle-end/79971
32117 * gimple-expr.c (useless_type_conversion_p): Preserve
32118 TYPE_SATURATING for fixed-point types.
32120 2017-03-09 Richard Biener <rguenther@suse.de>
32123 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
32124 alignment of BLKmode params.
32126 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32129 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
32130 (VALL_NO_V2Q): Likewise.
32132 * config/aarch64/aarch64-simd.md
32133 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
32135 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
32136 VALL_NO_V2Q mode iterator.
32137 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
32139 2017-03-09 Martin Liska <mliska@suse.cz>
32141 PR tree-optimization/79631
32142 * tree-chkp-opt.c (chkp_is_constant_addr): Call
32143 tree_int_cst_sign_bit just for INTEGER constants.
32145 2017-03-09 Martin Liska <mliska@suse.cz>
32149 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
32152 2017-03-09 Marek Polacek <polacek@redhat.com>
32155 * tree.c (inchash::add_expr): Handle TREE_VEC.
32157 2017-03-09 Martin Liska <mliska@suse.cz>
32160 (chkp_narrow_size_and_offset): New function.
32161 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
32162 (void chkp_parse_bit_field_ref): New function.
32163 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
32164 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
32166 2017-03-09 Martin Liska <mliska@suse.cz>
32169 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
32170 (chkp_find_bounds_1): Remove gcc_unreachable.
32172 2017-03-09 Jakub Jelinek <jakub@redhat.com>
32175 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
32176 BUILT_IN_SYNC*, determine the access type from the size suffix and
32177 always build a MEM_REF with that type. Handle forgotten
32178 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
32181 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
32182 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
32183 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
32184 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
32185 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
32186 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
32187 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
32188 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
32189 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
32190 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
32191 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
32192 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
32193 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
32194 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
32195 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
32196 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
32197 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
32198 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
32199 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
32200 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
32201 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
32202 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
32203 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
32204 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
32205 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
32206 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
32207 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
32208 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
32209 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
32210 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
32211 definitions outside of __OPTIMIZE__ guarded section.
32214 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
32215 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
32216 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
32217 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
32220 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32222 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
32223 ("vfenez<mode>"): Add missing constraints.
32225 2017-03-08 Martin Sebor <msebor@redhat.com>
32228 * config/nds32/nds32.c (nds32_option_override):
32229 Fix misspelled diagnostic.
32231 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32234 * gimplify.c (gimplify_omp_for): Replace index var in outer
32235 taskloop statement with an artificial variable and add
32236 OMP_CLAUSE_PRIVATE clause for it.
32238 2017-03-08 Richard Biener <rguenther@suse.de>
32240 PR tree-optimization/79955
32241 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
32242 for accesses that are completely outside of the variable.
32244 2017-03-08 Andrew Haley <aph@redhat.com>
32246 PR tree-optimization/79943
32247 * tree-ssa-loop-split.c (compute_new_first_bound): When
32248 calculating the new upper bound, (END-BEG) should be added, not
32251 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32253 * config/avr/avr.md (setmemhi): Make sure match_dup
32254 operand number comes before match_scratch.
32256 2017-03-08 Richard Biener <rguenther@suse.de>
32258 PR tree-optimization/79920
32259 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
32260 with ncopies == 1 to ...
32261 (vect_transform_slp_perm_load): ... here. Properly compute
32262 all element loads by iterating VF times over the group. Do
32263 not handle ncopies (computed in a broken way) in
32264 vect_create_mask_and_perm.
32266 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32269 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
32270 is a uniform vector, use uniform_vector_p return value instead of
32271 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
32273 2017-03-07 Marek Polacek <polacek@redhat.com>
32275 PR middle-end/79809
32276 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
32277 (alloca_call_type): Likewise.
32279 2017-03-07 Martin Liska <mliska@suse.cz>
32281 * gcov.c (process_args): Put comment to correct location.
32283 2017-03-07 Martin Liska <mliska@suse.cz>
32285 PR middle-end/68270
32286 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
32287 Use array_at_struct_end_p instead of DECL_CHAIN (field).
32288 (chkp_narrow_bounds_for_field): Likewise.
32289 (chkp_parse_array_and_component_ref): Pass one more argument to
32292 2017-03-07 Richard Biener <rguenther@suse.de>
32294 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
32297 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
32299 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
32300 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
32302 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32305 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
32306 to end of description.
32307 (PARAM_MAX_STORES_TO_MERGE): Likewise.
32309 2017-03-07 Jakub Jelinek <jakub@redhat.com>
32311 PR rtl-optimization/79901
32312 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
32314 (*avx512f_<code><mode>3<mask_name>): ... this.
32315 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
32316 iterator instead of VI8_AVX2_AVX512BW.
32318 PR rtl-optimization/79901
32319 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
32320 min/max expander, expand it using expand_vec_cond_expr.
32323 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
32326 2017-03-06 Jakub Jelinek <jakub@redhat.com>
32329 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
32330 to void * for PCH reasons.
32331 * dwarf2out.c (output_loc_operands, output_die): Cast
32332 v.val_vec.array to unsigned char *.
32334 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
32337 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
32340 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
32342 PR rtl-optimization/79571
32343 * lra-constraints.c (process_alt_operands): Calculate static
32344 reject and subtract it from overall when only addresses will be
32347 2017-03-06 Julia Koval <julia.koval@intel.com>
32350 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
32351 incoming stack boundary to 128 for 64-bit targets.
32353 2017-03-06 Richard Biener <rguenther@suse.de>
32355 PR tree-optimization/79894
32356 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
32357 to NULL after folding it.
32359 2017-03-06 Richard Biener <rguenther@suse.de>
32361 PR tree-optimization/79824
32362 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
32363 check disabling peeling for gaps.
32365 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
32367 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
32368 attributes): Document gettimeofday.
32370 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
32372 * config/s390/s390.c (s390_option_override_internal): Set
32373 PARAM_MIN_VECT_LOOP_BOUND
32375 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
32377 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
32378 * config/s390/s390.md: Likewise.
32380 2017-03-06 Jakub Jelinek <jakub@redhat.com>
32383 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
32384 (<avx2_avx512>_perm<mode>): Rename to ...
32385 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
32387 (<avx512>_perm<mode>_mask): Rename to ...
32388 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
32390 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
32391 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
32392 instead of VI8F_256_512.
32393 (avx512f_perm<mode>): New define_expand.
32394 (avx512f_perm<mode>_mask): Likewise.
32395 (avx512f_perm<mode>_1<mask_name>): New define_insn.
32396 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
32398 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32400 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
32401 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
32403 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
32405 2017-03-06 Martin Liska <mliska@suse.cz>
32408 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
32409 when having a SSA NAME w/o VAR_DECL assigned to it.
32411 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32413 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
32414 msa_dpsub_<su>_d): Fix MODE for vec_select.
32416 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32418 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
32420 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
32422 2017-03-06 Richard Biener <rguenther@suse.de>
32424 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
32425 * plugin.c (register_plugin_info): Likewise.
32426 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
32428 2017-03-05 Jakub Jelinek <jakub@redhat.com>
32430 * config/i386/sse.md (sse_storehps, sse_storelps,
32431 avx_<castmode><avxsizesuffix>_<castmode>,
32432 avx512f_<castmode><avxsizesuffix>_<castmode>,
32433 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
32434 in condition that at least one operand is not a MEM.
32436 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32438 PR middle-end/79805
32439 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
32440 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
32442 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
32443 gimple_call_nothrow_p flag based on whether original builtin can throw.
32444 If it can, emit following stmts on the fallthrough edge.
32445 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
32446 don't create new bb if inserting just debug stmts on the edge, try to
32447 insert them on the fallthru bb or just reset debug stmts.
32449 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
32452 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
32453 restore recog_data (including the operand rtxes inside it) around
32454 the call to get_insn_template.
32456 2017-03-03 Martin Sebor <msebor@redhat.com>
32458 PR tree-optimization/79699
32459 * context.c (context::~context): Free MPFR caches to avoid
32460 a memory leak on program exit.
32462 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32464 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
32465 Use wide_int::ulow () instead of .elt (0).
32467 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32469 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
32470 (*pushxf): Limit oF constraint to 32bit targets and add oC
32471 constraint for 64bit targets.
32472 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
32473 (*pushdf): Change rmF constraint to rmC.
32475 2017-03-03 Martin Liska <mliska@suse.cz>
32477 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
32478 Remove unused variable.
32480 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32483 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
32484 is a memory operand, increase num_memory.
32485 (ix86_expand_args_builtin): Likewise.
32487 2017-03-03 Jan Hubicka <jh@suse.cz>
32490 * ipa-devirt.c (maybe_record_node): Properly handle
32491 __cxa_pure_virtual visibility.
32493 2017-03-03 Martin Liska <mliska@suse.cz>
32495 PR tree-optimization/79803
32496 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
32498 (pass_loop_prefetch::execute): Disabled optimization if an
32499 assumption about L1 cache size is not met.
32501 2017-03-03 Martin Liska <mliska@suse.cz>
32503 PR rtl-optimization/79574
32504 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
32505 (hash_scan_set): Likewise.
32506 (dump_hash_table): Likewise.
32507 (hoist_code): Likewise.
32509 2017-03-03 Richard Biener <rguenther@suse.de>
32511 * fixed-value.c (fixed_from_string): Restore use of elt (1)
32512 in place of uhigh ().
32513 (fixed_convert_from_real): Likewise.
32515 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32518 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
32520 2017-03-03 Richard Biener <rguenther@suse.de>
32522 PR middle-end/79818
32523 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
32524 TYPE_OVERFLOW_UNDEFINED check.
32526 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32528 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
32530 (vector_ae_<mode>_p): Likewise.
32531 (vector_nez_<mode>_p): Likewise.
32532 (vector_ne_v2di_p): Likewise.
32533 (vector_ae_v2di_p): Likewise.
32534 (vector_ne_<mode>_p): Likewise.
32535 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
32537 (vsx_tsqrt<mode>2_fe): Likewise.
32539 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
32542 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
32544 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32546 PR rtl-optimization/79780
32547 * cprop.c (one_cprop_pass): When second and further conditional trap
32548 in a single basic block is turned into an unconditional trap, turn it
32549 into a deleted note to avoid RTL verification failures.
32551 2017-03-02 Richard Biener <rguenther@suse.de>
32553 * fold-const.c (const_binop): Use ulow () instead of elt (0).
32555 2017-03-02 Richard Biener <rguenther@suse.de>
32557 PR tree-optimization/79345
32559 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
32560 param and abort the walk, returning -1 if it is hit.
32561 (walk_aliased_vdefs): Take a limit param and pass it on.
32562 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
32563 defaulting to 0 and return a signed int.
32564 * tree-ssa-uninit.c (struct check_defs_data): New struct.
32565 (check_defs): New helper.
32566 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
32567 about uninitialized memory.
32568 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
32569 bogus uninitialized warning.
32570 (fixed_convert_from_real): Likewise.
32572 2017-03-02 Bin Cheng <bin.cheng@arm.com>
32574 PR tree-optimization/66768
32575 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
32576 iv_use if base object can't be determined.
32578 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32580 PR tree-optimization/79345
32581 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
32582 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
32583 (get_pattern_stats): Initialize it.
32584 * genemit.c (gen_expand): Verify match_scratch numbers come after
32585 match_operand/match_dup numbers.
32586 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
32587 match_scratch numbers.
32588 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
32590 * config/s390/s390.md (trunctdsd2): Likewise.
32592 2017-03-02 Richard Biener <rguenther@suse.de>
32594 * wide-int.h (wide_int_storage::operator=): Implement in terms
32597 2017-03-02 Richard Biener <rguenther@suse.de>
32599 PR tree-optimization/79777
32600 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
32601 the to insert expression to sth existing.
32603 2017-03-01 Martin Sebor <msebor@redhat.com>
32605 PR middle-end/79692
32606 * gimple-ssa-sprintf.c
32607 (directive::known_width_and_precision): New function.
32608 (format_integer): Use it.
32609 (get_mpfr_format_length): Consider the full range of precision
32610 when computing %g output with the # flag. Set the likely byte
32611 count to 3 rather than 1 when precision is indeterminate.
32612 (format_floating): Correct the lower bound of precision.
32614 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32616 * doc/invoke.texi: Document default code model for 64-bit Linux.
32618 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32621 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
32622 udiv rather than div since input pattern is unsigned.
32624 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
32626 * config/i386/i386.c (print_reg): Warn for values of
32627 unsupported size in integer register.
32629 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
32632 * config/rs6000/predicates.md (current_file_function_operand): Do
32633 not allow self calls to be local if the function is replaceable.
32635 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
32638 * config/rs6000/altivec.h (vec_ctz and others): Change the
32639 preprocessor macro that controls conditional compilation from
32640 _ARCH_PWR9 to __POWER9_VECTOR__.
32641 (vec_all_ne): Change parameterization of __altivec_scalar_pred
32642 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
32643 control (instead of _ARCH_PWR9 control) so that template
32644 definition uses power9-specific function.
32645 (vec_any_eq): Likewise.
32646 (vec_all_ne): Change macro definition to use a power9-specific
32647 expansion under #ifdef __POWER9_VECTOR__ control (instead of
32648 _ARCH_PWR9 control).
32649 (vec_any_eq) Likewise.
32650 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
32651 expansion for CMPNEF to remove support for xvcmpnesp instruction.
32652 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
32653 support for xvcmpnedp instruction.
32654 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
32655 macro expansion so that Power9 implementation of vec_all_ne does
32656 not use the AltiVec predicate framework.
32657 (VCMPNEH_P): Likewise.
32658 (VCMPNEW_P): Likewise.
32659 (VCMPNED_P): Likewise.
32660 (VCMPNEFP_P): Likewise.
32661 (VCMPNEDP_P): Likewise.
32662 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
32663 implementation of vec_any_eq to not use AltiVec predicate
32665 (VCMPAEH_P): Likewise.
32666 (VCMPAEW_P): Likewise.
32667 (VCMPAED_P): Likewise.
32668 (VCMPAEFP_P): Likewise.
32669 (VCMPAEDP_P): Likewise.
32670 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
32671 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
32672 not use the AltiVec predicate framework.
32673 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
32674 of vec_any_eq to not use AltiVec predicate framework.
32675 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32676 support for predefined __POWER9_VECTOR__ macro to indicate that
32677 Power9 instruction selection is enabled.
32678 (altivec_overloaded_builtins): Remove extraneous
32679 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
32680 function argument types RS6000_BTI_bool_V16QI and
32681 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
32682 entry for overloaded function argument types RS6000_BTI_bool_V4SI
32683 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
32684 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
32685 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
32686 Power9 for implementations of vec_cmpne. Change the signature for
32687 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
32688 (representing vec_all_ne) to remove the previously described first
32689 argument of type RS6000_BTI_INTSI, as this was an artifact of
32690 reliance on the AltiVec predicate framework, which is no longer
32691 used in the implementation of these functions. Add
32692 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
32693 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
32694 since, unlike the AltiVec predicate framework implementation, we
32695 do not share function descriptors between vec_alle and vec_anyeq.
32696 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
32697 set of modes that receive special treatment even when
32698 TARGET_P9_VECTOR is true. The special treatment emits code that
32699 does not depend on Power9 instructions.
32700 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
32701 define_expand to not rely on AltiVec predicate framework.
32702 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
32704 (vector_ne_v2di_p): Change this define_expand to not rely on
32705 AltiVec predicate framework.
32706 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
32708 (vector_ne_<mode>_p): Change this define_expand to not rely on
32709 AltiVec predicate framework.
32710 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
32712 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
32713 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
32714 define_insn pattern.
32715 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
32716 define_insn pattern because the xvcmpne<VSs>. instruction is not
32718 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
32719 instruction is not supported.
32721 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32723 * config/nvptx/nvptx.c: Include intl.h.
32725 2017-03-01 Martin Jambor <mjambor@suse.cz>
32728 * ipa-prop.h (ipa_bits): Removed field known.
32729 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
32730 to pointers. Adjusted their comments to warn about their sharing.
32731 (ipcp_transformation_summary): Change bits to a vector of pointers.
32732 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
32733 (ipa_get_ipa_bits_for_value): Declare.
32734 * tree-vrp.h (value_range): Mark as GTY((for_user)).
32735 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
32736 (ipa_bits_hash_table): Likewise.
32737 (ipa_vr_ggc_hash_traits): Likewise.
32738 (ipa_vr_hash_table): Likewise.
32739 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
32740 being pointers and vr_known being removed.
32741 (ipa_set_jf_unknown): Likewise.
32742 (ipa_get_ipa_bits_for_value): New function.
32743 (ipa_set_jfunc_bits): Likewise.
32744 (ipa_get_value_range): New overloaded functions.
32745 (ipa_set_jfunc_vr): Likewise.
32746 (ipa_compute_jump_functions_for_edge): Use the above functions to
32747 construct bits and vr parts of jump functions.
32748 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
32749 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
32751 (ipcp_grow_transformations_if_necessary): Also allocate
32752 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
32754 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
32755 them. Fix too long lines.
32756 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
32757 vr_known being removed.
32758 (ipa_read_jump_function): Use new setter functions to construct bits
32759 and vr parts of jump functions or set them to NULL.
32760 (write_ipcp_transformation_info): Adjust for bits being pointers.
32761 (read_ipcp_transformation_info): Likewise.
32762 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
32764 Include gt-ipa-prop.h.
32765 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
32767 (ipcp_store_bits_results): Likewise.
32768 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
32769 Do not write to existing jump functions but use a temporary instead.
32771 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32774 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
32775 attempt to use its first operand as BIT_FIELD_REF base.
32777 2017-03-01 Richard Biener <rguenther@suse.de>
32779 PR middle-end/79721
32780 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
32781 interpolating formula in wrapping arithmetic.
32782 (chrec_apply): Convert chrec_evaluate return value to wanted type.
32784 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32786 PR tree-optimization/79734
32787 * tree-vect-generic.c (expand_vector_condition): Optimize
32788 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
32789 Handle VEC_COND_EXPR where comparison has different inner width from
32790 type's inner width.
32792 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
32794 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
32795 markup, and similar issues. Remove @opindex entries for things
32796 that aren't options. Add missing -mmpy-option entries.
32798 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32800 PR tree-optimization/79737
32801 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
32802 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
32803 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
32804 instead of byte_size. Formatting fix.
32805 (shift_bytes_in_array_right): Formatting fix.
32807 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
32810 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
32811 condition on optimize for the leaf function test.
32813 2017-02-28 Martin Liska <mliska@suse.cz>
32816 * read-rtl-function.c (function_reader::handle_unknown_directive):
32817 Bail out when one uses -flto.
32819 2017-02-28 Martin Liska <mliska@suse.cz>
32821 * common.opt: Replace space with tabular for options of <number>
32823 * config/i386/i386.opt: Show <number> value for
32824 -mlarge-data-threshold.
32825 * opts.c (print_filtered_help): Do not display number in hexadecimal
32828 2017-02-28 Martin Liska <mliska@suse.cz>
32830 * common.opt: Fix --help=option -Q for options which are of
32833 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
32835 * config/i386/i386.c (print_reg): Error out for values
32836 of 8-bit size in invalid integer register.
32838 2017-02-28 Martin Sebor <msebor@redhat.com>
32840 PR tree-optimization/79691
32841 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
32843 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32846 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
32847 gcc_unreachable with output_operand_lossage.
32849 2017-02-28 Richard Biener <rguenther@suse.de>
32851 PR tree-optimization/79740
32852 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
32854 (visit_nary_op): Insert the nary into the hashtable if we
32855 pattern-matched sth.
32856 * tree-ssa-pre.c (eliminate_insert): Robustify.
32858 2017-02-28 Richard Biener <rguenther@suse.de>
32860 PR middle-end/79731
32861 * fold-const.c (decode_field_reference): Reject out-of-bound
32864 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32866 * config/i386/i386.c: Include intl.h.
32867 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
32868 instead of just cond ? "..." : "...".
32869 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
32870 * coverage.c (read_counts_file): Likewise.
32871 * omp-offload.c: Include intl.h.
32872 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
32873 of just cond ? "..." : "...".
32874 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
32875 of just cond ? "..." : "...".
32877 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
32880 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
32882 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
32883 'tune for' CPU name.
32884 * config/arm/arm-cpu-data.h: Regenerated.
32886 2017-02-28 Richard Biener <rguenther@suse.de>
32888 PR tree-optimization/79732
32889 * tree-inline.c (expand_call_inline): Do not shadow var.
32891 2017-02-28 Richard Biener <rguenther@suse.de>
32893 PR tree-optimization/79723
32894 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
32895 address-space properly.
32897 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
32899 * doc/optinfo.texi (Optimization groups): Fix option used for
32901 * doc/invoke.texi (-fopt-info): Document "omp".
32902 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
32903 (OPTGROUP_ALL): Add OPTGROUP_OMP.
32904 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
32905 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
32906 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
32908 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
32910 * dumpfile.c (optgroup_options): Instead of "openmp", associate
32911 OPTGROUP_OMP with "omp".
32913 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
32916 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
32917 for arithmetic shift of unsigned V2DI.
32919 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
32921 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
32922 arc/linux.h headers.
32923 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
32924 (LINK_SPEC): Likewise.
32925 (ARC_TLS_EXTRA_START_SPEC): Likewise.
32926 (EXTRA_SPECS): Likewise.
32927 (STARTFILE_SPEC): Likewise.
32928 (ENDFILE_SPEC): Likewise.
32929 (LIB_SPEC): Likewise.
32930 (TARGET_SDATA_DEFAULT): Likewise.
32931 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
32932 (MULTILIB_DEFAULTS): Likewise.
32933 (DWARF2_UNWIND_INFO): Likewise.
32934 * config/arc/big.h: New file.
32935 * config/arc/elf.h: Likewise.
32936 * config/arc/linux.h: Likewise.
32937 * config/arc/t-uClibc: Remove.
32939 2017-02-27 Bin Cheng <bin.cheng@arm.com>
32941 PR tree-optimization/77536
32942 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
32943 (tree_transform_and_unroll_loop): Use above function to compute the
32944 estimated niter of unrolled loop and use it when scaling profile.
32945 Also use count info rather than frequency if it's non-zero.
32946 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
32947 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
32948 (vect_transform_loop): Call above function.
32950 2017-02-27 Richard Biener <rguenther@suse.de>
32952 PR tree-optimization/45397
32953 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
32954 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
32955 (visit_nary_op): Add pattern matching for CSEing sign-changed
32956 or truncated operations with wider ones.
32958 2017-02-27 Richard Biener <rguenther@suse.de>
32960 PR tree-optimization/79690
32961 * tree-vect-stmts.c (vectorizable_store): Use vector type
32962 built from the DR with address-space.
32964 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
32966 * doc/invoke.texi (Optimize Options): Refine the description
32967 of asan-use-after-return.
32969 2017-02-25 Alan Modra <amodra@gmail.com>
32971 PR rtl-optimization/79584
32972 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
32973 base, not ad->base_term, the reg within base. Remove assertion
32974 that ad->base == ad->base_term. Replace gen_int_mode using
32975 bogus mode with const0_rtx.
32977 2017-02-25 Jakub Jelinek <jakub@redhat.com>
32979 PR middle-end/79396
32980 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
32981 FMA_EXPR like tcc_binary or tcc_unary.
32983 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
32986 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
32988 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
32989 (output_loc_operands): Handle DW_OP_call_ref and
32990 DW_OP_GNU_variable_value.
32991 (struct variable_value_struct): New type.
32992 (struct variable_value_hasher): Likewise.
32993 (variable_value_hash): New variable.
32994 (string_types): Remove.
32995 (copy_loc_descr): New function.
32996 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
32997 (prepend_loc_descr_to_each): New function.
32998 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
32999 instead of add_loc_descr_to_each if the first argument is single
33000 location list and the second has multiple.
33001 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
33002 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
33003 when looking for variable value which doesn't have other location info.
33004 (loc_list_from_tree): Formatting fix.
33005 (gen_array_type_die): Simplify DW_AT_string_length handling.
33006 (adjust_string_types): Remove.
33007 (gen_subprogram_die): Don't call adjust_string_types nor test/set
33008 string_types. Call resolve_variable_values.
33009 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
33010 (resolve_addr_in_expr): Likewise. Add A argument.
33011 (copy_deref_exprloc): Remove deref argument. Adjust for the
33012 original expression being DW_OP_GNU_variable_value with optionally
33013 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
33014 optionally after it.
33015 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
33016 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
33017 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
33018 (variable_value_hasher::hash, variable_value_hasher::equal): New
33020 (resolve_variable_value_in_expr, resolve_variable_value,
33021 resolve_variable_values, note_variable_value_in_expr,
33022 note_variable_value): New functions.
33023 (dwarf2out_early_finish): Call note_variable_value on all toplevel
33026 2017-02-24 Jakub Jelinek <jakub@redhat.com>
33029 * opts.h (handle_generated_option): Add GENERATED_P argument.
33030 * opts-common.c (handle_option): Adjust function comment.
33031 (handle_generated_option): Add GENERATED_P argument, pass it to
33033 (control_warning_option): Pass false to handle_generated_option
33035 * opts.c (maybe_default_option): Pass true to handle_generated_option
33037 * optc-gen.awk: Likewise.
33039 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33041 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
33042 a REG, look at the REG it is a SUBREG of.
33043 (splitter for cmpeqsi_t): Ditto.
33045 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33047 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
33048 the special USEs with the pattern of the insn, not the insn itself.
33050 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
33053 * doc/invoke.texi: Document -mload-store-pairs.
33055 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33056 Sandra Loosemore <sandra@codesourcery.com>
33058 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
33059 argument isn't a CONST_INT.
33060 (nios2_alternate_compare_const): Assert op is a CONST_INT.
33061 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
33062 (nios2_validate_compare): Bypass alternate compare logic if *op2
33063 is not a CONST_INT.
33064 (ldstwm_operation_p): Return false if first_base is not a REG or
33065 if first_offset is not a CONST_INT.
33067 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33069 * config/cris/cris.md: Use correct operand in a define_peephole2.
33071 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33073 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
33075 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33077 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
33078 this_insn if it is an INSN or JUMP_INSN.
33079 (force_offsettable): Look at base, not at addr.
33080 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
33081 on things that aren't necessarily CONST_INTs.
33083 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
33085 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
33086 -mfpmath=sse is the default also for x86-32 targets with SSE2
33087 instruction set when @option{-ffast-math} is enabled
33089 2017-02-24 Jeff Law <law@redhat.com>
33091 PR rtl-optimizatoin/79286
33092 * ira.c (update_equiv_regs): Drop may_trap_p exception to
33095 2017-02-24 Richard Biener <rguenther@suse.de>
33097 PR tree-optimization/79389
33098 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
33101 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
33103 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
33104 function comment to reflect reality.
33105 (loop_exits_before_overflow): Fix typo in function description.
33107 2017-02-24 Richard Biener <rguenther@suse.de>
33109 PR tree-optimization/79389
33110 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
33111 properly that a threading opportunity exists. Detect conditional
33112 copy/constant propagation opportunities.
33114 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
33116 * config/visium/visium.md (type): Add trap.
33117 (b): New mode attribute.
33118 (*btst): Rename into...
33119 (*btst<mode>): ...this and adjust.
33120 (*cbranchsi4_btst_insn): Rename into...
33121 (*cbranch<mode>4_btst_insn): ...this and adjust.
33122 (trap): New define_insn.
33124 2017-02-23 Jakub Jelinek <jakub@redhat.com>
33126 PR tree-optimization/79389
33127 * ifcvt.c (struct noce_if_info): Add rev_cond field.
33128 (noce_reversed_cond_code): New function.
33129 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
33130 reversed_comparison_code. Formatting fix.
33131 (noce_try_store_flag): Test rev_cond != NULL in addition to
33132 reversed_comparison_code.
33133 (noce_try_store_flag_constants): Likewise.
33134 (noce_try_store_flag_mask): Likewise.
33135 (noce_try_addcc): Use rev_cond if non-NULL instead of
33136 reversed_comparison_code.
33137 (noce_try_cmove_arith): Likewise. Formatting fixes.
33138 (noce_try_minmax, noce_try_abs): Clear rev_cond.
33139 (noce_find_if_block): Initialize rev_cond.
33140 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
33141 instead of false as last argument never attempt to reverse it
33144 2017-02-23 Bin Cheng <bin.cheng@arm.com>
33146 PR tree-optimization/79663
33147 * tree-predcom.c (combine_chains): Process refs in reverse order
33148 only for ZERO length chains, and add explaining comment.
33150 2017-02-23 Jeff Law <law@redhat.com>
33152 PR tree-optimization/79578
33153 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
33154 in call to operand_equal_p.
33156 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
33159 * config/i386/cpuid.h: Fix another undefined behavior.
33161 2017-02-23 Richard Biener <rguenther@suse.de>
33163 PR tree-optimization/79683
33164 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
33165 vector types for data-refs.
33167 2017-02-23 Martin Liska <mliska@suse.cz>
33169 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
33171 2017-02-23 Jakub Jelinek <jakub@redhat.com>
33173 PR middle-end/79665
33174 * internal-fn.c (get_range_pos_neg): Moved to ...
33175 * tree.c (get_range_pos_neg): ... here. No longer static.
33176 * tree.h (get_range_pos_neg): New prototype.
33177 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
33178 are known to be in between 0 and signed maximum inclusive, try to
33179 expand both unsigned and signed divmod and use the cheaper one from
33182 2017-02-22 Jeff Law <law@redhat.com>
33184 PR tree-optimization/79578
33185 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
33186 to compare base operands.
33188 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
33191 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
33192 gpc_reg_operand instead of fpr_reg_operand.
33194 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
33196 * config/mips/mips.c (mips_return_in_memory): Force FP
33197 vector types to be returned in memory for o32 ABI.
33199 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33201 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
33202 instead of DW_TAG_member for static data member declarations and don't
33203 set no_linkage_name for static inline data members.
33204 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
33207 2017-02-22 Martin Liska <mliska@suse.cz>
33209 * doc/invoke.texi: Replace inequality signs with square brackets
33212 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
33215 * lra-constraints.c (simplify_operand_subreg): Handle
33216 WORD_REGISTER_OPERATIONS targets.
33218 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33221 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
33222 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
33223 elimination by swapping fld*.
33225 2017-02-22 Richard Biener <rguenther@suse.de>
33227 PR tree-optimization/79673
33228 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
33229 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
33230 irrelevant address-space qualifiers and avoiding a
33231 ADDR_SPACE_CONVERT_EXPR from fold_convert.
33233 2017-02-22 Richard Biener <rguenther@suse.de>
33235 PR tree-optimization/79666
33236 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
33237 to not symbolically negate if that may introduce undefined
33240 2017-02-22 Martin Liska <mliska@suse.cz>
33243 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
33244 * data-streamer-out.c (streamer_write_gcov_count_stream):
33246 * value-prof.c (stream_out_histogram_value): Make assert more
33247 precise based on type of counter.
33249 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
33252 * config/i386/i386.md (standard_x87sse_constant_load splitter):
33253 Use nonimmediate_operand instead of memory_operand for operand 1.
33254 (float-extend standard_x87sse_constant_load splitter): Ditto.
33256 2017-02-21 Jeff Law <law@redhat.com>
33258 PR tree-optimization/79621
33259 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
33260 blocks with edges to themselves.
33262 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33265 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
33266 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
33267 Use gimple_call_builtin_p.
33270 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
33271 on temporarily removed DEBUG_INSNs.
33273 PR tree-optimization/79649
33274 * tree-loop-distribution.c (classify_partition): Give up on
33275 non-generic address space loads/stores.
33277 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
33279 * doc/loop.texi (Loop manipulation): Remove nonexistent
33280 tree_ssa_loop_version from the documentation.
33281 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
33283 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33286 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
33287 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
33288 * config/rs6000/rs6000.c: Include except.h.
33289 (rs6000_expand_split_stack_prologue): Call
33290 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
33292 2017-02-21 Martin Jambor <mjambor@suse.cz>
33295 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
33296 have been analyzed.
33298 2017-02-21 Martin Jambor <mjambor@suse.cz>
33300 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
33301 for backward compatibility only.
33302 * doc/invoke.texi (Option Summary): Remove all references to
33303 -fipa-cp-alignment.
33305 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
33309 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33311 * lra-constraints.c (curr_insn_transform): Handle
33312 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
33314 2017-02-21 Martin Liska <mliska@suse.cz>
33316 * config/i386/i386.opt: Replace -masm-dialect with -masm.
33318 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
33320 PR translation/79638
33321 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
33323 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
33326 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
33327 (arm_function_ok_for_sibcall): Return false for an indirect call by
33328 descriptor if all the argument registers are used.
33329 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
33330 alignment of the function.
33332 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33334 PR tree-optimization/61441
33335 * simplify-rtx.c (simplify_const_unary_operation): For
33336 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
33337 the sNaN unmodified.
33339 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33341 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33342 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33343 instead of SYSTEM_HEADER_DIR.
33345 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
33346 Martin Liška <mliska@suse.cz>
33348 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
33349 Fix typos and grammar, use active voice, and clarify.
33351 2017-02-20 Marek Polacek <polacek@redhat.com>
33353 PR middle-end/79537
33354 * gimplify.c (gimplify_expr): Handle unused *&&L;.
33357 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
33359 2017-02-20 Jakub Jelinek <jakub@redhat.com>
33362 * config/i386/i386.c (ix86_expand_builtin): Handle
33363 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
33364 ix86_builtins_isa[fcode].isa as a requirement of those
33365 flags and any other flag in the bitmask.
33366 (ix86_init_mmx_sse_builtins): Use 0 instead of
33367 ~OPTION_MASK_ISA_64BIT as mask.
33368 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
33369 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
33370 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
33371 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
33373 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33376 * lra-constraints.c (split_reg): Check requested split mode
33377 is supported by the register.
33379 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33381 * lra-constraints.c (simplify_operand_subreg): Remove early
33384 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33387 * lra-constraints.c (curr_insn_transform): Tighten condition
33388 for converting SUBREG reloads from OP_OUT to OP_INOUT.
33390 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33393 * lra-constraints.c (curr_insn_transform): Handle
33394 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
33396 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
33399 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
33401 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
33403 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
33406 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
33409 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33411 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
33414 2017-02-18 Jakub Jelinek <jakub@redhat.com>
33416 * final.c (last_columnnum, override_columnnum): New variables.
33417 (final_start_function): Set last_columnnum, pass it to begin_prologue
33418 hook and pass 0 to dwarf2out_begin_prologue.
33419 (final_scan_insn): Update override_columnnum. Pass last_columnnum
33420 to source_line debug hook.
33421 (notice_source_line): Compute last_columnnum and for debug_column_info
33422 return true on column changes.
33423 * debug.h (struct gcc_debug_hooks): Add column argument to
33424 source_line and begin_prologue hooks.
33425 (debug_nothing_int_charstar_int_bool): Remove prototype.
33426 (debug_nothing_int_int_charstar,
33427 debug_nothing_int_int_charstar_int_bool): New prototypes.
33428 (dwarf2out_begin_prologue): Add column argument.
33429 * debug.c (do_nothing_debug_hooks): Adjust source_line and
33430 begin_prologue hooks.
33431 (debug_nothing_int_charstar_int_bool): Remove.
33432 (debug_nothing_int_int_charstar,
33433 debug_nothing_int_int_charstar_int_bool): New functions.
33434 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
33435 through to dwarf2out_source_line.
33436 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
33437 (dwarf2out_source_line): Add column argument, emit it if requested.
33438 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
33440 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33441 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33442 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
33443 through to dwarf2out_begin_prologue.
33444 (vmsdbgout_source_line): Add column argument, pass it through to
33445 dwarf2out_source_line.
33446 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
33447 dbxout_source_line caller.
33448 (dbxout_source_line): Add column argument.
33450 * common.opt (gno-column-info, gcolumn-info): New options.
33451 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
33452 (check_die): Also test for multiple DW_AT_decl_column attributes.
33453 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
33454 DW_AT_decl_column if requested.
33455 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
33457 (gen_variable_die): Likewise.
33458 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
33459 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
33462 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
33463 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
33464 (ix86_handle_option): Handle OPT_m3dnowa.
33465 * doc/invoke.texi (-m3dnowa): Document.
33466 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
33467 -m3dnowa instead of -m3dnow -march=athlon.
33470 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
33471 instead of gcc_assert for K, r and R code checks. Formatting fixes.
33473 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33476 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
33477 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
33478 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
33479 generator for vsx_xxpermdi_<mode>_be.
33480 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
33481 force big-endian semantics.
33482 (vsx_xxpermdi_<mode>_be): New define_expand with same
33483 implementation as previous version of vsx_xxpermdi_<mode>.
33485 2017-02-17 Jakub Jelinek <jakub@redhat.com>
33487 PR tree-optimization/79327
33488 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
33489 variable, its initialization and use.
33491 2017-02-17 Julia Koval <julia.koval@intel.com>
33493 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
33494 (OPTION_MASK_ISA_PKU_UNSET): New.
33495 (ix86_handle_option): Handle -mrdpid.
33496 * config/i386/cpuid.h (bit_RDPID): New.
33497 * config/i386/driver-i386.c (host_detect_local_cpu):
33498 Detect RDPID feature.
33499 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
33500 * config/i386/i386-c.c (ix86_target_macros_internal):
33502 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
33503 (ix86_valid_target_attribute_inner_p): Add "rdpid".
33504 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
33505 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
33506 * config/i386/i386.md (define_insn "rdpid"): New.
33507 * config/i386/i386.opt Add -mrdpid.
33508 * config/i386/immintrin.h (_rdpid_u32): New.
33510 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
33512 PR rtl-optimization/79541
33513 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
33514 instead of transforming it into USE.
33516 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
33518 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
33519 If HONOR_SNANS (SFmode) force the input to a register.
33520 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
33521 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
33522 an frsp or similar insn.
33524 2017-02-17 Martin Liska <mliska@suse.cz>
33526 PR rtl-optimization/79577
33527 * params.def (selsched-max-sched-times): Increase minimum to 1.
33529 2017-02-17 Martin Liska <mliska@suse.cz>
33531 PR rtl-optimization/79574
33532 * gcse.c (want_to_gcse_p): Prevent integer overflow.
33534 2017-02-17 Martin Liska <mliska@suse.cz>
33536 PR tree-optimization/79529
33537 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
33538 ssa_defined_default_def_p to handle cases which are implicitly
33540 * tree-ssa.c (ssa_defined_default_def_p): New function.
33541 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
33542 which are implicitly defined.
33543 * tree-ssa.h (ssa_defined_default_def_p): Declare.
33545 2017-02-17 Richard Biener <rguenther@suse.de>
33547 PR middle-end/79576
33548 * params.def (max-ssa-name-query-depth): Limit to 10.
33550 2017-02-17 Richard Biener <rguenther@suse.de>
33552 PR tree-optimization/79552
33553 * tree-ssa-structalias.c (visit_loadstore): Properly verify
33556 2017-02-17 Richard Biener <rguenther@suse.de>
33559 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
33561 2017-02-17 Marek Polacek <polacek@redhat.com>
33563 PR middle-end/79536
33564 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
33565 (fold_negate_expr): New wrapper.
33567 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
33569 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
33570 Correct terminology and de-emphasize pre-standard behavior.
33572 2017-02-16 Alan Modra <amodra@gmail.com>
33574 PR rtl-optimization/79286
33575 * ira.c (def_dominates_uses): New function.
33576 (update_equiv_regs): Don't create an equivalence for insns that
33577 may trap where the register def does not dominate the use.
33579 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
33581 PR rtl-optimization/78127
33582 * lra.c (lra): Call lra_eliminate before finish the loop after
33585 2017-02-16 Richard Biener <rguenther@suse.de>
33587 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
33589 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
33590 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
33591 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
33592 (isl_val_int_from_wi): New function.
33593 (extract_affine_gmp): Rename to ...
33594 (extract_affine_wi): ... this, take a widest_int.
33595 (extract_affine_int): Just wrap extract_affine_wi.
33596 (add_param_constraints): Use isl_val_int_from_wi.
33597 (add_loop_constraints): Likewise, and extract_affine_wi.
33599 2017-02-15 Jeff Law <law@redhat.com>
33601 PR middle-end/79521
33602 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
33603 ira_init_register_move_cost_if_necessary.
33605 2017-02-15 Martin Sebor <msebor@redhat.com>
33607 PR middle-end/32003
33608 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
33609 removed in a prior commit.
33611 2017-02-15 Bin Cheng <bin.cheng@arm.com>
33613 PR tree-optimization/79347
33614 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
33615 counters during peeling.
33617 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
33619 * Makefile.in (site.exp): Remove "set ISLVER".
33621 2017-02-15 Jakub Jelinek <jakub@redhat.com>
33624 * real.c (real_from_integer): Call real_convert even for decimal.
33626 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
33629 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
33631 2017-02-14 Andrew Pinski <apinski@cavium.com>
33633 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
33634 cores and change the partno/implementer to be correct.
33635 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
33636 the 'B" as the implementer.
33637 * config/aarch64/aarch64-tune.md: Regenerate.
33639 2017-02-14 Carl Love <cel@us.ibm.com>
33641 * config/rs6000/rs6000.c: Add case statement entry to make the
33642 xvcvuxdsp built-in argument unsigned.
33643 * config/rs6000/vsx.md: Fix the source and return operand types so they
33644 match the instruction definitions from the ISA document. Fix typo
33645 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
33648 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
33651 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
33652 member early_clobber_alts.
33653 * lra-lives.c (reg_early_clobber_p): New.
33654 (process_bb_lives): Use it.
33655 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
33656 (debug_operand_data): Initialize early_clobber_alts.
33657 (setup_operand_alternative): Set up early_clobber_alts.
33658 (collect_non_operand_hard_regs): Ditto. Pass early clobber
33659 alternatives to new_insn_reg.
33660 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
33662 (lra_update_insn_regno_info): Pass the new arg.
33664 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33666 PR middle-end/79505
33667 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
33668 (new_oacc_loop_raw): Don't clear already cleared fields.
33671 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
33672 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
33673 _mm512_prefetch_i64gather_ps): New inline functions and macros.
33675 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
33678 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
33680 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
33683 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
33684 the extra instruction to the right place to store 128-bit constant
33687 2017-02-14 Martin Sebor <msebor@redhat.com>
33689 PR middle-end/79448
33690 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
33691 warning for strings of unknown length.
33693 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
33695 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
33697 2017-02-14 Jeff Law <law@redhat.com>
33700 * ira-costs.c (scan_one_insn): Initialize register move costs
33701 for pseudos seen in USE/CLOBBER insns.
33703 PR tree-optimization/79095
33704 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
33705 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
33706 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
33707 if the operands are known to be not equal, then the resulting range
33709 (intersect_ranges): If the new range is ~[0,0] and the old range is
33710 wide, then prefer ~[0,0].
33711 * tree-vrp.c (overflow_comparison_p_1): New function.
33712 (overflow_comparison_p): New function.
33713 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
33714 if NAME is used in an overflow test.
33715 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
33716 overflow check that can be expressed as an equality test, then adjust
33717 ops to be that equality test.
33719 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33721 * config/s390/s390-builtin-types.def: Remove flags argument.
33722 * config/s390/s390.c (s390_init_builtins): Likewise.
33724 2017-02-14 Martin Liska <mliska@suse.cz>
33726 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
33727 vector. Fix trailing white spaces.
33729 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
33731 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
33734 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33736 PR rtl-optimization/68664
33737 * config/arm/arm.c (arm_sched_can_speculate_insn):
33738 New function. Declare prototype.
33739 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
33741 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33743 PR rtl-optimization/68664
33744 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
33746 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
33748 2017-02-14 Amit Pawar <amit.pawar@amd.com>
33750 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
33751 max skip bytes for function, loop and jump.
33753 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33755 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
33756 ABS_EXPR for gimple dump.
33758 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33761 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
33763 PR tree-optimization/79408
33764 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
33765 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
33766 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
33767 also if rhs1 is INTEGER_CST.
33769 2017-02-14 Richard Biener <rguenther@suse.de>
33771 PR middle-end/79432
33772 * tree-into-ssa.c (insert_phi_nodes): When the function can
33773 have abnormal edges rewrite SSA names with broken use-def
33774 dominance out of SSA and register them for PHI insertion.
33776 2017-02-13 Martin Sebor <msebor@redhat.com>
33778 PR middle-end/79496
33779 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
33780 clearing info.nowrite flag when snprintf size argument is a range.
33782 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33784 * cprop.c (cprop_jump): Add missing space in string literal.
33785 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
33786 (get_constraint_for_component_ref): Likewise.
33787 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
33788 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
33789 * lra-constraints.c (process_alt_operands): Likewise.
33790 * ipa-inline.c (inline_small_functions): Likewise.
33791 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
33792 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
33793 * trans-mem.c (diagnose_tm_1_op): Likewise.
33794 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
33795 (grid_parallel_clauses_gridifiable): Likewise.
33797 * config/nvptx/mkoffload.c (process): Add space in between
33800 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
33801 "MOD4_SSE_REGS" and "ALL_REGS".
33803 * spellcheck.c (test_data): Add , in between "foo" and "food".
33805 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33808 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
33809 boundary crossing check and subsequent code generation agree.
33811 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33813 * config/aarch64/aarch64.c (has_memory_op): Delete.
33814 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
33817 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33819 PR rtl-optimization/79388
33820 PR rtl-optimization/79450
33821 * combine.c (distribute_notes): When removing TEM_INSN for which
33822 corresponding dest has last value recorded, invalidate that last
33825 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33827 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
33828 of explicit '@'. Add missing assembly comment marker on branch costs
33831 2017-02-13 Nathan Sidwell <nathan@acm.org>
33833 * gengtype-lex.l (<in_struct>): Add '/'.
33835 2017-02-13 Martin Liska <mliska@suse.cz>
33838 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
33840 2017-02-13 Richard Biener <rguenther@suse.de>
33842 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
33844 * configure: Re-generate.
33845 * config.in: Likewise.
33846 * graphite-dependences.c: Simplify as if
33847 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
33848 * graphite-isl-ast-to-gimple.c: Likewise.
33849 * graphite-optimize-isl.c: Likewise.
33850 * graphite-poly.c: Likewise.
33851 * graphite-sese-to-poly.c: Likewise.
33852 * graphite.h: Likewise.
33853 * toplev.c: Include isl/version.h and use isl_version () for
33854 printing the ISL version.
33855 * doc/install.texi: Update ISL requirement.
33857 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33859 * doc/standards.texi (Standards): Update reference to
33862 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33864 * doc/extend.texi (Named Address Spaces): sourceware.org now
33866 * doc/install.texi (Binaries): Ditto.
33869 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33871 * doc/cpp.texi: Replace "stringify"/"stringification" with C
33872 standard terminology "stringize"/"stringizing" throughout.
33873 * doc/cppinternals.texi: Likewise.
33875 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33877 * doc/extend.texi: Fix some spelling mistakes and typos.
33878 * doc/invoke.texi: Likewise.
33880 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
33883 * params.def (inline-min-speedup) Change from 10 to 8.
33885 2017-02-11 Jakub Jelinek <jakub@redhat.com>
33887 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
33890 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
33893 * ipa-inline-analysis.c (get_minimal_bb): New function.
33894 (record_modified): Use it.
33895 (remap_edge_change_prob): Handle also ancestor functions.
33897 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
33899 * doc/contrib.texi (Contributors): Remove broken link into
33900 the Mauve CVS repository.
33902 2017-02-11 Jakub Jelinek <jakub@redhat.com>
33904 PR middle-end/79454
33905 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
33906 result computation whenever lhs doesn't have vector mode, not
33907 just when it has BLKmode.
33909 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
33911 * doc/makefile.texi (profiledbootstrap): Refer to the
33912 installation instructions only in textual form.
33914 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33917 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
33919 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
33921 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
33922 (Specific): Update mingw-w64 reference.
33924 (Specific): Remove broken link to Renesas RX processor.
33926 2017-02-10 Richard Biener <rguenther@suse.de>
33928 * toplev.c (process_options): Do not mention obsolete graphite
33929 options when printing sorry message about missing graphite support.
33930 Mention -floop-nest-optimize.
33932 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
33934 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
33935 (vtst_p16): Likewise.
33936 (vtstq_p8): Likewise.
33937 (vtstq_p16): Likewise.
33939 (vtstq_p64): Likewise.
33940 * config/arm/arm_neon.h (vgetq_lane_p64): New.
33941 (vset_lane_p64): New.
33942 (vsetq_lane_p64): New.
33944 2017-02-10 Jakub Jelinek <jakub@redhat.com>
33946 PR tree-optimization/79411
33947 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
33948 stmt operands are SSA_NAMEs used in abnormal phis.
33949 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
33952 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
33955 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
33958 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
33960 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
33962 2017-02-09 Jakub Jelinek <jakub@redhat.com>
33964 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
33968 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
33969 not arbitrary TREE_CONSTANT.
33972 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
33973 "omp declare target link" attribute unless is_global_var.
33974 * omp-offload.c (find_link_var_op): Likewise.
33976 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
33977 Chung-Lin Tang <cltang@codesourcery.com>
33979 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
33981 (gimplify_adjust_omp_clauses): Don't delete TILE.
33982 (gimplify_omp_for): Deal with TILE.
33983 * internal-fn.c (expand_GOACC_TILE): New function.
33984 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
33986 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
33987 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
33989 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
33990 avoid DIV for outermost collapse var.
33991 (expand_oacc_for): Insert tile element loop as needed. Adjust.
33992 Remove out of date comments, fix whitespace.
33993 * omp-general.c (omp_extract_for_data): Deal with tiling.
33994 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
33995 adjust OLF_DIM_BASE value.
33996 (struct omp_for_data): Add tiling field.
33997 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
33998 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
33999 for auto loops. Remove default auto determining, moved to
34000 oacc_loop_fixed_partitions.
34001 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
34002 stmts, add e_mask field.
34003 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
34004 (oacc_thread_numbers): Use oacc_dim_call.
34005 (oacc_xform_tile): New.
34006 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
34007 (finish_oacc_loop): Adjust for ifns vector.
34008 (oacc_loop_discover_walk): Append loop abstraction sites to list,
34009 add case for GOACC_TILE fns.
34010 (oacc_loop_xform_loop): Delete.
34011 (oacc_loop_process): Iterate over call list directly, and add
34012 handling for GOACC_TILE fns.
34013 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
34015 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
34016 vector partitioning to outer loops. Assign 2 partitions to loops
34017 when available. Add TILE handling.
34018 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
34019 (execite_oacc_device_lower): Process GOACC_TILE fns,
34020 ignore unknown specs.
34021 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
34022 * tree.c (omp_clause_num_ops): Adjust TILE ops.
34023 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
34025 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
34027 * configure.ac (ACX_BUGURL): Update.
34028 * configure: Regenerate.
34030 2017-02-09 Richard Biener <rguenther@suse.de>
34032 PR tree-optimization/69823
34033 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
34034 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
34036 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
34038 * config/arc/arc-c.def: Add __NPS400__ definition.
34039 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
34040 (TARGET_NPS400): Define.
34042 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
34044 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
34046 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
34047 pointer, arch_info.
34048 (arc_cpu_types): Fill the arch_info field with a pointer into the
34049 arc_arch_types table.
34050 (arc_selected_cpu): Declare.
34051 * config/arc/arc.c (arc_selected_cpu): Make global.
34052 (arc_selected_arch): Delete.
34053 (arc_base_cpu): Delete.
34054 (arc_override_options): Remove references to deleted variables,
34055 update access to arch information.
34056 (ARC_OPT): Update access to arch information.
34057 (ARC_OPTX): Likewise.
34058 * config/arc/arc.h (arc_base_cpu): Remove declaration.
34059 (TARGET_ARC600): Update access to arch information.
34060 (TARGET_ARC601): Likewise.
34061 (TARGET_ARC700): Likewise.
34062 (TARGET_EM): Likewise.
34063 (TARGET_HS): Likewise.
34064 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
34067 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
34070 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
34071 condition/operands for integer GE/LE/GEU/LEU operations.
34073 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
34075 PR translation/79397
34076 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
34079 2017-02-08 Martin Jambor <mjambor@suse.cz>
34082 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
34083 whether allocation happened.
34084 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
34085 nothing was allocated.
34087 2017-02-08 Jakub Jelinek <jakub@redhat.com>
34089 PR tree-optimization/79408
34090 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
34091 constant, but SSA_NAME with a known integer range, use the minimum
34092 of that range instead of op1 to determine if modulo can be replaced
34093 with its first operand.
34095 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34097 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
34099 2017-02-08 Richard Biener <rguenther@suse.de>
34101 PR tree-optimization/71824
34102 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34103 Check all loops contained in the merged region.
34105 2017-02-07 Andrew Pinski <apinski@cavium.com>
34107 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
34109 2017-02-07 Andrew Pinski <apinski@cavium.com>
34111 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
34112 (thunderxt88): Likewise.
34113 (thunderxt81): Disable LSE and change v8.1 to v8.
34114 (thunderxt83): Likewise.
34116 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34117 Richard Biener <rguenther@suse.de>
34119 PR middle-end/79399
34120 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
34121 type from int to size_t.
34122 * ira-costs.c (struct_costs_size): Change type from int to size_t.
34124 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34126 PR rtl-optimization/79386
34127 * cprop.c (bypass_conditional_jumps): Initialize
34128 bypass_last_basic_block already before splitting bbs after
34129 unconditional traps...
34130 (bypass_conditional_jumps): ... rather than here.
34133 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
34134 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
34135 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
34136 fix -masm=intel patterns.
34138 2017-02-07 Richard Biener <rguenther@suse.de>
34140 PR tree-optimization/79256
34141 PR middle-end/79278
34142 * builtins.c (get_object_alignment_2): Use min_align_of_type
34143 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
34144 and ADJUST_FIELD_ALIGN.
34146 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
34148 * doc/tm.texi: Regenerate.
34149 * stor-layout.c (layout_decl): Adjust.
34150 (update_alignment_for_field): Likewise.
34151 (place_field): Likewise.
34152 (min_align_of_type): Likewise.
34153 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
34154 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
34155 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
34156 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
34157 * config/frv/frv.c (frv_adjust_field_align): Likewise.
34158 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
34159 * config/i386/i386.c (x86_field_alignment): Likewise.
34160 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
34161 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
34162 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
34163 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
34164 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
34165 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
34169 2017-01-30 Richard Biener <rguenther@suse.de>
34171 PR tree-optimization/79256
34172 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34173 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34176 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
34178 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
34179 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
34180 builtins to SImode and emit a zero-extend, if necessary.
34182 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34184 * docs/invoke.texi (RISC-V Options): Alphabetize.
34186 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34188 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
34191 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34193 * config/riscv/riscv.c: New file.
34194 * common/config/riscv/riscv-common.c: Likewise.
34195 * config.gcc: Likewise.
34196 * config/riscv/constraints.md: Likewise.
34197 * config/riscv/elf.h: Likewise.
34198 * config/riscv/generic.md: Likewise.
34199 * config/riscv/linux.h: Likewise.
34200 * config/riscv/multilib-generator: Likewise.
34201 * config/riscv/peephole.md: Likewise.
34202 * config/riscv/pic.md: Likewise.
34203 * config/riscv/predicates.md: Likewise.
34204 * config/riscv/riscv-builtins.c: Likewise.
34205 * config/riscv/riscv-c.c: Likewise.
34206 * config/riscv/riscv-ftypes.def: Likewise.
34207 * config/riscv/riscv-modes.def: Likewise.
34208 * config/riscv/riscv-opts.h: Likewise.
34209 * config/riscv/riscv-protos.h: Likewise.
34210 * config/riscv/riscv.h: Likewise.
34211 * config/riscv/riscv.md: Likewise.
34212 * config/riscv/riscv.opt: Likewise.
34213 * config/riscv/sync.md: Likewise.
34214 * config/riscv/t-elf-multilib: Likewise.
34215 * config/riscv/t-linux: Likewise.
34216 * config/riscv/t-linux-multilib: Likewise.
34217 * config/riscv/t-riscv: Likewise.
34218 * configure.ac: Likewise.
34219 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
34220 Waterman as RISC-V maintainers.
34221 * doc/install.texi: Add RISC-V entries.
34222 * doc/invoke.texi: Add RISC-V options section.
34223 * doc/md.texi: Add RISC-V constraints section.
34224 * configure: Regenerated.
34226 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
34229 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
34230 false values to be constant vectors with all 0 or all 1 bits set.
34231 (vcondu<mode><mode>): Likewise.
34232 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
34234 (fpmask_comparison_operator): Update comment.
34235 (vecint_comparison_operator): New predicate.
34236 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
34237 vector conditionals when the true and false values are constant
34238 vectors with all 0 bits or all 1 bits set.
34240 2017-02-06 Martin Sebor <msebor@redhat.com>
34242 PR tree-optimization/79376
34243 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
34245 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
34247 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
34248 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
34249 to simplify split condition.
34251 2017-02-06 Jakub Jelinek <jakub@redhat.com>
34253 * omp-expand.c (oxpand_omp_atomic_fetch_op,
34254 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
34257 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
34259 PR rtl-optimization/68664
34260 * target.def (can_speculate_insn): New hook.
34261 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
34262 * doc/tm.texi: Regenerate.
34263 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
34264 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
34265 (rs6000_sched_can_speculate_insn): New function.
34267 2017-02-06 Jakub Jelinek <jakub@redhat.com>
34269 PR tree-optimization/79284
34270 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
34271 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
34272 vectorizable_mask_load_store, vectorizable_operation,
34273 vect_is_simple_cond, get_same_sized_vectype): Use it instead
34274 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
34275 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
34276 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
34277 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
34278 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
34279 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
34280 is_gimple_assign (stmt). Replace another such test with
34281 is_gimple_assign (stmt).
34283 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
34286 * config/avr/avr.c (rtl-iter.h): Include it.
34287 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
34288 (avr_legitimate_combined_insn): ...and implementation.
34290 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
34292 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
34293 * config/s390/s390.c (s390_const_operand_ok)
34294 (s390_canonicalize_comparison, s390_extract_part)
34295 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
34296 (s390_contiguous_bitmask_p, s390_rtx_costs)
34297 (legitimize_pic_address): Likewise.
34298 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
34299 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
34300 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
34301 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
34302 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
34304 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
34306 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
34307 REGNO($0) == REGNO($1).
34309 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34311 * config/s390/linux.h(SIZE_TYPE): Add comment.
34313 2017-02-06 Julian Brown <julian@codesourcery.com>
34314 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34315 Virendra Pathak <virendra.pathak@broadcom.com>
34317 * config/aarch64/aarch64-cores.def: Change the scheduler
34319 * config/aarch64/aarch64.md: Include thunderx2t99.md.
34320 * config/aarch64/thunderx2t99.md: New file.
34322 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34324 * doc/standards.texi (Go Language): Update link to language
34327 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
34329 * tree-eh.c (lower_resx): Sanitize profile.
34330 (cleanup_empty_eh_move_lp): Likewise.
34332 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
34335 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
34337 * cfgloopmanip.h (loop_version): Update prototype.
34338 * modulo-sched.c (sms_schedule): Update call of loop_version.
34339 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
34340 * tree-parloops.c (gen_parallel_loop): Likewise.
34341 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
34342 * tree-ssa-loop-split.c (split_loop): Likewise.
34343 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
34344 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
34346 2017-02-05 Martin Liska <mliska@suse.cz>
34349 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
34351 (print_operand_address): Initialize a struct to zero.
34353 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34355 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
34356 garbage collector only in textual form.
34358 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34360 * doc/extend.texi (x86 specific memory model extensions for
34361 transactional memory): Simplify a phrase.
34363 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
34366 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
34367 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
34368 (atomic_storedi_1): Likewise.
34370 2017-02-04 Jakub Jelinek <jakub@redhat.com>
34372 PR tree-optimization/79338
34373 * tree-parloops.c (gather_scalar_reductions): Don't call
34374 vect_analyze_loop_form for loop->inner before destroying loop's
34377 2017-02-03 Martin Sebor <msebor@redhat.com>
34379 PR tree-optimization/79327
34380 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
34381 when precision has resulted in leading zeros.
34382 (format_integer): Adjust the likely counter to assume an unknown
34383 argument that may be zero is non-zero.
34385 2017-02-03 Jason Merrill <jason@redhat.com>
34388 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
34389 avoid copying non-taken branch.
34391 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34393 PR tree-optimization/79340
34394 * tree-vect-loop.c (vectorizable_reduction): Release
34395 vec_defs elements after safe_splicing them into other vectors.
34398 PR tree-optimization/79327
34399 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
34400 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
34402 (format_integer): Use wide_int_to_tree instead of build_int_cst
34403 + to_?hwi. If argmin is NULL, just set argmin and argmax to
34404 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
34405 of shortest and longest sequence.
34407 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
34409 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
34410 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
34412 2017-02-03 Walter Lee <walt@tilera.com>
34415 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
34416 after initial stackframe link reg save.
34417 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
34419 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34422 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
34423 wu for stxssp alternative.
34425 2017-02-03 Martin Sebor <msebor@redhat.com>
34427 PR tree-optimization/79352
34428 * gimple-fold.c (get_range_strlen): Add argument.
34429 (get_range_strlen): Change return type to bool.
34430 (get_maxval_strlen): Pass in a dummy argument.
34431 * gimple-fold.h (get_range_strlen): Change return type to bool.
34432 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
34433 * tree.h (array_at_struct_end_p): Add argument.
34434 * tree.c (array_at_struct_end_p): Handle it.
34436 2017-02-03 Martin Liska <mliska@suse.cz>
34439 * multiple_target.c (create_dispatcher_calls): Redirect edge
34440 from a caller of a dispatcher.
34441 (expand_target_clones): Make the clones local.
34442 (ipa_target_clone): Do both target clones and resolvers.
34443 (ipa_dispatcher_calls): Remove the pass.
34444 (pass_dispatcher_calls::gate): Likewise.
34445 (make_pass_dispatcher_calls): Likewise.
34446 * passes.def (pass_target_clone): Put as very first IPA early
34449 2017-02-03 Martin Liska <mliska@suse.cz>
34451 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
34452 in case of a function with ifunc attribute.
34454 2017-02-03 Martin Liska <mliska@suse.cz>
34456 * cgraph.c (cgraph_node::dump): Dump function version info.
34457 * symtab.c (symtab_node::dump_base): Add missing new line.
34459 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34461 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
34462 (ifcombine_ifandif): Use it.
34464 2017-02-03 Martin Liska <mliska@suse.cz>
34466 * doc/invoke.texi: Document default value for
34467 use-after-scope-direct-emission-threshold.
34469 2017-02-03 Martin Liska <mliska@suse.cz>
34471 PR tree-optimization/79339
34472 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
34473 (format_floating): Likewise.
34475 2017-02-03 Martin Liska <mliska@suse.cz>
34478 * ipa-prop.c (ipa_node_params_t::insert): Remove current
34480 (ipa_node_params_t::remove): Likewise.
34481 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
34482 initialization from removed ipa_node_params_t::insert.
34483 (ipa_node_params::~ipa_node_params): Move from removed
34484 ipa_node_params_t::release.
34485 * symbol-summary.h (symbol_summary::m_released): New member.
34486 Do not release a summary twice. Do not allow to call finalizer
34487 for types of a summary that live in GGC memory.
34489 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34491 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
34494 2017-02-02 Martin Sebor <msebor@redhat.com>
34496 PR middle-end/79275
34497 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
34498 (format_string): Tighten up the range of output for non-constant
34499 strings and correct the expected range for wide non-constant strings.
34501 2017-02-02 Martin Sebor <msebor@redhat.com>
34503 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
34505 PR middle-end/32003
34506 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
34508 (-fdump-tree-@var): Add to index and document how to come up
34509 with pass-specific option and dump file names.
34510 (-fdump-passes): Clarify where to look for output.
34512 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34514 PR middle-end/77445
34515 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
34516 statistics of the analyzed path; allow threading for speed when
34517 any of BBs along the path are optimized for speed.
34519 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
34521 PR middle-end/78468
34522 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
34523 settings of the virtual registers.
34526 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
34528 * explow.c (get_dynamic_stack_size): Take known alignment of stack
34529 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
34532 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34534 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
34535 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
34537 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34539 * config/s390/s390.md: Add missing comments with the expanded
34541 * config/s390/vector.md: Likewise.
34542 * config/s390/vx-builtins.md: Likewise.
34544 2017-02-02 Jakub Jelinek <jakub@redhat.com>
34547 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
34548 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
34549 conditions on a single line.
34551 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34553 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
34554 __S390_VX__ to __VX__.
34556 2017-02-01 Andrew Pinski <apinski@cavium.com>
34558 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
34559 stmt_info to record_stmt_cost.
34560 (vect_get_known_peeling_cost): Pass stmt_info if known to
34562 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
34563 cpu_vector_cost field into
34564 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
34565 field into vec_int_stmt_cost and vec_fp_stmt_cost.
34566 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
34567 splitting of scalar_stmt_cost and vec_stmt_cost.
34568 (thunderx_vector_cost): Likewise.
34569 (cortexa57_vector_cost): LIkewise.
34570 (exynosm1_vector_cost): Likewise.
34571 (xgene1_vector_cost): Likewise.
34572 (thunderx2t99_vector_cost): Improve after the splitting of the two
34574 (aarch64_builtin_vectorization_cost): Update for the splitting of
34575 scalar_stmt_cost and vec_stmt_cost.
34577 2017-02-01 Torvald Riegel <triegel@redhat.com>
34578 Richard Henderson <rth@redhat.com>
34580 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
34581 conditional on existance of a fast atomic load.
34582 * optabs-query.c (can_atomic_load_p): New function.
34583 * optabs-query.h (can_atomic_load_p): Declare it.
34584 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
34585 no fast atomic load is available for the particular size of access.
34586 (expand_atomic_compare_and_swap): Likewise.
34587 (expand_atomic_load): Likewise.
34588 (expand_atomic_store): Likewise.
34589 (expand_atomic_fetch_op): Likewise.
34590 * testsuite/lib/target-supports.exp
34591 (check_effective_target_sync_int_128): Remove x86 because it provides
34592 no fast atomic load.
34593 (check_effective_target_sync_int_128_runtime): Likewise.
34595 2017-02-01 Richard Biener <rguenther@suse.de>
34597 * graphite.c: Include tree-vectorizer.h for find_loop_location.
34598 (graphite_transform_loops): Provide opt-info for optimized nests.
34599 * tree-parloop.c (parallelize_loops): Provide opt-info for
34600 parallelized loops.
34602 2017-02-01 Richard Biener <rguenther@suse.de>
34604 PR middle-end/79315
34605 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
34606 was not set before.
34608 2017-02-01 Richard Biener <rguenther@suse.de>
34610 PR tree-optimization/71824
34611 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34612 Verify the loops are valid in the merged SESE region.
34613 (scop_detection::can_represent_loop_1): Check analyzing the
34614 evolution of the number of iterations in the region succeeds.
34616 2017-01-31 Ian Lance Taylor <iant@golang.org>
34618 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
34619 REG_ARGS_SIZE note to 32-bit push insns and call insn.
34621 2017-01-31 David Malcolm <dmalcolm@redhat.com>
34623 PR preprocessor/79210
34624 * input.c (get_substring_ranges_for_loc): Replace line_width
34625 assertion with error-handling.
34627 2017-01-31 Richard Biener <rguenther@suse.de>
34629 PR tree-optimization/77318
34630 * graphite-sese-to-poly.c (extract_affine): Fix assert.
34631 (create_pw_aff_from_tree): Take loop parameter.
34632 (add_condition_to_pbb): Pass loop of the condition to
34633 create_pw_aff_from_tree.
34635 2017-01-31 Jakub Jelinek <jakub@redhat.com>
34637 * config/s390/s390.c (s390_asan_shadow_offset): New function.
34638 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
34640 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
34644 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
34646 (convert_int_to_float128): Likewise.
34647 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
34648 (convert_int_to_float128): Likewise.
34649 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
34650 (UNSPEC_IEEE128_CONVERT): Likewise.
34651 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
34652 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
34653 Use local variables for IBM extended format.
34654 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
34655 (fix_trunc<mode>si2_fprs): Likewise.
34656 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
34657 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
34658 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
34659 to know that we can now have integers of all sizes in vector
34661 (fix<uns>_<mode>di2_hw): Likewise.
34662 (float<uns>_<mode>si2_hw): Likewise.
34663 (fix_<mode>si2_hw): Likewise.
34664 (fixuns_<mode>si2_hw): Likewise.
34665 (float<uns>_<mode>di2_hw): Likewise.
34666 (float_<mode>di2_hw): Likewise.
34667 (float_<mode>si2_hw): Likewise.
34668 (floatuns_<mode>di2_hw): Likewise.
34669 (floatuns_<mode>si2_hw): Likewise.
34670 (xscvqp<su>wz_<mode>): Delete, no longer used.
34671 (xscvqp<su>dz_<mode>): Likewise.
34672 (xscv<su>dqp_<mode>): Likewise.
34673 (ieee128_mfvsrd_64bit): Likewise.
34674 (ieee128_mfvsrd_32bit): Likewise.
34675 (ieee128_mfvsrwz): Likewise.
34676 (ieee128_mtvsrw): Likewise.
34677 (ieee128_mtvsrd_64bit): Likewise.
34678 (ieee128_mtvsrd_32bit): Likewise.
34680 2017-01-31 Martin Liska <mliska@suse.cz>
34683 * ipa-prop.c (ipa_free_all_node_params): Call release method
34684 instead of ~sumbol_summary to not to trigger double times
34687 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
34689 PR tree-optimization/71691
34690 * bitmap.h (class auto_bitmap): New.
34691 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
34692 is_maybe_undefined instead of ssa_undefined_value_p.
34694 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34696 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
34697 __S390_ARCH_LEVEL__ to __ARCH__.
34699 2017-01-31 Jakub Jelinek <jakub@redhat.com>
34701 PR tree-optimization/79267
34702 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
34703 if should_remove_lhs_p is true.
34705 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
34708 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
34709 (add_alignment_attribute): New.
34710 (base_type_die): Add alignment attribute.
34711 (subrange_type_die): Likewise.
34712 (modified_type_die): Likewise.
34713 (gen_array_type_die): Likewise.
34714 (gen_descr_array_type_die: Likewise.
34715 (gen_enumeration_type_die): Likewise.
34716 (gen_subprogram_die): Likewise.
34717 (gen_variable_die): Likewise.
34718 (gen_field_die): Likewise.
34719 (gen_ptr_to_mbr_type_die): Likewise.
34720 (gen_struct_or_union_type_die): Likewise.
34721 (gen_subroutine_type_die): Likewise.
34722 (gen_typedef_die): Likewise.
34723 (base_type_cmp): Compare alignment attribute.
34725 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34728 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
34729 (setb_unsigned) New pattern for setb with CCUNS.
34730 * config/rs6000/rs6000.c (expand_block_compare): Use a different
34731 subfc./subfe sequence to avoid overflow problems. Generate a
34732 shorter sequence with cmpld/setb for power9.
34733 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
34734 for generating subfc. instruction.
34735 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
34736 now uses this instruction.
34738 2017-01-30 Ian Lance Taylor <iant@google.com>
34741 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
34742 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
34744 2017-01-30 Martin Sebor <msebor@redhat.com>
34746 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
34747 Move constant to the right of a relational operator.
34748 (get_mpfr_format_length, format_character, format_string): Ditto.
34749 (should_warn_p, maybe_warn): Same.
34751 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
34753 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
34756 * asan.c (get_translation_unit_decl): Remove function.
34757 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
34759 2017-01-30 Martin Liska <mliska@suse.cz>
34761 PR gcov-profile/79259
34762 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
34763 -fprofile-generate.
34765 2017-01-30 Martin Liska <mliska@suse.cz>
34768 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
34769 Initialize variables with NULL value.
34771 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
34774 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
34776 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
34778 2017-01-30 Richard Biener <rguenther@suse.de>
34780 PR tree-optimization/79276
34781 * tree-vrp.c (process_assert_insertions): Properly adjust common
34782 when removing a duplicate.
34784 2017-01-30 Richard Biener <rguenther@suse.de>
34786 PR tree-optimization/79256
34787 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34788 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34790 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
34792 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
34795 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
34796 ("*r<noxa>sbg_<mode>_sll_bitmask")
34797 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
34798 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
34799 Use contiguous_bitmask_nowrap_operand.
34801 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34804 * config/rs6000/altivec.h (vec_xl): Revise #define.
34805 (vec_xst): Likewise.
34807 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
34809 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
34811 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
34813 PR rtl-optimization/79194
34814 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
34815 traps before call to bypass_conditional_jumps.
34817 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34819 PR tree-optimization/71374
34820 * lra-constraints.c (check_conflict_input_operands): New.
34821 (match_reload): Use it.
34823 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34826 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
34827 account to calculate conflict_set.
34829 2017-01-27 Bin Cheng <bin.cheng@arm.com>
34831 PR rtl-optimization/78559
34832 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
34833 other_insn in combine.
34835 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
34837 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
34838 uint16_type_node for BT_UINT16.
34840 2017-01-27 David Malcolm <dmalcolm@redhat.com>
34842 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
34843 "RTL Tests" to menu.
34844 (GIMPLE Tests): New node.
34845 (RTL Tests): New node.
34847 2017-01-27 Richard Biener <rguenther@suse.de>
34849 PR tree-optimization/79245
34850 * tree-loop-distribution.c (distribute_loop): Apply cost
34851 modeling also to detected patterns.
34853 2017-01-27 Richard Biener <rguenther@suse.de>
34855 PR tree-optimization/71433
34856 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
34857 (compare_assert_loc): New function.
34858 (process_assert_insertions): Sort and optimize assert locations
34859 to remove duplicates and push down identical assertions on
34860 edges to their destination block.
34862 2017-01-27 Richard Biener <rguenther@suse.de>
34864 PR tree-optimization/79244
34865 * tree-vrp.c (remove_range_assertions): Forcefully propagate
34866 out SSA names even if abnormal.
34868 2017-01-27 Jakub Jelinek <jakub@redhat.com>
34870 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
34871 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
34872 instead of MPFR_RNDN.
34874 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
34877 * arm.c (arm_option_override): Don't call build_target_option_node
34878 until after doing all option overrides.
34879 (arm_valid_target_attribute_tree): Likewise.
34881 2017-01-27 Martin Liska <mliska@suse.cz>
34883 * doc/invoke.texi (-fprofile-arcs): Document profiling support
34884 for {cd}tors and C++ {cd}tors.
34886 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
34888 * config/s390/s390.md ("*setmem_long_and")
34889 ("*setmem_long_and_31z"): Use zero_extend instead of and.
34891 2017-01-26 Martin Sebor <msebor@redhat.com>
34893 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
34896 2017-01-26 Martin Sebor <msebor@redhat.com>
34898 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
34899 HAVE_DFmode before using XFmode or DFmode.
34900 (parse_directive): Avoid using the z length modifier to avoid
34901 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
34903 PR middle-end/78703
34904 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
34905 to accept adjustment as an array.
34906 (get_int_range): New function.
34907 (struct directive): Make width and prec arrays.
34908 (directive::set_width, directive::set_precision): Call get_int_range.
34909 (format_integer, format_floating): Handle width and precision ranges.
34910 (format_string, parse_directive): Same.
34912 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34915 * dwarf2out.c (generate_skeleton_bottom_up): For children with
34916 comdat_type_p set, just clone them, but keep the children in the
34920 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
34921 which have direct callers with -fvar-tracking-assignments enabled
34923 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
34924 inside of type units.
34926 2017-01-26 Martin Sebor <msebor@redhat.com>
34928 PR middle-end/78703
34929 * gimple-ssa-sprintf.c (struct result_range): Add likely and
34931 (struct format_result): Replace number_chars, number_chars_min,
34932 and number_chars_max with a single member of struct result_range.
34934 (format_result::operator+=): Adjust.
34935 (struct fmtresult): Remove bounded. Handle likely and unlikely
34937 (fmtresult::adjust_for_width_or_precision): New function.
34938 (fmtresult:type_max_digits): New function.
34939 (bytes_remaining): Handle likely and unlikely counters.
34940 (min_bytes_remaining): Remove.
34941 (format_percent): Simplify.
34942 (format_integer, format_floating): Set likely and unlikely counters.
34943 (get_string_length, format_character, format_string): Same.
34944 (format_plain, should_warn_p): New function.
34945 (maybe_warn): Call should_warn_p. Update diagnostic messages
34946 and handle those for all directives, including plain strings.
34947 (format_directive): Handle likely and unlikely counters.
34948 Remove unnecessary quoting from diagnostics. Add an informational
34950 (add_bytes): Remove.
34951 (pass_sprintf_length::compute_format_length): Simplify.
34952 (try_substitute_return_value): Handle likely and unlikely counters.
34954 2017-01-26 Carl Love <cel@us.ibm.com>
34956 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
34957 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
34959 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
34962 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
34963 endianess for subregs into account.
34964 * lra-constraints.c (lra_constraints): Do risky transformations
34965 always on the first iteration.
34966 * lra-lives.c (check_pseudos_live_through_calls): Add arg
34967 last_call_used_reg_set.
34968 (process_bb_lives): Define and use last_call_used_reg_set.
34969 * lra.c (lra): Always continue after lra_constraints on the first
34972 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
34974 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
34976 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
34978 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34980 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
34981 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
34982 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
34983 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
34984 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
34985 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
34986 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
34987 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
34988 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
34990 2017-01-26 Marek Polacek <polacek@redhat.com>
34993 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
34994 for the third operand.
34996 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34998 PR middle-end/79236
34999 * omp-low.c (struct omp_context): Add simt_stmt field.
35000 (scan_omp_for): Return omp_context *.
35001 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
35002 context to the _simt_ SIMD stmt.
35003 (lower_omp_for): For combined SIMD with sibling _simt_
35004 SIMD, make sure to use the same decls in _looptemp_
35005 clauses as in the sibling.
35007 2017-01-26 David Sherwood <david.sherwood@arm.com>
35009 PR middle-end/79212
35010 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
35013 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35016 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
35017 emit fld b; fld a; if possible.
35019 * brig-builtins.def: Update copyright years.
35020 * config/arm/arm_acle_builtins.def: Update copyright years.
35022 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
35025 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
35026 constraint instead of o for the stxsd instruction.
35028 2017-01-25 Carl Love <cel@us.ibm.com>
35030 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
35031 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
35033 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
35035 * doc/invoke.texi (C++ Dialect Options): Fix typo.
35037 2017-01-25 Richard Biener <rguenther@suse.de>
35039 PR tree-optimization/69264
35040 * target.def (vector_alignment_reachable): Improve documentation.
35041 * doc/tm.texi: Regenerate.
35042 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
35044 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
35045 earlier changes with respect to TYPE_USER_ALIGN.
35046 (vector_alignment_reachable_p): Likewise. Improve dumping.
35048 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35051 * config/arm/arm.md (xordi3): Force constant operand into a register
35054 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35056 * doc/invoke.texi (-fstore-merging): Correct default optimization
35057 levels at which it is enabled.
35058 (-O): Move -fstore-merging from list to...
35061 2017-01-25 Richard Biener <rguenther@suse.de>
35064 * omp-expand.c: Include debug.h.
35065 (expand_omp_taskreg): Make sure to generate early debug before
35066 outlining anything from a function.
35067 (expand_omp_target): Likewise.
35068 (grid_expand_target_grid_body): Likewise.
35070 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
35073 * asan.c (get_translation_unit_decl): New function.
35074 (asan_add_global): Extract modules file name from globals
35075 TRANSLATION_UNIT_DECL name.
35077 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
35080 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
35081 for long calls with APCS frame and VFP.
35083 2017-01-24 David Malcolm <dmalcolm@redhat.com>
35085 * cfg.c (original_copy_tables_initialized_p): New function.
35086 * cfg.h (original_copy_tables_initialized_p): New decl.
35087 * cfgrtl.c (relink_block_chain): Guard the call to
35088 free_original_copy_tables with a call to
35089 original_copy_tables_initialized_p.
35090 * cgraph.h (symtab_node::native_rtl_p): New decl.
35091 * cgraphunit.c (symtab_node::native_rtl_p): New function.
35092 (symtab_node::needed_p): Don't assert for early assembly output
35093 for __RTL functions.
35094 (cgraph_node::finalize_function): Set "force_output" for __RTL
35096 (cgraph_node::analyze): Bail out early for __RTL functions.
35097 (analyze_functions): Update assertion to support __RTL functions.
35098 (cgraph_node::expand): Bail out early for __RTL functions.
35099 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
35101 * function.h (struct function): Update comment for field
35103 * gimple-expr.c: Include "tree-pass.h".
35104 (gimple_has_body_p): Return false for __RTL functions.
35105 * Makefile.in (OBJS): Add run-rtl-passes.o.
35106 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
35108 (gcc::pass_manager::get_clean_slate): New accessor.
35109 * passes.c: Include "insn-addr.h".
35110 (should_skip_pass_p): Add logging. Update logic for running
35111 "expand" to be compatible with both __GIMPLE and __RTL. Guard
35112 property-provider override so it is only done for gimple passes.
35114 (skip_pass): New function.
35115 (execute_one_pass): Call skip_pass when skipping passes.
35116 * read-md.c (md_reader::read_char): Support filtering
35117 the input to a subset of line numbers.
35118 (md_reader::md_reader): Initialize fields
35119 m_first_line and m_last_line.
35120 (md_reader::read_file_fragment): New function.
35121 * read-md.h (md_reader::read_file_fragment): New decl.
35122 (md_reader::m_first_line): New field.
35123 (md_reader::m_last_line): New field.
35124 * read-rtl-function.c (function_reader::create_function): Only
35125 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
35126 curr_properties. Set DECL_INITIAL to a dummy block.
35127 (read_rtl_function_body_from_file_range): New function.
35128 * read-rtl-function.h (read_rtl_function_body_from_file_range):
35130 * run-rtl-passes.c: New file.
35131 * run-rtl-passes.h: New file.
35133 2017-01-24 Jeff Law <law@redhat.com>
35135 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
35138 2017-01-24 Bin Cheng <bin.cheng@arm.com>
35140 PR tree-optimization/79159
35141 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
35142 (record_nonwrapping_iv): Improve boundary using above function if no
35143 value range information.
35145 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
35146 Martin Jambor <mjambor@suse.cz>
35148 * brig-builtins.def: New file.
35149 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
35150 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
35151 (DEF_HSAIL_SAT_BUILTIN): Likewise.
35152 (DEF_HSAIL_INTR_BUILTIN): Likewise.
35153 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
35154 * builtin-types.def (BT_INT8): New.
35155 (BT_INT16): Likewise.
35156 (BT_UINT8): Likewise.
35157 (BT_UINT16): Likewise.
35158 (BT_FN_ULONG): Likewise.
35159 (BT_FN_UINT_INT): Likewise.
35160 (BT_FN_UINT_ULONG): Likewise.
35161 (BT_FN_UINT_LONG): Likewise.
35162 (BT_FN_UINT_PTR): Likewise.
35163 (BT_FN_ULONG_PTR): Likewise.
35164 (BT_FN_INT8_FLOAT): Likewise.
35165 (BT_FN_INT16_FLOAT): Likewise.
35166 (BT_FN_UINT32_FLOAT): Likewise.
35167 (BT_FN_UINT16_FLOAT): Likewise.
35168 (BT_FN_UINT8_FLOAT): Likewise.
35169 (BT_FN_UINT64_FLOAT): Likewise.
35170 (BT_FN_UINT16_UINT32): Likewise.
35171 (BT_FN_UINT32_UINT16): Likewise.
35172 (BT_FN_UINT16_UINT16_UINT16): Likewise.
35173 (BT_FN_INT_PTR_INT): Likewise.
35174 (BT_FN_UINT_PTR_UINT): Likewise.
35175 (BT_FN_LONG_PTR_LONG): Likewise.
35176 (BT_FN_ULONG_PTR_ULONG): Likewise.
35177 (BT_FN_VOID_UINT64_UINT64): Likewise.
35178 (BT_FN_UINT8_UINT8_UINT8): Likewise.
35179 (BT_FN_INT8_INT8_INT8): Likewise.
35180 (BT_FN_INT16_INT16_INT16): Likewise.
35181 (BT_FN_INT_INT_INT): Likewise.
35182 (BT_FN_UINT_FLOAT_UINT): Likewise.
35183 (BT_FN_FLOAT_UINT_UINT): Likewise.
35184 (BT_FN_ULONG_UINT_UINT): Likewise.
35185 (BT_FN_ULONG_UINT_PTR): Likewise.
35186 (BT_FN_ULONG_ULONG_ULONG): Likewise.
35187 (BT_FN_UINT_UINT_UINT): Likewise.
35188 (BT_FN_VOID_UINT_PTR): Likewise.
35189 (BT_FN_UINT_UINT_PTR: Likewise.
35190 (BT_FN_UINT32_UINT64_PTR): Likewise.
35191 (BT_FN_INT_INT_UINT_UINT): Likewise.
35192 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
35193 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
35194 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
35195 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
35196 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
35197 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
35198 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
35199 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
35200 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
35201 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
35202 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
35203 * doc/frontends.texi: List BRIG FE.
35204 * doc/install.texi (Testing): Add BRIG tesring requirements.
35205 * doc/invoke.texi (Overall Options): Mention BRIG.
35206 * doc/standards.texi (Standards): Doucment BRIG HSA version.
35208 2017-01-24 Richard Biener <rguenther@suse.de>
35210 PR translation/79208
35211 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
35213 2017-01-24 Martin Jambor <mjambor@suse.cz>
35216 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
35217 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
35218 and known_contexts.
35220 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
35222 PR middle-end/79123
35223 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
35224 casts from signed to unsigned really don't have a range.
35226 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
35228 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
35229 GMP_RNDx for compatiblity.
35231 2017-01-24 Martin Liska <mliska@suse.cz>
35234 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
35235 that would prevent us to call alloca with -1 as argument.
35237 2017-01-24 Jakub Jelinek <jakub@redhat.com>
35239 * dwarf2out.c (output_compilation_unit_header, output_file_names):
35240 Avoid -Wformat-security warning.
35242 2017-01-23 Andrew Pinski <apinski@cavium.com>
35244 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
35247 2017-01-23 Martin Sebor <msebor@redhat.com>
35249 PR middle-end/78703
35250 * gimple-ssa-sprintf.c (warn_level): New global.
35251 (format_integer): Use it here and throughout the rest of the file.
35252 Use the same switch to compute sign as base.
35253 (maybe_warn): New function.
35254 (format_directive): Factor out warnings into maybe_warn.
35255 Add debugging output. Use warn_level.
35256 (add_bytes): Use warn_level.
35257 (pass_sprintf_length::compute_format_length): Add debugging output.
35258 (try_substitute_return_value): Same.
35259 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
35261 PR middle-end/78703
35262 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
35263 (struct fmtresult, format_integer, format_floating): Adjust.
35264 (fmtresult::fmtresult): Set max correctly in two argument ctor.
35265 (get_string_length, format_string,format_directive): Same.
35266 (pass_sprintf_length::compute_format_length): Same.
35267 (try_substitute_return_value): Simplify slightly.
35269 PR middle-end/78703
35270 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
35271 (fmtresult::operator+=): Outlined.
35272 (struct fmtresult): Add ctors.
35273 (struct conversion_spec): Rename...
35274 (struct directive): ...to this. Add and remove data members.
35275 (directive::set_width, directive::set_precision): New functions.
35276 (format_percent): Use fmtresult ctor.
35277 (get_width_and_precision): Remove.
35278 (format_integer): Make naming changes. Avoid computing width and
35280 (format_floating): Same. Adjust indentation.
35281 (format_character, format_none): New functions.
35282 (format_string): Moved character handling to format_character.
35283 (format_directive): Remove arguments, change return type.
35284 (parse_directive): New function.
35285 (pass_sprintf_length::compute_format_length): Move directive
35286 parsing to parse_directive.
35288 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35290 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
35291 (assign_assembler_name_if_needed): ... this.
35292 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
35293 (assign_assembler_name_if_needed): ... this.
35294 (free_lang_data_in_cgraph): Adjust callers.
35295 * cgraphunit.c (cgraph_node::analyze): Likewise.
35296 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
35299 2017-01-23 Richard Biener <rguenther@suse.de>
35301 PR tree-optimization/79088
35302 PR tree-optimization/79188
35303 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
35304 resetting loop bounds after last path deletion. Reset loop
35305 bounds of the target loop, make code match the comments.
35306 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
35307 Make sure loops need no fixups.
35309 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35311 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
35312 exponent support with double type for first argument.
35313 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
35314 type returned by __builtin_vec_extract_sig,
35315 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
35316 functions from "vector int" to "vector unsigned int" or from
35317 "vector long long int" to "vector unsigned long long int".
35318 Changed type returned by __builtin_vec_extract_exp,
35319 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
35320 functions from "vector int" to "vector unsigned int" or from
35321 "vector long long int" to "vector unsigned long long int".
35322 Changed return type of __builtin_vec_test_data_class,
35323 __builtin_vec_test_data_class_sp, and
35324 __builtin_vec_test_data_class_dp from "vector int" to
35325 "vector bool int" or from "vector long long int" to "vector bool
35326 long long int" and changed second argument type from "unsigned
35327 int" to "int". Added new overloaded function forms "vector float
35328 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
35329 "vector float __builtin_vec_insert_exp_sp (vector float, vector
35330 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
35331 double, vector unsigned long long int)" and "vector double
35332 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
35333 long int)". Changed return type of
35334 __builtin_scalar_test_data_class and
35335 __builtin_scalar_test_data_class_sp and
35336 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
35337 int" and changed second argument from "unsigned int" to "int".
35338 Changed type returned by __builtin_scalar_test_neg,
35339 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
35340 from "int" to "bool int". Added new overloaded function form
35341 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
35342 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
35343 exponent double-precision with floating point first argument.
35344 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
35345 documentation of scalar_test_data_class, scalar_test_neg,
35346 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
35347 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
35348 vec_test_data_class built-in functions to reflect refinements in
35349 their type signatures.
35351 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
35353 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
35355 (aarch64_elf_asm_destructor): Likewise.
35357 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
35359 PR rtl-optimization/78634
35360 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
35361 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
35362 * ifcvt.c (noce_try_cmove): Add missing cost check.
35364 PR rtl-optimization/71724
35365 * combine.c (if_then_else_cond): Look for situations where it is
35366 beneficial to undo the work of one of the recursive calls.
35368 2017-01-23 Bin Cheng <bin.cheng@arm.com>
35370 PR tree-optimization/70754
35371 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
35372 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
35373 combined stmt before it if not NULL.
35374 (combine_chains): Process refs reversely and compute dominance point
35377 2017-01-23 Martin Liska <mliska@suse.cz>
35379 PR tree-optimization/79196
35380 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
35381 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
35383 (strlen_optimize_stmt): Call the renamed function.
35385 2017-01-23 Michael Matz <matz@suse.de>
35387 PR tree-optimization/78384
35388 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
35390 2017-01-23 Richard Biener <rguenther@suse.de>
35392 PR tree-optimization/79186
35393 * tree-vrp.c (register_new_assert_for): Make sure we've seen
35394 both incoming edges before moving an assert.
35396 2017-01-23 Martin Jambor <mjambor@suse.cz>
35398 * ipa-prop.c (load_from_param_1): Removed.
35399 (load_from_unmodified_param): Bits from load_from_param_1 put back
35401 (load_from_param): Removed.
35402 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
35403 with stmt. Reverted back to use of load_from_unmodified_param.
35405 2017-01-23 Martin Jambor <mjambor@suse.cz>
35408 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
35409 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
35410 field a pointer to garbage collected vector, mark lattices and
35411 ipcp_orig_node with GTY((skip)).
35412 (ipa_get_param_count): Adjust to descriptors being a pointer.
35413 (ipa_get_param): Likewise.
35414 (ipa_get_type): Likewise.
35415 (ipa_get_param_move_cost): Likewise.
35416 (ipa_set_param_used): Likewise.
35417 (ipa_get_controlled_uses): Likewise.
35418 (ipa_set_controlled_uses): Likewise.
35419 (ipa_is_param_used): Likewise.
35420 (ipa_node_params_t): Move into garbage collector. New methods insert
35422 (ipa_node_params_sum): Annotate wth GTY(()).
35423 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
35425 (ipa_load_from_parm_agg): Adjust declaration.
35426 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
35427 * ipa-profile.c (ipa_profile): Likewise.
35428 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
35429 (ipa_populate_param_decls): Make descriptors parameter garbage
35431 (ipa_dump_param): Adjust to descriptors being a pointer.
35432 (ipa_alloc_node_params): Likewise.
35433 (ipa_initialize_node_params): Likewise.
35434 (load_from_param_1): Make descriptors parameter garbage collected.
35435 (load_from_unmodified_param): Likewise.
35436 (load_from_param): Likewise.
35437 (ipa_load_from_parm_agg): Likewise.
35438 (ipa_node_params::~ipa_node_params): Removed.
35439 (ipa_free_all_node_params): Remove call to delete operator.
35440 (ipa_node_params_t::insert): New.
35441 (ipa_node_params_t::remove): Likewise.
35442 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
35443 copy known_csts and known_contexts vectors.
35444 (ipa_read_node_info): Adjust to descriptors being a pointer.
35445 (ipcp_modif_dom_walker): Make m_descriptors field garbage
35447 (ipcp_transform_function): Make descriptors variable garbage
35450 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
35452 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
35453 * config/i386/avx512dqintrin.h: Ditto.
35454 * config/i386/avx512fintrin.h: Ditto.
35455 * config/i386/i386.c: Handle new builtins.
35456 * config/i386/i386-builtin.def: Add new builtins.
35457 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
35458 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
35460 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35461 Martin Liska <mliska@suse.cz>
35463 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
35464 * asan.c (asan_expand_poison_ifn): Support stores and use
35465 appropriate ASAN report function.
35466 * internal-fn.c (expand_ASAN_POISON_USE): New function.
35467 * internal-fn.def (ASAN_POISON_USE): Declare.
35468 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
35469 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
35470 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
35471 ASAN_POISON calls w/o LHS.
35472 * tree-ssa.c (execute_update_addresses_taken): Create clobber
35473 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
35474 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
35475 * gimplify.c (asan_poison_variables): Add attribute
35476 use_after_scope_memory to variables that really needs to live
35478 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
35479 having the attribute.
35481 2017-01-23 Martin Liska <mliska@suse.cz>
35483 * asan.c (create_asan_shadow_var): New function.
35484 (asan_expand_poison_ifn): Likewise.
35485 * asan.h (asan_expand_poison_ifn): New declaration.
35486 * internal-fn.c (expand_ASAN_POISON): Likewise.
35487 * internal-fn.def (ASAN_POISON): New builtin.
35488 * sanopt.c (pass_sanopt::execute): Expand
35489 asan_expand_poison_ifn.
35490 * tree-inline.c (copy_decl_for_dup_finish): Make function
35492 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
35493 * tree-ssa.c (is_asan_mark_p): New function.
35494 (execute_update_addresses_taken): Rewrite local variables
35495 (identified just by use-after-scope as addressable) into SSA.
35497 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35499 * doc/install.texi (Specific): opensource.apple.com uses https
35500 now. Remove trailing slash.
35502 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35504 * README.Portability: Remove note on an Irix compatibility issue.
35506 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
35508 * gcov.c (INCLUDE_ALGORITHM): Define.
35509 (INCLUDE_VECTOR): Define.
35510 No longer include <vector> and <algorithm> directly.
35512 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
35514 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
35516 * doc/invoke.texi (Code Gen Options): Ditto.
35518 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
35521 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
35523 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
35525 rtl-optimization/79125
35526 * cprop.c (local_cprop_pass): Handle cases where we make an
35527 unconditional trap.
35529 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
35533 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
35534 read from, for big endian.
35536 2017-01-20 Jiong Wang <jiong.wang@arm.com>
35538 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
35539 register pauth builtins for LP64 only.
35541 2017-01-20 Marek Polacek <polacek@redhat.com>
35544 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
35547 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35549 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
35551 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
35552 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
35553 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
35555 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35558 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
35559 in big-endian mode when they are not a single duplicated value.
35561 2017-01-20 Richard Biener <rguenther@suse.de>
35563 * BASE-VER: Bump to 7.0.1.
35565 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35567 * omp-low.c (omplow_simd_context): New struct. Use it...
35568 (lower_rec_simd_input_clauses): ...here and...
35569 (lower_rec_input_clauses): ...here to hold common data. Adjust all
35570 references to idx, lane, max_vf, is_simt.
35572 2017-01-20 Graham Markall <graham.markall@embecosm.com>
35574 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
35577 2017-01-20 Martin Jambor <mjambor@suse.cz>
35579 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
35580 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
35582 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
35583 (GTFILES): Rename hsa.c to hsa-common.c.
35584 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
35585 * hsa-dump.c: Likewise.
35586 * hsa-gen.c: Likewise.
35587 * hsa-regalloc.c: Likewise.
35588 * ipa-hsa.c: Likewise.
35589 * omp-expand.c: Likewise.
35590 * omp-low.c: Likewise.
35591 * toplev.c: Likewise.
35593 2017-01-20 Marek Polacek <polacek@redhat.com>
35596 * doc/invoke.texi: Document -Wduplicated-branches.
35597 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
35598 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
35599 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
35600 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
35601 return 0 only when not OEP_LEXICOGRAPHIC.
35602 (fold_build_cleanup_point_expr): Use the expression
35603 location when building CLEANUP_POINT_EXPR.
35604 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
35605 * tree.c (add_expr): Handle error_mark_node.
35607 2017-01-20 Martin Liska <mliska@suse.cz>
35610 * tree-profile.c (init_ic_make_global_vars): Do not call
35612 (gimple_init_gcov_profiler): Likewise.
35614 2017-01-20 Martin Liska <mliska@suse.cz>
35617 * cgraph.h (maybe_create_reference): Remove argument and
35619 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
35621 * ipa-cp.c (create_specialized_node): Likewise.
35622 * symtab.c (symtab_node::maybe_create_reference): Handle
35623 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
35625 2017-01-20 Martin Liska <mliska@suse.cz>
35627 * read-rtl-function.c (function_reader::create_function): Use
35628 build_decl instread of build_decl_stat.
35630 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
35632 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
35633 * config/i386/avx512dqintrin.h: Ditto.
35634 * config/i386/avx512fintrin.h: Ditto.
35635 * config/i386/i386-builtin-types.def: Add new types.
35636 * config/i386/i386.c: Handle new types.
35637 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
35638 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
35639 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
35640 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
35641 (__builtin_ia32_kshiftridi): New.
35642 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
35644 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
35648 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
35649 define to rs6000_init_stack_protect_guard.
35650 (rs6000_init_stack_protect_guard): New function.
35652 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
35653 Yunqiang Su <yunqiang.su@imgtec.com>
35655 * config.gcc (supported_defaults): Add madd4.
35656 (with_madd4): Add validation.
35657 (all_defaults): Add madd4.
35658 * config/mips/mips.opt (mmadd4): New option.
35659 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35661 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35663 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
35664 (ISA_HAS_FUSED_MADD4): Likewise.
35665 * doc/invoke.texi (-mmadd4): Document the new option.
35666 * doc/install.texi (--with-madd4): Document the new option.
35668 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35670 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
35671 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
35672 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
35673 (aarch64_init_pauth_hint_builtins): New.
35674 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
35675 (aarch64_expand_builtin): Expand new builtins.
35677 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35679 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
35680 * combine-stack-adj.c (no_unhandled_cfa): Handle
35681 REG_CFA_TOGGLE_RA_MANGLE.
35682 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
35683 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
35684 info for return address signing.
35685 (aarch64_expand_epilogue): Likewise.
35687 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35689 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
35690 * config/aarch64/aarch64-protos.h
35691 (aarch64_return_address_signing_enabled): New declaration.
35692 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
35694 (aarch64_expand_prologue): Sign return address before it's pushed onto
35696 (aarch64_expand_epilogue): Authenticate return address fetched from
35698 (aarch64_override_options): Sanity check for ILP32 and ISA level.
35699 (aarch64_attributes): New function attributes for "sign-return-address".
35700 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
35701 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
35702 ("*do_return"): Generate combined instructions according to key index.
35703 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
35704 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
35706 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
35707 * config/aarch64/aarch64.opt (msign-return-address=): New.
35708 * doc/extend.texi (AArch64 Function Attributes): Documents
35709 "sign-return-address=".
35710 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
35712 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
35714 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
35715 overall option summary.
35717 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35719 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
35720 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
35721 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
35722 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
35724 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
35726 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
35727 -mpower9-minmax by default for -mcpu=power9.
35728 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
35729 128-bit floating point.
35731 2017-01-20 Alan Modra <amodra@gmail.com>
35733 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
35734 optimizing for size.
35736 2017-01-20 Alan Modra <amodra@gmail.com>
35739 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
35740 for strcmp and strncmp from corresponding builtin decl.
35742 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
35744 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
35745 instead of i386/rtems-64.h.
35746 * config/i386/rtems-64.h: Remove.
35748 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
35752 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
35754 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
35756 2017-01-19 Tamar Christina <tamar.christina@arm.com>
35758 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
35759 Change int to HOST_WIDE_INT.
35760 * config/aarch64/aarch64-protos.h
35761 (aarch64_simd_gen_const_vector_dup): Likewise.
35762 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
35764 2017-01-19 David Malcolm <dmalcolm@redhat.com>
35766 * langhooks-def.h (lhd_type_for_size): New decl.
35767 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
35768 * langhooks.c (lhd_type_for_size): New function, taken from
35771 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
35773 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
35774 define_bypass for CR latency.
35775 (power9-cracked-alu): Update bypass latency and remove power9-branch.
35776 (power9-alu2): Add define_bypass for CR latency.
35778 (power9-mul): Update insn latency.
35779 (power9-mul-compare): Update insn latency, bypass latency and remove
35782 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35784 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
35786 * config/aarch64/aarch64.md
35787 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
35788 aarch64_nopcrelative_literal_loads.
35789 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
35791 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
35793 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
35794 TARGET_LOONGSON_3A.
35795 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
35797 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
35800 * config.gcc (supported_defaults): Add lxc1-sxc1.
35801 (with_lxc1_sxc1): Add validation.
35802 (all_defaults): Add lxc1-sxc1.
35803 * config/mips/mips.opt (mlxc1-sxc1): New option.
35804 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35806 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35807 __mips_no_lxc1_sxc1.
35808 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
35809 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
35810 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
35812 2017-01-19 Richard Biener <rguenther@suse.de>
35814 PR tree-optimization/72488
35815 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
35816 sure to restore SSA info.
35817 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
35819 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
35821 PR rtl-optimization/79121
35822 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
35823 of the inner type when shifting an extended value.
35825 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
35828 * symtab.c (symtab_node::equal_address_to): Fix comparing of
35829 interposable aliases.
35831 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
35834 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
35835 Use the evmergelohi instruction.
35836 (mov_si<mode>_e500_subreg4_2_le): Likewise.
35837 (mov_sitf_e500_subreg8_2_be): Likewise.
35838 (mov_sitf_e500_subreg12_2_le): Likewise.
35839 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
35840 (mov_si<mode>_e500_subreg4_2_be): Likewise.
35841 (mov_sitf_e500_subreg8_2_le): Likewise.
35842 (mov_sitf_e500_subreg12_2_be): Likewise.
35844 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35846 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
35847 attribute from vecsimple to vecperm.
35848 (altivec_vbpermq2): Likewise.
35850 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35853 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
35855 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35856 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
35857 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
35858 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
35859 case where N arg is SIZE_MAX.
35860 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
35861 (cmpstrsi): Add pattern.
35863 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
35865 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
35866 __builtin_vec_revb builtins.
35867 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
35868 built-in functions to support generation of the ISA 3.0 XXBR<x>
35869 vector byte reverse instructions.
35870 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
35871 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
35872 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
35873 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
35874 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
35875 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
35876 (P9V_BUILTIN_VEC_REVB): Likewise.
35877 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
35878 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
35879 (p9_xxbrq_v16qi): Likewise.
35880 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
35881 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
35882 (p9_xxbrh_v8hi): Likewise.
35883 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
35884 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
35885 vec_revb built-in functions.
35887 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
35889 PR rtl-optimization/78952
35890 * config/i386/i386.md (any_extract): New code iterator.
35891 (*insvqi_2): Use any_extract for source operand.
35892 (*insvqi_3): Use any_shiftrt for source operand.
35894 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
35896 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
35898 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
35900 2017-01-18 Matthias Klose <doko@ubuntu.com>
35902 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
35904 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35906 * config/rs6000/altivec.h (vec_bperm): Change #define.
35907 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
35908 (altivec_vbpermq2): New define_insn.
35909 (altivec_vbpermd): Likewise.
35910 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
35911 function interface.
35912 (VBPERMD): Likewise.
35913 (VBPERM): New polymorphic function interface.
35914 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
35915 Add entries for P9V_BUILTIN_VEC_VBPERM.
35916 * doc/extend.texi: Add interfaces for vec_bperm.
35918 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35920 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
35921 first letter of error messages.
35922 (s390_resolve_overloaded_builtin): Likewise.
35923 * config/s390/s390.c (s390_expand_builtin): Likewise.
35924 (s390_invalid_arg_for_unprototyped_fn): Likewise.
35925 (s390_valid_target_attribute_inner_p): Likewise.
35926 * config/s390/s390.md ("tabort"): Likewise.
35928 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
35930 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
35931 (ISA_AVOID_DIV_HILO): New macro.
35932 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
35933 (ISA_HAS_DDIV): Likewise.
35935 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35937 * doc/invoke.texi (fabi-version): Correct number of occurrences.
35939 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35941 * doc/invoke.texi (fabi-version): Spelling fix.
35943 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35946 * doc/invoke.texi (fabi-version): Mention mangling fix for
35949 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35952 * doc/invoke.texi (fabi-version): Document discriminator mangling.
35954 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
35957 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
35958 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
35960 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
35962 (stack_protect_test): Ditto.
35963 * config/rs6000/rs6000.opt (mstack-protector-guard=,
35964 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
35966 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
35967 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
35968 -mstack-protector-guard-offset=.
35969 (RS/6000 and PowerPC Options): Ditto.
35971 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
35973 * config/i386/i386.h (MASK_CLASS_P): New define.
35974 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
35975 there are no registers from different register sets also when
35976 mask registers are used. Update function comment.
35977 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
35978 to (*k/*r) and (*k/*km) alternatives.
35980 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
35982 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
35983 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
35984 (EH_RETURN_HANDLER_RTX): New define.
35985 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
35986 Force frame pointer in EH return functions.
35987 (aarch64_expand_epilogue): Add barrier for eh_return.
35988 (aarch64_final_eh_return_addr): Remove.
35989 (aarch64_eh_return_handler_rtx): New function.
35990 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
35992 (aarch64_eh_return_handler_rtx): New prototype.
35994 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35996 * config/rs6000/altivec.h (vec_rlmi): New #define.
35997 (vec_vrlnm): Likewise.
35998 (vec_rlnm): Likewise.
35999 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
36000 (UNSPEC_VRLNM): Likewise.
36001 (VIlong): New mode iterator.
36002 (altivec_vrl<VI_char>mi): New define_insn.
36003 (altivec_vrl<VI_char>nm): Likewise.
36004 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
36006 (VRLDNM): Likewise.
36007 (RLNM): New polymorphic function entry.
36008 (VRLWMI): New monomorphic function entry.
36009 (VRLDMI): Likewise.
36010 (RLMI): New polymorphic function entry.
36011 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
36012 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
36013 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
36016 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36019 * dwarf2out.c (field_byte_offset): Restore the
36020 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
36021 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
36022 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
36025 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36028 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
36030 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36033 * dwarf2out.c (add_data_member_location_attribute): For constant
36034 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
36035 instead of DW_AT_data_member_location, DW_AT_bit_offset and
36036 DW_AT_byte_size attributes.
36038 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36040 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
36041 after forcing to constant memory when the code model is medium.
36043 2017-01-17 Julia Koval <julia.koval@intel.com>
36046 * config/i386/avx512fintrin.h
36047 (_mm512_i32gather_ps): Change __addr type to void const*.
36048 (_mm512_mask_i32gather_ps): Ditto.
36049 (_mm512_i32gather_pd): Ditto.
36050 (_mm512_mask_i32gather_pd): Ditto.
36051 (_mm512_i64gather_ps): Ditto.
36052 (_mm512_mask_i64gather_ps): Ditto.
36053 (_mm512_i64gather_pd): Ditto.
36054 (_mm512_mask_i64gather_pd): Ditto.
36055 (_mm512_i32gather_epi32): Ditto.
36056 (_mm512_mask_i32gather_epi32): Ditto.
36057 (_mm512_i32gather_epi64): Ditto.
36058 (_mm512_mask_i32gather_epi64): Ditto.
36059 (_mm512_i64gather_epi32): Ditto.
36060 (_mm512_mask_i64gather_epi32): Ditto.
36061 (_mm512_i64gather_epi64): Ditto.
36062 (_mm512_mask_i64gather_epi64): Ditto.
36063 (_mm512_i32scatter_ps): Change __addr type to void*.
36064 (_mm512_mask_i32scatter_ps): Ditto.
36065 (_mm512_i32scatter_pd): Ditto.
36066 (_mm512_mask_i32scatter_pd): Ditto.
36067 (_mm512_i64scatter_ps): Ditto.
36068 (_mm512_mask_i64scatter_ps): Ditto.
36069 (_mm512_i64scatter_pd): Ditto.
36070 (_mm512_mask_i64scatter_pd): Ditto.
36071 (_mm512_i32scatter_epi32): Ditto.
36072 (_mm512_mask_i32scatter_epi32): Ditto.
36073 (_mm512_i32scatter_epi64): Ditto.
36074 (_mm512_mask_i32scatter_epi64): Ditto.
36075 (_mm512_i64scatter_epi32): Ditto.
36076 (_mm512_mask_i64scatter_epi32): Ditto.
36077 (_mm512_i64scatter_epi64): Ditto.
36078 (_mm512_mask_i64scatter_epi64): Ditto.
36079 * config/i386/avx512pfintrin.h
36080 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
36081 (_mm512_mask_prefetch_i32gather_ps): Ditto.
36082 (_mm512_mask_prefetch_i64gather_pd): Ditto.
36083 (_mm512_mask_prefetch_i64gather_ps): Ditto.
36084 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
36085 (_mm512_prefetch_i32scatter_ps): Ditto.
36086 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
36087 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
36088 (_mm512_prefetch_i64scatter_pd): Ditto.
36089 (_mm512_prefetch_i64scatter_ps): Ditto.
36090 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
36091 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
36092 * config/i386/avx512vlintrin.h
36093 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
36094 (_mm_mmask_i32gather_ps): Ditto.
36095 (_mm256_mmask_i32gather_pd): Ditto.
36096 (_mm_mmask_i32gather_pd): Ditto.
36097 (_mm256_mmask_i64gather_ps): Ditto.
36098 (_mm_mmask_i64gather_ps): Ditto.
36099 (_mm256_mmask_i64gather_pd): Ditto.
36100 (_mm_mmask_i64gather_pd): Ditto.
36101 (_mm256_mmask_i32gather_epi32): Ditto.
36102 (_mm_mmask_i32gather_epi32): Ditto.
36103 (_mm256_mmask_i32gather_epi64): Ditto.
36104 (_mm_mmask_i32gather_epi64): Ditto.
36105 (_mm256_mmask_i64gather_epi32): Ditto.
36106 (_mm_mmask_i64gather_epi32): Ditto.
36107 (_mm256_mmask_i64gather_epi64): Ditto.
36108 (_mm_mmask_i64gather_epi64): Ditto.
36109 (_mm256_i32scatter_ps): Change __addr type to void*.
36110 (_mm256_mask_i32scatter_ps): Ditto.
36111 (_mm_i32scatter_ps): Ditto.
36112 (_mm_mask_i32scatter_ps): Ditto.
36113 (_mm256_i32scatter_pd): Ditto.
36114 (_mm256_mask_i32scatter_pd): Ditto.
36115 (_mm_i32scatter_pd): Ditto.
36116 (_mm_mask_i32scatter_pd): Ditto.
36117 (_mm256_i64scatter_ps): Ditto.
36118 (_mm256_mask_i64scatter_ps): Ditto.
36119 (_mm_i64scatter_ps): Ditto.
36120 (_mm_mask_i64scatter_ps): Ditto.
36121 (_mm256_i64scatter_pd): Ditto.
36122 (_mm256_mask_i64scatter_pd): Ditto.
36123 (_mm_i64scatter_pd): Ditto.
36124 (_mm_mask_i64scatter_pd): Ditto.
36125 (_mm256_i32scatter_epi32): Ditto.
36126 (_mm256_mask_i32scatter_epi32): Ditto.
36127 (_mm_i32scatter_epi32): Ditto.
36128 (_mm_mask_i32scatter_epi32): Ditto.
36129 (_mm256_i32scatter_epi64): Ditto.
36130 (_mm256_mask_i32scatter_epi64): Ditto.
36131 (_mm_i32scatter_epi64): Ditto.
36132 (_mm_mask_i32scatter_epi64): Ditto.
36133 (_mm256_i64scatter_epi32): Ditto.
36134 (_mm256_mask_i64scatter_epi32): Ditto.
36135 (_mm_i64scatter_epi32): Ditto.
36136 (_mm_mask_i64scatter_epi32): Ditto.
36137 (_mm256_i64scatter_epi64): Ditto.
36138 (_mm256_mask_i64scatter_epi64): Ditto.
36139 (_mm_i64scatter_epi64): Ditto.
36140 (_mm_mask_i64scatter_epi64): Ditto.
36141 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
36142 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
36143 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
36144 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
36145 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
36146 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
36147 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
36148 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
36149 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
36150 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
36151 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
36152 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
36153 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
36154 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
36155 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
36156 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
36157 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
36158 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
36159 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
36160 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
36161 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
36162 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
36163 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
36164 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
36165 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
36166 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
36167 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
36168 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
36169 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
36170 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
36171 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
36172 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
36173 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
36174 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
36175 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
36176 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
36177 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
36178 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
36179 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
36180 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
36181 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
36182 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
36183 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
36184 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
36185 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
36186 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
36187 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
36188 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
36189 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
36190 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
36191 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
36192 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
36193 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
36194 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
36195 definitions accordingly.
36197 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
36198 Kuan-Lin Chen <kuanlinchentw@gmail.com>
36201 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
36204 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
36207 * ira-conflicts.c (ira_build_conflicts): Update total conflict
36208 hard regs for inner regno.
36210 2017-01-17 Martin Liska <mliska@suse.cz>
36213 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
36214 assumption and add comment.
36216 2017-01-17 Nathan Sidwell <nathan@acm.org>
36218 * ipa-visibility.c (localize_node): New function, broken out of ...
36219 (function_and_variable_visibility): ... here. Call it.
36221 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36223 PR middle-end/77445
36224 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
36225 correctly set frequency of oudgoing edge.
36226 (duplicate_thread_path): Fix profile updating.
36228 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36231 * configure.ac: Add GCC_BASE_VER.
36232 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
36233 version from BASE-VER file.
36234 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
36235 (gcc.o): Depend on $(BASEVER).
36236 * common.opt (dumpfullversion): New option.
36237 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
36238 * doc/invoke.texi: Document -dumpfullversion.
36239 * doc/install.texi: Document --with-gcc-major-version-only.
36240 * configure: Regenerated.
36242 2017-01-17 Richard Biener <rguenther@suse.de>
36244 PR tree-optimization/71433
36245 * tree-vrp.c (register_new_assert_for): Merge same asserts
36246 on all incoming edges.
36247 (process_assert_insertions_for): Handle insertions at the
36250 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
36252 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
36253 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
36255 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
36258 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
36261 2017-01-17 Alan Modra <amodra@gmail.com>
36264 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
36265 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
36266 symbolic stack limit when pic.
36268 2017-01-16 Martin Sebor <msebor@redhat.com>
36270 PR tree-optimization/78608
36271 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
36273 2017-01-16 Jeff Law <law@redhat.com>
36276 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
36277 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
36278 for several include directories that may be relative to sysroot.
36279 * config/i386/x-mingw32 (gplus_includedir): Define.
36280 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
36281 (native_system_includedir): Likewise.
36282 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
36283 override if TARGET_SYSTEM_ROOT is defined.
36284 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
36286 PR tree-optimization/79090
36287 PR tree-optimization/33562
36288 PR tree-optimization/61912
36289 PR tree-optimization/77485
36290 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
36291 and computed trims into the dump file.
36293 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36295 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
36297 2017-01-16 Jakub Jelinek <jakub@redhat.com>
36300 * gimplify.c (gimplify_init_constructor): If want_value and
36301 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
36305 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
36306 sequence. Formatting fixes.
36307 (doloop_optimize): Formatting fixes.
36310 * gcc.c (debug_level_greater_than_spec_func): New function.
36311 (static_spec_functions): Add debug-level-gt spec function.
36312 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
36314 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
36315 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
36316 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
36317 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
36318 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
36319 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
36321 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
36323 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
36324 QImode fixups to general and mask registers only.
36326 2017-01-16 Carl Love <cel@us.ibm.com>
36328 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
36329 for built-in functions
36330 vector signed char vec_nabs (vector signed char)
36331 vector signed short vec_nabs (vector signed short)
36332 vector signed int vec_nabs (vector signed int)
36333 vector signed long long vec_nabs (vector signed long long)
36334 vector float vec_nabs (vector float)
36335 vector double vec_nabs (vector double)
36336 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
36338 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
36339 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
36340 * doc/extend.texi: Update the documentation file for the new built-in
36343 2017-01-16 Martin Sebor <msebor@redhat.com>
36345 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
36348 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36350 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
36351 UNSPEC_VSX__XXSPLTD to require special splat handling.
36353 2017-01-16 David Malcolm <dmalcolm@redhat.com>
36356 * system.h: Poison strndup.
36358 2017-01-16 Alan Modra <amodra@gmail.com>
36361 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
36364 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
36366 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
36368 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
36370 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
36371 call recog here. Assert that INSN_CODE (insn) is non-negative.
36373 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
36376 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
36378 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
36379 in the currently scheduled RTL fragment.
36381 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
36383 PR rtl-optimization/78751
36384 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
36387 2017-01-14 Jeff Law <law@redhat.com>
36389 PR tree-optimization/79090
36390 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
36391 variable length stores.
36392 (compute_trims): Delete dead assignment to *trim_tail.
36393 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
36396 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
36398 PR rtl-optimization/78626
36399 PR rtl-optimization/78727
36400 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
36401 of a block, and split such blocks after everything else is finished.
36403 2017-01-14 Alan Modra <amodra@gmail.com>
36406 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
36407 target legitimate_combined_insn.
36408 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
36409 (rs6000_legitimate_combined_insn): New function.
36410 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
36412 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
36413 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
36414 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
36416 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
36418 * doc/frontends.texi (G++ and GCC): Remove references to Java.
36420 2017-01-13 Jeff Law <law@redhat.com>
36422 PR tree-optimization/33562
36423 PR tree-optimization/61912
36424 PR tree-optimization/77485
36425 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
36427 (delete_dead_assignment): Likewise.
36428 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
36429 statement to delete_dead_call and delete_dead_assignment.
36431 2017-01-13 David Malcolm <dmalcolm@redhat.com>
36434 * substring-locations.c (format_warning_va): Strengthen case 1 so
36435 that both endpoints of the substring must be within the format
36436 range for just the substring to be printed.
36438 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
36440 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
36441 * config/i386/i386.c (ix86_target_string): Add missing options
36442 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
36443 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
36444 flags_other and ix86_target_other to flags2_other. Display unknown
36446 (ix86_valid_target_attribute_inner_p): Add missing options and
36447 reorder options by implied ISAs, as in ix86_target_string.
36449 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36451 * hash-table.h (hash_table::too_empty_p): New function.
36452 (hash_table::expand): Use it.
36453 (hash_table::traverse): Likewise.
36454 (hash_table::empty_slot): Use sizeof (value_type) instead of
36455 sizeof (PTR) to convert bytes to elements. Shrink the table
36456 if the current size is excessive for the current number of
36459 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36461 * ira-costs.c (record_reg_classes): Break from the inner loop
36462 early once alt_fail is known to be true. Update outer loop
36463 handling accordingly.
36465 2017-01-13 Jeff Law <law@redhat.com>
36467 * tree-ssa-dse.c (decrement_count): New function.
36468 (increment_start_addr, maybe_trim_memstar_call): Likewise.
36469 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
36470 when we know the partially dead statement is a mem* function.
36472 PR tree-optimization/61912
36473 PR tree-optimization/77485
36474 * tree-ssa-dse.c: Include expr.h.
36475 (maybe_trim_constructor_store): New function.
36476 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
36478 PR tree-optimization/33562
36479 PR tree-optimization/61912
36480 PR tree-optimization/77485
36481 * doc/invoke.texi: Document new dse-max-object-size param.
36482 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
36483 * tree-ssa-dse.c: Include params.h.
36484 (dse_store_status): New enum.
36485 (initialize_ao_ref_for_dse): New, partially extracted from
36487 (valid_ao_ref_for_dse, normalize_ref): New.
36488 (setup_live_bytes_from_ref, compute_trims): Likewise.
36489 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
36490 (maybe_trim_partially_dead_store): Likewise.
36491 (maybe_trim_complex_store): Likewise.
36492 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
36493 Track what bytes live from the original store. Return tri-state
36494 for dead, partially dead or live.
36495 (dse_dom_walker): Add constructor, destructor and new private members.
36496 (delete_dead_call, delete_dead_assignment): New extracted from
36498 (dse_optimize_stmt): Make a member of dse_dom_walker.
36499 Use initialize_ao_ref_for_dse.
36501 PR tree-optimization/33562
36502 PR tree-optimization/61912
36503 PR tree-optimization/77485
36504 * sbitmap.h (bitmap_count_bits): Prototype.
36505 (bitmap_clear_range, bitmap_set_range): Likewise.
36506 * sbitmap.c (bitmap_clear_range): New function.
36507 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
36509 2017-01-13 Martin Liska <mliska@suse.cz>
36512 * function.c (set_cfun): Add new argument force.
36513 * function.h (set_cfun): Likewise.
36514 * ipa-inline-transform.c (inline_call): Use the function when
36515 strict alising from is dropped for function we inline to.
36517 2017-01-13 Richard Biener <rguenther@suse.de>
36519 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
36520 for dumping GIMPLE INTEGER_CSTs.
36522 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
36524 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
36525 to 201112L since C++17.
36527 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
36530 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
36531 if -fsanitize=kernel-address is present.
36533 2017-01-13 Richard Biener <rguenther@suse.de>
36535 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
36536 as _Literal ( type ) number in case usual suffixes do not
36537 preserve all information.
36539 2017-01-13 Richard Biener <rguenther@suse.de>
36541 PR tree-optimization/77283
36542 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
36543 and ssa-iterators.h.
36544 (is_feasible_trace): Implement a cost model based on joiner
36547 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
36550 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
36551 char or short to __float128/_Float128 directly.
36553 2017-01-12 Martin Sebor <msebor@redhat.com>
36555 to -Wformat-overflow.
36556 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
36557 (min_bytes_remaining): Same.
36558 (get_string_length): Same.
36559 (format_string): Same.
36560 (format_directive): Same.
36562 (pass_sprintf_length::handle_gimple_call): Same.
36564 2017-01-12 Jakub Jelinek <jakub@redhat.com>
36566 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
36567 info.nowrite calls with no lhs that can't throw. Return bool
36568 whether gsi_remove has been called or not.
36569 (pass_sprintf_length::handle_gimple_call): Return bool whether
36570 try_substitute_return_value called gsi_remove. Formatting fix.
36571 (pass_sprintf_length::execute): Don't use gsi_remove if
36572 handle_gimple_call returned true.
36575 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
36576 be removed due to side-effects, don't remove following barrier nor
36577 turn the successor edge into fallthru edge.
36579 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36582 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
36583 element-reversing loads and stores as not swappable.
36585 2017-01-12 Nathan Sidwell <nathan@acm.org>
36586 Nicolai Stange <nicstange@gmail.com>
36588 * combine.c (try_combine): Don't ignore result of overlap checking
36589 loop. Combine overlap & asm check into single loop.
36591 2017-01-12 Richard Biener <rguenther@suse.de>
36593 * tree-pretty-print.c (dump_generic_node): Provide -gimple
36594 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
36596 2017-01-12 Richard Biener <rguenther@suse.de>
36598 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
36599 and TS_TARGET_OPTION directly derive from TS_BASE.
36600 * tree-core.h (tree_optimization_option): Derive from tree_base.
36601 (tree_target_option): Likewise.
36603 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
36605 * config/i386/i386.c (memory_address_length): Increase len
36606 only when rip_relative_addr_p returns false.
36608 2017-01-11 Julia Koval <julia.koval@intel.com>
36610 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
36611 (OPTION_MASK_ISA_SGX_SET): New.
36612 (ix86_handle_option): Handle OPT_msgx.
36613 * config.gcc: Added sgxintrin.h.
36614 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
36615 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
36616 * config/i386/i386.c (ix86_target_string): Add -msgx.
36618 (ix86_option_override_internal): Handle new options.
36619 (ix86_valid_target_attribute_inner_p): Add sgx.
36620 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
36621 * config/i386/i386.opt: Add msgx.
36622 * config/i386/sgxintrin.h: New file.
36623 * config/i386/x86intrin.h: Add sgxintrin.h.
36625 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36628 * fold-const.c (maybe_nonzero_address): Return 1 for function
36630 (tree_single_nonzero_warnv_p): Don't handle function local objects
36634 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
36637 2017-01-11 David Malcolm <dmalcolm@redhat.com>
36640 * opts.c: Include "spellcheck.h"
36641 (struct string_fragment): New struct.
36642 (struct edit_distance_traits<const string_fragment &>): New
36644 (get_closest_sanitizer_option): New function.
36645 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
36647 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36649 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
36651 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
36652 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
36653 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
36654 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
36655 for initial die_offset if dwarf_split_debug_info.
36656 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
36657 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
36659 (output_skeleton_debug_sections): Formatting fix. Use
36660 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
36661 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
36663 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
36665 * config/arm/cortex-a53.md: Add bypasses for
36666 cortex_a53_r2f_cvt.
36667 (cortex_a53_r2f): Only use for transfers.
36668 (cortex_a53_f2r): Likewise.
36669 (cortex_a53_r2f_cvt): Add reservation for conversions.
36670 (cortex_a53_f2r_cvt): Likewise.
36672 2017-01-11 Tamar Christina <tamar.christina@arm.com>
36674 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
36675 to all inlined functions, change static to extern.
36677 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
36680 * config/arm/arm.c (legitimize_pic_address): Handle reference to
36682 (arm_assemble_integer): Likewise.
36684 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
36686 * config.gcc: Use new awk script to check CPU, FPU and architecture
36687 parameters for --with-... options.
36688 * config/arm/parsecpu.awk: New file
36689 * config/arm/arm-cpus.in: New file.
36690 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
36692 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
36694 * config/arm/t-arm: Update dependency rules.
36695 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
36696 of processing .def files.
36697 * config/arm/genopt.sh: Deleted.
36698 * config/arm/gentune.sh: Deleted.
36699 * config/arm/arm-cores.def: Deleted.
36700 * config/arm/arm-arches.def: Deleted.
36701 * config/arm/arm-fpus.def: Deleted.
36702 * config/arm/arm-tune.md: Regenerated.
36703 * config/arm/arm-tables.opt: Regenerated.
36704 * config/arm/arm-cpu.h: New generated file.
36705 * config/arm/arm-cpu-data.h: New generated file.
36706 * config/arm/arm-cpu-cdata.h: New generated file.
36708 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
36711 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
36713 (input_varpool_node): Unpack dynamically_initialized bit.
36715 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
36717 PR rtl-optimization/79032
36718 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
36719 the alignment of the adjusted memory reference against that of MODE,
36720 instead of the alignment of the original memory reference.
36722 2017-01-11 Martin Jambor <mjambor@suse.cz>
36724 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
36726 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
36727 decorated functions.
36729 2017-01-11 Richard Biener <rguenther@suse.de>
36731 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
36732 set range/nonnull info for PHI results. Do not set it on
36733 stmts marked for removal.
36735 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
36737 * expr.c (store_field): In the bitfield case, fetch the return value
36738 from the registers before applying a single big-endian adjustment.
36739 Always do a final load for a BLKmode value not larger than a word.
36741 2017-01-10 David Malcolm <dmalcolm@redhat.com>
36744 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
36745 that we correctly handle column numbers greater than
36746 LINE_MAP_MAX_COLUMN_NUMBER.
36748 2017-01-10 Martin Sebor <msebor@redhat.com>
36750 PR middle-end/78245
36751 * gimple-ssa-sprintf.c (get_destination_size): Call
36752 {init,fini}object_sizes.
36753 * tree-object-size.c (addr_object_size): Adjust.
36754 (pass_through_call): Adjust.
36755 (pass_object_sizes::execute): Adjust.
36756 * tree-object-size.h (fini_object_sizes): Declare.
36758 2017-01-10 Martin Sebor <msebor@redhat.com>
36760 PR tree-optimization/78775
36761 * builtins.c (get_size_range): Move...
36762 * calls.c: ...to here.
36763 (alloc_max_size): Accept zero argument.
36764 (operand_signed_p): Remove.
36765 (maybe_warn_alloc_args_overflow): Call get_size_range.
36766 * calls.h (get_size_range): Declare.
36768 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
36770 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
36771 from TI's devices.csv file as of September 2016.
36772 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
36774 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
36776 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
36777 * doc/invoke.texi: Likewise.
36778 * doc/md.texi: Likewise.
36779 * doc/objc.texi: Likewise.
36781 2017-01-10 Joshua Conner <joshconner@google.com>
36783 * config/arm/fuchsia-elf.h: New file.
36784 * config/fuchsia.h: New file.
36785 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
36786 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
36788 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
36790 2016-01-10 Richard Biener <rguenther@suse.de>
36792 PR tree-optimization/79034
36793 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
36794 Propagate out degenerate PHIs in the joiner.
36796 2017-01-10 Martin Liska <mliska@suse.cz>
36798 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
36799 (sort_congruence_classes_by_decl_uid): Likewise.
36800 (sort_congruence_class_groups_by_decl_uid): Likewise.
36801 (sem_item_optimizer::merge_classes): Sort class, groups in these
36802 classes and members in the groups by DECL_UID of declarations.
36803 This would make merge operations stable.
36805 2017-01-10 Martin Liska <mliska@suse.cz>
36807 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
36808 usage of m_classes_vec.
36809 (sem_item_optimizer::~sem_item_optimizer): Likewise.
36810 (sem_item_optimizer::get_group_by_hash): Likewise.
36811 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
36812 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
36813 (sem_item_optimizer::verify_classes): Likewise.
36814 (sem_item_optimizer::process_cong_reduction): Likewise.
36815 (sem_item_optimizer::dump_cong_classes): Likewise.
36816 (sem_item_optimizer::merge_classes): Likewise.
36817 * ipa-icf.h (congruence_class_hash): Rename from
36818 congruence_class_group_hash. Remove declaration of m_classes_vec.
36820 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
36822 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
36823 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
36824 * config.gcc: Add avx512vpopcntdqintrin.h.
36825 * config/i386/avx512vpopcntdqintrin.h: New.
36826 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
36827 * config/i386/i386-builtin-types.def: Add new types.
36828 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
36829 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
36830 __builtin_ia32_vpopcountq_v8di_mask): New.
36831 * config/i386/i386-c.c (ix86_target_macros_internal): Define
36832 __AVX512VPOPCNTDQ__.
36833 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
36834 (PTA_AVX512VPOPCNTDQ): Define.
36835 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
36836 TARGET_AVX512VPOPCNTDQ_P): Define.
36837 * config/i386/i386.opt: Add mavx512vpopcntdq.
36838 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
36839 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
36841 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36843 PR middle-end/77484
36844 * predict.def (PRED_CALL): Set to 67.
36846 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
36848 * expr.c (store_field): In the bitfield case, if the value comes from
36849 a function call and is of an aggregate type returned in registers, do
36850 not modify the field mode; extract the value in all cases if the mode
36851 is BLKmode and the size is not larger than a word.
36853 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
36856 * config/i386/cpuid.h: Fix undefined behavior.
36858 2017-01-04 Jeff Law <law@redhat.com>
36860 PR tree-optimization/79007
36861 PR tree-optimization/67955
36862 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
36863 conservative for pt.null when flag_non_call_exceptions is on.
36865 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36867 PR translation/79019
36868 PR translation/79020
36869 * params.def (PARAM_INLINE_MIN_SPEEDUP,
36870 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
36871 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
36873 * config/avr/avr.opt (maccumulate-args): Likewise.
36874 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
36875 * common.opt (freport-bug): Likewise.
36876 * cif-code.def (CIF_FINAL_ERROR): Likewise.
36877 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
36878 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
36879 translatable string.
36880 * config/i386/i386.c (function_value_32): Likewise.
36881 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
36882 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
36884 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
36885 * common/config/msp430/msp430-common.c (msp430_handle_option):
36887 * symtab.c (symtab_node::verify_base): Likewise.
36888 * opts.c (set_debug_level): Likewise.
36889 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
36890 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
36891 missing whitespace to translatable strings.
36892 * config/avr/avr.md (bswapsi2): Fix typo in comment.
36893 * config/sh/superh.h: Likewise.
36894 * config/i386/xopintrin.h: Likewise.
36895 * config/i386/znver1.md: Likewise.
36896 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
36897 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
36898 * double-int.h (struct double_int): Likewise.
36899 * double-int.c (div_and_round_double): Likewise.
36900 * wide-int.cc: Likewise.
36901 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
36902 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
36903 * cfgcleanup.c (crossjumps_occured): Renamed to ...
36904 (crossjumps_occurred): ... this.
36905 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
36908 PR tree-optimization/78899
36909 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
36910 returning bool return struct loop *, NULL for failure and the new
36912 (versionable_outer_loop_p): Don't version outer loop if it has
36913 dont_vectorized bit set.
36914 (tree_if_conversion): When versioning outer loop, ensure
36915 tree_if_conversion is performed also on the inner loop of the
36916 non-vectorizable outer loop copy.
36917 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
36918 LOOP_VECTORIZED in inner loop of the scalar outer loop and
36919 prevent vectorization of it.
36920 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
36921 the outer loop vectorization of the non-scalar version is attempted
36922 before vectorization of the inner loop in scalar version. If
36923 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
36924 vectorization of its inner loop.
36925 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
36926 has 2 inner loops, rename also on edges from bb whose single pred
36927 is outer_loop->header. Fix typo in function comment.
36929 2017-01-09 Martin Sebor <msebor@redhat.com>
36932 * asan.c (asan_emit_stack_protection): Increase local buffer size
36933 to avoid snprintf truncation warning.
36935 2017-01-09 Andrew Pinski <apinski@cavium.com>
36937 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
36938 to reference thunderx2t99 for the tuning structure
36939 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
36941 (thunderx2t99_extra_costs): This.
36942 * config/aarch64/aarch64-tune.md: Regenerate.
36943 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
36944 (vulcan_addrcost_table): This.
36945 (vulcan_regmove_cost): Rename to ...
36946 (thunderx2t99_regmove_cost): This.
36947 (vulcan_vector_cost): Rename to ...
36948 (thunderx2t99_vector_cost): this.
36949 (vulcan_branch_cost): Rename to ...
36950 (thunderx2t99_branch_cost): This.
36951 (vulcan_tunings): Rename to ...
36952 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
36953 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
36955 2017-01-09 Martin Jambor <mjambor@suse.cz>
36959 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
36960 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
36961 (propagate_vr_accross_jump_function): Use the above function for all
36962 value range computations for pass-through jump functions and type
36963 converasion from explicit value range values.
36964 (ipcp_propagate_stage): Do not attempt to deduce types of formal
36965 parameters from TYPE_ARG_TYPES.
36966 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
36967 (ipa_write_node_info): Stream type of the actual argument.
36968 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
36970 2017-01-09 Martin Liska <mliska@suse.cz>
36973 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
36974 (lookup_compiler): Do not show error message with have_E.
36976 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36978 PR tree-optimization/78938
36979 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
36980 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
36981 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
36982 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
36985 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36987 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
36990 2017-01-09 Richard Biener <rguenther@suse.de>
36992 PR tree-optimization/78997
36993 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
36994 name condition properly.
36996 2017-01-09 Richard Biener <rguenther@suse.de>
36999 * dwarf2out.c (is_cxx): New overload with context.
37000 (is_naming_typedef_decl): Use it.
37002 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37004 * invoke.texi (Option Summary): Correct spacing in option lists
37005 and add line breaks to fix over-long lines.
37007 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37009 PR middle-end/17660
37011 * extend.texi (Common Variable Attributes): Add xref to GCC
37012 Internals manual to explain mode attribute keywords.
37014 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37017 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
37018 and Preprocessor Options.
37019 (Options for Linking): Document -pthread here....
37020 (RS/6000 and PowerPC Options): ...not here.
37021 (Solaris 2 Options): ...or here.
37022 * doc/cppopts.texi: Document -pthread.
37024 2017-01-08 Martin Sebor <msebor@redhat.com>
37026 PR middle-end/77708
37027 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
37028 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
37029 New member functions.
37030 (format_directive): Used them.
37032 (pass_sprintf_length::handle_gimple_call): Same.
37033 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
37034 to avoid truncation for any argument.
37035 (extract_affine_mul): Same.
37036 * tree.c (get_file_function_name): Same.
37038 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37040 PR middle-end/77484
37041 * predict.def (PRED_INDIR_CALL): Set to 86.
37043 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37045 PR preprocessor/54124
37046 * doc/cppopts.texi: Reformat -d subtable to list the full name
37047 of the options. Add cross-reference to the docs for the general
37048 compiler -d options.
37049 * doc/invoke.texi (Developer Options): Add cross-reference to the
37050 preprocessor-specific -d option documentation.
37052 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37054 PR preprocessor/13498
37055 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
37056 redudant material, and reflect new command-line options.
37057 (System Headers): Likewise.
37059 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37061 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
37062 -isystem, and -idirafter. Copy-edit.
37063 * doc/cppopts.texi: Copy-edit. Remove contradiction about
37064 default for -ftrack-macro-expansion. Delete obsolete and
37065 badly-formatted implementation details about -fdebug-cpp output.
37066 * doc/cppwarnopts.texi: Copy-edit.
37068 2017-01-07 David Malcolm <dmalcolm@redhat.com>
37071 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37072 that the transition from a max line width >= 1<<10 to narrower
37073 lines works correctly.
37075 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
37077 * doc/options.texi (PerFunction): New.
37078 * opt-functions.awk (switch_flags): Map both Optimization and
37079 PerFunction to CL_OPTIMIZATION.
37080 * opth-gen.awk: Test for PerFunction flag along with
37082 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
37083 it only when the latter is present. Skip those that don't in
37084 the hash function generator.
37085 * common.opt (fvar-tracking): Mark as PerFunction instead of
37087 (fvar-tracking-assignments): Likewise.
37088 (fvar-tracking-assignments-toggle): Likewise.
37089 (fvar-tracking-uninit): Likewise.
37091 2017-01-07 Jakub Jelinek <jakub@redhat.com>
37093 PR translation/79018
37094 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
37097 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
37100 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
37101 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
37102 TARGET_LONG_JUMP_TABLE_OFFSETS.
37103 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
37104 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
37105 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
37106 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
37107 * config/m68k/m68k.md (tablejump expander): Likewise.
37108 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
37109 TARGET_LONG_JUMP_TABLE_OFFSETS.
37110 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
37111 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
37113 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
37114 David Holsgrove <david.holsgrove@xilinx.com>
37116 * common/config/microblaze/microblaze-common.c
37117 (TARGET_EXCEPT_UNWIND_INFO): Remove.
37118 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
37120 * config/microblaze/microblaze.c (microblaze_must_save_register)
37121 (microblaze_expand_epilogue, microblaze_return_addr): Handle
37123 (microblaze_eh_return): New function.
37124 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
37125 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
37126 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
37127 * config/microblaze/microblaze.md (eh_return): New pattern.
37129 2017-01-06 Jakub Jelinek <jakub@redhat.com>
37131 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
37132 GCC_DIAGNOSTIC_STRINGIFY): Define.
37134 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
37136 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37138 * config/arm/arm.md (<mcrr>): New.
37140 * config/arm/arm.c (arm_arch5te): New.
37141 (arm_option_override): Set arm_arch5te.
37142 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
37144 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
37145 (arm_mcrr_qualifiers): ... this. New.
37146 (MRRC_QUALIFIERS): Define to...
37147 (arm_mrrc_qualifiers): ... this. New.
37148 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
37150 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
37151 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
37152 (MRRCI, mrrc, MRRC): New.
37153 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
37154 VUNSPEC_MRRC2): New.
37156 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37158 * config/arm/arm.md (<mcr>): New.
37160 * config/arm/arm.c (arm_coproc_builtin_available): Add
37161 support for mcr, mrc, mcr2 and mrc2.
37162 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
37163 (arm_mcr_qualifiers): ... this. New.
37164 (MRC_QUALIFIERS): Define to ...
37165 (arm_mrc_qualifiers): ... this. New.
37166 (MCR_QUALIFIERS): Define to ...
37167 (arm_mcr_qualifiers): ... this. New.
37168 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
37170 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
37171 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
37172 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
37173 VUNSPEC_MRC2): New.
37175 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37177 * config/arm/arm.md (*ldc): New.
37181 * config/arm/arm.c (arm_coproc_builtin_available): Add
37182 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
37183 (arm_coproc_ldc_stc_legitimate_address): New.
37184 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
37185 'qualifier_const_pointer'.
37186 (LDC_QUALIFIERS): Define to...
37187 (arm_ldc_qualifiers): ... this. New.
37188 (STC_QUALIFIERS): Define to...
37189 (arm_stc_qualifiers): ... this. New.
37190 * config/arm/arm-protos.h
37191 (arm_coproc_ldc_stc_legitimate_address): New.
37192 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
37193 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
37194 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
37195 stc2, stcl, stc2l): New.
37196 * config/arm/constraints.md (Uz): New.
37197 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
37198 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
37199 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
37200 VUNSPEC_STC2L): New.
37202 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37204 * config/arm/arm.md (<cdp>): New.
37205 * config/arm/arm.c (neon_const_bounds): Rename this ...
37206 (arm_const_bounds): ... this.
37207 (arm_coproc_builtin_available): New.
37208 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
37209 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
37210 (CDP_QUALIFIERS): Define to...
37211 (arm_cdp_qualifiers): ... this. New.
37213 (arm_expand_builtin_args): Add case for 6 arguments.
37214 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
37215 (arm_const_bounds): ... this.
37216 (arm_coproc_builtin_available): New.
37217 * config/arm/arm_acle.h (__arm_cdp): New.
37219 * config/arm/arm_acle_builtins.def (cdp): New.
37221 * config/arm/iterators.md (CDPI,CDP,cdp): New.
37222 * config/arm/neon.md: Rename all 'neon_const_bounds' to
37223 'arm_const_bounds'.
37224 * config/arm/types.md (coproc): New.
37225 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
37226 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
37227 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
37228 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
37230 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37232 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
37233 (UBINOP_QUALIFIERS): New.
37235 (acle_builtin_data): New. Change comment.
37236 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
37237 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
37238 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
37239 arm_acle_builtins.def.
37240 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
37241 (arm_init_acle_builtins): New.
37242 (CRC32_BUILTIN): Remove.
37243 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
37244 crc32cb, crc32ch and crc32cw.
37245 (arm_init_crc32_builtins): Remove.
37246 (arm_init_builtins): Use arm_init_acle_builtins rather
37247 than arm_init_crc32_builtins.
37248 (arm_expand_acle_builtin): New.
37249 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
37250 * config/arm/arm_acle_builtins.def: New.
37252 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37254 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
37255 (arm_builtin_datum): ... this.
37256 (arm_init_neon_builtin): Rename to ...
37257 (arm_init_builtin): ... this. Add a new parameters PREFIX
37258 and USE_SIG_IN_NAME.
37259 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
37260 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
37261 'arm_builtin_datum'.
37262 (arm_init_vfp_builtins): Likewise.
37263 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
37264 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
37265 (arm_expand_neon_args): Rename to ...
37266 (arm_expand_builtin_args): ... this. Rename builtin_arg
37267 enum values and differentiate between ARG_BUILTIN_MEMORY
37268 and ARG_BUILTIN_NEON_MEMORY.
37269 (arm_expand_neon_builtin_1): Rename to ...
37270 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
37271 values, arm_expand_builtin_args and add bool parameter NEON.
37272 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
37273 (arm_expand_vfp_builtin): Likewise.
37274 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
37276 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37278 PR middle-end/77484
37279 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
37280 * predict.c (tree_estimate_probability_bb): Reverse direction of
37281 polymorphic call predictor.
37283 2017-01-06 David Malcolm <dmalcolm@redhat.com>
37285 * passes.c (execute_one_pass): Split out pass-skipping logic into...
37286 (determine_pass_name_match): ...this new function and...
37287 (should_skip_pass_p): ...this new function.
37289 2017-01-06 Nathan Sidwell <nathan@acm.org>
37291 * ipa-visibility.c (function_and_variable_visibility): Reformat
37292 comments and long lines. Remove extrneous if.
37293 * symtab.c (symtab_node::make_decl_local): Fix code format.
37294 (symtab_node::set_section_for_node): Fix comment typo.
37296 2017-01-06 Martin Liska <mliska@suse.cz>
37299 * lra-constraints.c: Rename invariant to lra_invariant.
37300 * predict.c (set_even_probabilities): Initialize e to NULL.
37302 2017-01-05 Martin Sebor <msebor@redhat.com>
37304 PR tree-optimization/78910
37305 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
37306 (format_integer): Correct off-by-one error in the handling
37307 of precision with negative numbers in signed conversions..
37309 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
37311 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
37313 2017-01-05 Jakub Jelinek <jakub@redhat.com>
37315 PR tree-optimization/71016
37316 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
37317 factor_out_conditional_conversion. Formatting fix.
37318 (factor_out_conditional_conversion): Add cond_stmt argument.
37319 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
37320 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
37323 2017-01-05 David Malcolm <dmalcolm@redhat.com>
37325 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
37326 read-rtl-function.o, and selftest-rtl.o.
37327 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
37328 (selftest::aarch64_test_loading_full_dump): New function.
37329 (selftest::aarch64_run_selftests): New function.
37330 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
37331 selftest::aarch64_run_selftests.
37332 * config/i386/i386.c
37333 (selftest::ix86_test_loading_dump_fragment_1): New function.
37334 (selftest::ix86_test_loading_call_insn): New function.
37335 (selftest::ix86_test_loading_full_dump): New function.
37336 (selftest::ix86_test_loading_unspec): New function.
37337 (selftest::ix86_run_selftests): Call the new functions.
37338 * emit-rtl.c (maybe_set_max_label_num): New function.
37339 * emit-rtl.h (maybe_set_max_label_num): New decl.
37340 * function.c (instantiate_decls): Guard call to
37341 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
37342 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
37344 * gensupport.c (gen_reader::gen_reader): Pass "false"
37345 for new "compact" param of rtx_reader.
37346 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
37347 rather than an empty string for NULL strings.
37348 * read-md.c: Potentially include config.h rather than bconfig.h.
37349 Wrap include of errors.h with #ifdef GENERATOR_FILE.
37350 (have_error): New global, copied from errors.c.
37351 (md_reader::read_name): Rename to...
37352 (md_reader::read_name_1): ...this, adding "out_loc" param,
37353 and converting "missing name or number" to returning false, rather
37355 (md_reader::read_name): Reimplement in terms of read_name_1.
37356 (md_reader::read_name_or_nil): New function.
37357 (md_reader::read_string): Handle "(nil)" by returning NULL.
37358 (md_reader::md_reader): Add new param "compact".
37359 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
37360 (md_reader::read_file): New method.
37361 * read-md.h (md_reader::md_reader): Add new param "compact".
37362 (md_reader::read_file): New method.
37363 (md_reader::is_compact): New accessor.
37364 (md_reader::read_name): Convert return type from void to file_location.
37365 (md_reader::read_name_or_nil): New decl.
37366 (md_reader::read_name_1): New decl.
37367 (md_reader::m_compact): New field.
37368 (noop_reader::noop_reader): Pass "false" for new "compact" param
37370 (rtx_reader::rtx_reader): Add new "compact" param.
37371 (rtx_reader::read_rtx_operand): Make virtual and convert return
37372 type from void to rtx.
37373 (rtx_reader::read_until): New decl.
37374 (rtx_reader::handle_any_trailing_information): New virtual function.
37375 (rtx_reader::postprocess): New virtual function.
37376 (rtx_reader::finalize_string): New virtual function.
37377 (rtx_reader::m_in_call_function_usage): New field.
37378 (rtx_reader::m_reuse_rtx_by_id): New field.
37379 * read-rtl-function.c: New file.
37380 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
37381 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
37382 (selftest::verify_three_block_rtl_cfg): New decl.
37383 * read-rtl-function.h: New file.
37384 * read-rtl.c: Potentially include config.h rather than bconfig.h.
37385 For host, include function.h, memmodel.h, and emit-rtl.h.
37386 (one_time_initialization): New function.
37387 (struct compact_insn_name): New struct.
37388 (compact_insn_names): New array.
37389 (find_code): Handle insn codes in compact dumps.
37390 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
37391 (bind_subst_iter_and_attr): Likewise.
37392 (add_condition_to_string): Likewise.
37393 (add_condition_to_rtx): Likewise.
37394 (apply_attribute_uses): Likewise.
37395 (add_current_iterators): Likewise.
37396 (apply_iterators): Likewise.
37397 (initialize_iterators): Guard usage of apply_subst_iterator with
37398 #ifdef GENERATOR_FILE.
37399 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
37400 (md_reader::read_mapping): Likewise.
37401 (add_define_attr_for_define_subst): Likewise.
37402 (add_define_subst_attr): Likewise.
37403 (read_subst_mapping): Likewise.
37404 (check_code_iterator): Likewise.
37405 (rtx_reader::read_rtx): Likewise. Move one-time initialization
37407 (one_time_initialization): New function.
37408 (rtx_reader::read_until): New method.
37409 (read_flags): New function.
37410 (parse_reg_note_name): New function.
37411 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
37412 Handle reuse_rtx ids.
37413 Wrap iterator lookup within #ifdef GENERATOR_FILE.
37414 Add parsing support for RTL dumps, mirroring the special-cases in
37415 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
37416 values, and calling handle_any_trailing_information.
37417 (rtx_reader::read_rtx_operand): Convert return type from void
37418 to rtx, returning return_rtx. Handle case 'e'. Call
37419 finalize_string on XSTR and XTMPL fields.
37420 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
37421 "(nil)" values were omitted. Call the postprocess vfunc on the
37423 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
37424 class ctor. Initialize m_in_call_function_usage. Call
37425 one_time_initialization.
37426 * rtl-tests.c (selftest::test_uncond_jump): Call
37427 set_new_first_and_last_insn.
37428 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
37429 * selftest-rtl.c: New file.
37430 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
37431 (selftest::get_insn_by_uid): New decl.
37432 * selftest-run-tests.c (selftest::run_tests): Call
37433 read_rtl_function_c_tests.
37434 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
37435 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
37438 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
37440 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
37441 operands in a special way. Assert that pos+len <= mode precision.
37443 2017-01-05 Jakub Jelinek <jakub@redhat.com>
37445 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
37446 3 argument Alias with unlimited for the negative form.
37447 (fno-vect-cost-model): Removed.
37449 2017-01-05 Martin Liska <mliska@suse.cz>
37451 * hsa-gen.c (gen_hsa_divmod): New function.
37452 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
37454 2017-01-05 Martin Liska <mliska@suse.cz>
37457 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
37460 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37462 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
37463 small constant length operands.
37465 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37467 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
37468 between loop iterations.
37470 2017-01-05 Martin Liska <mliska@suse.cz>
37473 * gimplify.c (gimplify_decl_expr): Compare to
37474 asan_poisoned_variables instread of checking flags.
37475 (gimplify_target_expr): Likewise.
37476 (gimplify_expr): Likewise.
37477 (gimplify_function_tree): Conditionally initialize
37478 asan_poisoned_variables.
37480 2017-01-04 Jeff Law <law@redhat.com>
37482 PR tree-optimizatin/78812
37483 * rtl.h (contains_mem_rtx_p): Prototype.
37484 * ifcvt.c (containts_mem_rtx_p): Move from here to...
37485 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
37486 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
37487 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
37488 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
37490 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37492 * input.c (assert_char_at_range): Default-initialize actual_range.
37494 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37496 * df-scan.c (df_ref_create_structure): Make regno unsigned,
37497 to match the caller.
37499 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37501 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
37502 insns after final jump in test to emit dummy move.
37504 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37506 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
37507 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
37509 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37511 * multiple_target.c (create_dispatcher_calls): Init e_next.
37512 * tree-ssa-loop-split.c (split_loop): Init border.
37513 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
37516 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
37521 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
37522 (altivec_register_operand): Do not return true if the operand
37523 contains a SUBREG mixing SImode and SFmode.
37524 (vsx_register_operand): Likewise.
37525 (vsx_reg_sfsubreg_ok): New predicate.
37526 (vfloat_operand): Do not return true if the operand contains a
37527 SUBREG mixing SImode and SFmode.
37528 (vint_operand): Likewise.
37529 (vlogical_operand): Likewise.
37530 (gpc_reg_operand): Likewise.
37531 (int_reg_operand): Likewise.
37532 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
37533 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
37534 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
37536 (rs6000_emit_move_si_sf_subreg): New helper function.
37537 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
37538 fixup SUBREGs involving SImode and SFmode.
37539 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
37540 numbers for the new peephole2 optimization.
37541 (peephole2 for SFmode unions): New peephole2 to optimize cases in
37542 the GLIBC math library that do AND/IOR/XOR operations on single
37543 precision floating point.
37544 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
37545 target macros to say whether we need to avoid SUBREGs mixing
37547 (TARGET_ALLOW_SF_SUBREG): Likewise.
37548 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
37549 (UNSPEC_SI_FROM_SF): Likewise.
37550 (iorxor): Change spacing.
37551 (and_ior_xor): New iterator for AND, IOR, and XOR.
37552 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
37553 (movdi_from_sf_zero_ext): Likewise.
37554 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
37555 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
37556 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
37557 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
37558 (fms<mode>4): Likewise.
37559 (fnma<mode>4): Likewise.
37560 (fnms<mode>4): Likewise.
37561 (nfma<mode>4): Likewise.
37562 (nfms<mode>4): Likewise.
37564 2017-01-04 Marek Polacek <polacek@redhat.com>
37567 * doc/invoke.texi: Document -Wpointer-compare.
37569 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37571 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
37574 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
37575 descriptions for -gdwarf-5 and emit them as uleb128 instead of
37578 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
37581 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
37582 documentation of the powerpc_popcntb_ok attribute.
37583 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
37584 code to issue warning messages if a requested CPU configuration is
37585 not supported by the binary (assembler and loader) toolchain.
37586 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
37587 made to define a built-in function that has been disabled.
37588 (paired_init_builtins): Add assertion to prevent ICE if attempt is
37589 made to define a built-in function that has been disabled.
37590 (altivec_init_builtins): Add comment explaining why definition
37591 of the DST built-in functions is not preceded by an assertion
37592 check. Add assertions to prevent ICE if attempts are made to
37593 define an altivec predicate or an abs* built-in function that has
37595 (htm_init_builtins): Add comment explaining why definition of the
37596 htm built-in functions is not preceded by an assertion check.
37598 2017-01-04 Jeff Law <law@redhat.com>
37600 PR tree-optimizatin/67955
37601 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
37602 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
37603 the points-to solution does not include pt_null. Use DECL_PT_UID
37606 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
37608 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
37609 Use gen_int_mode instead of gen_lopwart for const_int operands.
37611 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37613 PR tree-optimization/71563
37614 * match.pd: Simplify X << Y into X if Y is known to be 0 or
37615 out of range value - has low bits known to be zero.
37617 2017-01-04 Alan Modra <amodra@gmail.com>
37619 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
37620 * configure: Regenerate.
37621 * config.in: Regenerate.
37623 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37626 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
37627 a substring of the message, but strcmp with the whole message. Ifdef
37628 ENABLE_NLS, translate the message first using dgettext.
37630 2017-01-03 Jeff Law <law@redhat.com>
37632 PR tree-optimizatin/78856
37633 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
37634 (mark_threaded_blocks): Remove code to truncate thread paths that
37635 cross multiple loop headers. Instead invalidate the cached loop
37636 iteration information and handle case of a thread path walking
37637 into an irreducible region.
37639 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
37642 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
37643 assertions. Add support for doing the signbit if the IEEE 128-bit
37644 floating point value is in a GPR.
37645 * config/rs6000/rs6000.md (Fsignbit): Delete.
37646 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
37647 Update the length attribute if the value is in a GPR.
37648 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
37649 the sign or zero extension instruction, since the value is always 0/1.
37650 (signbit<mode>2_dm2): Delete using <Fsignbit>.
37653 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
37654 extracting SImode to a GPR register so that we can generate a
37655 store, limit the vector to be in a traditional Altivec register
37656 for the vextuwrx instruction.
37658 2017-01-03 Ian Lance Taylor <iant@google.com>
37660 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
37662 2017-01-03 Martin Sebor <msebor@redhat.com>
37664 PR tree-optimization/78696
37665 * gimple-ssa-sprintf.c (format_floating): Correct handling of
37666 precision. Use MPFR for %f for greater fidelity. Correct handling
37668 (pass_sprintf_length::compute_format_length): Set width and precision
37669 specified by asrerisk to void_node for vararg functions.
37670 (try_substitute_return_value): Adjust dump output.
37672 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
37674 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
37676 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
37678 * doc/invoke.texi (SPARC options): Document -mlra as the default.
37679 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
37680 -mlra/-mno-lra was passed to the compiler.
37682 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
37684 PR rtl-optimization/65618
37685 * emit-rtl.c (try_split): Move initialization of "before" and
37686 "after" to just before the call to emit_insn_after_setloc.
37688 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
37690 * doc/md.texi (Standard Names): Remove reference to Java frontend.
37692 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
37694 * dwarf2out.c (gen_enumeration_type_die): When
37695 -gno-strict-dwarf, add a DW_AT_encoding attribute.
37697 2017-01-03 Jakub Jelinek <jakub@redhat.com>
37699 PR tree-optimization/78965
37700 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
37701 Change first argument from const call_info & to call_info &. For %n
37702 set info.nowrite to false.
37704 PR middle-end/78901
37705 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
37706 possibly throwing calls.
37708 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
37709 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
37710 and fns handling, rather than in a separate case SSA_NAME.
37712 2017-01-02 Jeff Law <law@redhat.com>
37714 * config/darwin-driver.c (darwin_driver_init): Const-correctness
37715 fixes for first_period and second_period variables.
37717 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
37720 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
37721 (*insvqi_1): New insn pattern.
37722 (*insvqi_1_mem_rex64): Ditto.
37723 (*insvqi_2): Ditto.
37724 (*insvqi_3): Rename from *insvqi.
37726 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
37728 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
37730 * doc/cfg.texi (Edges): Remove reference to Java.
37731 (Maintaining the CFG): Ditto.
37733 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37735 PR middle-end/77674
37736 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
37737 transparent aliases.
37739 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37741 PR middle-end/77484
37742 * predict.def (PRED_CALL): Update hitrate.
37743 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
37744 * predict.c (tree_estimate_probability_bb): Split CALL predictor
37745 into direct/indirect/polymorphic variants.
37747 2017-01-01 Jakub Jelinek <jakub@redhat.com>
37749 Update copyright years.
37751 * gcc.c (process_command): Update copyright notice dates.
37752 * gcov-dump.c (print_version): Ditto.
37753 * gcov.c (print_version): Ditto.
37754 * gcov-tool.c (print_version): Ditto.
37755 * gengtype.c (create_file): Ditto.
37756 * doc/cpp.texi: Bump @copying's copyright year.
37757 * doc/cppinternals.texi: Ditto.
37758 * doc/gcc.texi: Ditto.
37759 * doc/gccint.texi: Ditto.
37760 * doc/gcov.texi: Ditto.
37761 * doc/install.texi: Ditto.
37762 * doc/invoke.texi: Ditto.
37764 Copyright (C) 2017 Free Software Foundation, Inc.
37766 Copying and distribution of this file, with or without modification,
37767 are permitted in any medium without royalty provided the copyright
37768 notice and this notice are preserved.