1 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4 2017-11-29 Martin Aberg <maberg@gaisler.com>
6 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
7 to prevent b2bst errata sequence.
8 (sqrtdf2_fix): Likewise.
10 2017-12-06 Jakub Jelinek <jakub@redhat.com>
12 PR tree-optimization/81945
13 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
14 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
15 to dest_cfun has orig_loop_num set, either remap it to the new
16 loop number if the loop got moved too, or clear it.
18 2017-12-05 Steve Ellcey <sellcey@cavium.com>
20 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
22 (thunderx2t99_nothing): New insn reservation.
23 (thunderx2t99_mrs): New insn reservation.
24 (thunderx2t99_multiple): New insn reservation.
25 (thunderx2t99_alu_basi): Add bfx to reservation.
26 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
28 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
31 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
33 2017-12-05 Bin Cheng <bin.cheng@arm.com>
35 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
36 tree-ssa-pre.c::remove_dead_inserted_code.
37 * tree-ssa-dce.h: New file.
38 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
39 (remove_dead_inserted_code): Move and rename to function
40 tree-ssa-dce.c::simple_dce_from_worklist.
41 (pass_pre::execute): Update use.
43 2017-12-05 Richard Biener <rguenther@suse.de>
45 PR tree-optimization/83277
46 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
47 to code-gen liveout vars.
49 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
51 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
52 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
53 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
56 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
58 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
60 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
61 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
64 2017-12-05 Richard Biener <rguenther@suse.de>
66 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL):
68 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
69 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
70 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
72 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
74 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
76 2017-12-05 Martin Liska <mliska@suse.cz>
77 Jakub Jelinek <jakub@redhat.com>
79 * doc/invoke.texi: Document the options.
80 * flag-types.h (enum sanitize_code): Add
81 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
82 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
83 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
84 * opts.c: Define new sanitizer options.
85 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
86 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
88 2017-12-05 Julia Koval <julia.koval@intel.com>
90 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
91 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
92 (ix86_handle_option): Handle -mavx512vnni.
93 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
94 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
95 * config/i386/i386-c (__AVX512VNNI__): New.
96 * config/i386/i386.c (ix86_target_string): Handle new option.
97 (ix86_valid_target_attribute_inner_p): Handle new option.
98 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
99 * config/i386/i386.opt (mavx512vnni): New option.
101 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
104 * athlon.md: Disable for generic.
105 * haswell.md: Enable for generic.
106 * i386.c (ix86_sched_init_global): Add core hooks for generic.
107 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
109 (ix86_adjust_cost): Move generic to haswell path.
111 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
113 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
114 instead of MEM_P in a couple more places. Fix formatting issues.
116 2017-12-04 Jim Wilson <jimw@sifive.com>
118 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
119 instead of GP_REG_LAST-1.
120 (riscv_adjust_libcall_cfi_prologue): Likewise.
121 (riscv_adjust_libcall_cri_epilogue): Likewise.
122 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
125 2017-12-04 Luis Machado <luis.machado@linaro.org>
127 * ipa-pure-const.c (check_decl): Add missing newline.
128 (state_from_flags): Likewise.
130 2017-12-04 Jeff Law <law@redhat.com>
132 PR tree-optimizatin/78496
133 * gimple-ssa-evrp-analyze.h
134 (evrp_range_analyzer::get_vr_values): Simplify.
135 * gimple-ssa-evrp-analyze.c: Corresponding changes.
136 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
137 and gimple-ssa-evrp-analyze.h.
138 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
139 (simplify_stmt_for_jump_threading): Copy a blob of code from
140 tree-vrp.c to use ranges to simplify statements.
141 (dom_opt_dom_walker::before_dom_children): Call
142 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
143 (dom_opt_dom_walker::after_dom_children): Similarly for
144 evrp_range_analyzer::leave.
145 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
148 * gimple-ssa-evrp-analyze.c
149 (evrp_range_analyzer::extract_range_from_stmt): Always use
150 vr_values::update_value_range so preexisting range info is
151 medged with new range info, even if the new range is VR_VARYING.
153 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
155 * combine.c: Adjust comment.
156 (use_crosses_set_p): Delete.
157 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
158 (try_combine): Ditto.
160 2017-12-04 Richard Biener <rguenther@suse.de>
162 PR tree-optimization/83255
163 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
164 Re-add zero-iteration check.
166 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
168 PR rtl-optimization/83245
169 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
170 hard registers as earlyclobber, also if not in an asm.
172 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
176 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
179 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
180 rs6000_cpu based on cpu_index, not tune_index.
182 2017-12-04 Richard Biener <rguenther@suse.de>
184 PR tree-optimization/83238
185 * graphite-scop-detection.c (scop_detection::merge_sese): Make
186 code match comment, rejecting invalid SESE regions.
188 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
190 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
191 require base operand is a REG_POINTER prior to reload on targets
192 with non-equivalent space registers.
194 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
196 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
197 (update_specialized_profile): Fix updating of counts.
198 (perhaps_add_new_callers): Likewise.
200 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
203 * x86-tune.def: Remove obsolette FIXMEs.
204 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
205 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
206 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
208 (X86_TUNE_PAD_RETURNS): Disable for generic.
210 2017-12-02 Jakub Jelinek <jakub@redhat.com>
212 PR tree-optimization/83170
213 PR tree-optimization/83241
214 * gimple-ssa-store-merging.c
215 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
216 gimple_vuse (ins_stmt) in case it has changed.
217 (imm_store_chain_info::output_merged_store): Likewise.
219 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
223 * tree-cfg.c (pass_warn_function_return::execute): Handle
224 __builtin_ubsan_handle_missing_return like __builtin_unreachable
225 with BUILTINS_LOCATION.
229 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
230 is BLKmode for vector field with vector raw mode, use TYPE_MODE
231 instead of DECL_MODE.
233 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
234 last argument to rtx pointer.
235 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
236 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
237 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
238 depending on the chosen ISAs.
239 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
240 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
241 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
243 * config/i386/sse.md (mov<mode>_internal): Likewise.
244 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
246 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
248 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
250 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
252 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
253 parameters from prototype.
254 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
255 parameters. Don't print a comment.
256 (emit_fusion_gpr_load): Adjust.
257 (emit_fusion_load_store): Adjust.
258 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
259 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
260 comment on the second line.
262 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
265 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
266 rs6000_cpu based on cpu_index, not tune_index.
268 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
270 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
271 which_alternative instead of which_alternative + 1.
272 (output_asm_insn): Print an extra tab if the template is short.
274 2017-12-01 Jim Wilson <jimw@sifive.com>
276 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
278 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
279 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
281 * doc/tm.texi: Regenerate.
283 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
286 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
287 whether we can allocate pseudos before trying to fix an address.
288 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
289 memory address is indexed or indirect.
290 (floatuns_<mode>si2_hw2): Likewise.
292 2017-12-01 Jason Merrill <jason@redhat.com>
294 * Makefile.in (TAGS): Add c-family/*.cc.
296 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
298 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
299 (call_value_insn): Likewise.
300 (sibcall_insn): Likewise.
301 (sibcall_value_insn): Likewise.
302 (movsi_aarch64): Likewise.
303 (movdi_aarch64): Likewise.
304 (add_losym_): Likewise.
305 (ldr_got_small_): Likewise.
306 (ldr_got_small_sidi): Likewise.
307 (ldr_got_small_28k_): Likewise.
308 (ldr_got_small_28k_sidi): Likewise.
309 * config/aarch64/aarch64.c (aarch64_print_address_internal):
310 Move output_addr_const to symbolic case. Add error check.
312 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
314 * config/s390/predicates.md (plus16_Q_operand): New predicate.
315 * config/s390/s390.md: Disable MVC merging peephole if it would
316 disable operand forwarding.
317 (new peephole2): Split MVCs if it would turn them into up to 2
320 2017-12-01 Richard Biener <rguenther@suse.de>
322 PR tree-optimization/83232
323 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
324 detection of same access. Instead of breaking the group here
325 do not consider the duplicate. Add comment explaining real fix.
327 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
329 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
331 2017-12-01 Sudakshina Das <sudi.das@arm.com>
333 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
335 2017-12-01 Jakub Jelinek <jakub@redhat.com>
337 * function.h (struct function): Remove cilk_frame_decl,
338 is_cilk_function and calls_cilk_spawn fields.
339 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
341 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
342 cilk_elemental field.
343 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
344 * target.def: Adjust comment.
345 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
346 Don't test cilk_elemental.
348 PR tree-optimization/83233
349 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
350 bswap_stat name for the struct.
353 * tree.h (SWITCH_BREAK_LABEL_P): Define.
354 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
355 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
356 SWITCH_BREAK_LABEL_P set on the label.
357 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
358 added for default case if it was missing and not all cases covered.
359 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
360 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
362 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
363 to avoid -Wimplicit-fallthrough warning.
364 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
365 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
367 PR tree-optimization/83221
368 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
370 (init_reassoc): Formatting fix.
373 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
374 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
378 * tree-cfg.c: Include asan.h.
379 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
381 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
383 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
385 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
387 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
388 vpcmpeqd instruction.
390 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
392 * config/i386/i386.c (standard_sse_constant_opcode): Fix
393 registers type for 128bit mode.
395 2017-11-30 Jakub Jelinek <jakub@redhat.com>
397 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
399 * gengtype-state.c (read_state_pair): Likewise.
400 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
401 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
402 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
403 * ifcvt.c (noce_try_store_flag_constants): Likewise.
404 * tree-ssa-ccp.c (ccp_finalize): Likewise.
405 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
406 * builtins.c (fold_builtin_3): Likewise.
407 * graphite-scop-detection.c
408 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
409 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
412 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
415 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
416 parameter. Return true if ignore_decl_rtl_set_p is true and other
417 conditions are satisfied.
418 * asan.h (asan_protect_global): Add new parameter.
419 * varasm.c (categorize_decl_for_section): Pass true as second parameter
420 to asan_protect_global calls.
422 2017-11-30 Jim Wilson <jimw@sifive.com>
424 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
425 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
426 -mstrict-align, add info on default value. Delete redundant lines for
427 -mabi. Add missing -mexplicit-relocs docs.
429 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
431 * config/arc/arc.md (trap): New pattern.
433 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
435 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
436 instruction to end into a delay slot.
437 * config/arc/arc.md (cond_delay_insn): Check if the instruction
438 can be placed into a delay slot against reg_note.
440 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
442 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
443 labels number of usages.
445 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
447 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
449 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
451 2017-11-30 Jakub Jelinek <jakub@redhat.com>
454 * internal-fn.c (expand_mul_overflow): Optimize unsigned
455 multiplication by power of 2 constant into two shifts + comparison.
457 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
460 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
462 2017-11-30 Richard Biener <rguenther@suse.de>
464 PR tree-optimization/83202
465 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
466 allow_peel argument and guard peeling.
467 (canonicalize_loop_induction_variables): Likewise.
468 (canonicalize_induction_variables): Pass false.
469 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
470 peeling from cunrolli.
472 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
474 * combine.c (try_combine): Print a message to dump file whenever
475 I0, I1, or I2 cannot be combined into I3.
477 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
479 PR rtl-optimization/83156
480 PR rtl-optimization/82621
481 * combine.c (try_combine): Don't split an I2 if one of the dests is
482 set again before I3. Allow unused dests.
484 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
486 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
488 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
490 PR rtl-optimization/80818
491 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
492 recursively. Use insn code for clobber.
493 (lra_set_insn_recog_data): Pass the new arg to
494 collect_non_operand_hard_regs.
495 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
497 (lra_update_insn_regno_info): Pass insn to
498 add_regs_to_insn_regno_info.
500 2017-11-29 Jim Wilson <jimw@sifive.com>
501 Andrew Waterman <andrew@sifive.com>
503 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
504 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
505 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
506 use SINGLE_SHIFT_COST.
507 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
508 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
510 2017-11-29 Julia Koval <julia.koval@intel.com>
512 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
513 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
514 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
515 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
516 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
517 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
518 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
519 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
520 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
521 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
522 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
523 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
524 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
525 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
526 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
527 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
528 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
529 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
530 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
531 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
532 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
533 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
534 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
535 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
536 __builtin_ia32_vpshldv_v2di_mask,
537 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
538 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
539 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
541 2017-11-29 Julia Koval <julia.koval@intel.com>
543 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
544 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
545 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
546 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
547 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
548 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
549 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
550 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
551 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
552 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
553 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
554 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
555 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
556 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
557 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
558 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
559 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
560 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
561 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
562 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
563 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
565 2017-11-29 Daniel Cederman <cederman@gaisler.com>
567 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
568 movsi_pic_gotdata_op instruction as a load for the UT699 errata
571 2017-11-29 Martin Aberg <maberg@gaisler.com>
573 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
574 to prevent b2bst errata sequence.
575 (sqrtdf2_fix): Likewise.
577 2017-11-29 Daniel Cederman <cederman@gaisler.com>
579 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
580 (div_sqrt_insn_p): New function.
581 (sparc_do_work_around_errata): Insert NOP instructions to
582 prevent sequences that could trigger the TN-0013 errata for
583 certain LEON3 processors.
584 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
585 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
586 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
587 (in_branch_delay): Prevent div and sqrt in delay slot if
589 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
591 2017-11-29 Daniel Cederman <cederman@gaisler.com>
593 * config/sparc/sparc.c (atomic_insn_p): New function.
594 (sparc_do_work_around_errata): Insert NOP instructions to
595 prevent sequences that could trigger the TN-0010 errata for
597 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
598 instruction referable in atomic_insns_p.
600 2017-11-29 Daniel Cederman <cederman@gaisler.com>
602 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
603 (atomic_compare_and_swap_leon3_1): Likewise.
606 2017-11-29 Daniel Cederman <cederman@gaisler.com>
608 * config/sparc/sparc.c (fpop_insn_p): New function.
609 (sparc_do_work_around_errata): Insert NOP instructions to
610 prevent sequences that could trigger the TN-0012 errata for
612 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
613 * config/sparc/sparc.md (fix_gr712rc): New attribute.
614 (in_branch_annul_delay): Prevent floating-point instructions
615 in delay slot of annulled integer branch.
617 2017-11-29 Richard Biener <rguenther@suse.de>
619 PR tree-optimization/83202
620 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
622 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
623 nodes vectorized to the same stmts multiple times.
624 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
625 (vect_analyze_slp_instance): Adjust.
626 (scalar_stmts_to_slp_tree_map_t): New typedef.
627 (vect_schedule_slp_instance): Add a map recording the SLP node
628 representing the vectorized stmts for a set of scalar stmts.
629 Avoid code-generating redundancies.
630 (vect_schedule_slp): Allocate map and pass it down.
632 2017-11-29 Nathan Sidwell <nathan@acm.org>
635 * tree.c (build_complex_type): Fix canonicalization. Only fill in
638 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
640 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
641 cases for printing LDP/STP memory addresses.
642 (aarch64_print_address_internal): Renamed from
643 aarch64_print_operand_address, added parameter, add Pmode check.
644 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
645 (aarch64_print_operand_address): Indirect to
646 aarch64_print_address_internal.
647 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
648 'y' operand output specifier.
650 2017-11-29 Jakub Jelinek <jakub@redhat.com>
653 * tree.c (build_simple_mem_ref_loc): Handle
654 get_addr_base_and_unit_offset returning a MEM_REF.
657 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
660 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
663 2017-11-28 David Malcolm <dmalcolm@redhat.com>
665 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
667 (layout::move_to_column): Likewise.
669 2017-11-28 Jakub Jelinek <jakub@redhat.com>
672 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
673 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
675 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
676 Martin Jambor <mjambor@suse.cz>
679 * tree.h (expr_type_first_operand_type_p): Declare
680 * tree.c (expr_type_first_operand_type_p): New function.
681 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
682 (ipa_value_from_jfunc): Adjust declaration.
683 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
684 Use it as result type for arithmetics, unless it is NULL in which case
685 be more conservative.
686 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
687 ipa_get_jf_pass_through_result.
688 (propagate_vals_across_pass_through): Likewise.
689 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
690 is to propagate_vals_across_pass_through.
691 (propagate_constants_across_call): Pass PARM_TYPE to
692 propagate_scalar_across_jump_function.
693 (find_more_scalar_values_for_callers_subset): Pass parameter type to
694 ipa_value_from_jfunc.
695 (cgraph_edge_brings_all_scalars_for_node): Likewise.
696 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
697 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
698 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
699 target_type, pass it to ipa_value_from_jfunc.
700 (update_indirect_edges_after_inlining): Pass parameter type to
701 try_make_edge_direct_simple_call.
703 2017-11-28 Jeff Law <law@redhat.com>
705 * gimple-ssa-evrp-analyze.c
706 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
707 refine ranges if scev_initialized_p returns true.
708 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
710 2017-11-28 Julia Koval <julia.koval@intel.com>
712 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
713 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
714 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
715 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
716 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
717 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
718 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
719 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
720 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
721 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
722 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
723 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
724 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
725 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
726 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
727 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
728 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
729 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
730 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
731 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
732 __builtin_ia32_vpshrd_v2di_mask): New builtins.
733 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
735 2017-11-28 Julia Koval <julia.koval@intel.com>
737 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
738 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
739 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
740 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
741 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
742 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
743 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
744 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
745 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
746 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
747 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
748 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
749 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
750 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
751 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
752 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
753 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
754 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
755 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
756 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
757 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
758 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
759 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
760 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
761 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
762 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
763 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
764 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
765 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
766 __builtin_ia32_vpshld_v2di_mask): New builtins.
767 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
768 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
770 2017-11-28 Richard Biener <rguenther@suse.de>
772 PR tree-optimization/80776
773 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
775 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
777 (evrp_range_analyzer::record_ranges_from_incoming_edges):
778 If the incoming edge is an effective fallthru because the other
779 edge only reaches a __builtin_unreachable () then record ranges
780 derived from the controlling condition in SSA info.
781 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
782 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
784 2017-11-28 Olivier Hainque <hainque@adacore.com>
786 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
787 on mingw build hosts.
789 2017-11-28 Jakub Jelinek <jakub@redhat.com>
792 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
794 2017-11-28 Richard Biener <rguenther@suse.de>
797 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
798 copies generated from memcpy use a character array as reference
801 2017-11-28 Julia Koval <julia.koval@intel.com>
802 Sebastian Peryt <sebastian.peryt@intel.com>
804 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
805 c-family/c-cilkplus.o, c-family/array-notation-common.o,
806 cilk-common.o, cilk.h, cilk-common.c): Remove.
808 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
809 * builtins.c (is_builtin_name): Remove cilkplus condition.
810 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
811 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
812 cilk-builtins.def, cilkplus.def): Remove.
813 * cif-code.def (CILK_SPAWN): Remove.
814 * cilk-builtins.def: Delete.
815 * cilk-common.c: Ditto.
817 * cilkplus.def: Ditto.
818 * config/darwin.h (fcilkplus): Delete.
819 * cppbuiltin.c: Ditto.
820 * doc/extend.texi: Remove cilkplus doc.
821 * doc/generic.texi: Ditto.
822 * doc/invoke.texi: Ditto.
823 * doc/passes.texi: Ditto.
824 * gcc.c (fcilkplus): Remove.
825 * gengtype.c (cilk.h): Remove.
826 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
828 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
830 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
832 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
833 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
835 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
836 inline_read_section): Ditto.
837 * ipa-inline-analysis.c (cilk.h): Remove.
838 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
839 * lto-wrapper.c (merge_and_complain, append_compiler_options,
840 append_linker_options): Remove condition for fcilkplus.
841 * lto/lto-lang.c (cilk.h): Remove.
842 (lto_init): Remove condition for fcilkplus.
843 * omp-expand.c (expand_cilk_for_call): Delete.
844 (expand_omp_taskreg, expand_omp_for_static_chunk,
845 expand_omp_for): Remove cilkplus
847 (expand_cilk_for): Delete.
848 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
849 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
850 execute_lower_omp, diagnose_sb_0): Ditto.
851 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
852 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
853 * tree-nested.c: Ditto.
854 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
855 (dump_generic_node): Ditto.
856 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
857 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
859 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
861 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
863 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
864 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
865 for vector mode and !TARGET_SIMD.
867 2017-11-28 Jakub Jelinek <jakub@redhat.com>
869 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
871 * tree.h (SWITCH_LABELS): Remove.
872 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
873 assert SWITCH_BODY is non-NULL.
874 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
876 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
878 PR tree-optimization/80788
879 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
880 has TREE_OVERFLOW set, call drop_tree_overflow.
882 2017-11-28 Richard Biener <rguenther@suse.de>
884 PR tree-optimization/83158
885 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more
888 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
891 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
892 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
894 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
896 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
897 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
899 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
902 * varasm.c (compare_constant): Take the mode of the constants into
903 account when comparing floating point constants.
905 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
907 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
908 from explicit instantiation of debug_helper.
909 * vec.h (DEFINE_DEBUG_VEC): Ditto.
911 2017-11-27 Richard Biener <rguenther@suse.de>
913 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
916 2017-11-27 Richard Biener <rguenther@suse.de>
918 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
919 (type_hash_canon): Also clear int_cst_hash_table entry for
921 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
923 2017-11-27 Tamar Christina <tamar.christina@arm.com>
925 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
926 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
927 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
929 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
931 * hash-map.h (gt_cleare_cache): Avoid UB.
933 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
935 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
936 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
937 (decide_unroll_runtime_iterations): Likewise.
938 (decide_unroll_stupid): Likewise.
940 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
943 * config/i386/i386.md: Add a loop with incssp.
945 2017-11-27 Martin Jambor <mjambor@suse.cz>
947 PR tree-optimization/81248
948 * tree-sra.c (splice_param_accesses): Remove size check.
949 (decide_one_param_reduction): Fix size check.
950 * gimple-pretty-print.c (dump_profile): Silence warning.
951 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
953 2017-11-27 Jakub Jelinek <jakub@redhat.com>
956 * dbxout.c (lastlineno): New variable.
957 (dbx_debug_hooks): Use dbxout_switch_text_section as
958 switch_text_section debug hook.
959 (dbxout_function_end): Switch to current_function_section
960 rather than function_section. If crtl->has_bb_partition,
961 output just one N_FUN, depending on in_cold_section_p.
962 (dbxout_source_line): Remember last lineno in lastlineno.
963 (dbxout_switch_text_section): New function.
964 (dbxout_function_decl): Adjust dbxout_block caller.
965 (dbx_block_with_cold_children): New function.
966 (dbxout_block): Return true if any LBRAC/RBRAC have been
967 emitted. Use dbx_block_with_cold_children at depth == 0
968 in second partition. Add PARENT_BLOCKNUM argument, pass
969 it optionally adjusted to children. Output LBRAC/RBRAC
970 around recursive call only if the block is in the current
971 partition, if not and anything was output, emit empty
973 * final.c (final_scan_insn): Compute cold_function_name
974 before calling switch_text_section debug hook. Call
975 that hook even if dwarf2out_do_frame if not emitting
979 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
982 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
984 PR rtl-optimization/82488
985 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
987 2017-11-26 Julia Koval <julia.koval@intel.com>
989 * config/i386/i386.c (processor_target_table): Add skylake_cost for
991 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
994 2017-11-26 Julia Koval <julia.koval@intel.com>
996 * config/i386/driver-i386.c (host_detect_local_cpu):
997 Detect skylake-avx512.
999 2017-11-26 Julia Koval <julia.koval@intel.com>
1001 * config.gcc: Add -march=cannonlake.
1002 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
1003 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
1004 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
1005 (PTA_CANNONLAKE): New.
1006 (processor_target_table): Add cannonlake.
1007 (ix86_option_override_internal): Ditto.
1008 (fold_builtin_cpu): Ditto.
1009 (get_builtin_code_for_version): Handle cannonlake.
1010 (M_INTEL_COREI7_CANNONLAKE): New.
1011 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
1012 * doc/invoke.texi: Add -march=cannonlake.
1014 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
1016 * plugin.c (add_new_plugin): Use platform-specific library extensions.
1017 (try_init_one_plugin): Alternative implementation for MinGW.
1018 * Makefile.in (plugin_implib): New.
1019 (gengtype-lex.c): Fix broken AIX workaround.
1020 * configure: Regenerate.
1021 * doc/plugins.texi: Document support for MinGW.
1023 2017-11-25 Jakub Jelinek <jakub@redhat.com>
1025 PR rtl-optimization/81553
1026 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
1027 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
1028 is a shift where C1 has different mode than the whole shift, use C1's
1029 mode for MULT rather than the shift's mode.
1032 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
1033 builtins not enabled in the currently selected ISA.
1035 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
1037 PR tree-optimization/71026
1038 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
1039 (insert_reciprocals): Change to insert reciprocals before a division
1040 by a square and to insert the square of a reciprocal.
1041 (execute_cse_reciprocals_1): Change to consider division by a square.
1042 (register_division_in): Add importance parameter.
1044 2017-11-24 Richard Biener <rguenther@suse.de>
1046 PR tree-optimization/82402
1047 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
1048 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
1050 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
1052 * match.pd (0-ptr): New transformation.
1054 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
1057 * ipa-inline.c (inline_small_functions): Set current badnes correctly
1058 when skipping checking.
1060 2017-11-24 Richard Biener <rguenther@suse.de>
1062 PR tree-optimization/83128
1063 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
1064 (vn_reference_lookup_3): Likewise.
1066 2017-11-24 Jakub Jelinek <jakub@redhat.com>
1069 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
1070 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
1073 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
1074 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
1075 check for the latter with gimple_call_builtin_p. Do not handle
1076 BUILT_IN_STPNCPY_CHK which is not a pass through call.
1078 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
1080 * config/arm/arm_neon.h: Fix pragma GCC push_options before
1083 2017-11-23 Julia Koval <julia.koval@intel.com>
1085 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
1086 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
1087 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
1088 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
1089 _mm512_maskz_expandloadu_epi16): New intrinsics.
1090 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
1091 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
1092 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
1093 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
1094 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
1095 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
1096 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
1097 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
1098 _mm256_maskz_expandloadu_epi8): New intrinsics.
1099 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
1100 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
1101 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
1102 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
1103 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
1104 * config/i386/sse.md (VI248_VLBW): New iterator.
1105 (expand<mode>_mask, expand<mode>_maskz): New patterns.
1107 2017-11-23 Julia Koval <julia.koval@intel.com>
1109 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
1110 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
1111 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
1112 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
1113 _mm512_mask_compressstoreu_epi16): New.
1114 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
1115 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
1116 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
1117 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
1118 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
1119 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
1120 _mm256_mask_compressstoreu_epi8): New.
1121 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
1122 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
1123 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
1124 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
1125 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
1126 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
1127 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
1128 __builtin_ia32_compresshi128_mask,
1129 __builtin_ia32_compressstoreuqi512_mask,
1130 __builtin_ia32_compressstoreuhi512_mask,
1131 __builtin_ia32_compressstoreuqi256_mask,
1132 __builtin_ia32_compressstoreuqi128_mask,
1133 __builtin_ia32_compressstoreuhi256_mask,
1134 __builtin_ia32_compressstoreuhi128_mask): New builtins.
1135 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
1137 (ix86_expand_special_args_builtin): Handle new types.
1138 (s4fma_expand): Handle new builtin array.
1139 * config/i386/immintrin.h: Include new headers.
1140 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
1141 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
1143 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1145 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
1146 fixes. Declare temp and g variables at the top in order to avoid
1147 {} in most of the cases.
1149 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
1151 * match.pd (ptr-0): New transformation.
1153 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
1155 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
1156 (struct addr_mode_cost_table): New.
1157 (struct tune_params): Add field addr_mode_costs.
1158 * config/arm/arm.c (generic_addr_mode_costs): New.
1159 (arm_slowmul_tune): Initialise addr_mode_costs field.
1160 (arm_fastmul_tune): Likewise.
1161 (arm_strongarm_tune): Likewise.
1162 (arm_xscale_tune): Likewise.
1163 (arm_9e_tune): Likewise.
1164 (arm_marvell_pj4_tune): Likewise.
1165 (arm_v6t2_tune): Likewise.
1166 (arm_cortex_tune): Likewise.
1167 (arm_cortex_a8_tune): Likewise.
1168 (arm_cortex_a7_tune): Likewise.
1169 (arm_cortex_a15_tune): Likewise.
1170 (arm_cortex_a35_tune): Likewise.
1171 (arm_cortex_a53_tune): Likewise.
1172 (arm_cortex_a57_tune): Likewise.
1173 (arm_exynosm1_tune): Likewise.
1174 (arm_xgene1_tune): Likewise.
1175 (arm_cortex_a5_tune): Likewise.
1176 (arm_cortex_a9_tune): Likewise.
1177 (arm_cortex_a12_tune): Likewise.
1178 (arm_cortex_a73_tune): Likewise.
1179 (arm_v7m_tune): Likewise.
1180 (arm_cortex_m7_tune): Likewise.
1181 (arm_v6m_tune): Likewise.
1182 (arm_fa726te_tune): Likewise.
1183 (arm_mem_costs): Use table lookup to calculate cost of addressing
1186 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
1188 * config/arm/arm.c (arm_mem_costs): New function.
1189 (arm_rtx_costs_internal): Use arm_mem_costs.
1191 2017-11-23 Mark Wielaard <mark@klomp.org>
1193 * dwarf2out.c (init_sections_and_labels): Use generation to create
1194 unique ranges_section_label and ranges_base_label. Return generation.
1195 (output_rnglists): Add generation argument. Use generation to create
1196 unique ranges labels.
1197 (dwarf2out_finish): Get generation from init_sections_and_labels
1198 and pass generation to output_rnglists.
1200 2017-11-23 Mike Stump <mikestump@comcast.net>
1201 Eric Botcazou <ebotcazou@adacore.com>
1203 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
1204 * cfgloop.h (struct loop): Add unroll field.
1205 * function.h (struct function): Add has_unroll bitfield.
1206 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
1207 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
1208 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
1209 (pass_rtl_unroll_loops::gate): Likewise.
1210 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
1211 for which loop->unroll==1.
1212 (decide_unroll_constant_iterations): Use note for consistency and
1213 take loop->unroll into account. Return early if loop->unroll is set.
1214 Fix thinko in existing test.
1215 (decide_unroll_runtime_iterations): Use note for consistency and
1216 take loop->unroll into account.
1217 (decide_unroll_stupid): Likewise.
1218 * lto-streamer-in.c (input_cfg): Read loop->unroll.
1219 * lto-streamer-out.c (output_cfg): Write loop->unroll.
1220 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
1222 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
1223 (print_loop): Print loop->unroll if set.
1224 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
1225 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
1226 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
1228 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
1229 loop->unroll is set and smaller than the trip count. Otherwise bypass
1230 entirely the heuristics if loop->unroll is set. Remove dead note.
1231 Fix off-by-one bug in other note.
1232 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
1233 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
1235 (tree_unroll_loops_completely): Make static.
1236 (pass_complete_unroll::execute): Use correct type for variable.
1237 (pass_complete_unrolli::execute): Fix formatting.
1238 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
1240 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
1242 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
1243 enable when TARGET_PREFER_AVX128 is set.
1245 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1247 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
1249 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
1250 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
1252 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1254 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
1255 frequencies. Fix estimation of aggregate parameters.
1257 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1259 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
1262 2017-11-23 Tom de Vries <tom@codesourcery.com>
1264 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
1266 2017-11-23 Tom de Vries <tom@codesourcery.com>
1268 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
1271 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
1274 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
1275 sibcall_value_pcrel_fdpic): Use local variable instead of
1277 (calli_tbr_rel): Add missing operand 2.
1278 (call_valuei_tbr_rel): Add missing operand 3.
1280 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1283 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
1284 bitpos/bitsize covering the whole destination, use store_expr only if
1285 the complex mode is the same. Otherwise, use expand_normal and if
1286 it returns CONCAT, subreg each part separately instead of trying to
1287 subreg the whole result.
1289 2017-11-23 Richard Biener <rguenther@suse.de>
1291 PR tree-optimization/23094
1292 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
1293 come here from walking over backedges in the first iteration.
1294 (vn_reference_lookup_3): Skip clobbers that store the same value.
1296 2017-11-23 Richard Biener <rguenther@suse.de>
1298 PR tree-optimization/81403
1299 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
1300 a block we need a leader relative to.
1301 (phi_translate_1): For nary processing require a leader from
1302 get_representative_for given we run expression simplification
1303 using match-and-simplify. Remove previous fix.
1305 2017-11-22 Jeff Law <law@redhat.com>
1307 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
1308 Use new method allocate_value_range rather than accessing the
1309 vrp_value_range_pool data member directly.
1310 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
1311 to use extract_range_from_stmt method to avoid need for
1312 extract_range_from_assignment method.
1313 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
1314 method rather than setting values_propgated data member directly.
1315 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
1316 and values propagated data members and extract_range_from_assignment
1317 method. Reorder private data members to conform to standards.
1318 Add new methods set_lattice_propagation_complete and
1319 allocate_value_range.
1321 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
1323 PR rtl-optimization/83030
1324 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
1325 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
1326 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
1327 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
1328 CROSSING_JUMP_P flag is already set before setting it.
1329 * cfgrtl.c (fixup_partition_crossing): Likewise.
1330 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
1333 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1335 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
1336 Handle the case where both arguments are using gen_const_vec_series.
1338 2017-11-22 David Malcolm <dmalcolm@redhat.com>
1341 * pretty-print.c (pp_format): Move quoting implementation to
1342 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
1343 to pass address of "quote" local.
1344 (pp_begin_quote): New function.
1345 (pp_end_quote): New function.
1346 * pretty-print.h (printer_fn): Convert penultimate param from bool
1348 (pp_begin_quote): New decl.
1349 (pp_end_quote): New decl.
1350 * tree-diagnostic.c (default_tree_printer): Convert penultimate
1351 param from bool to bool *.
1352 * tree-diagnostic.h (default_tree_printer): Likewise.
1354 2017-11-22 Jeff Law <law@redhat.com>
1356 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
1357 Initialize vr_values.
1358 (evrp_range_analyzer::try_find_new_range): Call methods attached to
1359 vr_values via vr_values class instance rather than delegators.
1360 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
1361 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1362 (evrp_range_analyzer::push_value_range): Likewise.
1363 (evrp_range_analyzer::pop_value_range): Likewise.
1364 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
1365 most delegators. Those remaining are exposed as public interfaces.
1366 Make vr_values a pointer and private.
1367 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
1369 (evrp_range_analyzer::get_vr_value): New method.
1370 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
1371 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
1372 (class evrp_dom_walker): Attach evrp_folder class, initialize
1373 it in the ctor. Remove temporary delegators.
1374 (evrp_dom_walker::before_dom_children): Call methods in attached
1375 evrp_range_analyzer class via class instance pointer. Use
1376 free value_range_constant_singleton to remove need for
1377 op_with_constant_singleton_value delegator method. Do not
1378 create a vrp_prop class instance for every call! Narrow
1379 scope of a couple variables.
1380 (evrp_dom_walker::cleanup): Call methods in attached
1381 evrp_range_analyzer class via class instance pointer.
1382 * vr-values.h (class vr_values): Privatize many methods and
1385 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1387 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
1388 * cfgexpand.c (expand_debug_expr): Remove handling for them.
1389 * expr.c (expand_expr_real_2): Likewise.
1390 * fold-const.c (const_unop): Likewise.
1391 * optabs-tree.c (optab_for_tree_code): Likewise.
1392 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1393 * tree-inline.c (estimate_operator_cost): Likewise.
1394 * tree-pretty-print.c (dump_generic_node): Likewise.
1395 (op_code_prio): Likewise.
1396 (op_symbol_code): Likewise.
1397 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
1398 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
1399 * internal-fn.c (direct_internal_fn_optab): New function.
1400 (direct_internal_fn_array, direct_internal_fn_supported_p
1401 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
1402 * fold-const-call.c (fold_const_reduction): New function.
1403 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
1405 * tree-vect-loop.c: Include internal-fn.h.
1406 (reduction_code_for_scalar_code): Rename to...
1407 (reduction_fn_for_scalar_code): ...this and return an internal
1409 (vect_model_reduction_cost): Take an internal_fn rather than
1411 (vect_create_epilog_for_reduction): Likewise. Build calls rather
1413 (vectorizable_reduction): Use internal functions rather than tree
1414 codes for the reduction operation. Update calls to the functions
1416 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
1417 Use calls to internal functions rather than REDUC tree codes.
1418 * config/aarch64/aarch64-simd.md: Update comment accordingly.
1420 2017-11-22 Olivier Hainque <hainque@adacore.com>
1422 * config/vxworks.c (vxworks_override_options): Pick default
1423 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
1424 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
1425 DWARF_GNAT_ENCODINGS_DEFAULT.
1426 * config/vxworksae.h: Likewise.
1428 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
1430 PR tree-optimization/83104
1431 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
1434 2017-11-22 Marek Polacek <polacek@redhat.com>
1435 H.J. Lu <hongjiu.lu@intel.com>
1436 Jason Merrill <jason@redhat.com>
1441 * calls.c (initialize_argument_information): Call
1442 warn_parameter_passing_abi target hook.
1443 (store_one_arg): Use 0 for empty record size. Don't push 0 size
1444 argument onto stack.
1445 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
1446 * common.opt: Update -fabi-version description.
1447 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
1448 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
1450 (ix86_is_empty_record): New function.
1451 (ix86_warn_parameter_passing_abi): New function.
1452 (TARGET_EMPTY_RECORD_P): Redefine.
1453 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
1454 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
1455 * doc/tm.texi: Regenerated.
1456 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
1457 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
1458 * dwarf2out.c (get_ultimate_context): Move to tree.c.
1459 * explow.c (hard_function_value): Call arg_int_size_in_bytes
1460 instead of int_size_in_bytes.
1461 * expr.c (copy_blkmode_to_reg): Likewise.
1462 * function.c (aggregate_value_p): Return 0 for empty types.
1463 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi target hook.
1464 (locate_and_pad_parm): Call arg size_in_bytes instead
1466 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
1467 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
1468 * target.def (empty_record_p, warn_parameter_passing_abi): New target
1470 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
1471 (std_gimplify_va_arg_expr): Skip empty records. Call
1472 arg_size_in_bytes instead size_in_bytes.
1473 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
1474 * tree-core.h (tree_type_common): Add empty_flag.
1475 (tree_decl_common): Update comments.
1476 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
1478 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1479 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
1481 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1482 * tree.c (default_is_empty_type): New function.
1483 (default_is_empty_record): New function.
1484 (arg_int_size_in_bytes): New function.
1485 (arg_size_in_bytes): New function.
1486 (get_ultimate_context): New function.
1487 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
1488 TRANSLATION_UNIT_WARN_EMPTY_P.
1489 (default_is_empty_record, arg_int_size_in_bytes,
1490 arg_size_in_bytes, get_ultimate_context): Declare.
1492 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1494 * config/arm/arm.c (cmse_clear_registers): New function.
1495 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
1496 code by call to cmse_clear_registers.
1497 (cmse_nonsecure_entry_clear_before_return): Likewise.
1499 2017-11-22 Tamar Christina <tamar.christina@arm.com>
1501 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
1502 (vdot_s32, vdotq_s32): New.
1503 (vdot_lane_u32, vdotq_lane_u32): New.
1504 (vdot_lane_s32, vdotq_lane_s32): New.
1507 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1510 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
1511 for unsigned values with fewer HWIs than the precision.
1512 (test_overflow): New function.
1513 (wide_int_cc_tests): Call it.
1515 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1517 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
1518 has the same mode class as Pmode.
1520 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1522 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
1523 padding_bits_to_clear_ptr.
1524 (cmse_nonsecure_entry_clear_before_return): Likewise.
1526 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1528 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
1529 auto_sbitap instead of integer bitfield to control register needing
1532 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1534 PR tree-optimization/83044
1535 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
1536 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
1537 that need the upper bound. Subtract offset from
1538 get_addr_base_and_unit_offset only if positive and subtract it
1539 before division by eltsize rather than after it.
1542 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
1543 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
1544 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
1548 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
1549 expand_binop with *mul_widen_optab, make sure at least one of the
1550 operands doesn't have VOIDmode.
1553 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
1555 PR rtl-optimization/82044
1556 PR tree-optimization/82042
1557 * dse.c (record_store): Check for overflow.
1558 (check_mem_read_rtx): Properly check for overflow if width == -1, call
1559 add_wild_read instead of clear_rhs_from_active_local_stores on
1560 overflow and log it into dump_file.
1562 2017-11-22 Richard Biener <rguenther@suse.de>
1564 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
1565 fake edges to exit when looking for a place to insert.
1566 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
1568 (insert_into_preds_of_block): Commit edge insertion immediately,
1569 assert that doesn't require new BBs.
1570 (fini_pre): Release expressions.
1571 (pass_pre::execute): Shuffle things around a bit, if the fn
1572 is too large do not compute AVAIL either as this is really the
1575 2017-11-22 Richard Biener <rguenther@suse.de>
1577 PR tree-optimization/83089
1578 * tree-if-conv.c (pass_if_conversion::execute): If anything
1579 changed reset SCEV and free the number of iteration estimates.
1581 2017-11-21 Martin Sebor <msebor@redhat.com>
1583 PR tree-optimization/82945
1584 * calls.h (warn_nonstring_bound): Remove unused function.
1586 2017-11-21 Martin Sebor <msebor@redhat.com>
1588 PR tree-optimization/82945
1589 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
1590 * calls.h (maybe_warn_nonstring_arg): Declare new function.
1591 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
1593 (initialize_argument_information): Call maybe_warn_nonstring_arg.
1594 * calls.h (get_attr_nonstring_decl): Declare new function.
1595 * doc/extend.texi (attribute nonstring): Update.
1596 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
1597 get_attr_nonstring_decl and handle it.
1598 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
1599 detection of nul-termination.
1600 (strlen_to_stridx): Change to a pointer.
1601 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
1602 (pass_strlen::execute): Same.
1604 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
1606 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
1607 for the new option -mprefer-vector-width=[none|128|256|512].
1608 * config/i386/i386.c (ix86_target_string): remove old style options
1609 -mprefer-avx256 and make -mprefer-avx128 as alias.
1610 (ix86_option_override_internal): Apply defaults for the
1611 -mprefer-vector-width=[128|256] option.
1612 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
1613 Implement macros to work with -mprefer-vector-width=.
1614 * config/i386/i386.opt: Implemented option
1615 -mprefer-vector-width=[none|128|256|512].
1616 * doc/invoke.texi: Documentation for
1617 -mprefer-vector-width=[none|128|256|512].
1619 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
1621 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
1623 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
1625 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
1627 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
1629 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
1631 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
1632 * expr.c (expand_expr_real_2): Likewise.
1633 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
1634 fold_binary_loc): Likewise.
1635 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
1636 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
1637 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
1638 MINUS_EXPR transformations.
1639 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
1640 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
1641 * tree-inline.c (estimate_operator_cost): Likewise.
1642 * tree-pretty-print.c (dump_generic_node, op_code_prio,
1643 op_symbol_code): Likewise.
1644 * tree-vect-stmts.c (vectorizable_operation): Likewise.
1645 * vr-values.c (extract_range_from_binary_expr): Likewise.
1646 * varasm.c (initializer_constant_valid_p_1): Likewise.
1647 * tree.def: New tree code POINTER_DIFF_EXPR.
1649 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
1651 * config/i386/i386.md (*bswap<mode>2_movbe): Add
1652 integer suffix to movbe mnemonic.
1653 (*bswaphi2_movbe): Ditto.
1654 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
1656 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1659 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
1660 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
1661 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
1663 2017-11-21 Martin Liska <mliska@suse.cz>
1665 * tree-inline.c (expand_call_inline): Remove not needed
1668 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
1669 Jakub Jelinek <jakub@redhat.com>
1672 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
1673 Remove static keyword from f variable.
1675 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1677 PR tree-optimization/83086
1678 * gimple-ssa-store-merging.c
1679 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
1680 rather than n.base_addr.
1682 2017-11-21 Martin Liska <mliska@suse.cz>
1684 PR rtl-optimization/82044
1685 PR tree-optimization/82042
1686 * dse.c (check_mem_read_rtx): Check for overflow.
1688 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1690 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
1693 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
1695 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
1696 and its corresponding call arg location note.
1698 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
1699 Andrew Burgess <andrew.burgess@embecosm.com>
1701 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
1703 (arc_return_slot_offset): Likewise.
1704 (arc_eh_return_address_location): New declaration.
1705 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
1706 (MUST_SAVE_REGISTER): Add exception handler case.
1707 (MUST_SAVE_RETURN_ADDR): Likewise.
1708 (arc_frame_pointer_required): Likewise.
1709 (arc_frame_pointer_needed): New function.
1710 (arc_compute_frame_size): Changed.
1711 (arc_expand_prologue): Likewise.
1712 (arc_expand_epilogue): Likewise.
1713 (arc_initial_elimination_offset): Likewise.
1714 (arc_return_slot_offset): Delete.
1715 (arc_eh_return_address_location): New function.
1716 (arc_builtin_setjmp_frame_value): Likewise.
1717 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
1718 (EH_RETURN_STACKADJ_RTX): Define.
1719 (EH_RETURN_HANDLER_RTX): Likewise.
1720 * config/arc/arc.md (eh_return): Delete.
1722 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1724 * print-tree.h (debug_vec_tree): Remove prototype.
1725 * gdbinit.in (pvt): Remove macro.
1727 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1729 PR tree-optimization/83047
1730 * gimple-ssa-store-merging.c
1731 (imm_store_chain_info::output_merged_store): If the loads with the
1732 same vuse are in different basic blocks, for load_gsi pick a load
1733 location that is dominated by the other loads.
1736 * config/i386/i386.c (ix86_memmodel_check): Start
1737 -Winvalid-memory-model diagnostics with lowercase letter.
1740 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
1741 set in_cold_section_p to the partition containing loc_list->first.
1742 When seeing loc_list->last_before_switch node, update secname and
1743 perform range_across_switch second partition handling only after that.
1746 * run-rtl-passes.c: Include debug.h.
1747 (run_rtl_passes): Call debug_hooks->assembly_start.
1748 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
1752 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
1753 OPTAB_DIRECT in calls to expand_simple_binop.
1755 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1758 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
1760 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1763 * spellcheck.h (best_match::blithely_get_best_candidate): New
1766 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1768 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
1769 with lower case letter.
1771 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
1773 * config/i386/i386.md (bswaphi2): New expander.
1774 (*bswaphi2_movbe): New insn pattern.
1775 (bswaphi -> rorhi pepehole2): New peephole pattern.
1777 2017-11-20 Jeff Law <law@redhat.com>
1779 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
1780 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
1781 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
1782 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
1783 gimple-ssa-evrp-analyze.h.
1785 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
1786 set BB_VISITED here.
1787 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
1789 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1791 * tree-predcom.c: Add general comment on Store-Store chains.
1792 (split_data_refs_to_components): Postpone clearing eliminate_store_p
1794 (get_chain_last_ref_at): Rename into...
1795 (get_chain_last_write_at): ...this.
1796 (get_chain_last_write_before_load): New function.
1797 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
1798 CT_STORE_STORE when write reference is added.
1799 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
1800 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
1801 (initialize_root_vars_store_elim_1): Ditto.
1802 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
1803 definition is created.
1804 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
1805 chain by replacing it with dominant stored value.
1807 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1809 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
1811 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
1813 * vr-values.c (extract_range_from_binary_expr): Use a full range
1816 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
1818 * config/arm/arm.md (R4_REGNUM): Define constant.
1819 (nonsecure_call_internal): Remove r4 clobber.
1820 (nonsecure_call_value_internal): Likewise.
1821 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
1822 clobber and resequence match_operands.
1823 (nonsecure_call_value_reg_thumb1_v5): Likewise.
1824 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
1825 (nonsecure_call_value_reg_thumb2): Likewise.
1827 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1829 PR tree-optimization/78821
1830 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
1831 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
1832 to the address of the base rather than the base itself.
1833 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
1834 (find_bswap_or_nop_finalize): New function.
1835 (find_bswap_or_nop): Use it.
1836 (bswap_replace): Return a tree rather than bool, change first
1837 argument from gimple * to gimple_stmt_iterator, allow inserting
1838 into an empty sequence, allow ins_stmt to be NULL - then emit
1839 all stmts into gsi. Fix up MEM_REF address gimplification.
1840 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
1841 (struct store_immediate_info): Add N and INS_STMT non-static
1843 (store_immediate_info::store_immediate_info): Initialize them
1844 from newly added ctor args.
1845 (merged_store_group::apply_stores): Formatting fixes. Sort by
1847 (stmts_may_clobber_ref_p): For stores call also
1848 refs_anti_dependent_p.
1849 (gather_bswap_load_refs): New function.
1850 (imm_store_chain_info::try_coalesce_bswap): New method.
1851 (imm_store_chain_info::coalesce_immediate_stores): Use it.
1852 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
1853 (imm_store_chain_info::output_merged_store): Fail if number of
1854 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
1855 and NOP_EXPR rhs_code.
1856 (pass_store_merging::process_store): Compute n and ins_stmt, if
1857 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
1858 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
1860 (pass_store_merging::execute): Calculate dominators.
1862 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
1863 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
1864 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
1865 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
1866 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
1867 class pass_optimize_bswap, bswap_replace,
1868 pass_optimize_bswap::execute): Moved to ...
1869 * gimple-ssa-store-merging.c: ... this file.
1870 Include optabs-tree.h.
1871 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
1872 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
1873 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
1874 anonymous namespace, remove static keywords.
1875 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
1876 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
1878 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
1881 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
1883 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
1885 * vec.h (debug_helper): New function.
1886 (DEFINE_DEBUG_VEC): New macro.
1887 * hash-set.h (debug_helper): New function.
1888 (DEFINE_DEBUG_HASH_SET): New macro.
1889 * cfg.c (debug_slim (edge)): New function.
1890 Call DEFINE_DEBUG_VEC for edges.
1891 Call DEFINE_DEBUG_HASH_SET for edges.
1892 * cfghooks.c (debug_slim (basic_block)): New function.
1893 Call DEFINE_DEBUG_VEC for basic blocks.
1894 Call DEFINE_DEBUG_HASH_SET for basic blocks.
1895 * print-tree.c (debug_slim): New function to handle trees.
1896 Call DEFINE_DEBUG_VEC for trees.
1897 Call DEFINE_DEBUG_HASH_SET for trees.
1898 (debug (vec<tree, va_gc>) &): Remove.
1899 (debug (<vec<tree, va_gc>) *): Remove.
1900 * print-rtl.c (debug_slim): New function to handle const_rtx.
1901 Call DEFINE_DEBUG_VEC for rtx_def.
1902 Call DEFINE_DEBUG_VEC for rtx_insn.
1903 Call DEFINE_DEBUG_HASH_SET for rtx_def.
1904 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
1905 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
1906 (debug (vec<rtx_insn *> *ptr): Remove.
1907 (debug_insn_vector): Remove.
1908 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
1910 2017-11-20 Tom de Vries <tom@codesourcery.com>
1912 PR rtl-optimization/82020
1913 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
1914 IF_THEN_ELSE condition.
1916 2017-11-19 Jeff Law <law@redhat.com>
1918 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1919 of degenerates resulting from ignoring an edge.
1921 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1924 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
1926 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1929 * profile-count.c (profile_count::to_sreal_scale): Fix return value
1930 for uninitialied counts.
1932 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1935 * tree-inline.c (estimate_num_insns): Set to 1 at least.
1937 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1940 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
1942 2017-11-19 Tom de Vries <tom@codesourcery.com>
1944 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
1947 2017-11-19 Tom de Vries <tom@codesourcery.com>
1949 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
1952 2017-11-19 Tom de Vries <tom@codesourcery.com>
1954 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1957 2017-11-19 Tom de Vries <tom@codesourcery.com>
1959 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
1961 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
1963 2017-11-19 Tom de Vries <tom@codesourcery.com>
1965 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
1967 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
1968 semicolon after MCORE_EXPORT_NAME call.
1970 2017-11-19 Tom de Vries <tom@codesourcery.com>
1973 * vmsdbgout.c (vmsdbgout_early_finish): New function.
1974 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
1976 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
1978 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
1979 (cgraph_edge::clone): Cleanup updating of profile.
1980 * ipa-cp.c (update_profiling_info): Likewise.
1981 * ipa-inline-transform.c (inline_transform): Likewise.
1982 * ipa-inline.c (inline_small_functions): Add missing space to dump.
1983 * ipa-split.c (execute_split_functions): Do not split when function
1985 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
1986 * profile-count.c (profile_count::dump): Add global0.
1987 (profile_count::to_cgraph_frequency): Do not ICE when entry is
1989 (profile_count::to_sreal_scale): Likewise.
1990 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
1991 (profile_count::combine_with_ipa_count): New function.
1992 * profile-count.h (profile_guessed_global0adjusted): New.
1993 (profile_count::adjusted_zero): New.
1994 (profile_count::global0adjusted): New.
1995 (profile_count::combine_with_ipa_count): New.
1996 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
1997 correct profile of return block of split functions.
1998 (copy_cfg_body): Remove unused profile_count.
1999 (copy_body): Likewise.
2000 (expand_call_inline): Update.
2001 (tree_function_versioning): Update.
2003 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
2005 * hash-set.h (hash_set::empty): New.
2006 * tree-ssa-threadbackward.h: Delete.
2007 * tree-ssa-threadbackward.c (class thread_jumps): New.
2008 Move max_threaded_paths into class.
2009 (fsm_find_thread_path): Remove arguments that are now in class.
2010 (profitable_jump_thread_path): Rename to...
2011 (thread_jumps::profitable_jump_thread_path): ...this.
2012 (convert_and_register_jump_thread_path): Rename to...
2013 (thread_jumps::convert_and_register_current_path): ...this.
2014 (check_subpath_and_update_thread_path): Rename to...
2015 (thread_jumps::check_subpath_and_update_thread_path): ...this.
2016 (register_jump_thread_path_if_profitable): Rename to...
2017 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
2018 (handle_phi): Rename to...
2019 (thread_jumps::handle_phi): ...this.
2020 (handle_assignment): Rename to...
2021 (thread_jumps::handle_assignment): ...this.
2022 (fsm_find_control_statement_thread_paths): Rename to...
2023 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
2024 (find_jump_threads_backwards): Rename to...
2025 (thread_jumps::find_jump_threads_backwards): ...this.
2026 Initialize path local data.
2027 (pass_thread_jumps::execute): Call find_jump_threads_backwards
2028 from within thread_jumps class.
2029 (pass_early_thread_jumps::execute): Same.
2031 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2033 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
2035 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2037 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
2038 to canonical location.
2040 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2043 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
2045 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
2047 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
2049 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
2050 rcrt1.o%s/grcrt1.o%s for -static-pie.
2052 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2054 * i386.c (ix86_multiplication_cost, ix86_division_cost,
2055 ix86_shift_rotate_cost): Break out from ...
2056 (ix86_rtx_costs): ... here.
2057 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
2060 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2062 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
2063 when entry block was promoted unlikely.
2064 (estimate_bb_frequencies): Increase frequency scale.
2065 * profile-count.h (profile_count): Export precision info.
2067 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2069 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
2070 disturbin profile of entry block.
2072 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2074 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
2077 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2079 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
2080 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
2082 2017-11-17 Jeff Law <law@redhat.com>
2084 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
2085 from evrp_dom_walker class. Various methods moved into new class.
2086 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
2087 (evrp_range_analyzer::enter): New method.
2088 (evrp_range_analyzer::leave): New method.
2089 (evrp_dom_walker): Remove delegators no longer needed by this class.
2090 Replace vr_values data member with evrp_range_analyzer
2092 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
2093 method extracted from evrp_dom_walker::before_dom_children.
2094 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
2095 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
2097 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
2098 Add private copy constructor and move assignment operators.
2099 Privatize methods and class data where trivially possible.
2100 (evrp_dom_walker::cleanup): New function, extracted from
2101 execute_early_vrp. Simplify access to class data.
2103 * vr-values.h (get_output_for_vrp): Prototype.
2104 * vr-values.c (get_output_for_vrp): New function extracted from
2105 vrp_visit_assignment_or_call and extract_range_from_stmt.
2106 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
2108 2017-11-17 Luis Machado <luis.machado@linaro.org>
2110 * config/aarch64/aarch64.c
2111 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
2112 (qdf24xx_tunings) <autoprefetcher_model>: Set to
2113 tune_params::AUTOPREFETCHER_WEAK.
2115 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2118 * config/arm/arm.c (arm_valid_target_attribute_rec):
2119 Parse "arch=" and "+<ext>".
2120 (arm_valid_target_attribute_tree): Re-init global options.
2121 (arm_option_override): Make non-static.
2122 (arm_options_perform_arch_sanity_checks): Make errors fatal.
2123 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
2124 (__ARM_FEATURE_CRC32): Support undef.
2125 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
2126 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
2128 2017-11-17 David Malcolm <dmalcolm@redhat.com>
2130 * gdbinit.in (break-on-diagnostic): New command.
2132 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2134 * config/i386/i386.c (ix86_expand_epilogue): Change simple
2135 return to indirect jump for EH return if control-flow
2136 protection is enabled. Change explicit 'false' argument in
2137 pro_epilogue_adjust_stack with a value of
2139 * config/i386/i386.md (simple_return_indirect_internal):
2140 Remove SImode restriction to support 64-bit.
2142 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2144 * combine.c (added_notes_insn): New.
2145 (try_combine): Handle added_notes_insn like added_links_insn.
2146 Rewrite return value code.
2147 (distribute_notes): Set added_notes_insn to the earliest insn we added
2150 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2152 PR rtl-optimization/82621
2153 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
2154 dest of one of those SETs is unused.
2156 2017-11-17 Richard Biener <rguenther@suse.de>
2159 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
2160 * tree-pretty-print.c (dump_generic_node): Handle
2161 annot_expr_parallel_kind.
2162 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
2163 * gimplify.c (gimple_boolify): Likewise.
2165 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2167 * config.gcc (extra_headers): Add cet.h for x86 targets.
2168 * config/i386/cet.h: New file.
2169 * doc/install.texi: Add --enable-cet/--disable-cet.
2171 2017-11-17 Richard Biener <rguenther@suse.de>
2173 PR tree-optimization/83017
2174 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
2175 (gen_parallel_loop): Properly count iterations.
2176 (parallelize_loops): Handle loop->can_be_parallel independent
2177 of flag_loop_parallelize_all. Make static profitability test match
2179 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
2180 * invoke.texi (parloops-min-per-thread): Document.
2182 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
2184 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
2185 upstreaming review comments.
2187 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2189 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
2190 with fast unaligned access.
2191 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
2193 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
2195 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
2196 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
2198 2017-11-17 Richard Biener <rguenther@suse.de>
2200 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
2201 folding of references.
2203 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
2206 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
2207 of replacing call to strncmp with corresponding call to strcmp when
2210 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
2212 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
2213 option prefer-avx256 for skylake-avx512 configuration.
2214 * config/i386/i386.c (ix86_option_override_internal): Ditto.
2215 (get_builtin_code_for_version): Ditto.
2217 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
2218 Monk Chiang <sh.chiang04@gmail.com>
2220 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
2221 (FIXED_REGISTERS): Reserve more register numbers.
2222 (CALL_USED_REGISTERS): Likewise.
2223 (REG_ALLOC_ORDER): Likewise.
2224 (REG_CLASS_CONTENTS): Likewise.
2225 (REGISTER_NAMES): Likewise.
2227 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
2228 Kito Cheng <kito.cheng@gmail.com>
2230 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
2232 * config/nds32/iterators.md: Add vector mode iterators and attributes.
2234 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
2236 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
2237 parameter list for vec_splats.
2239 2017-11-16 Joseph Myers <joseph@codesourcery.com>
2241 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
2243 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
2245 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
2248 * doc/install.texi: Document incompatibility of Perl >=5.6.26
2249 with the required version of automake 1.11.6.
2251 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
2253 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
2255 (DU_C2_3_power9): Correct reservation combinations.
2256 (FP_DIV_power9, VEC_DIV_power9): New.
2257 (power9-alu): Split out rotate/shift...
2258 (power9-rot): ...to here, correct dispatch resource.
2259 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
2261 (power9-fp): Correct latency.
2262 (power9-sdiv): Add div/sqrt resource.
2263 (power9-ddiv): Correct latency, add div/sqrt resource.
2264 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
2265 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
2267 (power9-qpdiv, power9-qpmul): Adjust resource usage.
2269 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2271 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
2272 switch statement mapping KF built-ins to TF built-ins if we don't
2273 have the proper ISA 3.0 assembler support.
2275 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2277 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
2278 (lower_emutls_function_body): Do not compute it.
2280 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2282 * ipa-split.c (split_bb_info): Turn time to sreal.
2283 (split_point): Likewise.
2284 (dump_split_point): Likewise.
2285 (fine_split_points): Likewise.
2286 (execute_split_functions): Only zero split_bbs; turn time to sreals.
2288 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2290 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
2293 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2295 * predict.c (combine_predictions_for_bb): Preserve zero predicted
2297 (expensive_function_p): Remove useless assert.
2298 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
2300 2017-11-16 Martin Sebor <msebor@redhat.com>
2302 PR tree-optimization/82588
2303 PR tree-optimization/82583
2304 * tree-vrp.c (check_array_ref): Handle flexible array members,
2305 string literals, and inner indices.
2306 (search_for_addr_array): Add detail to diagnostics.
2308 2017-11-16 Nathan Sidwell <nathan@acm.org>
2312 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
2313 (SET_DECL_ASSEMBLER_NAME): Forward to
2314 overwrite_decl_assembler_name.
2315 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
2316 (overwrite_decl_assembler_name): Declare.
2317 * tree.c (overwrite_decl_assembler_name): New.
2318 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
2319 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
2320 (LANG_HOOKS_INITIALIZER): Add it.
2321 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
2322 * langhooks.c (lhd_set_decl_assembler_name): Use
2323 SET_DECL_ASSEMBLER_NAME.
2324 (lhd_overwrite_decl_assembler_name): Default implementation.
2326 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2327 Jackson Woodruff <jackson.woodruff@arm.com>
2329 PR tree-optimization/71026
2330 * match.pd: Canonicalize constant multiplies in division.
2332 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2334 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
2335 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
2336 Remove OPT_fomit_frame_pointer entry.
2337 * common/config/arc/arc-common.c: Likewise.
2338 * common/config/arm/arm-common.c: Likewise.
2339 * common/config/avr/avr-common.c: Likewise.
2340 * common/config/c6x/c6x-common.c: Likewise.
2341 * common/config/cr16/cr16-common.c: Likewise.
2342 * common/config/cris/cris-common.c: Likewise.
2343 * common/config/epiphany/epiphany-common.c: Likewise.
2344 * common/config/fr30/fr30-common.c: Likewise.
2345 * common/config/frv/frv-common.c: Likewise.
2346 * common/config/ia64/ia64-common.c: Likewise.
2347 * common/config/iq2000/iq2000-common.c: Likewise.
2348 * common/config/lm32/lm32-common.c: Likewise.
2349 * common/config/m32r/m32r-common.c: Likewise.
2350 * common/config/mcore/mcore-common.c: Likewise.
2351 * common/config/microblaze/microblaze-common.c: Likewise.
2352 * common/config/mips/mips-common.c: Likewise.
2353 * common/config/mmix/mmix-common.c: Likewise.
2354 * common/config/mn10300/mn10300-common.c: Likewise.
2355 * common/config/nios2/nios2-common.c: Likewise.
2356 * common/config/pa/pa-common.c: Likewise.
2357 * common/config/pdp11/pdp11-common.c: Likewise.
2358 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
2359 * common/config/riscv/riscv-common.c: Likewise.
2360 * common/config/rs6000/rs6000-common.c: Likewise.
2361 * common/config/rx/rx-common.c: Likewise.
2362 * common/config/s390/s390-common.c: Likewise.
2363 * common/config/sh/sh-common.c: Likewise.
2364 * common/config/sparc/sparc-common.c: Likewise.
2365 * common/config/tilegx/tilegx-common.c: Likewise.
2366 * common/config/tilepro/tilepro-common.c: Likewise.
2367 * common/config/v850/v850-common.c: Likewise.
2368 * common/config/visium/visium-common.c: Likewise.
2369 * common/config/xstormy16/xstormy16-common.c: Likewise.
2370 * common/config/xtensa/xtensa-common.c: Likewise.
2371 * invoke.texi (-fomit-frame-pointer): Update documentation.
2373 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2375 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
2377 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2379 * cfg.c (scale_bbs_frequencies_int,
2380 cale_bbs_frequencies_gcov_type): Remove.
2381 * cfg.h (scale_bbs_frequencies_int,
2382 cale_bbs_frequencies_gcov_type): Remove.
2384 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2386 * tree-ssa-loop-manip.c
2387 (scale_dominated_blocks_in_loop): Update to profile counts.
2388 (tree_transform_and_unroll_loop): Likewise.
2390 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2392 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
2393 scale_bbs_frequencies_int.
2395 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2397 * final.c (compute_alignments): Use counts rather than frequencies.
2399 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2401 * cfgloopanal.c: Include sreal.h
2402 (average_num_loop_insns): Use counts and sreal for accounting.
2404 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2406 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
2409 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2410 Kito Cheng <kito.cheng@gmail.com>
2412 * config/nds32/constraints.md: Provide more constraints.
2413 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
2414 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
2415 support constraints usage.
2417 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2419 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
2421 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2422 Kito Cheng <kito.cheng@gmail.com>
2424 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
2425 * config/nds32/nds32.opt: Refine the layout.
2426 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
2427 TARGET_EXT_STRING): Support new options.
2428 * config/nds32/nds32.h: Likewise.
2429 * config/nds32/nds32.md: Likewise.
2430 * config/nds32/nds32-predicates.c: Likewise.
2431 * config/nds32/constraints.md: Likewise.
2432 * common/config/nds32/nds32-common.c: Likewise.
2434 2017-11-16 Julia Koval <julia.koval@intel.com>
2437 * config/i386/gfniintrin.h: Add sse check.
2438 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
2440 2017-11-16 Julia Koval <julia.koval@intel.com>
2442 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
2443 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
2444 (ix86_handle_option): Handle -mavx512vbmi2.
2445 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
2446 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
2447 * config/i386/i386-c.c (__AVX512VBMI2__): New.
2448 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
2449 (ix86_valid_target_attribute_inner_p): Ditto.
2450 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
2451 * config/i386/i386.opt (mavx512vbmi2): New option.
2452 * doc/invoke.texi: Add new option.
2454 2017-11-16 Julia Koval <julia.koval@intel.com>
2456 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
2457 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
2458 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
2459 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
2460 _mm512_gf2p8mul_epi8): New intrinsics.
2461 * config/i386/i386-builtin-types.def
2462 (V64QI_FTYPE_V64QI_V64QI): New type.
2463 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
2464 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
2465 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
2466 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
2467 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
2468 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2470 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
2472 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
2473 explicitly as a stream of bytes.
2475 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2477 * config/rs6000/altivec.h (vec_xst_be): New #define.
2478 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
2479 and externalize from *altivec_vperm_<mode>_internal.
2480 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
2482 (XL_BE_V8HI): Likewise.
2483 (XL_BE_V4SI): Likewise.
2484 (XL_BE_V4SI): Likewise.
2485 (XL_BE_V2DI): Likewise.
2486 (XL_BE_V4SF): Likewise.
2487 (XL_BE_V2DF): Likewise.
2488 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
2489 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
2490 all array entries with these keys: VSX_BUILTIN_VEC_XL,
2491 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
2492 VSX_BUILTIN_VEC_XST_BE.
2493 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
2494 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
2496 (altivec_init_builtins): Replace conditional calls to def_builtin
2497 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
2498 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
2499 with unconditional calls. Remove calls to def_builtin for
2500 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
2501 __builtin_vec_xst_be.
2502 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
2503 to define_expand, and add alternate RTL generation for P8.
2504 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
2505 vsx_ld_elemrev_v8hi.
2506 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
2507 add alternate RTL generation for P8.
2508 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
2509 vsx_ld_elemrev_v16qi.
2510 (vsx_st_elemrev_v8hi): Convert define_insn
2511 to define_expand, and add alternate RTL generation for P8.
2512 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
2513 vsx_st_elemrev_v8hi.
2514 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
2515 add alternate RTL generation for P8.
2516 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
2517 vsx_st_elemrev_v16qi.
2519 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
2522 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
2523 TARGET_AVX512ER check.
2524 (ix86_option_override_internal): Set MASK_VZEROUPPER if
2525 neither -mzeroupper nor -mno-zeroupper is used and
2526 TARGET_EMIT_VZEROUPPER is set.
2527 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
2528 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
2530 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
2532 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2533 folding of vector compares.
2534 (fold_build_vec_cmp): New helper function.
2535 (fold_compare_helper): New helper function.
2536 (builtin_function_type): Add compare builtins to the list of functions
2537 having unsigned arguments. Cosmetic updates to comment indentation.
2538 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
2539 the not+eq combination.
2541 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2543 PR tree-optimization/82726
2544 PR tree-optimization/70754
2545 * tree-predcom.c (order_drefs_by_pos): New function.
2546 (combine_chains): Move code setting has_max_use_after to...
2547 (try_combine_chains): ...here. New parameter. Sort combined chains
2548 according to position information.
2549 (tree_predictive_commoning_loop): Update call to above function.
2550 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
2552 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2554 PR tree-optimization/82726
2556 2017-01-23 Bin Cheng <bin.cheng@arm.com>
2558 PR tree-optimization/70754
2559 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
2560 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
2561 combined stmt before it if not NULL.
2562 (combine_chains): Process refs reversely and compute dominance point
2566 2017-02-23 Bin Cheng <bin.cheng@arm.com>
2568 PR tree-optimization/79663
2569 * tree-predcom.c (combine_chains): Process refs in reverse order
2570 only for ZERO length chains, and add explaining comment.
2572 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2574 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
2575 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
2576 * doc/invoke.texi (ARM Options): Add armv8.3-a.
2578 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2580 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
2582 2017-11-15 Martin Liska <mliska@suse.cz>
2584 * tree-cfg.c (pass_warn_function_return::execute):
2585 Compare warn_return_type for greater than zero.
2587 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
2591 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
2592 to return true on Xeon and not on Xeon Phi.
2593 (ix86_check_avx256_register): Changed to ...
2594 (ix86_check_avx_upper_register): ... this. Add extra check for
2595 VALID_AVX512F_REG_OR_XI_MODE.
2596 (ix86_avx_u128_mode_needed): Changed
2597 ix86_check_avx256_register to ix86_check_avx_upper_register.
2598 (ix86_check_avx256_stores): Changed to ...
2599 (ix86_check_avx_upper_stores): ... this. Changed
2600 ix86_check_avx256_register to ix86_check_avx_upper_register.
2601 (ix86_avx_u128_mode_after): Changed
2602 avx_reg256_found to avx_upper_reg_found. Changed
2603 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
2604 (ix86_avx_u128_mode_entry): Changed
2605 ix86_check_avx256_register to ix86_check_avx_upper_register.
2606 (ix86_avx_u128_mode_exit): Ditto.
2607 * config/i386/i386.h: (host_detect_local_cpu): New define.
2609 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2611 * config/arm/xgene1.md (xgene1): Split into automatons
2612 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
2613 (xgene1_f_load): Adjust reservations and/or types.
2614 (xgene1_f_store): Likewise.
2615 (xgene1_load_pair): Likewise.
2616 (xgene1_store_pair): Likewise.
2617 (xgene1_fp_load1): Likewise.
2618 (xgene1_load1): Likewise.
2619 (xgene1_store1): Likewise.
2620 (xgene1_move): Likewise.
2621 (xgene1_alu): Likewise.
2622 (xgene1_simd): Likewise.
2623 (xgene1_bfm): Likewise.
2624 (xgene1_neon_load1): Likewise.
2625 (xgene1_neon_store1): Likewise.
2626 (xgene1_neon_logic): Likewise.
2627 (xgene1_neon_st1): Likewise.
2628 (xgene1_neon_ld1r): Likewise.
2629 (xgene1_alu_cond): Added.
2630 (xgene1_shift_reg): Likwise.
2631 (xgene1_bfx): Likewise.
2632 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
2634 2017-11-15 Jakub Jelinek <jakub@redhat.com>
2637 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
2639 (can_widen_mult_without_libcall): New function.
2640 (expand_mul_overflow): If only checking unsigned mul overflow,
2641 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
2642 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
2643 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
2644 (expand_DIVMOD): Formatting fix.
2645 * expmed.h (expand_mult): Add NO_LIBCALL argument.
2646 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
2647 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
2649 PR tree-optimization/82977
2650 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
2651 constructed temporary to strlen_to_stridx.put.
2653 2017-11-15 Martin Liska <mliska@suse.cz>
2655 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
2656 * configure: Regenerate.
2658 2017-11-15 Martin Liska <mliska@suse.cz>
2661 * config/sh/sh-mem.cc: Use proper probability for
2664 2017-11-14 Jeff Law <law@redhat.com>
2666 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
2667 the red zone for stack_clash_protection_final_dynamic_probe targets
2668 when the total dynamic stack size is zero bytes.
2670 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
2671 blocks is post order.
2673 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
2675 * dumpfile.h (TDF_COMPARE_DEBUG): New.
2676 * final.c (rest_of_clean_state): Set it for the
2677 -fcompare-debug dump.
2678 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
2679 class when TDF_COMPARE_DEBUG is set.
2681 * dwarf2out.c (gen_producer_string): Discard
2684 2017-11-15 Joseph Myers <joseph@codesourcery.com>
2687 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
2688 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
2689 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
2690 (__TGMATH_REAL_2_3): Remove macros.
2691 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
2692 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
2694 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
2696 (remquo): Define using __TGMATH_REAL_3.
2698 2017-11-14 Jeff Law <law@redhat.com>
2700 * vr-values.c: New file with contents extracted from tree-vrp.c.
2701 * Makefile.in (OBJS): Add vr-values.o
2702 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
2703 (set_value_range, set_and_canonicalize_value_range): Likewise.
2704 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
2705 (value_range_constant_singleton, symbolic_range_p): Likewise.
2706 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
2707 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
2708 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
2709 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
2710 (find_case_label_range, find_case_label_index): Likewise.
2711 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
2712 (range_int_cst_singleton_p, value_inside_range): Likewise.
2713 (get_single_symbol): Likewise.
2714 (switch_update): Move structure definition here.
2715 (to_remove_edges, to_update_switch_stmts): Provide externs.
2716 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
2717 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
2718 (vrp_val_is_min, set_value_range): Likewise.
2719 (set_and_canonicalize_value_range, copy_value_range): Likewise.
2720 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
2721 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
2722 (range_is_nonnull, range_int_cst_p): Likewwise.
2723 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
2724 (get_single_symbol, operand_less_p): Likewise
2725 (compare_values_warnv, compare_values): Likewise.
2726 (value_inside_range, value_range_constant_singleton): Likewise.
2727 (zero_nonzero_bitgs_from_vr): Likewise.
2728 (extract_range_from_binary_expr_1): Likewise.
2729 (overflow_comparison_p): Likewise.
2730 (to_remove_edges, to_update_switch_stmts): Likewise.
2731 (find_case_label-index, find_case_label_range): Likewise.
2732 (switch_update, set_value_range_to_nonnegative): Remove.
2733 (set_value_range_to_truthvalue): Likewise.
2734 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
2735 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
2736 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
2737 (find_case_label_ranges, test_for_singularity): Likewise.
2738 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
2739 (x_vr_values): Move to its remaining use site.
2741 2017-11-10 Jeff Law <law@redhat.com>
2743 * vr-values.h (VR_INITIALIZER): Move #define here.
2744 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
2745 * Makefile.in (OBJS): Add tree-evrp.o
2746 * tree-vrp.h (assert_info): Move structure definition here.
2747 (set_value_range_to_varying): Prototype.
2748 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2749 (infer_value_range, register_edge_assert_for): Likewise.
2750 (stmt_interesting_for_vrp): Likewise.
2751 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
2752 (set_value_range_to_varying): No longer static.
2753 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2754 (infer_value_range, register_edge_assert_for): Likewise.
2756 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2758 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
2759 generate the XXBRD instruction.
2761 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
2762 (rs6000_builtin_type_compatible): Treat _Float128 and long double
2763 as being compatible if -mabi=ieeelongdouble.
2764 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
2765 to setup float128 built-ins with hardware support.
2766 (BU_FLOAT128_HW_2): Likewise.
2767 (BU_FLOAT128_HW_3): Likewise.
2768 (BU_FLOAT128_HW_VSX_1): Likewise.
2769 (BU_FLOAT128_HW_VSX_2): Likewise.
2770 (scalar_extract_expq): Change float128 built-in functions to
2771 accommodate having both KFmode and TFmode functions. Use the
2772 KFmode variant as the default.
2773 (scalar_extract_sigq): Likewise.
2774 (scalar_test_neg_qp): Likewise.
2775 (scalar_insert_exp_q): Likewise.
2776 (scalar_insert_exp_qp): Likewise.
2777 (scalar_test_data_class_qp): Likewise.
2778 (sqrtf128_round_to_odd): Delete processing the round to odd
2779 built-in functions as special built-in functions, and define them
2780 as float128 built-ins. Use the KFmode variant as the default.
2781 (truncf128_round_to_odd): Likewise.
2782 (addf128_round_to_odd): Likewise.
2783 (subf128_round_to_odd): Likewise.
2784 (mulf128_round_to_odd): Likewise.
2785 (divf128_round_to_odd): Likewise.
2786 (fmaf128_round_to_odd): Likewise.
2787 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
2788 support for KFmode and TFmode xststdcqp calls.
2789 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
2790 point, switch the built-in handlers for the get/set float128
2791 exponent, get float128 mantissa, float128 test built-ins, and the
2792 float128 round to odd built-in functions. Eliminate creating the
2793 float128 round to odd built-in functions as special built-ins.
2794 (rs6000_init_builtins): Eliminate special creation of the float128
2795 round to odd built-in functions.
2796 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
2797 function insns to support both TFmode and KFmode variants.
2798 (xsxsigqp_<mode>): Likewise.
2799 (xsiexpqpf_<mode>): Likewise.
2800 (xsiexpqp_<mode>): Likewise.
2801 (xststdcqp_<mode>): Likewise.
2802 (xststdcnegqp_<mode>): Likewise.
2803 (xststdcqp_<mode>): Likewise.
2805 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2807 * tree-ssa-threadupdate.c (compute_path_counts): Remove
2808 unused path_in_freq_ptr parameter.
2809 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
2811 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2813 * ipa-inline.c (edge_badness): Dump sreal frequency.
2814 (compute_inlined_call_time): Match natural implementaiton ...
2815 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
2816 forgotten division by CGRAPH_FREQ_BASE.
2818 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2820 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
2821 Solaris 11. Update comment.
2822 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
2824 * config/sol2.h (STARTFILE_SPEC): Likewise.
2825 * configure: Regenerate.
2827 2017-11-14 Carl Love <cel@us.ibm.com>
2829 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
2830 le_ and be_ prefixes to swap* variables. Remove
2831 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
2834 2017-11-14 Jason Merrill <jason@redhat.com>
2836 Support GTY((cache)) on hash_map.
2837 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
2838 (ggc_cache_remove): Override it instead of ggc_mx.
2839 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
2840 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
2841 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
2842 (simple_cache_map_traits): Override maybe_mx.
2843 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
2844 (hash_map): Friend gt_cleare_cache.
2845 (gt_cleare_cache): New.
2846 * tree.h (tree_cache_traits): New hash_map traits class.
2847 (tree_cache_map): New typedef.
2849 2017-11-14 Richard Biener <rguenther@suse.de>
2851 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
2852 paramter and handling.
2853 (cleanup_control_flow_bb): Likewise.
2854 (cleanup_control_flow_pre): New helper performing a DFS walk
2855 to call cleanup_control_flow_bb in PRE order.
2856 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
2857 via cleanup_control_flow_pre.
2859 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
2861 * config/aarch64/aarch64-simd.md
2862 (aarch64_simd_bsl<mode>_internal): Remove DImode.
2863 (*aarch64_simd_bsl<mode>_alt): Likewise.
2864 (aarch64_simd_bsldi_internal): New.
2865 (aarch64_simd_bsldi_alt): Likewise.
2867 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2869 * tracer.c (better_p): Do not compare frequencies.
2870 * reg-stack.c (better_edge): Likewise.
2871 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
2874 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2876 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
2877 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
2878 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
2879 * lto-streamer-in.c (input_function): Use update_max_bb_count.
2880 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
2881 * predict.c (maybe_hot_frequency_p): Inline to ...
2882 (maybe_hot_count_p): ... here; rewrite to counts.
2883 (counts_to_freqs): Rename to ...
2884 (update_max_bb_count): ... this one.
2885 (expensive_function_p): Use counts.
2886 (estimate_bb_frequencies): Update.
2887 (rebuild_frequencies): Update.
2888 * predict.h (counts_to_freqs): Rename to ...
2889 (update_max_bb_count): ... this one.
2890 * profile.c (compute_branch_probabilities): Add debug info
2891 * tree-inline.c (expand_call_inline): Update debug info.
2892 (optimize_inline_calls): Use update_max_bb_count..
2893 (tree_function_versioning): Use update_max_bb_count..
2894 * value-prof.c (gimple_value_profile_transformations):
2895 Do not use update_max_bb_count.
2897 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2899 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
2900 always use frequencies.
2902 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2904 * bb-reorder.c: Remove frequencies from comments.
2905 (better_edge_p): Use profile counts.
2906 (find_traces): Dump profile counts.
2907 (rotate_loop): Use profile counts.
2908 (find_traces_1_round): Likewise.
2909 (connect_better_edge_p): Use counts instead of probabilities for
2911 (copy_bb_p): Drop early check for non-0 frequency.
2912 (sanitize_hot_paths): Update comments.
2914 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2916 * ipa-split.c (struct split_point): Add count.
2917 (consider_split): Do not compute incoming frequency; compute incoming
2918 count and store it to split_point.
2919 (split_function): Set count of the call to split part correctly.
2921 2017-11-13 Carl Love <cel@us.ibm.com>
2923 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
2925 2017-11-13 Tom Tromey <tom@tromey.com>
2927 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
2929 2017-11-13 Carl Love <cel@us.ibm.com>
2931 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
2932 Add support for builtins:
2933 unsigned int vec_first_{,miss}_match_{,or_eos}index,
2934 vector {un,}signed {char,int,short},
2935 vector {un,}signed {char,int,short}) arguments.
2936 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
2937 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
2938 Add BU_P9V_AV_2 expansions for the builtins.
2939 * config/rs6000/altivec.h (vec_first_match_index,
2940 vec_first_mismatch_index, vec_first_match_or_eos_index,
2941 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
2942 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
2943 new extern declaration.
2944 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
2945 * config/rs6000/vsx.md (first_match_index_<mode>,
2946 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
2947 first_mismatch_or_eos_index_<mode>): Add define expand.
2948 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
2949 * doc/extend.texi: Update the built-in documenation file for the new
2952 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
2954 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
2955 into the min/max operations for _Float<N> and _Float<N>X types.
2957 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
2960 * dwarf2out.c (do_eh_frame): New static variable.
2961 (dwarf2out_begin_prologue): Set it.
2962 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
2964 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2966 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
2968 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
2970 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
2973 * debug.h (dwarf2out_do_eh_frame): Declare.
2974 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
2975 (dwarf2out_do_frame): Use it.
2976 (dwarf2out_do_cfi_asm): Likewise.
2977 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
2978 (dwarf2out_assembly_start): Likewise.
2979 (dwarf2out_begin_prologue): Fix comment.
2980 * toplev.c (compile_file): Always call dwarf2out_frame_finish
2981 if the target needs either debug or unwind DWARF2 info.
2982 * lto-opts.c (lto_write_options): Do not save -fexceptions,
2983 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
2984 -ftrapping-math, -ftrapv and -fwrapv.
2986 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2988 * cgraph.c (cgraph_edge::sreal_frequency): New function.
2989 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
2990 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
2991 (estimate_edge_size_and_time): Likewise.
2992 (ipa_merge_fn_summary_after_inlining): Likewise.
2993 * ipa-inline.c (cgraph_freq_base_rec): Remove.
2994 (compute_uninlined_call_time): Use sreal_frequency.
2995 (compute_inlined_call_time): Likewise.
2996 (ipa_inline): Do not initialize cgraph_freq_base_rec.
2997 * profile-count.c: Include sreal.h.
2998 (profile_count::to_sreal_scale): New.
2999 * profile-count.h: Forward declare sreal.
3000 (profile_count::to_sreal_scale): Declare.
3002 2017-11-13 Nathan Sidwell <nathan@acm.org>
3004 * diagnostic.c (maybe_line_and_column): New.
3005 (diagnostic_get_location_text): Use it.
3006 (diagnostic_report_current_module): Likewise.
3007 (test_diagnostic_get_location_text): Add tests.
3009 2017-11-13 Luis Machado <luis.machado@linaro.org>
3011 * doc/md.texi (Specifying processor pipeline description): Fix
3012 incorrect latency for the div instruction example.
3014 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3016 PR tree-optimization/78821
3017 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
3018 that bit_not_p is the same.
3019 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
3020 (split_group): Count precisely bit_not_p bits in each statement.
3021 (invert_op): New function.
3022 (imm_store_chain_info::output_merged_store): Use invert_op to
3023 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
3024 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
3026 2017-11-13 Martin Liska <mliska@suse.cz>
3028 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
3029 (struct source_info): Likewise.
3030 (add_branch_counts): Likewise.
3031 (add_line_counts): Likewise.
3032 (function_summary): Likewise.
3033 (output_intermediate_line): Likewise.
3034 (generate_results): Likewise.
3036 2017-11-13 Martin Liska <mliska@suse.cz>
3038 * gcov.c (struct block_info): Remove typedef for block_t.
3039 (struct line_info): Likewise.
3040 (line_info::has_block): Likewise.
3041 (EXIT_BLOCK): Likewise.
3042 (unblock): Likewise.
3043 (circuit): Likewise.
3044 (get_cycles_count): Likewise.
3045 (process_file): Likewise.
3046 (read_graph_file): Likewise.
3047 (solve_flow_graph): Likewise.
3048 (find_exception_blocks): Likewise.
3049 (add_line_counts): Likewise.
3050 (accumulate_line_info): Likewise.
3051 (output_line_details): Likewise.
3053 2017-11-13 Martin Liska <mliska@suse.cz>
3055 * gcov.c (struct arc_info): Remove typedef for arc_t.
3056 (struct line_info): Likewise.
3057 (add_branch_counts): Likewise.
3058 (output_branch_count): Likewise.
3059 (function_info::~function_info): Likewise.
3060 (circuit): Likewise.
3061 (output_intermediate_line): Likewise.
3062 (read_graph_file): Likewise.
3063 (solve_flow_graph): Likewise.
3064 (find_exception_blocks): Likewise.
3065 (add_line_counts): Likewise.
3066 (accumulate_line_info): Likewise.
3067 (output_line_details): Likewise.
3068 (output_function_details): Likewise.
3070 2017-11-13 Martin Liska <mliska@suse.cz>
3072 * gcov.c (struct function_info): Remove typedef for function_t.
3073 (struct source_info): Likewise.
3074 (source_info::get_functions_at_location): Likewise.
3075 (solve_flow_graph): Likewise.
3076 (find_exception_blocks): Likewise.
3077 (add_line_counts): Likewise.
3078 (output_intermediate_file): Likewise.
3079 (process_file): Likewise.
3080 (generate_results): Likewise.
3081 (release_structures): Likewise.
3082 (read_graph_file): Likewise.
3083 (read_count_file): Likewise.
3084 (accumulate_line_counts): Likewise.
3085 (output_lines): Likewise.
3087 2017-11-13 Martin Liska <mliska@suse.cz>
3089 * gcov.c (function_info::function_info): Remove num_counts
3090 and add vector<gcov_type>.
3091 (function_info::~function_info): Use the vector.
3092 (process_file): Likewise.
3093 (read_graph_file): Likewise.
3094 (read_count_file): Likewise.
3095 (solve_flow_graph): Likewise.
3097 2017-11-13 Martin Liska <mliska@suse.cz>
3099 * gcov.c (function_info::is_artificial): New function.
3100 (process_file): Erase all artificial early.
3101 (generate_results): Skip as all artificial are already
3104 2017-11-13 Martin Liska <mliska@suse.cz>
3106 * gcov.c (read_graph_file): Store to global vector of functions.
3107 (read_count_file): Iterate the vector.
3108 (process_file): Likewise.
3109 (generate_results): Likewise.
3110 (release_structures): Likewise.
3112 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3114 PR tree-optimization/82954
3115 * gimple-ssa-store-merging.c
3116 (imm_store_chain_info::coalesce_immediate_stores): If
3117 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
3119 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
3121 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
3122 Upddate call to ENDIAN_LANE_N.
3123 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
3124 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
3125 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
3126 and use aarch64_endian_lane_rtx.
3127 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
3129 2017-11-12 Tom de Vries <tom@codesourcery.com>
3131 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
3133 2017-11-12 Tom de Vries <tom@codesourcery.com>
3135 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
3137 2017-11-12 Tom de Vries <tom@codesourcery.com>
3139 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
3141 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3142 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3143 * defaults.h (ASM_OUTPUT_LABELREF): Same.
3145 2017-11-11 Martin Sebor <msebor@redhat.com>
3148 * doc/extend.texi (attribute nonstring): Remove spurious argument.
3151 * prefic.c (translate_name): Replace strncpy with memcpy to
3152 avoid -Wstringop-truncation.
3154 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3156 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
3158 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3160 * predict.c (maybe_hot_frequency_p): Do not use cfun.
3162 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3164 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
3167 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
3169 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
3170 enable generating XXBRH if the value is in a vector register.
3171 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
3172 value is in a vector register.
3173 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
3174 register to register bswap64's instead of doing the GPR sequence
3175 used on previous machines.
3176 (bswapdi2_xxbrd): New insn.
3177 (bswapdi2_reg): Disallow on ISA 3.0.
3178 (register to register bswap64 splitter): Do not split the insn on
3179 ISA 3.0 systems that use XXBRD.
3181 2017-11-10 Martin Sebor <msebor@redhat.com>
3184 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
3186 (find_subframework_file): Same.
3188 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3190 * auto-profile.c (afdo_indirect_call): Drop frequency.
3191 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
3192 (cgraph_node::create_edge): Drop frequency argument.
3193 (cgraph_node::create_indirect_edge): Drop frequency argument.
3194 (cgraph_edge::make_speculative): Drop frequency arguments.
3195 (cgraph_edge::resolve_speculation): Do not update frequencies
3196 (cgraph_edge::dump_edge_flags): Do not dump frequency.
3197 (cgraph_node::dump): Check consistency in IPA mode.
3198 (cgraph_edge::maybe_hot_p): Use IPA counter.
3199 (cgraph_edge::verify_count_and_frequency): Rename to ...
3200 (cgraph_edge::verify_count): ... this one; drop frequency checking.
3201 (cgraph_node::verify_node): Update.
3202 * cgraph.h (struct cgraph_edge): Drop frequency.
3203 (cgraph_edge::frequency): New function.
3204 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
3206 (cgraph_edge::rebuild_edges): Likewise.
3207 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
3208 (duplicate_thunk_for_node): Do not pass frequency.
3209 (cgraph_node::create_clone): Scale only counts.
3210 (cgraph_node::create_virtual_clone): Do not pass frequency.
3211 (cgraph_node::create_edge_including_clones): Do not pass frequency.
3212 (cgraph_node::create_version_clone): Do not pass frequency.
3213 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
3214 (cgraph_node::expand_thunk): Do not pass frequency.
3215 (cgraph_node::create_wrapper): Do not pass frequency.
3216 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
3218 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
3219 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
3220 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
3221 (gather_caller_stats): Use frequency function.
3222 (ipcp_cloning_candidate_p): Use frequency function.
3223 (ipcp_propagate_stage): Use frequency function.
3224 (get_info_about_necessary_edges): Use frequency function.
3225 (update_profiling_info): Update only IPA profile.
3226 (update_specialized_profile): Use frequency functoin.
3227 (perhaps_add_new_callers): Update only IPA profile.
3228 * ipa-devirt.c (ipa_devirt): Use IPA profile.
3229 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
3230 (dump_ipa_call_summary): Use frequency function.
3231 (estimate_edge_size_and_time): Use frequency function.
3232 (ipa_merge_fn_summary_after_inlining): Use frequency function.
3233 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
3234 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
3235 (update_noncloned_counts): ... ths one; scale counts only.
3236 (clone_inlined_nodes): Do not scale frequency.
3237 (inline_call): Do not pass frequency.
3238 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
3239 (compute_inlined_call_time): Use IPA profile.
3240 (want_inline_small_function_p): Use IPA profile.
3241 (want_inline_self_recursive_call_p): Use IPA profile.
3242 (edge_badness): Use IPA profile.
3243 (lookup_recursive_calls): Use IPA profile.
3244 (recursive_inlining): Do not pass frequency.
3245 (resolve_noninline_speculation): Do not update frequency.
3246 (inline_small_functions): Collect max of IPA profile.
3247 (dump_overall_stats): Dump IPA porfile.
3248 (dump_inline_stats): Dump IPA porfile.
3249 (ipa_inline): Collect IPA stats.
3250 * ipa-inline.h (clone_inlined_nodes): Update prototype.
3251 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
3252 (ipa_propagate_frequency): Use frequency function.
3253 (ipa_profile): Cleanup.
3254 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
3255 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
3256 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
3257 (input_node): Do not stream frequency.
3258 (input_edge): Do not stream frequency.
3259 (merge_profile_summaries): Scale only IPA profiles.
3260 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
3261 * predict.c (drop_profile): Do not recompute frequency.
3262 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
3263 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
3264 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
3265 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
3266 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
3267 * tree-inline.c (copy_bb): Do not scale frequency.
3268 (expand_call_inline): Do not scale frequency.
3269 (tree_function_versioning): Do not scale frequency.
3270 * ubsan.c (ubsan_create_edge): Do not pass frequency.
3272 2017-11-10 Julia Koval <julia.koval@intel.com>
3274 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
3275 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
3276 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
3277 (_mm256_maskz_gf2p8affine_epi64_epi8)
3278 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
3279 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
3280 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
3281 (__builtin_ia32_vgf2p8affineqb_v32qi)
3282 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
3283 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
3285 2017-11-10 Tamar Christina <tamar.christina@arm.com>
3289 (arm_option_override): Refactor.
3290 (arm_option_reconfigure_globals): New.
3291 (arm_options_perform_arch_sanity_checks): New.
3292 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
3294 (arm_options_perform_arch_sanity_checks): Likewise
3296 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
3298 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
3299 (power9-qpmul): New.
3300 * rs6000/rs6000.md ("type" attr): Add qmul.
3301 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
3302 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
3303 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
3305 2017-11-10 Martin Sebor <msebor@redhat.com>
3308 * builtins.c (compute_objsize): Handle arrays that
3309 compute_builtin_object_size likes to fail for. Make extern.
3310 * builtins.h (compute_objsize): Declare.
3311 (check_strncpy_sizes): New function.
3312 (expand_builtin_strncpy): Call check_strncpy_sizes.
3313 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
3314 -Wstringop-truncation.
3315 (gimple_fold_builtin_strncat): Same.
3316 * gimple.c (gimple_build_call_from_tree): Set call location.
3317 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
3318 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
3319 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
3320 (handle_builtin_strlen): Use strlen_to_stridx.
3321 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
3323 Use strlen_to_stridx.
3324 (pass_strlen::execute): Release strlen_to_stridx.
3325 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
3326 (-Wstringop-truncation): Document new option.
3328 2017-11-10 Martin Liska <mliska@suse.cz>
3330 PR gcov-profile/82702
3331 * gcov.c (main): Handle intermediate files in a different
3333 (get_gcov_intermediate_filename): New function.
3334 (output_gcov_file): Remove support of intermediate files.
3335 (generate_results): Allocate intermediate file.
3336 (release_structures): Clean-up properly fn_end.
3337 (output_intermediate_file): Start iterating with line 1.
3339 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3341 PR tree-optimization/82929
3342 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3343 ops_swapped_p non-static data member.
3344 (store_immediate_info::store_immediate_info): Clear it.
3345 (imm_store_chain_info::coalesce_immediate_stores): If swapping
3346 ops set ops_swapped_p.
3347 (count_multiple_uses): Handle ops_swapped_p.
3349 2017-11-10 Martin Liska <mliska@suse.cz>
3351 * coverage.c (coverage_init): Stream information about
3352 support of has_unexecuted_blocks.
3353 * doc/gcov.texi: Document that.
3354 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
3355 * gcov.c (read_graph_file): Likewise.
3356 (output_line_beginning): Fix a small issue with
3359 2017-11-10 Bin Cheng <bin.cheng@arm.com>
3361 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
3362 reference of trivial component.
3364 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3367 * gimple-ssa-store-merging.c
3368 (pass_store_merging::terminate_all_aliasing_chains): For
3369 gimple_store_p stmts also call refs_output_dependent_p.
3371 PR rtl-optimization/82913
3372 * compare-elim.c (try_merge_compare): Punt if def_insn is not
3375 2017-11-09 Jeff Law <law@redhat.com>
3377 * vr-values.h: New file with vr_values class.
3378 * tree-vrp.c: Include vr-values.h
3379 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
3380 data objects into the vr_values class.
3381 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
3382 (get_value_range): Make it a member function within vr_values class.
3383 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
3384 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
3385 (op_with_constant_singleton_value_range): Likewise.
3386 (extract_range_for_var_from_comparison_expr): Likewise.
3387 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
3388 (extract_range_from_binary_expr): Likewise.
3389 (extract_range_from_unary_expr): Likewise.
3390 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
3391 (check_for_binary_op_overflow, extract_range_basic): Likewise.
3392 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
3393 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
3394 (compare_name_with_value, compare_names): Likewise.
3395 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
3396 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
3397 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
3398 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
3399 (extract_range_from_phi_node): Likewise.
3400 (simplify_truth_ops_using_ranges): Likewise.
3401 (simplify_div_or_mod_using_ranges): Likewise.
3402 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
3403 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
3404 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
3405 (simplify_float_conversion_using_ranges): Likewise.
3406 (simplify_internal_call_using_ranges): Likewise.
3407 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
3408 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
3409 poitner via x_vr_values for calls into gimple folder.
3410 (vrp_initialize_lattice): Make this the vr_values ctor.
3411 (vrp_free_lattice): Make this the vr_values dtor.
3412 (set_vr_value): New function.
3413 (class vrp_prop): Add vr_values data member. Add various member
3414 functions as well as member functions that delegate to vr_values.
3415 (check_array_ref): Make a member function within vrp_prop class.
3416 (search_for_addr_array, vrp_initialize): Likewise.
3417 (vrp_finalize): Likewise. Revamp to avoid direct access to
3418 vr_value, values_propagated, etc.
3419 (check_array_bounds): Extract vrp_prop class instance pointer from
3420 walk info structure. Use it to call member functions.
3421 (check_all_array_refs): Make a member function within vrp_prop class.
3422 Smuggle class instance pointer via walk info structure.
3423 (x_vr_values): New local static.
3424 (vrp_valueize): Use x_vr_values to get class instance.
3425 (vr_valueize_1): Likewise.
3426 (class vrp_folder): Add vr_values data member. Add various member
3427 functions as well as member functions that delegate to vr_values.
3428 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
3429 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
3430 class instance from vr_values. Use it to call member functions.
3431 (vrp_dom_walker): Add vr_values data member.
3432 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
3433 instance via x_vr_values.
3434 (identify_jump_threads): Accept vr_values as argument. Store
3435 it into the walker structure.
3436 (evrp_dom_walker): Add vr_values class data member. Add various
3438 (evrp_dom_walker::try_find_new_range): Use vr_values data
3439 member to access the memory allocator.
3440 (evrp_dom_walker::before_dom_children): Store vr_values class
3441 instance into the vrp_folder class.
3442 (evrp_dom_walker::push_value_range): Rework to avoid direct
3443 access to num_vr_values and vr_value.
3444 (evrp_dom_walker::pop_value_range): Likewise.
3445 (execute_early_vrp): Remove call to vrp_initialize_lattice.
3446 Use vr_values to get to dump_all_value_ranges member function.
3447 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
3448 and simplify_cond_using_ranges_2 via vrp_prop class instance.
3449 Pass vr_values class instance down to identify_jump_threads.
3450 Remove call to vrp_free_lattice.
3451 (debug_all_value_ranges): Remove.
3453 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
3454 (vrp_folder): Likewise.
3456 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
3457 Get it from the existing bitmap instead.
3458 (vrp_intersect_ranges_1): Likewise.
3460 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3462 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3464 (store_immediate_info::store_immediate_info): Add bitnotp argument,
3465 set bit_not_p to it.
3466 (imm_store_chain_info::coalesce_immediate_stores): Break group
3467 if bit_not_p is different.
3468 (count_multiple_uses, split_group,
3469 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
3470 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
3471 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
3472 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
3473 to store_immediate_info ctor.
3475 2017-11-09 Jim Wilson <jimw@sifive.com>
3477 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
3478 (scan_prog_file): Likewise.
3480 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3482 * bb-reorder.c (max_entry_frequency): Remove.
3483 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
3484 connect_traces, push_to_next_round_p): Remove prototypes.
3485 (find_traces_1_round): Use counts only.
3486 (push_to_next_round_p): Likewise.
3487 (find_traces): Likewise.
3488 (rotate_loop): Likewise.
3489 (find_traces_1_round): Likewise.
3490 (connect_traces): Likewise.
3491 (edge_order): Likewise.
3493 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
3495 * config/arm/arm.c (output_return_instruction): Add comments to
3496 indicate requirement for cmse_nonsecure_entry return to account
3497 for the size of clearing instruction output here.
3498 (thumb_exit): Likewise.
3499 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
3500 return in hardfloat mode.
3502 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
3504 * config/rs6000/rs6000.c (machine_function): Add a bool,
3505 "toc_is_wrapped_separately".
3506 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
3507 if it wasn't explicitly set or unset, we are optimizing for speed, and
3508 doing separate shrink-wrapping.
3509 (rs6000_get_separate_components): Enable the TOC component if
3510 saving the TOC register in the prologue.
3511 (rs6000_components_for_bb): Handle the TOC component.
3512 (rs6000_emit_prologue_components): Store the TOC register where needed.
3513 (rs6000_set_handled_components): Mark TOC as handled, if handled.
3514 (rs6000_emit_prologue): Don't save the TOC if that is already done.
3516 2017-11-09 Martin Jambor <mjambor@suse.cz>
3518 * ipa-param-manipulation.c: New file.
3519 * ipa-param-manipulation.h: Likewise.
3520 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
3521 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
3522 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
3523 (ipa_parm_adjustment): Likewise.
3524 (ipa_parm_adjustment_vec): Likewise.
3525 (ipa_get_vector_of_formal_parms): Moved declaration to
3526 ipa-param-manipulation.h.
3527 (ipa_get_vector_of_formal_parm_types): Likewise.
3528 (ipa_modify_formal_parameters): Likewise.
3529 (ipa_modify_call_arguments): Likewise.
3530 (ipa_combine_adjustments): Likewise.
3531 (ipa_dump_param_adjustments): Likewise.
3532 (ipa_modify_expr): Likewise.
3533 (ipa_get_adjustment_candidate): Likewise.
3534 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
3535 ipa-param-manipulation.c.
3536 (ipa_get_vector_of_formal_parm_types): Likewise.
3537 (ipa_modify_formal_parameters): Likewise.
3538 (ipa_modify_call_arguments): Likewise.
3539 (ipa_modify_expr): Likewise.
3540 (get_ssa_base_param): Likewise.
3541 (ipa_get_adjustment_candidate): Likewise.
3542 (index_in_adjustments_multiple_times_p): Likewise.
3543 (ipa_combine_adjustments): Likewise.
3544 (ipa_dump_param_adjustments): Likewise.
3545 * tree-sra.c: Also include ipa-param-manipulation.h
3546 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
3549 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3550 Alan Hayward <alan.hayward@arm.com>
3551 David Sherwood <david.sherwood@arm.com>
3553 * doc/sourcebuild.texi (vect_masked_store): Document.
3555 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3556 Alan Hayward <alan.hayward@arm.com>
3557 David Sherwood <david.sherwood@arm.com>
3559 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
3561 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3562 Alan Hayward <alan.hayward@arm.com>
3563 David Sherwood <david.sherwood@arm.com>
3565 * doc/sourcebuild.texi (vect_variable_length): Document.
3567 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3568 Alan Hayward <alan.hayward@arm.com>
3569 David Sherwood <david.sherwood@arm.com>
3571 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
3573 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3574 Alan Hayward <alan.hayward@arm.com>
3575 David Sherwood <david.sherwood@arm.com>
3577 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
3579 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3580 Alan Hayward <alan.hayward@arm.com>
3581 David Sherwood <david.sherwood@arm.com>
3583 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
3584 previously undocumented selectors.
3585 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
3587 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3589 * doc/rtl.texi (const_vector): Say that elements can be
3590 const_wide_ints too.
3591 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
3592 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
3593 (gen_vec_duplicate): Use it instead of CONSTANT_P.
3594 * optabs.c (expand_vector_broadcast): Likewise.
3596 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3597 Alan Hayward <alan.hayward@arm.com>
3598 David Sherwood <david.sherwood@arm.com>
3600 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
3601 scaled index even if the unscaled address was invalid.
3602 Don't increase the complexity of using a scale in that case.
3604 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3605 Alan Hayward <alan.hayward@arm.com>
3606 David Sherwood <david.sherwood@arm.com>
3608 * doc/rtl.texi: Rewrite the subreg rules so that they partition
3609 the inner register into REGMODE_NATURAL_SIZE bytes rather than
3610 UNITS_PER_WORD bytes.
3611 * emit-rtl.c (validate_subreg): Divide subregs into blocks
3612 based on REGMODE_NATURAL_SIZE of the inner mode.
3613 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
3614 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
3615 * expmed.c (lowpart_bit_field_p): Divide the value up into
3616 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
3617 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
3618 whether something is likely to occupy more than one register.
3620 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3623 * ipa-inline-transform.c (update_noncloned_frequencies): Use
3624 profile_count::adjust_for_ipa_scaling.
3625 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
3626 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
3628 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
3630 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3632 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
3633 (split_group): Add total_orig and total_new arguments, estimate the
3634 number of statements related to the store group without store merging
3635 and with store merging.
3636 (imm_store_chain_info::output_merged_store): Adjust split_group
3637 callers, punt if estimated number of statements with store merging
3638 is not smaller than estimated number of statements without it.
3640 (handled_load): Remove has_single_use checks.
3641 (pass_store_merging::process_store): Likewise.
3643 2017-11-09 Richard Biener <rguenther@suse.de>
3645 PR tree-optimization/82902
3646 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
3648 2017-11-09 Martin Liska <mliska@suse.cz>
3651 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
3654 2017-11-09 Martin Liska <mliska@suse.cz>
3656 PR tree-optimization/82669
3657 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
3659 2017-11-09 Martin Liska <mliska@suse.cz>
3661 PR gcov-profile/48463
3662 * coverage.c (coverage_begin_function): Output also end locus
3663 of a function and information whether the function is
3665 * gcov-dump.c (tag_function): Parse and print the information.
3666 * gcov.c (INCLUDE_MAP): Add include.
3667 (INCLUDE_SET): Likewise.
3668 (struct line_info): Move earlier in the source file because
3669 of vector<line_info> in function_info structure.
3670 (line_info::line_info): Likewise.
3671 (line_info::has_block): Likewise.
3672 (struct source_info): Add new member index.
3673 (source_info::get_functions_at_location): New function.
3674 (function_info::group_line_p): New function.
3675 (output_intermediate_line): New function.
3676 (output_intermediate_file): Use the mentioned function.
3677 (struct function_start): New.
3678 (struct function_start_pair_hash): Likewise.
3679 (process_file): Add code that identifies group functions.
3680 Assign lines either to global or function scope.
3681 (generate_results): Skip artificial functions.
3682 (find_source): Assign index for each source file.
3683 (read_graph_file): Read new flag artificial and end_line.
3684 (add_line_counts): Assign it either to global of function scope.
3685 (accumulate_line_counts): Isolate core of the function to
3686 accumulate_line_info and call it for both function and global
3688 (accumulate_line_info): New function.
3689 (output_line_beginning): Fix GNU coding style.
3690 (print_source_line): New function.
3691 (output_line_details): Likewise.
3692 (output_function_details): Likewise.
3693 (output_lines): Iterate both source (global) scope and function
3695 (struct function_line_start_cmp): New class.
3696 * doc/gcov.texi: Reflect changes in documentation.
3698 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3701 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
3702 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
3703 and similarly for not instead of neg.
3705 2017-11-08 Andi Kleen <ak@linux.intel.com>
3707 * config/i386/i386.opt: Add -mforce-indirect-call.
3708 * config/i386/predicates.md: Check for flag_force_indirect_call.
3709 * doc/invoke.texi: Document -mforce-indirect-call
3711 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
3713 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
3715 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
3716 config/riscv/riscv.c (predict.h): New include.
3717 (riscv_slow_unaligned_access_p): No longer static.
3718 (riscv_block_move_straight): Add require.
3719 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3721 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3724 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
3725 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
3726 nonimmediate_operand predicate for operand 1 instead of
3729 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3731 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
3733 * config/aarch64/constraints.md (Uml): New constraint.
3734 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
3737 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3739 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
3740 of two vec_duplicates into a vec_concat.
3742 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3744 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3745 Simplify vec_merge of vec_duplicate and vec_concat.
3746 * config/aarch64/constraints.md (Utq): New constraint.
3747 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
3750 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3752 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3753 Simplify vec_merge of vec_duplicate and const_vector.
3754 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
3756 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
3757 mode iterator. Update predicate on operand 1 to
3758 handle non-const_vec constants. Delete constraints.
3759 (*aarch64_combinez_be<mode>): Likewise for operand 2.
3761 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3763 PR tree-optimization/78821
3764 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
3766 (store_operand_info::store_operand_info): Initialize it to false.
3767 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
3768 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
3769 store in the group, and if chain_info is non-NULL, to ignore altogether
3771 (compatible_load_p): Fail if bit_not_p does not match.
3772 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
3773 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
3774 (pass_store_merging::process_store): Adjust
3775 terminate_all_aliasing_chains calls to pass NULL in all current spots,
3776 call terminate_all_aliasing_chains newly when adding a store into
3777 a chain with non-NULL chain_info.
3779 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3781 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
3783 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3785 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3787 (aarch64_layout_frame): Initialise emit_frame_chain.
3788 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
3789 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
3791 2017-11-08 Martin Liska <mliska@suse.cz>
3793 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
3794 of gimple_call_internal_p.
3796 2017-11-07 Tom de Vries <tom@codesourcery.com>
3798 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
3800 2017-11-07 Tom de Vries <tom@codesourcery.com>
3802 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3805 2017-11-08 Martin Liska <mliska@suse.cz>
3808 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
3810 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3812 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
3814 2017-11-07 Jakub Jelinek <jakub@redhat.com>
3817 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
3818 (*cmp<mode>_ccz_1): New insn with $k alternative.
3821 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
3822 RTX_COMM_COMPARE as commutative as well.
3823 (ix86_binary_operator_ok): Formatting fix.
3824 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
3825 *<code><mode>3<mask_name><round_saeonly_name>,
3826 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
3827 *<s>mul<mode>3_highpart<mask_name>,
3828 *vec_widen_umult_even_v16si<mask_name>,
3829 *vec_widen_umult_even_v8si<mask_name>,
3830 *vec_widen_umult_even_v4si<mask_name>,
3831 *vec_widen_smult_even_v16si<mask_name>,
3832 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
3833 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
3834 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
3835 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
3836 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
3837 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
3838 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
3839 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
3840 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
3841 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
3842 ix86_binary_operator_ok. Formatting fixes.
3843 (*<plusminus_insn><mode>3<mask_name><round_name>,
3844 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
3847 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
3849 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
3850 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
3851 eq and ne if TARGET_ISEL.
3852 (cmp): New code_iterator.
3853 (UNS, UNSU_, UNSIK): New code_attrs.
3854 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
3855 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
3857 ("eq<mode>3"): ... this.
3858 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
3860 ("ne<mode>3"): ... this.
3862 2017-11-07 Julia Koval <julia.koval@intel.com>
3865 * common/config/i386/i386-common.c
3866 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
3867 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
3868 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
3869 * config/i386/i386.opt: Ditto.
3870 * config/i386/i386.c (ix86_target_string): Ditto.
3871 (ix86_option_override_internal): Ditto.
3872 (ix86_init_mpx_builtins): Move MPX to args2.
3873 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
3874 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3875 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3876 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3877 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3878 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3879 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
3881 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
3884 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
3885 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
3886 (zero-extendsidi peephole2): Remove peephole.
3888 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3891 * stor-layout.c: Include attribs.h.
3892 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
3893 explicit lookup of "aligned" attribute.
3895 2017-11-07 Andrew Waterman <andrew@sifive.com>
3897 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
3898 (riscv_expand_block_move): Likewise.
3899 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
3901 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
3902 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
3903 * config/riscv/riscv.c (riscv_block_move_straight): New function.
3904 (riscv_adjust_block_mem): Likewise.
3905 (riscv_block_move_loop): Likewise.
3906 (riscv_expand_block_move): Likewise.
3907 * config/riscv/riscv.md (movmemsi): New pattern.
3909 2017-11-07 Michael Clark <michaeljclark@mac.com>
3911 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
3912 (MUSL_DYNAMIC_LINKER): Likewise.
3914 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
3916 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
3917 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
3920 2017-11-07 Richard Biener <rguenther@suse.de>
3922 * match.pd: Fix build.
3924 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
3925 Jackson Woodruff <jackson.woodruff@arm.com>
3927 PR tree-optimization/71026
3928 * match.pd: Canonicalize negate in division.
3930 2017-11-07 Sudakshina Das <sudi.das@arm.com>
3933 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
3935 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
3937 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
3938 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
3940 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
3942 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
3943 non-scalar integral types.
3944 * match.pd (negate_expr_p): Handle MINUS_EXPR.
3945 (-(A-B), -(~A)): New transformations.
3947 2017-11-07 Tom de Vries <tom@codesourcery.com>
3949 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
3950 semicolon after "do {} while (0)".
3951 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3952 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3953 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3954 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3955 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3957 2017-11-07 Tom de Vries <tom@codesourcery.com>
3959 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
3960 after "do {} while (0)".
3961 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3962 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3963 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3964 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3965 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3967 2017-11-07 Tom de Vries <tom@codesourcery.com>
3970 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
3972 (arm_rtx_costs_internal): Add missing semicolon after
3973 HANDLE_NARROW_SHIFT_ARITH call.
3975 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
3977 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3978 disable isel if it was not set explicitly.
3980 2017-11-06 James Bowman <james.bowman@ftdichip.com>
3982 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
3983 (add_type_attribute) likewise.
3985 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
3987 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
3988 of struct ix86_frame.
3989 (ix86_initial_elimination_offset): Likewise.
3990 (ix86_expand_split_stack_prologue): Likewise.
3992 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
3994 * tree-vrp.h (enum value_range_type): Update stale comment.
3996 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3997 Alan Hayward <alan.hayward@arm.com>
3998 David Sherwood <david.sherwood@arm.com>
4000 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
4001 (aarch64_expand_vec_perm_const): Take the number of units too.
4002 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
4003 (aarch64_expand_vec_perm_const): Likewise.
4004 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
4005 (vec_perm<mode>): Update accordingly.
4007 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4008 Alan Hayward <alan.hayward@arm.com>
4009 David Sherwood <david.sherwood@arm.com>
4011 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
4012 Take the number of units too.
4013 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
4014 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
4015 but check for a vector mode before rather than after the call.
4016 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
4017 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
4018 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
4019 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
4020 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
4021 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
4022 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
4023 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
4024 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
4025 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
4026 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
4027 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
4028 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
4029 (aarch64_sqdmull2_n<mode>): Update accordingly.
4031 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4032 Alan Hayward <alan.hayward@arm.com>
4033 David Sherwood <david.sherwood@arm.com>
4035 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
4036 the number of units too.
4037 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
4038 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
4039 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
4040 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
4041 (vec_store_lanesxi<mode>): Update accordingly.
4043 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4044 Alan Hayward <alan.hayward@arm.com>
4045 David Sherwood <david.sherwood@arm.com>
4047 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
4048 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
4049 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
4050 of units rather than the mode.
4051 * config/aarch64/iterators.md (nunits): New mode attribute.
4052 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4053 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
4054 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
4055 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
4056 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
4057 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
4058 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
4059 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
4060 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
4061 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
4062 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4063 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
4064 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
4065 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
4066 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
4067 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
4068 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
4069 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
4070 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
4071 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
4072 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
4073 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
4074 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
4075 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4076 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
4077 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
4078 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4079 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
4080 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
4081 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
4082 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
4083 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
4084 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
4085 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
4086 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
4087 (aarch64_simd_vec_setv2di): Likewise.
4089 2017-11-06 Carl Love <cel@us.ibm.com>
4091 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
4093 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
4094 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
4096 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
4098 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
4099 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
4100 Add power 8 macro expansions.
4101 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
4102 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
4103 power 8 instructions. (VSX_XXBR): Add iterator.
4105 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
4107 * config/arm/arm.md (predicable_short_it): Change default to "no",
4108 improve documentation, remove uses that are identical to the default.
4109 (enabled_for_depr_it): Rename to enabled_for_short_it.
4110 * gcc/config/arm/arm-fixed.md (predicable_short_it):
4111 Remove default uses.
4112 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
4113 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
4114 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
4115 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
4117 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
4120 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
4121 float128 helper macros, which are no longer used after deleting
4122 the old 'q' built-in functions, and moving the round to odd
4123 built-in functions to being special built-in functions.
4124 (BU_FLOAT128_2): Likewise.
4125 (BU_FLOAT128_1_HW): Likewise.
4126 (BU_FLOAT128_2_HW): Likewise.
4127 (BU_FLOAT128_3_HW): Likewise.
4128 (FABSQ): Delete old 'q' built-in functions.
4129 (COPYSIGNQ): Likewise.
4130 (SQRTF128_ODD): Move round to odd built-in functions to be
4131 special built-in functions, so that we can handle
4132 -mabi=ieeelongdouble.
4133 (TRUNCF128_ODD): Likewise.
4134 (ADDF128_ODD): Likewise.
4135 (SUBF128_ODD): Likewise.
4136 (MULF128_ODD): Likewise.
4137 (DIVF128_ODD): Likewise.
4138 (FMAF128_ODD): Likewise.
4139 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
4140 built-in names to 'f128'.
4141 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
4142 old 'q' built-in functions, as the machine independent code for
4143 'f128' built-in functions handles this.
4144 (rs6000_expand_builtin): Add expansion for float128 round to odd
4145 functions, keying off on -mabi=ieeelongdouble of whether to use
4146 the KFmode or TFmode variant.
4147 (rs6000_init_builtins): Initialize the _Float128 round to odd
4149 * doc/extend.texi (PowerPC Built-in Functions): Document the old
4150 _Float128 'q' built-in functions are now mapped into the new
4151 'f128' built-in functions.
4153 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
4155 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
4156 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
4158 2017-11-06 David Malcolm <dmalcolm@redhat.com>
4161 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
4162 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
4163 * toplev.c: Include "ipa-fnsummary.h".
4164 (toplev::finalize): Call ipa_fnsummary_c_finalize.
4166 2017-11-06 Jakub Jelinek <jakub@redhat.com>
4168 PR tree-optimization/82838
4169 * gimple-ssa-store-merging.c
4170 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
4171 on a separate gimple_seq which is then appended to seq.
4173 2017-11-06 Jeff Law <law@redhat.com>
4176 * config/i386/i386.c (PROBE_INTERVAL): Remove.
4177 (get_probe_interval): New functions.
4178 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
4179 (ix86_adjust_stack_and_probe): Likewise.
4180 (output_adjust_stack_and_probe): Likewise.
4181 (ix86_emit_probe_stack_range): Likewise.
4182 (ix86_expand_prologue): Likewise.
4184 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4186 PR tree-optimization/82816
4187 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
4188 if the modes of the two types are the same.
4189 (convert_plusminus_to_widen): Likewise.
4191 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4193 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
4195 (usadv16qi): New define_expand.
4196 (usadv8hi): New define_expand.
4198 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
4201 * ipa-inline-transform.c (update_noncloned_frequencies): Always
4203 (inline_transform): Likewise.
4204 * predict.c (counts_to_freqs): Remove useless conditional.
4205 * profile-count.h (profile_count::apply_scale): Move sanity check.
4206 * tree-inline.c (copy_bb): Always scale.
4207 (copy_cfg_body): Likewise.
4209 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
4212 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
4214 (*cmp_ite0): Add enabled_for_depr_it attribute.
4215 (*cmp_ite1): Likewise.
4217 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
4219 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
4222 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4224 * tree-vrp.c (vrp_int_const_binop): Return true on success and
4225 return the value by pointer.
4226 (extract_range_from_multiplicative_op_1): Update accordingly.
4227 Return as soon as an operation fails.
4229 2017-11-05 Tom de Vries <tom@codesourcery.com>
4232 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
4233 (DEF_SANITIZER_BUILTIN): ... here.
4234 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
4235 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
4237 2017-11-05 Tom de Vries <tom@codesourcery.com>
4240 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
4242 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
4243 ASM_OUTPUT_BEFORE_CASE_LABEL call.
4244 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
4246 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4247 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4248 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4250 2017-11-05 Tom de Vries <tom@codesourcery.com>
4253 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
4256 2017-11-04 Michael Clark <michaeljclark@mac.com>
4258 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
4259 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
4261 (*addsi3_extended): Likewise.
4262 (*addsi3_extended2): Likewise.
4263 (<optab>si3): Likewise.
4264 (<optab>di3): Likewise.
4265 (<optab><mode>3): Likewise.
4266 (<*optabe>si3_internal): Likewise.
4267 (zero_extendqi<SUPERQI:mode>2): Likewise.
4268 (*add<mode>hi3): Likewise.
4269 (*xor<mode>hi3): Likewise.
4270 (<optab>di3): Likewise.
4271 (*<optab>si3_extend): Likewise.
4272 (*sge<u>_<X:mode><GPR:mode>): Likewise.
4273 (*slt<u>_<X:mode><GPR:mode>): Likewise.
4274 (*sle<u>_<X:mode><GPR:mode>): Likewise.
4276 2017-11-04 Andrew Waterman <andrew@sifive.com>
4278 * config/riscv/riscv.c (riscv_option_override): Conditionally set
4279 TARGET_STRICT_ALIGN based upon -mtune argument.
4281 2017-11-04 Andrew Waterman <andrew@sifive.com>
4283 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
4285 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
4287 * config/i386/i386.c (choose_basereg): Use optional scratch
4288 register and add assertion.
4289 (x86_emit_outlined_ms2sysv_save): Use scratch register when
4290 needed, and don't allocate stack.
4291 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
4292 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
4293 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
4295 2017-11-03 Jeff Law <law@redhat.com>
4297 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
4298 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
4299 to probe at the start of a noreturn function.
4301 2017-11-03 Jakub Jelinek <jakub@redhat.com>
4303 PR tree-optimization/78821
4304 * gimple-ssa-store-merging.c: Update the file comment.
4305 (MAX_STORE_ALIAS_CHECKS): Define.
4306 (struct store_operand_info): New type.
4307 (store_operand_info::store_operand_info): New constructor.
4308 (struct store_immediate_info): Add rhs_code and ops data members.
4309 (store_immediate_info::store_immediate_info): Add rhscode, op0r
4310 and op1r arguments to the ctor, initialize corresponding data members.
4311 (struct merged_store_group): Add load_align_base and load_align
4313 (merged_store_group::merged_store_group): Initialize them.
4314 (merged_store_group::do_merge): Update them.
4315 (merged_store_group::apply_stores): Pick the constant for
4316 encode_tree_to_bitpos from one of the two operands, or skip
4317 encode_tree_to_bitpos if neither operand is a constant.
4318 (class pass_store_merging): Add process_store method decl. Remove
4319 bool argument from terminate_all_aliasing_chains method decl.
4320 (pass_store_merging::terminate_all_aliasing_chains): Remove
4321 var_offset_p argument and corresponding handling.
4322 (stmts_may_clobber_ref_p): New function.
4323 (compatible_load_p): New function.
4324 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
4325 if there is overlap and rhs_code is not INTEGER_CST. For
4326 non-overlapping stores terminate group if rhs is not mergeable.
4327 (get_alias_type_for_stmts): Change first argument from
4328 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
4329 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
4330 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
4332 (get_location_for_stmts): Change first argument from
4333 auto_vec<gimple *> & to vec<gimple *> &.
4334 (struct split_store): Remove orig_stmts data member, add orig_stores.
4335 (split_store::split_store): Create orig_stores rather than orig_stmts.
4336 (find_constituent_stmts): Renamed to ...
4337 (find_constituent_stores): ... this. Change second argument from
4338 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
4339 to info structures rather than the statements.
4340 (split_group): Rename ALLOW_UNALIGNED argument to
4341 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
4342 it. Adjust find_constituent_stores caller.
4343 (imm_store_chain_info::output_merged_store): Handle rhs_code other
4344 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
4345 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
4346 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
4347 (mem_valid_for_store_merging): New function.
4348 (handled_load): New function.
4349 (pass_store_merging::process_store): New method.
4350 (pass_store_merging::execute): Use process_store method. Adjust
4351 terminate_all_aliasing_chains caller.
4353 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4355 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4356 Return true for more constants, symbols and label references.
4357 (aarch64_valid_floating_const): Remove unused function.
4359 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4362 * config/aarch64/aarch64.c (aarch64_layout_frame):
4363 Undo forcing of LR at bottom of frame.
4365 2017-11-03 Jeff Law <law@redhat.com>
4368 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
4369 for int_registers_saved.
4371 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
4372 extracted from tree-ssa-dom.c.
4373 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
4374 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
4375 (record_equivalences_from_incoming_edge): Add additional argument
4376 to single_pred_edge_ignoring_loop_edges call.
4377 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
4378 (uncprop_dom_walker::before_dom_children): Add additional argument
4379 to single_pred_edge_ignoring_loop_edges call.
4380 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
4381 single_pred_edge_ignoring_loop_edges rather than open coding.
4382 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
4384 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
4386 * match.pd (-(-A)): Rewrite.
4388 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4390 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
4391 (rs6000_emit_int_cmove): New declaration.
4392 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
4393 (rs6000_emit_sISEL): Delete.
4394 (rs6000_emit_int_cmove): Make non-static.
4395 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
4396 instead of rs6000_emit_sISEL.
4398 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
4400 * asan.c (create_cond_insert_point): Maintain profile.
4401 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
4403 * basic-block.h (struct basic_block_def): Remove frequency.
4404 (EDGE_FREQUENCY): Use to_frequency
4405 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
4407 (find_traces): Update to use to_frequency.
4408 (find_traces_1_round): Likewise; use only IPA counts.
4409 (bb_to_key): Likewise.
4410 (connect_traces): Use IPA counts only.
4411 (copy_bb_p): Update to use to_frequency.
4412 (fix_up_crossing_landing_pad): Likewise.
4413 (sanitize_hot_paths): Likewise.
4414 * bt-load.c (basic_block_freq): Likewise.
4415 * cfg.c (init_flow): Set count_max to uninitialized.
4416 (check_bb_profile): Remove frequencies; check counts.
4417 (dump_bb_info): Do not dump frequencies.
4418 (update_bb_profile_for_threading): Update counts only.
4419 (scale_bbs_frequencies_int): Likewise.
4420 (MAX_SAFE_MULTIPLIER): Remove.
4421 (scale_bbs_frequencies_gcov_type): Update counts only.
4422 (scale_bbs_frequencies_profile_count): Update counts only.
4423 (scale_bbs_frequencies): Update counts only.
4424 * cfg.h (struct control_flow_graph): Add count-max.
4425 (update_bb_profile_for_threading): Update prototype.
4426 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
4427 (find_many_sub_basic_blocks): Likewise.
4428 * cfgcleanup.c (try_forward_edges): Likewise.
4429 (try_crossjump_to_edge): Likewise.
4430 * cfgexpand.c (expand_gimple_cond): Likewise.
4431 (expand_gimple_tailcall): Likewise.
4432 (construct_init_block): Likewise.
4433 (construct_exit_block): Likewise.
4434 * cfghooks.c (verify_flow_info): Check consistency of counts.
4435 (dump_bb_for_graph): Do not dump frequencies.
4436 (split_block_1): Do not update frequencies.
4437 (split_edge): Do not update frequencies.
4438 (make_forwarder_block): Do not update frequencies.
4439 (duplicate_block): Do not update frequencies.
4440 (account_profile_record): Do not update frequencies.
4441 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
4442 for global heuristics.
4443 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
4444 (expected_loop_iterations_unbounded): Use counts only.
4445 * cfgloopmanip.c (scale_loop_profile): Simplify.
4446 (create_empty_loop_on_edge): Simplify
4448 (duplicate_loop_to_header_edge): Simplify
4449 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
4450 (update_br_prob_note): Take care of removing note when profile
4452 (relink_block_chain): Do not dump frequency.
4453 (rtl_account_profile_record): Use to_frequency.
4454 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
4455 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
4456 (cgraph_update_edges_for_call_stmt_node): Likewise.
4457 (cgraph_edge::verify_count_and_frequency): Update.
4458 (cgraph_node::verify_node): Temporarily disable frequency verification.
4459 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
4460 to_cgraph_frequency.
4461 (cgraph_edge::rebuild_edges): Convert to ipa counts.
4462 * cgraphunit.c (init_lowered_empty_function): Do not initialize
4464 (cgraph_node::expand_thunk): Update profile.
4465 * except.c (dw2_build_landing_pads): Do not update frequency.
4466 * final.c (compute_alignments): Use to_frequency.
4467 (dump_basic_block_info): Do not dump frequency.
4468 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
4469 (dump_gimple_bb_header): Do not dump frequency.
4470 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
4472 * gimple-streamer-in.c (input_bb): Do not stream frequency.
4473 * gimple-streamer-out.c (output_bb): Do not stream frequency.
4474 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
4475 (init_before_recovery): Do not update frequency.
4476 (sched_create_recovery_edges): Do not update frequency.
4477 * hsa-gen.c (convert_switch_statements): Do not update frequency.
4478 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
4479 (ipa_cp_c_finalize): Set max_count to uninitialized.
4480 * ipa-fnsummary.c (get_minimal_bb): Use counts.
4481 (param_change_prob): Use counts.
4482 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
4484 * ipa-split.c (consider_split): Use to_frequency.
4485 (split_function): Use to_frequency.
4486 * ira-build.c (loop_compare_func): Likewise.
4487 (mark_loops_for_removal): Likewise.
4488 (mark_all_loops_for_removal): Likewise.
4489 * loop-doloop.c (doloop_modify): Do not update frequency.
4490 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
4492 * lto-streamer-in.c (input_function): Update count_max.
4493 * omp-expand.c (expand_omp_taskreg): Update count_max.
4494 * omp-simd-clone.c (simd_clone_adjust): Update profile.
4495 * predict.c (maybe_hot_frequency_p): Use to_frequency.
4496 (maybe_hot_count_p): Use ipa counts only.
4497 (maybe_hot_bb_p): Simplify.
4498 (maybe_hot_edge_p): Simplify.
4499 (probably_never_executed): Do not take frequency argument.
4500 (probably_never_executed_bb_p): Do not pass frequency.
4501 (probably_never_executed_edge_p): Likewise.
4502 (combine_predictions_for_bb): Check that profile is nonzero.
4503 (propagate_freq): Do not set frequency.
4504 (drop_profile): Simplify.
4505 (counts_to_freqs): Simplify.
4506 (expensive_function_p): Use to_frequency.
4507 (propagate_unlikely_bbs_forward): Simplify.
4508 (determine_unlikely_bbs): Simplify.
4509 (estimate_bb_frequencies): Add hack to silence graphite issues.
4510 (compute_function_frequency): Use ipa counts.
4511 (pass_profile::execute): Update.
4512 (rebuild_frequencies): Use counts only.
4513 (force_edge_cold): Use counts only.
4514 * profile-count.c (profile_count::dump): Dump new count types.
4515 (profile_count::differs_from_p): Check compatiblity.
4516 (profile_count::to_frequency): New function.
4517 (profile_count::to_cgraph_frequency): New function.
4518 * profile-count.h (struct function): Declare.
4519 (enum profile_quality): Add profile_guessed_local and
4520 profile_guessed_global0.
4521 (class profile_proability): Decrease number of bits to 29;
4522 update from_reg_br_prob_note and to_reg_br_prob_note.
4523 (class profile_count: Update comment; decrease number of bits
4524 to 61. Check compatibility.
4525 (profile_count::compatible_p): New private member function.
4526 (profile_count::ipa_p): New member function.
4527 (profile_count::operator<): Handle global zero correctly.
4528 (profile_count::operator>): Handle global zero correctly.
4529 (profile_count::operator<=): Handle global zero correctly.
4530 (profile_count::operator>=): Handle global zero correctly.
4531 (profile_count::nonzero_p): New member function.
4532 (profile_count::force_nonzero): New member function.
4533 (profile_count::max): New member function.
4534 (profile_count::apply_scale): Handle IPA scalling.
4535 (profile_count::guessed_local): New member function.
4536 (profile_count::global0): New member function.
4537 (profile_count::ipa): New member function.
4538 (profile_count::to_frequency): Declare.
4539 (profile_count::to_cgraph_frequency): Declare.
4540 * profile.c (OVERLAP_BASE): Delete.
4541 (compute_frequency_overlap): Delete.
4542 (compute_branch_probabilities): Do not use compute_frequency_overlap.
4543 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
4544 * sched-ebb.c (rank): Use counts only.
4545 * shrink-wrap.c (handle_simple_exit): Use counts only.
4546 (try_shrink_wrapping): Use counts only.
4547 (place_prologue_for_one_component): Use counts only.
4548 * tracer.c (find_best_predecessor): Use to_frequency.
4549 (find_trace): Use to_frequency.
4550 (tail_duplicate): Use to_frequency.
4551 * trans-mem.c (expand_transaction): Do not update frequency.
4552 * tree-call-cdce.c: Do not update frequency.
4553 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
4554 (gimple_merge_blocks): Likewise.
4555 (gimple_split_edge): Likewise.
4556 (gimple_duplicate_sese_region): Likewise.
4557 (gimple_duplicate_sese_tail): Likewise.
4558 (move_sese_region_to_fn): Likewise.
4559 (gimple_account_profile_record): Likewise.
4560 (insert_cond_bb): Likewise.
4561 * tree-complex.c (expand_complex_div_wide): Likewise.
4562 * tree-eh.c (lower_resx): Update profile.
4563 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
4565 (initialize_cfun): Do not initialize frequencies
4566 (freqs_to_counts): Delete.
4567 (copy_cfg_body): Ignore count parameter.
4568 (copy_body): Update.
4569 (expand_call_inline): Update count_max.
4570 (optimize_inline_calls): Update count_max.
4571 (tree_function_versioning): Update count_max.
4572 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
4573 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
4575 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
4576 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
4577 (try_peel_loop): Likewise.
4578 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
4580 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
4581 (tree_transform_and_unroll_loop): Do not use frequencies
4582 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
4583 Use reliable prediction only.
4584 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
4585 * tree-ssa-sink.c (select_best_block): Use to_frequency.
4586 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
4587 probability scaling.
4588 * tree-ssa-threadupdate.c (create_block_for_threading): Do
4589 not update frequency
4590 (any_remaining_duplicated_blocks): Likewise.
4591 (update_profile): Likewise.
4592 (estimated_freqs_path): Delete.
4593 (freqs_to_counts_path): Delete.
4594 (clear_counts_path): Delete.
4595 (ssa_fix_duplicate_block_edges): Likewise.
4596 (duplicate_thread_path): Likewise.
4597 * tree-switch-conversion.c (gen_inbound_check): Use counts.
4598 * tree-tailcall.c (decrease_profile): Do not update frequency.
4599 (eliminate_tail_call): Likewise.
4600 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
4601 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
4602 (optimize_mask_stores): Likewise.
4603 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4604 * ubsan.c (ubsan_expand_null_ifn): Update profile.
4605 (ubsan_expand_ptr_ifn): Update profile.
4606 * value-prof.c (gimple_ic): Simplify.
4607 * value-prof.h (gimple_ic): Update prototype.
4608 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
4609 * ipa-inline.c (compute_uninlined_call_time): Be sure that
4611 (want_inline_self_recursive_call_p): Likewise.
4612 (resolve_noninline_speculation): Only cummulate defined counts.
4613 (inline_small_functions): Use nonzero_p.
4614 (ipa_inline): Do not access freed node.
4616 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4618 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4619 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
4621 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
4623 * config/riscv/riscv.c (riscv_legitimize_move): Handle
4624 non-legitimate address.
4626 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4628 * config/rs6000/rs6000.md (*lt0_disi): Delete.
4629 (*lt0_<mode>di, *lt0_<mode>si): New.
4631 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4633 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
4634 TARGET_PAIRED_FLOAT.
4636 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
4637 Jim Wilson <jim.wilson@linaro.org>
4639 * config/aarch64/aarch64-cores.def (saphira): New CPU.
4640 * config/aarch64/aarch64-tune.md: Regenerated.
4641 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
4642 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
4644 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
4646 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
4647 (arc_expand_prologue): Restore blink for millicode.
4648 * config/arc/linux.h (LINK_EH_SPEC): Defined.
4650 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
4653 * config/i386/i386.c (ix86_vector_duplicate_value): Use
4654 gen_vec_duplicate after forcing the scalar into a register.
4656 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
4658 * combine (try_combine): Print the insns input to try_combine to the
4661 2017-11-02 Steve Ellcey <sellcey@cavium.com>
4664 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
4665 Remove second argument from aarch64_process_target_attr call.
4666 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
4668 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
4670 (aarch64_handle_attr_arch): Remove second argument.
4671 (aarch64_handle_attr_cpu): Ditto.
4672 (aarch64_handle_attr_tune): Ditto.
4673 (aarch64_handle_attr_isa_flags): Ditto.
4674 (aarch64_process_one_target_attr): Ditto.
4675 (aarch64_process_target_attr): Ditto.
4676 (aarch64_option_valid_attribute_p): Remove second argument.
4677 on aarch64_process_target_attr call.
4679 2017-11-02 David Malcolm <dmalcolm@redhat.com>
4681 * diagnostic.c: Include "selftest-diagnostic.h".
4682 (selftest::assert_location_text): New function.
4683 (selftest::test_diagnostic_get_location_text): New function.
4684 (selftest::diagnostic_c_tests): Call it.
4686 2017-11-02 David Malcolm <dmalcolm@redhat.com>
4688 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
4689 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
4690 (class selftest::test_diagnostic_context): Move to...
4691 * selftest-diagnostic.c: New file.
4692 * selftest-diagnostic.h: New file.
4694 2017-11-02 James Bowman <james.bowman@ftdichip.com>
4696 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
4697 offset range for FT32B.
4698 * config/ft32/ft32.h: option "mcompress" enables relaxation.
4699 * config/ft32/ft32.md: Add TARGET_NOPM.
4700 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
4701 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
4703 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
4705 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
4707 2017-11-02 Jeff Law <law@redhat.com>
4709 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
4710 virtual keyword on FINAL OVERRIDE members.
4712 * tree-ssa-propagate.h (ssa_propagation_engine): Group
4713 virtuals together. Add virtual destructor.
4714 (substitute_and_fold_engine): Similarly.
4716 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
4718 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
4720 2017-11-02 Richard Biener <rguenther@suse.de>
4722 PR tree-optimization/82795
4723 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
4725 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4727 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
4729 (gcc_GAS_CHECK_FEATURE): Remove.
4730 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
4731 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
4732 * configure: Regenerate.
4734 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
4736 * config/arc/arc.c (hwloop_optimize): Account for empty
4739 2017-11-02 Richard Biener <rguenther@suse.de>
4742 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
4743 Truncate ARRAY_REF index and element size.
4745 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4747 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
4749 2017-11-01 Jeff Law <law@redhat.com>
4751 * tree-ssa-ccp.c (ccp_folder): New class derived from
4752 substitute_and_fold_engine.
4753 (ccp_folder::get_value): New member function.
4754 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
4755 (ccp_fold_stmt): Remove prototype.
4756 (ccp_finalize): Call substitute_and_fold from the ccp_class.
4757 * tree-ssa-copy.c (copy_folder): New class derived from
4758 substitute_and_fold_engine.
4759 (copy_folder::get_value): Renamed from get_value.
4760 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
4761 * tree-vrp.c (vrp_folder): New class derived from
4762 substitute_and_fold_engine.
4763 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
4764 (vrp_folder::get_value): New member function.
4765 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
4766 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
4767 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
4768 provide a class interface to folder/substitute routines.
4769 (ssa_prop_fold_stmt_fn): Remove typedef.
4770 (ssa_prop_get_value_fn): Likewise.
4771 (subsitute_and_fold): Remove prototype.
4772 (replace_uses_in): Likewise.
4773 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
4774 Renamed from replace_uses_in. Call the virtual member function
4775 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
4776 (substitute_and_fold_dom_walker): Remove initialization of
4777 data member entries for calbacks. Add substitute_and_fold_engine
4778 member and initialize it.
4779 (substitute_and_fold_dom_walker::before_dom_children0: Use the
4780 member functions for get_value, replace_phi_args_in c
4781 replace_uses_in, and fold_stmt calls.
4782 (substitute_and_fold_engine::substitute_and_fold): Renamed from
4783 substitute_and_fold. Remove assert. Update ctor call.
4785 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
4786 (ssa_prop_visit_phi_fn): Likewise.
4787 (class ssa_propagation_engine): New class to provide an interface
4789 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
4791 (ssa_prop_visit_phi): Likewise.
4792 (ssa_propagation_engine::simulate_stmt): Moved into class.
4793 Call visit_phi/visit_stmt from the class rather than via
4794 file scoped static variables.
4795 (ssa_propagation_engine::simulate_block): Moved into class.
4796 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
4797 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
4798 set file scoped statics for the visit_stmt/visit_phi callbacks.
4799 * tree-complex.c (complex_propagate): New class derived from
4800 ssa_propagation_engine.
4801 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
4802 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
4803 (tree_lower_complex): Call ssa_propagate via the complex_propagate
4805 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
4806 ssa_propagation_engine.
4807 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
4808 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
4809 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
4810 * tree-ssa-copy.c (copy_prop): New class derived from
4811 ssa_propagation_engine.
4812 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
4813 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
4814 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
4815 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
4816 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
4817 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
4818 (execute_vrp): Call ssa_propagate from the vrp_prop class.
4820 2017-11-01 Jakub Jelinek <jakub@redhat.com>
4822 PR rtl-optimization/82778
4823 PR rtl-optimization/82597
4824 * compare-elim.c (struct comparison): Add in_a_setter field.
4825 (find_comparison_dom_walker::before_dom_children): Remove killed
4826 bitmap and df_simulate_find_defs call, instead walk the defs.
4827 Compute last_setter and initialize in_a_setter. Merge definitions
4828 with first initialization for a few variables.
4829 (try_validate_parallel): Use insn_invalid_p instead of
4830 recog_memoized. Return insn rather than just the pattern.
4831 (try_merge_compare): Fix up comment. Don't uselessly test if
4832 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
4834 (execute_compare_elim_after_reload): Remove df_chain_add_problem
4837 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4838 Alan Hayward <alan.hayward@arm.com>
4839 David Sherwood <david.sherwood@arm.com>
4841 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
4842 aarch64_hard_regno_nregs to get the number of registers
4845 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4846 Alan Hayward <alan.hayward@arm.com>
4847 David Sherwood <david.sherwood@arm.com>
4849 * config/aarch64/constraints.md (Upl): Rename to...
4851 * config/aarch64/aarch64.md
4852 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
4855 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4856 Alan Hayward <alan.hayward@arm.com>
4857 David Sherwood <david.sherwood@arm.com>
4859 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
4860 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
4863 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4864 Alan Hayward <alan.hayward@arm.com>
4865 David Sherwood <david.sherwood@arm.com>
4867 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
4868 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
4869 (aarch64_evpc_dup): Generate rtl direcly, rather than using
4871 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
4872 of a single element.
4873 * config/aarch64/iterators.md: Add a comment above the permute
4874 unspecs to say that they are generated directly by
4875 aarch64_expand_vec_perm_const.
4876 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
4878 2017-11-01 Nathan Sidwell <nathan@acm.org>
4880 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
4882 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4884 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
4885 and medany code models, and describe what they do.
4887 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4889 Revert accidental duplicate:
4891 * combine.c (can_change_dest_mode): Reject changes in
4892 REGMODE_NATURAL_SIZE.
4894 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
4896 PR rtl-optimization/64682
4897 PR rtl-optimization/69567
4898 PR rtl-optimization/69737
4899 PR rtl-optimization/82683
4900 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
4901 register mentioned in the note, drop the note, unless it came from I3,
4902 in which case it should go to I3 again.
4904 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4906 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
4907 and return false if not.
4908 (clear_bytes_written_by, live_bytes_read): Update accordingly.
4910 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4912 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
4913 range is known to be empty.
4915 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4916 Alan Hayward <alan.hayward@arm.com>
4917 David Sherwood <david.sherwood@arm.com>
4919 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
4920 and CONST_VECTOR_NUNITS instead of computing the number of units from
4921 the byte sizes of the vector and element.
4922 (simplify_binary_operation_1): Likewise.
4923 (simplify_const_binary_operation): Likewise.
4924 (simplify_ternary_operation): Likewise.
4926 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4927 Alan Hayward <alan.hayward@arm.com>
4928 David Sherwood <david.sherwood@arm.com>
4930 * var-tracking.c (INT_MEM_OFFSET): Replace with...
4931 (int_mem_offset): ...this new function.
4932 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
4933 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
4934 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
4937 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4938 Alan Hayward <alan.hayward@arm.com>
4939 David Sherwood <david.sherwood@arm.com>
4941 * lower-subreg.c (interesting_mode_p): New function.
4942 (compute_costs, find_decomposable_subregs, decompose_register)
4943 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
4944 (resolve_clobber, dump_choices): Use it.
4946 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4947 Alan Hayward <alan.hayward@arm.com>
4948 David Sherwood <david.sherwood@arm.com>
4950 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
4952 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4953 Alan Hayward <alan.hayward@arm.com>
4954 David Sherwood <david.sherwood@arm.com>
4956 * alias.c (find_base_value, find_base_term): Only process integer
4957 truncations. Check the precision rather than the size.
4959 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4960 Alan Hayward <alan.hayward@arm.com>
4961 David Sherwood <david.sherwood@arm.com>
4963 * machmode.h (is_narrower_int_mode): New function
4964 * optabs.c (expand_float, expand_fix): Use it.
4965 * dwarf2out.c (rotate_loc_descriptor): Likewise.
4967 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4968 Alan Hayward <alan.hayward@arm.com>
4969 David Sherwood <david.sherwood@arm.com>
4971 * rtl.h (narrower_subreg_mode): New function.
4972 * ira-color.c (update_costs_from_allocno): Use it.
4974 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4975 Alan Hayward <alan.hayward@arm.com>
4976 David Sherwood <david.sherwood@arm.com>
4978 * optabs-query.h (convert_optab_p): New function, split out from...
4979 (convert_optab_handler): ...here.
4980 (widening_optab_handler): Delete.
4981 (find_widening_optab_handler): Remove permit_non_widening parameter.
4982 (find_widening_optab_handler_and_mode): Likewise. Provide an
4983 override that operates on mode class wrappers.
4984 * optabs-query.c (widening_optab_handler): Delete.
4985 (find_widening_optab_handler_and_mode): Remove permit_non_widening
4986 parameter. Assert that the two modes are the same class and that
4987 the "from" mode is narrower than the "to" mode. Use
4988 convert_optab_handler instead of widening_optab_handler.
4989 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
4990 instead of widening_optab_handler.
4991 * expr.c (expand_expr_real_2): Update calls to
4992 find_widening_optab_handler.
4993 * optabs.c (expand_widen_pattern_expr): Likewise.
4994 (expand_binop_directly): Take the insn_code as a parameter.
4995 (expand_binop): Only call find_widening_optab_handler for
4996 conversion optabs; use optab_handler otherwise. Update calls
4997 to find_widening_optab_handler and expand_binop_directly.
4998 Use convert_optab_handler instead of widening_optab_handler.
4999 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
5000 find_widening_optab_handler and use scalar_mode rather than
5002 (convert_plusminus_to_widen): Likewise.
5004 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5005 Alan Hayward <alan.hayward@arm.com>
5006 David Sherwood <david.sherwood@arm.com>
5008 * machmode.h (fixed_size_mode): New class.
5009 * rtl.h (get_pool_mode): Return fixed_size_mode.
5010 * gengtype.c (main): Add fixed_size_mode.
5011 * target.def (get_raw_result_mode): Return a fixed_size_mode.
5012 (get_raw_arg_mode): Likewise.
5013 * doc/tm.texi: Regenerate.
5014 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
5015 * targhooks.c (default_get_reg_raw_mode): Likewise.
5016 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
5017 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
5018 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
5019 (msp430_get_raw_result_mode): Likewise.
5020 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
5021 * dbxout.c (dbxout_parms): Require fixed-size modes.
5022 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
5023 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5024 * omp-low.c (lower_oacc_reductions): Likewise.
5025 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
5026 (simplify_subreg): Update accordingly.
5027 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
5028 (force_const_mem): Update accordingly. Return NULL_RTX for modes
5029 that aren't fixed-size.
5030 (get_pool_mode): Return a fixed_size_mode.
5031 (output_constant_pool_2): Take a fixed_size_mode.
5033 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5034 Alan Hayward <alan.hayward@arm.com>
5035 David Sherwood <david.sherwood@arm.com>
5037 * doc/rtl.texi (vec_series): Document.
5038 (const): Say that the operand can be a vec_series.
5039 * rtl.def (VEC_SERIES): New rtx code.
5040 * rtl.h (const_vec_series_p_1): Declare.
5041 (const_vec_series_p): New function.
5042 * emit-rtl.h (gen_const_vec_series): Declare.
5043 (gen_vec_series): Likewise.
5044 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
5045 (gen_vec_series): Likewise.
5046 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
5047 * simplify-rtx.c (simplify_unary_operation): Handle negations
5049 (simplify_binary_operation_series): New function.
5050 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
5051 (test_vector_ops_series): New function.
5052 (test_vector_ops): Call it.
5053 * config/powerpcspe/altivec.md (altivec_lvsl): Use
5054 gen_const_vec_series.
5055 (altivec_lvsr): Likewise.
5056 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
5058 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5059 Alan Hayward <alan.hayward@arm.com>
5060 David Sherwood <david.sherwood@arm.com>
5062 * doc/rtl.texi (const): Update description of address constants.
5063 Say that vector constants are allowed too.
5064 * common.md (E, F): Use CONSTANT_P instead of checking for
5066 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
5067 checking for CONST_VECTOR.
5068 * expmed.c (make_tree): Use build_vector_from_val for a CONST
5070 * expr.c (expand_expr_real_2): Check for vector modes instead
5071 of checking for CONST_VECTOR.
5072 * rtl.h (const_vec_p): New function.
5073 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
5074 (unwrap_const_vec_duplicate): Handle them here too.
5076 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5077 David Malcolm <dmalcolm@redhat.com>
5078 Alan Hayward <alan.hayward@arm.com>
5079 David Sherwood <david.sherwood@arm.com>
5081 * rtl.h (vec_duplicate_p): New function.
5082 * selftest-rtl.c (assert_rtx_eq_at): New function.
5083 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
5084 (assert_rtx_eq_at): Declare.
5085 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
5086 * selftest-run-tests.c (selftest::run_tests): Call it.
5087 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
5088 (simplify_unary_operation_1): Recursively handle vector duplicates.
5089 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
5091 (simplify_subreg): Handle subregs of vector duplicates.
5092 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
5093 (selftest::simplify_rtx_c_tests): New functions.
5095 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5096 Alan Hayward <alan.hayward@arm.com>
5097 David Sherwood <david.sherwood@arm.com>
5099 * emit-rtl.h (gen_const_vec_duplicate): Declare.
5100 (gen_vec_duplicate): Likewise.
5101 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
5103 (gen_const_vector): ...here.
5104 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
5105 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
5106 whose elements are all equal.
5107 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
5108 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
5109 (simplify_relational_operation): Likewise.
5110 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
5112 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
5113 (aarch64_expand_vector_init): Likewise.
5114 * config/arm/arm.c (neon_vdup_constant): Likewise.
5115 (neon_expand_vector_init): Likewise.
5116 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
5117 (arm_block_set_unaligned_vect): Likewise.
5118 (arm_block_set_aligned_vect): Likewise.
5119 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
5120 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
5121 (expand_vec_perm_even_odd_pack): Likewise.
5122 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
5123 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
5124 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
5125 gen_const_vec_duplicate.
5126 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
5127 * config/mips/mips.c (mips_gen_const_int_vector): Use
5128 gen_const_vec_duplicate.
5129 (mips_expand_vector_init): Use CONST0_RTX.
5130 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
5131 (define_split): Use gen_const_vec_duplicate.
5132 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
5133 (define_split): Use gen_const_vec_duplicate.
5134 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
5135 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
5136 * config/spu/spu.c (spu_const): Likewise.
5138 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5139 Alan Hayward <alan.hayward@arm.com>
5140 David Sherwood <david.sherwood@arm.com>
5142 * combine.c (can_change_dest_mode): Reject changes in
5143 REGMODE_NATURAL_SIZE.
5145 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
5147 * configure.ac (--enable-libssp): New.
5148 (gcc_cv_libc_provides_ssp): Check for explicit setting before
5149 trying to determine target-specific default. Adjust indentation.
5150 * configure: Regenerated.
5151 * doc/install.texi (Configuration): Expand --disable-libssp
5154 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
5156 config/i386/i386.c (ix86_expand_epilogue): Correct stack
5159 2017-10-31 Martin Jambor <mjambor@suse.cz>
5162 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
5164 2017-10-31 David Malcolm <dmalcolm@redhat.com>
5166 * auto-profile.c (autofdo_source_profile::read): Use
5167 UNKNOWN_LOCATION rather than 0.
5168 * diagnostic-core.h (warning_at_rich_loc): Rename to...
5169 (warning_at): ...this overload.
5170 (warning_at_rich_loc_n): Rename to...
5171 (warning_n): ...this overload.
5172 (error_at_rich_loc): Rename to...
5173 (error_at): ...this overload.
5174 (pedwarn_at_rich_loc): Rename to...
5175 (pedwarn): ...this overload.
5176 (permerror_at_rich_loc): Rename to...
5177 (permerror): ...this overload.
5178 (inform_at_rich_loc): Rename to...
5179 (inform): ...this overload.
5180 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
5181 (diagnostic_n_impl_richloc): Rename to...
5182 (diagnostic_n_impl): ...this rich_location *-based decl.
5183 (inform_at_rich_loc): Rename to...
5184 (inform): ...this, and add an assertion.
5185 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
5186 (warning_at_rich_loc): Rename to...
5187 (warning_at): ...this, and add an assertion.
5188 (warning_at_rich_loc_n): Rename to...
5189 (warning_n): ...this, and add an assertion.
5190 (warning_n): Update location_t-based implementation for removal of
5191 location_t-based diagnostic_n_impl.
5192 (pedwarn_at_rich_loc): Rename to...
5193 (pedwarn): ...this, and add an assertion.
5194 (permerror_at_rich_loc): Rename to...
5195 (permerror): ...this, and add an assertion.
5196 (error_n): Update for removal of location_t-based diagnostic_n_impl.
5197 (error_at_rich_loc): Rename to...
5198 (error_at): ...this, and add an assertion.
5199 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
5200 (driver::do_spec_on_infiles): Likewise.
5201 * substring-locations.c (format_warning_va): Update for renaming
5202 of inform_at_rich_loc.
5204 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
5206 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
5207 _Float<N>X built-in functions so that the variant without the
5208 "__builtin_" prefix is only enabled for the GNU C and Objective C
5209 languages when they are in non-strict ANSI/ISO mode.
5210 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
5211 * target.def (floatn_builtin_p): Add a target hook to control
5212 whether _Float<N> and _Float<N>X built-in functions without the
5213 "__builtin_" prefix are enabled, and return true for C and
5214 Objective C in the default hook. Include langhooks.h in
5216 * targhooks.h (default_floatn_builtin_p): Likewise.
5217 * targhooks.c (default_floatn_builtin_p): Likewise.
5218 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
5219 floatn_builtin_p target hook.
5220 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
5222 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
5223 Eric Botcazou <ebotcazou@adacore.com>
5225 PR rtl-optimization/81803
5226 * lra-constraints.c (curr_insn_transform): Also reload the whole
5227 register for a strict subreg no wider than a word if this is for
5228 a WORD_REGISTER_OPERATIONS target.
5230 2017-10-31 Jason Merrill <jason@redhat.com>
5232 * gdbinit.in: Skip over inlines from timevar.h.
5234 2017-10-31 Martin Liska <mliska@suse.cz>
5236 * doc/gcov.texi: Document new option.
5237 * gcov.c (print_usage): Likewise print it.
5238 (process_args): Support the argument.
5239 (format_count): New function.
5240 (format_gcov): Use the function.
5242 2017-10-31 Martin Liska <mliska@suse.cz>
5244 * gcov.c (struct name_map): do not use typedef.
5245 Define operator== and operator<.
5246 (name_search): Remove.
5247 (name_sort): Remove.
5248 (main): Do not allocate names.
5249 (process_file): Add vertical space.
5250 (generate_results): Use std::find.
5251 (release_structures): Do not release memory.
5252 (find_source): Use std::find.
5254 2017-10-31 Martin Liska <mliska@suse.cz>
5256 * gcov.c (struct line_info): Remove it's typedef.
5257 (line_info::line_info): Add proper ctor.
5258 (line_info::has_block): Do not use a typedef.
5259 (struct source_info): Do not use typedef.
5260 (circuit): Likewise.
5261 (get_cycles_count): Likewise.
5262 (output_intermediate_file): Iterate via vector iterator.
5263 (add_line_counts): Use std::vector methods.
5264 (accumulate_line_counts): Likewise.
5265 (output_lines): Likewise.
5267 2017-10-31 Martin Liska <mliska@suse.cz>
5269 * gcov.c (struct source_info): Remove typedef.
5270 (source_info::source_info): Add proper ctor.
5271 (accumulate_line_counts): Use struct, not it's typedef.
5272 (output_gcov_file): Likewise.
5273 (output_lines): Likewise.
5274 (main): Do not allocate an array.
5275 (output_intermediate_file): Use size of vector container.
5276 (process_file): Resize the vector.
5277 (generate_results): Do not preallocate, use newly added vector
5279 (release_structures): Do not release sources.
5280 (find_source): Use vector methods.
5281 (add_line_counts): Do not use typedef.
5283 2017-10-31 Martin Liska <mliska@suse.cz>
5285 * doc/gcov.texi: Document that.
5286 * gcov.c (add_line_counts): Mark lines with a non-executed
5288 (output_line_beginning): Handle such lines.
5289 (output_lines): Pass new argument.
5290 (output_intermediate_file): Print it in intermediate format.
5292 2017-10-31 Martin Liska <mliska@suse.cz>
5294 * color-macros.h: New file.
5295 * diagnostic-color.c: Factor out color related to macros to
5297 * doc/gcov.texi: Document -k option.
5298 * gcov.c (INCLUDE_STRING): Include string.h.
5299 (print_usage): Add -k option.
5300 (process_args): Parse it.
5301 (pad_count_string): New function.
5302 (output_line_beginning): Likewise.
5303 (DEFAULT_LINE_START): New macro.
5304 (output_lines): Support color output.
5306 2017-10-31 Martin Liska <mliska@suse.cz>
5308 PR gcov-profile/82633
5309 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
5310 their interaction with GCOV infrastructure.
5311 * configure.ac: Add -fkeep-{inline,static}-functions to
5313 * configure: Regenerate.
5315 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
5318 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
5320 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
5323 * config/rs6000/rs6000.md (allocate_stack): Force update interval
5324 into a register if it does not fit into an immediate offset field.
5326 2017-10-31 Olivier Hainque <hainque@adacore.com>
5328 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
5330 2017-10-31 Julia Koval <julia.koval@intel.com>
5332 * config.gcc: Add gfniintrin.h.
5333 * config/i386/gfniintrin.h: New.
5334 * config/i386/i386-builtin-types.def
5335 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5336 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5337 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5338 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5339 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5340 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
5341 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
5342 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
5343 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
5344 V64QI_FTYPE_V64QI_V64QI_INT): New types.
5345 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5346 * config/i386/immintrin.h: Include gfniintrin.h.
5347 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
5349 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
5351 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
5353 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5355 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
5356 (arm_ashldi3_1bit): Remove pattern.
5357 (ashrdi3): Remove shift by 1 expansion.
5358 (arm_ashrdi3_1bit): Remove pattern.
5359 (lshrdi3): Remove shift by 1 expansion.
5360 (arm_lshrdi3_1bit): Remove pattern.
5361 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
5362 cost of ashldi3 by 1.
5363 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
5364 (<shift>di3_neon): Likewise.
5366 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5368 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
5369 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
5370 and (*aarch64_simd_mov<VQ:mode>).
5371 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
5372 pattern alternative.
5373 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
5374 attributes to match pattern alternative.
5376 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
5378 * config.gcc (powerpc*-*-*): Add emmintrin.h.
5379 * config/rs6000/emmintrin.h: New file.
5380 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
5382 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5384 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
5385 * (movdi_vfp_cortexa8): Remove pattern.
5387 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5389 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
5391 (Specific, alpha*-dec-osf5.1): Remove.
5392 (Specific, mips-sgi-irix5): Remove.
5393 (Specific, mips-sgi-irix6): Remove.
5395 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5398 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
5399 arguments to clear_bit_region_be.
5401 2017-10-30 Jim Wilson <wilson@tuliptree.org>
5403 * gimplify.c: Include memmodel.h.
5405 2017-10-30 Martin Jambor <mjambor@suse.cz>
5407 * omp-grid.c (grid_attempt_target_gridification): Also insert a
5408 condition whether loop should be executed at all.
5410 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
5412 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5413 gimple folding of vec_madd() intrinsics.
5414 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
5415 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
5416 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
5418 2017-10-30 Richard Biener <rguenther@suse.de>
5420 PR tree-optimization/82762
5422 2017-10-23 Richard Biener <rguenther@suse.de>
5424 PR tree-optimization/82129
5426 2017-08-01 Richard Biener <rguenther@suse.de>
5428 PR tree-optimization/81181
5429 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5430 (compute_antic): ... end of iteration here.
5432 2017-10-30 Joseph Myers <joseph@codesourcery.com>
5434 * doc/invoke.texi (C Dialect Options): Document -std=c17,
5435 -std=iso9899:2017 and -std=gnu17.
5436 * doc/standards.texi (C Language): Document C17 support.
5437 * doc/cpp.texi (Overview): Mention -std=c17.
5438 (Standard Predefined Macros): Document C11 and C17 values of
5439 __STDC_VERSION__. Do not refer to C99 support as incomplete.
5440 * doc/extend.texi (Inline): Do not list individual options for
5441 standards newer than C99.
5442 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
5444 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
5447 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
5449 * asan.c (asan_finish_file): Align asan globals array by shadow
5452 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5455 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
5456 (struct store_immediate_info): Add bitregion_start and bitregion_end
5458 (store_immediate_info::store_immediate_info): Add brs and bre
5459 arguments and initialize bitregion_{start,end} from those.
5460 (struct merged_store_group): Add bitregion_start, bitregion_end,
5461 align_base and mask fields. Drop unnecessary struct keyword from
5462 struct store_immediate_info. Add do_merge method.
5463 (clear_bit_region_be): Use memset instead of loop storing zeros.
5464 (merged_store_group::do_merge): New method.
5465 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
5466 stores as long as the surrounding bitregions have no gaps.
5467 (merged_store_group::merge_overlapping): Use do_merge.
5468 (merged_store_group::apply_stores): Test that bitregion_{start,end}
5469 is byte aligned, rather than requiring that start and width are
5470 byte aligned. Drop unnecessary struct keyword from
5471 struct store_immediate_info. Allocate and populate also mask array.
5472 Make start of the arrays relative to bitregion_start rather than
5473 start and size them according to bitregion_{end,start} difference.
5474 (struct imm_store_chain_info): Drop unnecessary struct keyword from
5475 struct store_immediate_info.
5476 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
5477 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
5478 struct keyword from struct store_immediate_info.
5479 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
5480 between stores as long as the surrounding bitregions have no gaps.
5482 (struct split_store): Add orig non-static data member.
5483 (split_store::split_store): Initialize orig to false.
5484 (find_constituent_stmts): Return store_immediate_info *, non-NULL
5485 if there is exactly a single original stmt. Change stmts argument
5486 to pointer from reference, if NULL, don't push anything to it. Add
5487 first argument, use it to optimize skipping over orig stmts that
5488 are known to be before bitpos already. Simplify.
5489 (split_group): Return unsigned int count how many stores are or
5490 would be needed rather than a bool. Add allow_unaligned argument.
5491 Change split_stores argument from reference to pointer, if NULL,
5492 only do a dry run computing how many stores would be produced.
5493 Rewritten algorithm to use both alignment and misalign if
5494 !allow_unaligned and handle bitfield stores with gaps.
5495 (imm_store_chain_info::output_merged_store): Set start_byte_pos
5496 from bitregion_start instead of start. Compute allow_unaligned
5497 here, if true, do 2 split_group dry runs to compute which one
5498 produces fewer stores and prefer aligned if equal. Punt if
5499 new count is bigger or equal than original before emitting any
5500 statements, rather than during that. Remove no longer needed
5501 new_ssa_names tracking. Replace num_stmts with
5502 split_stores.length (). Use 32-bit stack allocated entries
5503 in split_stores auto_vec. Try to reuse original store lhs/rhs1
5504 if possible. Handle bitfields with gaps.
5505 (pass_store_merging::execute): Ignore bitsize == 0 stores.
5506 Compute bitregion_{start,end} for the stores and construct
5507 store_immediate_info with that. Formatting fixes.
5509 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
5512 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
5513 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
5515 2017-10-29 Jim Wilson <wilson@tuliptree.org>
5517 * gimplify.c: Include tm_p.h.
5519 * common.opt (gcoff): Re-add as ignored option.
5520 (gcoff1, gcoff2, gcoff3): Likewise.
5522 * Makefile.in (OBJS): Delete sdbout.o.
5523 (GTFILES): Delete $(srcdir)/sdbout.c.
5524 * debug.h: Delete sdb_debug_hooks.
5525 * final.c: Delete sdbout.h include.
5526 (final_scan_insn): Delete SDB_DEBUG check.
5527 (rest_of_clean_state): Likewise.
5528 * output.h: Delete sdb_begin_function_line.
5531 * toplev.c: Delete sdbout.h include.
5532 (process_options): Delete SDB_DEBUG check.
5533 * tree-core.h (tree_type_common): Delete pointer field of
5535 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
5536 TYPE_SYMTAB_POINTER.
5537 * tree.h (TYPE_SYMTAB_POINTER): Delete.
5538 (TYPE_SYMTAB_IS_POINTER): Delete.
5539 (TYPE_SYMTAB_IS_DIE): Renumber.
5540 * xcoffout.c: Refer to former sdbout.c file.
5541 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
5543 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
5544 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
5545 (Debugging Options): Delete -gcoff.
5546 (-gstabs): Delete SDB reference.
5548 (-gcoff@var{level}): Delete.
5549 * doc/passes.texi (Debugging information output): Delete SDB and
5550 sdbout.c references.
5551 * doc/tm.texi: Regenerate.
5552 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
5553 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
5555 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
5556 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
5557 (SDB_DEBUGGING_INFO): Delete.
5558 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
5559 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
5560 * target.def (output_source_filename): Delete COFF reference.
5562 * common.opt (gcoff): Delete.
5563 (gxcoff+): Update Negative chain.
5564 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
5566 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
5567 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
5568 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
5570 (expand_function_start): Change sdb reference to past tense.
5571 (expand_function_end): Change sdb reference to past tense.
5572 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
5573 * opts.c (debug_type_names): Delete coff entry.
5574 (common_handle_option): Delete OPT_gcoff case.
5575 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
5577 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
5578 * config/cris/cris.h: Delete SDB reference in comment.
5579 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
5580 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
5581 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
5582 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
5584 (ix86_expand_prologue): Likewise.
5585 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
5586 * config/ia64/ia64.h: Likewise.
5587 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
5588 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
5590 * config/mmix/mmix.h: Likewise.
5591 * config/nds32/nds32.c: Likewise.
5592 * config/stormy/storym16.h: Likewise.
5593 * config/visium/visium.h: Likewise.
5594 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
5596 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
5598 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
5599 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
5600 FRAME_POINTER_REGNUM point at high end of local var area.
5602 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
5604 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
5605 Move comment around. Do not reset best_edge for a copiable
5606 destination if the copy would cause a partition change.
5607 (better_edge_p): Remove redundant check.
5609 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
5611 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
5613 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5615 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
5616 for math functions that have _Float<N> and _Float<N>X variants.
5617 (mathfn_built_in_2): Add support for math functions that have
5618 _Float<N> and _Float<N>X variants.
5619 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
5620 (expand_builtin_mathfn_ternary): Add support for fma with
5621 _Float<N> and _Float<N>X variants.
5622 (expand_builtin): Likewise.
5623 (fold_builtin_3): Likewise.
5624 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
5625 create math function _Float<N> and _Float<N>X variants as external
5627 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
5628 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
5629 the __builtin_ prefix and if not strict ansi, without the prefix.
5630 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
5631 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
5632 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
5633 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
5634 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
5635 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
5636 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
5637 function signatures for fma _Float<N> and _Float<N>X variants.
5638 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
5639 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
5640 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
5641 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
5642 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
5643 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
5644 * gencfn-macros.c (print_case_cfn): Add support for math functions
5645 that have _Float<N> and _Float<N>X variants.
5646 (print_define_operator_list): Likewise.
5647 (fltfn_suffixes): Likewise.
5649 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
5650 for math functions that have _Float<N> and _Float<N>X variants.
5651 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
5652 and _Float<N>X variants.
5653 (COPYSIGN): Likewise.
5656 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
5657 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
5659 (integer_valued_read_call_p): Likewise.
5660 * fold-const-call.c (fold_const_call_ss): Likewise.
5661 (fold_const_call_sss): Add support for copysign, fmin, and fmax
5662 _Float<N> and _Float<N>X variants.
5663 (fold_const_call_ssss): Add support for fma _Float<N> and
5664 _Float<N>X variants.
5665 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
5666 support for copysign and fma _Float<N> and _Float<N>X variants.
5667 (backprop::process_builtin_call_use): Likewise.
5668 * tree-call-cdce.c (can_test_argument_range); Add support for
5669 sqrt _Float<N> and _Float<N>X variants.
5670 (edom_only_function): Likewise.
5671 (get_no_error_domain): Likewise.
5672 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
5673 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
5674 copysign _Float<N> and _Float<N>X variants.
5675 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
5676 handled by machine independent code.
5677 (FMAF128): Likewise.
5678 * doc/cpp.texi (Common Predefined Macros): Document defining
5679 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
5680 fma _Float<N> and _Float<N>X variants.
5682 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
5685 * config/i386/i386-modes.def (CCFPU): Remove definition.
5686 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
5687 (ix86_cc_modes_compatible): Ditto.
5688 (ix86_expand_carry_flag_compare): Ditto.
5689 (ix86_expand_int_movcc): Ditto.
5690 (ix86_expand_int_addcc): Ditto.
5691 (ix86_reverse_condition): Ditto.
5692 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
5693 Return true/false for unordered/ordered fp comparisons.
5694 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
5695 (ix86_prepare_fp_compare_args): Update for rename.
5696 (ix86_expand_fp_compare): Update for rename. Generate unordered
5697 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
5698 (ix86_expand_sse_compare_and_jump): Ditto.
5699 * config/i386/predicates.md (fcmov_comparison_operator):
5700 Remove CCFPU mode handling.
5701 (ix86_comparison_operator): Ditto.
5702 (ix86_carry_flag_operator): Ditto.
5703 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
5704 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
5705 (*cmpu<mode>_cc_i387): Ditto.
5706 (FPCMP): Remove mode iterator.
5707 (unord): Remove mode attribute.
5708 (unord_subst): New define_subst transformation
5709 (unord): New define_subst attribute.
5711 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
5712 (*cmpi<unord>xf_i387): Ditto.
5713 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
5714 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
5715 using unord_subst transformation.
5716 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
5717 (round_saeonly): Also handle CCFP mode.
5718 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
5719 Remove UNSPEC_SAHF unspec handling.
5721 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5723 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
5725 2017-10-27 Jeff Law <law@redhat.com>
5727 * gimple-ssa-sprintf.c: Include domwalk.h.
5728 (class sprintf_dom_walker): New class, derived from dom_walker.
5729 (sprintf_dom_walker::before_dom_children): New function.
5730 (struct call_info): Moved into sprintf_dom_walker class
5731 (compute_formath_length, handle_gimple_call): Likewise.
5732 (sprintf_length::execute): Call the dominator walker rather
5733 than walking the statements.
5735 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
5736 gimple statement locations.
5737 (check_array_bounds): Corresponding changes. Get the statement's
5738 location directly from wi->stmt.
5740 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
5743 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
5745 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5747 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
5748 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
5750 2017-10-27 Jakub Jelinek <jakub@redhat.com>
5753 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
5754 * config/i386/i386.c (maybe_get_pool_constant): Removed.
5755 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
5756 maybe_get_pool_constant.
5757 * config/i386/predicates.md (zero_extended_scalar_load_operand):
5760 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5762 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
5763 2.26 caveat. Update gas and gld versions.
5764 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
5767 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5769 * cgraph.h (set_malloc_flag): Declare.
5770 * cgraph.c (set_malloc_flag_1): New function.
5771 (set_malloc_flag): Likewise.
5772 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
5773 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
5775 (read_ipa_call_summary): Add support for reading is_return_callee.
5776 (write_ipa_call_summary): Stream is_return_callee.
5777 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
5778 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
5779 ipa-prop.h, ipa-fnsummary.h.
5780 (pure_const_names): Change to static.
5781 (malloc_state_e): Define.
5782 (malloc_state_names): Define.
5783 (funct_state_d): Add field malloc_state.
5784 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
5785 (check_retval_uses): New function.
5786 (malloc_candidate_p): Likewise.
5787 (analyze_function): Add support for malloc attribute.
5788 (pure_const_write_summary): Stream malloc_state.
5789 (pure_const_read_summary): Add support for reading malloc_state.
5790 (dump_malloc_lattice): New function.
5791 (propagate_malloc): New function.
5792 (warn_function_malloc): New function.
5793 (ipa_pure_const::execute): Call propagate_malloc and
5794 ipa_free_fn_summary.
5795 (pass_local_pure_const::execute): Add support for malloc attribute.
5796 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
5797 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
5799 2017-10-27 Martin Liska <mliska@suse.cz>
5801 PR gcov-profile/82457
5802 * doc/invoke.texi: Document that one needs a non-strict ISO mode
5803 for fork-like functions to be properly instrumented.
5805 2017-10-27 Richard Biener <rguenther@suse.de>
5808 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
5809 info when we redirected EH.
5811 2017-10-26 Michael Collison <michael.collison@arm.com>
5813 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
5815 (<optab>_trunchf<GPI:mode>2: New pattern.
5816 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
5817 * config/aarch64/iterators.md (wv): New mode attribute.
5818 (vf, VF): New mode attributes.
5819 (vgp, VGP): New mode attributes.
5820 (s): Update attribute with SImode and DImode prefixes.
5822 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
5824 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
5825 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
5826 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
5827 (nios2_option_overide): Initialize it. Don't allow R0-relative
5828 addressing with PIC.
5829 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
5830 (nios2_symbolic_constant_p): Likewise.
5831 (nios2_legitimate_address_p): Likewise.
5832 (nios2_r0rel_section_name_p): New.
5833 (nios2_symbol_ref_in_r0rel_data_p): New.
5834 (nios2_emit_move_sequence): Handle r0rel_constant_p.
5835 (r0rel_constant_p): New.
5836 (nios2_print_operand_address): Handle r0rel_constant_p.
5837 (nios2_cdx_narrow_form_p): Likewise.
5838 * config/nios2/nios2.opt (mr0rel-sec=): New option.
5839 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
5840 (Nios II Options): Document -mr0rel-sec.
5842 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
5844 * config/nios2/nios2.c: Include xregex.h.
5845 (nios2_gprel_sec_regex): New.
5846 (nios2_option_overide): Initialize it. Don't allow GP-relative
5847 addressing with PIC.
5848 (nios2_small_section_name_p): Check for regex match.
5849 * config/nios2/nios2.opt (mgprel-sec=): New option.
5850 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
5851 (Nios II Options): Document -mgprel-sec.
5853 2017-10-26 Jim Wilson <wilson@tuliptree.org>
5855 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
5857 2017-10-26 Tom de Vries <tom@codesourcery.com>
5859 PR tree-optimization/82707
5860 * gimple.c (gimple_copy): Fix unsharing of
5861 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
5863 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
5865 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
5866 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
5867 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
5868 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
5869 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
5870 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
5871 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
5872 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
5873 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
5874 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
5875 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
5876 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
5877 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
5878 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
5879 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
5880 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
5881 _mm512_mask_cmpunord_ps_mask): New intrinsics.
5883 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5885 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
5887 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
5888 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
5889 warning to rs6000.c. Remove the Undocumented flag, since it has
5891 (-mabi=ibmlongdouble): Likewise.
5892 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
5893 already set, set the default format for long double.
5894 (rs6000_debug_reg_global): Print whether long double is IBM or
5896 (rs6000_option_override_internal): Rework setting long double
5897 format. Only warn if the user is changing the long double default
5898 and they did not use -Wno-psabi.
5899 * doc/invoke.texi (PowerPC options): Update the documentation for
5900 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
5902 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5903 Alan Hayward <alan.hayward@arm.com>
5904 David Sherwood <david.sherwood@arm.com>
5906 * rtl.h (wider_subreg_mode): New function.
5907 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
5908 rather than an unsigned int *.
5909 * ira-color.c (regno_max_ref_width): Replace with...
5910 (regno_max_ref_mode): ...this new variable.
5911 (coalesced_pseudo_reg_slot_compare): Update accordingly.
5912 Use wider_subreg_mode.
5913 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
5914 rather than an unsigned int *.
5915 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
5916 (process_alt_operands): Likewise.
5917 (invariant_p): Likewise.
5918 * lra-spills.c (assign_mem_slot): Likewise.
5919 (add_pseudo_to_slot): Likewise.
5920 * lra.c (collect_non_operand_hard_regs): Likewise.
5921 (add_regs_to_insn_regno_info): Likewise.
5922 * reload1.c (regno_max_ref_width): Replace with...
5923 (regno_max_ref_mode): ...this new variable.
5924 (reload): Update accordingly. Update call to
5925 ira_sort_regnos_for_alter_reg.
5926 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
5927 (init_eliminable_invariants): Update to use regno_max_ref_mode.
5928 (scan_paradoxical_subregs): Likewise.
5930 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5932 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
5933 (aarch64_frame): Add emit_frame_chain boolean.
5934 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5935 Move eh_return case to aarch64_layout_frame.
5936 (aarch64_layout_frame): Initialize emit_frame_chain.
5937 (aarch64_expand_prologue): Use emit_frame_chain.
5939 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5941 * config/aarch64/aarch64.c (aarch64_layout_frame):
5942 Ensure LR is always stored at the bottom of the callee-saves.
5943 Remove rarely used frame layout which saves callee-saves at top of
5944 frame, so the store of LR can be used as a valid probe in all cases.
5946 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5948 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
5949 Improve unaligned TImode/TFmode base/offset split.
5951 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5952 Alan Hayward <alan.hayward@arm.com>
5953 David Sherwood <david.sherwood@arm.com>
5955 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
5956 * combine.c (find_single_use_1): Likewise.
5957 (expand_field_assignment): Likewise.
5958 (move_deaths): Likewise.
5959 * lra-constraints.c (simplify_operand_subreg): Likewise.
5960 (curr_insn_transform): Likewise.
5961 * lra.c (collect_non_operand_hard_regs): Likewise.
5962 (add_regs_to_insn_regno_info): Likewise.
5963 * rtlanal.c (reg_referenced_p): Likewise.
5964 (covers_regno_no_parallel_p): Likewise.
5966 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5968 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
5969 Don't print any bits outside the precision of the value.
5970 * wide-int.cc (test_printing): Add some new tests.
5972 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5974 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
5975 supports -xbrace_comment option.
5976 * configure: Regenerate.
5977 * config.in: Regenerate.
5978 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
5979 (ASM_CPU_SPEC): Use it.
5981 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5983 * target.def (static_rtx_alignment): New hook.
5984 * targhooks.h (default_static_rtx_alignment): Declare.
5985 * targhooks.c (default_static_rtx_alignment): New function.
5986 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
5987 * doc/tm.texi: Regenerate.
5988 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
5989 instead of targetm.constant_alignment. Remove call to
5991 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5992 (cris_preferred_mininum_alignment): New function, split out from...
5993 (cris_constant_alignment): ...here.
5994 (cris_static_rtx_alignment): New function.
5995 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
5997 (ix86_constant_alignment): ...here.
5998 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5999 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6000 (mmix_static_rtx_alignment): New function.
6001 * config/spu/spu.c (spu_static_rtx_alignment): New function.
6002 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6004 2017-10-26 Tamar Christina <tamar.christina@arm.com>
6007 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
6008 Add flag_trapping_math and flag_fp_int_builtin_inexact.
6010 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
6012 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
6013 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
6014 mark as a sign-extending load.
6015 (local_pic_load_u): Define.
6017 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
6020 * fold-const.c (operand_equal_for_comparison_p): Also return true
6021 if ARG0 is a simple variant of ARG1 with narrower precision.
6022 (fold_ternary_loc): Always pass unstripped operands to the predicate.
6024 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
6026 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
6028 * i386.h (processor_costs): Add gather_static, gather_per_elt,
6029 scatter_static, scatter_per_elt.
6030 * x86-tune-costs.h: Add new cost entries.
6032 2017-10-25 Richard Biener <rguenther@suse.de>
6034 * tree-ssa-sccvn.h (vn_eliminate): Declare.
6035 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
6036 class pass_fre): Move to ...
6037 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
6038 class pass_fre): ... here and adjust for statistics.
6040 2017-10-25 Jakub Jelinek <jakub@redhat.com>
6043 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
6044 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
6045 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
6046 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
6049 2017-10-25 Richard Biener <rguenther@suse.de>
6051 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
6052 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
6053 eliminate_push_avail, eliminate_insert): Move inside...
6054 (class eliminate_dom_walker): ... this class in preparation
6056 (fini_eliminate): Remove by merging with ...
6057 (eliminate): ... this function. Adjust for class changes.
6058 (pass_pre::execute): Remove fini_eliminate call.
6059 (pass_fre::execute): Likewise.
6061 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6064 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
6065 (VPERMI2, VPERMI2I): New mode iterators.
6066 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
6067 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
6069 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
6070 mode iterator. Remove 3 old define_insn patterns.
6071 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
6072 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
6073 VPERMI2 mode iterator, remove the other two expanders.
6074 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
6075 to use VPERMI2 mode iterator, add another alternative for vpermi2*
6076 instructions, remove the other two patterns.
6077 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
6078 mode iterator, remove the other two patterns.
6079 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
6080 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
6081 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
6082 and adjust argument order accordingly.
6083 (ix86_expand_vec_perm): Adjust caller.
6084 (expand_vec_perm_1): Likewise.
6085 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
6086 (expand_vec_perm_vpermt2_vpshub2): ... this.
6087 (ix86_expand_vec_perm_const_1): Adjust caller.
6088 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
6091 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
6092 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
6093 (vec_shl_<mode>): Remove unused expander.
6094 (avx512bw_<shift_insn><mode>3): New define_insn.
6095 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
6096 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
6098 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
6101 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
6104 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6106 PR rtl-optimization/82396
6107 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
6108 (autopref_multipass_init): Simplify initialization.
6109 (autopref_rank_data): Simplify sort order.
6110 * gcc/sched-int.h (autopref_multipass_data_): Remove
6111 multi_mem_insn_p, min_offset and max_offset.
6113 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6116 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6117 Check special value of flag_omit_frame_pointer.
6118 (aarch64_can_eliminate): Likewise.
6119 (aarch64_override_options_after_change_1): Simplify handling of
6120 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
6122 2017-10-24 Richard Biener <rguenther@suse.de>
6124 PR tree-optimization/82697
6125 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
6126 zero for conditional load and unconditional store.
6128 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6130 * doc/install.texi: Document bootstrap-cet.
6132 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6135 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
6136 ENDBR instruction at function entrance if function is only
6139 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6142 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
6143 patterns to better describe from which operation the CF is computed.
6144 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
6145 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
6146 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
6147 is 0, use _0 suffixed expanders instead of emitting a comparison
6150 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
6152 * config/i386/i386.md(*movsf_internal, *movdf_internal):
6153 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
6155 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
6158 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
6159 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
6160 * loop-iv.c (iv_get_reaching_def): Likewise.
6161 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
6162 variable is promoted and the partition contains undefined values.
6164 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6166 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
6168 (nios2_address_cost): Define.
6169 (nios2_legitimize_address): Recognize (exp + constant) directly.
6170 (TARGET_ADDRESS_COST): Define.
6172 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6174 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
6175 (nios2_symbolic_memory_operand_p): Declare.
6176 (nios2_split_large_constant): Declare.
6177 (nios2_split_symbolic_memory_operand): Declare.
6178 * config/nios2/nios2.c: Adjust includes.
6179 (nios2_symbolic_constant_allowed): New.
6180 (nios2_symbolic_constant_p): New.
6181 (nios2_plus_symbolic_constant_p): New.
6182 (nios2_valid_addr_expr_p): Recognize addresses involving
6184 (nios2_legitimate_address_p): Likewise, also LO_SUM.
6185 (nios2_symbolic_memory_operand_p): New.
6186 (nios2_large_constant_p): New.
6187 (nios2_split_large_constant): New.
6188 (nios2_split_plus_large_constant): New.
6189 (nios2_split_symbolic_memory_operand): New.
6190 (nios2_legitimize_address): Code refactoring. Handle addresses
6191 involving symbolic constants.
6192 (nios2_emit_move_sequence): Likewise.
6193 (nios2_print_operand): Improve error output.
6194 (nios2_print_operand_address): Handle LO_SUM.
6195 (nios2_cdx_narrow_form_p): Likewise.
6196 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
6197 operands involving symbolic constants.
6198 (movhi_internal, movsi_internal): Likewise.
6199 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
6200 (extendhisi2, extendqi<mode>2): Likewise.
6202 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6204 * tree-pass.h (PROP_rtl_split_insns): Define.
6205 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
6207 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6209 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
6211 2017-10-23 Jakub Jelinek <jakub@redhat.com>
6214 * target.def (const_not_ok_for_debug_p): Default to
6215 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
6216 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
6217 * targhooks.c (default_const_not_ok_for_debug_p): New function.
6218 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
6219 which targetm.const_not_ok_for_debug_p returned true.
6220 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
6222 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
6224 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
6225 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
6226 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
6228 (ix86_const_not_ok_for_debug_p): New function.
6229 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
6230 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
6232 2017-10-23 David Malcolm <dmalcolm@redhat.com>
6235 * system.h: Conditionally include "unique-ptr.h" if
6236 INCLUDE_UNIQUE_PTR is defined.
6237 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
6238 of defining INCLUDE_UNIQUE_PTR before including "system.h".
6240 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
6242 * config/rl78/rl78.md: New define_expand "subdi3".
6244 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
6247 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
6248 DF_REF_INSN if DF_REF_INSN_INFO is false.
6250 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
6252 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
6253 xmm_move instead of sse_move.
6254 (sse_store_index): New function.
6255 (ix86_register_move_cost): Be more sensible about mismatch stall;
6256 model AVX moves correctly; make difference between sse->integer and
6258 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
6259 moves; make difference between SSE and AVX.
6260 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
6261 and zmm_move. Increase size of sse load and store tables;
6262 add unaligned load and store tables; add ssemmx_to_integer.
6263 * x86-tune-costs.h: Update all entries according to real
6264 move latencies from Agner Fog's manual and chip documentation.
6266 2017-10-23 Jakub Jelinek <jakub@redhat.com>
6269 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
6270 * config/i386/constraints.md (Wf): New constraint.
6271 * config/i386/i386.md (UNSPEC_SBB): New unspec.
6272 (cmp<dwi>_doubleword): Removed.
6273 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
6274 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
6275 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
6276 expand with cmp<dwi>_doubleword. For LTU and GEU use
6277 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
6279 * common.opt (gcolumn-info): Enable by default.
6280 * doc/invoke.texi (gcolumn-info): Document new default.
6282 2017-10-23 Richard Biener <rguenther@suse.de>
6284 PR tree-optimization/82672
6285 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
6286 Fold the stmt if we propagated into it.
6288 2017-10-23 Richard Biener <rguenther@suse.de>
6290 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
6291 (bitmap_remove_expr_from_set): ... this. All callers call this
6292 for non-constant values.
6293 (bitmap_set_subtract): Rename to...
6294 (bitmap_set_subtract_expressions): ... this. Adjust and
6296 (bitmap_set_contains_value): Remove superfluous check.
6297 (bitmap_set_replace_value): Inline into single caller ...
6298 (bitmap_value_replace_in_set): ... here and simplify.
6299 (dependent_clean): Merge into ...
6300 (clean): ... this using an overload. Adjust.
6301 (prune_clobbered_mems): Adjust.
6302 (compute_antic_aux): Likewise.
6303 (compute_partial_antic_aux): Likewise.
6305 2017-10-23 Richard Biener <rguenther@suse.de>
6307 PR tree-optimization/82129
6309 2017-08-01 Richard Biener <rguenther@suse.de>
6311 PR tree-optimization/81181
6312 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6313 (compute_antic): ... end of iteration here.
6315 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6317 * target.def (starting_frame_offset): New hook.
6318 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
6319 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
6320 * doc/tm.texi.in: Regenerate.
6321 * hooks.h (hook_hwi_void_0): Declare.
6322 * hooks.c (hook_hwi_void_0): New function.
6323 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
6324 STARTING_FRAME_OFFSET.
6325 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
6326 * reload1.c (reload): Likewise.
6327 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
6328 instead of STARTING_FRAME_OFFSET.
6329 * function.c (try_fit_stack_local): Likewise.
6330 (assign_stack_local_1): Likewise
6331 (instantiate_virtual_regs): Likewise.
6332 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
6333 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
6334 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
6335 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
6336 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
6337 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
6338 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
6339 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
6340 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
6341 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
6342 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
6343 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
6344 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
6345 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
6346 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
6347 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
6348 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
6349 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
6350 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
6351 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
6352 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
6353 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
6354 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
6355 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
6356 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
6357 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
6358 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
6359 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
6360 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
6361 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
6362 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
6363 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
6364 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
6365 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
6366 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
6367 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
6368 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
6369 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
6370 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
6371 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
6372 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
6373 * config/avr/avr.c (avr_starting_frame_offset): Make static and
6374 return a HOST_WIDE_INT.
6375 (avr_builtin_setjmp_frame_value): Use it instead of
6376 STARTING_FRAME_OFFSET.
6377 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6378 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
6379 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
6381 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6382 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
6383 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
6384 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6385 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
6386 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
6387 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6388 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
6389 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
6390 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6391 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
6392 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
6394 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6395 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
6396 * config/mips/mips.c (mips_compute_frame_info): Refer to
6397 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6398 (mips_starting_frame_offset): New function.
6399 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6400 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
6401 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
6402 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
6403 and return a HOST_WIDE_INT.
6404 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6405 (mmix_initial_elimination_offset): Refer to
6406 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6407 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
6408 * config/pa/pa.c (pa_starting_frame_offset): New function.
6409 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
6410 (pa_expand_prologue): Likewise.
6411 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6412 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
6413 !FRAME_GROWS_DOWNWARD handling to...
6414 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6415 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
6416 !FRAME_GROWS_DOWNWARD handling to...
6417 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6418 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
6419 !FRAME_GROWS_DOWNWARD handling to...
6420 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6421 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
6423 (rs6000_starting_frame_offset): New function.
6424 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
6425 !FRAME_GROWS_DOWNWARD handling to...
6426 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6427 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
6428 !FRAME_GROWS_DOWNWARD handling to...
6429 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6430 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
6431 !FRAME_GROWS_DOWNWARD handling to...
6432 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6433 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
6434 (rs6000_starting_frame_offset): New function.
6435 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
6436 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
6437 * config/vax/vax.c (vax_starting_frame_offset): New function.
6438 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
6439 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6440 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
6441 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
6442 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6443 * system.h (STARTING_FRAME_OFFSET): Poison.
6445 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6447 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
6448 SCALAR_TYPE_MODE instead of TYPE_MODE.
6450 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6451 Alan Hayward <alan.hayward@arm.com>
6452 David Sherwood <david.sherwood@arm.com>
6454 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
6456 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6457 Alan Hayward <alan.hayward@arm.com>
6458 David Sherwood <david.sherwood@arm.com>
6460 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
6462 2017-10-23 Richard Biener <rguenther@suse.de>
6464 PR tree-optimization/82129
6465 * tree-ssa-pre.c (bitmap_set_and): Remove.
6466 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
6467 canonicalizing expressions in the set to those with lowest
6468 ID rather than taking that from the first edge.
6470 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6472 * combine.c (rtx_equal_for_field_assignment_p): Use
6473 byte_lowpart_offset.
6475 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6476 Alan Hayward <alan.hayward@arm.com>
6477 David Sherwood <david.sherwood@arm.com>
6479 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
6480 to a SUBREG_PROMOTED_VAR.
6482 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6483 Alan Hayward <alan.hayward@arm.com>
6484 David Sherwood <david.sherwood@arm.com>
6486 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
6487 (expand_debug_source_expr): Likewise.
6488 * combine.c (combine_simplify_rtx): Likewise.
6489 * cse.c (fold_rtx): Likewise.
6490 * optabs.c (expand_float): Likewise.
6491 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6492 (simplify_binary_operation_1): Likewise.
6494 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6495 Alan Hayward <alan.hayward@arm.com>
6496 David Sherwood <david.sherwood@arm.com>
6498 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
6499 (record_promoted_value): Likewise.
6500 * expr.c (expand_expr_real_2): Likewise.
6501 * ree.c (update_reg_equal_equiv_notes): Likewise.
6502 (combine_set_extension): Likewise.
6503 * rtlanal.c (low_bitmask_len): Likewise.
6504 * simplify-rtx.c (neg_const_int): Likewise.
6505 (simplify_binary_operation_1): Likewise.
6507 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6508 Alan Hayward <alan.hayward@arm.com>
6509 David Sherwood <david.sherwood@arm.com>
6511 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
6512 * regcprop.c (maybe_mode_change): Likewise.
6513 * reload1.c (alter_reg): Likewise.
6515 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6517 * inchash.h (inchash::hash::add_wide_int): New function.
6518 * lto-streamer-out.c (hash_tree): Use it.
6520 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6522 * inchash.h (inchash::hash::add_wide_int): Rename to...
6523 (inchash::hash::add_hwi): ...this.
6524 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
6525 (polymorphic_call_target_hasher::hash): Likewise.
6526 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
6527 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
6528 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
6529 * lto-streamer-out.c (hash_tree): Likewise.
6530 * optc-save-gen.awk: Likewise.
6531 * tree.c (add_expr): Likewise.
6533 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6536 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
6537 for ordered inequality comparisons even with TARGET_IEEE_FP.
6539 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6542 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
6543 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
6544 Expand with cmp<dwi>_doubleword.
6546 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6548 * extend.texi: Add x86 specific to 'nocf_check' attribute.
6549 List CET intrinsics.
6550 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
6551 specific to -fcf-protection option.
6553 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6555 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
6556 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
6557 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
6558 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
6559 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
6560 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
6561 (extra_objs): Add cet.o for Linux/x86 targets.
6562 (tmake_file): Add i386/t-cet for Linux/x86 targets.
6563 * config/i386/cet.c: New file.
6564 * config/i386/cetintrin.h: Likewise.
6565 * config/i386/t-cet: Likewise.
6566 * config/i386/cpuid.h (bit_SHSTK): New.
6567 (bit_IBT): Likewise.
6568 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
6569 pass IBT and SHSTK bits.
6570 * config/i386/i386-builtin-types.def
6571 (VOID_FTYPE_UNSIGNED_PVOID): New.
6572 (VOID_FTYPE_UINT64_PVOID): Likewise.
6573 * config/i386/i386-builtin.def: Add CET intrinsics.
6574 * config/i386/i386-c.c (ix86_target_macros_internal): Add
6575 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
6576 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
6577 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
6579 * config/i386/i386.c (rest_of_insert_endbranch): New.
6580 (pass_data_insert_endbranch): Likewise.
6581 (pass_insert_endbranch): Likewise.
6582 (make_pass_insert_endbranch): Likewise.
6583 (ix86_notrack_prefixed_insn_p): Likewise.
6584 (ix86_target_string): Add -mibt, -mshstk flags.
6585 (ix86_option_override_internal): Add flag_cf_protection
6587 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
6588 (ix86_print_operand): Add 'notrack' prefix output.
6589 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
6590 (ix86_expand_builtin): Expand CET intrinsics.
6591 (x86_output_mi_thunk): Add 'endbranch' instruction.
6592 * config/i386/i386.h (TARGET_IBT): New.
6593 (TARGET_IBT_P): Likewise.
6594 (TARGET_SHSTK): Likewise.
6595 (TARGET_SHSTK_P): Likewise.
6596 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
6597 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
6598 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
6599 (builtin_setjmp_setup): New pattern.
6600 (builtin_longjmp): Likewise.
6601 (rdssp<mode>): Likewise.
6602 (incssp<mode>): Likewise.
6603 (saveprevssp): Likewise.
6604 (rstorssp): Likewise.
6605 (wrss<mode>): Likewise.
6606 (wruss<mode>): Likewise.
6607 (setssbsy): Likewise.
6608 (clrssbsy): Likewise.
6609 (nop_endbr): Likewise.
6610 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
6612 * config/i386/immintrin.h: Include <cetintrin.h>.
6613 * config/i386/linux-common.h
6614 (file_end_indicate_exec_stack_and_cet): New prototype.
6615 (TARGET_ASM_FILE_END): New.
6617 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
6619 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
6620 latencies instead of having separate table; make difference between
6621 integer and float costs.
6622 * i386.h (processor_costs): Remove scalar_stmt_cost,
6623 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
6624 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
6626 * x86-tune-costs.h: Remove entries which has been removed in
6627 procesor_costs from all tables; make cond_taken_branch_cost
6628 and cond_not_taken_branch_cost COST_N_INSNS based.
6630 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
6632 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
6634 2017-10-20 Jakub Jelinek <jakub@redhat.com>
6636 * config/i386/i386.md (isa): Remove fma_avx512f.
6637 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
6638 <avx512>_fmadd_<mode>_mask3<round_name>,
6639 <avx512>_fmsub_<mode>_mask<round_name>,
6640 <avx512>_fmsub_<mode>_mask3<round_name>,
6641 <avx512>_fnmadd_<mode>_mask<round_name>,
6642 <avx512>_fnmadd_<mode>_mask3<round_name>,
6643 <avx512>_fnmsub_<mode>_mask<round_name>,
6644 <avx512>_fnmsub_<mode>_mask3<round_name>,
6645 <avx512>_fmaddsub_<mode>_mask<round_name>,
6646 <avx512>_fmaddsub_<mode>_mask3<round_name>,
6647 <avx512>_fmsubadd_<mode>_mask<round_name>,
6648 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
6649 (*vec_widen_umult_even_v16si<mask_name>,
6650 *vec_widen_smult_even_v16si<mask_name>): Likewise.
6651 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
6653 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6655 * extend.texi: Add 'nocf_check' documentation.
6656 * gimple.texi: Add second parameter to
6657 gimple_build_call_from_tree.
6658 * invoke.texi: Add -fcf-protection documentation.
6659 * rtl.texi: Add REG_CALL_NOTRACK documenation.
6661 2017-10-20 Richard Biener <rguenther@suse.de>
6663 PR tree-optimization/82473
6664 * tree-vect-loop.c (vectorizable_reduction): Properly get at
6665 the largest input type.
6667 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6669 * c-attribs.c (handle_nocf_check_attribute): New function.
6670 (c_common_attribute_table): Add 'nocf_check' handling.
6671 * gimple-parser.c: Add second argument NULL to
6672 gimple_build_call_from_tree.
6673 * attrib.c (comp_type_attributes): Check nocf_check attribute.
6674 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
6676 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
6677 * common.opt: Add fcf-protection flag.
6678 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
6679 * flag-types.h: Add enum cf_protection_level.
6680 * gimple.c (gimple_build_call_from_tree): Add second parameter.
6681 Add 'nocf_check' attribute propagation to gimple call.
6682 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
6683 (gimple_build_call_from_tree): Update prototype.
6684 (gimple_call_nocf_check_p): New function.
6685 (gimple_call_set_nocf_check): Likewise.
6686 * gimplify.c: Add second argument to gimple_build_call_from_tree.
6687 * ipa-icf.c: Add nocf_check attribute in statement hash.
6688 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
6689 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
6690 * toplev.c (process_options): Add flag_cf_protection handling.
6692 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6694 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
6696 2017-10-20 Richard Biener <rguenther@suse.de>
6698 PR tree-optimization/82603
6699 * tree-if-conv.c (predicate_mem_writes): Make sure to only
6700 remove false predicated stores.
6702 2017-10-20 Richard Biener <rguenther@suse.de>
6704 * graphite-isl-ast-to-gimple.c
6705 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
6706 Remove return value and simplify, dump copied stmt after lhs
6708 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
6709 Reduce dump verbosity.
6710 (gsi_insert_earliest): Likewise.
6711 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
6712 * graphite.c (print_global_statistics): Adjust dumping.
6713 (print_graphite_scop_statistics): Likewise.
6714 (print_graphite_statistics): Do not dump loops here.
6715 (graphite_transform_loops): But here.
6717 2017-10-20 Nicolas Roche <roche@adacore.com>
6719 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
6720 * configure: Regenerate.
6722 2017-10-20 Jakub Jelinek <jakub@redhat.com>
6725 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
6726 functions when optimizing replace GIMPLE_RETURN stmts with
6727 calls to __builtin_unreachable ().
6730 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
6731 for -fsanitize=thread link of executables.
6732 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
6733 link of executables.
6736 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
6738 (<shift_insn><mode>3<mask_name>): Change the last of the 3
6739 define_insns for logical vector shifts to use VI248_AVX512BW
6740 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
6741 condition, useless isa and prefix attributes. Change the first
6742 2 of these define_insns to ...
6743 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
6744 define_insn for avx512vl.
6745 (<shift_insn><mode>3): ... and this, new define_insn without
6746 masking for non-avx512vl.
6749 * config/i386/sse.md (*andnot<mode>3,
6750 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
6751 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
6752 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
6753 not applied use empty suffix even for TARGET_AVX512VL.
6754 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
6755 is applied, supply evex,evex or evex,evex,evex instead of just
6758 2017-10-20 Julia Koval <julia.koval@intel.com>
6760 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
6761 (OPTION_MASK_ISA_GFNI_UNSET): New.
6762 (ix86_handle_option): Handle OPT_mgfni.
6763 * config/i386/cpuid.h (bit_GFNI): New.
6764 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
6765 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
6766 * config/i386/i386.c (ix86_target_string): Add -mgfni.
6767 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
6768 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
6769 * config/i386/i386.opt: Add mgfni.
6771 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
6773 * config/msp430/msp430.c (msp430_option_override): Disable
6774 -fdelete-null-pointer-checks.
6775 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
6777 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6779 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
6780 of x87 and SSE instructions.
6782 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6784 * asan.c (create_cond_insert_point): Do not update edge count.
6785 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
6786 (afdo_propagate_circuit): Likewise.
6787 (afdo_calculate_branch_prob): Likewise.
6788 (afdo_annotate_cfg): Likewise.
6789 * basic-block.h (struct edge_def): Remove count.
6790 (edge_def::count): New accessor.
6791 * bb-reorder.c (rotate_loop): Update.
6792 (find_traces_1_round): Update.
6793 (connect_traces): Update.
6794 (sanitize_hot_paths): Update.
6795 * cfg.c (unchecked_make_edge): Update.
6796 (make_single_succ_edge): Update.
6797 (check_bb_profile): Update.
6798 (dump_edge_info): Update.
6799 (update_bb_profile_for_threading): Update.
6800 (scale_bbs_frequencies_int): Update.
6801 (scale_bbs_frequencies_gcov_type): Update.
6802 (scale_bbs_frequencies_profile_count): Update.
6803 (scale_bbs_frequencies): Update.
6804 * cfganal.c (connect_infinite_loops_to_exit): Update.
6805 * cfgbuild.c (compute_outgoing_frequencies): Update.
6806 (find_many_sub_basic_blocks): Update.
6807 * cfgcleanup.c (try_forward_edges): Update.
6808 (try_crossjump_to_edge): Update
6809 * cfgexpand.c (expand_gimple_cond): Update
6810 (expand_gimple_tailcall): Update
6811 (construct_exit_block): Update
6812 * cfghooks.c (verify_flow_info): Update
6813 (redirect_edge_succ_nodup): Update
6814 (split_edge): Update
6815 (make_forwarder_block): Update
6816 (duplicate_block): Update
6817 (account_profile_record): Update
6818 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6819 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6820 * cfgloopmanip.c (scale_loop_profile): Update.
6822 (lv_adjust_loop_entry_edge): Update.
6823 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
6824 (force_nonfallthru_and_redirect): Update.
6825 (purge_dead_edges): Update.
6826 (rtl_flow_call_edges_add): Update.
6827 * cgraphunit.c (init_lowered_empty_function): Update.
6828 (cgraph_node::expand_thunk): Update.
6829 * gimple-pretty-print.c (dump_probability): Update.
6830 (dump_edge_probability): Update.
6831 * gimple-ssa-isolate-paths.c (isolate_path): Update.
6832 * haifa-sched.c (sched_create_recovery_edges): Update.
6833 * hsa-gen.c (convert_switch_statements): Update.
6834 * ifcvt.c (dead_or_predicable): Update.
6835 * ipa-inline-transform.c (inline_transform): Update.
6836 * ipa-split.c (split_function): Update.
6837 * ipa-utils.c (ipa_merge_profiles): Update.
6838 * loop-doloop.c (add_test): Update.
6839 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
6840 * lto-streamer-in.c (input_cfg): Update.
6841 (input_function): Update.
6842 * lto-streamer-out.c (output_cfg): Update.
6843 * modulo-sched.c (sms_schedule): Update.
6844 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6845 * predict.c (maybe_hot_edge_p): Update.
6846 (unlikely_executed_edge_p): Update.
6847 (probably_never_executed_edge_p): Update.
6848 (dump_prediction): Update.
6849 (drop_profile): Update.
6850 (propagate_unlikely_bbs_forward): Update.
6851 (determine_unlikely_bbs): Update.
6852 (force_edge_cold): Update.
6853 * profile.c (compute_branch_probabilities): Update.
6854 * reg-stack.c (better_edge): Update.
6855 * shrink-wrap.c (handle_simple_exit): Update.
6856 * tracer.c (better_p): Update.
6857 * trans-mem.c (expand_transaction): Update.
6858 (split_bb_make_tm_edge): Update.
6859 * tree-call-cdce.c: Update.
6860 * tree-cfg.c (gimple_find_sub_bbs): Update.
6861 (gimple_split_edge): Update.
6862 (gimple_duplicate_sese_region): Update.
6863 (gimple_duplicate_sese_tail): Update.
6864 (gimple_flow_call_edges_add): Update.
6865 (insert_cond_bb): Update.
6866 (execute_fixup_cfg): Update.
6867 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
6868 * tree-complex.c (expand_complex_div_wide): Update.
6869 * tree-eh.c (lower_resx): Update.
6870 (unsplit_eh): Update.
6871 (cleanup_empty_eh_move_lp): Update.
6872 * tree-inline.c (copy_edges_for_bb): Update.
6873 (freqs_to_counts): Update.
6874 (copy_cfg_body): Update.
6875 * tree-ssa-dce.c (remove_dead_stmt): Update.
6876 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6877 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
6878 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
6879 (unloop_loops): Update.
6880 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6881 * tree-ssa-loop-split.c (connect_loops): Update.
6882 (split_loop): Update.
6883 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6884 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
6885 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
6886 * tree-ssa-reassoc.c (branch_fixup): Update.
6887 * tree-ssa-tail-merge.c (replace_block_by): Update.
6888 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
6889 (compute_path_counts): Update.
6890 (update_profile): Update.
6891 (recompute_probabilities): Update.
6892 (update_joiner_offpath_counts): Update.
6893 (estimated_freqs_path): Update.
6894 (freqs_to_counts_path): Update.
6895 (clear_counts_path): Update.
6896 (ssa_fix_duplicate_block_edges): Update.
6897 (duplicate_thread_path): Update.
6898 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
6899 (case_bit_test_cmp): Update.
6900 (collect_switch_conv_info): Update.
6901 (gen_inbound_check): Update.
6902 (do_jump_if_equal): Update.
6903 (emit_cmp_and_jump_insns): Update.
6904 * tree-tailcall.c (decrease_profile): Update.
6905 (eliminate_tail_call): Update.
6906 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6907 (vect_do_peeling): Update.
6908 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6909 * ubsan.c (ubsan_expand_null_ifn): Update.
6910 (ubsan_expand_ptr_ifn): Update.
6911 * value-prof.c (gimple_divmod_fixed_value): Update.
6912 (gimple_mod_pow2): Update.
6913 (gimple_mod_subtract): Update.
6914 (gimple_ic): Update.
6915 (gimple_stringop_fixed_value): Update.
6917 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6920 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
6922 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
6924 PR rtl-optimization/82395
6925 * ira-color.c (allocno_priority_compare_func): Fix comparison step
6926 based on non_spilled_static_chain_regno_p.
6928 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6930 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
6931 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
6932 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
6934 2017-10-19 Martin Sebor <msebor@redhat.com>
6936 PR tree-optimization/82596
6937 * tree.c (array_at_struct_end_p): Handle STRING_CST.
6939 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
6941 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
6942 (get_mem_refs_of_builtin_call): Likewise.
6943 * builtins.c (expand_builtin_apply): Adjust call to
6944 allocate_dynamic_stack_space.
6945 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
6946 the third argument to allocate_dynamic_stack_space, otherwise -1.
6947 (expand_builtin): Deal with all alloca variants.
6948 (is_inexpensive_builtin): Likewise.
6949 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
6950 * calls.c (special_function_p): Deal with all alloca variants.
6951 (initialize_argument_information): Adjust call to
6952 allocate_dynamic_stack_space.
6953 (expand_call): Likewise.
6954 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
6955 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
6956 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
6957 use it for the stack usage computation.
6958 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
6959 * function.c (gimplify_parameters): Call build_alloca_call_expr.
6960 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
6961 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
6962 (in_loop_p): Remove first argument and useless check.
6963 (pass_walloca::execute): Remove useless test and adjust call to above.
6964 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
6965 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
6966 (gimplify_call_expr): Deal with all alloca variants.
6967 * hsa-gen.c (gen_hsa_alloca): Likewise.
6968 (gen_hsa_insns_for_call): Likewise.
6969 * ipa-pure-const.c (special_builtin_state): Likewise.
6970 * tree-chkp.c (chkp_build_returned_bound): Likewise.
6971 * tree-object-size.c (alloc_object_size): Likewise.
6972 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
6973 (call_may_clobber_ref_p_1): Likewise.
6974 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
6975 (ccp_fold_stmt): Likewise.
6976 (optimize_stack_restore): Likewise.
6977 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
6978 (mark_all_reaching_defs_necessary_1): Likewise.
6979 (propagate_necessity): Likewise.
6980 (eliminate_unnecessary_stmts): Likewise.
6981 * tree.c (build_common_builtin_nodes): Build
6982 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
6983 (build_alloca_call_expr): New function.
6984 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
6985 (CASE_BUILT_IN_ALLOCA): Likewise.
6986 (build_alloca_call_expr): Declare.
6987 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
6989 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
6992 * dwarf2out.c (new_die_raw): New static inline function.
6993 (new_die): Use it to create the DIE.
6994 (add_AT_external_die_ref): Likewise.
6995 (clone_die): Likewise.
6996 (clone_as_declaration): Likewise.
6997 (dwarf2out_vms_debug_main_pointer): Likewise.
6998 (base_type_die): Likewise. Remove early return for corner cases.
6999 Do not call add_pubtype on the DIE here.
7000 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
7001 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
7002 typedefs for base types with DW_AT_endianity. Make sure a DIE with
7003 native order exists for base types, attach the DIE manually and call
7004 add_pubtype on it. Do not equate a reverse order DIE to the type.
7006 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
7008 * config/arm/arm.c (align_ok_ldrd_strd): New function.
7009 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
7011 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
7013 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7015 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
7016 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
7017 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
7018 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
7019 * opts.c (sanitizer_opts): Add builtin.
7020 * ubsan.c (instrument_builtin): New function.
7021 (pass_ubsan::execute): Call it.
7022 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
7023 * doc/invoke.texi: Document -fsanitize=builtin.
7025 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
7026 builtins, store max (log2 (align), 0) into uchar field instead of
7027 align into uptr field.
7028 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
7029 store uchar 0 field instead of uptr 0 field.
7030 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
7031 instead of passing one address of struct with 2 locations pass
7032 two addresses of structs with 1 location each.
7033 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
7034 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
7035 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
7036 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
7037 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
7038 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
7039 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
7040 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
7042 2017-10-19 Martin Liska <mliska@suse.cz>
7045 * file-find.c (remove_prefix): Remove.
7046 * file-find.h (remove_prefix): Likewise.
7047 * gcc-ar.c: Remove smartness of lookup.
7049 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
7051 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
7052 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
7053 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
7055 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7058 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
7059 (setcc + and to xor + setcc): New peephole2.
7061 2017-10-19 Tom de Vries <tom@codesourcery.com>
7063 * doc/sourcebuild.texi (Test Directives, Variants of
7064 dg-require-support): Add dg-require-stack-size.
7066 2017-10-19 Martin Liska <mliska@suse.cz>
7069 * gimplify.c (gimplify_decl_expr): Do not instrument variables
7070 that have a large alignment.
7071 (gimplify_target_expr): Likewise.
7073 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
7075 PR rtl-optimization/82602
7076 * ira.c (rtx_moveable_p): Return false for volatile asm.
7078 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
7081 * config/i386/i386-modes.def (CCGZ): New CC mode.
7082 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
7083 * config/i386/predicates.md (ix86_comparison_operator):
7085 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7086 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
7087 with double-word subtraction.
7088 (put_condition_code): Handle CCGZmode.
7090 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
7092 * wide-int.cc (debug (const wide_int &)): New.
7093 (debug (const wide_int *)): New.
7094 (debug (const widest_int &)): New.
7095 (debug (const widest_int *)): New.
7097 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
7100 * lra-constraints.c (curr_insn_transform): Use non-input operand
7101 instead of output one for matched reload.
7103 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7105 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
7106 (tree-ssa-loop-ivopts.h): New header file.
7107 (struct builtin_info): New fields.
7108 (classify_builtin_1): Compute and record base and offset parts for
7109 memset builtin partition by calling strip_offset.
7110 (offset_cmp, fuse_memset_builtins): New functions.
7111 (finalize_partitions): Fuse adjacent memset partitions by calling
7113 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
7114 Expose the interface.
7115 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
7117 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7119 PR tree-optimization/82574
7120 * tree-loop-distribution.c (find_single_drs): New parameter. Check
7121 that data reference must be executed exactly once per iteration
7122 against the outermost loop in nest.
7123 (classify_partition): Update call to above function.
7125 2017-10-18 Richard Biener <rguenther@suse.de>
7127 PR tree-optimization/82591
7128 * graphite.c (graphite_transform_loops): Move code gen message
7130 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7131 Here. Handle scop_to_isl_ast failing.
7132 (scop_to_isl_ast): Limit the number of ISL operations.
7134 2017-10-18 Richard Biener <rguenther@suse.de>
7136 * graphite-isl-ast-to-gimple.c
7137 (translate_isl_ast_to_gimple::set_rename): Simplify.
7138 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
7139 (graphite_copy_stmts_from_block): ... here.
7140 (copy_bb_and_scalar_dependences): Simplify.
7141 (add_parameters_to_ivs_params): Canonicalize.
7142 (generate_entry_out_of_ssa_copies): Simplify.
7143 * graphite-sese-to-poly.c (extract_affine_name): Simplify
7144 by passing in ISL dimension.
7145 (parameter_index_in_region_1): Rename to ...
7146 (parameter_index_in_region): ... this.
7147 (extract_affine): Adjust assert, pass down parameter index.
7148 (add_param_constraints): Use range-info when available.
7149 (build_scop_context): Adjust.
7150 * sese.c (new_sese_info): Adjust.
7151 (free_sese_info): Likewise.
7152 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
7153 Remove unused typedefs.
7154 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
7156 2017-10-18 Martin Liska <mliska@suse.cz>
7158 * combine.c (simplify_compare_const): Add gcc_fallthrough.
7160 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7162 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
7163 (s390_sched_init): Do not reset s390_sched_state if we entered the
7164 current basic block via a fallthru edge and all others are unlikely.
7166 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7168 * config/s390/s390.c (NUM_SIDES): New variable.
7169 (LONGRUNNING_THRESHOLD): New variable.
7170 (LATENCY_FACTOR): New variable.
7171 (s390_sched_score): Decrease score for long-running instructions on
7173 (s390_sched_variable_issue): Perform bookkeeping for long-running
7176 2017-10-18 Richard Biener <rguenther@suse.de>
7178 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
7179 Simplify with removal of the parameter rename map.
7180 (set_rename): Likewise.
7181 (should_copy_to_new_region): Likewise.
7182 (graphite_copy_stmts_from_block): Likewise.
7183 (copy_bb_and_scalar_dependences): Remove initialization of
7184 unused copied_bb_map.
7186 (copy_internal_parameters): Likewise.
7187 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
7188 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
7189 Use INTEGRAL_TYPE_P.
7190 (parameter_index_in_region_1): Rename to ...
7191 (assign_parameter_index_in_region): ... this. Assert we have
7192 a parameter we handle.
7193 (scan_tree_for_params): Adjust.
7194 * sese.h (parameter_rename_map_t): Remove.
7195 (struct sese_info_t): Remove unused parameter_rename_map and
7196 copied_bb_map members.
7197 * sese.c (new_sese_info): Adjust.
7198 (free_sese_info): Likewise.
7200 2017-10-18 Martin Liska <mliska@suse.cz>
7203 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
7204 on an abnormal edge.
7206 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
7208 * doc/invoke.texi (ffunction-sections and fdata-sections):
7211 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
7213 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
7214 the use statement can throw internally.
7216 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
7218 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
7219 any RTX present on the RHS of a SET.
7220 * compare-elim.c (try_eliminate_compare): Restore comment.
7222 2017-10-17 Jakub Jelinek <jakub@redhat.com>
7224 * langhooks.h (struct lang_hooks): Document that tree_size langhook
7225 may be also called on tcc_type nodes.
7226 * langhooks.c (lhd_tree_size): Likewise.
7228 2017-10-17 David Malcolm <dmalcolm@redhat.com>
7230 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
7231 format_warning_at_substring.
7232 (maybe_warn): Convert source_range * param to a location_t. Pass
7233 UNKNOWN_LOCATION rather than NULL to fmtwarn.
7234 (format_directive): Remove code to extract source_ranges and
7235 source_range * in favor of just a location_t.
7236 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
7238 * substring-locations.c (format_warning_va): Convert
7239 source_range * param to a location_t.
7240 (format_warning_at_substring): Likewise.
7241 * substring-locations.h (format_warning_va): Likewise.
7242 (format_warning_at_substring): Likewise.
7244 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
7246 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
7248 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
7249 and scatter/gather ops.
7251 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
7252 vec_gather_load and vec_scatter_store.
7253 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
7254 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
7255 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
7256 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
7257 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
7258 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
7260 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
7262 * reg-stack.c (compare_for_stack_reg): Add bool argument.
7263 Detect FTST instruction and handle its register pops. Only pop
7264 second operand if can_pop_second_op is true.
7265 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
7266 set can_pop_second_op to false in the compare_for_stack_reg call.
7268 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
7269 output_fp_compare for stack register operands.
7270 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
7271 instructions here. Do not emit stack register pops here. Assert
7272 that FCOMPP pops next to top stack register. Rewrite function.
7274 2017-10-17 Nathan Sidwell <nathan@acm.org>
7277 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
7278 use DECL_ASSEMBLER_NAME_RAW.
7281 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
7284 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
7285 Wilco Dijkstra <wilco.dijkstra@arm.com>
7287 * builtins.c (expand_builtin_update_setjmp_buf): Add a
7288 converstion to Pmode from the buf_addr.
7290 2017-10-17 Richard Biener <rguenther@suse.de>
7292 * graphite-dependences.c (scop_get_reads_and_writes): Change
7293 output parameters to references.
7295 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
7297 PR 71026/tree-optimization
7298 * fold-const.c (distribute_real_division): Removed.
7299 (fold_binary_loc): Remove calls to distribute_real_divison.
7301 2017-10-17 Richard Biener <rguenther@suse.de>
7303 * graphite-scop-detection.c
7304 (scop_detection::stmt_has_simple_data_refs_p): Always use
7305 the full nest as region.
7306 (try_generate_gimple_bb): Likewise.
7307 * sese.c (scalar_evolution_in_region): Simplify now that
7308 SCEV can handle instantiation in regions.
7309 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
7310 in the non-loop part of a function if requested.
7312 2017-10-17 Richard Biener <rguenther@suse.de>
7314 PR tree-optimization/82563
7315 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
7317 (graphite_regenerate_ast_isl): Call it.
7318 * graphite-scop-detection.c (build_scops): Remove entry edge split.
7320 2017-10-17 Jakub Jelinek <jakub@redhat.com>
7322 PR tree-optimization/82549
7323 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
7324 Formatting fixes. Instead of calling make_bit_field_ref with negative
7327 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
7329 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
7330 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
7331 _mm_maskz_reduce_ss): New.
7332 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
7333 __builtin_ia32_reducess_mask): Ditto..
7334 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
7335 * config/i386/sse.md (reduces<mode>): Renamed to ...
7336 (reduces<mode><mask_scalar_name>): ... this.
7337 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
7339 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
7340 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
7342 2017-10-16 David Malcolm <dmalcolm@redhat.com>
7344 * Makefile.in (OBJS): Add unique-ptr-tests.o.
7345 * selftest-run-tests.c (selftest::run_tests): Call
7346 selftest::unique_ptr_tests_cc_tests.
7347 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
7348 * unique-ptr-tests.cc: New file.
7350 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
7353 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7355 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7356 (make_hard_regno_born, make_hard_regno_dead): Update
7357 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
7359 2017-10-16 Jeff Law <law@redhat.com>
7361 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7363 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
7365 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
7367 2017-10-16 Olivier Hainque <hainque@adacore.com>
7369 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
7370 with_cpu if we were configured for an e500v2 target cpu name.
7372 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
7374 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
7375 * doc/invoke.texi: Document +nodsp as a valid extension for
7378 2017-10-16 Martin Liska <mliska@suse.cz>
7380 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
7381 (test_set_range): Likewise.
7382 (test_range_functions): Rename to ...
7383 (test_bit_in_range): ... this.
7384 (sbitmap_c_tests): Add new test.
7386 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7388 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
7390 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
7391 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
7393 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7395 * config/aarch64/aarch64-builtins.c
7396 (aarch64_types_quadopu_lane_qualifiers): New.
7397 (TYPES_QUADOPU_LANE): New.
7398 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
7399 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
7400 (aarch64_<sur>dot_laneq<vsi2qi>): New.
7401 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
7402 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
7403 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
7404 (Vdottype, DOTPROD): New.
7405 (sur): Add SDOT and UDOT.
7407 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7409 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
7410 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
7411 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
7413 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
7414 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
7415 Enable TARGET_DOTPROD.
7416 (cortex-a75.cortex-a55): Likewise.
7417 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
7419 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7421 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
7422 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
7424 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
7426 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
7427 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
7428 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
7429 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
7430 * config/arm/types.md (neon_dot, neon_dot_q): New.
7431 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
7433 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7435 * config/arm/arm.h (TARGET_DOTPROD): New.
7436 * config/arm/arm.c (arm_arch_dotprod): New.
7437 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
7438 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
7439 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
7440 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
7441 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
7442 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
7443 * doc/invoke.texi (armv8.2-a): Document dotprod
7445 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7447 * i386.c (ix86_vec_cost): New function.
7448 (ix86_rtx_costs): Handle vector operations better.
7449 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
7450 * x86-tune-costs.h: Add new costs to all tables.
7452 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7454 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
7456 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
7457 divsd, sqrtss and sqrtsd
7458 * x86-tune-costs.h: Add new entries to all costs.
7459 (znver1_cost): Fix to match real instruction latencies.
7461 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7462 Michael Collison <michael.collison@arm.com>
7464 * compare-elim.c: Include emit-rtl.h.
7465 (can_merge_compare_into_arith): New function.
7466 (try_validate_parallel): Likewise.
7467 (try_merge_compare): Likewise.
7468 (try_eliminate_compare): Call the above when no previous clobber
7470 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
7473 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7477 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
7478 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
7482 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
7483 to be any operand_equal_p operands. For & (B - 1) require
7484 B to be power of 2. Recognize
7485 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
7487 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7490 * optabs.c (expand_memory_blockage): Fix call of
7491 targetm.have_memory_blockage.
7493 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7496 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
7497 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
7498 objects to extra_objs instead of overwriting it.
7500 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7502 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
7503 Use any_fp_register_operand as operand[3] predicate. Simplify
7504 equality test for operands[2] and operands[4] memory location.
7505 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
7506 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
7507 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
7508 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
7509 any_fp_register_operand as operand[1] predicate. Simplify
7510 equality test for operands[0] and operands[3] memory location.
7511 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
7512 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
7513 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
7515 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7517 * target-insns.def: Add memory_blockage.
7518 * optabs.c (expand_memory_blockage): New function.
7519 (expand_asm_memory_barrier): Rename ...
7520 (expand_asm_memory_blockage): ... to this.
7521 (expand_mem_thread_fence): Call expand_memory_blockage
7522 instead of expand_asm_memory_barrier.
7523 (expand_mem_singnal_fence): Ditto.
7524 (expand_atomic_load): Ditto.
7525 (expand_atomic_store): Ditto.
7526 * doc/md.texi (Standard Pattern Names For Generation):
7527 Document memory_blockage instruction pattern.
7529 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
7531 * config/rl78/rl78.c (rl78_emit_libcall): New function.
7532 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
7533 * config/rl78/rl78.md: New define_expand "adddi3".
7535 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7537 * cfghooks.c (verify_flow_info): Disable check that all probabilities
7540 2017-10-13 Jeff Law <law@redhat.com>
7542 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
7544 2017-10-13 Jakub Jelinek <jakub@redhat.com>
7547 * internal-fn.c (expand_mul_overflow): If both operands have
7548 the same highpart of -1 or 0 and the topmost bit of lowpart
7549 is different, overflow is if res <= 0 rather than res < 0.
7551 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
7553 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
7554 TARGET_P9_VECTOR code for unaligned_load case.
7556 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7558 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
7560 2017-10-13 Nathan Sidwell <nathan@acm.org>
7562 * tree-core.h (tree_contains_struct): Make bool.
7563 * tree.c (tree_contains_struct): Likewise.
7564 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
7565 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
7566 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
7567 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
7568 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
7570 2017-10-13 Richard Biener <rguenther@suse.de>
7572 * graphite-isl-ast-to-gimple.c
7573 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
7574 parameters and dominance check.
7575 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
7576 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
7577 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
7578 Do not update SSA form here or do intermediate IL verification.
7579 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
7580 (graphite_initialize): Remove check on the number of loops in
7581 the function and inline into graphite_transform_loops.
7582 (graphite_finalize): Inline into graphite_transform_loops.
7583 (graphite_transform_loops): Perform SSA update and IL verification
7585 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
7587 2017-10-13 Richard Biener <rguenther@suse.de>
7589 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
7590 graphite_expression_type_precision): Avoid global constructor
7592 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
7593 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
7594 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
7595 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
7596 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
7597 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
7599 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
7602 * config/i386/i386.h (ix86_red_zone_size): New.
7603 * config/i386/i386.md (push peephole2s): Replace
7604 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
7606 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7607 Alan Hayward <alan.hayward@arm.com>
7608 David Sherwood <david.sherwood@arm.com>
7610 * combine.c (can_change_dest_mode): Reject changes in
7611 REGMODE_NATURAL_SIZE.
7613 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7614 Alan Hayward <alan.hayward@arm.com>
7615 David Sherwood <david.sherwood@arm.com>
7617 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
7618 (expand_debug_source_expr): Likewise.
7619 * combine.c (combine_simplify_rtx): Likewise.
7620 * cse.c (fold_rtx): Likewise.
7621 * fwprop.c (canonicalize_address): Likewise.
7622 * targhooks.c (default_shift_truncation_mask): Likewise.
7624 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7625 Alan Hayward <alan.hayward@arm.com>
7626 David Sherwood <david.sherwood@arm.com>
7628 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
7629 (widened_mode): Likewise.
7630 (expand_unop): Likewise.
7631 * ree.c (transform_ifelse): Likewise.
7632 (merge_def_and_ext): Likewise.
7633 (combine_reaching_defs): Likewise.
7634 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7636 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7637 Alan Hayward <alan.hayward@arm.com>
7638 David Sherwood <david.sherwood@arm.com>
7640 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
7641 * combine.c (gen_lowpart_for_combine): Likewise.
7642 * dwarf2out.c (rtl_for_decl_location): Likewise.
7643 * final.c (alter_subreg): Likewise.
7644 * rtlhooks.c (gen_lowpart_general): Likewise.
7645 (gen_lowpart_if_possible): Likewise.
7647 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7648 Alan Hayward <alan.hayward@arm.com>
7649 David Sherwood <david.sherwood@arm.com>
7651 * calls.c (expand_call): Use subreg_lowpart_offset.
7652 * cse.c (cse_insn): Likewise.
7653 * regcprop.c (copy_value): Likewise.
7654 (copyprop_hardreg_forward_1): Likewise.
7656 2017-10-13 Jakub Jelinek <jakub@redhat.com>
7659 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
7660 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
7661 =Q constraints to +Q and into insn condition add check
7662 that operands[0] and operands[1] are equal.
7663 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
7664 =Q constraints to +Q and into insn condition add check
7665 that operands[0] is equal to either operands[1] or operands[2].
7668 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
7669 instead of handling MINUS_EXPR twice (once for each argument),
7670 canonicalize operand order and handle just once, use rtype where
7671 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
7674 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
7675 any values of __C while still being pattern recognizable as a simple
7678 2017-10-13 Richard Biener <rguenther@suse.de>
7680 PR tree-optimization/82451
7682 2017-10-02 Richard Biener <rguenther@suse.de>
7684 PR tree-optimization/82355
7685 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7686 a mapping for the enclosing loop but avoid generating one for
7688 (copy_bb_and_scalar_dependences): Remove premature codegen
7689 error on PHIs in blocks duplicated into multiple places.
7690 * graphite-scop-detection.c
7691 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7692 in the region use it as loop and nest to analyze the DR in.
7693 (try_generate_gimple_bb): Likewise.
7694 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7695 (add_loop_constraints): For blocks in a loop not in the region
7696 create a dimension with a single iteration.
7697 * sese.h (gbb_loop_at_index): Remove assert.
7699 * cfgloop.c (loop_preheader_edge): For the loop tree root
7700 return the single successor of the entry block.
7701 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7702 Reset the SCEV hashtable and niters.
7703 * graphite-scop-detection.c
7704 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
7705 assert that we only have POLYNOMIAL_CHREC that vary in loops
7706 contained in the region.
7707 (scop_detection::graphite_can_represent_expr): Adjust.
7708 (scop_detection::stmt_has_simple_data_refs_p): For loops
7709 not in the region set loop to NULL. The nest is now the
7710 entry edge to the region.
7711 (try_generate_gimple_bb): Likewise.
7712 * sese.c (scalar_evolution_in_region): Adjust for
7713 instantiate_scev change.
7714 * tree-data-ref.h (graphite_find_data_references_in_stmt):
7715 Make nest parameter the edge into the region.
7716 (create_data_ref): Likewise.
7717 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
7718 entry edge into a region and adjust instantiate_scev calls.
7719 (create_data_ref): Likewise.
7720 (graphite_find_data_references_in_stmt): Likewise.
7721 (find_data_references_in_stmt): Pass the loop preheader edge
7722 from the nest argument.
7723 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
7724 parameter the edge into the region.
7725 (instantiate_parameters): Use the loop preheader edge as entry.
7726 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
7728 (get_instantiated_value_entry): Make instantiate_below parameter
7729 the edge into the region.
7730 (instantiate_scev_name): Likewise. Adjust dominance checks,
7731 when we cannot use loop-based instantiation instantiate by
7732 walking use-def chains.
7733 (instantiate_scev_poly): Adjust.
7734 (instantiate_scev_binary): Likewise.
7735 (instantiate_scev_convert): Likewise.
7736 (instantiate_scev_not): Likewise.
7737 (instantiate_array_ref): Remove.
7738 (instantiate_scev_3): Likewise.
7739 (instantiate_scev_2): Likewise.
7740 (instantiate_scev_1): Likewise.
7741 (instantiate_scev_r): Do not blindly handle N-operand trees.
7742 Do not instantiate array-refs. Handle all constants and invariants.
7743 (instantiate_scev): Make instantiate_below parameter
7744 the edge into the region.
7745 (resolve_mixers): Use the loop preheader edge for the region
7746 parameter to instantiate_scev_r.
7747 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
7749 2017-10-13 Richard Biener <rguenther@suse.de>
7751 PR tree-optimization/82525
7752 * graphite-isl-ast-to-gimple.c
7753 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
7755 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
7756 Fail code generation when we cannot represent the isl integer.
7757 (binary_op_to_tree): Elide modulo operations that are no-ops
7758 in the type we code generate. Remove now superfluous code
7761 2017-10-13 Richard Biener <rguenther@suse.de>
7763 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
7764 (scop_detection::harmful_loop_in_region): Remove premature
7765 IV type restriction.
7766 (scop_detection::graphite_can_represent_scev): We can handle
7767 pointer IVs just fine.
7769 2017-10-13 Alan Modra <amodra@gmail.com>
7771 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
7772 "Clobbers and Scratch Registers". Add paragraph on
7773 alternative to clobbers for scratch registers and OpenBLAS
7776 2017-10-13 Alan Modra <amodra@gmail.com>
7778 * doc/extend.texi (Clobbers): Correct vax example. Delete old
7779 example of a memory input for a string of known length. Move
7780 commentary out of table. Add a number of new examples
7781 covering array memory inputs.
7783 2017-10-12 Martin Liska <mliska@suse.cz>
7785 PR tree-optimization/82493
7786 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
7787 (test_range_functions): New function.
7788 (sbitmap_c_tests): Likewise.
7789 * selftest-run-tests.c (selftest::run_tests): Run new tests.
7790 * selftest.h (sbitmap_c_tests): New function.
7792 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7794 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7796 * config/rs6000/amo.h: Fix spacing issue.
7798 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7801 * config/i386/i386.md (*ashl<mode>3_mask_1,
7802 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
7803 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
7806 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
7808 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
7809 (profile_probability): Set max_probability
7810 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
7812 * profile-count.c (profile_probability::differs_from_p): Do not
7813 rely on max_probaiblity == 10000
7815 2017-10-12 Jeff Law <law@redhat.com>
7817 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
7820 2017-10-12 Martin Sebor <msebor@redhat.com>
7824 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
7825 (handle_alias_pairs): Call it.
7826 * common.opt (-Wattribute-alias): New option.
7827 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
7828 * doc/invoke.texi (-Wattribute-alias): Document.
7830 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
7833 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
7835 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7837 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7838 (make_hard_regno_born, make_hard_regno_dead): Update
7839 bb_killed_pseudos and bb_gen_pseudos.
7841 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
7843 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
7845 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
7847 * config/alpha/alpha.c (alpha_split_conditional_move):
7848 Use std::swap instead of manually swapping.
7849 (alpha_stdarg_optimize_hook): Ditto.
7850 (alpha_canonicalize_comparison): Ditto.
7852 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7854 * tree-loop-distribution.c (struct builtin_info): New struct.
7855 (struct partition): Refactor fields into struct builtin_info.
7856 (partition_free): Free struct builtin_info.
7857 (build_size_arg_loc, build_addr_arg_loc): Delete.
7858 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
7859 information from struct builtin_info.
7860 (find_single_drs): New function refactored from classify_partition.
7861 Also moved builtin validity checks to this function.
7862 (compute_access_range, alloc_builtin): New functions.
7863 (classify_builtin_st, classify_builtin_ldst): New functions.
7864 (classify_partition): Refactor code into functions find_single_drs,
7865 classify_builtin_st and classify_builtin_ldst.
7866 (distribute_loop): Don't do runtime alias check when distributing
7868 (find_seed_stmts_for_distribution): New function.
7869 (pass_loop_distribution::execute): Refactor code finding seed
7870 stmts into above function. Support distribution for the innermost
7871 two-level loop nest. Adjust dump information.
7873 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7875 * tree-loop-distribution.c: Adjust the general comment.
7876 (NUM_PARTITION_THRESHOLD): New macro.
7877 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
7878 (classify_partition): Skip builtin pattern of loop nest's inner loop.
7879 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
7880 in call to build_partition_graph.
7881 (finalize_partitions): New parameter. Make loop distribution more
7882 conservative by fusing more partitions.
7883 (distribute_loop): Don't do runtime alias check in case of loop nest
7885 (find_seed_stmts_for_distribution): New function.
7886 (prepare_perfect_loop_nest): New function.
7887 (pass_loop_distribution::execute): Refactor code finding seed stmts
7888 and loop nest into above functions. Support loop nest distribution.
7889 Adjust dump information accordingly.
7891 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7893 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
7894 and set PTYPE_SEQUENTIAL for merged partition.
7896 2017-10-12 Richard Biener <rguenther@suse.de>
7898 PR tree-optimization/69728
7900 2017-09-19 Richard Biener <rguenther@suse.de>
7902 PR tree-optimization/69728
7903 * graphite-sese-to-poly.c (schedule_error): New global.
7904 (add_loop_schedule): Handle empty domain by failing the
7906 (build_original_schedule): Handle schedule_error.
7908 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
7909 domain by returning an unchanged schedule.
7911 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7913 * genrecog.c (validate_pattern): For VEC_SELECT verify that
7914 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
7916 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
7918 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
7921 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7924 * expr.c (store_field): Don't optimize away bitsize == 0 store
7925 from CALL_EXPR with addressable return type.
7927 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
7929 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
7930 * config/rs6000/rs6000.md (sel): Delete mode attribute.
7931 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
7932 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
7933 TARGET_ISEL instead of TARGET_ISEL<sel>.
7935 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
7937 * config/rs6000/rs6000.c
7938 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
7940 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
7942 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
7944 (reg_or_cint_operand): Fix comment.
7945 (reg_or_zero_operand): New predicate.
7946 * config/rs6000/rs6000-protos.h (output_isel): Delete.
7947 * config/rs6000/rs6000.c (output_isel): Delete.
7948 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
7949 instead of reg_or_cint_operand. Output instruction directly (not via
7951 (isel_unsigned_<mode>): Ditto.
7952 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
7953 gpc_reg_operand. Add an instruction alternative for this. Output
7954 instruction directly.
7955 (*isel_reversed_unsigned_<mode>): Ditto.
7957 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7959 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
7960 (TARGET_CANONICALIZE_COMPARISON): Define.
7962 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
7965 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
7966 Check whether the dest is REG before adding REG_EQUIV note.
7968 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
7971 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7973 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7974 (make_hard_regno_born, make_hard_regno_dead): Update
7975 bb_killed_pseudos and bb_gen_pseudos.
7977 2017-10-11 Nathan Sidwell <nathan@acm.org>
7979 * incpath.h (enum incpath_kind): Name enum, prefix values.
7980 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
7981 * incpath.c (heads, tails): Use INC_MAX.
7982 (add_env_var_paths, add_standard_paths): Use incpath_kind.
7983 (merge_include_chains, split_quote_chain,
7984 register_include_chains): Update incpath_kind names.
7985 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
7986 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
7988 (add_framework_path, darwin_register_objc_includes): Likewise.
7989 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
7991 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7993 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
7994 Do not use float_operator operator predicate.
7995 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
7996 * config/i386/predicates.md (float_operator): Remove predicate.
7998 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8000 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
8001 (*jccxf_i387): Ditto.
8002 (*jcc<mode>_i387): Ditto.
8003 (*jccu<mode>_i387): Ditto.
8004 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
8005 (*jcc_*_i387 splitters): Remove.
8006 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
8007 * config/i386/i386.c (ix86_split_fp_branch): Remove.
8008 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
8011 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8013 * profile-count.h (slow_safe_scale_64bit): New function.
8014 (safe_scale_64bit): New inline.
8015 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
8016 * profile-count.c: Include wide-int.h
8017 (slow_safe_scale_64bit): New.
8019 2017-10-11 Nathan Sidwell <nathan@acm.org>
8021 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
8022 HAS_DECL_ASSEMBLER_NAME_P.
8023 * gimple-expr.c (gimple_decl_printable_name: Check
8024 HAS_DECL_ASSEMBLER_NAME_P too.
8025 * ipa-utils.h (type_in_anonymous_namespace_p): Check
8026 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
8027 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
8028 * passes.c (rest_of_decl_compilation): Check
8029 HAS_DECL_ASSEMBLER_NAME_P too.
8030 * recog.c (verify_changes): Likewise.
8031 * tree-pretty-print.c (dump_decl_name): Likewise.
8032 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
8034 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
8035 (SET_DECL_ASSEMBLER_NAME): Use it.
8036 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
8037 (COPY_DECL_ASSEMBLER_NAME): Likewise.
8038 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
8040 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8042 * config.gcc (i386, x86_64): Add extra objects.
8043 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
8044 (ix86_min_insn_size): Declare.
8045 (ix86_issue_rate): Declare.
8046 (ix86_adjust_cost): Declare.
8047 (ia32_multipass_dfa_lookahead): Declare.
8048 (ix86_macro_fusion_p): Declare.
8049 (ix86_macro_fusion_pair_p): Declare.
8050 (ix86_bd_has_dispatch): Declare.
8051 (ix86_bd_do_dispatch): Declare.
8052 (ix86_core2i7_init_hooks): Declare.
8053 (ix86_atom_sched_reorder): Declare.
8054 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
8055 (COSTS_N_BYTES): Move to x86-tune-costs.h.
8056 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
8057 (rip_relative_addr_p): Rename to ...
8058 (ix86_rip_relative_addr_p): ... this one; export.
8059 (memory_address_length): Update.
8060 (ix86_issue_rate): Move to x86-tune-sched.c.
8061 (ix86_flags_dependent): Move to x86-tune-sched.c.
8062 (ix86_agi_dependent): Move to x86-tune-sched.c.
8063 (exact_dependency_1): Move to x86-tune-sched.c.
8064 (exact_store_load_dependency): Move to x86-tune-sched.c.
8065 (ix86_adjust_cost): Move to x86-tune-sched.c.
8066 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
8067 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
8068 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
8069 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
8070 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
8071 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
8072 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
8073 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
8074 (min_insn_size): Rename to ...
8075 (ix86_min_insn_size): ... this one; export.
8076 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
8077 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
8078 (core2i7_first_cycle_multipass_backtrack): Move to
8079 x86-tune-sched-core.c.
8080 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
8081 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
8082 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
8083 (ix86_avoid_jump_mispredicts): Update.
8084 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
8085 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
8086 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
8087 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
8088 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
8089 (MAX_INSN): Move to ix86-tune-sched-bd.c.
8090 (MAX_IMM): Move to ix86-tune-sched-bd.c.
8091 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
8092 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
8093 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
8094 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
8095 (MAX_STORE): Move to ix86-tune-sched-bd.c.
8096 (BIG): Move to ix86-tune-sched-bd.c.
8097 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
8098 (enum insn_path): Move to ix86-tune-sched-bd.c.
8099 (get_mem_group): Move to ix86-tune-sched-bd.c.
8100 (is_cmp): Move to ix86-tune-sched-bd.c.
8101 (dispatch_violation): Move to ix86-tune-sched-bd.c.
8102 (is_branch): Move to ix86-tune-sched-bd.c.
8103 (is_prefetch): Move to ix86-tune-sched-bd.c.
8104 (init_window): Move to ix86-tune-sched-bd.c.
8105 (allocate_window): Move to ix86-tune-sched-bd.c.
8106 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
8107 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
8108 (process_end_window): Move to ix86-tune-sched-bd.c.
8109 (allocate_next_window): Move to ix86-tune-sched-bd.c.
8110 (find_constant): Move to ix86-tune-sched-bd.c.
8111 (get_num_immediates): Move to ix86-tune-sched-bd.c.
8112 (has_immediate): Move to ix86-tune-sched-bd.c.
8113 (get_insn_path): Move to ix86-tune-sched-bd.c.
8114 (get_insn_group): Move to ix86-tune-sched-bd.c.
8115 (count_num_restricted): Move to ix86-tune-sched-bd.c.
8116 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
8117 (add_insn_window): Move to ix86-tune-sched-bd.c.
8118 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
8119 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
8120 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
8121 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
8122 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
8123 (do_dispatch): Move to ix86-tune-sched-bd.c.
8124 (has_dispatch): Move to ix86-tune-sched-bd.c.
8125 * config/i386/t-i386: Add new object files.
8126 * config/i386/x86-tune-costs.h: New file.
8127 * config/i386/x86-tune-sched-atom.c: New file.
8128 * config/i386/x86-tune-sched-bd.c: New file.
8129 * config/i386/x86-tune-sched-core.c: New file.
8130 * config/i386/x86-tune-sched.c: New file.
8132 2017-10-11 Liu Hao <lh_mouse@126.com>
8134 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
8135 the generic version below instead.
8136 (should_colorize): Recognize Windows consoles as terminals
8138 * pretty-print.c [__MINGW32__] (write_all): New function.
8139 [__MINGW32__] (find_esc_head): Likewise.
8140 [__MINGW32__] (find_esc_terminator): Likewise.
8141 [__MINGW32__] (eat_esc_sequence): Likewise.
8142 [__MINGW32__] (mingw_ansi_fputs): New function that handles
8144 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
8147 2017-10-11 Richard Biener <rguenther@suse.de>
8149 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
8150 Properly call analyze_scalar_evolution with the loop of the stmt.
8152 2017-10-11 Richard Biener <rguenther@suse.de>
8154 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
8155 * tree-core.h (tree_base): Add chrec_var union member.
8156 * tree.h (CHREC_VAR): Remove.
8157 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
8158 * tree-chrec.h (build_polynomial_chrec): Adjust.
8159 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
8160 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
8162 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
8164 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
8165 * match.pd: ... here.
8166 ((T) X == (T) Y): Relax condition.
8168 2017-10-11 Bin Cheng <bin.cheng@arm.com>
8170 PR tree-optimization/82472
8171 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
8173 (break_alias_scc_partitions): Update postorder number.
8175 2017-10-11 Martin Liska <mliska@suse.cz>
8178 * opts.c (parse_no_sanitize_attribute): Do not use error_value
8180 * opts.h (parse_no_sanitize_attribute): Remove last argument.
8182 2017-10-11 Martin Liska <mliska@suse.cz>
8184 * print-rtl.c (print_insn): Move declaration of idbuf
8185 to same scope as name.
8187 2017-10-11 Martin Liska <mliska@suse.cz>
8192 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8195 2017-10-11 Martin Liska <mliska@suse.cz>
8198 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8201 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
8203 * config.gcc (default_gnu_indirect_function): Default to yes for
8204 arm*-*-linux* with glibc.
8206 2017-10-11 Richard Biener <rguenther@suse.de>
8208 * tree-scalar-evolution.c (get_scalar_evolution): Handle
8209 default-defs and types we do not want to analyze.
8210 (interpret_loop_phi): Replace unreachable code with an assert.
8211 (compute_scalar_evolution_in_loop): Remove and inline ...
8212 (analyze_scalar_evolution_1): ... here, replacing condition with
8213 what makes the intent clearer. Remove handling of cases
8214 get_scalar_evolution now handles.
8216 2017-10-10 Jim Wilson <wilson@tuliptree.org>
8218 PR rtl-optimization/81434
8219 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
8220 comment for main loop. In sched_group_found if, also add checks for
8221 pass and min_cost_group.
8223 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
8225 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
8226 (rs6000_insn_cost): New function.
8227 * config/rs6000/rs6000.md (cost): New attribute.
8229 2017-10-10 Jakub Jelinek <jakub@redhat.com>
8230 H.J. Lu <hongjiu.lu@intel.com>
8234 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
8235 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
8236 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
8237 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
8238 to OPTION_MASK_ISA_AVX512VL - builtins that have both
8239 OPTION_MASK_ISA_MMX and some other bit set require both
8240 mmx and the ISAs without the mmx bit.
8241 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
8242 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
8243 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
8244 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
8245 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
8246 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
8247 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
8248 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
8249 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
8250 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
8251 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
8252 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
8253 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
8254 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
8255 Add OPTION_MASK_ISA_MMX.
8257 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
8259 * config.gcc (armv7*-*-freebsd*): New target.
8260 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
8262 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
8264 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
8265 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
8268 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
8270 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
8271 a template parameter.
8272 (WIDE_INT_REF_FOR): Update accordingly.
8273 * tree.h (wi::int_traits <const_tree>): Delete.
8274 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
8275 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
8276 (wi::tree_to_wide_ref): New typedef.
8277 (wi::to_wide): New function.
8278 * calls.c (get_size_range): Use wi::to_wide when operating on
8280 * cgraph.c (cgraph_node::create_thunk): Likewise.
8281 * config/i386/i386.c (ix86_data_alignment): Likewise.
8282 (ix86_local_alignment): Likewise.
8283 * dbxout.c (stabstr_O): Likewise.
8284 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
8285 * expr.c (const_vector_from_tree): Likewise.
8286 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
8287 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
8288 (fold_negate_expr_1, int_const_binop_1, const_binop)
8289 (fold_convert_const_int_from_real, optimize_bit_field_compare)
8290 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
8291 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
8292 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
8293 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
8294 (fold_not_const, round_up_loc): Likewise.
8295 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
8296 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
8297 (alloca_call_type): Likewise.
8298 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
8299 * godump.c (go_output_typedef): Likewise.
8300 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
8301 * internal-fn.c (get_min_precision): Likewise.
8302 * ipa-cp.c (ipcp_store_vr_results): Likewise.
8303 * ipa-polymorphic-call.c
8304 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
8305 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
8306 (ipa_modify_call_arguments): Likewise.
8307 * match.pd: Likewise.
8308 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
8309 * print-tree.c (print_node_brief, print_node): Likewise.
8310 * stmt.c (expand_case): Likewise.
8311 * stor-layout.c (layout_type): Likewise.
8312 * tree-affine.c (tree_to_aff_combination): Likewise.
8313 * tree-cfg.c (group_case_labels_stmt): Likewise.
8314 * tree-data-ref.c (dr_analyze_indices): Likewise.
8315 (prune_runtime_alias_test_list): Likewise.
8316 * tree-dump.c (dequeue_and_dump): Likewise.
8317 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
8318 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
8319 * tree-pretty-print.c (dump_generic_node): Likewise.
8320 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
8321 (simple_iv_with_niters): Likewise.
8322 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
8323 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
8324 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
8325 * tree-ssa-loop-niter.c (split_to_var_and_offset)
8326 (refine_value_range_using_guard, number_of_iterations_ne_max)
8327 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
8328 (get_cst_init_from_scev, record_nonwrapping_iv)
8329 (scev_var_range_cant_overflow): Likewise.
8330 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
8331 * tree-ssa-pre.c (compute_avail): Likewise.
8332 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
8333 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
8334 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
8335 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
8336 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
8337 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
8338 (dump_case_nodes, try_switch_expansion): Likewise.
8339 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
8340 (vect_do_peeling): Likewise.
8341 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
8342 * tree-vect-stmts.c (vectorizable_load): Likewise.
8343 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
8344 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
8345 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
8346 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
8347 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
8348 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
8349 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
8350 (evrp_dom_walker::before_dom_children): Likewise.
8351 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
8352 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
8353 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
8354 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
8355 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
8356 (get_range_pos_neg): Likewise.
8357 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
8358 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
8359 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
8360 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
8361 * config/avr/avr.c (avr_fold_builtin): Likewise.
8362 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
8363 * config/msp430/msp430.c (msp430_attr): Likewise.
8364 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
8365 * config/powerpcspe/powerpcspe-c.c
8366 (altivec_resolve_overloaded_builtin): Likewise.
8367 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
8368 (rs6000_expand_ternop_builtin): Likewise.
8369 * config/rs6000/rs6000-c.c
8370 (altivec_resolve_overloaded_builtin): Likewise.
8371 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
8372 (rs6000_expand_ternop_builtin): Likewise.
8373 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
8375 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8377 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
8378 when copying loop nest with only one inner loop.
8380 2017-10-10 Richard Biener <rguenther@suse.de>
8382 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
8383 blocks if SCEV is active.
8384 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
8386 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
8387 (scev_initialize): Assert we are not yet initialized.
8389 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8391 * tree-loop-distribution.c (generate_loops_for_partition): Remove
8392 inner loop's exit stmt by making it always exit the loop, otherwise
8393 we would generate an infinite empty loop.
8395 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8397 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
8398 renaming variables in new preheader if it's deleted.
8400 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8402 * tree-loop-distribution.c (struct partition): Remove unused field
8403 loops of the structure.
8404 (partition_alloc, partition_free): Ditto.
8405 (build_rdg_partition_for_vertex): Ditto.
8407 2017-10-09 Jeff Law <law@redhat.com>
8409 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
8410 return type to match prototype and documentation.
8412 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8414 * config/rs6000/rs6000.c (processor_costs): Move to ...
8415 * config/rs6000/rs6000.h: ... here.
8416 (rs6000_cost): Declare.
8418 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
8420 * except.c (setjmp_fn): New global variable.
8421 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
8422 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
8423 if DONT_USE_BUILTIN_SETJMP is defined.
8425 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8427 * target.def (insn_cost): New hook.
8428 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
8429 * doc/tm.texi: Regenerate.
8430 * rtlanal.c (insn_cost): Use the new hook.
8432 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8434 * combine.c (combine_validate_cost): Compute the new insn_cost,
8435 not just pattern_cost.
8436 (try_combine): Adjust comment.
8438 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8440 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
8442 * combine.c (uid_insn_cost): Adjust comment.
8443 (combine_validate_cost): Adjust comment. Use pattern_cost instead
8445 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
8446 * dse.c (find_shift_sequence): Ditto.
8447 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
8448 (bb_valid_for_noce_process_p): Use pattern_cost.
8449 * rtl.h (insn_rtx_cost): Delete.
8450 (pattern_cost): New prototype.
8451 (insn_cost): New prototype.
8452 * rtlanal.c (insn_rtx_cost): Rename to...
8453 (pattern_cost): ... this.
8456 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
8458 * config/i386/i386.md (*jcc_2): Remove insn pattern.
8459 (*jcc<mode>_0_r_i387): Ditto.
8460 (*jccxf_r_i387): Ditto.
8461 (*jcc<mode>_r_i387): Ditto.
8462 (*jccu<mode>_r_i387): Ditto.
8463 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
8464 (*jcc): Rename from *jcc_1.
8466 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8468 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
8469 deferred rescans after the lvx/stvx recombination pre-pass.
8471 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
8473 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
8474 memory operation instruction support.
8475 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
8476 (rs6000-ibm-aix[789]*): Likewise.
8477 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
8478 Document new functions.
8480 2017-10-09 Richard Biener <rguenther@suse.de>
8482 PR tree-optimization/82397
8483 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
8484 equality only for semantically equal trees.
8486 2017-10-09 Richard Biener <rguenther@suse.de>
8488 PR tree-optimization/82449
8489 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
8490 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
8491 allow constant addresses.
8492 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
8495 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8497 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
8500 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8503 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
8506 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8509 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
8511 2017-10-09 Jakub Jelinek <jakub@redhat.com>
8514 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
8515 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
8517 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
8519 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
8520 (WI_BINARY_PREDICATE_RESULT): Likewise.
8521 (wi::binary_traits::operator_result): New type.
8522 (wi::binary_traits::predicate_result): Likewise.
8523 (generic_wide_int::operator~, unary generic_wide_int::operator-)
8524 (generic_wide_int::operator==, generic_wide_int::operator!=)
8525 (generic_wide_int::operator&, generic_wide_int::and_not)
8526 (generic_wide_int::operator|, generic_wide_int::or_not)
8527 (generic_wide_int::operator^, generic_wide_int::operator+
8528 (binary generic_wide_int::operator-, generic_wide_int::operator*):
8530 (operator~, unary operator-, operator==, operator!=, operator&)
8531 (operator|, operator^, operator+, binary operator-, operator*): New
8533 * expr.c (get_inner_reference): Use wi::bit_and_not.
8534 * fold-const.c (fold_binary_loc): Likewise.
8535 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
8536 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
8537 (bit_value_binop): Likewise.
8538 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
8539 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
8540 (extract_range_from_binary_expr_1): Likewise.
8541 (masked_increment): Likewise.
8542 (simplify_bit_ops_using_ranges): Likewise.
8544 2017-10-09 Martin Jambor <mjambor@suse.cz>
8547 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
8548 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
8549 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
8551 (hsa_fixup_mov_insn_type): New function.
8552 (hsa_op_with_type::get_in_type): Use it.
8553 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
8554 immediates in an assert.
8555 (hsa_op_with_type::extend_int_to_32bit): New method.
8556 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
8557 types. Convert to dest type if necessary.
8558 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
8559 (reg_for_gimple_ssa): Pass false as min32int to
8560 hsa_type_for_scalar_tree_type.
8561 (gen_hsa_addr): Fixup type when creating addresable temporary.
8562 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
8563 (gen_hsa_unary_operation): Extend operands and convert to dest type if
8564 necessary. Call hsa_fixup_mov_insn_type.
8565 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
8566 extend operands and convert to dest type if necessary.
8567 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
8568 to dest type if necessary.
8569 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
8570 if conversion nt necessary and size matches.
8571 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
8572 to dest type if necessary.
8573 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
8574 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
8576 (gen_hsa_clrsb): Likewise.
8577 (gen_hsa_ffs): Likewise.
8578 (gen_hsa_divmod): Extend operands and convert to dest type if
8580 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
8582 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8584 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
8585 Remove empty default arguments. Use a brace block as output
8587 (conditional return): Ditto.
8589 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
8590 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
8592 (group_ending_nop): Ditto.
8593 (doloop_end): Ditto.
8594 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
8595 (splitters for those): Ditto.
8597 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8599 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
8600 a conditional jump (and the compare for it) so that pc_rtx is the
8602 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
8603 for the deleted and renamed ctr<mode>_internal[234] patterns.
8604 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
8605 Delete second conditional return pattern.
8606 (ctr<mode>_internal2): Delete this second bdnz pattern.
8607 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
8608 (ctr<mode>_internal4): Delete this second bdz pattern.
8610 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
8612 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
8613 (always_initialized_rtx_for_ssa_name_p): New predicate.
8614 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
8615 (finish_out_of_ssa): Free new field of SA.
8616 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
8617 * tree-ssa-coalesce.c: Include tree-ssa.h.
8618 (get_parm_default_def_partitions): Remove extern keyword.
8619 (get_undefined_value_partitions): New function.
8620 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
8621 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
8623 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
8625 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
8627 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
8629 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
8631 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
8632 for targets that preffer 128bit.
8634 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
8636 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
8638 2017-10-08 Olivier Hainque <hainque@adacore.com>
8640 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
8641 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
8643 (thumb_set_return_address): Likewise.
8645 2017-10-08 Olivier Hainque <hainque@adacore.com>
8647 * common/config/arm/arm-common.c (arm_except_unwind_info):
8648 Handle DWARF2_UNWIND_INFO.
8650 2017-10-07 Michael Collison <michael.collison@arm.com>
8652 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
8655 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
8657 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
8658 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
8659 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
8660 defined, force the creation of a new block for a dispatch label.
8662 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
8664 * invoke.texi (Wsuggest-attribute=cold): Document.
8665 * common.opt (Wsuggest-attribute=cold): New
8666 * ipa-pure-const.c (warn_function_cold): New function.
8667 * predict.c (compute_function_frequency): Use it.
8668 * predict.h (warn_function_cold): Declare.
8670 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
8672 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
8675 2017-10-06 Martin Liska <mliska@suse.cz>
8677 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
8678 keyword for member functions.
8679 (struct sanopt_tree_couple): New struct.
8680 (struct sanopt_tree_couple_hash): New function.
8681 (struct sanopt_ctx): Add new hash_map.
8682 (has_dominating_ubsan_ptr_check): New function.
8683 (record_ubsan_ptr_check_stmt): Likewise.
8684 (maybe_optimize_ubsan_ptr_ifn): Likewise.
8685 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
8686 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
8688 2017-10-06 Sudakshina Das <sudi.das@arm.com>
8691 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
8692 aarch64_simd_valid_immediate on CONST_VECTORs.
8693 (aarch64_reg_or_bic_imm): Likewise.
8695 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
8697 PR rtl-optimization/82396
8698 * haifa-sched.c (ready_sort_real): Disable qsort checking.
8700 2017-10-06 Sebastian Pop <sebpop@gmail.com>
8702 * graphite-dependences.c (scop_get_reads): Move code to...
8703 (scop_get_must_writes): Move code to...
8704 (scop_get_may_writes): Move code to...
8705 (scop_get_reads_and_writes): ... here.
8706 (scop_get_dependences): Call scop_get_reads_and_writes.
8708 2017-10-06 Jakub Jelinek <jakub@redhat.com>
8710 PR tree-optimization/82434
8711 * fold-const.h (can_native_encode_type_p,
8712 can_native_encode_string_p): Remove.
8713 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
8714 don't encode anything, just return what would be otherwise returned.
8715 (native_encode_fixed, native_encode_complex, native_encode_vector):
8717 (native_encode_string): Likewise. Inline by hand
8718 can_native_encode_string_p.
8719 (can_native_encode_type_p): Remove.
8720 (can_native_encode_string_p): Remove.
8721 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
8722 STRING_CSTs using can_native_encode_string_p, test all
8723 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
8724 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
8725 argument from native_encode_expr.
8726 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
8727 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
8730 2017-10-06 Richard Biener <rguenther@suse.de>
8732 PR tree-optimization/82397
8733 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
8734 operand_equal_p but rely on data_ref_compare_tree for detecting
8736 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
8737 to match up with dr_group_sort_cmp.
8739 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8742 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
8744 * config/s390/s390-builtin-types.def: Regenerate.
8746 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8749 * config/s390/s390-builtin-types.def: Regenerate.
8750 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
8751 Change flag from B_VXE to B_VX.
8752 (s390_vec_min_dbl): Remove B_VXE flag.
8754 2017-10-06 Richard Biener <rguenther@suse.de>
8756 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
8757 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
8758 translate_isl_ast_to_gimple::is_valid_rename,
8759 translate_isl_ast_to_gimple::get_rename,
8760 translate_isl_ast_to_gimple::get_def_bb_for_const,
8761 translate_isl_ast_to_gimple::get_new_name,
8762 translate_isl_ast_to_gimple::collect_all_ssa_names,
8763 translate_isl_ast_to_gimple::copy_loop_phi_args,
8764 translate_isl_ast_to_gimple::collect_all_ssa_names,
8765 translate_isl_ast_to_gimple::copy_loop_phi_args,
8766 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
8767 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
8768 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
8769 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
8770 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
8771 translate_isl_ast_to_gimple::copy_cond_phi_args,
8772 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
8773 translate_isl_ast_to_gimple::edge_for_new_close_phis,
8774 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
8775 translate_isl_ast_to_gimple::rename_uses,
8776 translate_isl_ast_to_gimple::rename_all_uses): Remove.
8777 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
8778 (set_rename_for_each_def): Likewise.
8779 (graphite_copy_stmts_from_block): Handle debug stmt resetting
8780 here. Handle rewriting SCEV analyzable uses here.
8781 (copy_bb_and_scalar_dependences): Generate code for PHI
8783 (graphite_regenerate_ast_isl): Adjust.
8784 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
8785 (add_write, add_read): New functions.
8786 (build_cross_bb_scalars_def): Use it and simplify.
8787 (build_cross_bb_scalars_use): Likewise.
8788 (graphite_find_cross_bb_scalar_vars): Inline into...
8789 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
8790 simulating out-of-SSA. Compute liveout and add dependencies.
8791 (build_scops): Force an empty entry block.
8792 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
8794 (sese_build_liveouts): Declare.
8795 (sese_trivially_empty_bb_p): Likewise.
8796 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
8797 compute liveout and debug_liveout.
8798 (sese_bad_liveouts_use): Remove.
8799 (sese_reset_debug_liveouts_bb): Likewise.
8800 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
8801 (sese_build_liveouts): Build liveout and debug_liveout and store
8803 (new_sese_info): Adjust.
8804 (free_sese_info): Likewise.
8805 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
8806 do not build liveout here.
8807 (move_sese_in_condition): Adjust region entry.
8808 (scev_analyzable_p): Match up with chrec_apply requirements.
8809 (sese_trivially_empty_bb_p): New.
8810 * tree-into-ssa.c (get_reaching_def): Properly support generating
8811 default-defs for incremental rewrite of anonymous names.
8813 2017-10-06 Richard Biener <rguenther@suse.de>
8815 * graphite-sese-to-poly.c (extract_affine): For casts increasing
8816 precision do not perform modulo reduction.
8818 2017-10-06 Richard Biener <rguenther@suse.de>
8820 PR tree-optimization/82436
8821 * tree-vect-slp.c (vect_supported_load_permutation_p): More
8822 conservatively choose the vectorization factor when checking
8823 whether we can perform the required load permutation.
8824 (vect_transform_slp_perm_load): Assert when we may not fail.
8826 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
8828 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
8829 message for incompatible -msdata=* and -mcall-* options.
8831 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8833 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
8834 rate for post-reload scheduling.
8836 2017-10-05 Tamar Christina <tamar.christina@arm.com>
8838 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
8840 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8842 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
8843 to improve monte carlo in scimark.
8845 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8847 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
8848 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
8849 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
8850 pentium4_cost, nocona_cost): Set reassociation width to 1.
8851 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
8852 width to 2 for fp operations and 1 otherwise.
8853 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
8855 (atom_cost): Set reassociation width to 2.
8856 (slm_cost, generic_cost): Set fp reassociation width
8857 to 2 and 1 otherwise.
8858 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
8859 (core_cost): Set fp reassociation width to 4 and vector to 2.
8860 (ix86_reassociation_width): Rewrite using cost table; special case
8861 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
8862 and TARGET_AVX128_OPTIMAL.
8863 * config/i386/i386.h (processor_costs): Add
8864 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
8865 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
8866 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
8867 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
8868 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
8869 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
8871 2017-10-05 Nathan Sidwell <nathan@acm.org>
8873 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
8875 2017-10-05 Tamar Christina <tamar.christina@arm.com>
8877 * config/arm/arm.c (arm_test_fpu_data): New.
8878 (arm_run_selftests): Call arm_test_fpu_data.
8880 2017-10-04 Nathan Sidwell <nathan@acm.org>
8882 * toplev.c (toplev::main): Remove excess parens on pretty_printer
8884 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
8886 2017-10-04 Sudakshina Das <sudi.das@arm.com>
8888 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
8889 check type for aarch64_simd_valid_immediate.
8890 (aarch64_output_simd_mov_immediate): Update prototype.
8891 (aarch64_simd_valid_immediate): Update prototype.
8892 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
8893 support for ORR-immediate.
8894 (and<mode>3): modified pattern to add support for BIC-immediate.
8895 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
8896 now checks for valid immediate for BIC and ORR based on new enum
8898 (aarch64_output_simd_mov_immediate): Function now used to output
8899 BIC/ORR imm as well based on new enum argument.
8900 * config/aarch64/constraints.md (Do): New vector immediate constraint.
8902 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
8903 (aarch64_reg_or_bic_imm): Likewise.
8905 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8907 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
8908 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
8910 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
8914 PR rtl-optimization/82396
8915 * haifa-sched.c (autopref_multipass_init): Simplify
8917 (autopref_rank_data): Simplify sort order.
8918 * sched-int.h (autopref_multipass_data_): Remove
8919 multi_mem_insn_p, min_offset and max_offset.
8921 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8923 * doc/sourcebuild.texi: Document vect_peeling_profitable.
8925 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8927 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
8930 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8932 * doc/sourcebuild.texi: Document vect_long_mult.
8934 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
8936 PR tree-optimization/82413
8937 * fold-const.c (build_range_check): Use widest_int when comparing
8938 the maximum ETYPE value with HIGH.
8940 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
8942 PR rtl-optimization/82396
8943 * haifa-sched.c (autopref_multipass_init): Simplify
8945 (autopref_rank_data): Simplify sort order.
8946 * sched-int.h (autopref_multipass_data_): Remove
8947 multi_mem_insn_p, min_offset and max_offset.
8949 2017-10-04 Jakub Jelinek <jakub@redhat.com>
8951 PR tree-optimization/82381
8952 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
8953 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
8956 PR tree-optimization/82374
8957 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
8958 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
8959 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
8960 current_function_decl to the new decl.
8962 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
8964 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
8965 helper macro for IEEE float128 hardware built-in functions.
8966 (SQRTF128_ODD): Add built-in functions with the round-to-odd
8968 (TRUNCF128_ODD): Likewise.
8969 (ADDF128_ODD): Likewise.
8970 (SUBF128_ODD): Likewise.
8971 (MULF128_ODD): Likewise.
8972 (DIVF128_ODD): Likewise.
8973 (FMAF128_ODD): Likewise.
8974 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
8975 UNSPEC_TRUNC_ROUND_TO_ODD.
8976 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
8977 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
8978 floating point round to odd instructions.
8979 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
8980 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
8981 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
8982 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
8983 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
8984 (trunc<mode>sf2_hw): Change the truncate with round to odd
8985 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
8986 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
8987 to odd hardware instructions.
8988 (sub<mode>3_odd): Likewise.
8989 (mul<mode>3_odd): Likewise.
8990 (div<mode>3_odd): Likewise.
8991 (sqrt<mode>2_odd): Likewise.
8992 (fma<mode>4_odd): Likewise.
8993 (fms<mode>4_odd): Likewise.
8994 (nfma<mode>4_odd): Likewise.
8995 (nfms<mode>4_odd): Likewise.
8996 (trunc<mode>df2_odd): Change the truncate with round to odd
8997 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
8999 * doc/extend.texi (PowerPC built-in functions): Update documentation
9000 for existing IEEE float128-bit built-in functions. Add built-in
9001 functions that generate the IEEE 128-bit floating point round to
9004 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
9006 PR rtl-optimization/77729
9007 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
9008 to (X&(C1&~C2))|C2 transformations.
9010 2017-10-03 Martin Jambor <mjambor@suse.cz>
9012 PR tree-optimization/82363
9013 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
9014 mismatch, mark lacc written regardless of racc.
9016 2017-10-03 Jakub Jelinek <jakub@redhat.com>
9018 PR tree-optimization/82381
9019 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
9020 stmt_to_insert nor wheather SSA_NAMEs are default defs.
9021 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
9022 fallthrough into reassoc_stmt_dominates_stmt_p.
9025 * combine.c (combine_instructions): Don't combine in unreachable
9028 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9031 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
9032 function to not use the have_cpu variable. Do not set cpu_index,
9033 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
9035 (rs6000_valid_attribute_p): Remove duplicate initializations of
9036 old_optimize and func_optimize.
9037 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
9038 (rs6000_activate_target_options): Make global.
9039 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
9042 2017-10-02 Jakub Jelinek <jakub@redhat.com>
9044 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
9045 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
9046 Set *poffset to 0 and *psize and *pmax_size to -1 if
9047 *poffset + *psize overflows in HOST_WIDE_INT.
9049 PR tree-optimization/82387
9050 PR tree-optimization/82388
9051 PR tree-optimization/82389
9052 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
9053 instead of live_bytes non-NULL.
9055 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
9058 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
9061 2017-10-02 Richard Biener <rguenther@suse.de>
9063 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
9064 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
9065 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
9067 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
9069 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
9070 requested precision matches the type's.
9071 * calls.c (alloc_max_size): Calculate the new candidate size as
9072 a widest_int and use wi::to_widest when comparing it with the
9073 current candidate size.
9074 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
9075 zero rather than integer_zero_node.
9076 * match.pd: Check for a no-op conversion before using wi::add
9077 rather than after. Use tree_to_uhwi when summing small shift
9078 counts into an unsigned int.
9080 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
9081 Alan Hayward <alan.hayward@arm.com>
9082 David Sherwood <david.sherwood@arm.com>
9085 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
9086 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
9087 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
9088 POINTER_AND_FP_REGS.
9090 2017-10-02 Richard Biener <rguenther@suse.de>
9092 PR tree-optimization/82355
9093 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
9094 a mapping for the enclosing loop but avoid generating one for
9096 (copy_bb_and_scalar_dependences): Remove premature codegen
9097 error on PHIs in blocks duplicated into multiple places.
9098 * graphite-scop-detection.c
9099 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
9100 in the region use it as loop and nest to analyze the DR in.
9101 (try_generate_gimple_bb): Likewise.
9102 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
9103 (add_loop_constraints): For blocks in a loop not in the region
9104 create a dimension with a single iteration.
9105 * sese.h (gbb_loop_at_index): Remove assert.
9107 2017-10-01 Kevin Buettner <kevinb@redhat.com>
9109 * omp-expand.c (adjust_context_scope): New function.
9110 (expand_parallel_call): Call adjust_context_scope.
9112 2017-10-01 Jeff Law <law@redhat.com>
9114 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
9115 dom_opt_dom_walker class with direct access to private members.
9116 Add comments. Call test_for_singularity.
9117 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
9118 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
9119 m_dummy_cond anymore.
9120 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
9122 (pass_dominator:execute): Build the dummy_cond here and pass it
9123 to the dom_opt_dom_walker ctor.
9124 (test_for_singularity): New function.
9126 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
9127 Maya Rashish <coypu@sdf.org>
9129 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
9130 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
9131 (alpha*-*-netbsd*) Use nbsd_tm_file.
9132 (arm*-*-netbsdelf*) Likewise.
9133 (i[34567]86-*-netbsdelf*) Likewise.
9134 (x86_64-*-netbsd*) Likewise.
9135 (mips*-*-netbsd*) Likewise.
9136 (powerpc-*-netbsd*) Likewise.
9137 (sh*-*-netbsd*) Likewise.
9138 (sparc-*-netbsdelf*) Likewise.
9139 (sparc64-*-netbsd*) Likewise.
9140 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
9142 (vax-*-netbsdelf*) Likewise.
9143 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
9144 (UINT_FAST8_TYPE) Likewise.
9145 (INT_FAST16_TYPE) Check CHAR_FAST16.
9146 (UINT_FAST16_TYPE) Likewise.
9148 2017-09-30 Jakub Jelinek <jakub@redhat.com>
9151 * config/i386/i386.md
9152 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
9153 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
9154 *divmodsi4_zext_2): New define_insn_and_split.
9155 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
9156 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
9157 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
9158 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
9159 New define_insn_and_split.
9160 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
9161 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
9162 operands[1] having DImode when mode is SImode.
9164 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
9165 always SImode for DIV and MOD in REG_EQUAL notes.
9167 2017-09-29 Yury Gribov <tetra2005@gmail.com>
9170 * match.pd: Fix handling of NaNs in pattern.
9172 2017-09-29 Jeff Law <law@redhat.com>
9174 * sbitmap.c (bitmap_bit_in_range_p): New function.
9175 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
9176 * tree-ssa-dse.c (live_bytes_read): New function.
9177 (dse_classify_store): Ignore reads of dead bytes.
9179 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
9180 typos and whitespace errors.
9181 * config/i386/predicates.md (address_no_seg_operand): Likewise.
9182 * config/s390/s390.c (s390_emit_prologue): Likewise.
9184 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
9187 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
9188 with a symbol for LRA.
9190 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
9192 PR rtl-optimization/82338
9193 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
9195 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
9197 * genmodes.c (calc_wider_mode): Suppress qsort macro.
9198 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
9199 (qsort_chk): Declare.
9200 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
9201 (qsort_chk): New function.
9203 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9205 PR tree-optimization/82337
9206 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
9207 phi definition if the PHI result appears in an abnormal PHI.
9208 (find_basis_for_base_expr): Don't record a basis if the LHS of the
9209 basis appears in an abnormal PHI.
9211 2017-09-29 Richard Biener <rguenther@suse.de>
9213 * graphite-isl-ast-to-gimple.c
9214 (translate_isl_ast_to_gimple::set_codegen_error): New function.
9215 (binary_op_to_tree): Use it.
9216 (get_rename_from_scev): Likewise.
9217 (copy_loop_phi_nodes): Likewise.
9218 (copy_bb_and_scalar_dependences): Likewise.
9219 (translate_pending_phi_nodes): Likewise.
9221 2017-09-29 Jakub Jelinek <jakub@redhat.com>
9224 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
9225 for movabsq $(i32 << shift), r64.
9227 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
9229 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
9230 index when encoding %esp as %rsp to avoid 0x67 prefix.
9232 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
9234 * config/i386/i386.md (*movsf_internal, *movdf_internal):
9235 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
9237 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
9239 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
9240 Extensions with more than 16 double VFP registers.
9241 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
9242 to_clear_mask and all code related to it. Replace the remaining
9243 entry by a sbitmap and adapt code accordingly.
9245 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
9247 * brig-builtins.def: Change pure attributes to const.
9249 2017-09-28 Joseph Myers <joseph@codesourcery.com>
9251 * config.gcc (default_gnu_indirect_function): Default to yes for
9252 sparc*-*-linux* with glibc.
9254 2017-09-28 Joseph Myers <joseph@codesourcery.com>
9256 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
9257 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
9258 when creating .init_array and .fini_array sections with priority
9261 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
9264 * config/aarch64/aarch64.c
9265 (aarch64_builtin_support_vector_misalignment): Always return false
9266 when misalignment is unknown.
9268 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
9270 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
9271 this function to return false if the definition used by the swap
9272 instruction is artificial, or if the memory address from which the
9273 constant value is loaded is not represented by a base address held
9274 in a register or if the base address register is a frame or stack
9275 pointer. Additionally, return false if the base address of the
9276 loaded constant is a SYMBOL_REF but is not considered to be a
9278 (replace_swapped_load_constant): New function.
9279 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
9280 loaded constant vector with a load of a swapped constant vector.
9282 2017-09-27 Carl Love <cel@us.ibm.com>
9284 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
9285 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
9286 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
9289 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
9291 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
9292 first, always call autopref_rank_data otherwise.
9294 2017-09-27 Richard Biener <rguenther@suse.de>
9296 * graphite-scop-detection.c (find_scop_parameters): Move
9297 loop bound handling ...
9298 (gather_bbs::before_dom_children): ... here, avoiding the need
9299 to build scop_info->loop_nest.
9300 (record_loop_in_sese): Remove.
9301 * sese.h (sese_info_t::loop_nest): Remove.
9302 * sese.c (new_sese_info): Do not allocate loop_nest.
9303 (free_sese_info): Do not free loop_nest.
9305 2017-09-27 Jakub Jelinek <jakub@redhat.com>
9308 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
9309 lhs from calls if the lhs has addressable type.
9311 2017-09-27 Richard Biener <rguenther@suse.de>
9313 * graphite.h (scop::max_alias_set): New member.
9314 * graphite-scop-detection.c: Remove references to non-existing
9315 --param in comments.
9316 (build_alias_sets): Record the maximum alias set used for drs.
9317 (build_scops): Support zero as unlimited for
9318 --param graphite-max-arrays-per-scop.
9319 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
9321 (build_poly_sr_1): ... here. Compute alias set based on the
9322 maximum alias set used for drs rather than
9323 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
9325 2017-09-27 Richard Biener <rguenther@suse.de>
9327 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
9328 --param loop-block-tile-size=0 to disable tiling.
9330 2017-09-27 Richard Biener <rguenther@suse.de>
9332 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
9333 (graphite-max-nb-scop-params): Document special value zero.
9334 * domwalk.h (dom_walker::STOP): New symbolical constant.
9335 (dom_walker::dom_walker): Add optional parameter for bb to
9337 (dom_walker::~dom_walker): Declare.
9338 (dom_walker::before_dom_children): Document STOP return value.
9339 (dom_walker::m_user_bb_to_rpo): New member.
9340 (dom_walker::m_bb_to_rpo): Likewise.
9341 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
9342 mapping here if not provided by the user.
9343 (dom_walker::~dom_walker): Free bb to RPO mapping if not
9344 provided by the user.
9345 (dom_walker::STOP): Define.
9346 (dom_walker::walk): Do not compute bb to RPO mapping here.
9347 Support STOP return value from before_dom_children to stop
9349 * graphite-optimize-isl.c (optimize_isl): If the schedule
9350 is the same still generate code if -fgraphite-identity
9351 or -floop-parallelize-all are given.
9352 * graphite-scop-detection.c: Include cfganal.h.
9353 (gather_bbs::gather_bbs): Get and pass through bb to RPO
9355 (gather_bbs::before_dom_children): Return STOP for BBs
9357 (build_scops): Compute bb to RPO mapping and pass it to
9358 the domwalk. Treat --param graphite-max-nb-scop-params=0
9359 as not limiting the number of params.
9360 * graphite.c (graphite_initialize): Remove limit on the
9361 number of basic-blocks in a function.
9362 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
9363 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
9364 default value of 10.
9366 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9368 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
9369 Adjust code to eliminate needing to do the shift right 32-bits
9370 operation after XSCVDPSPN.
9372 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9374 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
9375 ((X / Y) != 0 -> X >= Y): Likewise.
9377 2017-09-26 Carl Love <cel@us.ibm.com>
9379 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
9380 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
9381 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
9382 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
9383 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
9384 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
9385 definitions and overloading.
9386 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
9387 statement for P9V_BUILTIN_XST_LEN_R.
9388 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
9389 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
9390 define_expand and define_insn for the instructions and builtins.
9391 * doc/extend.texi: Update the built-in documentation file for the new
9393 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
9394 define_insn for the instructions
9396 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
9399 * gcc/config/netbsd-protos.h: New file.
9400 * gcc/config/netbsd.c: New file.
9401 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
9402 * gcc/config/t-netbsd: New file.
9403 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
9404 (tmake_file) Add t-netbsd.
9405 (extra_objs) Add netbsd.o.
9407 2017-09-26 Janus Weil <janus@gcc.gnu.org>
9411 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
9413 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9415 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
9416 sign extension from a vector register to a GPR by doing a 32-bit
9417 direct move and then an EXTSW.
9418 (extendsi<mode>2 splitter): Likewise.
9419 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
9420 right or vector extract after doing XSCVDPSPN. Use
9421 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
9423 (movdi_from_sf_zero_ext): Likewise.
9424 (reload_gpr_from_vsxsf): Likewise.
9425 (p8_mfvsrd_4_disf): Delete, no longer used.
9426 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
9427 then needing to move the SFmode to a GPR to use the XSCVDPSP
9428 instruction instead of FRSP and XSCVDPSPN.
9429 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
9430 it is adjacent to the other XSCVSPDP insns.
9431 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
9432 SFmode to be in traditional Altivec registers.
9433 (vsx_xscvdpspn): Eliminate useless alternative constraint.
9434 (vsx_xscvspdpn): Likewise.
9435 (vsx_xscvspdpn_scalar): Likewise.
9437 2017-09-26 Martin Jambor <mjambor@suse.cz>
9439 * tree-sra.c (compare_access_positions): Put integral types first,
9440 stabilize sorting of integral types, remove conditions putting
9441 non-full-precision integers last.
9442 (sort_and_splice_var_accesses): Disable scalarization if a
9443 non-integert would be represented by a non-full-precision integer.
9445 2017-09-26 Joseph Myers <joseph@codesourcery.com>
9447 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
9448 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
9449 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
9450 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
9451 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
9452 conditionals inside the function instead of around it. Call
9453 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
9454 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
9456 2017-09-26 Richard Biener <rguenther@suse.de>
9458 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
9460 (scop_detection::build_scop_breadth): ... this. Removed.
9461 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
9462 (scop_detection::harmful_stmt_in_bb): Likewise.
9463 (scop_detection::graphite_can_represent_stmt): Likewise.
9464 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
9465 (scop_detection::can_represent_loop): Remove recursion, fold in ...
9466 (scop_detection::can_represent_loop_1): ... this. Removed.
9467 (scop_detection::harmful_loop_in_region): Simplify after inlining
9468 the above and remove more quadraticness.
9469 (build_scops): Adjust.
9470 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
9473 2017-09-26 Jakub Jelinek <jakub@redhat.com>
9476 * config/i386/i386.c (ix86_print_operand_address_as): Only test
9477 REGNO (base) == SP_REG if base is a REG.
9480 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
9481 if it is different SSA_NAME.
9482 (optimize_range_tests_cmp_bitwise): New function.
9483 (optimize_range_tests): Call it.
9485 2017-09-26 Richard Biener <rguenther@suse.de>
9487 PR tree-optimization/82321
9488 * graphite.c (canonicalize_loop_closed_ssa): Properly check
9489 for the def being inside the loop.
9491 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9493 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
9495 * config/s390/s390-builtins.def: Fix constraint on op4.
9497 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9499 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
9500 independent expanders.
9501 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
9502 ("vec_ordered", "vec_unordered"): New expanders.
9504 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9506 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
9509 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9511 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
9512 vec_unpacks_lo_v16qi.
9513 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
9515 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9517 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
9518 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
9519 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
9521 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9523 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
9525 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
9527 ("*vec_slb<mode>"): New insn pattern.
9528 ("vec_shr_<mode>"): New expander.
9529 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
9530 and force the shift count operand to V16QImode.
9531 ("vec_srb<mode>"): Set shift count mode to V16QI.
9533 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9535 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
9536 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
9537 ("vec_widen_smult_hi_<mode>"): New expander definitions.
9539 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
9542 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
9544 2017-09-26 Richard Biener <rguenther@suse.de>
9546 PR tree-optimization/82320
9547 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
9550 2017-09-25 Jeff Law <law@redhat.com>
9552 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
9553 prototype for new argument.
9554 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
9555 mostly extracted from rs6000_emit_allocate_stack.
9556 (rs6000_emit_probe_stack_range_stack_clash): New function.
9557 (rs6000_emit_allocate_stack): Call
9558 rs6000_emit_probe_stack_range_stack_clash as needed.
9559 (rs6000_emit_probe_stack_range): Add additional argument
9560 to call to gen_probe_stack_range{si,di}.
9561 (output_probe_stack_range): New.
9562 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
9563 (output_probe_stack_range_stack_clash): New.
9564 (rs6000_emit_prologue): Emit notes into dump file as requested.
9565 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
9566 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
9567 Add additional operand and pass it to output_probe_stack_range.
9569 2017-09-25 Bin Cheng <bin.cheng@arm.com>
9571 PR tree-optimization/82163
9572 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
9573 (checking_verify_loop_closed_ssa): New parameter.
9574 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
9575 (check_loop_closed_ssa_stmt): Delete.
9576 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
9577 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
9578 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
9581 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
9583 * brig-builtins.def: Treat HSAIL barrier builtins as
9584 setjmp/longjump style functions.
9586 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9588 * target.def (constant_alignment): New hook.
9589 * defaults.h (CONSTANT_ALIGNMENT): Delete.
9590 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
9591 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
9592 * doc/tm.texi: Regenerate.
9593 * targhooks.h (default_constant_alignment): Declare.
9594 (constant_alignment_word_strings): Likewise.
9595 * targhooks.c (default_constant_alignment): New function.
9596 (constant_alignment_word_strings): Likewise.
9597 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
9598 instead of CONSTANT_ALIGNMENT.
9599 * varasm.c (align_variable, get_variable_align, build_constant_desc)
9600 (force_const_mem): Likewise.
9601 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
9602 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
9603 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
9604 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9605 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
9607 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
9608 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9609 constant_alignment_word_strings.
9610 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
9611 (CONSTANT_ALIGNMENT): Likewise.
9612 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9613 (arm_constant_alignment): New function.
9614 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
9615 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9616 constant_alignment_word_strings.
9617 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
9618 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9619 constant_alignment_word_strings.
9620 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
9621 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9622 (cris_constant_alignment): New function.
9623 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
9624 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9625 (epiphany_constant_alignment): New function.
9626 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
9627 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9628 constant_alignment_word_strings.
9629 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
9630 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9631 constant_alignment_word_strings.
9632 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
9633 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9634 constant_alignment_word_strings.
9635 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
9636 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
9637 * config/i386/i386.c (ix86_constant_alignment): Make static.
9638 Use the same interface as the target hook.
9639 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9640 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
9641 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9642 constant_alignment_word_strings.
9643 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
9644 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
9645 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9646 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
9647 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9648 constant_alignment_word_strings.
9649 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
9650 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9651 constant_alignment_word_strings.
9652 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
9653 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9654 constant_alignment_word_strings.
9655 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
9656 * config/microblaze/microblaze.c (microblaze_constant_alignment):
9658 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9659 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
9660 * config/mips/mips.c (mips_constant_alignment): New function.
9661 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9662 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
9663 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
9664 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9665 (mmix_constant_alignment): Make static. Use the same interface
9667 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
9668 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9669 constant_alignment_word_strings.
9670 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
9671 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9672 constant_alignment_word_strings.
9673 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
9674 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9675 constant_alignment_word_strings.
9676 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
9677 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9678 (rs6000_constant_alignment): New function.
9679 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
9680 * config/riscv/riscv.c (riscv_constant_alignment): New function.
9681 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9682 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
9683 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9684 (rs6000_constant_alignment): New function.
9685 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
9686 * config/s390/s390.c (s390_constant_alignment): New function.
9687 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9688 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
9689 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9690 constant_alignment_word_strings.
9691 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
9692 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9693 (sparc_constant_alignment): New function.
9694 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
9695 * config/spu/spu.c (spu_constant_alignment): New function.
9696 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9697 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
9698 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9699 constant_alignment_word_strings.
9700 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
9701 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9702 constant_alignment_word_strings.
9703 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
9704 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9705 constant_alignment_word_strings.
9706 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
9707 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9708 (visium_constant_alignment): New function.
9709 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
9710 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9711 (xtensa_constant_alignment): New function.
9712 * system.h (CONSTANT_ALIGNMENT): Poison.
9714 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
9716 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9717 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
9718 (rs6000_builtin_valid_without_lhs): New helper function.
9719 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9720 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
9722 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9724 * target.h (vec_perm_indices): Use unsigned short rather than
9726 (auto_vec_perm_indices): Likewise.
9727 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
9728 Use unsigned int rather than unsigned char.
9729 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
9731 2017-09-25 Richard Biener <rguenther@suse.de>
9733 * cfgloop.h (sort_sibling_loops): Declare.
9734 * cfgloop.c (sort_sibling_loops_cmp): New helper.
9735 (sort_sibling_loops): New function sorting the sibling loop list
9737 * graphite.c (graphite_transform_loops): Sort sibling loops.
9739 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
9741 * target.def (vec_perm_const_ok): Change sel parameter to
9743 * optabs-query.c (can_vec_perm_p): Update accordingly.
9744 * doc/tm.texi: Regenerate.
9745 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
9746 auto_vec_perm_indices and remove separate nelt field.
9747 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
9748 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
9749 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
9750 (aarch64_expand_vec_perm_const): Update accordingly.
9751 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
9752 to vec_perm_indices.
9753 * config/arm/arm.c (expand_vec_perm_d): Change perm to
9754 auto_vec_perm_indices and remove separate nelt field.
9755 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
9756 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
9757 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
9759 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
9760 to vec_perm_indices.
9761 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
9762 sel to vec_perm_indices.
9763 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
9764 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
9765 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
9767 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
9770 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
9773 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
9774 on the FUNCTION_DECL function context if it has a DIE that is a
9777 2017-09-25 Richard Biener <rguenther@suse.de>
9779 PR tree-optimization/82285
9780 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
9783 2017-09-25 Tom de Vries <tom@codesourcery.com>
9787 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
9790 2017-09-25 Richard Biener <rguenther@suse.de>
9792 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
9793 ISL errors other than isl_error_quota happen. Dump if the
9794 schedule is the same.
9795 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
9796 errors instead of aborting inside ISL.
9798 2017-09-25 Iain Sandoe <iain@codesourcery.com>
9801 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
9802 of libgcc_eh for m64.
9803 * config/i386/darwin64.h: Likewise.
9805 2017-09-25 Richard Biener <rguenther@suse.de>
9808 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
9809 attribute for incomplete types nor twice for complete ones.
9811 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
9814 * config/i386/i386.c (ix86_print_operand_address_as): Encode
9815 %esp as %rsp to avoid 0x67 prefix if there is no index or base
9818 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
9821 * config/i386/i386.opt (mprefer-avx256): Use
9822 ix86_target_flags variable.
9823 * config/i386/i386.c (ix86_target_string): Move
9824 -mprefer-avx256 to flag2_opts.
9826 2017-09-22 Jakub Jelinek <jakub@redhat.com>
9829 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
9830 and x != -1 | y != -1 into (x & y) != -1.
9832 2017-09-22 Steve Ellcey <sellcey@cavium.com>
9834 * config.gcc: Add new case statement to set
9835 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
9836 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
9837 s390x-*-linux* case statements. Added aarch64 to the list of
9838 supported architectures.
9840 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9842 PR tree-optimization/82289
9843 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
9844 STMT_VINFO_RELEVANT_P.
9846 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9847 Alan Hayward <alan.hayward@arm.com>
9848 David Sherwood <david.sherwood@arm.com>
9850 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
9851 for VR_RANGE only; don't allow VR_ANTI_RANGE.
9852 (extract_range_from_binary_expr_1): Don't call
9853 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
9855 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9856 Alan Hayward <alan.hayward@arm.com>
9857 David Sherwood <david.sherwood@arm.com>
9859 * target.def (preferred_vector_alignment): New hook.
9860 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
9862 * doc/tm.texi: Regenerate.
9863 * targhooks.h (default_preferred_vector_alignment): Declare.
9864 * targhooks.c (default_preferred_vector_alignment): New function.
9865 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
9867 (DR_TARGET_ALIGNMENT): New macro.
9868 (aligned_access_p): Update commentary.
9869 (vect_known_alignment_in_bytes): New function.
9870 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
9872 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
9873 Calculate the misalignment based on the target alignment rather than
9875 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
9876 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
9877 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
9878 the target alignment, rather than masking the element misalignment
9879 with the number of elements in a vector. Also use the target
9880 alignment when calculating the maximum number of peels.
9881 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
9882 instead of TYPE_ALIGN_UNIT.
9883 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
9884 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
9885 (vect_create_addr_base_for_vector_ref): Update call accordingly.
9886 (vect_create_data_ref_ptr): Likewise.
9887 (vect_setup_realignment): Realign by ANDing with
9888 -DR_TARGET_MISALIGNMENT.
9889 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
9890 the number of peels based on DR_TARGET_ALIGNMENT.
9891 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
9892 with the guaranteed alignment boundary when deciding whether
9894 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
9895 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
9896 (ensure_base_align): Remove stmt_info parameter. Get the
9897 target base alignment from DR_TARGET_ALIGNMENT.
9898 (vectorizable_store): Update call accordingly. Interpret
9899 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
9901 (vectorizable_load): Likewise.
9903 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9904 Alan Hayward <alan.hayward@arm.com>
9905 David Sherwood <david.sherwood@arm.com>
9907 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
9908 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
9909 (vect_enhance_data_refs_alignment): Likewise.
9911 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9913 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9914 error. Only quit immediately if parsing is complete.
9915 (BEGIN): Initialize fatal_err and parse_done.
9916 (begin fpu, end fpu): Check number of arguments.
9917 (begin arch, end arch): Likewise.
9918 (begin cpu, end cpu): Likewise.
9919 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9920 (optalias): Likewise.
9922 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9924 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9925 * config/arm/arm-isa.h: Delete. Move definitions to ...
9926 * arm-cpus.in: ... here. Use new feature and fgroup values.
9927 * config/arm/arm.c (arm_option_override): Use lower case for feature
9929 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9930 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9931 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9933 (print_isa_bits_for): New function.
9934 (gen_isa): New function.
9935 (gen_comm_data): Use print_isa_bits_for.
9936 (define feature): New keyword.
9937 (define fgroup): New keyword.
9938 * config/arm/t-arm (TM_H): Remove.
9939 (GTM_H): Add arm-isa.h.
9940 (arm-isa.h): Add rule to generate file.
9941 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9942 case for feature bit names.
9944 2017-09-22 Richard Biener <rguenther@suse.de>
9946 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
9948 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
9949 print of no dependency loops ...
9950 * graphite.c (graphite_transform_loops): ... here.
9951 (canonicalize_loop_closed_ssa_form): Work from inner to outer
9953 (same_close_phi_node, remove_duplicate_close_phi,
9954 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
9955 (canonicalize_loop_closed_ssa): ... here and simplify.
9956 * graphite-optimize-isl.c: Include tree-vectorizer.h.
9957 (optimize_isl): Use dump_printf_loc to tell when we stopped
9958 optimizing because of an ISL timeout.
9960 2017-09-22 Richard Biener <rguenther@suse.de>
9962 PR tree-optimization/82291
9963 * tree-if-conv.c (predicate_mem_writes): Make sure to
9964 remove writes in blocks predicated with false.
9966 2017-09-22 Richard Biener <rguenther@suse.de>
9968 * sese.c: Include cfganal.h.
9969 (if_region_set_false_region): Remove.
9970 (create_if_region_on_edge): Likewise.
9971 (move_sese_in_condition): Re-implement without destroying
9974 2017-09-22 Richard Biener <rguenther@suse.de>
9976 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
9977 Verify both BBs contain loop PHI nodes before dispatching to
9979 (graphite_regenerate_ast_isl): Do not recompute dominators,
9980 do not verify three times. Restructure for clarity.
9981 * graphite-scop-detection.c (same_close_phi_node,
9982 remove_duplicate_close_phi, make_close_phi_nodes_unique,
9983 defined_in_loop_p, canonicalize_loop_closed_ssa,
9984 canonicalize_loop_closed_ssa_form): Simplify, remove excess
9985 checking and SSA rewrite, move to ...
9986 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
9987 (graphite_initialize): Do not pass in ctx, do not reset the
9988 SCEV cache, compute only dominators.
9989 (graphite_transform_loops): Allocate ISL ctx after
9990 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
9991 Maintain post-dominators only around build_scops.
9992 * sese.c (if_region_set_false_region): Make static. Free
9993 and recompute dominators.
9994 (move_sese_in_condition): Assert we don't get called with
9995 post-dominators computed.
9996 * sese.h (if_region_set_false_region): Remove.
9998 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
10000 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
10001 mode attribute for TARGET_AVX512VL.
10003 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
10005 * config/i386/i386.opt (mprefer-avx256): New option.
10006 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
10008 (ix86_preferred_simd_mode): Return 256-bit AVX modes
10009 for TARGET_PREFER_AVX256.
10010 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
10012 2017-09-21 Jeff Law <law@redhat.com>
10014 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
10015 Fix dump output if the only stack space is for pushed registers.
10017 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10019 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
10022 2017-09-21 Martin Sebor <msebor@redhat.com>
10025 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
10026 code (in C++) or code that triggers warnings.
10028 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
10030 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
10032 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
10034 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
10035 * sched-rgn.c: Ditto.
10036 * sel-sched-ir.c: Ditto.
10038 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
10040 * toplev.h (set_random_seed): Adjust return type.
10041 * toplev.c (init_local_tick): Move eager initialization of random_seed
10042 to get_random_seed. Adjust comment.
10043 (init_random_seed): Inline to get_random_seed, delete.
10044 (get_random_seed): Initialize random_seed lazily.
10045 (set_random_seed): Do not return previous value.
10046 (print_switch_value): Do not call get_random_seed.
10048 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
10050 * cgraph.c (delete_function_version): New, broken out from...
10051 (cgraph_node::delete_function_version): ...here. Rename to
10052 cgraph_node::delete_function_version_by_decl. Update all uses.
10053 (cgraph_node::remove): Call delete_function_version.
10055 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10058 * tree-inline.c (expand_call_inline): Emit clobber stmts for
10059 VAR_DECLs to which addressable non-volatile parameters are mapped
10060 and for id->retvar after the return value assignment. Clear
10061 id->retval and id->retbnd after inlining.
10063 2017-09-21 Richard Biener <rguenther@suse.de>
10065 PR tree-optimization/82276
10066 PR tree-optimization/82244
10067 * tree-vrp.c (build_assert_expr_for): Set
10068 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
10070 (remove_range_assertions): Revert earlier change.
10072 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
10075 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
10077 2017-09-21 Richard Biener <rguenther@suse.de>
10079 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10080 Restore valid IL after code generation errors.
10081 * graphite.c (graphite_transform_loops): Diagnose code
10082 generation issues as MSG_MISSED_OPTIMIZATION and continue
10083 with processing SCOPs.
10085 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10086 Alan Hayward <alan.hayward@arm.com>
10087 David Sherwood <david.sherwood@arm.com>
10089 * calls.c (compute_argument_addresses): Use simplify_gen_binary
10090 rather than choosing between plus_constant and gen_rtx_<CODE>.
10091 * expr.c (emit_push_insn): Likewise.
10092 (expand_expr_real_2): Likewise.
10094 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10095 Alan Hayward <alan.hayward@arm.com>
10096 David Sherwood <david.sherwood@arm.com>
10098 * loop-unroll.c (split_iv): Call copy_rtx on the step.
10100 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10101 Alan Hayward <alan.hayward@arm.com>
10102 David Sherwood <david.sherwood@arm.com>
10104 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
10105 calling tree_to_uhwi.
10107 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10108 Alan Hayward <alan.hayward@arm.com>
10109 David Sherwood <david.sherwood@arm.com>
10111 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
10112 INTEGER_CST rather than a negative test for ADDR_EXPR.
10114 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10115 Alan Hayward <alan.hayward@arm.com>
10116 David Sherwood <david.sherwood@arm.com>
10118 * tree-vrp.c (extract_range_from_binary_expr_1): Check
10119 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
10121 2017-09-21 Richard Biener <rguenther@suse.de>
10123 PR tree-optimization/71351
10124 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
10125 graphite_create_new_loop_guard): Remove, fold remaining parts
10127 (translate_isl_ast_node_for): ... here and simplify.
10129 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10132 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
10133 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
10134 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
10135 alternative always use QI mode, for -Os imov (=R,R) alternative
10136 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
10139 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10140 Jeff Law <law@redhat.com>
10142 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
10143 (allocate_stack_space): New function, partially extracted from
10144 s390_emit_prologue.
10145 (s390_emit_prologue): Track offset to most recent stack probe.
10146 Code to allocate space moved into allocate_stack_space.
10147 Dump actions when no stack is allocated.
10148 (s390_prologue_plus_offset): New function.
10149 (s390_emit_stack_probe): Likewise.
10151 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
10153 * common.opt (Wa, Wl, Wp, g, gz=): Add
10155 (gno-column-info): Remove.
10156 (gcolumn-info): Drop RejectNegative.
10157 (gno-): New prefix.
10158 (gno-record-gcc-switches): Remove.
10159 (grecord-gcc-switches): Drop RejectNegative.
10160 (gno-split-dwarf): Remove.
10161 (gsplit-dwarf): Drop RejectNegative.
10162 (gno-strict-dwarf): Remove.
10163 (gstrict-dwarf): Drop RejectNegative.
10164 * config/darwin.opt (gfull, gused): Add RejectNegative.
10165 * dwarf2out.c (gen_producer_string): Drop
10166 gno-record-gcc-switches handler.
10167 * optc-gen.awk: Add g to prefixes with negative forms.
10168 * opts-common.c (remapping_prefix_p): New.
10169 (find_opt): Check it.
10170 (generate_canonical_option): Test g prefix.
10171 (option_map): Add -gno- mapping.
10172 (add_misspelling_candidates): Check remapping_prefix_p.
10174 2017-09-20 Jeff Law <law@redhat.com>
10176 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
10177 thinko in stack clash protection support.
10179 * explow.c (compute_stack_clash_protection_loop_data): Use
10180 CONST_INT_P instead of explicit test. Verify object is a
10181 CONST_INT_P before looking at INTVAL.
10182 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
10183 instead of explicit test.
10185 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
10188 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
10189 address instead of to r1 and r11.
10191 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
10193 * config.gcc: Support "knm".
10194 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
10195 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
10197 * config/i386/i386.c (m_KNM): Define.
10198 (processor_target_table): Add "knm".
10200 (ix86_option_override_internal): Add "knm".
10201 (ix86_issue_rate): Add PROCESSOR_KNM.
10202 (ix86_adjust_cost): Ditto.
10203 (ia32_multipass_dfa_lookahead): Ditto.
10204 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
10205 (fold_builtin_cpu): Add M_INTEL_KNM.
10206 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
10207 (processor_type): Add PROCESSOR_KNM.
10208 * config/i386/x86-tune.def: Add m_KNM.
10209 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
10211 2017-09-20 Richard Biener <rguenther@suse.de>
10213 PR tree-optimization/80213
10214 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
10215 are allowed in empty BBs as well.
10216 (canonicalize_loop_closed_ssa): Also look for other complex
10218 (scop_detection::get_sese): Include the loop-closed PHI block
10220 (scop_detection::merge_sese): Remove code adding extra blocks.
10221 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
10222 (build_scops): Assert the final returned scop is invalid.
10224 2017-09-20 Richard Biener <rguenther@suse.de>
10226 PR tree-optimization/82264
10227 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
10229 (vn_phi_lookup): Likewise.
10230 (vn_phi_insert): Likewise.
10232 2017-09-20 Jakub Jelinek <jakub@redhat.com>
10234 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
10235 that fits into uhwi or shwi, add DW_AT_const_value regardless
10236 of early_dwarf without going through RTL, using add_AT_unsigned
10239 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
10240 (DEBUG_LTO_ABBREV_SECTION): Likewise.
10241 (DEBUG_LTO_MACINFO_SECTION): Likewise.
10242 (DEBUG_MACRO_SECTION): Likewise.
10243 (DEBUG_LTO_MACRO_SECTION): Likewise.
10244 (DEBUG_STR_DWO_SECTION): Likewise.
10245 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
10246 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
10247 (DEBUG_LTO_DWO_LINE_SECTION): Define.
10248 (DEBUG_LTO_LINE_STR_SECTION): Define.
10249 (init_sections_and_labels): Initialize debug_line_str_section
10250 variable. Initialize debug_loc_section for -gdwarf-5 to
10251 DEBUG_LOCLISTS_SECTION. Formatting fixes.
10253 2017-09-20 Richard Biener <rguenther@suse.de>
10255 * graphite-sese-to-poly.c (extract_affine): Properly handle
10256 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
10258 2017-09-20 Richard Biener <rguenther@suse.de>
10260 PR tree-optimization/81373
10261 * graphite-scop-detection.c (build_cross_bb_scalars_def):
10262 Force SESE live-out defs to be handled even if they are
10265 2017-09-19 Jeff Law <law@redhat.com>
10267 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
10268 nothing for stack adjustments with REG_STACK_CHECK.
10269 * sched-deps.c (parse_add_or_inc): Reject insns with
10270 REG_STACK_CHECK from dependency breaking.
10271 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
10272 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
10273 * reg-notes.def (STACK_CHECK): New note.
10275 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
10276 (ix86_expand_prologue): Dump stack clash info as needed.
10277 Call ix86_adjust_stack_and_probe_stack_clash as needed.
10279 * function.c (dump_stack_clash_frame_info): New function.
10280 * function.h (dump_stack_clash_frame_info): Prototype.
10281 (enum stack_clash_probes): New enum.
10283 * config/alpha/alpha.c (alpha_expand_prologue): Also check
10284 flag_stack_clash_protection.
10285 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
10286 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
10287 (arm_frame_pointer_required): Likewise.
10288 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
10289 (ia64_expand_prologue): Likewise.
10290 * config/mips/mips.c (mips_expand_prologue): Likewise.
10291 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
10292 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
10293 (sparc_flat_expand_prologue): Likewise.
10294 * config/spu/spu.c (spu_expand_prologue): Likewise.
10296 * explow.c: Include "params.h".
10297 (anti_adjust_stack_and_probe_stack_clash): New function.
10298 (get_stack_check_protect): Likewise.
10299 (compute_stack_clash_protection_loop_data): Likewise.
10300 (emit_stack_clash_protection_loop_start): Likewise.
10301 (emit_stack_clash_protection_loop_end): Likewise.
10302 (allocate_dynamic_stack_space): Use get_stack_check_protect.
10303 Use anti_adjust_stack_and_probe_stack_clash.
10304 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
10305 (emit_stack_clash_protection_loop_start): Likewise.
10306 (emit_stack_clash_protection_loop_end): Likewise.
10307 * rtl.h (get_stack_check_protect): Prototype.
10308 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
10309 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
10310 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
10312 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
10314 * doc/tm.texi: Rebuilt.
10315 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
10316 get_stack_check_protect.
10317 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
10318 * config/arm/arm.c (arm_expand_prologue): Likewise.
10319 (arm_frame_pointer_required): Likewise.
10320 * config/i386/i386.c (ix86_expand_prologue): Likewise.
10321 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
10322 * config/mips/mips.c (mips_expand_prologue): Likewise.
10323 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
10324 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
10325 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
10326 (sparc_flat_expand_prologue): Likewise.
10328 * common.opt (-fstack-clash-protection): New option.
10329 * flag-types.h (enum stack_check_type): Note difference between
10330 -fstack-check= and -fstack-clash-protection.
10331 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
10332 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
10333 * toplev.c (process_options): Issue warnings/errors for cases
10334 not handled with -fstack-clash-protection.
10335 * doc/invoke.texi (-fstack-clash-protection): Document new option.
10336 (-fstack-check): Note additional problem with -fstack-check=generic.
10337 Note that -fstack-check is primarily for Ada and refer users
10338 to -fstack-clash-protection for stack-clash-protection.
10339 Document new params for stack clash protection.
10341 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
10343 * config/i386/i386.c (ix86_split_long_move): Do not handle
10344 address used for LEA in a special way.
10346 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
10348 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
10350 2017-09-19 Martin Sebor <msebor@redhat.com>
10353 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
10354 of incompatible types.
10356 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
10358 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10359 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
10360 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10361 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
10363 2017-09-19 Richard Biener <rguenther@suse.de>
10365 PR tree-optimization/82244
10366 * tree-vrp.c (remove_range_assertions): Do not propagate
10367 a constant to abnormals but replace the assert with a copy.
10369 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
10371 PR rtl-optimization/57878
10372 PR rtl-optimization/68988
10373 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
10374 avoidance test involving non_reload_pseudos. Move frequency test
10375 below the general fragmentation avoidance test.
10377 2017-09-19 Richard Biener <rguenther@suse.de>
10379 PR tree-optimization/69728
10380 * graphite-sese-to-poly.c (schedule_error): New global.
10381 (add_loop_schedule): Handle empty domain by failing the
10383 (build_original_schedule): Handle schedule_error.
10385 2017-09-19 Richard Biener <rguenther@suse.de>
10387 * graphite-scop-detection.c (scop_detection::can_represent_loop):
10388 Do not iterate to sibling loops but only to siblings of inner
10391 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
10394 * config/m68k/m68k.md (moveq feeding equality comparison): Check
10395 that the registers are different.
10397 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
10399 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
10400 to processor_model and "amdfam17h" to arch_names_table.
10401 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
10403 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10406 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
10408 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10409 Alan Hayward <alan.hayward@arm.com>
10410 David Sherwood <david.sherwood@arm.com>
10412 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
10414 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
10416 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
10417 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
10418 vect_schedule_slp_instance.
10419 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
10420 Update call to vect_slp_analyze_node_operations. Simplify return
10422 (vect_slp_analyze_bb_1): Update call accordingly.
10423 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
10424 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
10425 (vect_schedule_slp): Update call accordingly.
10427 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10428 Alan Hayward <alan.hayward@arm.com>
10429 David Sherwood <david.sherwood@arm.com>
10431 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
10432 with types that aren't in fact scalar.
10434 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10436 * tree-vect-slp.c (vect_record_max_nunits): New function,
10438 (vect_build_slp_tree_1): ...here.
10439 (vect_build_slp_tree_2): Call it for phis too.
10441 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10443 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
10444 to vect_get_vec_def_for_operand when getting the mask operand.
10446 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10447 Alan Hayward <alan.hayward@arm.com>
10448 David Sherwood <david.sherwood@arm.com>
10450 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
10453 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10454 Alan Hayward <alan.hayward@arm.com>
10455 David Sherwood <david.sherwood@arm.com>
10457 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
10458 calculation for vector booleans.
10460 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10461 Alan Hayward <alan.hayward@arm.com>
10462 David Sherwood <david.sherwood@arm.com>
10464 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
10466 (vect_transform_stmt): ...here.
10467 (vect_analyze_stmt): Use it instead of calling
10468 vectorizable_live_operation directly.
10470 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
10472 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
10473 non-SIMT targets in acc vector loops.
10475 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
10477 * configure.ac: Add arc and check if assembler supports gdwarf2.
10478 * configure: Regenerate.
10480 2017-09-18 Richard Biener <rguenther@suse.de>
10482 PR tree-optimization/82220
10483 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
10484 epilogue niters from the min_profitable_iters compute.
10486 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10489 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
10490 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
10491 (ix86_init_pic_reg): Revert 2017-09-01 changes.
10493 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
10496 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
10497 switching to a new text section.
10499 2017-09-18 Richard Biener <rguenther@suse.de>
10501 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
10503 (build_alias_set): Reject aliases with no access function.
10505 2017-09-18 Richard Biener <rguenther@suse.de>
10507 PR tree-optimization/79622
10508 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
10510 (build_cross_bb_scalars_use): Likewise.
10512 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
10514 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
10516 2017-09-18 Alan Modra <amodra@gmail.com>
10519 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
10520 stack_pointer_rtx for count 0. Update comments. Break up
10521 large rtl expression.
10523 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
10525 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
10526 Increase to 20 bytes.
10527 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
10528 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
10529 or avx version of the stub.
10531 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
10534 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
10535 compute the minimum stack alignment. Also update preferred stack
10536 boundary for leaf functions.
10538 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
10540 PR tree-optimization/82228
10541 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
10544 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10546 * common/config/nds32/nds32-common.c
10547 (nds32_option_optimization_table): Refine formatting.
10548 (nds32_option_optimization_table): Use -fsched-pressure and
10549 -fomit-frame-pointer for specific optimization level.
10551 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10553 * config/nds32/nds32.c: Refine formatting and comments.
10554 * config/nds32/nds32.h: Likewise.
10555 * config/nds32/nds32.md: Likewise.
10556 * config/nds32/nds32-cost.c: Likewise.
10557 * config/nds32/nds32-isr.c: Likewise.
10558 * config/nds32/nds32-md-auxiliary.c: Likewise.
10559 * config/nds32/nds32-multiple.md: Likewise.
10560 * config/nds32/nds32-predicates.c: Likewise.
10562 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
10563 Jakub Jelinek <jakub@redhat.com>
10565 Add support for -std=c++2a.
10566 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
10568 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
10570 2017-09-15 Steve Ellcey <sellcey@cavium.com>
10573 * doc/extend.texi (Common Function Attributes): Add
10574 references to ARM, AArch64, and S/390 specific attributes.
10575 (Function Specific Option Pragmas): Add AArch64 and S/390
10576 to list of back ends that support the target pragma.
10578 2017-09-15 Nathan Sidwell <nathan@acm.org>
10580 * doc/standards.texi: Fix C++17 description. Update URLs for
10583 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
10585 * common.opt (Wcast-align=strict): New warning option.
10586 * doc/invoke.texi: Document -Wcast-align=strict.
10588 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
10590 * cgraph.h (cgraph_thunk_info): Add comments.
10591 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
10592 assert for VIRTUAL_* arguments stricter.
10594 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
10596 PR tree-optimization/71026
10597 * match.pd: Move RDIV patterns from fold-const.c
10598 * fold-const.c (distribute_real_division): Removed.
10599 (fold_binary_loc): Remove calls to distribute_real_divison.
10601 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10603 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
10604 c++1z and gnu++1z as deprecated. Change other references to
10605 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
10606 Change -Wc++1z-compat to -Wc++17-compat.
10607 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
10608 * dwarf2out.c (highest_c_language): Handle C++17.
10609 (gen_compile_unit_die): Likewise.
10611 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10613 PR rtl-optimization/82192
10614 * combine.c (make_extraction): Don't look through non-paradoxical
10615 SUBREGs or TRUNCATE if pos + len is or might be bigger than
10618 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10619 Alan Hayward <alan.hayward@arm.com>
10620 David Sherwood <david.sherwood@arm.com>
10622 * target.def (function_arg_offset): New hook.
10623 * targhooks.h (default_function_arg_offset): Declare.
10624 * targhooks.c (default_function_arg_offset): New function.
10625 * function.c (locate_and_pad_parm): Use
10626 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
10627 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
10628 (TARGET_FUNCTION_ARG_OFFSET): ...this.
10629 * doc/tm.texi: Regenerate.
10630 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
10631 * config/spu/spu.c (spu_function_arg_offset): New function.
10632 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
10633 * system.h (FUNCTION_ARG_OFFSET): Poison.
10635 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10636 Alan Hayard <alan.hayward@arm.com>
10637 David Sherwood <david.sherwood@arm.com>
10639 * target.def (truly_noop_truncation): New hook.
10640 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
10641 than TRULY_NOOP_TRUNCATION.
10642 * hooks.h (hook_bool_uint_uint_true): Declare.
10643 * hooks.c (hook_bool_uint_uint_true): New function.
10644 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
10645 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
10646 * doc/tm.texi: Regenerate.
10647 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
10648 rather than TRULY_NOOP_TRUNCATION in comments.
10649 (simplify_comparison): Likewise.
10650 (record_truncated_value): Likewise.
10651 * expmed.c (extract_bit_field_1): Likewise.
10652 (extract_split_bit_field): Likewise.
10653 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
10654 instead of TRULY_NOOP_TRUNCATION.
10655 * function.c (assign_parm_setup_block): Likewise.
10656 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
10657 * rtlhooks.c: Include target.h.
10658 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
10659 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
10660 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
10661 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
10662 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
10663 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
10664 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
10665 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
10666 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
10667 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
10668 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
10669 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
10670 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
10671 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
10672 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
10673 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
10674 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
10675 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
10676 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
10677 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
10678 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
10679 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
10680 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
10681 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
10682 * config/mips/mips.c (mips_truly_noop_truncation): New function.
10683 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10684 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10685 rather than TRULY_NOOP_TRUNCATION in comments.
10686 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
10687 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
10688 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
10689 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
10690 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
10691 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
10692 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
10693 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
10694 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
10695 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
10696 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
10697 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10698 rather than TRULY_NOOP_TRUNCATION in comments.
10699 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
10700 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
10701 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
10702 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
10703 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
10704 TRULY_NOOP_TRUNCATION condition.
10705 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
10706 (TRULY_NOOP_TRUNCATION): Delete.
10707 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
10708 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
10709 * config/spu/spu.c (spu_truly_noop_truncation): New function.
10710 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10711 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
10712 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
10713 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
10714 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10715 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10716 rather than TRULY_NOOP_TRUNCATION in comments.
10717 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
10718 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
10719 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
10720 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
10721 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
10722 * system.h (TRULY_NOOP_TRUNCATION): Poison.
10724 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
10727 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
10728 (*cmp_ior): Likewise.
10729 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
10730 (*ior_scc_scc_cmp): Likewise.
10731 (*and_scc_scc): Likewise.
10732 (*and_scc_scc_cmp): Likewise.
10734 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10735 Alan Hayard <alan.hayward@arm.com>
10736 David Sherwood <david.sherwood@arm.com>
10738 * target.def (can_change_mode_class): New hook.
10739 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
10740 (hard_regno_nregs): Likewise.
10741 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
10742 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
10743 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
10744 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
10745 (LOAD_EXTEND_OP): Update accordingly.
10746 * doc/tm.texi: Regenerate.
10747 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
10748 CANNOT_CHANGE_MODE_CLASS.
10749 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
10750 (REG_CAN_CHANGE_MODE_P): ...this new macro.
10751 * combine.c (simplify_set): Update accordingly.
10752 * emit-rtl.c (validate_subreg): Likewise.
10753 * recog.c (general_operand): Likewise.
10754 * regcprop.c (mode_change_ok): Likewise.
10755 * reload1.c (choose_reload_regs): Likewise.
10756 (inherit_piecemeal_p): Likewise.
10757 * rtlanal.c (simplify_subreg_regno): Likewise.
10758 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
10759 instead of CANNOT_CHANGE_MODE_CLASS.
10760 (reload_cse_simplify_operands): Likewise.
10761 * reload.c (push_reload): Use targetm.can_change_mode_class
10762 instead of CANNOT_CHANGE_MODE_CLASS.
10763 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
10764 REG_CANNOT_CHANGE_MODE_P.
10765 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10766 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
10767 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10768 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10769 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10770 (arm_can_change_mode_class): New function.
10771 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10772 than CANNOT_CHANGE_MODE_CLASS in comments.
10773 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10774 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
10775 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
10776 (ix86_can_change_mode_class): ...this new function, inverting the
10777 sense of the return value.
10778 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10779 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10780 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10781 (ia64_can_change_mode_class): New function.
10782 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10783 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
10784 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
10785 (m32c_can_change_mode_class): ...this new function, inverting the
10786 sense of the return value.
10787 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10788 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10789 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
10790 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
10791 (mips_can_change_mode_class): ...this new function, inverting the
10792 sense of the return value.
10793 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10794 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10795 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10796 (msp430_can_change_mode_class): New function.
10797 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10798 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
10799 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10800 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10801 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10802 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
10803 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10804 (pa_cannot_change_mode_class): Replace with...
10805 (pa_can_change_mode_class): ...this new function, inverting the
10806 sense of the return value.
10807 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10808 than CANNOT_CHANGE_MODE_CLASS in comments.
10809 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10810 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
10811 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10812 (pdp11_cannot_change_mode_class): Replace with...
10813 (pdp11_can_change_mode_class): ...this new function, inverting the
10814 sense of the return value.
10815 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10816 * config/powerpcspe/powerpcspe-protos.h
10817 (rs6000_cannot_change_mode_class_ptr): Delete.
10818 * config/powerpcspe/powerpcspe.c
10819 (rs6000_cannot_change_mode_class_ptr): Delete.
10820 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10821 (rs6000_option_override_internal): Assign to
10822 targetm.can_change_mode_class instead of
10823 rs6000_cannot_change_mode_class_ptr.
10824 (rs6000_cannot_change_mode_class): Replace with...
10825 (rs6000_can_change_mode_class): ...this new function, inverting the
10826 sense of the return value.
10827 (rs6000_debug_cannot_change_mode_class): Replace with...
10828 (rs6000_debug_can_change_mode_class): ...this new function.
10829 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10830 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
10831 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10832 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10833 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
10835 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
10836 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10837 (rs6000_option_override_internal): Assign to
10838 targetm.can_change_mode_class instead of
10839 rs6000_cannot_change_mode_class_ptr.
10840 (rs6000_cannot_change_mode_class): Replace with...
10841 (rs6000_can_change_mode_class): ...this new function, inverting the
10842 sense of the return value.
10843 (rs6000_debug_cannot_change_mode_class): Replace with...
10844 (rs6000_debug_can_change_mode_class): ...this new function.
10845 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10846 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
10847 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
10848 (s390_can_change_mode_class): ...this new function, inverting the
10849 sense of the return value.
10850 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10851 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10852 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
10853 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10854 (sh_cannot_change_mode_class): Replace with...
10855 (sh_can_change_mode_class): ...this new function, inverting the
10856 sense of the return value.
10857 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10858 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10859 (sparc_can_change_mode_class): New function.
10860 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10861 * config/spu/spu.c (spu_can_change_mode_class): New function.
10862 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10863 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10864 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10865 (visium_can_change_mode_class): New function.
10866 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
10868 2017-09-15 Richard Biener <rguenther@suse.de>
10870 PR tree-optimization/82217
10871 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
10872 but not undefined case.
10874 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10877 * postreload.c (reload_cse_simplify_operands): Skip
10878 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
10880 2017-09-15 Richard Biener <rguenther@suse.de>
10882 PR tree-optimization/68823
10883 * graphite-scop-detection.c (build_alias_set): If we have a
10884 possible dependence check whether we can handle them by just
10885 looking at the DRs DR_ACCESS_FNs.
10886 (build_scops): If build_alias_set fails, fail the SCOP.
10888 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
10890 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
10891 to support float128 built-in functions that require the ISA 3.0
10893 (BU_FLOAT128_3_HW): Likewise.
10894 (SQRTF128): Add support for the IEEE 128-bit square root and fma
10895 built-in functions.
10896 (FMAF128): Likewise.
10898 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
10899 support for built-in functions that need the ISA 3.0 IEEE 128-bit
10900 floating point instructions.
10901 (rs6000_invalid_builtin): Likewise.
10902 (rs6000_builtin_mask_names): Likewise.
10903 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
10904 (RS6000_BTM_FLOAT128_HW): Likewise.
10905 (RS6000_BTM_COMMON): Likewise.
10906 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
10908 * doc/extend.texi (RS/6000 built-in functions): Document the
10909 IEEE 128-bit floating point square root and fused multiply-add
10910 built-in functions.
10912 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
10914 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
10915 reg (r2) isn't in the set of registers defined in the prologue.
10917 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10918 Alan Hayward <alan.hayward@arm.com>
10919 David Sherwood <david.sherwood@arm.com>
10921 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
10922 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
10923 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
10924 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
10925 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
10927 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
10928 max_vectorization_factor.
10929 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
10931 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10932 Alan Hayward <alan.hayward@arm.com>
10933 David Sherwood <david.sherwood@arm.com>
10935 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
10936 (vect_worthwhile_without_simd_p): Declare.
10937 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
10938 (vectorizable_reduction): Use it.
10939 * tree-vect-stmts.c (vectorizable_shift): Likewise.
10940 (vectorizable_operation): Likewise.
10942 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10943 Alan Hayward <alan.hayward@arm.com>
10944 David Sherwood <david.sherwood@arm.com>
10946 * tree-vectorizer.h (vect_get_num_copies): New function.
10947 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
10948 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10949 (vectorizable_induction): Likewise.
10950 (vectorizable_live_operation): Likewise.
10951 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10952 (vectorizable_bswap): Likewise.
10953 (vectorizable_call): Likewise.
10954 (vectorizable_conversion): Likewise.
10955 (vectorizable_assignment): Likewise.
10956 (vectorizable_shift): Likewise.
10957 (vectorizable_operation): Likewise.
10958 (vectorizable_store): Likewise.
10959 (vectorizable_load): Likewise.
10960 (vectorizable_condition): Likewise.
10961 (vectorizable_comparison): Likewise.
10962 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
10964 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10965 Alan Hayward <alan.hayward@arm.com>
10966 David Sherwood <david.sherwood@arm.com>
10968 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
10969 of vect_init_vector.
10971 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10972 Alan Hayward <alan.hayward@arm.com>
10973 David Sherwood <david.sherwood@arm.com>
10975 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
10976 an inline wrapper that provides a location.
10977 (gimple_build_vector): Likewise.
10978 * gimple-fold.c (gimple_build_vector_from_val): New function.
10979 (gimple_build_vector): Likewise.
10980 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
10981 functions to build the initial value. Always return a gimple value.
10982 (get_initial_defs_for_reduction): Likewise. Only compute
10984 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
10985 vect_init_vector on the results from get_initial_def(s)_for_reduction.
10986 (vectorizable_induction): Use gimple_build_vector rather than
10989 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10990 Alan Hayward <alan.hayward@arm.com>
10991 David Sherwood <david.sherwood@arm.com>
10993 * target.h (vec_perm_indices): New typedef.
10994 (auto_vec_perm_indices): Likewise.
10995 * optabs-query.h: Include target.h
10996 (can_vec_perm_p): Take a vec_perm_indices *.
10997 * optabs-query.c (can_vec_perm_p): Likewise.
10998 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
10999 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11000 * tree-vect-generic.c (lower_vec_perm): Likewise.
11001 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
11002 (vect_grouped_load_supported): Likewise.
11003 (vect_shift_permute_load_chain): Likewise.
11004 (vect_permute_store_chain): Use auto_vec_perm_indices.
11005 (vect_permute_load_chain): Likewise.
11006 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
11007 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
11008 Update uses of can_vec_perm_p.
11009 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
11010 mode with a number of elements. Take a vec_perm_indices *.
11011 (vect_create_epilog_for_reduction): Update accordingly.
11012 Use auto_vec_perm_indices.
11013 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
11014 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
11015 (vect_transform_slp_perm_load): Likewise.
11016 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
11017 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
11018 (vect_gen_perm_mask_checked): Likewise.
11019 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
11020 (vect_gen_perm_mask_checked): Likewise.
11021 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
11022 (vectorizable_store): Likewise.
11023 (vectorizable_load): Likewise.
11024 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
11025 (vectorizable_bswap): Likewise.
11027 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11028 Alan Hayward <alan.hayward@arm.com>
11029 David Sherwood <david.sherwood@arm.com>
11031 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
11032 * tree.c (build_vector): Likewise.
11033 (build_vector_from_ctor): Update accordingly.
11034 (build_vector_from_val): Likewise.
11035 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
11036 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11037 * tree-vect-generic.c (add_rshift): Likewise.
11038 (expand_vector_divmod): Likewise.
11039 (optimize_vector_constructor): Likewise.
11040 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
11041 (vect_transform_slp_perm_load): Likewise.
11042 (vect_schedule_slp_instance): Likewise.
11043 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
11044 (vectorizable_call): Likewise.
11045 (vect_gen_perm_mask_any): Likewise. Add elements in order.
11046 * expmed.c (make_tree): Likewise.
11047 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
11048 a vector passed to build_vector.
11049 (fold_convert_const): Likewise.
11050 (exact_inverse): Likewise.
11051 (fold_ternary_loc): Likewise.
11052 (fold_relational_const): Likewise.
11053 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
11054 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
11055 (const_unop): Likewise. Store the reduction accumulator in a
11056 variable rather than an array.
11057 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
11058 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
11059 the new vector, rather than constructing it after the input arrays.
11060 (native_interpret_vector): Use auto_vec<tree> when building
11061 a vector passed to build_vector. Add elements in order.
11062 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
11063 auto_vec<tree> when building a vector passed to build_vector.
11064 (vect_create_epilog_for_reduction): Likewise.
11065 (vectorizable_induction): Likewise.
11066 (get_initial_def_for_reduction): Likewise. Fix indentation of
11068 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
11070 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
11071 passed to build_vector.
11073 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11074 Alan Hayward <alan.hayward@arm.com>
11075 David Sherwood <david.sherwood@arm.com>
11077 * tree-core.h (tree_base::u): Add an "nelts" field.
11078 (tree_vector): Use VECTOR_CST_NELTS as the length.
11079 * tree.c (tree_size): Likewise.
11080 (make_vector): Initialize VECTOR_CST_NELTS.
11081 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
11082 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
11083 TYPE_VECTOR_SUBPARTS.
11084 * expr.c (const_vector_mask_from_tree): Consistently use "units"
11085 as the number of units, setting it from VECTOR_CST_NELTS.
11086 (const_vector_from_tree): Likewise.
11087 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
11088 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
11089 (fold_negate_expr_1): Likewise.
11090 (fold_convert_const): Likewise.
11091 (const_binop): Likewise. Differentiate the number of output and
11093 (const_unop): Likewise.
11094 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
11095 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
11096 in cases that did the opposite.
11098 2017-09-14 Richard Biener <rguenther@suse.de>
11100 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
11103 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
11105 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
11107 2017-09-14 Jakub Jelinek <jakub@redhat.com>
11110 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
11111 if and where to split a bb, except for splitting before debug insn
11112 sequences followed by non-label real insn. Delete debug insns
11113 in between basic blocks.
11115 * combine.c (make_compound_operation_int): Formatting fixes.
11117 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
11118 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11119 * config/netbsd.h (LINK_EH_SPEC): Likewise.
11120 * config/sol2.h (LINK_EH_SPEC): Likewise.
11121 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11122 * config/s390/linux.h (LINK_SPEC): Likewise.
11123 * config/freebsd.h (LINK_EH_SPEC): Likewise.
11124 * config/openbsd.h (LINK_EH_SPEC): Likewise.
11125 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11126 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
11127 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
11128 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11129 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
11130 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
11132 2017-09-13 Jakub Jelinek <jakub@redhat.com>
11134 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
11136 (ENDFILE_LINUX_SPEC): Likewise.
11137 (LINK_EH_SPEC): Likewise.
11138 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
11139 (LINK_OS_LINUX_SPEC32): Likewise.
11140 (LINK_OS_LINUX_SPEC64): Likewise.
11141 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
11142 (LINK_OS_LINUX_SPEC): Likewise.
11144 2017-09-13 Martin Liska <mliska@suse.cz>
11146 PR middle-end/82154
11147 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
11148 CASE_HIGH is NULL_TREE.
11150 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
11151 Alan Hayward <alan.hayward@arm.com>
11152 David Sherwood <david.sherwood@arm.com>
11154 * target.def (secondary_memory_needed): New hook.
11155 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
11156 instead of SECONDARY_MEMORY_NEEDED.
11157 (secondary_memory_needed_mode): Likewise.
11158 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
11159 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
11160 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
11161 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
11162 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
11163 * doc/tm.texi: Regenerate.
11164 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
11165 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
11166 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11167 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
11168 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
11169 * config/i386/i386.c (inline_secondary_memory_needed): Put the
11170 mode argument first and change the reg_class arguments to reg_class_t.
11171 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
11172 Make static. Update the call to inline_secondary_memory_needed.
11173 (ix86_register_move_cost): Update the call to
11174 inline_secondary_memory_needed.
11175 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11176 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
11178 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
11179 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11181 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
11182 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
11183 * config/mips/mips.c (mips_secondary_memory_needed): Make static
11184 and match hook interface. Add comment from mips.h.
11185 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11186 * config/mmix/mmix.md (truncdfsf2): Refer to
11187 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11189 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
11190 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
11191 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11192 (pa_secondary_memory_needed): New function.
11193 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
11194 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
11195 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11196 (pdp11_secondary_memory_needed): Make static and match hook interface.
11197 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
11198 * config/powerpcspe/powerpcspe-protos.h
11199 (rs6000_secondary_memory_needed_ptr): Delete.
11200 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
11202 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11203 (rs6000_option_override_internal): Assign to
11204 targetm.secondary_memory_needed rather than
11205 rs6000_secondary_memory_needed_ptr.
11206 (rs6000_secondary_memory_needed): Match hook interface.
11207 (rs6000_debug_secondary_memory_needed): Likewise.
11208 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
11209 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
11210 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
11211 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11212 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
11213 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
11215 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
11216 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11217 (rs6000_option_override_internal): Assign to
11218 targetm.secondary_memory_needed rather than
11219 rs6000_secondary_memory_needed_ptr.
11220 (rs6000_secondary_memory_needed): Match hook interface.
11221 (rs6000_debug_secondary_memory_needed): Likewise.
11222 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
11223 * config/s390/s390.c (s390_secondary_memory_needed): New function.
11224 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11225 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
11226 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11227 (sparc_secondary_memory_needed): New function.
11228 * lra-constraints.c (check_and_process_move): Refer to
11229 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11231 (curr_insn_transform): Likewise.
11232 (process_alt_operands): Use targetm.secondary_memory_needed
11233 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11234 (check_secondary_memory_needed_p): Likewise.
11235 (choose_split_class): Likewise.
11236 * reload.c: Unconditionally include code that was previously
11237 conditional on SECONDARY_MEMORY_NEEDED.
11238 (push_secondary_reload): Use targetm.secondary_memory_needed
11239 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11240 (push_reload): Likewise.
11241 * reload1.c: Unconditionally include code that was previously
11242 conditional on SECONDARY_MEMORY_NEEDED.
11243 (choose_reload_regs): Use targetm.secondary_memory_needed
11244 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11245 (gen_reload): Likewise.
11246 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
11248 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
11249 Alan Hayward <alan.hayward@arm.com>
11250 David Sherwood <david.sherwood@arm.com>
11252 * target.def (secondary_memory_needed_mode): New hook:
11253 * targhooks.c (default_secondary_memory_needed_mode): Declare.
11254 * targhooks.h (default_secondary_memory_needed_mode): New function.
11255 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
11256 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
11257 * doc/tm.texi: Regenerate.
11258 * lra-constraints.c (check_and_process_move): Use
11259 targetm.secondary_memory_needed_mode instead of
11260 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
11261 (curr_insn_transform): Likewise.
11262 * reload.c (get_secondary_mem): Likewise.
11263 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11264 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
11266 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11267 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11268 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
11269 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11270 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
11272 * config/powerpcspe/powerpcspe-protos.h
11273 (rs6000_secondary_memory_needed_mode): Delete.
11274 * config/powerpcspe/powerpcspe.c
11275 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11276 (rs6000_secondary_memory_needed_mode): Make static.
11277 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11278 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
11280 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
11282 (rs6000_secondary_memory_needed_mode): Make static.
11283 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11284 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
11285 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11286 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11287 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
11289 (sparc_secondary_memory_needed_mode): New function.
11290 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
11292 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
11294 * config/aarch64/constraints.md (Umq): New constraint.
11295 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
11297 (mov<mode>): Update condition.
11299 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11301 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
11302 when bitposition is the same.
11304 2017-09-13 Richard Biener <rguenther@suse.de>
11306 * dwarf2out.c (output_die_symbol): Remove.
11307 (output_die): Do not output a DIEs symbol.
11309 2017-09-13 Richard Biener <rguenther@suse.de>
11311 PR middle-end/82128
11312 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
11313 default-def to avoid breaking iterator update with the weird
11314 interaction with cgraph_update_edges_for_call_stmt_node.
11316 2017-09-13 Richard Biener <rguenther@suse.de>
11318 * tree-cfg.c (verify_gimple_assign_binary): Add verification
11319 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
11320 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
11321 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
11323 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
11325 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
11326 Disable pc relative literal load irrespective of
11327 TARGET_FIX_ERR_A53_84341 for default.
11329 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
11331 * config/sparc/sparc.c (output_return): Output the source location of
11332 the insn in the delay slot, if any.
11333 (output_sibcall): Likewise.
11335 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
11338 * common.opt (-static-pie): New alias.
11339 (shared): Negate static-pie.
11340 (-no-pie): Update help text.
11342 (static-pie): New option.
11343 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
11344 -static-pie support.
11345 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
11346 (LINK_EH_SPEC): Likewise.
11347 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11348 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11349 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11350 * gcc.c (LINK_COMMAND_SPEC): Likewise.
11351 (init_gcc_specs): Likewise.
11352 (init_spec): Likewise.
11353 (display_help): Update help message for -pie.
11354 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
11357 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
11359 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
11360 (movdi_aarch64): Likewise.
11361 (movti_aarch64): Likewise.
11363 2017-09-12 Simon Wright <simon@pushface.org>
11366 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
11367 calculation of the minor version, always output as 0.
11369 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11372 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
11373 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
11374 on it early, rather than manual conversion late. For
11375 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
11376 instead of performing manual conversion.
11378 2017-09-12 Carl Love <cel@us.ibm.com>
11380 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
11381 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
11383 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11384 VMULOSW): Add definitions.
11385 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11386 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11387 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11388 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11389 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11391 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11393 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
11395 (movti_aarch64): Likewise.
11396 (movdf_aarch64): Likewise.
11397 (movtf_aarch64): Likewise.
11398 (load_pairdi): Likewise.
11399 (store_pairdi): Likewise.
11400 (load_pairdf): Likewise.
11401 (store_pairdf): Likewise.
11402 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
11403 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
11404 (ldr_got_small_<mode>): Likewise.
11405 (ldr_got_small_28k_<mode>): Likewise.
11406 (ldr_got_tiny): Likewise.
11407 * config/aarch64/iterators.md (ldst_sz): New.
11408 (ldpstp_sz): Likewise.
11409 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
11411 (thunderx_load): Split load_8 to load_16.
11412 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
11414 (thunderx2t99_storepair_basic): Split store_8 to store_16.
11415 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
11416 (xgene1_store_pair): Split store_8 to store_16.
11417 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
11418 (falkor_st_0_st_sd): Split store_8 to store_16.
11420 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11422 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
11423 and store1/2/3/4 to store_4/8/12/16.
11424 * config/aarch64/aarch64.md: Update for rename.
11425 * config/arm/arm.md: Likewise.: Likewise.
11426 * config/arm/arm.c: Likewise.
11427 * config/arm/thumb1.md: Likewise.
11428 * config/arm/thumb2.md: Likewise.
11429 * config/arm/vfp.md: Likewise.
11430 * config/arm/arm-generic.md: Likewise.
11431 * config/arm/arm1020e.md: Likewise.
11432 * config/arm/arm1026ejs.md: Likewise.
11433 * config/arm/arm1136jfs.md: Likewise.
11434 * config/arm/arm926ejs.md: Likewise.
11435 * config/arm/cortex-a15.md: Likewise.
11436 * config/arm/cortex-a17.md: Likewise.
11437 * config/arm/cortex-a5.md: Likewise.
11438 * config/arm/cortex-a53.md: Likewise.
11439 * config/arm/cortex-a57.md: Likewise.
11440 * config/arm/cortex-a7.md: Likewise.
11441 * config/arm/cortex-a8.md: Likewise.
11442 * config/arm/cortex-a9.md: Likewise.
11443 * config/arm/cortex-m4.md: Likewise.
11444 * config/arm/cortex-m7.md: Likewise.
11445 * config/arm/cortex-r4.md: Likewise.
11446 * config/arm/exynos-m1.md: Likewise.
11447 * config/arm/fa526.md: Likewise.
11448 * config/arm/fa606te.md: Likewise.
11449 * config/arm/fa626te.md: Likewise.
11450 * config/arm/fa726te.md: Likewise.
11451 * config/arm/fmp626.md: Likewise.
11452 * config/arm/iwmmxt.md: Likewise.
11453 * config/arm/ldmstm.md: Likewise.
11454 * config/arm/marvell-pj4.md: Likewise.
11455 * config/arm/xgene1.md: Likewise.
11456 * config/aarch64/thunderx.md: Likewise.
11457 * config/aarch64/thunderx2t99.md: Likewise.
11458 * config/aarch64/falkor.md: Likewise.
11460 2017-09-12 Martin Liska <mliska@suse.cz>
11462 * attribs.c (private_lookup_attribute): New function.
11463 * attribs.h (private_lookup_attribute): Declared here.
11464 (lookup_attribute): Called from this place.
11466 2017-09-12 Richard Biener <rguenther@suse.de>
11468 PR tree-optimization/82157
11469 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
11470 stmts with side-effects.
11472 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11473 Alan Hayward <alan.hayward@arm.com>
11474 David Sherwood <david.sherwood@arm.com>
11476 * target.def (hard_regno_nregs): New hook.
11477 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
11478 * targhooks.h (default_hard_regno_nregs): Declare.
11479 * targhooks.c (default_hard_regno_nregs): New function.
11480 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
11481 (TARGET_HARD_REGNO_NREGS): ...this hook.
11482 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
11483 (CLASS_MAX_NREGS): Likewise.
11484 * doc/tm.texi: Regenerate.
11485 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
11486 instead of HARD_REGNO_NREGS.
11487 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
11488 HARD_REGNO_NREGS in the comment.
11489 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
11490 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
11491 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
11492 Return an unsigned int.
11493 (TARGET_HARD_REGNO_NREGS): Redefine.
11494 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
11495 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
11496 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11497 (arc_hard_regno_nregs): New function.
11498 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
11499 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
11500 (arm_hard_regno_nregs): New function.
11501 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
11502 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
11503 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
11504 (TARGET_HARD_REGNO_NREGS): Redefine.
11505 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
11506 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
11507 (HARD_REGNO_NREGS): Delete.
11508 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
11509 (cr16_hard_regno_nregs): New function.
11510 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
11511 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
11512 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
11513 (cris_hard_regno_nregs): New function.
11514 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
11515 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
11516 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11517 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
11518 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
11519 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
11520 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
11521 (frv_hard_regno_nregs): Make static. Take and return an
11523 (frv_class_max_nregs): Remove outdated copy of documentation.
11524 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
11525 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
11526 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
11527 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
11528 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
11529 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
11530 (TARGET_HARD_REGNO_NREGS): Redefine.
11531 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
11532 (CLASS_MAX_NREGS): Update comment.
11533 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
11534 (ia64_hard_regno_nregs): New function.
11535 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
11536 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
11537 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
11538 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
11539 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
11541 (m32c_hard_regno_nregs): Likewise. Make static.
11542 (TARGET_HARD_REGNO_NREGS): Redefine.
11543 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
11544 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
11545 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
11546 (m68k_hard_regno_nregs): New function.
11547 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
11548 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
11549 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
11550 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
11551 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
11552 Take and return an unsigned int.
11553 (TARGET_HARD_REGNO_NREGS): Redefine.
11554 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
11555 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11556 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
11557 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
11558 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
11559 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
11560 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
11561 (msp430_hard_regno_nregs): Make static. Take and return an
11563 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
11564 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
11565 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
11566 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
11567 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
11568 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
11569 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
11570 (TARGET_HARD_REGNO_NREGS): Redefine.
11571 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
11572 (PA_HARD_REGNO_NREGS): ...this.
11573 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
11574 (PA_HARD_REGNO_NREGS): ...this.
11575 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
11576 (pa_hard_regno_nregs): New function.
11577 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
11578 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
11579 (pdp11_hard_regno_nregs): New function.
11580 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
11581 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
11582 (rs6000_hard_regno_nregs_hook): New function.
11583 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
11584 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
11585 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
11586 Take and return an unsigned int. Move earlier in file.
11587 (TARGET_HARD_REGNO_NREGS): Redefine.
11588 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
11589 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
11590 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
11591 (rl78_hard_regno_nregs): Make static. Take and return an
11593 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
11594 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
11595 (rs6000_hard_regno_nregs_hook): New function.
11596 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
11597 * config/rx/rx.c (rx_hard_regno_nregs): New function.
11598 (TARGET_HARD_REGNO_NREGS): Redefine.
11599 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
11600 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
11601 instead of HARD_REGNO_NREGS.
11602 (s390_hard_regno_nregs): New function.
11603 (s390_hard_regno_mode_ok): Add comment from s390.h.
11604 (TARGET_HARD_REGNO_NREGS): Redefine.
11605 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
11606 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
11607 (sh_hard_regno_nregs): New function.
11608 (sh_pass_in_reg_p): Use it.
11609 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
11610 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11611 (sparc_hard_regno_nregs): New function.
11612 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
11613 * config/spu/spu.c (spu_hard_regno_nregs): New function.
11614 (spu_function_arg_advance): Use it, supplying a valid register number.
11615 (TARGET_HARD_REGNO_NREGS): Redefine.
11616 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
11617 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
11618 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
11619 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
11620 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
11621 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
11622 (CLASS_MAX_NREGS): Remove copy of old documentation.
11623 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
11624 (visium_hard_regno_nregs): New function.
11625 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
11626 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
11627 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
11628 xtensa_hard_regno_nregs): New function.
11629 * system.h (HARD_REGNO_NREGS): Poison.
11631 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11633 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
11634 hard_regno_nregs instead of HARD_REGNO_NREGS.
11635 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
11636 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
11637 (c6x_expand_epilogue): Likewise.
11638 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
11639 (frv_read_iacc_argument): Likewise.
11640 * config/sh/sh.c: Include regs.h.
11641 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
11642 (regs_used): Likewise.
11643 (output_stack_adjust): Likewise.
11644 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
11645 * expmed.c: Include regs.h.
11646 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
11647 * ree.c: Include regs.h.
11648 (combine_reaching_defs): Use hard_regno_nregs instead of
11650 (add_removable_extension): Likewise.
11652 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11654 * regs.h (hard_regno_nregs): Turn into a function.
11655 (end_hard_regno): Update accordingly.
11656 * caller-save.c (setup_save_areas): Likewise.
11657 (save_call_clobbered_regs): Likewise.
11658 (replace_reg_with_saved_mem): Likewise.
11659 (insert_restore): Likewise.
11660 (insert_save): Likewise.
11661 * combine.c (can_change_dest_mode): Likewise.
11662 (move_deaths): Likewise.
11663 (distribute_notes): Likewise.
11664 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
11665 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
11666 (rs6000_split_multireg_move): Likewise.
11667 (rs6000_register_move_cost): Likewise.
11668 (rs6000_memory_move_cost): Likewise.
11669 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
11670 (rs6000_split_multireg_move): Likewise.
11671 (rs6000_register_move_cost): Likewise.
11672 (rs6000_memory_move_cost): Likewise.
11673 * cselib.c (cselib_reset_table): Likewise.
11674 (cselib_lookup_1): Likewise.
11675 * emit-rtl.c (set_mode_and_regno): Likewise.
11676 * function.c (aggregate_value_p): Likewise.
11677 * ira-color.c (setup_profitable_hard_regs): Likewise.
11678 (check_hard_reg_p): Likewise.
11679 (calculate_saved_nregs): Likewise.
11680 (assign_hard_reg): Likewise.
11681 (improve_allocation): Likewise.
11682 (calculate_spill_cost): Likewise.
11683 * ira-emit.c (modify_move_list): Likewise.
11684 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
11685 (ira_hard_reg_in_set_p): Likewise.
11686 * ira.c (setup_reg_mode_hard_regset): Likewise.
11687 (clarify_prohibited_class_mode_regs): Likewise.
11688 (check_allocation): Likewise.
11689 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11690 (lra_setup_reg_renumber): Likewise.
11691 (setup_try_hard_regno_pseudos): Likewise.
11692 (spill_for): Likewise.
11693 (assign_hard_regno): Likewise.
11694 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
11695 * lra-constraints.c (in_class_p): Likewise.
11696 (lra_constraint_offset): Likewise.
11697 (simplify_operand_subreg): Likewise.
11698 (lra_constraints): Likewise.
11699 (split_reg): Likewise.
11700 (split_if_necessary): Likewise.
11701 (invariant_p): Likewise.
11702 (inherit_in_ebb): Likewise.
11703 * lra-lives.c (process_bb_lives): Likewise.
11704 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
11705 (get_hard_regs): Likewise.
11706 (do_remat): Likewise.
11707 * lra-spills.c (assign_spill_hard_regs): Likewise.
11708 * mode-switching.c (create_pre_exit): Likewise.
11709 * postreload.c (reload_combine_recognize_pattern): Likewise.
11710 * recog.c (peep2_find_free_register): Likewise.
11711 * regcprop.c (kill_value_regno): Likewise.
11712 (set_value_regno): Likewise.
11713 (copy_value): Likewise.
11714 (maybe_mode_change): Likewise.
11715 (find_oldest_value_reg): Likewise.
11716 (copyprop_hardreg_forward_1): Likewise.
11717 * regrename.c (check_new_reg_p): Likewise.
11718 (regrename_do_replace): Likewise.
11719 * reload.c (push_reload): Likewise.
11720 (combine_reloads): Likewise.
11721 (find_dummy_reload): Likewise.
11722 (operands_match_p): Likewise.
11723 (find_reloads): Likewise.
11724 (find_equiv_reg): Likewise.
11725 (reload_adjust_reg_for_mode): Likewise.
11726 * reload1.c (count_pseudo): Likewise.
11727 (count_spilled_pseudo): Likewise.
11728 (find_reg): Likewise.
11729 (clear_reload_reg_in_use): Likewise.
11730 (free_for_value_p): Likewise.
11731 (allocate_reload_reg): Likewise.
11732 (choose_reload_regs): Likewise.
11733 (reload_adjust_reg_for_temp): Likewise.
11734 (emit_reload_insns): Likewise.
11735 (delete_output_reload): Likewise.
11736 * rtlanal.c (subreg_get_info): Likewise.
11737 * sched-deps.c (sched_analyze_reg): Likewise.
11738 * sel-sched.c (init_regs_for_mode): Likewise.
11739 (mark_unavailable_hard_regs): Likewise.
11740 (choose_best_reg_1): Likewise.
11741 (verify_target_availability): Likewise.
11742 * valtrack.c (dead_debug_insert_temp): Likewise.
11743 * var-tracking.c (track_loc_p): Likewise.
11744 (emit_note_insn_var_location): Likewise.
11745 * varasm.c (make_decl_rtl): Likewise.
11746 * reginfo.c (choose_hard_reg_mode): Likewise.
11747 (init_reg_modes_target): Refer directly to
11748 this_target_regs->x_hard_regno_nregs.
11750 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11752 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
11753 instead of hard_regno_nregs.
11755 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11757 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
11758 end_hard_regno instead of hard_regno_nregs.
11759 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
11760 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
11761 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
11762 * ira-color.c (improve_allocation): Likewise.
11763 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11764 * lra-lives.c (mark_regno_live): Likewise.
11765 (mark_regno_dead): Likewise.
11766 * lra-remat.c (operand_to_remat): Likewise.
11767 * lra.c (collect_non_operand_hard_regs): Likewise.
11768 * postreload.c (reload_combine_note_store): Likewise.
11769 (move2add_valid_value_p): Likewise.
11770 * reload.c (regno_clobbered_p): Likewise.
11772 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11774 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
11776 * config/v850/v850.c (v850_reorg): Likewise.
11777 * reload.c (refers_to_regno_for_reload_p): Likewise.
11778 (find_equiv_reg): Likewise.
11779 * reload1.c (reload_reg_reaches_end_p): Likewise.
11781 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11783 * caller-save.c (add_used_regs): Use REG_NREGS instead of
11785 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
11786 * config/arm/arm.c (output_move_neon): Likewise.
11787 (arm_attr_length_move_neon): Likewise.
11788 (neon_split_vcombine): Likewise.
11789 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
11790 (c6x_mark_reg_written): Likewise.
11791 (c6x_dwarf_register_span): Likewise.
11792 * config/i386/i386.c (ix86_save_reg): Likewise.
11793 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
11794 (rws_access_reg): Likewise.
11795 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
11796 * mode-switching.c (create_pre_exit): Likewise.
11797 * ree.c (combine_reaching_defs): Likewise.
11798 (add_removable_extension): Likewise.
11799 * regcprop.c (find_oldest_value_reg): Likewise.
11800 (copyprop_hardreg_forward_1): Likewise.
11801 * reload.c (reload_inner_reg_of_subreg): Likewise.
11802 (push_reload): Likewise.
11803 (combine_reloads): Likewise.
11804 (find_dummy_reload): Likewise.
11805 (reload_adjust_reg_for_mode): Likewise.
11806 * reload1.c (find_reload_regs): Likewise.
11807 (forget_old_reloads_1): Likewise.
11808 (reload_reg_free_for_value_p): Likewise.
11809 (reload_adjust_reg_for_temp): Likewise.
11810 (emit_reload_insns): Likewise.
11811 (delete_output_reload): Likewise.
11812 * sel-sched.c (choose_best_reg_1): Likewise.
11813 (choose_best_pseudo_reg): Likewise.
11815 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11816 Alan Hayward <alan.hayward@arm.com>
11817 David Sherwood <david.sherwood@arm.com>
11819 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
11820 * target.def (slow_unaligned_access): New hook.
11821 * targhooks.h (default_slow_unaligned_access): Declare.
11822 * targhooks.c (default_slow_unaligned_access): New function.
11823 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
11824 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
11825 * doc/tm.texi: Regenerate.
11826 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
11827 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
11828 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
11830 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
11831 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
11833 (rs6000_slow_unaligned_access): New function.
11834 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
11835 (expand_block_compare): Likewise.
11836 (expand_strn_compare): Likewise.
11837 (rs6000_rtx_costs): Likewise.
11838 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
11839 (riscv_slow_unaligned_access): Likewise.
11840 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
11841 (riscv_slow_unaligned_access_p): ...this and make static.
11842 (riscv_option_override): Update accordingly.
11843 (riscv_slow_unaligned_access): New function.
11844 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
11845 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
11846 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
11847 (rs6000_slow_unaligned_access): New function.
11848 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
11849 (rs6000_rtx_costs): Likewise.
11850 * config/rs6000/rs6000-string.c (expand_block_compare)
11851 (expand_strn_compare): Use targetm.slow_unaligned_access instead
11852 of SLOW_UNALIGNED_ACCESS.
11853 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
11854 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
11855 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
11856 of SLOW_UNALIGNED_ACCESS.
11857 * expmed.c (simple_mem_bitfield_p): Likewise.
11858 * expr.c (alignment_for_piecewise_move): Likewise.
11859 (emit_group_load_1): Likewise.
11860 (emit_group_store): Likewise.
11861 (copy_blkmode_from_reg): Likewise.
11862 (emit_push_insn): Likewise.
11863 (expand_assignment): Likewise.
11864 (store_field): Likewise.
11865 (expand_expr_real_1): Likewise.
11866 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
11867 * lra-constraints.c (simplify_operand_subreg): Likewise.
11868 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11869 * gimple-ssa-store-merging.c: Likewise in block comment at start
11871 * tree-ssa-strlen.c: Include target.h.
11872 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
11873 of SLOW_UNALIGNED_ACCESS.
11874 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
11876 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11878 PR rtl-optimization/82185
11879 * expmed.c (emit_store_flag_int): Only test tem if it has been
11882 2017-09-12 Richard Biener <rguenther@suse.de>
11884 PR middle-end/82149
11885 * match.pd ((FTYPE) N CMP CST): Fix typo.
11887 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
11889 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
11891 (mips_near_type_p): Add 'short_call' attribute as a synonym
11893 * doc/extend.texi (short_call): Document new function attribute.
11895 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11898 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
11899 assertion check that in the condition.
11900 (get_atomic_generic_size): Likewise. Before testing if parameter
11901 has pointer type, if it has array type, call for C++
11902 default_conversion to perform array-to-pointer conversion.
11904 2017-09-12 Richard Biener <rguenther@suse.de>
11906 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
11907 for operations we cannot scalarize.
11909 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
11911 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
11912 vectors heap vectors. Clean up comments.
11913 Make visited_bbs a reference.
11914 (profitable_jump_thread_path): Make GC
11915 vectors heap vectors. Clean up comments.
11917 (convert_and_register_jump_thread_path): Make GC vectors heap
11919 (check_subpath_and_update_thread_path): Same. Clean up comments.
11920 Make visited_bbs a reference.
11921 (handle_phi): Abstract common code to to
11922 register_jump_thread_path_if_profitable.
11923 Rename VAR_BB to DEF_BB.
11925 Make GC vectors heap vectors.
11926 Make visited_bbs a reference.
11927 (handle_assignment): Same.
11928 (register_jump_thread_path_if_profitable): New.
11929 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
11931 Make GC vectors heap vectors. Clean up comments.
11932 Make visited_bbs a reference.
11933 (find_jump_threads_backwards): Make visited_bbs live in the stack.
11934 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
11937 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
11940 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
11941 words of E_DImode object are reachable by xtensa_uimm8x4 access.
11943 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
11945 Revert r251800 and r251799.
11947 2017-09-11 Martin Jambor <mjambor@suse.cz>
11950 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
11951 arguments in advance.
11952 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
11953 use it to find predecessor edges.
11954 (naive_outof_ssa): Collect vector of predecessors.
11956 2017-09-08 Jason Merrill <jason@redhat.com>
11958 PR c++/70029 - ICE with ref-qualifier and -flto
11959 * langhooks.h (struct lang_hooks_for_types): Add
11960 copy_lang_qualifiers.
11961 * attribs.c (build_type_attribute_qual_variant): Use it.
11962 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
11964 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
11965 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
11967 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
11970 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
11971 (*mulsi3_sp64): New instruction.
11972 (mulsi3): New expander.
11974 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
11976 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
11978 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11980 * sancov.c: Include memmodel.h.
11982 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
11985 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
11988 2017-09-07 Carl Love <cel@us.ibm.com>
11990 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
11991 the sldi instruction.
11993 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
11995 * sancov.c: Include tm_p.h.
11997 2017-09-07 Jakub Jelinek <jakub@redhat.com>
12000 * output.h (switch_to_other_text_partition): New declaration.
12001 * varasm.c (switch_to_other_text_partition): New function.
12002 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
12003 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
12004 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
12005 to the other text partition before emitting LCL label and switch back
12006 after emitting the word after it.
12008 2017-09-07 Richard Biener <rguenther@suse.de>
12010 * passes.def (pass_split_crit_edges): Remove instance before PRE.
12011 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
12012 critical edges here, after loop init.
12013 (pass_data_pre): Remove PROP_no_crit_edges flags.
12014 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
12015 for valueization of call args to avoid leaking VN_TOP.
12016 (visit_use): Assert we do not visit default defs.
12017 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
12018 Use error_mark_node to more easily detect leaking VN_TOP.
12019 All default-defs are varying, not VN_TOP. Mark them visited.
12020 (run_scc_vn): Make code match comment.
12022 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12024 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
12025 OPTION_MASK_FLOAT128_KEYWORD.
12026 (POWERPC_MASKS): Likewise.
12027 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
12028 support for the -mfloat128-type option, and make -mfloat128
12029 default on PowerPC Linux systems. Define or undefine
12030 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
12031 Define __float128 to be __ieee128 if IEEE 128-bit support is
12032 enabled, or undefine it.
12033 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
12034 Delete defining __FLOAT128_TYPE__.
12035 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
12036 -mfloat128-type option and make -mfloat128 default on PowerPC
12038 (TARGET_FLOAT128_TYPE): Likewise.
12039 (-mfloat128-type): Likewise.
12040 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12041 Delete the -mfloat128-type option and make -mfloat128 default on
12042 PowerPC Linux systems. Always use __ieee128 to be the keyword for
12043 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
12044 128-bit floating point is enabled. Change tests from using
12045 -mfloat128-type to -mfloat128.
12046 (rs6000_mangle_type): Use the correct mangling for the __float128
12047 type even if normal long double is restricted to 64-bits.
12048 (floatn_mode): Enable the _Float128 type by default on VSX Linux
12050 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
12051 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
12052 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
12054 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
12055 documentation for -mfloat128.
12057 2017-09-06 Olivier Hainque <hainque@adacore.com>
12059 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
12061 2017-09-06 Wish Wu <wishwu007@gmail.com>
12062 Jakub Jelinek <jakub@redhat.com>
12064 * asan.c (initialize_sanitizer_builtins): Add
12065 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
12066 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
12067 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
12068 BT_FN_VOID_UINT64_PTR variables.
12069 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
12070 (BT_FN_VOID_UINT16_UINT16): Likewise.
12071 (BT_FN_VOID_UINT32_UINT32): Likewise.
12072 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
12073 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
12074 (BT_FN_VOID_UINT64_PTR): Likewise.
12075 * common.opt (flag_sanitize_coverage): New variable.
12076 (fsanitize-coverage=trace-pc): Remove.
12077 (fsanitize-coverage=): Add.
12078 * flag-types.h (enum sanitize_coverage_code): New enum.
12079 * fold-const.c (fold_range_test): Disable non-short-circuit
12080 optimization if flag_sanitize_coverage.
12081 (fold_truth_andor): Likewise.
12082 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
12083 * opts.c (COVERAGE_SANITIZER_OPT): Define.
12084 (coverage_sanitizer_opts): New array.
12085 (get_closest_sanitizer_option): Add OPTS argument, handle also
12086 OPT_fsanitize_coverage_.
12087 (parse_sanitizer_options): Adjusted to also handle
12088 OPT_fsanitize_coverage_.
12089 (common_handle_option): Add OPT_fsanitize_coverage_.
12090 * sancov.c (instrument_comparison, instrument_switch): New function.
12091 (sancov_pass): Add trace-cmp support.
12092 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
12093 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
12094 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
12095 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
12096 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
12097 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
12098 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
12099 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
12100 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
12101 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
12103 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12105 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12106 error. Only quit immediately if parsing is complete.
12107 (BEGIN): Initialize fatal_err and parse_done.
12108 (begin fpu, end fpu): Check number of arguments.
12109 (begin arch, end arch): Likewise.
12110 (begin cpu, end cpu): Likewise.
12111 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12112 (optalias): Likewise.
12114 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12116 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12117 * config/arm/arm-isa.h: Delete. Move definitions to ...
12118 * arm-cpus.in: ... here. Use new feature and fgroup values.
12119 * config/arm/arm.c (arm_option_override): Use lower case for feature
12121 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12122 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12123 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12125 (print_isa_bits_for): New function.
12126 (gen_isa): New function.
12127 (gen_comm_data): Use print_isa_bits_for.
12128 (define feature): New keyword.
12129 (define fgroup): New keyword.
12130 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
12131 (arm-isa.h): Add rule to generate file.
12132 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12133 case for feature bit names.
12135 2017-09-06 Richard Biener <rguenther@suse.de>
12137 * tree-ssa-pre.c (NECESSARY): Remove.
12138 (create_expression_by_pieces): Do not touch pass-local flags.
12139 (insert_into_preds_of_block): Likewise.
12140 (do_pre_regular_insertion): Likewise.
12141 (eliminate_insert): Likewise.
12142 (eliminate_dom_walker::before_dom_children): Likewise.
12143 (fini_eliminate): Do not look at inserted_exprs.
12144 (mark_operand_necessary): Remove.
12145 (remove_dead_inserted_code): Replace with simple work-list
12146 algorithm based on inserted_exprs and SSA uses.
12147 (pass_pre::execute): Re-order fini_eliminate and
12148 remove_dead_inserted_code.
12150 2017-09-06 Olivier Hainque <hainque@adacore.com>
12152 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
12153 for VxWorks 7. Adjust surrounding comments.
12155 2017-09-06 Richard Biener <rguenther@suse.de>
12157 * gimple-ssa-strength-reduction.c
12158 (find_candidates_dom_walker::before_dom_children): Also allow
12161 2017-09-06 Richard Biener <rguenther@suse.de>
12163 PR tree-optimization/82108
12164 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
12165 for gap in the non-permutation SLP case.
12167 2017-09-06 Martin Jambor <mjambor@suse.cz>
12169 PR tree-optimization/82078
12170 * tree-sra.c (sort_and_splice_var_accesses): Move call to
12171 add_access_to_work_queue...
12172 (build_accesses_from_assign): ...here.
12173 (propagate_all_subaccesses): Make sure racc is the group
12174 representative, if there is one.
12176 2017-09-06 Jakub Jelinek <jakub@redhat.com>
12178 PR middle-end/82095
12179 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
12182 2017-09-06 Richard Biener <rguenther@suse.de>
12184 * gimple-ssa-strength-reduction.c
12185 (find_candidates_dom_walker::before_doom_children): Use a
12186 type and not a mode check.
12188 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
12191 * config/arm/predicates.md (arm_general_adddi_operand): Create new
12193 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
12195 2017-09-05 Jeff Law <law@redhat.com>
12197 PR tree-optimization/64910
12198 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
12199 cases where we have 3 or more operands.
12201 2017-09-05 Jakub Jelinek <jakub@redhat.com>
12203 PR middle-end/81768
12204 * omp-low.c (lower_omp_for): Recompute tree invariant if
12205 gimple_omp_for_initial/final is ADDR_EXPR.
12207 PR middle-end/81768
12208 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
12209 into gimple val before gimplification fo the COND_EXPR.
12211 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
12213 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
12214 REGION_COPY argument.
12215 (thread_through_all_blocks): Remove unused argument to
12216 duplicate_thread_path.
12218 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12219 Alan Hayward <alan.hayward@arm.com>
12220 David Sherwood <david.sherwood@arm.com>
12222 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
12223 Take a scalar_mode rather than a machine_mode.
12224 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12225 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
12226 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12227 (aarch64_gen_adjusted_ldpstp): Likewise.
12228 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
12230 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12231 Alan Hayward <alan.hayward@arm.com>
12232 David Sherwood <david.sherwood@arm.com>
12234 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
12235 Take a scalar_int_mode instead of a machine_mode.
12236 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
12237 (aarch64_output_scalar_simd_mov_immediate): Likewise.
12238 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
12239 (aarch64_simd_attr_length_rglist): Delete.
12240 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
12241 a scalar_int_mode instead of a machine_mode.
12242 (aarch64_add_offset): Likewise.
12243 (aarch64_internal_mov_immediate): Likewise
12244 (aarch64_add_constant_internal): Likewise.
12245 (aarch64_add_constant): Likewise.
12246 (aarch64_movw_imm): Likewise.
12247 (aarch64_rtx_arith_op_extract_p): Likewise.
12248 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
12249 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
12250 Remove assert that the mode isn't a vector.
12251 (aarch64_output_scalar_simd_mov_immediate): Likewise.
12252 (aarch64_expand_mov_immediate): Update calls after above changes.
12253 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
12254 (aarch64_and_bitmask_imm): Check for scalar integer modes.
12255 (aarch64_move_imm): Likewise.
12256 (aarch64_can_const_movi_rtx_p): Likewise.
12257 (aarch64_strip_extend): Likewise.
12258 (aarch64_extr_rtx_p): Likewise.
12259 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
12260 a CONST_INT when the mode parameter is VOIDmode.
12261 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
12263 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12265 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
12266 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12267 (bitwise_type_for_mode): Update accordingly.
12269 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12271 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
12272 * stor-layout.c (mode_for_size_tree): Likewise.
12273 (mode_for_array): Update accordingly.
12274 (layout_decl): Likewise.
12275 (compute_record_mode): Likewise. Only set the mode once.
12277 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12279 * target.def (get_mask_mode): Change return type to opt_mode.
12281 * doc/tm.texi: Regenerate.
12282 * targhooks.h (default_get_mask_mode): Return an opt_mode.
12283 * targhooks.c (default_get_mask_mode): Likewise.
12284 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12285 * optabs-query.c (can_vec_mask_load_store_p): Update use of
12286 targetm.get_mask_mode.
12287 * tree.c (build_truth_vector_type): Likewise.
12289 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12291 * machmode.h (mode_for_vector): Return an opt_mode.
12292 * stor-layout.c (mode_for_vector): Likewise.
12293 (mode_for_int_vector): Update accordingly.
12294 (layout_type): Likewise.
12295 * config/i386/i386.c (emit_memmov): Likewise.
12296 (ix86_expand_set_or_movmem): Likewise.
12297 (ix86_expand_vector_init): Likewise.
12298 (ix86_get_mask_mode): Likewise.
12299 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
12301 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
12302 * expmed.c (extract_bit_field_1): Likewise.
12303 * expr.c (expand_expr_real_2): Likewise.
12304 * optabs-query.c (can_vec_perm_p): Likewise.
12305 (can_vec_mask_load_store_p): Likewise.
12306 * optabs.c (expand_vec_perm): Likewise.
12307 * targhooks.c (default_get_mask_mode): Likewise.
12308 * tree-vect-stmts.c (vectorizable_store): Likewise.
12309 (vectorizable_load): Likewise.
12310 (get_vectype_for_scalar_type_and_size): Likewise.
12312 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12314 * machmode.h (mode_for_int_vector): New function.
12315 * stor-layout.c (mode_for_int_vector): Likewise.
12316 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
12317 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
12318 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12319 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12320 (s390_expand_vcond): Likewise.
12322 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12324 * machmode.h (opt_machine_mode): New type.
12325 (opt_mode<T>): Allow construction from anything that can be
12327 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
12328 (mode_for_size): Return an opt_machine_mode.
12329 * stor-layout.c (mode_for_size): Likewise.
12330 (mode_for_size_tree): Update call accordingly.
12331 (bitwise_mode_for_mode): Likewise.
12332 (make_fract_type): Likewise.
12333 (make_accum_type): Likewise.
12334 * caller-save.c (replace_reg_with_saved_mem): Update call
12336 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12337 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12338 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12339 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12340 * expmed.c (extract_bit_field_1): Likewise.
12341 * reload.c (get_secondary_mem): Likewise.
12342 * varasm.c (assemble_integer): Likewise.
12343 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
12346 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12348 * machmode.h (decimal_float_mode_for_size): New function.
12349 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
12350 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
12351 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
12352 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
12353 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
12354 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
12356 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12358 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
12359 (get_builtin_sync_mode): Likewise.
12360 (expand_ifn_atomic_compare_exchange): Likewise.
12361 (expand_builtin_atomic_clear): Likewise.
12362 (expand_builtin_atomic_test_and_set): Likewise.
12363 (fold_builtin_atomic_always_lock_free): Likewise.
12364 * calls.c (compute_argument_addresses): Likewise.
12365 (emit_library_call_value_1): Likewise.
12366 (store_one_arg): Likewise.
12367 * combine.c (combine_instructions): Likewise.
12368 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
12369 * config/arm/arm.c (arm_function_value): Likewise.
12370 (aapcs_allocate_return_reg): Likewise.
12371 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
12372 * config/i386/i386.c (construct_container): Likewise.
12373 (ix86_gimplify_va_arg): Likewise.
12374 (ix86_expand_sse_cmp): Likewise.
12375 (emit_memmov): Likewise.
12376 (emit_memset): Likewise.
12377 (expand_small_movmem_or_setmem): Likewise.
12378 (ix86_expand_pextr): Likewise.
12379 (ix86_expand_pinsr): Likewise.
12380 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
12381 * config/microblaze/microblaze.c (microblaze_block_move_straight):
12383 * config/mips/mips.c (mips_function_value_1) Likewise.
12384 (mips_block_move_straight): Likewise.
12385 (mips_expand_ins_as_unaligned_store): Likewise.
12386 * config/powerpcspe/powerpcspe.c
12387 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12388 (rs6000_darwin64_record_arg_flush): Likewise.
12389 * config/rs6000/rs6000.c
12390 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12391 (rs6000_darwin64_record_arg_flush): Likewise.
12392 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
12393 (sparc_function_value_1): Likewise.
12394 * config/spu/spu.c (adjust_operand): Likewise.
12395 (spu_emit_branch_or_set): Likewise.
12396 (arith_immediate_p): Likewise.
12397 * emit-rtl.c (gen_lowpart_common): Likewise.
12398 * expr.c (expand_expr_real_1): Likewise.
12399 * function.c (assign_parm_setup_block): Likewise.
12400 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
12401 * reload1.c (alter_reg): Likewise.
12402 * stor-layout.c (mode_for_vector): Likewise.
12403 (layout_type): Likewise.
12405 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12407 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
12408 (spu_convert_move): Likewise.
12409 * lower-subreg.c (resolve_simple_move): Likewise.
12411 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12414 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
12415 define_insn to a define_expand.
12416 (altivec_vsum2sws_direct): New define_insn.
12417 (altivec_vsumsws): Convert from a define_insn to a define_expand.
12419 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12421 * config/arm/arm.c (arm_option_params_internal): Improve setting of
12424 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
12429 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
12430 realign stack if stack alignment needed is less than incoming
12433 2017-09-05 Marek Polacek <polacek@redhat.com>
12436 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
12439 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12441 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
12443 2017-09-05 Richard Biener <rguenther@suse.de>
12445 PR tree-optimization/82084
12446 * fold-const.c (can_native_encode_string_p): Handle wide characters.
12448 2017-09-05 Richard Biener <rguenther@suse.de>
12450 PR tree-optimization/82102
12451 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
12453 2017-09-05 Martin Liska <mliska@suse.cz>
12455 PR tree-optimization/82032
12456 * tree-cfg.c (generate_range_test): New function.
12457 * tree-cfg.h (generate_range_test): Declared here.
12458 * tree-cfgcleanup.c (convert_single_case_switch): New function.
12459 (cleanup_control_expr_graph): Use it.
12460 * tree-switch-conversion.c (try_switch_expansion): Remove
12462 (emit_case_nodes): Use generate_range_test.
12464 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12467 * config/i386/i386.md (*<btsc><mode>_mask): Add
12468 TARGET_USE_BT to insn constraint.
12469 (*btr<mode>_mask): Ditto.
12471 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
12473 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
12474 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
12476 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
12479 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
12480 TARGET_NEON and TARGET_IWMMXT.
12481 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
12482 TARGET_NEON and TARGET_IWMMXT.
12483 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
12485 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12487 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
12488 (ix86_rewrite_tls_address): Ditto.
12489 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
12490 (ix86_rewrite_tls_address_1): Ditto.
12491 (ix86_rewrite_tls_address): Ditto.
12492 * config/i386/predicates.md (tls_address_pattern): New predicate.
12493 * config/i386/i386.md (TLS address splitter): New splitter.
12495 2017-09-04 Richard Biener <rguenther@suse.de>
12497 PR tree-optimization/82084
12498 * fold-const.h (can_native_encode_string_p): Declare.
12499 * fold-const.c (can_native_encode_string_p): Factor out from ...
12500 (native_encode_string): ... here.
12501 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
12502 vectorizing stores from constants we later cannot handle.
12504 2017-09-04 Marek Polacek <polacek@redhat.com>
12507 * doc/invoke.texi: Update -Wtautological-compare documentation.
12509 2017-09-04 Jeff Law <law@redhat.com>
12511 PR tree-optimization/64910
12512 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
12513 swap the first and last operand if the last is a constant.
12515 2017-09-04 Marek Polacek <polacek@redhat.com>
12518 * convert.c (do_narrow): When sanitizing signed integer overflows,
12519 bail out for signed types.
12520 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
12522 2017-09-04 Richard Biener <rguenther@suse.de>
12524 PR tree-optimization/82060
12525 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
12526 Move devirtualization after stmt folding and before EH/AB/noreturn
12527 cleanup to get the stmt refs canonicalized. Use a bool instead
12528 of gimple_modified_p since that doesn't work for NOPs. Schedule
12529 NOPs generated by folding for removal.
12531 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12532 Alan Hayward <alan.hayward@arm.com>
12533 David Sherwood <david.sherwood@arm.com>
12535 * coretypes.h (pad_direction): New enum.
12536 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
12537 (FUNCTION_ARG_PADDING): Likewise.
12538 * target.def (function_arg_padding): New hook.
12539 * targhooks.h (default_function_arg_padding): Declare.
12540 * targhooks.c (default_function_arg_padding): New function.
12541 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
12542 (TARGET_FUNCTION_ARG_PADDING): ...this.
12543 * doc/tm.texi: Regenerate.
12544 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
12545 instead of direction.
12546 (compute_argument_addresses): Likewise.
12547 (load_register_parameters): Likewise.
12548 (emit_library_call_value_1): Likewise.
12549 (store_one_arg): Use targetm.calls.function_arg_padding instead
12550 of FUNCTION_ARG_PADDING.
12551 (must_pass_in_stack_var_size_or_pad): Likewise.
12552 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
12553 (emit_group_store): Likewise.
12554 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
12555 instead of FUNCTION_ARG_PADDING.
12556 (emit_push_insn): Likewise, and propagate enum change throughout
12558 * function.h (direction): Delete.
12559 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
12561 * function.c (assign_parm_find_stack_rtl): Likewise.
12562 (assign_parm_setup_block_p): Likewise.
12563 (assign_parm_setup_block): Likewise.
12564 (gimplify_parameters): Likewise.
12565 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
12566 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
12568 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
12569 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12570 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
12571 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
12572 (aarch64_function_arg_padding): ...this new function.
12573 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
12574 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12575 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
12576 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12577 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
12578 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12579 (arm_pad_arg_upward): Replace with...
12580 (arm_function_arg_padding): ...this new function.
12581 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
12583 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
12584 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
12585 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12586 (ia64_hpux_function_arg_padding): Replace with...
12587 (ia64_function_arg_padding): ...this new function. Use pad_direction
12588 instead of direction. Check for TARGET_HPUX.
12589 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
12590 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12591 (iq2000_function_arg_padding): New function.
12592 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
12593 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
12594 (mips_function_arg_padding): ...this new function.
12595 (mips_pad_reg_upward): Update accordingly.
12596 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12597 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
12598 targetm.calls.function_arg_padding.
12599 (FUNCTION_ARG_PADDING): Delete.
12600 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12601 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
12602 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12603 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
12604 (nios2_block_reg_padding): Return pad_direction instead of direction.
12605 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
12606 instead of direction.
12607 (nios2_function_arg_padding): Likewise. Make static.
12608 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12609 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
12610 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
12611 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
12612 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12613 (pa_function_arg_padding): Make static. Return pad_direction instead
12615 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
12616 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12617 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
12618 instead of direction. Use targetm.calls.function_arg_padding.
12619 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
12620 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
12621 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
12622 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
12623 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
12625 (function_arg_padding): Rename to...
12626 (rs6000_function_arg_padding): ...this. Make static. Return
12627 pad_direction instead of direction.
12628 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
12629 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
12630 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12631 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
12632 instead of direction. Use targetm.calls.function_arg_padding.
12633 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
12634 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
12635 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
12636 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
12637 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12638 (function_arg_padding): Rename to...
12639 (rs6000_function_arg_padding): ...this. Make static. Return
12640 pad_direction instead of direction.
12641 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
12642 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
12643 * config/s390/s390.c (s390_function_arg_padding): New function.
12644 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12645 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
12646 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
12647 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12648 (function_arg_padding): Rename to...
12649 (sparc_function_arg_padding): ...this. Make static. Return
12650 pad_direction instead of direction.
12651 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
12652 * config/spu/spu.c (spu_function_arg_padding): New function.
12653 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12654 * system.h (FUNCTION_ARG_PADDING): Poison.
12656 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12657 Alan Hayward <alan.hayward@arm.com>
12658 David Sherwood <david.sherwood@arm.com>
12660 * target.def (modes_tieable_p): New hook.
12661 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
12662 (TARGET_MODES_TIEABLE_P): ...this.
12663 * doc/tm.texi.in: Regenerate.
12664 * hooks.h (hook_bool_mode_mode_true): Declare.
12665 * hooks.c (hook_bool_mode_mode_true): New function.
12666 * combine.c (subst): Use targetm.modes_tieable_p instead of
12668 * dse.c (find_shift_sequence): Likewise.
12669 * expmed.c (extract_low_bits): Likewise.
12670 * lower-subreg.c: Include target.h.
12671 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
12673 * rtlanal.c (rtx_cost): Likewise.
12674 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
12675 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
12676 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
12677 (TARGET_MODES_TIEABLE_P): Redefine.
12678 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
12679 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
12680 (TARGET_MODES_TIEABLE_P): Redefine.
12681 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
12682 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
12683 (arc_modes_tieable_p): New function.
12684 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
12685 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
12686 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
12687 (arm_modes_tieable_p): Make static.
12688 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
12689 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
12690 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
12691 (TARGET_MODES_TIEABLE_P): Redefine.
12692 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
12693 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
12694 (TARGET_MODES_TIEABLE_P): Redefine.
12695 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
12696 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
12697 (cr16_modes_tieable_p): New function.
12698 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
12699 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
12700 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
12701 (TRULY_NOOP_TRUNCATION): Update comment.
12702 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
12703 (TRULY_NOOP_TRUNCATION): Update comment.
12704 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
12705 (frv_modes_tieable_p): New function.
12706 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
12707 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
12708 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
12709 (TARGET_MODES_TIEABLE_P): Redefine.
12710 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
12711 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
12712 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
12713 (TARGET_MODES_TIEABLE_P): Redefine.
12714 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
12715 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
12716 (ia64_modes_tieable_p): New function.
12717 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
12718 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
12719 (iq2000_modes_tieable_p): New function.
12720 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
12721 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
12722 (lm32_modes_tieable_p): New function.
12723 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
12724 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
12725 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
12726 (TARGET_MODES_TIEABLE_P): Redefine.
12727 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
12728 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
12729 (m32r_modes_tieable_p): New function.
12730 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
12731 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
12732 (m68k_modes_tieable_p): New function.
12733 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
12734 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
12735 (mcore_modes_tieable_p): New function.
12736 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
12737 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
12739 (TARGET_MODES_TIEABLE_P): Redefine.
12740 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
12741 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
12742 * config/mips/mips.c (mips_modes_tieable_p): Make static.
12743 (TARGET_MODES_TIEABLE_P): Redefine.
12744 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
12745 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
12746 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
12747 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
12748 (mn10300_modes_tieable_p): ...this and make static.
12749 (TARGET_MODES_TIEABLE_P): Redefine.
12750 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
12751 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
12752 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
12753 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
12754 (msp430_modes_tieable_p): Make static.
12755 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
12756 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
12757 (TARGET_MODES_TIEABLE_P): Redefine.
12758 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
12759 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
12760 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
12761 (TARGET_MODES_TIEABLE_P): Redefine.
12762 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
12763 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
12764 * config/pa/pa.c (pa_modes_tieable_p): Make static.
12765 (TARGET_MODES_TIEABLE_P): Redefine.
12766 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
12767 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
12768 (pdp11_modes_tieable_p): New function.
12769 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
12770 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
12771 (rs6000_modes_tieable_p): New function.
12772 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12773 * config/powerpcspe/powerpcspe.md: Update comment.
12774 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
12775 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
12776 (TARGET_MODES_TIEABLE_P): Redefine.
12777 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
12778 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
12779 (rl78_modes_tieable_p): New function.
12780 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
12781 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
12782 (rs6000_modes_tieable_p): New function.
12783 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12784 * config/rs6000/rs6000.md: Update comment.
12785 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
12786 * config/rx/rx.c (rx_modes_tieable_p): New function.
12787 (TARGET_MODES_TIEABLE_P): Redefine.
12788 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
12789 * config/s390/s390.c (s390_modes_tieable_p): New function.
12790 (TARGET_MODES_TIEABLE_P): Redefine.
12791 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
12792 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
12793 (sh_modes_tieable_p): New function.
12794 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
12795 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
12796 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
12797 (sparc_modes_tieable_p): Make static.
12798 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
12799 * config/spu/spu.c (spu_modes_tieable_p): New function.
12800 (TARGET_MODES_TIEABLE_P): Redefine.
12801 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
12802 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
12803 (TARGET_MODES_TIEABLE_P): Redefine.
12804 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
12805 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
12806 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
12807 * config/v850/v850.c (v850_modes_tieable_p): New function.
12808 (TARGET_MODES_TIEABLE_P): Redefine.
12809 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
12810 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
12811 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
12812 (visium_modes_tieable_p): New function.
12813 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
12814 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
12815 (xtensa_modes_tieable_p): New function.
12816 * system.h (MODES_TIEABLE_P): Poison.
12818 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12819 Alan Hayward <alan.hayward@arm.com>
12820 David Sherwood <david.sherwood@arm.com>
12822 * target.def (hard_regno_mode_ok): New hook.
12823 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
12824 (TARGET_HARD_REGNO_MODE_OK): ...this.
12825 * doc/tm.texi.in: Regenerate.
12826 * hooks.h (hook_bool_uint_mode_true): Declare.
12827 * hooks.c (hook_bool_uint_mode_true): New function.
12828 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
12829 HARD_REGNO_MODE_OK.
12830 * genpreds.c (write_insn_preds_c): Add an include of target.h.
12831 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
12832 instead of HARD_REGNO_MODE_OK.
12833 * caller-save.c: Include target.h.
12834 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
12835 HARD_REGNO_MODE_OK.
12836 * combine.c (can_combine_p): Likewise.
12837 (combinable_i3pat): Likewise.
12838 (can_change_dest_mode): Likewise.
12839 * expr.c (init_expr_target): Likewise.
12840 (convert_move): Likewise.
12841 (convert_modes): Likewise.
12842 * ira.c (setup_prohibited_class_mode_regs): Likewise.
12843 (setup_prohibited_mode_move_regs): Likewise.
12844 * ira.h (target_ira): Likewise.
12845 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12846 * lra-constraints.c (process_alt_operands): Likewise.
12847 (split_reg): Likewise.
12848 * recog.c (peep2_find_free_register): Likewise.
12849 * ree.c (combine_reaching_defs): Likewise.
12850 * regcprop.c (maybe_mode_change): Likewise.
12851 * reginfo.c (init_reg_sets_1): Likewise.
12852 (choose_hard_reg_mode): Likewise.
12853 (simplifiable_subregs): Likewise.
12854 * regrename.c (check_new_reg_p): Likewise.
12855 * reload.c (find_valid_class): Likewise.
12856 (find_valid_class_1): Likewise.
12857 (reload_inner_reg_of_subreg): Likewise.
12858 (push_reload): Likewise.
12859 (combine_reloads): Likewise.
12860 (find_dummy_reload): Likewise.
12861 (find_reloads): Likewise.
12862 * reload1.c (find_reg): Likewise.
12863 (set_reload_reg): Likewise.
12864 (allocate_reload_reg): Likewise.
12865 (choose_reload_regs): Likewise.
12866 (reload_adjust_reg_for_temp): Likewise.
12867 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
12868 (simplify_subreg_regno): Likewise.
12869 * sel-sched.c (init_regs_for_mode): Likewise.
12870 * varasm.c (make_decl_rtl): Likewise.
12871 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
12872 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
12873 HARD_REGNO_MODE_OK.
12874 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
12875 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
12876 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12877 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
12878 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
12879 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12880 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
12881 (arc_mode_class): Delete.
12882 (HARD_REGNO_MODE_OK): Delete.
12883 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12884 (arc_hard_regno_mode_ok): Rename old array to...
12885 (arc_hard_regno_mode_ok_modes): ...this.
12886 (arc_conditional_register_usage): Update accordingly.
12887 (arc_mode_class): Make static.
12888 (arc_hard_regno_mode_ok): New function.
12889 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
12890 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
12891 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12892 (arm_hard_regno_mode_ok): Make static.
12893 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
12894 HARD_REGNO_MODE_OK.
12895 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
12896 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
12897 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
12899 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12900 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
12901 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
12902 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
12903 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
12904 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12905 * config/bfin/predicates.md (valid_reg_operand): Use
12906 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
12907 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
12908 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
12909 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12910 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
12911 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
12912 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12913 (cr16_hard_regno_mode_ok): Make static and return a bool.
12914 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
12915 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12916 (cris_hard_regno_mode_ok): New function.
12917 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
12918 (epiphany_mode_class): Delete.
12919 (HARD_REGNO_MODE_OK): Delete.
12920 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
12921 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12922 (hard_regno_mode_ok): Rename to...
12923 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
12924 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
12925 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
12926 HARD_REGNO_MODE_OK.
12927 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
12928 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
12929 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12930 (frv_hard_regno_mode_ok): Make static and return a bool.
12931 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
12932 HARD_REGNO_MODE_OK.
12933 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
12934 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
12935 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
12936 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
12938 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12939 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
12940 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
12941 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
12943 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12944 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
12945 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12946 (ia64_hard_regno_mode_ok): New function.
12947 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
12948 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12949 (iq2000_hard_regno_mode_ok): New function.
12950 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
12951 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12952 (lm32_hard_regno_mode_ok): New function.
12953 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
12954 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
12955 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
12956 instead of HARD_REGNO_MODE_OK.
12957 (m32c_hard_regno_ok): Rename to...
12958 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
12959 (m32c_cannot_change_mode_class): Update accordingly.
12960 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12961 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
12962 (m32r_mode_class): Delete.
12963 (HARD_REGNO_MODE_OK): Delete.
12964 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12965 (m32r_hard_regno_mode_ok): Rename to...
12966 (m32r_hard_regno_modes): ...this.
12967 (m32r_mode_class): Make static.
12968 (m32r_hard_regno_mode_ok): New function.
12969 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
12970 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
12971 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12972 (m68k_hard_regno_mode_ok): Make static.
12973 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
12974 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12975 (mcore_hard_regno_mode_ok): New function.
12976 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
12977 (HARD_REGNO_MODE_OK): Delete.
12978 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
12980 (microblaze_hard_regno_mode_ok_p): ...this and make static.
12981 (microblaze_hard_regno_mode_ok): New function.
12982 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12983 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
12984 (mips_hard_regno_mode_ok): Delete.
12985 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
12986 (mips_hard_regno_mode_ok_p): ...this and make static.
12987 (mips_hard_regno_mode_ok_p): Rename to...
12988 (mips_hard_regno_mode_ok_uncached): ...this.
12989 (mips_hard_regno_mode_ok): New function.
12990 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
12991 of HARD_REGNO_MODE_OK.
12992 (mips_option_override): Update after above name changes.
12993 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12994 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
12995 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
12996 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
12997 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
12998 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12999 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
13000 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
13001 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
13002 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13003 (msp430_hard_regno_mode_ok): Make static and return a bool.
13004 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
13005 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
13006 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
13008 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13009 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
13010 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
13011 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
13012 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
13013 (PA_HARD_REGNO_MODE_OK): ...this
13014 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
13015 (PA_HARD_REGNO_MODE_OK): ...this.
13016 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13017 (pa_hard_regno_mode_ok): New function.
13018 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
13019 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13020 (pdp11_hard_regno_mode_ok): New function.
13021 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
13022 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
13024 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
13026 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13027 (rs6000_hard_regno_mode_ok): Rename to...
13028 (rs6000_hard_regno_mode_ok_uncached): ...this.
13029 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13030 (rs6000_hard_regno_mode_ok): New function.
13031 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
13032 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
13033 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
13034 (riscv_hard_regno_mode_ok): ...this and make static.
13035 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13036 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
13037 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
13038 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13039 (rl78_hard_regno_mode_ok): Make static and return bool.
13040 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
13041 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
13043 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
13044 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13045 (rs6000_hard_regno_mode_ok): Rename to...
13046 (rs6000_hard_regno_mode_ok_uncached): ...this.
13047 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13048 (rs6000_hard_regno_mode_ok): New function.
13049 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
13050 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
13051 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13052 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
13053 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
13054 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
13055 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13056 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
13057 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
13058 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13059 (sh_hard_regno_mode_ok): Make static.
13060 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
13061 instead of HARD_REGNO_MODE_OK.
13062 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
13063 (sparc_mode_class): Delete.
13064 (HARD_REGNO_MODE_OK): Delete.
13065 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13066 (hard_regno_mode_classes): Make static.
13067 (sparc_mode_class): Likewise.
13068 (sparc_hard_regno_mode_ok): New function.
13069 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
13070 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
13071 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
13073 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13074 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
13075 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
13076 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
13077 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
13078 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13079 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
13080 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
13081 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13082 (visium_hard_regno_mode_ok): New function.
13083 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
13084 instead of HARD_REGNO_MODE_OK.
13085 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
13086 (HARD_REGNO_MODE_OK): Delete.
13087 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
13088 (xtensa_hard_regno_mode_ok_p): ...this and make static.
13089 (xtensa_option_override): Update accordingly.
13090 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13091 (xtensa_hard_regno_mode_ok): New function.
13092 * system.h (HARD_REGNO_MODE_OK): Poison.
13094 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13095 Alan Hayward <alan.hayward@arm.com>
13096 David Sherwood <david.sherwood@arm.com>
13098 * target.def (hard_regno_call_part_clobbered): New hook.
13099 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
13100 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
13101 * doc/tm.texi: Regenerate.
13102 * hooks.h (hook_bool_uint_mode_false): Declare.
13103 * hooks.c (hook_bool_uint_mode_false): New function.
13104 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13105 * cselib.c (cselib_process_insn): Use
13106 targetm.hard_regno_call_part_clobbered instead of
13107 HARD_REGNO_CALL_PART_CLOBBERED.
13108 * ira-conflicts.c (ira_build_conflicts): Likewise.
13109 * ira-costs.c (ira_tune_allocno_costs): Likewise.
13110 * lra-constraints.c (need_for_call_save_p): Likewise.
13111 * lra-lives.c: Include target.h.
13112 (check_pseudos_live_through_calls): Use
13113 targetm.hard_regno_call_part_clobbered instead of
13114 HARD_REGNO_CALL_PART_CLOBBERED.
13115 * regcprop.c: Include target.h.
13116 (copyprop_hardreg_forward_1): Use
13117 targetm.hard_regno_call_part_clobbered instead of
13118 HARD_REGNO_CALL_PART_CLOBBERED.
13119 * reginfo.c (choose_hard_reg_mode): Likewise.
13120 * regrename.c (check_new_reg_p): Likewise.
13121 * reload.c (find_equiv_reg): Likewise.
13122 * reload1.c (emit_reload_insns): Likewise.
13123 * sched-deps.c (deps_analyze_insn): Likewise.
13124 * sel-sched.c (init_regs_for_mode): Likewise.
13125 (mark_unavailable_hard_regs): Likewise.
13126 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
13127 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13128 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
13130 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13131 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13132 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
13134 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
13136 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13137 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13138 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
13140 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13141 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13142 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
13144 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13145 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
13147 * config/powerpcspe/powerpcspe.c
13148 (rs6000_hard_regno_call_part_clobbered): New function.
13149 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13150 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13151 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
13153 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13154 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13155 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
13157 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13158 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13159 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
13161 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13162 Alan Hayward <alan.hayward@arm.com>
13163 David Sherwood <david.sherwood@arm.com>
13165 * rtl.h (subreg_memory_offset): Declare.
13166 * emit-rtl.c (subreg_memory_offset): New function.
13167 * expmed.c (store_bit_field_1): Use it.
13168 * expr.c (undefined_operand_subword_p): Likewise.
13169 * simplify-rtx.c (simplify_subreg): Likewise.
13171 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
13173 PR rtl-optimization/57448
13176 * optabs.c (expand_atomic_load): Place compiler memory barriers if
13177 using atomic_load pattern.
13178 (expand_atomic_store): Likewise.
13180 2017-09-04 Jakub Jelinek <jakub@redhat.com>
13183 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
13184 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
13185 handling. Use replace_call_with_value with NULL instead of
13186 gsi_replace, unlink_stmt_vdef and release_defs.
13188 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
13191 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
13193 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13196 * rtl.h (emit_library_call_value_1): Declare.
13197 (emit_library_call): Replace declaration with a series of overloads.
13198 Remove the parameter count argument.
13199 (emit_library_call_value): Likewise.
13200 * calls.c (emit_library_call_value_1): Make global. Replace varargs
13201 with an "rtx_mode_t *".
13202 (emit_library_call_value): Delete.
13203 (emit_library_call): Likewise.
13204 * asan.c (asan_emit_stack_protection): Update calls accordingly.
13205 (asan_emit_allocas_unpoison): Likewise.
13206 * builtins.c (expand_builtin_powi): Likewise.
13207 (expand_asan_emit_allocas_unpoison): Likewise.
13208 * cfgexpand.c (expand_main_function): Likewise.
13209 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
13210 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
13211 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
13212 * config/arm/arm.c (arm_trampoline_init): Likewise.
13213 (arm_call_tls_get_addr): Likewise.
13214 (arm_expand_divmod_libfunc): Likewise.
13215 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
13216 (smulsi3_highpart): Likewise.
13217 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
13218 (c6x_expand_compare): Likewise.
13219 (c6x_expand_movmem): Likewise.
13220 * config/frv/frv.c (frv_trampoline_init): Likewise.
13221 * config/i386/i386.c (ix86_trampoline_init): Likewise.
13222 (ix86_expand_divmod_libfunc): Likewise.
13223 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
13224 (ia64_expand_compare): Likewise.
13225 (ia64_profile_hook): Likewise.
13226 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
13227 (nonlocal_goto): Likewise.
13228 (restore_stack_nonlocal): Likewise.
13229 * config/m32r/m32r.c (block_move_call): Likewise.
13230 (m32r_trampoline_init): Likewise.
13231 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
13232 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
13233 (m68k_call_m68k_read_tp): Likewise.
13234 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
13235 (microblaze_expand_divide): Likewise.
13236 * config/mips/mips.h (mips_args): Likewise.
13237 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
13238 (MIPS_ICACHE_SYNC): Likewise.
13239 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
13240 (nios2_trampoline_init): Likewise.
13241 * config/pa/pa.c (hppa_tls_call): Likewise.
13242 (pa_trampoline_init): Likewise.
13243 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
13244 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
13245 (expand_strn_compare): Likewise.
13246 (rs6000_generate_compare): Likewise.
13247 (rs6000_expand_float128_convert): Likewise.
13248 (output_profile_hook): Likewise.
13249 (rs6000_trampoline_init): Likewise.
13250 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
13251 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
13252 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
13253 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
13254 (rs6000_generate_compare): Likewise.
13255 (rs6000_expand_float128_convert): Likewise.
13256 (output_profile_hook): Likewise.
13257 (rs6000_trampoline_init): Likewise.
13258 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
13259 * config/sh/sh.c (sh_trampoline_init): Likewise.
13260 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
13261 (sparc_emit_float_lib_cmp): Likewise.
13262 (sparc32_initialize_trampoline): Likewise.
13263 (sparc64_initialize_trampoline): Likewise.
13264 (sparc_profile_hook): Likewise.
13265 * config/spu/spu.c (ea_load_store): Likewise.
13266 * config/spu/spu.md (floatunssidf2): Likewise.
13267 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
13268 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
13269 * config/visium/visium.c (expand_block_move_4): Likewise.
13270 (expand_block_move_2): Likewise.
13271 (expand_block_move_1): Likewise.
13272 (expand_block_set_4): Likewise.
13273 (expand_block_set_2): Likewise.
13274 (expand_block_set_1): Likewise.
13275 (visium_trampoline_init): Likewise.
13276 (visium_profile_hook): Likewise.
13277 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
13278 (xtensa_setup_frame_addresses): Likewise.
13279 (xtensa_trampoline_init): Likewise.
13280 * except.c (sjlj_emit_function_enter): Likewise.
13281 (sjlj_emit_function_exit): Likewise.
13282 * explow.c (allocate_dynamic_stack_space): Likewise.
13283 (probe_stack_range): Likewise.
13284 * expr.c (convert_mode_scalar): Likewise.
13285 * optabs.c (expand_binop): Likewise.
13286 (expand_twoval_binop_libfunc): Likewise.
13287 (expand_unop): Likewise.
13288 (prepare_cmp_insn): Likewise.
13289 (prepare_float_lib_cmp): Likewise.
13290 (expand_float): Likewise.
13291 (expand_fix): Likewise.
13292 (expand_fixed_convert): Likewise.
13293 (maybe_emit_sync_lock_test_and_set): Likewise.
13294 (expand_atomic_compare_and_swap): Likewise.
13295 (expand_mem_thread_fence): Likewise.
13296 (expand_atomic_fetch_op): Likewise.
13298 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
13300 * doc/generic.texi (OpenACC): Adjust URL.
13301 * doc/invoke.texi (C Dialect Options): Ditto.
13303 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
13305 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
13306 predicate for operand 1. Add (m,<S>) constraint.
13307 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
13308 Prevent memory operand 1 with register operand 2.
13310 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
13312 PR rtl-optimization/82024
13313 * combine.c (try_combine): If the combination result is a PARALLEL,
13314 and we only need to retain the SET in there that would be placed
13315 at I2, check that we can place that at I3 instead, before doing so.
13317 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13320 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
13322 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
13323 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
13326 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
13327 Jeff Law <law@redhat.com>
13329 * varasm.c (bss_initializer_p): Do not put constants into .bss
13330 (categorize_decl_for_section): Handle bss_initializer_p returning
13331 false when DECL_INITIAL is NULL.
13333 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13336 * config/s390/s390.c (s390_can_inline_p): New function.
13338 2017-09-01 Jeff Law <law@redhat.com>
13340 PR tree-optimization/82052
13341 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
13342 Always initialize the returned slot after a hash table miss
13343 when INSERT is true.
13345 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
13347 * config/s390/s390.md (mem_signal_fence): Remove.
13348 * doc/md.texi (mem_signal_fence): Remove.
13349 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
13351 * target-insns.def (mem_signal_fence): Remove.
13353 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13356 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
13359 * asan.c (create_odr_indicator): Strip name encoding from assembler
13360 name before appending it after __odr_asan_.
13362 2017-09-01 Martin Liska <mliska@suse.cz>
13364 PR tree-optimization/82059
13365 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
13366 frequency only when an edge is redirected.
13368 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13370 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
13371 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
13372 (arc_conditional_register_usage): Remove ARC600 lp_count
13374 (arc_file_start): Emit Tag_ARC_CPU_variation.
13375 (arc_can_use_doloop_p): New conditions to use ZOLs.
13376 (hwloop_fail): New function.
13377 (hwloop_optimize): Likewise.
13378 (hwloop_pattern_reg): Likewise.
13379 (arc_doloop_hooks): New struct, to be used with reorg_loops.
13380 (arc_reorg_loops): New function, calls reorg_loops.
13381 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
13382 (arc600_corereg_hazard): Remove ZOL checking, case handled by
13384 (arc_loop_hazard): Remove function, functionality moved into
13386 (arc_hazard): Remove arc_loop_hazard call.
13387 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
13388 into hwloop_optimize.
13389 (arc_label_align): Remove ZOL handling.
13390 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
13391 * config/arc/arc.md (doloop_begin): Remove pattern.
13392 (doloop_begin_i): Likewise.
13393 (doloop_end_i): Likewise.
13394 (doloop_fallback): Likewise.
13395 (doloop_fallback_m): Likewise.
13396 (doloop_end): Reimplement expand.
13397 (arc_lp): New pattern for LP instruction.
13398 (loop_end): New pattern.
13399 (loop_fail): Likewise.
13400 (decrement_and_branch_until_zero): Likewise.
13401 * config/arc/arc.opt (mlpc-width): New option.
13402 * doc/invoke.texi (mlpc-width): Document option.
13404 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13406 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
13407 (arc_ccfsm_advance): Fix checking for delay slots.
13408 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
13410 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13412 * config/arc/arc.md (movqi_insn): Add stores to save constant long
13414 (movhi_insn): Update store instruction constraint which are saving
13415 6-bit short immediates.
13416 (movsi_insn): Consider also short scaled load operations.
13417 (zero_extendhisi2_i): Use Usd constraint instead of T.
13418 (extendhisi2_i): Add q constraint.
13419 (arc_clzsi2): Add type and length attributes.
13420 (arc_ctzsi2): Likewise.
13421 * config/arc/constraints.md (Usc): Update constraint, the
13422 assembler can parse two relocations for a single instruction.
13424 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13426 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
13427 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
13429 2017-08-31 Olivier Hainque <hainque@adacore.com>
13431 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
13432 match as powerpc-wrs-vxworks*.
13434 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
13436 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
13437 register constraint for by-element operand.
13438 (aarch64_mls_elt_merge<mode>): Likewise.
13440 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13442 * config/arc/arc.c (arc_can_follow_jump): Check for short
13445 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13447 * config.gcc: Use g.opt for arc.
13448 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
13449 functionality moved to ...
13450 (legitimate_scaled_address_p): New function, ...here.
13451 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
13452 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
13453 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
13455 (arc_override_options): Handle G option.
13456 (arc_output_pic_addr_const): Correct function definition.
13457 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
13458 (arc_decl_anon_ns_mem_p): Delete.
13459 (arc_in_small_data_p): Overhaul this function to take into
13460 consideration the value given via G option.
13461 (arc_rewrite_small_data_1): Renamed and corrected old
13462 arc_rewrite_small_data function.
13463 (arc_rewrite_small_data): New function.
13464 (small_data_pattern): Don't use pic_offset_table_rtx.
13465 * config/arc/arc.h (CC1_SPEC): Recognize G option.
13466 * config/arc/simdext.md (movmisalignv2hi): Use
13467 prepare_move_operands function.
13469 (movmisalign*): Likewise.
13470 * doc/invoke.texi (ARC options): Document -G option.
13472 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13474 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
13476 * config/arc/arc.c (arc_print_operand): Output scalled address for
13477 sdata whenever is possible.
13478 (arc_in_small_data_p): Allow sdata for 64bit datum when double
13479 load/stores are available.
13480 (compact_sda_memory_operand): Check for the alignment required by
13481 code density instructions.
13482 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
13484 * config/arc/constraints.md (Usd): Update constraint.
13485 (Us0): New constraint.
13486 (Usc): Update constraint.
13488 2017-08-31 Richard Biener <rguenther@suse.de>
13490 PR middle-end/82054
13491 * dwarf2out.c (dwarf2out_early_global_decl): Process each
13492 function only once.
13494 2017-08-31 Tamar Christina <tamar.christina@arm.com>
13496 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
13497 Resize type_signature.
13499 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13500 Alan Hayward <alan.hayward@arm.com>
13501 David Sherwood <david.sherwood@arm.com>
13503 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
13504 subregs whose inner modes can be stored in GPRs.
13505 (aarch64_classify_index): Likewise.
13507 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13508 Alan Hayward <alan.hayward@arm.com>
13509 David Sherwood <david.sherwood@arm.com>
13511 * config/aarch64/iterators.md (V_cmp_result): Rename to...
13512 (V_INT_EQUIV): ...this.
13513 (v_cmp_result): Rename to...
13514 (v_int_equiv): ...this.
13515 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
13516 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
13517 (copysign<mode>3): Likewise.
13518 (aarch64_simd_bsl<mode>_internal): Likewise.
13519 (aarch64_simd_bsl<mode>): Likewise.
13520 (vec_cmp<mode><mode>): Likewise.
13521 (vcond<mode><mode>): Likewise.
13522 (vcond<v_cmp_mixed><mode>): Likewise.
13523 (vcondu<mode><v_cmp_mixed>): Likewise.
13524 (aarch64_cm<optab><mode>): Likewise.
13525 (aarch64_cmtst<mode>): Likewise.
13526 (aarch64_fac<optab><mode>): Likewise.
13527 (vec_perm_const<mode>): Likewise.
13528 (vcond_mask_<mode><v_cmp_result>): Rename to...
13529 (vcond_mask_<mode><v_int_equiv>): ...this.
13530 (vec_cmp<mode><v_cmp_result>): Rename to...
13531 (vec_cmp<mode><v_int_equiv>): ...this.
13533 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13534 Alan Hayward <alan.hayward@arm.com>
13535 David Sherwood <david.sherwood@arm.com>
13537 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
13539 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
13540 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
13541 (UNSPEC_LD4_DREG): New unspecs.
13542 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
13543 (aarch64_ld2<mode>_dreg_be): Replace with...
13544 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
13546 (aarch64_ld3<mode>_dreg_le)
13547 (aarch64_ld3<mode>_dreg_be): Replace with...
13548 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
13550 (aarch64_ld4<mode>_dreg_le)
13551 (aarch64_ld4<mode>_dreg_be): Replace with...
13552 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
13555 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13557 PR tree-optimization/81987
13558 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
13559 insert an initializer in a location not dominated by the stride
13562 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13564 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
13565 on the entire header of the finally block in the fallthru case.
13567 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13569 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
13571 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
13573 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
13574 rs6000_emit_move_from_cr and call renamed function.
13575 (rs6000_emit_prologue): Call renamed functions.
13576 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
13577 movesi_from_cr, remove volatile CRs.
13579 2017-08-30 Jon Beniston <jon@beniston.com>
13580 Richard Biener <rguenther@suse.de>
13582 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
13583 of VECTOR_MODE_P check.
13584 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
13585 element vector types.
13587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13589 * df.h (df_read_modify_subreg_p): Remove in favor of...
13590 * rtl.h (read_modify_subreg_p): ...this new function. Take a
13591 const_rtx instead of an rtx.
13592 * cprop.c (local_cprop_find_used_regs): Update accordingly.
13593 * df-problems.c (df_word_lr_mark_ref): Likewise.
13594 * ira-lives.c (mark_pseudo_reg_live): Likewise.
13595 (mark_pseudo_reg_dead): Likewise.
13596 (mark_ref_dead): Likewise.
13597 * reginfo.c (init_subregs_of_mode): Likewise.
13598 * sched-deps.c (sched_analyze_1): Likewise.
13599 * df-scan.c (df_def_record_1): Likewise.
13600 (df_uses_record): Likewise.
13601 (df_read_modify_subreg_p): Remove in favor of...
13602 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
13603 const_rtx instead of an rtx.
13605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13606 Alan Hayward <alan.hayward@arm.com>
13607 David Sherwood <david.sherwood@arm.com>
13609 * rtl.h (partial_subreg_p): New function.
13610 * caller-save.c (save_call_clobbered_regs): Use it.
13611 * calls.c (expand_call): Likewise.
13612 * combine.c (combinable_i3pat): Likewise.
13613 (simplify_set): Likewise.
13614 (make_extraction): Likewise.
13615 (make_compound_operation_int): Likewise.
13616 (gen_lowpart_or_truncate): Likewise.
13617 (force_to_mode): Likewise.
13618 (make_field_assignment): Likewise.
13619 (reg_truncated_to_mode): Likewise.
13620 (record_truncated_value): Likewise.
13621 (move_deaths): Likewise.
13622 * cse.c (record_jump_cond): Likewise.
13623 (cse_insn): Likewise.
13624 * cselib.c (cselib_lookup_1): Likewise.
13625 * expmed.c (extract_bit_field_using_extv): Likewise.
13626 * function.c (assign_parm_setup_reg): Likewise.
13627 * ifcvt.c (noce_convert_multiple_sets): Likewise.
13628 * ira-build.c (create_insn_allocnos): Likewise.
13629 * lra-coalesce.c (merge_pseudos): Likewise.
13630 * lra-constraints.c (match_reload): Likewise.
13631 (simplify_operand_subreg): Likewise.
13632 (curr_insn_transform): Likewise.
13633 * lra-lives.c (process_bb_lives): Likewise.
13634 * lra.c (new_insn_reg): Likewise.
13635 (lra_substitute_pseudo): Likewise.
13636 * regcprop.c (mode_change_ok): Likewise.
13637 (maybe_mode_change): Likewise.
13638 (copyprop_hardreg_forward_1): Likewise.
13639 * reload.c (push_reload): Likewise.
13640 (find_reloads): Likewise.
13641 (find_reloads_subreg_address): Likewise.
13642 * reload1.c (alter_reg): Likewise.
13643 (eliminate_regs_1): Likewise.
13644 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13646 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
13648 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
13649 back to if statements, including unpack.
13651 2017-08-30 Martin Liska <mliska@suse.cz>
13653 PR inline-asm/82001
13654 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
13656 (func_checker::compare_asm_inputs_outputs): ... this function.
13657 (func_checker::compare_gimple_asm): Use the function to compare
13658 also ASM constrains.
13659 * ipa-icf-gimple.h: Rename the function.
13661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13662 Alan Hayward <alan.hayward@arm.com>
13663 David Sherwood <david.sherwood@arm.com>
13665 * coretypes.h (complex_mode): New type.
13666 * gdbhooks.py (build_pretty_printer): Handle it.
13667 * machmode.h (complex_mode): New class.
13668 (complex_mode::includes_p): New function.
13669 (is_complex_int_mode): Likewise.
13670 (is_complex_float_mode): Likewise.
13671 * genmodes.c (get_mode_class): Handle complex mode classes.
13672 * function.c (expand_function_end): Use is_complex_int_mode.
13674 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13675 Alan Hayward <alan.hayward@arm.com>
13676 David Sherwood <david.sherwood@arm.com>
13678 * coretypes.h (scalar_mode_pod): New typedef.
13679 * gdbhooks.py (build_pretty_printer): Handle it.
13680 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
13681 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
13682 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
13683 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
13684 as_a <scalar_mode>.
13686 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13687 Alan Hayward <alan.hayward@arm.com>
13688 David Sherwood <david.sherwood@arm.com>
13690 * machmode.h (mode_for_vector): Take a scalar_mode instead
13692 * stor-layout.c (mode_for_vector): Likewise.
13693 * explow.c (promote_mode): Use as_a <scalar_mode>.
13694 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
13696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13697 Alan Hayward <alan.hayward@arm.com>
13698 David Sherwood <david.sherwood@arm.com>
13700 * target.def (preferred_simd_mode): Take a scalar_mode
13701 instead of a machine_mode.
13702 * targhooks.h (default_preferred_simd_mode): Likewise.
13703 * targhooks.c (default_preferred_simd_mode): Likewise.
13704 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
13705 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
13706 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
13707 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
13708 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
13709 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
13710 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
13711 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
13713 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
13714 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
13715 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
13716 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
13717 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
13718 * doc/tm.texi: Regenerate.
13719 * optabs-query.c (can_vec_mask_load_store_p): Return false for
13722 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13723 Alan Hayward <alan.hayward@arm.com>
13724 David Sherwood <david.sherwood@arm.com>
13726 * target.def (scalar_mode_supported_p): Take a scalar_mode
13727 instead of a machine_mode.
13728 * targhooks.h (default_scalar_mode_supported_p): Likewise.
13729 * targhooks.c (default_scalar_mode_supported_p): Likewise.
13730 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
13731 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13732 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
13733 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
13734 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
13735 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
13736 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
13737 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
13738 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
13739 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
13740 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
13741 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
13743 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
13744 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
13745 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13746 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13747 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
13749 * doc/tm.texi: Regenerate.
13751 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13752 Alan Hayward <alan.hayward@arm.com>
13753 David Sherwood <david.sherwood@arm.com>
13755 * coretypes.h (opt_scalar_mode): New typedef.
13756 * gdbhooks.py (build_pretty_printers): Handle it.
13757 * machmode.h (mode_iterator::get_2xwider): Add overload for
13759 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
13761 * expr.c (convert_mode_scalar): Likewise.
13762 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13763 * optabs.c (expand_float): Likewise.
13764 (expand_fix): Likewise.
13765 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13768 Alan Hayward <alan.hayward@arm.com>
13769 David Sherwood <david.sherwood@arm.com>
13771 * optabs.c (expand_float): Explicitly check for scalars before
13772 using a branching expansion.
13773 (expand_fix): Likewise.
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 * expr.c (convert_mode): Split scalar handling out into...
13780 (convert_mode_scalar): ...this new function. Treat the modes
13783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13784 Alan Hayward <alan.hayward@arm.com>
13785 David Sherwood <david.sherwood@arm.com>
13787 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
13789 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
13791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13792 Alan Hayward <alan.hayward@arm.com>
13793 David Sherwood <david.sherwood@arm.com>
13795 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
13796 rather than a machine_mode.
13797 (fixed_from_string): Likewise.
13798 (fixed_convert): Likewise.
13799 (fixed_convert_from_int): Likewise.
13800 (fixed_convert_from_real): Likewise.
13801 (real_convert_from_fixed): Likewise.
13802 * fixed-value.c (fixed_from_double_int): Likewise.
13803 (fixed_from_string): Likewise.
13804 (fixed_convert): Likewise.
13805 (fixed_convert_from_int): Likewise.
13806 (fixed_convert_from_real): Likewise.
13807 (real_convert_from_fixed): Likewise.
13808 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
13810 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13811 Alan Hayward <alan.hayward@arm.com>
13812 David Sherwood <david.sherwood@arm.com>
13814 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
13815 of separate mode class checks. Do not allow vector modes here.
13816 (immed_wide_int_const): Use as_a <scalar_mode>.
13817 * explow.c (trunc_int_for_mode): Likewise.
13818 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
13819 (wi::shwi): Likewise.
13820 (wi::min_value): Likewise.
13821 (wi::max_value): Likewise.
13822 * dwarf2out.c (loc_descriptor): Likewise.
13823 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
13824 for CONST_WIDE_INT.
13826 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13827 Alan Hayward <alan.hayward@arm.com>
13828 David Sherwood <david.sherwood@arm.com>
13830 * tree.h (SCALAR_TYPE_MODE): New macro.
13831 * expr.c (expand_expr_addr_expr_1): Use it.
13832 (expand_expr_real_2): Likewise.
13833 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
13834 (fold_convert_const_fixed_from_int): Likewise.
13835 (fold_convert_const_fixed_from_real): Likewise.
13836 (native_encode_fixed): Likewise
13837 (native_encode_complex): Likewise
13838 (native_encode_vector): Likewise.
13839 (native_interpret_fixed): Likewise.
13840 (native_interpret_real): Likewise.
13841 (native_interpret_complex): Likewise.
13842 (native_interpret_vector): Likewise.
13843 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
13844 (simd_clone_adjust_argument_types): Likewise.
13845 (simd_clone_init_simd_arrays): Likewise.
13846 (simd_clone_adjust): Likewise.
13847 * stor-layout.c (layout_type): Likewise.
13848 * tree.c (build_minus_one_cst): Likewise.
13849 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13850 * tree-inline.c (estimate_move_cost): Likewise.
13851 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
13852 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
13853 (vectorizable_reduction): Likewise.
13854 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
13855 (vect_recog_mixed_size_cond_pattern): Likewise.
13856 (check_bool_pattern): Likewise.
13857 (adjust_bool_pattern): Likewise.
13858 (search_type_for_mask_1): Likewise.
13859 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
13860 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13861 (vectorizable_load): Likewise.
13862 (vectorizable_store): Likewise.
13863 * ubsan.c (ubsan_encode_value): Likewise.
13864 * varasm.c (output_constant): Likewise.
13866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13867 Alan Hayward <alan.hayward@arm.com>
13868 David Sherwood <david.sherwood@arm.com>
13870 * coretypes.h (scalar_mode): New class.
13871 * machmode.h (scalar_mode): Likewise.
13872 (scalar_mode::includes_p): New function.
13873 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
13874 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
13875 * genmodes.c (get_mode_class): Handle remaining scalar modes.
13876 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
13877 * expmed.c (store_bit_field_1): Likewise.
13878 (extract_bit_field_1): Likewise.
13879 * expr.c (write_complex_part): Likewise.
13880 (read_complex_part): Likewise.
13881 (emit_move_complex_push): Likewise.
13882 (expand_expr_real_2): Likewise.
13883 * function.c (assign_parm_setup_reg): Likewise.
13884 (assign_parms_unsplit_complex): Likewise.
13885 * optabs.c (expand_binop): Likewise.
13886 * rtlanal.c (subreg_get_info): Likewise.
13887 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13888 * varasm.c (output_constant_pool_2): Likewise.
13890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13891 Alan Hayward <alan.hayward@arm.com>
13892 David Sherwood <david.sherwood@arm.com>
13894 * expmed.c (extract_high_half): Use scalar_int_mode and remove
13896 (expmed_mult_highpart_optab): Likewise.
13897 (expmed_mult_highpart): Likewise.
13899 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13900 Alan Hayward <alan.hayward@arm.com>
13901 David Sherwood <david.sherwood@arm.com>
13903 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
13904 instead of a machine_mode.
13905 (builtin_memset_read_str): Likewise.
13906 * builtins.c (c_readstr): Likewise.
13907 (builtin_memcpy_read_str): Likewise.
13908 (builtin_strncpy_read_str): Likewise.
13909 (builtin_memset_read_str): Likewise.
13910 (builtin_memset_gen_str): Likewise.
13911 (expand_builtin_signbit): Use scalar_int_mode for local variables.
13912 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
13913 instead of a machine_mode.
13914 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
13916 (make_extraction): Likewise.
13917 (try_widen_shift_mode): Take and return scalar_int_modes instead
13919 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
13920 a scalar_int_mode instead of a machine_mode.
13921 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
13922 (avr_addr_space_pointer_mode): Likewise.
13923 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
13924 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
13925 (msp430_unwind_word_mode): Likewise.
13926 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
13927 (spu_addr_space_pointer_mode): Likewise.
13928 (spu_addr_space_address_mode): Likewise.
13929 (spu_libgcc_cmp_return_mode): Likewise.
13930 (spu_libgcc_shift_count_mode): Likewise.
13931 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
13932 (rl78_addr_space_pointer_mode): Likewise.
13933 (fl78_unwind_word_mode): Likewise.
13934 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
13936 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
13937 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
13938 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
13939 (mips_valid_pointer_mode): Likewise.
13940 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
13941 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
13942 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
13944 (ft32_addr_space_address_mode): Likewise.
13945 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
13946 scalar_int_mode instead of a machine_mode.
13947 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
13949 (m32c_addr_space_address_mode): Likewise.
13950 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
13951 (rs6000_eh_return_filter_mode): Likewise.
13952 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
13953 (rs6000_eh_return_filter_mode): Likewise.
13954 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
13955 (s390_libgcc_shift_count_mode): Likewise.
13956 (s390_unwind_word_mode): Likewise.
13957 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
13959 * target.def (mode_rep_extended): Likewise.
13960 (valid_pointer_mode): Likewise.
13961 (addr_space.valid_pointer_mode): Likewise.
13962 (eh_return_filter_mode): Return a scalar_int_mode rather than
13964 (libgcc_cmp_return_mode): Likewise.
13965 (libgcc_shift_count_mode): Likewise.
13966 (unwind_word_mode): Likewise.
13967 (addr_space.pointer_mode): Likewise.
13968 (addr_space.address_mode): Likewise.
13969 * doc/tm.texi: Regenerate.
13970 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
13972 (do_jump): Use scalar_int_mode for local variables.
13973 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
13974 rather than a machine_mode.
13975 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
13976 (scompare_loc_descriptor_wide): Likewise.
13977 (scompare_loc_descriptor_narrow): Likewise.
13978 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
13980 * except.c (sjlj_emit_dispatch_table): Likewise.
13981 (expand_builtin_eh_copy_values): Likewise.
13982 * explow.c (convert_memory_address_addr_space_1): Likewise.
13983 Take a scalar_int_mode rather than a machine_mode.
13984 (convert_memory_address_addr_space): Take a scalar_int_mode rather
13985 than a machine_mode.
13986 (memory_address_addr_space): Use scalar_int_mode for local variables.
13987 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
13988 rather than a machine_mode.
13989 * expmed.c (mask_rtx): Likewise.
13990 (init_expmed_one_conv): Likewise.
13991 (expand_mult_highpart_adjust): Likewise.
13992 (extract_high_half): Likewise.
13993 (expmed_mult_highpart_optab): Likewise.
13994 (expmed_mult_highpart): Likewise.
13995 (expand_smod_pow2): Likewise.
13996 (expand_sdiv_pow2): Likewise.
13997 (emit_store_flag_int): Likewise.
13998 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
14000 (extract_low_bits): Likewise.
14001 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
14003 * expr.c (pieces_addr::adjust): Likewise.
14004 (can_store_by_pieces): Likewise.
14005 (store_by_pieces): Likewise.
14006 (clear_by_pieces_1): Likewise.
14007 (expand_expr_addr_expr_1): Likewise.
14008 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
14009 (expand_expr_real_1): Likewise.
14010 (try_casesi): Likewise.
14011 * final.c (shorten_branches): Likewise.
14012 * fold-const.c (fold_convert_const_int_from_fixed): Change the
14013 type of "mode" to machine_mode.
14014 * internal-fn.c (expand_arith_overflow_result_store): Take a
14015 scalar_int_mode rather than a machine_mode.
14016 (expand_mul_overflow): Use scalar_int_mode for local variables.
14017 * loop-doloop.c (doloop_modify): Likewise.
14018 (doloop_optimize): Likewise.
14019 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
14020 than a machine_mode.
14021 (expand_doubleword_shift_condmove): Likewise.
14022 (expand_doubleword_shift): Likewise.
14023 (expand_doubleword_clz): Likewise.
14024 (expand_doubleword_popcount): Likewise.
14025 (expand_doubleword_parity): Likewise.
14026 (expand_absneg_bit): Use scalar_int_mode for local variables.
14027 (prepare_float_lib_cmp): Likewise.
14028 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
14029 rather than a machine_mode.
14030 (convert_memory_address_addr_space): Likewise.
14031 (get_mode_bounds): Likewise.
14032 (get_address_mode): Return a scalar_int_mode rather than a
14034 * rtlanal.c (get_address_mode): Likewise.
14035 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
14036 than a machine_mode.
14037 * targhooks.c (default_mode_rep_extended): Likewise.
14038 (default_valid_pointer_mode): Likewise.
14039 (default_addr_space_valid_pointer_mode): Likewise.
14040 (default_eh_return_filter_mode): Return a scalar_int_mode rather
14041 than a machine_mode.
14042 (default_libgcc_cmp_return_mode): Likewise.
14043 (default_libgcc_shift_count_mode): Likewise.
14044 (default_unwind_word_mode): Likewise.
14045 (default_addr_space_pointer_mode): Likewise.
14046 (default_addr_space_address_mode): Likewise.
14047 * targhooks.h (default_eh_return_filter_mode): Likewise.
14048 (default_libgcc_cmp_return_mode): Likewise.
14049 (default_libgcc_shift_count_mode): Likewise.
14050 (default_unwind_word_mode): Likewise.
14051 (default_addr_space_pointer_mode): Likewise.
14052 (default_addr_space_address_mode): Likewise.
14053 (default_mode_rep_extended): Take a scalar_int_mode rather than
14055 (default_valid_pointer_mode): Likewise.
14056 (default_addr_space_valid_pointer_mode): Likewise.
14057 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
14059 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
14060 rather than a machine_mode.
14061 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
14062 for local variables.
14063 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14064 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
14065 than a machine_mode.
14067 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14068 Alan Hayward <alan.hayward@arm.com>
14069 David Sherwood <david.sherwood@arm.com>
14071 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
14072 the mode argument to scalar_int_mode.
14073 (do_jump_by_parts_zero_rtx): Likewise.
14074 (do_jump_by_parts_equality_rtx): Likewise.
14075 (do_jump_by_parts_greater): Take a mode argument.
14076 (do_jump_by_parts_equality): Likewise.
14077 (do_jump_1): Update calls accordingly.
14079 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14080 Alan Hayward <alan.hayward@arm.com>
14081 David Sherwood <david.sherwood@arm.com>
14083 * is-a.h (safe_dyn_cast): New function.
14084 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
14085 (jump_table_for_label): Likewise.
14086 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
14087 instead of an rtx_insn *.
14088 (shorten_branches): Use dyn_cast instead of LABEL_P and
14089 JUMP_TABLE_DATA_P. Use jump_table_for_label and
14090 rtx_jump_table_data::get_data_mode.
14091 (final_scan_insn): Likewise.
14093 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14094 Alan Hayward <alan.hayward@arm.com>
14095 David Sherwood <david.sherwood@arm.com>
14097 * combine.c (try_combine): Use is_a <scalar_int_mode> when
14098 trying to combine a full-register integer set with a subreg
14101 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14102 Alan Hayward <alan.hayward@arm.com>
14103 David Sherwood <david.sherwood@arm.com>
14105 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
14106 that is always either address_mode or pointer_mode.
14108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14109 Alan Hayward <alan.hayward@arm.com>
14110 David Sherwood <david.sherwood@arm.com>
14112 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
14113 when the two are known to be equal.
14115 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14116 Alan Hayward <alan.hayward@arm.com>
14117 David Sherwood <david.sherwood@arm.com>
14119 * simplify-rtx.c (simplify_const_unary_operation): Use
14120 is_a <scalar_int_mode> instead of checking for a nonzero
14121 precision. Forcibly convert op_mode to a scalar_int_mode
14122 in that case. More clearly differentiate the operand and
14123 result modes and use the former when deciding what the value
14124 of a count-bits operation should be. Use is_int_mode instead
14125 of checking for a MODE_INT. Remove redundant check for whether
14126 this mode has a zero precision.
14128 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14129 Alan Hayward <alan.hayward@arm.com>
14130 David Sherwood <david.sherwood@arm.com>
14132 * optabs.c (widen_leading): Change the type of the mode argument
14133 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
14134 (widen_bswap): Likewise.
14135 (expand_parity): Likewise.
14136 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
14137 (expand_ffs): Likewise.
14138 (epand_unop): Check for scalar integer modes before calling the
14141 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14142 Alan Hayward <alan.hayward@arm.com>
14143 David Sherwood <david.sherwood@arm.com>
14145 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
14147 (expand_expr_real_1): Update call accordingly.
14149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14150 Alan Hayward <alan.hayward@arm.com>
14151 David Sherwood <david.sherwood@arm.com>
14153 * expmed.c (store_bit_field_using_insv): Add op0_mode and
14154 value_mode arguments. Use scalar_int_mode internally.
14155 (store_bit_field_1): Rename the new integer mode from imode
14156 to op0_mode and use it instead of GET_MODE (op0). Update calls
14157 to store_split_bit_field, store_bit_field_using_insv and
14158 store_fixed_bit_field.
14159 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
14160 Use scalar_int_mode internally. Use a bit count rather than a mode
14161 when calculating the largest bit size for get_best_mode.
14162 Update calls to store_split_bit_field and store_fixed_bit_field_1.
14163 (store_fixed_bit_field_1): Add mode and value_mode arguments.
14164 Remove assertion that OP0 has a scalar integer mode.
14165 (store_split_bit_field): Add op0_mode and value_mode arguments.
14166 Update calls to extract_fixed_bit_field.
14167 (extract_bit_field_using_extv): Add an op0_mode argument.
14168 Use scalar_int_mode internally.
14169 (extract_bit_field_1): Rename the new integer mode from imode to
14170 op0_mode and use it instead of GET_MODE (op0). Update calls to
14171 extract_split_bit_field, extract_bit_field_using_extv and
14172 extract_fixed_bit_field.
14173 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
14174 to extract_split_bit_field and extract_fixed_bit_field_1.
14175 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
14176 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
14177 on the target mode.
14178 (extract_split_bit_field): Add an op0_mode argument. Update call
14179 to extract_fixed_bit_field.
14181 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14182 Alan Hayward <alan.hayward@arm.com>
14183 David Sherwood <david.sherwood@arm.com>
14185 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
14186 * explow.c (hard_function_value): Likewise.
14187 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
14188 convert_to_mode call outside the loop.
14189 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
14190 for the mode iterator. Require the mode specified by max_pieces
14192 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
14194 (copy_blkmode_to_reg): Likewise.
14195 (set_storage_via_setmem): Likewise.
14196 * optabs.c (prepare_cmp_insn): Likewise.
14197 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14198 * stor-layout.c (finish_bitfield_representative): Likewise.
14200 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14201 Alan Hayward <alan.hayward@arm.com>
14202 David Sherwood <david.sherwood@arm.com>
14204 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
14205 * expr.c (convert_move): Use them.
14206 (convert_modes): Likewise.
14207 (store_expr_with_bounds): Likewise.
14209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14210 Alan Hayward <alan.hayward@arm.com>
14211 David Sherwood <david.sherwood@arm.com>
14213 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
14214 parameter for the mode of "x". Remove the "known_x", "known_mode"
14215 and "known_ret" arguments. Change the type of the mode argument
14216 to scalar_int_mode.
14217 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
14218 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
14219 (reg_num_sign_bit_copies_for_combine): Likewise.
14220 * rtlanal.c (nonzero_bits1): Likewise.
14221 (num_sign_bit_copies1): Likewise.
14222 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
14223 (reg_num_sign_bit_copies_general): Likewise.
14224 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
14225 (reg_nonzero_bits_general): Likewise.
14227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14228 Alan Hayward <alan.hayward@arm.com>
14229 David Sherwood <david.sherwood@arm.com>
14231 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
14232 than in subroutines. Return 1 for non-integer modes.
14233 (cached_num_sign_bit_copies): Change the type of the mode parameter
14234 to scalar_int_mode.
14235 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
14236 classes. Handle CONST_INT_P first and then check whether X also
14237 has a scalar integer mode. Check the same thing for inner registers
14238 of a SUBREG and for values that are being extended or truncated.
14240 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14241 Alan Hayward <alan.hayward@arm.com>
14242 David Sherwood <david.sherwood@arm.com>
14244 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
14245 in subroutines. Return the mode mask for non-integer modes.
14246 (cached_nonzero_bits): Change the type of the mode parameter
14247 to scalar_int_mode.
14248 (nonzero_bits1): Likewise. Remove early exit for other mode
14249 classes. Handle CONST_INT_P first and then check whether X
14250 also has a scalar integer mode.
14252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14253 Alan Hayward <alan.hayward@arm.com>
14254 David Sherwood <david.sherwood@arm.com>
14256 * expr.c (widest_int_mode_for_size): Make the comment match the code.
14257 Return a scalar_int_mode and assert that the size is greater than
14259 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
14260 (op_by_pieces_d::op_by_pieces_d): Likewise.
14261 (op_by_pieces_d::run): Likewise.
14262 (can_store_by_pieces): Likewise.
14264 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14265 Alan Hayward <alan.hayward@arm.com>
14266 David Sherwood <david.sherwood@arm.com>
14268 * combine.c (extract_left_shift): Add a mode argument and update
14270 (make_compound_operation_int): Change the type of the mode parameter
14271 to scalar_int_mode and update the call to extract_left_shift.
14273 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14274 Alan Hayward <alan.hayward@arm.com>
14275 David Sherwood <david.sherwood@arm.com>
14277 * combine.c (simplify_and_const_int): Change the type of the mode
14278 parameter to scalar_int_mode.
14279 (simplify_and_const_int_1): Likewise. Update recursive call.
14281 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14282 Alan Hayward <alan.hayward@arm.com>
14283 David Sherwood <david.sherwood@arm.com>
14285 * combine.c (simplify_compare_const): Check that the mode is a
14286 scalar_int_mode (rather than VOIDmode) before testing its
14288 (simplify_comparison): Move COMPARISON_P handling out of the
14289 loop and restrict the latter part of the loop to scalar_int_modes.
14290 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
14291 and when considering SUBREG_REGs. Use is_int_mode instead of
14292 checking GET_MODE_CLASS against MODE_INT.
14294 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14295 Alan Hayward <alan.hayward@arm.com>
14296 David Sherwood <david.sherwood@arm.com>
14298 * combine.c (try_widen_shift_mode): Move check for equal modes to...
14299 (simplify_shift_const_1): ...here. Use scalar_int_mode for
14300 shift_unit_mode and for modes involved in scalar shifts.
14302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14303 Alan Hayward <alan.hayward@arm.com>
14304 David Sherwood <david.sherwood@arm.com>
14306 * combine.c (force_int_to_mode): New function, split out from...
14307 (force_to_mode): ...here. Keep xmode up-to-date and use it
14308 instead of GET_MODE (x).
14310 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14311 Alan Hayward <alan.hayward@arm.com>
14312 David Sherwood <david.sherwood@arm.com>
14314 * optabs-query.h (extraction_insn::struct_mode): Change type to
14315 opt_scalar_int_mode and update comment.
14316 (extraction_insn::field_mode): Change type to scalar_int_mode.
14317 (extraction_insn::pos_mode): Likewise.
14318 * combine.c (make_extraction): Update accordingly.
14319 * optabs-query.c (get_traditional_extraction_insn): Likewise.
14320 (get_optab_extraction_insn): Likewise.
14321 * recog.c (simplify_while_replacing): Likewise.
14322 * expmed.c (narrow_bit_field_mem): Change the type of the mode
14323 parameter to opt_scalar_int_mode.
14325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14326 Alan Hayward <alan.hayward@arm.com>
14327 David Sherwood <david.sherwood@arm.com>
14329 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
14330 to a scalar_int_mode instead of a machine_mode.
14331 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
14332 (get_best_mode): Return a boolean and use a pointer argument to store
14333 the selected mode. Replace the limit mode parameter with a bit limit.
14334 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
14335 for the values returned by bit_field_mode_iterator::next_mode.
14336 (store_bit_field): Update call to get_best_mode.
14337 (store_fixed_bit_field): Likewise.
14338 (extract_fixed_bit_field): Likewise.
14339 * expr.c (optimize_bitfield_assignment_op): Likewise.
14340 * fold-const.c (optimize_bit_field_compare): Likewise.
14341 (fold_truth_andor_1): Likewise.
14342 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
14343 Update for new type of m_mode.
14344 (get_best_mode): As above.
14346 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14347 Alan Hayward <alan.hayward@arm.com>
14348 David Sherwood <david.sherwood@arm.com>
14350 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
14351 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
14352 (store_bit_field): Check is_a <scalar_int_mode> before calling
14353 strict_volatile_bitfield_p.
14354 (extract_bit_field): Likewise.
14356 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14357 Alan Hayward <alan.hayward@arm.com>
14358 David Sherwood <david.sherwood@arm.com>
14360 * target.def (cstore_mode): Return a scalar_int_mode.
14361 * doc/tm.texi: Regenerate.
14362 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
14363 * targhooks.h (default_cstore_mode): Likewise.
14364 * targhooks.c (default_cstore_mode): Likewise, using a forced
14366 * expmed.c (emit_cstore): Expect the target of the cstore to be
14369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14370 Alan Hayward <alan.hayward@arm.com>
14371 David Sherwood <david.sherwood@arm.com>
14373 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
14375 (niter_desc): Likewise mode.
14376 (iv_analyze): Add a mode parameter.
14378 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
14379 and change its type to scalar_int_mode.
14380 * loop-iv.c: Update commentary at head of file.
14381 (iv_constant): Pass the mode paraeter before the rtx it describes
14382 and change its type to scalar_int_mode. Remove VOIDmode handling.
14383 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
14384 (iv_extend): Likewise.
14385 (shorten_into_mode): Likewise.
14386 (iv_add): Use scalar_int_mode.
14387 (iv_mult): Likewise.
14388 (iv_shift): Likewise.
14389 (canonicalize_iv_subregs): Likewise.
14390 (get_biv_step_1): Pass the outer_mode parameter before the rtx
14391 it describes and change its mode to scalar_int_mode. Also change
14392 the type of the returned inner_mode to scalar_int_mode.
14393 (get_biv_step): Likewise, turning outer_mode from a pointer
14394 into a direct parameter. Update call to get_biv_step_1.
14395 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
14396 iv_constant and get_biv_step.
14397 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
14398 and change its type to scalar_int_mode. Don't initialise iv->mode
14399 to VOIDmode and remove later checks for its still being VOIDmode.
14400 Update calls to iv_analyze_op and iv_analyze_expr. Check
14401 is_a <scalar_int_mode> when changing the mode under consideration.
14402 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
14403 Update call to iv_analyze_expr.
14404 (iv_analyze_op): Add a mode parameter. Reject subregs whose
14405 inner register is not also a scalar_int_mode. Update call to
14407 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
14408 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
14409 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
14410 separate mode class checks. Update calls to iv_analyze. Remove
14411 fix-up of VOIDmodes after iv_analyze_biv.
14412 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
14413 don't have a scalar_int_mode. Update call to biv_p.
14415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14416 Alan Hayward <alan.hayward@arm.com>
14417 David Sherwood <david.sherwood@arm.com>
14419 * cfgexpand.c (convert_debug_memory_address): Use
14420 as_a <scalar_int_mode>.
14421 * combine.c (expand_compound_operation): Likewise.
14422 (make_extraction): Likewise.
14423 (change_zero_ext): Likewise.
14424 (simplify_comparison): Likewise.
14425 * cse.c (cse_insn): Likewise.
14426 * dwarf2out.c (minmax_loc_descriptor): Likewise.
14427 (mem_loc_descriptor): Likewise.
14428 (loc_descriptor): Likewise.
14429 * expmed.c (init_expmed_one_mode): Likewise.
14430 (synth_mult): Likewise.
14431 (emit_store_flag_1): Likewise.
14432 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
14433 of a comparison with size.
14434 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
14435 (reduce_to_bit_field_precision): Likewise.
14436 * function.c (expand_function_end): Likewise.
14437 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
14438 * loop-doloop.c (doloop_modify): Likewise.
14439 * optabs.c (expand_binop): Likewise.
14440 (expand_unop): Likewise.
14441 (expand_copysign_absneg): Likewise.
14442 (prepare_cmp_insn): Likewise.
14443 (maybe_legitimize_operand): Likewise.
14444 * recog.c (const_scalar_int_operand): Likewise.
14445 * rtlanal.c (get_address_mode): Likewise.
14446 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14447 (simplify_cond_clz_ctz): Likewise.
14448 * tree-nested.c (get_nl_goto_field): Likewise.
14449 * tree.c (build_vector_type_for_mode): Likewise.
14450 * var-tracking.c (use_narrower_mode): Likewise.
14452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14453 Alan Hayward <alan.hayward@arm.com>
14454 David Sherwood <david.sherwood@arm.com>
14456 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
14457 * builtins.c (expand_builtin_signbit): Use it.
14458 * cfgexpand.c (expand_debug_expr): Likewise.
14459 * dojump.c (do_jump): Likewise.
14460 (do_compare_and_jump): Likewise.
14461 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
14462 * expmed.c (make_tree): Likewise.
14463 * expr.c (expand_expr_real_2): Likewise.
14464 (expand_expr_real_1): Likewise.
14465 (try_casesi): Likewise.
14466 * fold-const-call.c (fold_const_call_ss): Likewise.
14467 * fold-const.c (unextend): Likewise.
14468 (extract_muldiv_1): Likewise.
14469 (fold_single_bit_test): Likewise.
14470 (native_encode_int): Likewise.
14471 (native_encode_string): Likewise.
14472 (native_interpret_int): Likewise.
14473 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
14474 * internal-fn.c (expand_addsub_overflow): Likewise.
14475 (expand_neg_overflow): Likewise.
14476 (expand_mul_overflow): Likewise.
14477 (expand_arith_overflow): Likewise.
14478 * match.pd: Likewise.
14479 * stor-layout.c (layout_type): Likewise.
14480 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14481 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
14482 * tree-ssanames.c (get_range_info): Likewise.
14483 * tree-switch-conversion.c (array_value_type) Likewise.
14484 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
14485 (vect_recog_divmod_pattern): Likewise.
14486 (vect_recog_mixed_size_cond_pattern): Likewise.
14487 * tree-vrp.c (extract_range_basic): Likewise.
14488 (simplify_float_conversion_using_ranges): Likewise.
14489 * tree.c (int_fits_type_p): Likewise.
14490 * ubsan.c (instrument_bool_enum_load): Likewise.
14491 * varasm.c (mergeable_string_section): Likewise.
14492 (narrowing_initializer_constant_valid_p): Likewise.
14493 (output_constant): Likewise.
14495 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14496 Alan Hayward <alan.hayward@arm.com>
14497 David Sherwood <david.sherwood@arm.com>
14499 * machmode.h (NARROWEST_INT_MODE): New macro.
14500 * expr.c (alignment_for_piecewise_move): Use it instead of
14501 GET_CLASS_NARROWEST_MODE (MODE_INT).
14502 (push_block): Likewise.
14503 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
14505 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14508 Alan Hayward <alan.hayward@arm.com>
14509 David Sherwood <david.sherwood@arm.com>
14511 * postreload.c (move2add_valid_value_p): Change the type of the
14512 mode parameter to scalar_int_mode.
14513 (move2add_use_add2_insn): Add a mode parameter and use it instead
14515 (move2add_use_add3_insn): Likewise.
14516 (reload_cse_move2add): Update accordingly.
14518 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14519 Alan Hayward <alan.hayward@arm.com>
14520 David Sherwood <david.sherwood@arm.com>
14522 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
14524 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
14525 * optabs.c (expand_unop): Likewise.
14527 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14528 Alan Hayward <alan.hayward@arm.com>
14529 David Sherwood <david.sherwood@arm.com>
14531 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
14532 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
14533 (popcount_loc_descriptor): Likewise.
14534 (bswap_loc_descriptor): Likewise.
14535 (rotate_loc_descriptor): Likewise.
14536 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
14537 calling the functions above.
14539 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14540 Alan Hayward <alan.hayward@arm.com>
14541 David Sherwood <david.sherwood@arm.com>
14543 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
14545 (try_combine): Likewise.
14546 (simplify_if_then_else): Likewise.
14547 * cse.c (cse_insn): Likewise.
14548 * dwarf2out.c (mem_loc_descriptor): Likewise.
14549 * emit-rtl.c (gen_lowpart_common): Likewise.
14550 * simplify-rtx.c (simplify_truncation): Likewise.
14551 (simplify_binary_operation_1): Likewise.
14552 (simplify_const_relational_operation): Likewise.
14553 (simplify_ternary_operation): Likewise.
14554 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
14556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14557 Alan Hayward <alan.hayward@arm.com>
14558 David Sherwood <david.sherwood@arm.com>
14560 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
14561 * reload.c (push_reload): Likewise.
14562 (find_reloads): Likewise.
14564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14565 Alan Hayward <alan.hayward@arm.com>
14566 David Sherwood <david.sherwood@arm.com>
14568 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
14569 (make_compound_operation_int): Likewise.
14570 (change_zero_ext): Likewise.
14571 * expr.c (convert_move): Likewise.
14572 (convert_modes): Likewise.
14573 * fwprop.c (forward_propagate_subreg): Likewise.
14574 * loop-iv.c (get_biv_step_1): Likewise.
14575 * optabs.c (widen_operand): Likewise.
14576 * postreload.c (move2add_valid_value_p): Likewise.
14577 * recog.c (simplify_while_replacing): Likewise.
14578 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14579 (simplify_binary_operation_1): Likewise. Remove redundant
14580 mode equality check.
14582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14583 Alan Hayward <alan.hayward@arm.com>
14584 David Sherwood <david.sherwood@arm.com>
14586 * combine.c (combine_simplify_rtx): Add checks for
14587 is_a <scalar_int_mode>.
14588 (simplify_if_then_else): Likewise.
14589 (make_field_assignment): Likewise.
14590 (simplify_comparison): Likewise.
14591 * ifcvt.c (noce_try_bitop): Likewise.
14592 * loop-invariant.c (canonicalize_address_mult): Likewise.
14593 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14595 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14596 Alan Hayward <alan.hayward@arm.com>
14597 David Sherwood <david.sherwood@arm.com>
14599 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
14600 is_a <scalar_int_mode> instead of != BLKmode.
14602 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14603 Alan Hayward <alan.hayward@arm.com>
14604 David Sherwood <david.sherwood@arm.com>
14606 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
14607 instead of != VOIDmode.
14608 * combine.c (if_then_else_cond): Likewise.
14609 (change_zero_ext): Likewise.
14610 * dwarf2out.c (mem_loc_descriptor): Likewise.
14611 (loc_descriptor): Likewise.
14612 * rtlanal.c (canonicalize_condition): Likewise.
14613 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
14615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14616 Alan Hayward <alan.hayward@arm.com>
14617 David Sherwood <david.sherwood@arm.com>
14619 * simplify-rtx.c (simplify_binary_operation_1): Use
14620 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
14622 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14623 Alan Hayward <alan.hayward@arm.com>
14624 David Sherwood <david.sherwood@arm.com>
14626 * wide-int.h (int_traits<unsigned char>) New class.
14627 (int_traits<unsigned short>) Likewise.
14628 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
14629 Use GET_MODE_UNIT_PRECISION and remove redundant test for
14631 * combine.c (set_nonzero_bits_and_sign_copies): Use
14632 is_a <scalar_int_mode>.
14633 (find_split_point): Likewise.
14634 (combine_simplify_rtx): Likewise.
14635 (simplify_logical): Likewise.
14636 (expand_compound_operation): Likewise.
14637 (expand_field_assignment): Likewise.
14638 (make_compound_operation): Likewise.
14639 (extended_count): Likewise.
14640 (change_zero_ext): Likewise.
14641 (simplify_comparison): Likewise.
14642 * dwarf2out.c (scompare_loc_descriptor): Likewise.
14643 (ucompare_loc_descriptor): Likewise.
14644 (minmax_loc_descriptor): Likewise.
14645 (mem_loc_descriptor): Likewise.
14646 (loc_descriptor): Likewise.
14647 * expmed.c (init_expmed_one_mode): Likewise.
14648 * lra-constraints.c (lra_constraint_offset): Likewise.
14649 * optabs.c (prepare_libcall_arg): Likewise.
14650 * postreload.c (move2add_note_store): Likewise.
14651 * reload.c (operands_match_p): Likewise.
14652 * rtl.h (load_extend_op): Likewise.
14653 * rtlhooks.c (gen_lowpart_general): Likewise.
14654 * simplify-rtx.c (simplify_truncation): Likewise.
14655 (simplify_unary_operation_1): Likewise.
14656 (simplify_binary_operation_1): Likewise.
14657 (simplify_const_binary_operation): Likewise.
14658 (simplify_const_relational_operation): Likewise.
14659 (simplify_subreg): Likewise.
14660 * stor-layout.c (bitwise_mode_for_mode): Likewise.
14661 * var-tracking.c (adjust_mems): Likewise.
14662 (prepare_call_arguments): Likewise.
14664 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14665 Alan Hayward <alan.hayward@arm.com>
14666 David Sherwood <david.sherwood@arm.com>
14668 * machmode.h (is_int_mode): New fuction.
14669 * combine.c (find_split_point): Use it.
14670 (combine_simplify_rtx): Likewise.
14671 (simplify_if_then_else): Likewise.
14672 (simplify_set): Likewise.
14673 (simplify_shift_const_1): Likewise.
14674 (simplify_comparison): Likewise.
14675 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
14676 * cse.c (notreg_cost): Likewise.
14677 (cse_insn): Likewise.
14678 * cselib.c (cselib_lookup_1): Likewise.
14679 * dojump.c (do_jump_1): Likewise.
14680 (do_compare_rtx_and_jump): Likewise.
14681 * dse.c (get_call_args): Likewise.
14682 * dwarf2out.c (rtl_for_decl_init): Likewise.
14683 (native_encode_initializer): Likewise.
14684 * expmed.c (emit_store_flag_1): Likewise.
14685 (emit_store_flag): Likewise.
14686 * expr.c (convert_modes): Likewise.
14687 (store_field): Likewise.
14688 (expand_expr_real_1): Likewise.
14689 * fold-const.c (fold_read_from_constant_string): Likewise.
14690 * gimple-ssa-sprintf.c (get_format_string): Likewise.
14691 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
14692 * optabs.c (expand_binop): Likewise.
14693 (expand_unop): Likewise.
14694 (expand_abs_nojump): Likewise.
14695 (expand_one_cmpl_abs_nojump): Likewise.
14696 * simplify-rtx.c (mode_signbit_p): Likewise.
14697 (val_signbit_p): Likewise.
14698 (val_signbit_known_set_p): Likewise.
14699 (val_signbit_known_clear_p): Likewise.
14700 (simplify_relational_operation_1): Likewise.
14701 * tree.c (vector_type_mode): Likewise.
14703 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14704 Alan Hayward <alan.hayward@arm.com>
14705 David Sherwood <david.sherwood@arm.com>
14707 * machmode.h (smallest_mode_for_size): Fix formatting.
14708 (smallest_int_mode_for_size): New function.
14709 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
14710 instead of smallest_mode_for_size.
14711 * combine.c (make_extraction): Likewise.
14712 * config/arc/arc.c (arc_expand_movmem): Likewise.
14713 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
14714 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14715 * config/s390/s390.c (s390_expand_insv): Likewise.
14716 * config/sparc/sparc.c (assign_int_registers): Likewise.
14717 * config/spu/spu.c (spu_function_value): Likewise.
14718 (spu_function_arg): Likewise.
14719 * coverage.c (get_gcov_type): Likewise.
14720 (get_gcov_unsigned_t): Likewise.
14721 * dse.c (find_shift_sequence): Likewise.
14722 * expmed.c (store_bit_field_1): Likewise.
14723 * expr.c (convert_move): Likewise.
14724 (store_field): Likewise.
14725 * internal-fn.c (expand_arith_overflow): Likewise.
14726 * optabs-query.c (get_best_extraction_insn): Likewise.
14727 * optabs.c (expand_twoval_binop_libfunc): Likewise.
14728 * stor-layout.c (layout_type): Likewise.
14729 (initialize_sizetypes): Likewise.
14730 * targhooks.c (default_get_mask_mode): Likewise.
14731 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
14733 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14734 Alan Hayward <alan.hayward@arm.com>
14735 David Sherwood <david.sherwood@arm.com>
14737 * machmode.h (opt_mode::else_blk): New function.
14738 (int_mode_for_mode): Declare.
14739 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
14740 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
14742 * cfgexpand.c (expand_debug_expr): Likewise.
14743 * combine.c (gen_lowpart_or_truncate): Likewise.
14744 (gen_lowpart_for_combine): Likewise.
14745 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
14746 * config/avr/avr.c (avr_to_int_mode): Likewise.
14747 (avr_out_plus_1): Likewise.
14748 (avr_out_plus): Likewise.
14749 (avr_out_round): Likewise.
14750 * config/i386/i386.c (ix86_split_to_parts): Likewise.
14751 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14752 (s390_expand_vcond): Likewise.
14753 * config/spu/spu.c (spu_split_immediate): Likewise.
14754 (spu_expand_mov): Likewise.
14755 * dse.c (get_stored_val): Likewise.
14756 * expmed.c (store_bit_field_1): Likewise.
14757 (convert_extracted_bit_field): Use int_mode_for_mode instead of
14759 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
14760 (extract_low_bits): Likewise.
14761 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
14762 handling rather than repeating the check.
14763 (emit_group_store): Likewise.
14764 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
14765 * optabs.c (expand_absneg_bit): Likewise.
14766 (expand_copysign_absneg): Likewise.
14767 (expand_copysign_bit): Likewise.
14768 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
14769 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
14770 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
14771 * var-tracking.c (prepare_call_arguments): Likewise.
14772 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
14773 int_mode_for_mode instead of mode_for_size.
14774 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14776 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14777 Alan Hayward <alan.hayward@arm.com>
14778 David Sherwood <david.sherwood@arm.com>
14780 * machmode.h (int_mode_for_size): New function.
14781 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
14782 instead of mode_for_size.
14783 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
14785 * combine.c (expand_field_assignment): Use int_mode_for_size
14786 instead of mode_for_size.
14787 (make_extraction): Likewise.
14788 (simplify_shift_const_1): Likewise.
14789 (simplify_comparison): Likewise.
14790 * dojump.c (do_jump): Likewise.
14791 * dwarf2out.c (mem_loc_descriptor): Likewise.
14792 * emit-rtl.c (init_derived_machine_modes): Likewise.
14793 * expmed.c (flip_storage_order): Likewise.
14794 (convert_extracted_bit_field): Likewise.
14795 * expr.c (copy_blkmode_from_reg): Likewise.
14796 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
14797 * internal-fn.c (expand_mul_overflow): Likewise.
14798 * lower-subreg.c (simple_move): Likewise.
14799 * optabs-libfuncs.c (init_optabs): Likewise.
14800 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14801 * tree.c (vector_type_mode): Likewise.
14802 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
14803 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
14804 * tree-vect-generic.c (expand_vector_parallel): Likewise.
14805 * tree-vect-stmts.c (vectorizable_load): Likewise.
14806 (vectorizable_store): Likewise.
14808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14809 Alan Hayward <alan.hayward@arm.com>
14810 David Sherwood <david.sherwood@arm.com>
14812 * coretypes.h (pod_mode): New type.
14813 (scalar_int_mode_pod): New typedef.
14814 * machmode.h (pod_mode): New class.
14815 (int_n_data_t::m): Change type to scalar_int_mode_pod.
14816 * genmodes.c (emit_mode_int_n): Update accordingly.
14817 * lower-subreg.h (target_lower_subreg): Change type to
14818 scalar_int_mode_pod.
14819 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
14820 scalar_int_mode_pod.
14822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14823 Alan Hayward <alan.hayward@arm.com>
14824 David Sherwood <david.sherwood@arm.com>
14826 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
14827 machine_mode to scalar_int_mode.
14828 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
14829 (rs6000_option_override_internal): Remove cast to int.
14830 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
14831 machine_mode to scalar_int_mode.
14832 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
14833 (rs6000_option_override_internal): Remove cast to int.
14834 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
14835 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
14837 * config/s390/s390.c (s390_expand_builtin): Likewise.
14838 * coretypes.h (scalar_int_mode): New type.
14839 (opt_scalar_int_mode): New typedef.
14840 * machmode.h (scalar_int_mode): New class.
14841 (scalar_int_mode::includes_p): New function.
14842 (byte_mode): Change type to scalar_int_mode.
14843 (word_mode): Likewise.
14844 (ptr_mode): Likewise.
14845 * emit-rtl.c (byte_mode): Likewise.
14846 (word_mode): Likewise.
14847 (ptr_mode): Likewise.
14848 (init_derived_machine_modes): Update accordingly.
14849 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
14850 and MODE_PARTIAL_INT.
14851 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
14852 opt_scalar_int_mode.
14854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14855 Alan Hayward <alan.hayward@arm.com>
14856 David Sherwood <david.sherwood@arm.com>
14858 * target.def (libgcc_floating_mode_supported_p): Take a
14860 * doc/tm.texi: Regenerate.
14861 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
14863 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
14864 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
14867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14868 Alan Hayward <alan.hayward@arm.com>
14869 David Sherwood <david.sherwood@arm.com>
14871 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
14872 * doc/tm.texi: Regenerate.
14873 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
14874 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
14875 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
14876 * targhooks.h (default_floatn_mode): Likewise.
14877 * targhooks.c (default_floatn_mode): Likewise.
14878 * tree.c (build_common_tree_nodes): Update accordingly.
14880 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14881 Alan Hayward <alan.hayward@arm.com>
14882 David Sherwood <david.sherwood@arm.com>
14884 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
14885 (mode_iterator::iterate_p): Likewise.
14886 (mode_iterator::get_wider): Likewise.
14887 * expr.c (init_expr_target): Use opt_scalar_float_mode.
14889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14890 Alan Hayward <alan.hayward@arm.com>
14891 David Sherwood <david.sherwood@arm.com>
14893 * coretypes.h (opt_scalar_float_mode): New typedef.
14894 * machmode.h (float_mode_for_size): New function.
14895 * emit-rtl.c (double_mode): Delete.
14896 (init_emit_once): Use float_mode_for_size.
14897 * stor-layout.c (layout_type): Likewise.
14898 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
14900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14901 Alan Hayward <alan.hayward@arm.com>
14902 David Sherwood <david.sherwood@arm.com>
14904 * output.h (assemble_real): Take a scalar_float_mode.
14905 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
14906 * config/arm/arm.md (consttable_4): Likewise.
14907 (consttable_8): Likewise.
14908 (consttable_16): Likewise.
14909 * config/mips/mips.md (consttable_float): Likewise.
14910 * config/s390/s390.c (s390_output_pool_entry): Likewise.
14911 * varasm.c (assemble_real): Take a scalar_float_mode.
14912 (output_constant_pool_2): Update accordingly.
14913 (output_constant): Likewise.
14915 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14916 Alan Hayward <alan.hayward@arm.com>
14917 David Sherwood <david.sherwood@arm.com>
14919 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
14920 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
14921 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
14922 (native_encode_real): Likewise.
14923 (native_interpret_real): Likewise.
14924 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
14925 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14928 Alan Hayward <alan.hayward@arm.com>
14929 David Sherwood <david.sherwood@arm.com>
14931 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
14932 <scalar_float_mode>. Simplify.
14933 (gen_extend_conv_libfunc): Likewise.
14935 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14936 Alan Hayward <alan.hayward@arm.com>
14937 David Sherwood <david.sherwood@arm.com>
14939 * coretypes.h (scalar_float_mode): New type.
14940 * machmode.h (mode_traits::from_int): Use machine_mode if
14941 USE_ENUM_MODES is defined.
14942 (is_a): New function.
14944 (dyn_cast): Likewise.
14945 (scalar_float_mode): New class.
14946 (scalar_float_mode::includes_p): New function.
14947 (is_float_mode): Likewise.
14948 * gdbhooks.py (MachineModePrinter): New class.
14949 (build_pretty_printer): Use it for scalar_float_mode.
14950 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
14951 (format_helper::format_helper): Turn into a template.
14952 * genmodes.c (get_mode_class): New function.
14953 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
14954 or machine_mode if none.
14955 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
14956 as_a <scalar_float_mode>.
14957 * dwarf2out.c (mem_loc_descriptor): Likewise.
14958 (insert_float): Likewise.
14959 (add_const_value_attribute): Likewise.
14960 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14961 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
14962 (expand_unop): Update accordingly.
14963 (expand_abs_nojump): Likewise.
14964 (expand_copysign_absneg): Take a scalar_float_mode.
14965 (expand_copysign_bit): Likewise.
14966 (expand_copysign): Update accordingly.
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_mode): New class.
14973 * machmode.h (opt_mode): Likewise.
14974 (opt_mode::else_void): New function.
14975 (opt_mode::require): Likewise.
14976 (opt_mode::exists): Likewise.
14977 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
14978 (GET_MODE_2XWIDER_MODE): Likewise.
14979 (mode_iterator::get_wider): Update accordingly.
14980 (mode_iterator::get_2xwider): Likewise.
14981 (mode_iterator::get_known_wider): Likewise, turning into a template.
14982 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
14983 forcing a wider mode to exist.
14984 * config/cr16/cr16.h (LONG_REG_P): Likewise.
14985 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14986 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
14987 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
14988 * lower-subreg.c (init_lower_subreg): Likewise.
14989 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
14990 on the final iteration.
14991 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
14992 a wider mode exists before asking for a move pattern.
14993 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
14994 forcing a wider mode to exist.
14995 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
14996 returning false if no such mode exists.
14997 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
14998 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
14999 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
15000 Avoid checking for a MODE_INT if we already know the mode is not a
15002 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
15003 forcing a wider mode to exist.
15004 (expmed_mult_highpart_optab): Likewise.
15005 (expmed_mult_highpart): Likewise.
15006 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
15008 * lto-streamer-in.c (lto_input_mode_table): Likewise.
15009 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
15010 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15011 * internal-fn.c (expand_mul_overflow): Update use of
15012 GET_MODE_2XWIDER_MODE.
15013 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15014 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
15015 GET_MODE_WIDER_MODE.
15016 (convert_plusminus_to_widen): Likewise.
15017 * tree-switch-conversion.c (array_value_type): Likewise.
15018 * var-tracking.c (emit_note_insn_var_location): Likewise.
15019 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15020 Return false inside rather than outside the loop if no wider mode
15022 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
15023 and GET_MODE_2XWIDER_MODE
15024 (can_compare_p): Use else_void.
15025 * gdbhooks.py (OptMachineModePrinter): New class.
15026 (build_pretty_printer): Use it for opt_mode.
15028 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15029 Alan Hayward <alan.hayward@arm.com>
15030 David Sherwood <david.sherwood@arm.com>
15032 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
15033 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
15035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15036 Alan Hayward <alan.hayward@arm.com>
15037 David Sherwood <david.sherwood@arm.com>
15039 * machmode.h (mode_traits): New structure.
15040 (get_narrowest_mode): New function.
15041 (mode_iterator::start): Likewise.
15042 (mode_iterator::iterate_p): Likewise.
15043 (mode_iterator::get_wider): Likewise.
15044 (mode_iterator::get_known_wider): Likewise.
15045 (mode_iterator::get_2xwider): Likewise.
15046 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
15047 (FOR_EACH_MODE): Likewise.
15048 (FOR_EACH_MODE_FROM): Likewise.
15049 (FOR_EACH_MODE_UNTIL): Likewise.
15050 (FOR_EACH_WIDER_MODE): Likewise.
15051 (FOR_EACH_2XWIDER_MODE): Likewise.
15052 * builtins.c (expand_builtin_strlen): Use new mode iterators.
15053 * combine.c (simplify_comparison): Likewise
15054 * config/i386/i386.c (type_natural_mode): Likewise.
15055 * cse.c (cse_insn): Likewise.
15056 * dse.c (find_shift_sequence): Likewise.
15057 * emit-rtl.c (init_derived_machine_modes): Likewise.
15058 (init_emit_once): Likewise.
15059 * explow.c (hard_function_value): Likewise.
15060 * expmed.c (extract_fixed_bit_field_1): Likewise.
15061 (extract_bit_field_1): Likewise.
15062 (expand_divmod): Likewise.
15063 (emit_store_flag_1): Likewise.
15064 * expr.c (init_expr_target): Likewise.
15065 (convert_move): Likewise.
15066 (alignment_for_piecewise_move): Likewise.
15067 (widest_int_mode_for_size): Likewise.
15068 (emit_block_move_via_movmem): Likewise.
15069 (copy_blkmode_to_reg): Likewise.
15070 (set_storage_via_setmem): Likewise.
15071 (compress_float_constant): Likewise.
15072 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15073 * optabs-query.c (get_best_extraction_insn): Likewise.
15074 * optabs.c (expand_binop): Likewise.
15075 (expand_twoval_unop): Likewise.
15076 (expand_twoval_binop): Likewise.
15077 (widen_leading): Likewise.
15078 (widen_bswap): Likewise.
15079 (expand_parity): Likewise.
15080 (expand_unop): Likewise.
15081 (prepare_cmp_insn): Likewise.
15082 (prepare_float_lib_cmp): Likewise.
15083 (expand_float): Likewise.
15084 (expand_fix): Likewise.
15085 (expand_sfix_optab): Likewise.
15086 * postreload.c (move2add_use_add2_insn): Likewise.
15087 * reg-stack.c (reg_to_stack): Likewise.
15088 * reginfo.c (choose_hard_reg_mode): Likewise.
15089 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15090 * stor-layout.c (mode_for_size): Likewise.
15091 (smallest_mode_for_size): Likewise.
15092 (mode_for_vector): Likewise.
15093 (finish_bitfield_representative): Likewise.
15094 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
15095 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
15096 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15097 * var-tracking.c (prepare_call_arguments): Likewise.
15099 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15100 Alan Hayward <alan.hayward@arm.com>
15101 David Sherwood <david.sherwood@arm.com>
15103 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
15104 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
15105 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
15106 * machmode.h (mode_size): Move earlier in file.
15107 (mode_precision): Likewise.
15108 (mode_inner): Likewise.
15109 (mode_nunits): Likewise.
15110 (mode_unit_size): Likewise.
15111 (unit_unit_precision): Likewise.
15112 (mode_wider): Likewise.
15113 (mode_2xwider): Likewise.
15114 (machine_mode): New class.
15115 (mode_to_bytes): New function.
15116 (mode_to_bits): Likewise.
15117 (mode_to_precision): Likewise.
15118 (mode_to_inner): Likewise.
15119 (mode_to_unit_size): Likewise.
15120 (mode_to_unit_precision): Likewise.
15121 (mode_to_nunits): Likewise.
15122 (GET_MODE_SIZE): Use mode_to_bytes.
15123 (GET_MODE_BITSIZE): Use mode_to_bits.
15124 (GET_MODE_PRECISION): Use mode_to_precision.
15125 (GET_MODE_INNER): Use mode_to_inner.
15126 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
15127 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
15128 (GET_MODE_NUNITS): Use mode_to_nunits.
15129 * system.h (ALWAYS_INLINE): New macro.
15130 * config/powerpcspe/powerpcspe-c.c
15131 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
15132 int for arg1_mode and arg2_mode.
15134 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15135 Alan Hayward <alan.hayward@arm.com>
15136 David Sherwood <david.sherwood@arm.com>
15138 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
15139 Prefix mode names with E_ in case statements.
15140 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15141 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
15142 (aarch64_split_simd_move): Likewise.
15143 (aarch64_gen_storewb_pair): Likewise.
15144 (aarch64_gen_loadwb_pair): Likewise.
15145 (aarch64_gen_store_pair): Likewise.
15146 (aarch64_gen_load_pair): Likewise.
15147 (aarch64_get_condition_code_1): Likewise.
15148 (aarch64_constant_pool_reload_icode): Likewise.
15149 (get_rsqrte_type): Likewise.
15150 (get_rsqrts_type): Likewise.
15151 (get_recpe_type): Likewise.
15152 (get_recps_type): Likewise.
15153 (aarch64_gimplify_va_arg_expr): Likewise.
15154 (aarch64_simd_container_mode): Likewise.
15155 (aarch64_emit_load_exclusive): Likewise.
15156 (aarch64_emit_store_exclusive): Likewise.
15157 (aarch64_expand_compare_and_swap): Likewise.
15158 (aarch64_gen_atomic_cas): Likewise.
15159 (aarch64_emit_bic): Likewise.
15160 (aarch64_emit_atomic_swap): Likewise.
15161 (aarch64_emit_atomic_load_op): Likewise.
15162 (aarch64_evpc_trn): Likewise.
15163 (aarch64_evpc_uzp): Likewise.
15164 (aarch64_evpc_zip): Likewise.
15165 (aarch64_evpc_ext): Likewise.
15166 (aarch64_evpc_rev): Likewise.
15167 (aarch64_evpc_dup): Likewise.
15168 (aarch64_gen_ccmp_first): Likewise.
15169 (aarch64_gen_ccmp_next): Likewise.
15170 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15171 (alpha_emit_xfloating_libcall): Likewise.
15172 (emit_insxl): Likewise.
15173 (alpha_arg_type): Likewise.
15174 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
15175 (arc_preferred_simd_mode): Likewise.
15176 (arc_secondary_reload): Likewise.
15177 (get_arc_condition_code): Likewise.
15178 (arc_print_operand): Likewise.
15179 (arc_legitimate_constant_p): Likewise.
15180 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15181 * config/arc/arc.md (casesi_load): Likewise.
15182 (casesi_compact_jump): Likewise.
15183 * config/arc/predicates.md (proper_comparison_operator): Likewise.
15184 (cc_use_register): Likewise.
15185 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15186 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
15187 (arm_init_iwmmxt_builtins): Likewise.
15188 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
15189 (neon_expand_vector_init): Likewise.
15190 (arm_attr_length_move_neon): Likewise.
15191 (maybe_get_arm_condition_code): Likewise.
15192 (arm_emit_vector_const): Likewise.
15193 (arm_preferred_simd_mode): Likewise.
15194 (arm_output_iwmmxt_tinsr): Likewise.
15195 (thumb1_output_casesi): Likewise.
15196 (thumb2_output_casesi): Likewise.
15197 (arm_emit_load_exclusive): Likewise.
15198 (arm_emit_store_exclusive): Likewise.
15199 (arm_expand_compare_and_swap): Likewise.
15200 (arm_evpc_neon_vuzp): Likewise.
15201 (arm_evpc_neon_vzip): Likewise.
15202 (arm_evpc_neon_vrev): Likewise.
15203 (arm_evpc_neon_vtrn): Likewise.
15204 (arm_evpc_neon_vext): Likewise.
15205 (arm_validize_comparison): Likewise.
15206 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
15207 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
15208 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
15209 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
15210 (c6x_preferred_simd_mode): Likewise.
15211 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
15212 (epiphany_rtx_costs): Likewise.
15213 * config/epiphany/predicates.md (proper_comparison_operator):
15215 * config/frv/frv.c (condexec_memory_operand): Likewise.
15216 (frv_emit_move): Likewise.
15217 (output_move_single): Likewise.
15218 (output_condmove_single): Likewise.
15219 (frv_hard_regno_mode_ok): Likewise.
15220 (frv_matching_accg_mode): Likewise.
15221 * config/h8300/h8300.c (split_adds_subs): Likewise.
15222 (h8300_rtx_costs): Likewise.
15223 (h8300_print_operand): Likewise.
15224 (compute_mov_length): Likewise.
15225 (output_logical_op): Likewise.
15226 (compute_logical_op_length): Likewise.
15227 (compute_logical_op_cc): Likewise.
15228 (h8300_shift_needs_scratch_p): Likewise.
15229 (output_a_shift): Likewise.
15230 (compute_a_shift_length): Likewise.
15231 (compute_a_shift_cc): Likewise.
15232 (expand_a_rotate): Likewise.
15233 (output_a_rotate): Likewise.
15234 * config/i386/i386.c (classify_argument): Likewise.
15235 (function_arg_advance_32): Likewise.
15236 (function_arg_32): Likewise.
15237 (function_arg_64): Likewise.
15238 (function_value_64): Likewise.
15239 (ix86_gimplify_va_arg): Likewise.
15240 (ix86_legitimate_constant_p): Likewise.
15241 (put_condition_code): Likewise.
15242 (split_double_mode): Likewise.
15243 (ix86_avx256_split_vector_move_misalign): Likewise.
15244 (ix86_expand_vector_logical_operator): Likewise.
15245 (ix86_split_idivmod): Likewise.
15246 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
15247 (ix86_build_const_vector): Likewise.
15248 (ix86_build_signbit_mask): Likewise.
15249 (ix86_match_ccmode): Likewise.
15250 (ix86_cc_modes_compatible): Likewise.
15251 (ix86_expand_branch): Likewise.
15252 (ix86_expand_sse_cmp): Likewise.
15253 (ix86_expand_sse_movcc): Likewise.
15254 (ix86_expand_int_sse_cmp): Likewise.
15255 (ix86_expand_vec_perm_vpermi2): Likewise.
15256 (ix86_expand_vec_perm): Likewise.
15257 (ix86_expand_sse_unpack): Likewise.
15258 (ix86_expand_int_addcc): Likewise.
15259 (ix86_split_to_parts): Likewise.
15260 (ix86_vectorize_builtin_gather): Likewise.
15261 (ix86_vectorize_builtin_scatter): Likewise.
15262 (avx_vpermilp_parallel): Likewise.
15263 (inline_memory_move_cost): Likewise.
15264 (ix86_tieable_integer_mode_p): Likewise.
15265 (x86_maybe_negate_const_int): Likewise.
15266 (ix86_expand_vector_init_duplicate): Likewise.
15267 (ix86_expand_vector_init_one_nonzero): Likewise.
15268 (ix86_expand_vector_init_one_var): Likewise.
15269 (ix86_expand_vector_init_concat): Likewise.
15270 (ix86_expand_vector_init_interleave): Likewise.
15271 (ix86_expand_vector_init_general): Likewise.
15272 (ix86_expand_vector_set): Likewise.
15273 (ix86_expand_vector_extract): Likewise.
15274 (emit_reduc_half): Likewise.
15275 (ix86_emit_i387_round): Likewise.
15276 (ix86_mangle_type): Likewise.
15277 (ix86_expand_round_sse4): Likewise.
15278 (expand_vec_perm_blend): Likewise.
15279 (canonicalize_vector_int_perm): Likewise.
15280 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
15281 (expand_vec_perm_1): Likewise.
15282 (expand_vec_perm_interleave3): Likewise.
15283 (expand_vec_perm_even_odd_pack): Likewise.
15284 (expand_vec_perm_even_odd_1): Likewise.
15285 (expand_vec_perm_broadcast_1): Likewise.
15286 (ix86_vectorize_vec_perm_const_ok): Likewise.
15287 (ix86_expand_vecop_qihi): Likewise.
15288 (ix86_expand_mul_widen_hilo): Likewise.
15289 (ix86_expand_sse2_abs): Likewise.
15290 (ix86_expand_pextr): Likewise.
15291 (ix86_expand_pinsr): Likewise.
15292 (ix86_preferred_simd_mode): Likewise.
15293 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
15294 * config/i386/sse.md (*andnot<mode>3): Likewise.
15295 (<mask_codefor><code><mode>3<mask_name>): Likewise.
15296 (*<code><mode>3): Likewise.
15297 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
15298 (ia64_expand_atomic_op): Likewise.
15299 (ia64_arg_type): Likewise.
15300 (ia64_mode_to_int): Likewise.
15301 (ia64_scalar_mode_supported_p): Likewise.
15302 (ia64_vector_mode_supported_p): Likewise.
15303 (expand_vec_perm_broadcast): Likewise.
15304 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
15305 (iq2000_function_arg_advance): Likewise.
15306 (iq2000_function_arg): Likewise.
15307 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
15308 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
15309 (m68k_libcall_value): Likewise.
15310 (m68k_function_value): Likewise.
15311 (sched_attr_op_type): Likewise.
15312 * config/mcore/mcore.c (mcore_output_move): Likewise.
15313 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
15315 (microblaze_function_arg): Likewise.
15316 * config/mips/mips.c (mips16_build_call_stub): Likewise.
15317 (mips_print_operand): Likewise.
15318 (mips_mode_ok_for_mov_fmt_p): Likewise.
15319 (mips_vector_mode_supported_p): Likewise.
15320 (mips_preferred_simd_mode): Likewise.
15321 (mips_expand_vpc_loongson_even_odd): Likewise.
15322 (mips_expand_vec_unpack): Likewise.
15323 (mips_expand_vi_broadcast): Likewise.
15324 (mips_expand_vector_init): Likewise.
15325 (mips_expand_vec_reduc): Likewise.
15326 (mips_expand_msa_cmp): Likewise.
15327 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
15328 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
15329 (cc_flags_for_mode): Likewise.
15330 * config/msp430/msp430.c (msp430_print_operand): Likewise.
15331 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
15332 (nds32_output_casesi_pc_relative): Likewise.
15333 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15334 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
15335 (nvptx_gen_unpack): Likewise.
15336 (nvptx_gen_pack): Likewise.
15337 (nvptx_gen_shuffle): Likewise.
15338 (nvptx_gen_wcast): Likewise.
15339 (nvptx_preferred_simd_mode): Likewise.
15340 * config/pa/pa.c (pa_secondary_reload): Likewise.
15341 * config/pa/predicates.md (base14_operand): Likewise.
15342 * config/powerpcspe/powerpcspe-c.c
15343 (altivec_resolve_overloaded_builtin): Likewise.
15344 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
15346 (rs6000_preferred_simd_mode): Likewise.
15347 (output_vec_const_move): Likewise.
15348 (rs6000_expand_vector_extract): Likewise.
15349 (rs6000_split_vec_extract_var): Likewise.
15350 (reg_offset_addressing_ok_p): Likewise.
15351 (rs6000_legitimate_offset_address_p): Likewise.
15352 (rs6000_legitimize_address): Likewise.
15353 (rs6000_emit_set_const): Likewise.
15354 (rs6000_const_vec): Likewise.
15355 (rs6000_emit_move): Likewise.
15356 (spe_build_register_parallel): Likewise.
15357 (rs6000_darwin64_record_arg_recurse): Likewise.
15358 (swap_selector_for_mode): Likewise.
15359 (spe_init_builtins): Likewise.
15360 (paired_init_builtins): Likewise.
15361 (altivec_init_builtins): Likewise.
15362 (do_load_for_compare): Likewise.
15363 (rs6000_generate_compare): Likewise.
15364 (rs6000_expand_float128_convert): Likewise.
15365 (emit_load_locked): Likewise.
15366 (emit_store_conditional): Likewise.
15367 (rs6000_output_function_epilogue): Likewise.
15368 (rs6000_handle_altivec_attribute): Likewise.
15369 (rs6000_function_value): Likewise.
15370 (emit_fusion_gpr_load): Likewise.
15371 (emit_fusion_p9_load): Likewise.
15372 (emit_fusion_p9_store): Likewise.
15373 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
15374 (fusion_gpr_mem_load): Likewise.
15375 (fusion_addis_mem_combo_load): Likewise.
15376 (fusion_addis_mem_combo_store): Likewise.
15377 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
15378 (fusion_gpr_mem_load): Likewise.
15379 (fusion_addis_mem_combo_load): Likewise.
15380 (fusion_addis_mem_combo_store): Likewise.
15381 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15383 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
15384 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15385 (rs6000_preferred_simd_mode): Likewise.
15386 (output_vec_const_move): Likewise.
15387 (rs6000_expand_vector_extract): Likewise.
15388 (rs6000_split_vec_extract_var): Likewise.
15389 (reg_offset_addressing_ok_p): Likewise.
15390 (rs6000_legitimate_offset_address_p): Likewise.
15391 (rs6000_legitimize_address): Likewise.
15392 (rs6000_emit_set_const): Likewise.
15393 (rs6000_const_vec): Likewise.
15394 (rs6000_emit_move): Likewise.
15395 (rs6000_darwin64_record_arg_recurse): Likewise.
15396 (swap_selector_for_mode): Likewise.
15397 (paired_init_builtins): Likewise.
15398 (altivec_init_builtins): Likewise.
15399 (rs6000_expand_float128_convert): Likewise.
15400 (emit_load_locked): Likewise.
15401 (emit_store_conditional): Likewise.
15402 (rs6000_output_function_epilogue): Likewise.
15403 (rs6000_handle_altivec_attribute): Likewise.
15404 (rs6000_function_value): Likewise.
15405 (emit_fusion_gpr_load): Likewise.
15406 (emit_fusion_p9_load): Likewise.
15407 (emit_fusion_p9_store): Likewise.
15408 * config/rx/rx.c (rx_gen_move_template): Likewise.
15409 (flags_from_mode): Likewise.
15410 * config/s390/predicates.md (s390_alc_comparison): Likewise.
15411 (s390_slb_comparison): Likewise.
15412 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
15413 (s390_vector_mode_supported_p): Likewise.
15414 (s390_cc_modes_compatible): Likewise.
15415 (s390_match_ccmode_set): Likewise.
15416 (s390_canonicalize_comparison): Likewise.
15417 (s390_emit_compare_and_swap): Likewise.
15418 (s390_branch_condition_mask): Likewise.
15419 (s390_rtx_costs): Likewise.
15420 (s390_secondary_reload): Likewise.
15421 (__SECONDARY_RELOAD_CASE): Likewise.
15422 (s390_expand_cs): Likewise.
15423 (s390_preferred_simd_mode): Likewise.
15424 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
15425 * config/sh/sh.c (sh_print_operand): Likewise.
15426 (dump_table): Likewise.
15427 (sh_secondary_reload): Likewise.
15428 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15429 * config/sh/sh.md (casesi_worker_1): Likewise.
15430 (casesi_worker_2): Likewise.
15431 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
15432 (fcc_comparison_operator): Likewise.
15433 * config/sparc/sparc.c (sparc_expand_move): Likewise.
15434 (emit_soft_tfmode_cvt): Likewise.
15435 (sparc_preferred_simd_mode): Likewise.
15436 (output_cbranch): Likewise.
15437 (sparc_print_operand): Likewise.
15438 (sparc_expand_vec_perm_bmask): Likewise.
15439 (vector_init_bshuffle): Likewise.
15440 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15441 (spu_vector_mode_supported_p): Likewise.
15442 (spu_expand_insv): Likewise.
15443 (spu_emit_branch_or_set): Likewise.
15444 (spu_handle_vector_attribute): Likewise.
15445 (spu_builtin_splats): Likewise.
15446 (spu_builtin_extract): Likewise.
15447 (spu_builtin_promote): Likewise.
15448 (spu_expand_sign_extend): Likewise.
15449 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15450 (tilegx_simd_int): Likewise.
15451 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
15452 (tilepro_simd_int): Likewise.
15453 * config/v850/v850.c (const_double_split): Likewise.
15454 (v850_print_operand): Likewise.
15455 (ep_memory_offset): Likewise.
15456 * config/vax/vax.c (vax_rtx_costs): Likewise.
15457 (vax_output_int_move): Likewise.
15458 (vax_output_int_add): Likewise.
15459 (vax_output_int_subtract): Likewise.
15460 * config/visium/predicates.md (visium_branch_operator): Likewise.
15461 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
15462 (visium_print_operand_address): Likewise.
15463 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15464 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
15465 (xtensa_expand_conditional_branch): Likewise.
15466 (xtensa_copy_incoming_a7): Likewise.
15467 (xtensa_output_literal): Likewise.
15468 * dfp.c (decimal_real_maxval): Likewise.
15469 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15472 Alan Hayward <alan.hayward@arm.com>
15473 David Sherwood <david.sherwood@arm.com>
15475 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
15476 (mode_nunits_inline): Likewise.
15477 (mode_inner_inline): Likewise.
15478 (mode_unit_size_inline): Likewise.
15479 (mode_unit_precision_inline): Likewise.
15480 (emit_insn_modes_h): Likewise. Also emit a #define of the
15482 (emit_mode_wider): Add an E_ prefix to mode names.
15483 (emit_mode_complex): Likewise.
15484 (emit_mode_inner): Likewise.
15485 (emit_mode_adjustments): Likewise.
15486 (emit_mode_int_n): Likewise.
15487 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
15488 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
15489 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
15490 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
15491 (CRC32_BUILTIN, ENTRY): Likewise.
15492 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
15493 (aarch64_pop_regs): Likewise.
15494 (aarch64_process_components): Likewise.
15495 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
15496 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
15497 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
15498 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
15499 * config/arm/arm.c (arm_init_libfuncs): Likewise.
15500 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
15502 * config/i386/i386-builtin.def (pcmpestr): Likewise.
15503 (pcmpistr): Likewise.
15504 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
15505 * config/mmix/mmix.c (mmix_output_condition): Likewise.
15506 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
15508 * config/rl78/rl78.c (mduc_regs): Likewise.
15509 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15510 (htm_expand_builtin): Likewise.
15511 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
15512 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
15513 * config/xtensa/xtensa.c (print_operand): Likewise.
15514 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
15515 (NUM_MODE_VECTOR_INT): Likewise.
15516 * genoutput.c (null_operand): Likewise.
15517 (output_operand_data): Likewise.
15518 * genrecog.c (print_parameter_value): Likewise.
15519 * lra.c (debug_operand_data): Likewise.
15521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15523 * dwarf2out.c (scompare_loc_descriptor_wide)
15524 (scompare_loc_descriptor_narrow): New functions, split out from...
15525 (scompare_loc_descriptor): ...here.
15526 * expmed.c (emit_store_flag_int): New function, split out from...
15527 (emit_store_flag): ...here.
15529 2017-08-30 Richard Biener <rguenther@suse.de>
15531 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
15532 (dwarf2out_early_finish): Move setting of AT_pubnames from
15533 early debug output to early finish.
15535 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15537 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
15538 and -mdata-region to the assembler.
15540 2017-08-30 Richard Biener <rguenther@suse.de>
15542 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
15544 (gen_subprogram_die): Add DW_AT_object_pointer only early.
15545 (dwarf2out_early_global_decl): Only generate a DIE for the
15546 abstract origin if it doesn't already exist or is a declaration DIE.
15547 (resolve_addr): Do not add the linkage name twice when
15548 generating a stub DIE for the DW_TAG_GNU_call_site target.
15550 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15552 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15553 Use machine_mode rather than int for arg1_mode.
15555 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
15558 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
15559 that the second argument of the built-in functions to unpack
15560 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
15561 switch statement instead a lot of if statements.
15562 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
15563 Allow 64-bit values to be in Altivec registers as well as
15564 traditional floating point registers.
15565 (pack<mode>, FMOVE128_VSX iterator): Likewise.
15567 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
15569 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
15570 MAX_REGS_PER_ADDRESS == 1.
15572 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
15574 * config/i386/i386.opt (flag_fentry): Do not init to -1.
15575 * config/i386/i386.c (ix86_option_override_internal): Simplify
15576 setting of opts->x_flag_entry.
15578 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15579 Jakub Jelinek <jakub@redhat.com>
15580 Richard Biener <rguenther@suse.de>
15582 PR tree-optimization/81503
15583 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
15584 folded constant fits in the target type; reorder tests for clarity.
15586 2017-08-29 Martin Liska <mliska@suse.cz>
15588 * passes.def: Include pass_lower_switch.
15589 * stmt.c (dump_case_nodes): Remove and move to
15590 tree-switch-conversion.
15591 (case_values_threshold): Likewise.
15592 (expand_switch_as_decision_tree_p): Likewise.
15593 (emit_case_decision_tree): Likewise.
15594 (expand_case): Likewise.
15595 (balance_case_nodes): Likewise.
15596 (node_has_low_bound): Likewise.
15597 (node_has_high_bound): Likewise.
15598 (node_is_bounded): Likewise.
15599 (emit_case_nodes): Likewise.
15600 (struct simple_case_node): New struct.
15601 (add_case_node): Remove.
15602 (emit_case_dispatch_table): Use vector instead of case_list.
15603 (reset_out_edges_aux): Remove.
15604 (compute_cases_per_edge): Likewise.
15605 (expand_case): Build list of simple_case_node.
15606 (expand_sjlj_dispatch_table): Use it.
15607 * tree-switch-conversion.c (struct case_node): Moved from
15608 stmt.c and adjusted.
15609 (emit_case_nodes): Likewise.
15610 (node_has_low_bound): Likewise.
15611 (node_has_high_bound): Likewise.
15612 (node_is_bounded): Likewise.
15613 (case_values_threshold): Likewise.
15614 (reset_out_edges_aux): Likewise.
15615 (compute_cases_per_edge): Likewise.
15616 (add_case_node): Likewise.
15617 (dump_case_nodes): Likewise.
15618 (balance_case_nodes): Likewise.
15619 (expand_switch_as_decision_tree_p): Likewise.
15620 (emit_jump): Likewise.
15621 (emit_case_decision_tree): Likewise.
15622 (try_switch_expansion): Likewise.
15623 (do_jump_if_equal): Likewise.
15624 (emit_cmp_and_jump_insns): Likewise.
15625 (fix_phi_operands_for_edge): New function.
15626 (record_phi_operand_mapping): Likewise.
15627 (class pass_lower_switch): New pass.
15628 (pass_lower_switch::execute): New function.
15629 (make_pass_lower_switch): Likewise.
15630 (conditional_probability):
15631 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
15632 * tree-pass.h: Add make_pass_lower_switch.
15634 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15637 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
15640 2017-08-29 Richard Biener <rguenther@suse.de>
15642 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
15643 we do not add a DW_AT_inline attribute twice.
15644 (gen_subprogram_die): Remove code setting DW_AT_inline on
15645 DECL_ABSTRACT_P nodes.
15647 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
15649 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
15650 calls to internal functions.
15651 (gimplify_modify_expr): Likewise.
15652 * tree-call-cdce.c (use_internal_fn): Likewise.
15653 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
15654 (convert_to_divmod): Set the nothrow flag.
15655 * tree-if-conv.c (predicate_mem_writes): Likewise.
15656 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15657 (vectorizable_call): Likewise.
15658 (vectorizable_store): Likewise.
15659 (vectorizable_load): Likewise.
15660 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
15661 (vect_recog_mask_conversion_pattern): Likewise.
15663 2017-08-29 Martin Liska <mliska@suse.cz>
15666 * gcc.c (driver_handle_option): Add new argument.
15667 * opts-common.c (handle_option): Pass
15668 target_option_override_hook.
15669 * opts-global.c (lang_handle_option): Add new option.
15670 (set_default_handlers): Add new argument.
15671 (decode_options): Likewise.
15672 * opts.c (target_handle_option): Likewise.
15673 (common_handle_option): Call target_option_override_hook.
15674 * opts.h (struct cl_option_handler_func): Add hook for
15675 target option override.
15676 (struct cl_option_handlers): Likewise.
15677 (set_default_handlers): Add new argument.
15678 (decode_options): Likewise.
15679 (common_handle_option): Likewise.
15680 (target_handle_option): Likewise.
15681 * toplev.c (toplev::main): Pass targetm.target_option.override
15684 2017-08-29 Richard Biener <rguenther@suse.de>
15685 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
15687 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
15688 life to the active subtree.
15690 2017-08-28 Jeff Law <law@redhat.com>
15692 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
15693 derive_equivalences.
15694 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
15695 Code moved into....
15696 (edge_info::derive_equivalences): New private member function
15698 * tree-ssa-dom.c (class edge_info): Changed from a struct
15699 to a class. Add ctor/dtor, methods and data members.
15700 (edge_info::edge_info): Renamed from allocate_edge_info.
15701 Initialize additional members.
15702 (edge_info::~edge_info): New.
15703 (free_dom_edge_info): Delete the edge info.
15704 (record_edge_info): Use new class & associated member functions.
15705 Tighten forms for testing for edge equivalences.
15706 (record_temporary_equivalences): Iterate over the simple
15707 equivalences rather than assuming there's only one per edge.
15708 (cprop_into_successor_phis): Iterate over the simple
15709 equivalences rather than assuming there's only one per edge.
15710 (optimize_stmt): Use operand_equal_p rather than pointer
15711 equality for mini-DSE code.
15713 2017-08-28 Nathan Sidwell <nathan@acm.org>
15715 * gcc.c (execute): Fold SIGPIPE handling into switch
15716 statement. Adjust internal error message.
15718 2017-08-28 Richard Biener <rguenther@suse.de>
15721 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
15722 Do nothing for removed DIEs.
15724 2017-08-28 Richard Biener <rguenther@suse.de>
15726 PR tree-optimization/81977
15727 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
15730 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
15733 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
15734 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
15735 using targetm.gen_mem_thread_fence.
15737 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
15740 * config/i386/i386.md (*<btsc><mode>): Change operand 2
15741 predicate to register_operand. Reorder operands.
15742 (*btr<mode>): Ditto.
15743 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
15744 (*btr<mode>_mask): Ditto.
15746 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
15748 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
15749 * config/rs6000/xmmintrin.h: New file.
15750 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
15752 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15755 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
15756 parameter and_insn and return it.
15757 (recombine_lvx_pattern): Insert a copy to ensure availability of
15758 the base register of the copied masking operation at the point of
15759 the instruction replacement.
15760 (recombine_stvx_pattern): Likewise.
15762 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
15764 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
15765 undocumented switches.
15766 (-mpower9-dform-vector): Likewise.
15767 (-mpower9-dform): Likewise.
15768 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
15769 comments to delete references to -mpower9-dform* switches.
15770 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
15771 Delete reference to -mpower9-dform* switches, test for
15772 -mpower9-vector instead.
15773 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
15774 (OTHER_P9_VECTOR_MASKS): Likewise.
15775 (POWERPC_MASKS): Likewise.
15776 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
15777 tests against -mpower9-dform* to -mpower9-vector. Delete code
15778 that checked for -mpower9-dform* consistancy with other options.
15779 Add test for -mpower9-misc to enable other power9 switches.
15780 (rs6000_init_hard_regno_mode_ok): Likewise.
15781 (rs6000_option_override_internal): Likewise.
15782 (rs6000_emit_prologue): Likewise.
15783 (rs6000_emit_epilogue): Likewise.
15784 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
15785 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
15786 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
15788 (emit_fusion_p9_store): Likewise.
15789 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
15790 resetting these macros if the assembler does not support ISA 3.0
15792 (TARGET_P9_DFORM_VECTOR): Likewise.
15793 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
15794 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
15796 2017-08-25 Alan Modra <amodra@gmail.com>
15798 PR rtl-optimization/81747
15799 * cse.c (cse_extended_basic_block): Don't attempt to record
15800 equivalences for degenerate conditional branches that branch
15801 to their fall-through.
15803 2017-08-24 Martin Sebor <msebor@redhat.com>
15805 PR middle-end/81908
15806 * gimple-fold.c (size_must_be_zero_p): New function.
15807 (gimple_fold_builtin_memory_op): Call it.
15809 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
15811 * config/rs6000/mm_malloc.h: New file.
15813 2017-08-24 Bin Cheng <bin.cheng@arm.com>
15815 PR tree-optimization/81913
15816 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
15817 analysis when either IVs in condition can wrap.
15819 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
15821 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
15822 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
15824 2017-08-24 Richard Biener <rguenther@suse.de>
15827 * targhooks.c (default_target_can_inline_p): Properly
15828 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
15829 is present and always compare.
15830 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
15831 infer -mfpmath=sse from TARGET_SSE_P.
15832 (ix86_can_inline_p): Properly use target_option_default_node when
15833 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
15835 2017-08-24 Richard Biener <rguenther@suse.de>
15838 * dwarf2out.c (output_die): Handle flag_generate_offload like
15840 (output_comp_unit): Likewise.
15841 (gen_array_type_die): Likewise.
15842 (dwarf2out_early_finish): Likewise.
15843 (note_variable_value_in_expr): Likewise.
15844 (dwarf2out_finish): Likewise. Adjust assert.
15845 * cgraphunit.c (symbol_table::compile): Move setting of
15846 flag_generate_offload earlier ...
15847 (symbol_table::finalize_compilation_unit): ... here, before
15848 early debug finalization.
15850 2017-08-24 Richard Biener <rguenther@suse.de>
15852 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
15853 and ipa-fnsummary.h.
15854 (ix86_can_inline_p): When ix86_fpmath flags do not match
15855 check whether the callee uses FP math at all.
15857 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
15859 PR middle-end/81931
15860 * tree-ssanames.c (get_nonzero_bits): Use element_precision
15861 instead of TYPE_PRECISION.
15863 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
15864 Alan Hayward <alan.hayward@arm.com>
15865 David Sherwood <david.sherwood@arm.com>
15867 * combine.c (make_extraction): Use subreg_offset_from_lsb.
15869 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
15871 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
15873 (ix86_frame::stack_realign_allocate): New field.
15874 (struct machine_frame_state): Modify comments.
15875 (machine_frame_state::sp_realigned_fp_end): New field.
15876 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
15877 layout calculation.
15878 (sp_valid_at): Add assertion to assure no attempt to access invalid
15879 offset of a realigned stack.
15880 (fp_valid_at): Likewise.
15881 (choose_baseaddr): Modify comments.
15882 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
15883 ix86_expand_prologue.
15884 (ix86_expand_prologue): Modify stack realignment and allocation.
15885 (ix86_expand_epilogue): Modify comments.
15886 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
15887 avx2_runtime, avx512f, and avx512f_runtime.
15889 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
15891 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
15892 (mstackrealign): Do not init to -1.
15893 * config/i386/i386.c (ix86_option_override_internal):
15894 Check opts_set, not opts when setting default value of
15895 opts->x_ix86_force_align_arg_pointer.
15897 2017-08-23 Richard Biener <rguenther@suse.de>
15899 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
15900 lang_hooks.decl_printable_name.
15901 * print-rtl-function.c (print_rtx_function): Likewise.
15902 * tree-pretty-print.c (dump_function_header): Likewise.
15904 2017-08-23 Richard Biener <rguenther@suse.de>
15907 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
15908 -g0 at compile-time.
15910 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15912 PR middle-end/19706
15913 * doc/sourcebuild.texi (Other hardware attributes):
15916 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15918 PR middle-end/19706
15919 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
15920 Fix single-use check.
15922 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
15924 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
15926 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
15928 * doc/install.texi: Modify to add more details on running selected
15931 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
15933 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
15934 is combined with -mabi=ms.
15935 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
15938 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15940 PR tree-optimization/81488
15941 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
15942 and cached_basis fields.
15943 (MAX_SPREAD): New constant.
15944 (alloc_cand_and_find_basis): Initialize new fields.
15945 (clear_visited): New function.
15946 (create_phi_basis_1): Rename from create_phi_basis, set visited
15947 and cached_basis fields.
15948 (create_phi_basis): New wrapper function.
15949 (phi_add_costs_1): Rename from phi_add_costs, add spread
15950 parameter, set visited field, short-circuit when limits reached.
15951 (phi_add_costs): New wrapper function.
15952 (record_phi_increments_1): Rename from record_phi_increments, set
15954 (record_phi_increments): New wrapper function.
15955 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
15956 (phi_incr_cost): New wrapper function.
15957 (all_phi_incrs_profitable_1): Rename from
15958 all_phi_incrs_profitable, set visited field.
15959 (all_phi_incrs_profitable): New wrapper function.
15961 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
15962 Alan Hayward <alan.hayward@arm.com>
15963 David Sherwood <david.sherwood@arm.com>
15965 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
15966 that takes the outer and inner modes.
15967 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
15968 comparison as the canonical test for a paradoxical subreg.
15969 * combine.c (simplify_set): Use paradoxical_subreg_p.
15970 (make_extraction): Likewise.
15971 (force_to_mode): Likewise.
15972 (rtx_equal_for_field_assignment_p): Likewise.
15973 (gen_lowpart_for_combine): Likewise.
15974 (simplify_comparison): Likewise.
15975 * cse.c (equiv_constant): Likewise.
15976 * expmed.c (store_bit_field_1): Likewise.
15977 * final.c (alter_subreg): Likewise.
15978 * fwprop.c (propagate_rtx): Likewise.
15979 (forward_propagate_subreg): Likewise.
15980 * ira-conflicts.c (ira_build_conflicts): Likewise.
15981 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
15982 * lra-constraints.c (curr_insn_transform): Likewise.
15983 (split_reg): Likewise.
15984 * lra-eliminations.c (move_plus_up): Likewise.
15985 (lra_eliminate_regs_1): Likewise.
15986 * recog.c (general_operand): Likewise.
15987 * ree.c (combine_reaching_defs): Likewise.
15988 * reload.c (push_reload): Likewise.
15989 (find_reloads): Likewise.
15990 * reload1.c (elimination_effects): Likewise.
15991 (compute_reload_subreg_offset): Likewise.
15992 (choose_reload_regs): Likewise.
15993 * rtlanal.c (subreg_lsb_1): Likewise.
15994 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15995 (simplify_subreg): Likewise.
15996 * var-tracking.c (track_loc_p): Likewise.
15997 * emit-rtl.c (byte_lowpart_offset): Likewise.
15998 (paradoxical_subreg_p): Delete out-of-line definition.
16000 2017-08-22 Jeff Law <law@redhat.com>
16002 PR tree-optimization/81741
16003 PR tree-optimization/71947
16004 * tree-ssa-dom.c: Include tree-inline.h.
16005 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
16006 equivalences if one is more expensive to compute than the other.
16007 * tree-ssa-scopedtables.h (class const_or_copies): Make
16008 record_const_or_copy_raw method private.
16009 (class avail_exprs_stack): New method simplify_binary_operation.
16010 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
16011 avail_exprs_stack::simplify_binary_operation as needed.
16012 (avail_exprs_stack::simplify_binary_operation): New function.
16014 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
16016 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
16017 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
16018 (DOT_SYMBOLS): Likewise.
16019 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
16020 (RELOCATABLE_NEEDS_FIXUP): Likewise.
16021 (RS6000_ABI_NAME): Likewise.
16022 (TARGET_CMODEL): Likewise.
16023 (TOC_SECTION_ASM_OP): Likewise.
16024 (SET_CMODEL): New macro.
16025 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
16027 2017-08-22 Richard Biener <rguenther@suse.de>
16029 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
16030 to allow for free-lang-data replacements similar to verify_type_variant.
16032 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16033 Alan Hayward <alan.hayward@arm.com>
16034 David Sherwood <david.sherwood@arm.com>
16036 * config/aarch64/aarch64.md (casesi): Use DImode rather than
16037 VOIDmode for the LABEL_REF.
16039 2017-08-22 Richard Biener <rguenther@suse.de>
16041 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
16043 2017-08-22 Richard Biener <rguenther@suse.de>
16045 * common.opt (feliminate-dwarf2-dups): Ignore.
16046 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
16047 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
16048 same_die_p_wrap, compute_section_prefix,
16049 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
16050 (comdat_symbol_id, comdat_symbol_number): Likewise.
16051 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
16053 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
16054 (output_die): Mark unreachable path unreachable.
16055 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
16056 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
16057 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
16058 (dwarf2out_early_finish): Likewise.
16060 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
16062 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
16064 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
16067 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
16068 not VAR_P. Filter attribute warnings with OPT_Wattributes.
16069 (avr_attribute_table) <io, io_low, address>: Initialize
16070 .decl_required with true.
16072 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
16074 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
16075 undocumented debugging options.
16076 (-mvsx-scalar-double): Likewise.
16077 (-mallow-df-permute): Likewise.
16078 (-mvectorize-builtins): Likewise.
16079 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16080 (rs6000_builtin_vectorized_function): Likewise.
16081 (rs6000_builtin_md_vectorized_function): Likewise.
16082 (rs6000_opt_vars): Likewise.
16084 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
16087 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
16088 (*btrq_imm): Rename from *btrq.
16089 (*btcq_imm): Rename from *btcq.
16090 (btsc): New code attribute.
16091 (*<btsc><mode>): New insn pattern.
16092 (*btr<mode>): Ditto.
16093 (*<btsc><mode>_mask): New insn_and_split pattern.
16094 (*btr<mode>_mask): Ditto.
16096 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16097 Alan Hayward <alan.hayward@arm.com>
16098 David Sherwood <david.sherwood@arm.com>
16100 * function.c (pad_below): Simplify padding calculation.
16102 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16103 Alan Hayward <alan.hayward@arm.com>
16104 David Sherwood <david.sherwood@arm.com>
16106 * target.def (function_prologue): Remove frame size argument.
16107 (function_epilogue): Likewise.
16108 * doc/tm.texi: Regenerate.
16109 * final.c (final_start_function): Update call to function_prologue.
16110 (final_end_function): Update call to function_epilogue.
16111 (default_function_pro_epilogue): Remove frame size argument.
16112 * output.h (default_function_pro_epilogue): Likewise.
16113 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
16114 (arm_output_function_prologue): Likewise.
16115 * config/frv/frv.c (frv_function_prologue): Likewise.
16116 (frv_function_epilogue): Likewise.
16117 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
16118 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
16119 (ia64_output_function_epilogue): Likewise.
16120 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
16121 (m32r_output_function_epilogue): Likewise.
16122 * config/microblaze/microblaze.c (microblaze_function_prologue)
16123 (microblaze_function_epilogue): Likewise.
16124 * config/mips/mips.c (mips_output_function_prologue): Likewise.
16125 (mips_output_function_epilogue): Likewise.
16126 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
16127 (mmix_target_asm_function_epilogue): Likewise.
16128 * config/msp430/msp430.c (msp430_start_function): Likewise.
16129 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
16130 (nds32_asm_function_epilogue): Likewise.
16131 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
16132 * config/pa/pa.c (pa_output_function_prologue): Likewise.
16133 (pa_output_function_epilogue): Likewise.
16134 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
16135 (rs6000_output_function_epilogue): Likewise.
16136 * config/rl78/rl78.c (rl78_start_function): Likewise.
16137 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
16138 (rs6000_output_function_epilogue): Likewise.
16139 * config/rx/rx.c (rx_output_function_prologue): Likewise.
16140 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
16141 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
16142 (sparc_asm_function_epilogue): Likewise.
16144 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16146 * tree.h (type_has_mode_precision_p): New function.
16147 * convert.c (convert_to_integer_1): Use it.
16148 * expr.c (expand_expr_real_2): Likewise.
16149 (expand_expr_real_1): Likewise.
16150 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
16151 * match.pd: Likewise.
16152 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
16153 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
16154 * tree-tailcall.c (process_assignment): Likewise.
16155 * tree-vect-loop.c (vectorizable_reduction): Likewise.
16156 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
16157 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
16158 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16159 (vectorizable_assignment): Likewise.
16160 (vectorizable_shift): Likewise.
16161 (vectorizable_operation): Likewise.
16162 * tree-vrp.c (register_edge_assert_for_2): Likewise.
16164 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
16166 * match.pd: Add pow (C, x) simplification.
16168 2017-08-21 Richard Biener <rguenther@suse.de>
16170 PR tree-optimization/81900
16171 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
16172 for blocks with abnormal predecessors.
16173 (compute_antic): Do not set visited flag prematurely.
16175 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
16178 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
16180 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16182 * stor-layout.h (vector_type_mode): Move to...
16183 * tree.h (vector_type_mode): ...here.
16184 * stor-layout.c (vector_type_mode): Move to...
16185 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
16187 2017-08-21 Richard Biener <rguenther@suse.de>
16189 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
16190 register_external_die hooks.
16191 (debug_false_tree_charstarstar_uhwistar): Declare.
16192 (debug_nothing_tree_charstar_uhwi): Likewise.
16193 * debug.c (do_nothing_debug_hooks): Adjust.
16194 (debug_false_tree_charstarstar_uhwistar): New do nothing.
16195 (debug_nothing_tree_charstar_uhwi): Likewise.
16196 * dbxout.c (dbx_debug_hooks): Adjust.
16197 (xcoff_debug_hooks): Likewise.
16198 * sdbout.c (sdb_debug_hooks): Likewise.
16199 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16200 * dwarf2out.c (macinfo_label_base): New global.
16201 (dwarf2out_register_external_die): New function for the
16202 register_external_die hook.
16203 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
16204 (dwarf2_debug_hooks): Use them.
16205 (dwarf2_lineno_debug_hooks): Adjust.
16206 (struct die_struct): Add with_offset flag.
16207 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
16208 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
16209 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
16210 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
16211 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
16212 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
16213 defining section names for the early LTO debug variants.
16214 (reset_indirect_string): New helper.
16215 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
16216 (print_dw_val): Add support for offsetted symbol references.
16217 (get_ultimate_context): Split out from is_cxx.
16218 (is_cxx): Use get_ultimate_context.
16219 (is_fortran): Add decl overload.
16220 (compute_comp_unit_symbol): Split out worker from
16221 compute_section_prefix.
16222 (compute_section_prefix): Call compute_comp_unit_symbol and
16223 set comdat_type_p here.
16224 (output_die): Skip DIE symbol output for the LTO added one.
16225 Handle DIE symbol references with offset.
16226 (output_comp_unit): Guard section name mangling properly.
16227 For LTO debug sections emit a symbol at the section beginning
16228 which we use to refer to its DIEs.
16229 (add_abstract_origin_attribute): For DIEs registered via
16230 dwarf2out_register_external_die directly refer to the early
16231 DIE rather than indirectly through the shadow one we created.
16232 Remove obsolete call to dwarf2out_abstract_function for
16233 non-function/block origins.
16234 (gen_array_type_die): When generating early LTO debug do
16235 not emit DW_AT_string_length.
16236 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
16237 late when in LTO. As suggested place a gcc_unreachable for
16238 the DECL_ABSTRACT_P case.
16239 (gen_subprogram_die): Avoid another specification DIE
16240 for early built declarations/definitions for the late LTO case.
16241 (gen_variable_die): Add type references for late duplicated VLA dies
16243 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
16244 we have the abstract instance already.
16245 (process_scope_var): Adjust decl DIE contexts in LTO which
16246 first puts them in limbo.
16247 (gen_decl_die): Do not generate type DIEs late apart from
16248 types for VLAs or for decls we do not yet have a DIE. Do not
16249 call dwarf2out_abstract_function late.
16250 (dwarf2out_early_global_decl): Make sure to create DIEs
16251 for abstract instances of a decl first.
16252 (dwarf2out_late_global_decl): Adjust comment.
16253 (output_macinfo_op): With multiple macro sections use
16254 macinfo_label_base to distinguish labels.
16255 (output_macinfo): Likewise. Update macinfo_label_base.
16256 Pass in the line info label.
16257 (note_variable_value_in_expr): When generating LTO resolve
16258 all variable values here by generating DIEs as needed.
16259 (init_sections_and_labels): Add early LTO debug flag parameter
16260 and generate different sections and names if set. Add generation
16261 counter for the labels so we can have multiple of them.
16262 (reset_dies): Helper to allow DIEs to be output multiple times.
16263 (dwarf2out_finish): When outputting DIEs to the fat part of an
16264 LTO object first reset DIEs.
16265 (dwarf2out_early_finish): Output early DIEs when generating LTO.
16266 (modified_type_die): Check for decl_ultimate_origin being self
16268 (gen_type_die_with_usage): Likewise.
16269 (gen_typedef_die): Allow decl_ultimate_origin being self.
16270 (set_decl_abstract_flags): Remove.
16271 (set_block_abstract_flags): Likewise.
16272 (dwarf2out_abstract_function): Treat the early generated DIEs
16273 as the abstract copy and only add DW_AT_inline and
16274 DW_AT_artificial here and call set_decl_origin_self.
16275 If the DIE has an abstract origin don't do anything.
16276 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
16277 if we have none yet (Go fails to build one, PR78628).
16278 (variably_modified_type_p): Prevent endless recursion for Ada
16279 cyclic pointer types.
16280 * lto-streamer-in.c: Include debug.h.
16281 (dref_queue): New global.
16282 (lto_read_tree_1): Stream in DIE references.
16283 (lto_input_tree): Register DIE references.
16284 (input_function): Stream DECL_DEBUG_ARGS.
16285 * lto-streamer-out.c: Include debug.h.
16286 (lto_write_tree_1): Output DIE references.
16287 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
16288 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
16289 (output_function): Stream DECL_DEBUG_ARGS.
16290 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
16291 Stream DECL_ABSTRACT_ORIGIN.
16292 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
16293 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
16294 DECL_CONTEXT for file-scope decls.
16295 * lto-streamer.h (struct dref_entry): Declare.
16296 (dref_queue): Likewise.
16297 * cfgexpand.c (pass_expand::execute): Do not call the
16298 outlining_inline_function hook here.
16299 * lto-wrapper.c (debug_obj): New global.
16300 (tool_cleanup): Unlink it if required.
16301 (debug_objcopy): New function.
16302 (run_gcc): Handle early debug sections in the IL files by
16303 extracting them to separate files, partially linkin them and
16304 feeding the result back as result to the linker.
16305 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
16306 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
16307 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
16308 sections into a separate segment.
16309 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
16311 (darwin_asm_dwarf_section): Likewise.
16312 (darwin_asm_output_dwarf_offset): Likewise.
16313 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
16315 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16316 Alan Hayward <alan.hayward@arm.com>
16317 David Sherwood <david.sherwood@arm.com>
16319 * read-md.h (md_reader::record_potential_iterator_use): Replace
16320 pointer argument with an rtx and an index.
16321 * read-rtl.c (iterator_group::apply_iterator): Likewise.
16322 (apply_mode_iterator): Likewise.
16323 (apply_code_iterator): Likewise.
16324 (apply_int_iterator): Likewise.
16325 (apply_subst_iterator): Likewise.
16326 (record_iterator_use): Likewise.
16327 (record_attribute_use): Likewise.
16328 (md_reader::record_potential_iterator_use): Likewise. Update calls
16329 to record_iterator_use and apply_iterator.
16330 (iterator_use): Replace ptr with x and index.
16331 (attribute_use): Likewise.
16332 (apply_attribute_uses): Update calls to apply_iterator.
16333 (apply_iterators): Likewise. Update initialization of iterator_use.
16334 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
16335 and record_potential_iterator_use.
16336 (rtx_reader::read_rtx_operand): Likewise.
16338 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16339 Alan Hayward <alan.hayward@arm.com>
16340 David Sherwood <david.sherwood@arm.com>
16342 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
16345 2017-08-21 Richard Biener <rguenther@suse.de>
16347 PR middle-end/81884
16348 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
16349 at struct end conservatively when comparing common bases.
16351 2017-08-21 Richard Biener <rguenther@suse.de>
16353 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
16354 (mem_ref_in_stmt): Remove.
16355 (determine_max_movement): Use ref index to get at the reference.
16356 (invariantness_dom_walker::before_dom_children): Deal with
16357 lim data already initialized.
16358 (gather_mem_refs_stmt): Initialize lim data and record ref index.
16360 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16362 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
16363 (TARGET_ISA_ROUND): Ditto.
16364 (TARGET_ROUND): Ditto.
16365 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
16366 * config/i386/i386.md: Ditto.
16367 * config/i386/sse.md: Ditto.
16368 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
16369 with OPTION_MASK_ISA_SSE4_1.
16371 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16374 * doc/extend.texi (x86 Built-in Functions): Correct the name of
16375 __builtin_ia32_lzcnt_u16.
16377 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
16380 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
16381 (rs6000_set_current_function): Rewrite function to use it.
16383 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
16386 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
16387 and TYPE_WARN_IF_NOT_ALIGN.
16388 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
16389 (handle_warn_if_not_align): New.
16390 (place_union_field): Call handle_warn_if_not_align.
16391 (place_field): Call handle_warn_if_not_align.
16392 Copy TYPE_WARN_IF_NOT_ALIGN.
16393 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
16394 (layout_type): Likewise.
16395 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
16397 (tree_decl_common): Add warn_if_not_align.
16398 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
16399 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
16400 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
16401 (DECL_WARN_IF_NOT_ALIGN): Likewise.
16402 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
16403 * doc/extend.texi: Document warn_if_not_aligned attribute.
16404 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
16406 2017-08-17 Martin Liska <mliska@suse.cz>
16409 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
16410 (get_data_dependence): Use it as pointer type.
16411 (distribute_loop): Likewise.
16413 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16415 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
16416 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
16417 (p8_vmrgow_<mode>_direct): New define_insn.
16418 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
16419 handle endianness for vmrgew and vmrgow permute patterns.
16421 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16423 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
16424 * config/rs6000/rs6000-cpus.def: Remove comment.
16425 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
16426 (POWERPC_MASKS): Likewise.
16427 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
16428 use of TARGET_VSX_TIMODE.
16429 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
16430 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
16431 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
16432 (rs6000_option_override_internal): Remove dead code.
16433 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
16434 (rs6000_legitimize_reload_address): Likewise.
16435 (rs6000_legitimate_address_p): Likewise.
16436 (rs6000_opt_masks): Delete "vsx-timode".
16437 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
16438 from function comment.
16439 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
16440 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
16441 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
16443 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
16444 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
16445 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
16446 (VSX_TI): Likewise.
16448 (define_peephole2): Likewise.
16450 2017-08-17 Martin Sebor <msebor@redhat.com>
16453 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
16454 past the end of an array.
16455 (test_pp_format): Add test cases.
16457 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
16459 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
16460 missing ECF_NOTHROW flags.
16462 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16465 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
16466 operands residing in integer registers.
16467 (*vsx_le_perm_load_<mode>): Likewise.
16468 (*vsx_le_perm_store_<mode>): Likewise.
16469 (define_peephole2): Add peepholes to optimize the above.
16471 2017-08-17 Marek Polacek <polacek@redhat.com>
16473 PR middle-end/81814
16474 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
16475 to mimic what shorten_compare did. Change the return type to bool.
16476 (fold_cond_expr_with_comparison): Update call to
16477 operand_equal_for_comparison_p.
16478 (fold_ternary_loc): Likewise.
16480 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
16482 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
16484 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
16486 2017-08-17 Richard Biener <rguenther@suse.de>
16488 * tree-ssa-structalias.c (solve_graph): When propagating
16489 to successors update the graphs succ edges and avoid duplicate work.
16491 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
16494 * config/i386/i386.c (ix86_option_override_internal): Save target
16495 specific options after ix86_stack_protector_guard_reg was changed.
16497 2017-08-17 Richard Biener <rguenther@suse.de>
16499 PR tree-optimization/81827
16500 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
16501 (new_var_info): Initialize it conservatively.
16502 (get_call_vi): Mark register vars.
16503 (new_scalar_tmp_constraint_exp): Likewise.
16504 (handle_rhs_call): Likewise.
16505 (handle_const_call): Likewise.
16506 (create_function_info_for): Likewise.
16507 (solve_constraints): Sort varinfos to separate register from
16508 non-register vars to pack points-to solution bitmaps during
16511 2017-08-17 Marek Polacek <polacek@redhat.com>
16513 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
16515 2017-08-17 Richard Biener <rguenther@suse.de>
16517 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
16518 to true when overflow is undefined and we saturated the result.
16520 2017-08-17 Alan Modra <amodra@gmail.com>
16523 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
16524 Don't use store multiple if only one reg needs saving.
16525 (interesting_frame_related_regno): New function.
16526 (rs6000_frame_related): Don't emit frame info for regs that
16528 (rs6000_emit_epilogue): Likewise.
16530 2017-08-16 Nathan Sidwell <nathan@acm.org>
16532 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
16533 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
16534 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
16535 * tree.c (free_lang_data_in_type): Use else-if chain. Always
16536 clear TYPE_LANG_1. Remove obsolete member-function stripping.
16537 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
16538 (verify_type): Adjust for TYPE_BINFO move.
16539 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
16540 process TYPE_BINFO directly.
16541 (hash_tree): Likewise.
16542 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16544 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16547 2017-08-16 David Malcolm <dmalcolm@redhat.com>
16549 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
16551 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
16554 * config/i386/i386.md (*anddi_1_btr): Change predicates of
16555 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
16556 Add ix86_binary_operator_ok to insn constraint.
16557 (*iordi_1_bts): Ditto.
16558 (*xordi_1_btc): Ditto.
16559 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
16560 Update corresponding peephole2 pattern.
16564 2017-08-16 Bin Cheng <bin.cheng@arm.com>
16566 PR tree-optimization/81832
16567 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
16568 copy loop header which has IFN_LOOP_DIST_ALIAS call.
16570 2017-08-16 Marek Polacek <polacek@redhat.com>
16573 * fold-const.c (fold_indirect_ref_1): Restore original behavior
16574 regarding size_zero_node.
16576 2017-08-16 Martin Liska <mliska@suse.cz>
16579 * config.gcc: Respect previously set extra_objs in case
16582 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
16584 PR tree-optimization/81835
16585 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
16586 the phi SSA_NAME. Check that the condition in a COND_EXPR does
16587 not depend on the phi.
16589 2017-08-16 Alan Modra <amodra@gmail.com>
16591 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
16594 2017-08-16 Alan Modra <amodra@gmail.com>
16596 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
16597 (save_reg_p): ..into this. Update all callers.
16598 (first_reg_to_save): Simplify.
16600 2017-08-16 Alan Modra <amodra@gmail.com>
16602 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
16605 2017-08-15 Joseph Myers <joseph@codesourcery.com>
16609 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
16610 constant count if that count is less than 32.
16612 2017-08-15 Nathan Sidwell <nathan@acm.org>
16614 * gcc.c (execute): Emit friendlier message if inferior is killed
16615 by an external cause.
16617 2017-08-15 Richard Biener <rguenther@suse.de>
16619 PR tree-optimization/81790
16620 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
16621 CONSTRUCTORs from simplifying and VN.
16623 2017-08-14 Martin Sebor <msebor@redhat.com>
16625 * builtin-attrs.def: Add comments.
16627 2017-08-14 Martin Sebor <msebor@redhat.com>
16630 * doc/extend.texi (attribute nonstring): Document new attribute.
16632 2017-08-14 Martin Sebor <msebor@redhat.com>
16635 * tree-diagnostic.c (default_tree_printer): Handle %G.
16636 * gimple-pretty-print.h (percent_G_format): Declare new function.
16637 * gimple-pretty-print.c (percent_G_format): Define.
16638 * tree-pretty-print.c (percent_K_format): Add argument.
16640 2017-08-14 Martin Sebor <msebor@redhat.com>
16642 PR translation/79998
16643 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
16644 Remove a stray space.
16646 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
16649 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
16650 (*iordi_1_bts): Ditto.
16651 (*xordi_1_btc): Ditto.
16653 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16656 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
16657 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16659 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
16660 quoted strings, and make more translator-friendly.
16661 (darwin_rs6000_override_options): Likewise.
16662 (rs6000_option_override_internal): Likewise.
16663 (rs6000_return_in_memory): Fix overlong line.
16664 (init_cmulative_args): Use quoted strings, and make more
16665 translator-friendly.
16666 (rs6000_pass_by_reference): Fix overlong line.
16667 (def_builtin): Use quoted strings.
16668 (altivec_expand_predicate_builtin): Use quoted strings, and make
16669 more translator-friendly.
16670 (htm_expand_builtin): Use quoted strings.
16671 (cpu_expand_builtin): Use quoted strings, and make more
16672 translator-friendly.
16673 (altivec_expand_builtin): Likewise.
16674 (paired_expand_predicate_builtin): Likewise.
16675 (rs6000_invalid_builtin): Likewise.
16676 (builtin_function_type): Use quoted strings.
16677 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
16678 more translator-friendly.
16679 (rs6000_trampoline_init): Likewise.
16680 (rs6000_handle_altivec_attribute): Likewise.
16681 (rs6000_inner_target_options): Use quoted strings.
16682 (rs6000_disable_incompatible_switches): Likewise.
16683 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
16684 strings, and make more translator-friendly.
16685 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
16687 2017-08-14 Bin Cheng <bin.cheng@arm.com>
16689 PR tree-optimization/81799
16690 * tree-loop-distribution.c (version_loop_by_alias_check): Force
16691 cond_expr to simple gimple operand.
16693 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
16695 PR middle-end/46932
16696 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
16698 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
16702 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
16703 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
16704 TARGET_GASISR_PROLOGUES.
16705 * config/avr/avr.c (avr_option_override): Same.
16706 (avr_pass_pre_proep::execute): Same.
16708 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
16711 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
16712 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
16714 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
16716 * config/i386/i386.md (*load_tp_<mode>): Redefine as
16717 define_insn_and_split. Split to a memory load from 0 in
16718 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
16719 using PTR mode iterator.
16720 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
16721 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
16722 (*add_tp_<mode>): Redefine as define_insn_and_split.
16723 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
16724 address space. Merge with *add_tp_x32 using PTR mode iterator.
16725 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
16726 Split to an add with a memory load from 0 in
16727 DEFAULT_TLS_SEG_REG address space.
16729 2017-08-12 Andrew Pinski <apinski@cavium.com>
16731 * config/aarch64/aarch64-option-extensions.def (rdma):
16732 Fix feature string to what Linux prints out in /proc/cpuinfo.
16734 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
16737 * dwarf2out.c (modified_type_die): For C typedef types that have
16738 an ultimate origin, process the ultimate origin instead of the
16740 (gen_typedef_die): Assert that input DECLs have no ultimate
16742 (gen_type_die_with_usage): For typedef variants that have an
16743 ultimate origin, just call gen_decl_die on the original DECL.
16744 (process_scope_var): Avoid creating DIEs for local typedefs and
16745 concrete static variables.
16747 2017-08-12 Alan Modra <amodra@gmail.com>
16751 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
16752 match gnu-user.h startfile.
16753 (ENDFILE_LINUX_SPEC): Similarly.
16755 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
16758 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
16760 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
16762 2017-08-11 Tamar Christina <tamar.christina@arm.com>
16763 * config/aarch64/aarch64.md (mov<mode>): Change.
16764 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16765 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
16766 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
16768 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
16770 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
16771 for storage order barriers.
16773 2017-08-11 Martin Liska <mliska@suse.cz>
16776 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
16777 variables of void type.
16779 2017-08-11 Martin Liska <mliska@suse.cz>
16781 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
16782 TARGET_SUPPORTS_ALIASES.
16783 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
16784 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
16785 (optimize_weakref): Likewise.
16786 * symtab.c (symtab_node::noninterposable_alias): Likewise.
16787 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
16788 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
16790 2017-08-11 Martin Liska <mliska@suse.cz>
16793 * config/i386/i386.c (make_resolver_func): Do complete
16794 refactoring of the function.
16796 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16799 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
16800 * config/i386/i386.c (ix86_stack_protect_guard): Use
16801 ix86_stack_protect_guard_symbol_str to generate varible declaration.
16802 * doc/invoke.texi (x86 Options): Document
16803 -mstack-protector-guard-symbol= option.
16805 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16807 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
16808 * config/i386/i386.c (ix86_split_stack_guard): New function.
16809 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
16810 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
16811 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
16812 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
16813 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
16814 (split_stack_space_check): Call ix86_split_stack_guard.
16816 2017-08-10 Martin Sebor <msebor@redhat.com>
16818 * print-tree.c (print_node): Print location using the established
16820 Replace spaces with colons.
16821 (debug_raw, debug): Ditto.
16823 2017-08-10 Martin Sebor <msebor@redhat.com>
16826 * pretty-print.c (pp_format): Correct the handling of %s precision.
16828 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
16831 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
16833 (ix86_finalize_stack_frame_flags): This. Also clear
16834 frame_pointer_needed if -fno-omit-frame-pointer is used without
16836 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
16837 with ix86_finalize_stack_frame_flags.
16838 (ix86_expand_epilogue): Likewise.
16839 (ix86_expand_split_stack_prologue): Likewise.
16840 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
16842 2017-08-10 Martin Liska <mliska@suse.cz>
16845 * c-attribs.c (handle_target_attribute):
16846 Report warning for an empty string argument of target attribute.
16848 2017-08-09 Jakub Jelinek <jakub@redhat.com>
16851 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
16853 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
16854 or DECL_NONLOCAL labels.
16855 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
16856 or DECL_NONLOCAL labels here.
16858 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
16860 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
16861 to indicate when early gimple folding has been disabled.
16862 (rs6000_gimple_fold_builtin): Add debug content.
16863 (rs6000_invalid_builtin): Fix whitespace.
16864 (rs6000_expand_builtin): Fix whitespace.
16865 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
16867 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
16870 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
16871 SAVE_MULTIPLE if not all the registers that saves, should be saved.
16873 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
16875 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
16876 (qdf24xx): Likewise.
16877 * config/aarch64/aarch64.md: Include falkor.md.
16878 * config/aarch64/falkor.md: New.
16880 2017-08-09 Marek Polacek <polacek@redhat.com>
16883 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
16884 * diagnostic.c (emit_diagnostic): Add a comment.
16885 (emit_diagnostic_valist): New function.
16887 2017-08-09 Marek Polacek <polacek@redhat.com>
16890 * input.c (make_location): New overload.
16891 * input.h (make_location): Declare.
16893 2017-08-08 Alan Modra <amodra@gmail.com>
16894 H.J. Lu <hongjiu.lu@intel.com>
16897 * gcc.c (NO_PIE_SPEC): Delete.
16898 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
16900 (LINK_PIE_SPEC): ..to here.
16901 (LINK_COMMAND_SPEC): Support -no-pie.
16902 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
16903 chain of crtbegin*.o selection, update for PIE_SPEC changes and
16905 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
16906 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
16907 (ENDFILE_CRTEND_SPEC): Similarly.
16909 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
16912 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
16913 (mstack-protector-guard-offset=): Ditto.
16914 * config/i386/i386.c (ix86_option_override): Handle
16915 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
16917 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
16918 ix86_stack_protect_guard_offset variables.
16919 (TARGET_STACK_PROTECT_GUARD): Always define.
16920 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
16921 and -mstack-protector-guard-offset= options.
16923 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16925 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
16926 boundary case for the last candidate.
16928 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16930 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
16931 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
16933 2017-08-08 Tamar Christina <tamar.christina@arm.com>
16935 PR middle-end/19706
16936 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
16937 * config/aarch64/aarch64-builtins.c
16938 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
16939 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
16940 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
16942 2017-08-08 Tamar Christina <tamar.christina@arm.com>
16943 Andrew Pinski <pinskia@gmail.com>
16945 PR middle-end/19706
16946 * internal-fn.def (XORSIGN): New.
16947 * optabs.def (xorsign_optab): New.
16948 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
16949 (convert_expand_mult_copysign): New.
16950 (pass_optimize_widening_mul::execute): Call
16951 convert_expand_mult_copysign.
16953 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16955 PR tree-optimization/81354
16956 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
16957 Insert on edges rather than explicitly creating landing pads.
16958 (analyze_candidates_and_replace): Commit edge inserts.
16960 2017-08-08 Richard Biener <rguenther@suse.de>
16962 PR middle-end/81719
16963 * tree-ssa-loop-niter.c: Include tree-dfa.h.
16964 (expand_simple_operations): Also look through ADDR_EXPRs with
16965 MEM_REF bases treating them as POINTER_PLUS_EXPR.
16967 2017-08-08 Richard Biener <rguenther@suse.de>
16969 PR tree-optimization/81723
16970 * tree-vect-slp.c (struct bst_traits): New hash traits.
16971 (bst_fail): New global.
16972 (vect_build_slp_tree_2): New worker, split out from ...
16973 (vect_build_slp_tree): ... this now wrapping it with using
16974 bst_fail set to cache SLP tree build fails. Properly handle
16976 (vect_analyze_slp_instance): Allocate and free bst_fail.
16978 2017-08-08 Martin Liska <mliska@suse.cz>
16981 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
16982 LABEL_DECLs that can be from a different function.
16984 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16986 PR tree-optimization/81744
16987 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
16988 loop's number of iterations.
16990 2017-08-08 Martin Liska <mliska@suse.cz>
16992 * asan.c: Include header files.
16993 * attribs.c (build_decl_attribute_variant): New function moved
16995 (build_type_attribute_qual_variant): Likewise.
16996 (cmp_attrib_identifiers): Likewise.
16997 (simple_cst_list_equal): Likewise.
16998 (omp_declare_simd_clauses_equal): Likewise.
16999 (attribute_value_equal): Likewise.
17000 (comp_type_attributes): Likewise.
17001 (build_type_attribute_variant): Likewise.
17002 (lookup_ident_attribute): Likewise.
17003 (remove_attribute): Likewise.
17004 (merge_attributes): Likewise.
17005 (merge_type_attributes): Likewise.
17006 (merge_decl_attributes): Likewise.
17007 (merge_dllimport_decl_attributes): Likewise.
17008 (handle_dll_attribute): Likewise.
17009 (attribute_list_equal): Likewise.
17010 (attribute_list_contained): Likewise.
17011 * attribs.h (lookup_attribute): New function moved from tree.[ch].
17012 (lookup_attribute_by_prefix): Likewise.
17013 * bb-reorder.c: Include header files.
17014 * builtins.c: Likewise.
17015 * calls.c: Likewise.
17016 * cfgexpand.c: Likewise.
17017 * cgraph.c: Likewise.
17018 * cgraphunit.c: Likewise.
17019 * convert.c: Likewise.
17020 * dwarf2out.c: Likewise.
17021 * final.c: Likewise.
17022 * fold-const.c: Likewise.
17023 * function.c: Likewise.
17024 * gimple-expr.c: Likewise.
17025 * gimple-fold.c: Likewise.
17026 * gimple-pretty-print.c: Likewise.
17027 * gimple.c: Likewise.
17028 * gimplify.c: Likewise.
17029 * hsa-common.c: Likewise.
17030 * hsa-gen.c: Likewise.
17031 * internal-fn.c: Likewise.
17032 * ipa-chkp.c: Likewise.
17033 * ipa-cp.c: Likewise.
17034 * ipa-devirt.c: Likewise.
17035 * ipa-fnsummary.c: Likewise.
17036 * ipa-inline.c: Likewise.
17037 * ipa-visibility.c: Likewise.
17039 * lto-cgraph.c: Likewise.
17040 * omp-expand.c: Likewise.
17041 * omp-general.c: Likewise.
17042 * omp-low.c: Likewise.
17043 * omp-offload.c: Likewise.
17044 * omp-simd-clone.c: Likewise.
17045 * opts-global.c: Likewise.
17046 * passes.c: Likewise.
17047 * predict.c: Likewise.
17048 * sancov.c: Likewise.
17049 * sanopt.c: Likewise.
17050 * symtab.c: Likewise.
17051 * toplev.c: Likewise.
17052 * trans-mem.c: Likewise.
17053 * tree-chkp.c: Likewise.
17054 * tree-eh.c: Likewise.
17055 * tree-into-ssa.c: Likewise.
17056 * tree-object-size.c: Likewise.
17057 * tree-parloops.c: Likewise.
17058 * tree-profile.c: Likewise.
17059 * tree-ssa-ccp.c: Likewise.
17060 * tree-ssa-live.c: Likewise.
17061 * tree-ssa-loop.c: Likewise.
17062 * tree-ssa-sccvn.c: Likewise.
17063 * tree-ssa-structalias.c: Likewise.
17064 * tree-ssa.c: Likewise.
17065 * tree-streamer-in.c: Likewise.
17066 * tree-vectorizer.c: Likewise.
17067 * tree-vrp.c: Likewise.
17068 * tsan.c: Likewise.
17069 * ubsan.c: Likewise.
17070 * varasm.c: Likewise.
17071 * varpool.c: Likewise.
17072 * tree.c: Remove functions moved to attribs.[ch].
17073 * tree.h: Likewise.
17074 * config/aarch64/aarch64.c: Add attrs.h header file.
17075 * config/alpha/alpha.c: Likewise.
17076 * config/arc/arc.c: Likewise.
17077 * config/arm/arm.c: Likewise.
17078 * config/avr/avr.c: Likewise.
17079 * config/bfin/bfin.c: Likewise.
17080 * config/c6x/c6x.c: Likewise.
17081 * config/cr16/cr16.c: Likewise.
17082 * config/cris/cris.c: Likewise.
17083 * config/darwin.c: Likewise.
17084 * config/epiphany/epiphany.c: Likewise.
17085 * config/fr30/fr30.c: Likewise.
17086 * config/frv/frv.c: Likewise.
17087 * config/ft32/ft32.c: Likewise.
17088 * config/h8300/h8300.c: Likewise.
17089 * config/i386/winnt.c: Likewise.
17090 * config/ia64/ia64.c: Likewise.
17091 * config/iq2000/iq2000.c: Likewise.
17092 * config/lm32/lm32.c: Likewise.
17093 * config/m32c/m32c.c: Likewise.
17094 * config/m32r/m32r.c: Likewise.
17095 * config/m68k/m68k.c: Likewise.
17096 * config/mcore/mcore.c: Likewise.
17097 * config/microblaze/microblaze.c: Likewise.
17098 * config/mips/mips.c: Likewise.
17099 * config/mmix/mmix.c: Likewise.
17100 * config/mn10300/mn10300.c: Likewise.
17101 * config/moxie/moxie.c: Likewise.
17102 * config/msp430/msp430.c: Likewise.
17103 * config/nds32/nds32-isr.c: Likewise.
17104 * config/nds32/nds32.c: Likewise.
17105 * config/nios2/nios2.c: Likewise.
17106 * config/nvptx/nvptx.c: Likewise.
17107 * config/pa/pa.c: Likewise.
17108 * config/pdp11/pdp11.c: Likewise.
17109 * config/powerpcspe/powerpcspe.c: Likewise.
17110 * config/riscv/riscv.c: Likewise.
17111 * config/rl78/rl78.c: Likewise.
17112 * config/rx/rx.c: Likewise.
17113 * config/s390/s390.c: Likewise.
17114 * config/sh/sh.c: Likewise.
17115 * config/sol2.c: Likewise.
17116 * config/sparc/sparc.c: Likewise.
17117 * config/spu/spu.c: Likewise.
17118 * config/stormy16/stormy16.c: Likewise.
17119 * config/tilegx/tilegx.c: Likewise.
17120 * config/tilepro/tilepro.c: Likewise.
17121 * config/v850/v850.c: Likewise.
17122 * config/vax/vax.c: Likewise.
17123 * config/visium/visium.c: Likewise.
17124 * config/xtensa/xtensa.c: Likewise.
17126 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17129 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
17130 constraints since the -mupper-regs-* switches have been
17132 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
17133 into a vector from a double word element that was extracted from
17134 another vector, and eliminate extra XXPERMDI instructions.
17135 (vsx_concat_<mode>_2): Likewise.
17136 (vsx_concat_<mode>_3): Likewise.
17137 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
17138 concat to allow optimizing inserts from previous extracts.
17140 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
17142 * config/i386/i386.c (ix86_stack_protect_guard): Generate
17143 memory reference to a SSP offset in TLS address space.
17144 (ix86_print_operand) <case '@'>: Remove.
17145 (ix86_print_operand_punct_valid_p): Remove '@' code.
17146 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
17147 UNSPEC_SP_TLS_TEST.
17148 (stack_tls_protect_set_<mode>): Remove.
17149 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
17150 (stack_tls_protect_test_<mode>): Remove.
17151 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
17153 2017-08-07 Olivier Hainque <hainque@adacore.com>
17156 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
17158 2017-08-07 Douglas Rupp <rupp@adacore.com>
17160 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
17161 variable was referenced as multidir in command.
17163 2017-08-07 Jakub Jelinek <jakub@redhat.com>
17166 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
17169 2017-08-07 Martin Liska <mliska@suse.cz>
17171 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
17172 * config/rl78/rl78.c: Add include of attribs.h.
17173 * config/sh/sh.c: Likewise.
17174 * config/v850/v850.c: Likewise.
17176 2017-08-07 Tom de Vries <tom@codesourcery.com>
17178 PR middle-end/78266
17179 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
17181 2017-08-07 Martin Liska <mliska@suse.cz>
17183 * config/mips/mips.c: Include attribs.h.
17185 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
17188 * doc/invoke.texi: Document change in behvaior for -Ofast for
17191 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
17193 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
17195 (aarch64_pop_regs): Likewise.
17196 (aarch64_gen_load_pair): Likewise.
17197 (aarch64_save_callee_saves): Likewise.
17198 (aarch64_restore_callee_saves): Likewise.
17200 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17202 * config/i386/i386.c: Revert the last change.
17204 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17207 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17209 (ix86_finalize_stack_frame_flags): This. Also clear
17210 frame_pointer_needed if -fno-omit-frame-pointer is used without
17212 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17213 with ix86_finalize_stack_frame_flags.
17214 (ix86_expand_epilogue): Likewise.
17215 (ix86_expand_split_stack_prologue): Likewise.
17217 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17220 * config/i386/i386.c (get_builtin_code_for_version): Set priority
17221 to P_AES for Westmere.
17223 2017-08-07 Jonathan Yong <10walls@gmail.com>
17225 * config/i386/mingw.opt (fset-stack-executable): Removed.
17226 * config/i386/cygming.opt (fset-stack-executable): Moved
17228 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
17230 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
17232 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
17234 2017-08-07 Marek Polacek <polacek@redhat.com>
17236 PR middle-end/81737
17237 * fold-const.c (fold_indirect_ref_1): Check type_domain.
17239 2017-08-07 Martin Liska <mliska@suse.cz>
17241 * attribs.h (canonicalize_attr_name): New function.
17242 (cmp_attribs): Move from c-format.c and adjusted.
17243 (is_attribute_p): Moved from tree.h.
17244 * tree-inline.c: Add new includes.
17245 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
17246 (private_is_attribute_p): Remove.
17247 (private_lookup_attribute): Likewise.
17248 (private_lookup_attribute_by_prefix): Simplify.
17249 (remove_attribute): Use is_attribute_p.
17250 * tree.h: Remove removed declarations.
17252 2017-08-07 Jakub Jelinek <jakub@redhat.com>
17254 PR middle-end/81698
17255 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
17256 instead of computing it in the function. Formatting fix.
17257 (expand_case): Don't rely on default_edge being the first edge,
17258 clear it if removing it, pass default_edge to
17259 emit_case_dispatch_table.
17260 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
17263 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
17265 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
17266 insn in the function, emit NOP after the insn.
17268 2017-08-06 Tom de Vries <tom@codesourcery.com>
17270 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
17273 2017-08-06 Tom de Vries <tom@codesourcery.com>
17275 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
17278 2017-08-04 Yury Gribov <tetra2005@gmail.com>
17280 PR tree-optimization/57371
17281 * match.pd: New pattern.
17283 2017-08-04 Marek Polacek <polacek@redhat.com>
17285 PR middle-end/81695
17286 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
17287 perform the computation in offset_int.
17289 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17291 PR tree-optimization/81136
17292 * tree-vectorizer.h: Include tree-hash-traits.h.
17293 (vec_base_alignments): New typedef.
17294 (vec_info): Add a base_alignments field.
17295 (vect_record_base_alignments): Declare.
17296 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
17298 (DR_IS_CONDITIONAL_IN_STMT): New macro.
17299 (create_data_ref): Add an is_conditional_in_stmt argument.
17300 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
17301 the is_conditional_in_stmt field.
17302 (data_ref_loc): Add an is_conditional_in_stmt field.
17303 (get_references_in_stmt): Set the is_conditional_in_stmt field.
17304 (find_data_references_in_stmt): Update call to create_data_ref.
17305 (graphite_find_data_references_in_stmt): Likewise.
17306 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
17307 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
17308 (vect_record_base_alignment): New function.
17309 (vect_record_base_alignments): Likewise.
17310 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
17311 for nested statements even if we fail to compute a misalignment.
17312 Use pooled base alignments for unconditional references.
17313 (vect_find_same_alignment_drs): Compare base addresses instead
17315 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
17316 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
17318 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17320 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
17321 Add an explicit name for the enum. Use auto_vec for slp_instances
17322 and grouped_stores.
17323 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
17325 (_bb_vec_info): Add a constructor and destructor.
17326 (vinfo_for_stmt): Return NULL for uids of -1 as well.
17327 (destroy_loop_vec_info): Delete.
17328 (vect_destroy_datarefs): Likewise.
17329 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
17330 (vec_info::vec_info): New function.
17331 (vec_info::~vec_info): Likewise.
17332 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
17333 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
17334 destroy_loop_vec_info.
17335 * tree-vect-loop.c (new_loop_vec_info): Replace with...
17336 (_loop_vec_info::_loop_vec_info): ...this.
17337 (destroy_loop_vec_info): Replace with...
17338 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
17339 the stmt_vec_infos. Leave handling of vec_info information to its
17340 destructor. Remove explicit vector releases.
17341 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
17342 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
17343 * tree-vect-slp.c (new_bb_vec_info): Replace with...
17344 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
17345 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
17346 (destroy_bb_vec_info): Replace with...
17347 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
17348 information to its destructor.
17349 (vect_slp_analyze_bb_1): Use new and delete instead of
17350 new_bb_vec_info and destroy_bb_vec_info.
17351 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
17354 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17356 * tree-data-ref.h (subscript): Add access_fn field.
17357 (data_dependence_relation): Add could_be_independent_p.
17358 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
17359 (same_access_functions): Move to tree-data-ref.c.
17360 * tree-data-ref.c (ref_contains_union_access_p): New function.
17361 (access_fn_component_p): Likewise.
17362 (access_fn_components_comparable_p): Likewise.
17363 (dr_analyze_indices): Add a reference to access_fn_component_p.
17364 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
17366 (constant_access_functions): Likewise.
17367 (add_other_self_distances): Likewise.
17368 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
17369 (initialize_data_dependence_relation): Use XCNEW and remove
17370 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
17371 of access functions that have the same type. Allow the
17372 subsequence to end with different bases in some circumstances.
17373 Record the chosen access functions in SUB_ACCESS_FN.
17374 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
17375 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
17376 (subscript_dependence_tester_1): Likewise dra and drb.
17377 (build_classic_dist_vector): Update calls accordingly.
17378 (subscript_dependence_tester): Likewise.
17379 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
17380 DDR_COULD_BE_INDEPENDENT_P.
17381 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
17382 comp_alias_ddrs instead of may_alias_ddrs.
17383 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
17385 (vect_analyze_data_ref_dependence): Use it if
17386 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
17387 distance vectors if that fails.
17388 (dependence_distance_ge_vf): New function.
17389 (vect_prune_runtime_alias_test_list): Use it. Don't clear
17390 LOOP_VINFO_MAY_ALIAS_DDRS.
17392 2017-08-04 Richard Biener <rguenther@suse.de>
17394 PR middle-end/81705
17395 * fold-const.c (fold_binary_loc): Properly restrict
17396 minus_var0 && minus_var1 case when associating undefined overflow
17399 2017-08-04 Tom de Vries <tom@codesourcery.com>
17401 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
17403 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17405 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17406 Don't start diagnostic messages with a capital letter.
17407 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17409 (rs6000_invalid_builtin): Likewise.
17410 (rs6000_trampoline_init): Likewise.
17412 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17415 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
17416 after setting changeable df flags.
17418 2017-08-03 Richard Biener <rguenther@suse.de>
17420 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
17421 up if the use is in USE - X.
17423 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
17425 * toplev.c (dumpfile.h): New include.
17426 (internal_error_reentered): New static function. Use it...
17427 (internal_error_function): ...here to handle reentered internal_error.
17429 2017-08-03 Richard Biener <rguenther@suse.de>
17431 PR middle-end/81148
17432 * fold-const.c (split_tree): Add minus_var and minus_con
17433 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
17434 here but always use minus_*.
17435 (associate_trees): Assert we never associate with MINUS_EXPR
17436 and NULL first operand. Do not recurse for PLUS_EXPR operands
17437 when associating as MINUS_EXPR either.
17438 (fold_binary_loc): Track minus_var and minus_con.
17440 2017-08-03 Tom de Vries <tom@codesourcery.com>
17443 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
17444 ACCEL_COMPILER, apply finish_options on
17445 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
17447 2017-08-03 Tom de Vries <tom@codesourcery.com>
17450 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
17451 function_entry_patch_area_size > 0.
17453 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17456 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
17457 instead of 10??LU, perform unit multiplication in wide_int,
17458 don't change alloc_object_size_limit if the limit is larger
17461 PR tree-optimization/81655
17462 PR tree-optimization/81588
17463 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
17464 the case when ranges[i].low and high are 1 for unsigned type with
17467 PR middle-end/81052
17468 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
17469 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
17471 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17473 * tree-vrp.h: Add include guard.
17475 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
17478 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
17479 (ud2): New insn pattern.
17480 * config/i386/i386.c (ix86_expand_epilogue):
17481 For naked functions, generate ud2 instead of trap insn.
17483 2017-08-02 Marek Polacek <polacek@redhat.com>
17486 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
17488 2017-08-02 Tom de Vries <tom@codesourcery.com>
17489 Cesar Philippidis <cesar@codesourcery.com>
17491 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
17492 Add missing edge probabilities.
17494 2017-08-02 Tamar Christina <tamar.christina@arm.com>
17496 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
17497 Correct endianness.
17499 2017-08-02 Jakub Jelinek <jakub@redhat.com>
17501 PR middle-end/79499
17502 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
17503 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
17504 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
17505 prologue_seq sequences - if any.
17507 2017-08-02 Richard Biener <rguenther@suse.de>
17509 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
17510 via vectors if supported, integer extracts via punning if supported
17511 or otherwise vector extracts.
17513 2017-08-02 Richard Biener <rguenther@suse.de>
17515 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
17517 (bitmap_insert_into_set): ... this.
17519 2017-08-02 Richard Biener <rguenther@suse.de>
17521 PR tree-optimization/81633
17523 2015-08-17 Alan Hayward <alan.hayward@arm.com>
17525 PR tree-optimization/71752
17526 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
17528 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
17530 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
17531 (machine_function::call_ms2sysv_pad_out): Remove field.
17532 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
17533 (ix86_compute_frame_layout): Likewise.
17535 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
17538 * config/i386/i386.c (ix86_set_func_type): Disallow naked
17539 attribute with interrupt attribute.
17541 2017-08-01 Andrew Pinski <apinski@cavium.com>
17543 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
17544 BIT_INSERT_EXPR's operand 1
17545 to see if the types precision matches.
17547 2017-08-01 Martin Liska <mliska@suse.cz>
17549 PR middle-end/70140
17550 * builtins.c (expand_builtin_memcpy_args): Remove.
17551 (expand_builtin_memcpy): Call newly added function
17552 expand_builtin_memory_copy_args.
17553 (expand_builtin_memcpy_with_bounds): Likewise.
17554 (expand_builtin_mempcpy): Remove last argument.
17555 (expand_builtin_mempcpy_with_bounds): Likewise.
17556 (expand_builtin_memory_copy_args): New function created from
17557 expand_builtin_mempcpy_args with small modifications.
17558 (expand_builtin_mempcpy_args): Remove.
17559 (expand_builtin_stpcpy): Remove unused argument.
17560 (expand_builtin): Likewise.
17561 (expand_builtin_with_bounds): Likewise.
17563 2017-08-01 Martin Liska <mliska@suse.cz>
17566 Make mempcpy more optimal (PR middle-end/70140).
17568 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17571 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
17572 __builtin_vec_cmpne verify both arguments are compatible vectors
17573 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
17574 verify arg1_type is a pointer or array type. For __builtin_vec_st,
17575 move computation of aligned to after checking the argument types.
17579 * config/rs6000/vsx.md (vextract_fp_from_shorth,
17580 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
17583 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
17584 Doug Rupp <rupp@adacore.com>
17585 Olivier Hainque <hainque@adacore.com>
17587 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
17588 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
17590 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
17591 for TARGET_OS_CPP_BUILTIN.
17592 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
17593 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
17595 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
17596 passing required abi options to the assembler for EABI configurations.
17597 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
17598 of .text.hot and .text.unlikely sections for kernel modules when
17599 using ARM style exceptions.
17600 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
17601 options. Add EXTRA_CC1_SPEC.
17602 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
17604 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
17606 (ARM_TARGET2_DWARF_FORMAT): Define.
17607 * config/arm/t-vxworks: Adjust multilib control to removal of the
17608 Diab command line options.
17610 2017-08-01 Martin Liska <mliska@suse.cz>
17612 PR gcov-profile/81561
17613 * gcov.c (unblock): Make unblocking safe as we need to preserve
17614 index correspondence of blocks and block_lists.
17616 2017-08-01 Richard Biener <rguenther@suse.de>
17618 PR tree-optimization/81181
17619 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
17620 (compute_antic): ... end of iteration here.
17622 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
17624 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
17625 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
17626 (ftree-slp-vectorize): Likewise.
17627 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
17628 can no longer be set independent of flag_tree_loop_vectorize.
17629 * omp-general.c (emp_max_vf): Likewise.
17630 * opts.c (enable_fdo_optimizations): Remove references to
17631 flag_tree_vectorize, these are now implicit.
17632 (common_handle_option): Remove handling for OPT_ftree_vectorize,
17633 and leave it for the options machinery.
17635 2017-08-01 Martin Liska <mliska@suse.cz>
17637 PR middle-end/70140
17638 * builtins.c (expand_builtin_memcpy_args): Remove.
17639 (expand_builtin_memcpy): Call newly added function
17640 expand_builtin_memory_copy_args.
17641 (expand_builtin_memcpy_with_bounds): Likewise.
17642 (expand_builtin_mempcpy): Remove last argument.
17643 (expand_builtin_mempcpy_with_bounds): Likewise.
17644 (expand_builtin_memory_copy_args): New function created from
17645 expand_builtin_mempcpy_args with small modifications.
17646 (expand_builtin_mempcpy_args): Remove.
17647 (expand_builtin_stpcpy): Remove unused argument.
17648 (expand_builtin): Likewise.
17649 (expand_builtin_with_bounds): Likewise.
17651 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
17654 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
17655 print "ds:" only for immediates in generic address space.
17657 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
17660 * config/i386/i386.c (ix86_funciton_naked): New prototype.
17661 (ix86_function_ok_for_sibcall): Return false for naked functions.
17663 2017-08-01 Richard Biener <rguenther@suse.de>
17665 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
17666 (compute_antic): Seed worklist with exit block predecessors.
17667 * cfganal.c (dfs_find_deadend): For a cycle return the source
17668 of the edge closing it.
17670 2017-08-01 Tamar Christina <tamar.christina@arm.com>
17672 * config/aarch64/aarch64.c
17673 (aarch64_can_const_movi_rtx_p): Move 0 check.
17675 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17677 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
17678 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
17680 * match.pd: Ditto in address comparison pattern.
17682 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17684 PR tree-optimization/81627
17685 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
17686 closed ssa form for store-store chain.
17688 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17690 PR tree-optimization/81620
17691 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
17692 for store-store chain.
17694 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17696 PR tree-optimization/81588
17697 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
17698 ranges[i].in_p, invert comparison code ccode. For >/>=,
17699 swap rhs1 and rhs2 and comparison code unconditionally,
17700 for </<= don't do that. Don't swap rhs1/rhs2 again if
17701 ranges[i].in_p, instead invert comparison code ccode if
17702 opcode or oe->rank is BIT_IOR_EXPR.
17705 * optabs.def (vec_extract_optab, vec_init_optab): Change from
17706 a direct optab to conversion optab.
17707 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
17708 with GET_MODE_INNER as last argument instead of optab_handler.
17709 * expmed.c (extract_bit_field_1): Likewise. Use vector from
17710 vector extraction if possible and optab is available.
17711 * expr.c (store_constructor): Use convert_optab_handler instead
17712 of optab_handler. Use vector initialization from smaller
17713 vectors if possible and optab is available.
17714 * tree-vect-stmts.c (vectorizable_load): Likewise.
17715 * doc/md.texi (vec_extract, vec_init): Document that the optabs
17716 now have two modes.
17717 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
17718 of vec_init from half-sized vectors with the same element mode.
17719 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
17720 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
17721 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
17722 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
17723 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
17724 after mode in gen_vec_extract* calls.
17725 (vec_extract<mode>): Renamed to ...
17726 (vec_extract<mode><ssescalarmodelower>): ... this.
17727 (vec_extract<mode><ssehalfvecmodelower>): New expander.
17728 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
17729 element mode after mode in gen_vec_init* calls.
17730 (VEC_INIT_HALF_MODE): New mode iterator.
17731 (vec_init<mode>): Renamed to ...
17732 (vec_init<mode><ssescalarmodelower>): ... this.
17733 (vec_init<mode><ssehalfvecmodelower>): New expander.
17734 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
17735 (vec_extractv2sfsf): ... this.
17736 (vec_initv2sf): Renamed to ...
17737 (vec_initv2sfsf): ... this.
17738 (vec_extractv2si): Renamed to ...
17739 (vec_extractv2sisi): ... this.
17740 (vec_initv2si): Renamed to ...
17741 (vec_initv2sisi): ... this.
17742 (vec_extractv4hi): Renamed to ...
17743 (vec_extractv4hihi): ... this.
17744 (vec_initv4hi): Renamed to ...
17745 (vec_initv4hihi): ... this.
17746 (vec_extractv8qi): Renamed to ...
17747 (vec_extractv8qiqi): ... this.
17748 (vec_initv8qi): Renamed to ...
17749 (vec_initv8qiqi): ... this.
17750 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
17751 (vec_init<mode>): Renamed to ...
17752 (vec_init<mode><VEC_base_l>): ... this.
17753 (vec_extract<mode>): Renamed to ...
17754 (vec_extract<mode><VEC_base_l>): ... this.
17755 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
17756 (vec_initv2sfsf): ... this.
17757 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
17758 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17759 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
17760 element mode after mode in gen_vec_init* calls.
17761 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
17762 (vec_init<mode><Vel>): ... this.
17763 (vec_extract<mode>): Renamed to ...
17764 (vec_extract<mode><Vel>): ... this.
17765 * config/aarch64/iterators.md (Vel): New mode attribute.
17766 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
17767 Add element mode after mode in gen_vec_extract* calls.
17768 * config/s390/vector.md (non_vec_l): New mode attribute.
17769 (vec_extract<mode>): Renamed to ...
17770 (vec_extract<mode><non_vec_l>): ... this.
17771 (vec_init<mode>): Renamed to ...
17772 (vec_init<mode><non_vec_l>): ... this.
17773 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
17774 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
17776 * config/arm/iterators.md (V_elem_l): New mode attribute.
17777 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
17778 (vec_extract<mode><V_elem_l>): ... this.
17779 (vec_extractv2di): Renamed to ...
17780 (vec_extractv2didi): ... this.
17781 (vec_init<mode>): Renamed to ...
17782 (vec_init<mode><V_elem_l>): ... this.
17783 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
17784 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
17785 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
17786 Add element mode after gen_vec_extract* calls.
17787 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
17788 (vec_init<mode><unitmode>): ... this.
17789 (vec_extract<mode>): Renamed to ...
17790 (vec_extract<mode><unitmode>): ... this.
17791 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
17792 (vec_init<mode><unitmode>): ... this.
17793 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
17794 (vec_initv2sfsf): ... this.
17795 (vec_extractv2sf): Renamed to ...
17796 (vec_extractv2sfsf): ... this.
17797 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
17798 Add element mode after gen_vec_extract* calls.
17799 * config/mips/mips.md (unitmode): New mode iterator.
17800 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
17801 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
17802 * config/spu/spu.md (inner_l): New mode attribute.
17803 (vec_init<mode>): Renamed to ...
17804 (vec_init<mode><inner_l>): ... this.
17805 (vec_extract<mode>): Renamed to ...
17806 (vec_extract<mode><inner_l>): ... this.
17807 * config/sparc/sparc.md (veltmode): New mode iterator.
17808 (vec_init<VMALL:mode>): Renamed to ...
17809 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
17810 * config/ia64/vect.md (vec_initv2si): Renamed to ...
17811 (vec_initv2sisi): ... this.
17812 (vec_initv2sf): Renamed to ...
17813 (vec_initv2sfsf): ... this.
17814 (vec_extractv2sf): Renamed to ...
17815 (vec_extractv2sfsf): ... this.
17816 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
17817 (vec_init<mode>): Renamed to ...
17818 (vec_init<mode><VEC_base_l>): ... this.
17819 (vec_extract<mode>): Renamed to ...
17820 (vec_extract<mode><VEC_base_l>): ... this.
17821 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
17822 (vec_initv2sfsf): ... this.
17823 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
17824 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17825 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
17826 gen_vec_init* calls.
17828 2017-08-01 Richard Biener <rguenther@suse.de>
17830 PR tree-optimization/81297
17831 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
17832 TREE_OVERFLOW from INTEGER_CSTs.
17834 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
17836 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
17838 2017-07-31 Carl Love <cel@us.ibm.com>
17840 * config/rs6000/rs6000-c: Add support for built-in functions
17841 vector signed char vec_xl_be (signed long long, signed char *);
17842 vector unsigned char vec_xl_be (signed long long, unsigned char *);
17843 vector signed int vec_xl_be (signed long long, signed int *);
17844 vector unsigned int vec_xl_be (signed long long, unsigned int *);
17845 vector signed long long vec_xl_be (signed long long, signed long long *);
17846 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
17847 vector signed short vec_xl_be (signed long long, signed short *);
17848 vector unsigned short vec_xl_be (signed long long, unsigned short *);
17849 vector double vec_xl_be (signed long long, double *);
17850 vector float vec_xl_be (signed long long, float *);
17851 * config/rs6000/altivec.h (vec_xl_be): Add #define.
17852 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
17853 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
17855 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
17856 (altivec_expand_builtin): Add switch statement to call
17857 altivec_expand_xl_be for each builtin.
17858 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
17859 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
17860 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
17861 __builtin_vsx_le_be_v16qi.
17862 * doc/extend.texi: Update the built-in documentation file for the
17863 new built-in functions.
17865 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17868 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
17870 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17872 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17874 * config.gcc: Add z14.
17875 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
17876 CPU model numbers for z13s and z14.
17877 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
17879 * config/s390/s390-opts.h (enum processor_type): Rename
17880 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
17881 * config/s390/s390.c (processor_table): Add field for CPU name to
17882 be passed to Binutils.
17883 (s390_asm_output_machine_for_arch): Use the new field in
17884 processor_table for Binutils.
17885 (s390_expand_builtin): Replace arch12 with z14.
17886 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
17887 (s390_get_sched_attrmask): Likewise.
17888 (s390_get_unit_mask): Likewise.
17889 * config/s390/s390.opt: Add z14 to processor_type enum.
17891 2017-07-31 Martin Jambor <mjambor@suse.cz>
17894 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
17895 regardless of optimization level.
17897 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
17898 Martin Liska <mliska@suse.cz>
17900 * predict.def: Remove old comment and adjust probability.
17901 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
17902 PREDICT statements.
17904 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17907 * config/i386/i386.c (ix86_function_naked): New function.
17908 (ix86_can_use_return_insn_p): Return false for naked functions.
17909 (ix86_expand_prologue): Skip prologue for naked functions.
17910 (ix86_expand_epilogue): Skip epilogue for naked functions
17911 and emit trap instruction.
17912 (ix86_warn_func_return): New function.
17913 (ix86_attribute_table): Add "naked" attribute specification.
17914 (TARGET_WARN_FUNC_RETURN): Define.
17915 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
17917 2017-07-31 Martin Liska <mliska@suse.cz>
17919 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
17920 (dump_gimple_bb_header): Always dump BB info.
17921 (pp_cfg_jump): Do not append info about BB when dumping a jump.
17923 2017-07-31 Martin Liska <mliska@suse.cz>
17926 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
17927 also with current_function_decl non-null equality.
17929 2017-07-31 Jakub Jelinek <jakub@redhat.com>
17932 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
17933 change type to the element type, instead add eltype variable and
17934 use it where we are interested in the element type.
17936 PR tree-optimization/81603
17937 * ipa-polymorphic-call.c
17938 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
17939 offset arithmetic in offset_int, bail out if the resulting bit offset
17940 doesn't fit into shwi.
17942 2017-07-31 Martin Liska <mliska@suse.cz>
17944 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
17945 (gimplify_save_expr): Fix comment.
17947 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
17950 * config/i386/i386.c (ix86_function_arg): Update arguments for
17952 (ix86_compute_frame_layout): Set the initial stack offset to
17953 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
17954 INCOMING_FRAME_SP_OFFSET.
17955 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
17956 stack before exception handler returns.
17957 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
17958 the 'ERROR_CODE' for exception handler.
17960 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
17962 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
17963 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
17964 (ASM_OUTPUT_REG_POP): Ditto.
17965 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
17966 instead of asm_fprintf to output pure string.
17968 2017-07-29 Jakub Jelinek <jakub@redhat.com>
17970 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
17971 to imported_module_or_decl hook.
17972 (debug_nothing_tree_tree_tree_bool): Remove.
17973 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
17974 * debug.c (do_nothing_debug_hooks): Use
17975 debug_nothing_tree_tree_tree_bool_bool instead of
17976 debug_nothing_tree_tree_tree_bool.
17977 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17978 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
17979 * sdbout.c (sdb_debug_hooks): Likewise.
17980 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
17981 (gen_namespace_die): Add DW_AT_export_symbols attribute if
17983 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
17984 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
17985 attribute, don't add anything.
17987 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17989 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
17990 (fold_build2_stat_loc): Likewise.
17991 (fold_build3_stat_loc): Likewise.
17992 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
17993 (fold_build1_loc): Remove macro.
17994 (fold_build2_loc): Likewise.
17995 (fold_build3_loc): Likewise.
17997 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17999 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
18000 (gimple_build_debug_bind_source_stat): Likewise.
18001 * gimple.h (gimple_build_debug_bind): Remove macro.
18002 (gimple_build_debug_bind_source): Likewise.
18004 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18006 * bitmap.c (bitmap_alloc): Adjust.
18007 (bitmap_gc_alloc): Likewise.
18008 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
18010 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18012 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
18013 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
18014 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
18015 (bitmap_gc_alloc_stat): Likewise.
18016 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
18018 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18020 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
18021 * rtl.h (shallow_copy_rtx): Remove macro.
18023 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18025 * emit-rtl.c (gen_raw_REG): Adjust.
18026 * gengenrtl.c (gendef): Likewise.
18027 * rtl.c (rtx_alloc_stat): Remove _stat from name.
18028 * rtl.h (rtx_alloc): Remove macro.
18030 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18032 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
18033 (build_tree_list_stat): Likewise.
18034 * tree.h (build_tree_list): Remove macro.
18035 (build_tree_list_vec): Likewise.
18037 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18039 * tree.c (make_vector_stat): Remove _stat from name.
18040 (build_vector_stat): Likewise.
18041 * tree.h (make_vector_stat): Remove macro.
18042 (build_vector_stat): Likewise.
18044 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18046 * tree.h (build_var_debug_value): Remove prototype.
18048 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18050 * tree.c (tree_cons_stat): Remove _stat from name.
18051 * tree.h (tree_cons): Remove macro.
18053 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18055 * tree.c (build_vl_exp_stat): Remove _stat from name.
18056 * tree.h (build_vl_exp): Remove macro.
18058 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18060 * tree.c (build_decl_stat): Remove _stat from name.
18061 * tree.h (build_decl): Remove macro.
18063 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18065 * gimple.c (gimple_build_with_ops_stat): Adjust.
18066 (gimple_alloc_stat): Remove _stat from name.
18067 * gimple.h (gimple_alloc): Remove macro.
18069 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18071 * tree.c (make_tree_vec_stat): Remove _stat from name.
18072 (grow_tree_vec_stat): Likewise.
18073 * tree.h (make_tree_vec_stat): Adjust prototype.
18074 (grow_tree_vec_stat): Likewise.
18075 (make_tree_vec): Remove macro.
18076 (grow_tree_vec): Likewise.
18078 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18080 * fold-const.c (fold_build1_stat_loc): Adjust.
18081 (fold_build2_stat_loc): Likewise.
18082 (fold_build3_stat_loc): Likewise.
18083 * tree.c (build0_stat): Remove _stat from name.
18084 (build1_stat): Likewise.
18085 (build2_stat): Likewise.
18086 (build3_stat): Likewise.
18087 (build4_stat): Likewise.
18088 (build5_stat): Likewise.
18089 * tree.h (build1_loc): Remove macro, and rename _stat function
18091 (build2_loc): Likewise.
18092 (build3_loc): Likewise.
18093 (build4_loc): Likewise.
18094 (build5_loc): Likewise.
18096 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18098 * tree.c (make_int_cst_stat): Remove _stat from name.
18099 * tree.h (make_int_cst_stat): Adjust prototype.
18100 (make_int_cst): Remove macro.
18102 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18104 * tree.c (make_tre_binfo_stat): Remove _stat from name.
18105 * tree.h (make_tree_binfo_stat): Adjust prototype.
18106 (make_tree_binfo): Remove.
18108 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18110 * tree.c (copy_node_stat): Rename to copy_node.
18111 (build_distinct_type_copy): Adjust.
18112 * tree.h (copy_node_stat): Adjust prototype.
18113 (copy_node): Remove macro.
18115 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18117 * tree.c (make_node_stat): rename to make_node.
18118 (build_tree_list_stat): Adjust.
18119 (build0_stat): Likewise.
18120 (build2_stat): Likewise.
18121 (build3_stat): Likewise.
18122 (build4_stat): Likewise.
18123 (build5_stat): Likewise.
18124 (build_decl_stat): Likewise.
18125 * tree.h (make_node_stat): Adjust prototype.
18126 (make_node): remove macro.
18128 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
18130 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
18131 (PPC_FEATURE2_SCV): Likewise.
18132 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
18134 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18136 * config/aarch64/aarch64.c
18137 (aarch64_internal_mov_immediate): Add new special pattern.
18138 * config/aarch64/aarch64.md (*movdi_aarch64):
18139 Add reg/32bit const mov case.
18141 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18142 Richard Sandiford <richard.sandiford@linaro.org>
18144 * config/aarch64/aarch64.md (mov<mode>): Generalize.
18145 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18146 Add integer and movi cases.
18147 (movi-split-hf-df-sf split, fp16): New.
18148 (enabled): Added TARGET_FP_F16INST.
18149 * config/aarch64/iterators.md (GPF_HF): New.
18150 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
18152 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18154 * config/aarch64/aarch64.c
18155 (aarch64_simd_container_mode): Add prototype.
18156 (aarch64_expand_mov_immediate): Add HI support.
18157 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
18158 (aarch64_can_const_movi_rtx_p): New.
18159 (aarch64_preferred_reload_class):
18160 Remove restrictions of using FP registers for certain SIMD operations.
18161 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
18162 (aarch64_valid_floating_const): Add integer move validation.
18163 (aarch64_simd_imm_scalar_p): Remove.
18164 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
18165 (aarch64_legitimate_constant_p): Expand list of supported cases.
18166 * config/aarch64/aarch64-protos.h
18167 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
18168 (aarch64_reinterpret_float_as_int): New.
18169 (aarch64_simd_imm_scalar_p): Remove.
18170 * config/aarch64/constraints.md (Uvi): New.
18171 (Dd): Split into Ds and new Dd.
18172 * config/aarch64/aarch64.md (*movsi_aarch64):
18174 (*movdi_aarch64): Add SIMD mov case.
18176 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18178 * tree-predcom.c: (struct chain): Handle store-store chain in which
18179 stores for elimination only store loop invariant values.
18180 (execute_pred_commoning_chain): Ditto.
18181 (prepare_initializers_chain_store_elim): Ditto.
18182 (prepare_finalizers): Ditto.
18183 (is_inv_store_elimination_chain): New function.
18184 (initialize_root_vars_store_elim_1): New function.
18186 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18188 * tree-predcom.c: Revise general description of the pass.
18189 (enum chain_type): New enum type for store elimination.
18190 (struct chain): New field supporting store elimination.
18191 (struct component): Ditto.
18192 (dump_chain): Dump store-stores chain.
18193 (release_chain): Release resources.
18194 (split_data_refs_to_components): Compute and create component
18195 contains only stores for elimination.
18196 (get_chain_last_ref_at): New function.
18197 (make_invariant_chain): Initialization.
18198 (make_rooted_chain): Specify chain type in parameter and record it.
18199 (add_looparound_copies): Skip for store-stores chain.
18200 (determine_roots_comp): Compute type of chain and pass it to
18202 (initialize_root_vars_store_elim_2): New function.
18203 (finalize_eliminated_stores): New function.
18204 (remove_stmt): Handle store for elimination.
18205 (execute_pred_commoning_chain): Execute predictive commoning on
18206 store-store chains.
18207 (determine_unroll_factor): Skip unroll for store-stores chain.
18208 (prepare_initializers_chain_store_elim): New function.
18209 (prepare_initializers_chain): Hanlde store-store chain.
18210 (prepare_finalizers_chain, prepare_finalizers): New function.
18211 (tree_predictive_commoning_loop): Return integer value indicating
18212 if loop is unrolled or lcssa form is corrupted.
18213 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
18215 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18217 * tree-predcom.c (initialize_root): Delete.
18218 (execute_pred_commoning_chain): Initialize root vars and replace
18219 reference of non-combined chain directly, rather than call above
18222 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18224 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
18225 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
18227 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18229 * tree-predcom.c (struct chain): New field init_seq.
18230 (release_chain): Release init_seq.
18231 (prepare_initializers_chain): Record intialization stmts in above
18233 (insert_init_seqs): New function.
18234 (tree_predictive_commoning_loop): Call insert_init_seqs.
18236 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18238 * tree-predcom.c (determine_roots_comp): Skip trivial components.
18240 2017-07-28 Richard Biener <rguenther@suse.de>
18242 * match.pd: Remove superfluous :c.
18243 * genmatch.c (simplify::id): Add member.
18244 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
18246 (current_id): New global.
18247 (dt_node::parent): Move from ...
18248 (dt_operand::parent): ... here. Add for_id member.
18249 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
18250 (decision_tree::find_node): Relax order requirement when
18251 merging DT_TRUE nodes to ones inbetween the current simplify
18252 and the one we try to merge with. Add diagnostic whenever
18253 we need to enforce pattern order by not merging.
18254 (decision_tree::insert): Set current_id.
18255 (decision_tree::print_node): Dump parent node and for_id.
18256 (parser::last_id): Add member.
18257 (parser::push_simplify): Assign unique id.
18258 (parser::parser): Initialize last_id.
18260 2017-07-28 Martin Liska <mliska@suse.cz>
18263 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
18264 gimple_build_debug_bind.
18266 2017-07-28 Richard Biener <rguenther@suse.de>
18268 PR tree-optimization/81502
18269 * match.pd: Add pattern combining BIT_INSERT_EXPR with
18271 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
18273 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
18274 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
18275 for BIT_FIELD_REF args.
18276 * fold-const.c (make_bit_field_ref): Likewise.
18277 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
18279 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18282 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
18283 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
18284 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
18285 Or it into SANITIZER_UNDEFINED.
18286 * ubsan.c: Include gimple-fold.h and varasm.h.
18287 (ubsan_expand_ptr_ifn): New function.
18288 (instrument_pointer_overflow): New function.
18289 (maybe_instrument_pointer_overflow): New function.
18290 (instrument_object_size): Formatting fix.
18291 (pass_ubsan::execute): Call instrument_pointer_overflow
18292 and maybe_instrument_pointer_overflow.
18293 * internal-fn.c (expand_UBSAN_PTR): New function.
18294 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
18295 * sanitizer.def (__ubsan_handle_pointer_overflow,
18296 __ubsan_handle_pointer_overflow_abort): New builtins.
18297 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
18298 * internal-fn.def (UBSAN_PTR): New internal function.
18299 * opts.c (sanitizer_opts): Add pointer-overflow.
18300 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
18301 * fold-const.c (build_range_check): Compute pointer range check in
18302 integral type if pointer arithmetics would be needed. Formatting
18305 2017-07-28 Martin Liska <mliska@suse.cz>
18308 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
18309 parameters that are of a variable-length.
18311 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
18313 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
18315 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
18316 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
18317 (CRT_CALL_STATIC_FUNCTION): Likewise.
18318 (ASM_DEFAULT_SPEC): New define.
18319 (ASM_SPEC32): Likewise.
18320 (ASM_SPEC64): Likewise.
18321 (ASM_SPEC_COMMON): Likewise.
18322 (ASM_SPEC): Likewise.
18323 (INVALID_64BIT): Likewise.
18324 (LINK_OS_DEFAULT_SPEC): Likewise.
18325 (LINK_OS_SPEC32): Likewise.
18326 (LINK_OS_SPEC64): Likewise.
18327 (POWERPC_LINUX): Likewise.
18328 (PTRDIFF_TYPE): Likewise.
18329 (RESTORE_FP_PREFIX): Likewise.
18330 (RESTORE_FP_SUFFIX): Likewise.
18331 (SAVE_FP_PREFIX): Likewise.
18332 (SAVE_FP_SUFFIX): Likewise.
18333 (SIZE_TYPE): Likewise.
18334 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
18335 (TARGET_64BIT): Likewise.
18336 (TARGET_64BIT): Likewise.
18337 (TARGET_AIX): Likewise.
18338 (WCHAR_TYPE_SIZE): Likewise.
18339 (WCHAR_TYPE): Undefine.
18340 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
18341 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
18342 (CPP_OS_RTEMS_SPEC): Delete.
18343 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
18344 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
18346 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
18348 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18350 PR tree-optimization/81578
18351 * tree-parloops.c (build_new_reduction): Bail out if
18352 reduction_code isn't one of the standard OpenMP reductions.
18353 Move the details printing after that decision.
18355 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18357 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
18358 related to reload_in_progress.
18359 (splat_input_operand): Likewise.
18360 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
18362 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
18364 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
18365 (TARGET_INSTANTIATE_DECLS): Likewise.
18366 (legitimate_indexed_address_p): Delete reload_in_progress code.
18367 (rs6000_debug_legitimate_address_p): Likewise.
18368 (rs6000_eliminate_indexed_memrefs): Likewise.
18369 (rs6000_emit_le_vsx_store): Likewise.
18370 (rs6000_emit_move_si_sf_subreg): Likewise.
18371 (rs6000_emit_move): Likewise.
18372 (register_to_reg_type): Likewise.
18373 (rs6000_pre_atomic_barrier): Likewise.
18374 (rs6000_machopic_legitimize_pic_address): Likewise.
18375 (rs6000_allocate_stack_temp): Likewise.
18376 (rs6000_address_for_fpconvert): Likewise.
18377 (rs6000_address_for_altivec): Likewise.
18378 (rs6000_secondary_memory_needed_rtx): Delete function.
18379 (rs6000_check_sdmode): Likewise.
18380 (rs6000_alloc_sdmode_stack_slot): Likewise.
18381 (rs6000_instantiate_decls): Likewise.
18382 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
18383 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
18384 Delete reload_in_progress.
18385 (*vec_reload_and_plus_<mptrsize>): Likewise.
18386 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
18387 (vsx_div_v2di): Likewise.
18388 (vsx_udiv_v2di): Likewise.
18390 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18392 * config/rs6000/rs6000.opt (mlra): Replace with stub.
18393 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
18394 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
18395 (rs6000_debug_reg_global): Delete print of LRA status.
18396 (rs6000_option_override_internal): Delete dead LRA related code.
18397 (rs6000_lra_p): Delete function.
18398 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
18400 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
18402 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
18403 * config/riscv/rtems.h: New file.
18405 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18406 Sudakshina Das <sudi.das@arm.com>
18408 * config/aarch64/aarch64.md
18409 (define_split for and<mode>3nr_compare): Move
18410 non aarch64_logical_operand to a register.
18411 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
18412 register immediate operand to a register.
18413 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
18415 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18417 PR middle-end/81564
18418 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
18420 2017-07-27 Richard Biener <rguenther@suse.de>
18422 PR tree-optimization/81573
18423 PR tree-optimization/81494
18424 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
18425 multi defuse cycle case.
18427 2017-07-27 Richard Biener <rguenther@suse.de>
18429 PR tree-optimization/81571
18430 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
18433 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
18435 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
18436 earlier and only if MASK_FPU is set. Adjust formatting.
18438 2017-07-27 Martin Liska <mliska@suse.cz>
18440 * opt-functions.awk: Add validation of value of Init.
18441 * optc-gen.awk: Pass new argument.
18443 2017-07-27 Martin Liska <mliska@suse.cz>
18445 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
18446 Fix wrong condition.
18448 2017-07-27 Martin Liska <mliska@suse.cz>
18450 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
18451 BBs and edges seen by autoFDO.
18453 2017-07-27 Richard Biener <rguenther@suse.de>
18455 PR tree-optimization/81502
18456 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
18457 with incompatible but same sized type.
18458 (execute_update_addresses_taken): Likewise.
18460 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
18462 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
18463 flag_tree_loop_vectorize rather than flag_tree_vectorize.
18465 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18468 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
18469 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
18470 Change s_operand to memory_operand.
18472 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
18474 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
18475 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
18476 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
18477 Emit instructions rather than returning an expression. Handle TFmode
18478 and KFmode by casting to TImode.
18479 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
18480 (rs6000_emit_le_vsx_store): Likewise.
18481 * config/rs6000/vsx.md (VSX_TI): New iterator.
18482 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
18483 (*vsx_le_undo_permute_<mode>): Likewise.
18484 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
18485 emit the split sequence.
18486 (*vsx_le_perm_store_<mode>): Likewise.
18488 2017-07-27 Jakub Jelinek <jakub@redhat.com>
18490 PR tree-optimization/81555
18491 PR tree-optimization/81556
18492 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
18493 if true, force CHANGED for the recursive invocation.
18494 (reassociate_bb): Remember original length of ops array, pass
18495 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
18497 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
18498 attributes for noipa attribute. For naked attribute use
18499 lookup_attribute first before lookup_attribute_spec.
18500 * final.c (rest_of_handle_final): Disable IPA RA for functions with
18502 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
18503 for functions with noipa attribute.
18504 (cgraph_externally_visible_p): Return true for functions with noipa
18506 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
18507 for functions with noipa attribute.
18508 * doc/extend.texi: Document noipa function attribute.
18509 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
18510 also for functions with noipa attribute.
18511 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
18513 2017-07-26 Andrew Pinski <apinski@cavium.com>
18515 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
18516 vec_unalign_load_cost and vec_unalign_store_cost.
18518 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
18520 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
18521 -mvsx-small-integer option.
18522 (ISA_3_0_MASKS_IEEE): Likewise.
18523 (OTHER_VSX_VECTOR_MASKS): Likewise.
18524 (POWERPC_MASKS): Likewise.
18525 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
18526 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
18527 code, only testing for DImode being allowed in non-VSX floating
18529 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
18530 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
18532 (rs6000_option_override_internal): Delete -mvsx-small-integer.
18533 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
18534 TARGET_P8_VECTOR test.
18535 (rs6000_secondary_reload_simple_move): Likewise.
18536 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
18537 since TARGET_P9_VECTOR was already tested.
18538 (rs6000_opt_masks): Remove -mvsx-small-integer.
18539 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
18540 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18542 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
18543 test for TARGET_VEXTRACTUB was used, and that uses
18545 (p9 extract splitter): Likewise.
18546 (vsx_extract_<mode>_di_p9): Likewise.
18547 (vsx_extract_<mode>_store_p9): Likewise.
18548 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
18549 for TARGET_P9_VECTOR was used. Delete code that is now dead with
18550 the elimination of TARGET_VSX_SMALL_INTEGER.
18551 (vsx_extract_<mode>_p8): Likewise.
18552 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
18553 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
18554 (vsx_set_<mode>_p9): Likewise.
18555 (vsx_set_v4sf_p9): Likewise.
18556 (vsx_set_v4sf_p9_zero): Likewise.
18557 (vsx_insert_extract_v4sf_p9): Likewise.
18558 (vsx_insert_extract_v4sf_p9_2): Likewise.
18559 * config/rs6000/rs6000.md (sign extend splitter): Change
18560 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
18561 (floatsi<mode>2_lfiwax_mem): Likewise.
18562 (floatunssi<mode>2_lfiwzx_mem): Likewise.
18563 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
18564 since a test for TARGET_P9_VECTOR was used.
18565 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18566 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
18567 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18568 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18569 TARGET_P8_VECTOR test.
18570 (fix_trunc<mode>si2_stfiwx): Likewise.
18571 (fix_trunc<mode>si2_internal): Likewise.
18572 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
18573 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18575 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18576 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18577 TARGET_P8_VECTOR test.
18578 (fixuns_trunc<mode>si2_stfiwx): Likewise.
18579 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
18580 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18582 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18583 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
18584 since a test for TARGET_P9_VECTOR was used.
18585 (splitter for loading small constants): Likewise.
18587 2017-07-26 Andrew Pinski <apinski@cavium.com>
18589 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
18592 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
18595 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
18596 (fp_valid_at): Likewise.
18598 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18600 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
18601 (qdf24xx_addrcost_table): Likewise.
18602 (cortexa57_tunings): Update to use generic_branch_cost.
18603 (cortexa72_tunings): Likewise.
18604 (cortexa73_tunings): Likewise.
18605 (qdf24xx_tunings): Likewise.
18607 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18609 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
18610 (thunderx2t99_branch_cost): Likewise.
18611 (cortexa35_tunings): Update to use generic_branch_cost.
18612 (cortexa53_tunings): Likewise.
18613 (cortexa57_tunings): Likewise.
18614 (cortexa72_tunings): Likewise.
18615 (cortexa73_tunings): Likewise.
18616 (thunderx2t99_tunings): Likewise.
18618 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
18620 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
18621 (sparc_option_override): Honour MASK_FSMULD.
18622 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
18623 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
18624 * config/sparc/sparc.opt (mfsmuld): New option.
18625 * doc/invoke.texi (mfsmuld): Document option.
18627 2017-07-26 Marek Polacek <polacek@redhat.com>
18629 PR middle-end/70992
18630 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
18632 2017-07-26 Richard Biener <rguenther@suse.de>
18634 * gimple-match-head.c (do_valueize): Return OP if valueize
18636 (get_def): New helper to get at the def stmt of a SSA name
18637 if valueize allows.
18638 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
18639 do_valueize to get at the def stmt.
18640 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
18642 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
18644 PR middle-end/46932
18645 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
18647 2017-07-26 Martin Liska <mliska@suse.cz>
18650 * function.c (expand_function_start): Make expansion of
18651 nonlocal_goto_save_area after parm_birth_insn.
18653 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
18655 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
18656 from all CPU target flags enable members.
18658 2017-07-26 Richard Biener <rguenther@suse.de>
18660 * genmatch.c (dt_simplify::gen): Make iterator vars const.
18661 (decision_tree::gen): Make 'type' const.
18662 (write_predicate): Likewise.
18664 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
18666 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
18667 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
18668 (rs6000_option_override_internal): Likewise.
18669 (rs6000_expand_vector_set): Likewise.
18670 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
18671 (TARGET_UPPER_REGS_SF): Likewise.
18672 (TARGET_UPPER_REGS_DI): Likewise.
18673 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
18674 (TARGET_DIRECT_MOVE_64BIT): Likewise.
18675 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
18676 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18677 (Splitters for DI constants in Altivec registers): Likewise.
18678 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
18679 (vsx_set_v4sf_p9): Likewise.
18680 (vsx_set_v4sf_p9_zero): Likewise.
18681 (vsx_insert_extract_v4sf_p9): Likewise.
18682 (vsx_insert_extract_v4sf_p9_2): Likewise.
18684 2017-07-25 Carl Love <cel@us.ibm.com>
18686 * doc/extend.texi: Update the built-in documentation file for the
18687 existing built-in functions
18688 vector signed char vec_cnttz (vector signed char);
18689 vector unsigned char vec_cnttz (vector unsigned char);
18690 vector signed short vec_cnttz (vector signed short);
18691 vector unsigned short vec_cnttz (vector unsigned short);
18692 vector signed int vec_cnttz (vector signed int);
18693 vector unsigned int vec_cnttz (vector unsigned int);
18694 vector signed long long vec_cnttz (vector signed long long);
18695 vector unsigned long long vec_cnttz (vector unsigned long long);
18697 2017-07-25 Andrew Pinski <apinski@cavium.com>
18699 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
18700 accesses where the use is for the first operand of a BIT_INSERT.
18702 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18705 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18706 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18708 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18710 * config/i386/gstabs.h: Delete.
18711 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
18713 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
18715 * config/i386/i386.c (ix86_decompose_address): Do not check for
18716 register RTX when looking at index_reg or base_reg.
18717 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
18719 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
18721 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
18722 to update EH info here.
18724 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18726 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
18728 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18730 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
18732 2017-07-25 Torsten Duwe <duwe@suse.de>
18734 * common.opt: Introduce -fpatchable-function-entry
18735 command line option, and its variables function_entry_patch_area_size
18736 and function_entry_patch_area_start.
18737 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
18738 including a two-value parser.
18739 * target.def (print_patchable_function_entry): New target hook.
18740 * targhooks.h (default_print_patchable_function_entry): New function.
18741 * targhooks.c (default_print_patchable_function_entry): Likewise.
18742 * toplev.c (process_options): Switch off IPA-RA if
18743 patchable function entries are being generated.
18744 * varasm.c (assemble_start_function): Look at the
18745 patchable-function-entry command line switch and current
18746 function attributes and maybe generate NOP instructions by
18747 calling the print_patchable_function_entry hook.
18748 * doc/extend.texi: Document patchable_function_entry attribute.
18749 * doc/invoke.texi: Document -fpatchable_function_entry
18750 command line option.
18751 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
18753 * doc/tm.texi: Re-generate.
18755 2017-07-25 Jakub Jelinek <jakub@redhat.com>
18758 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
18759 TARGET_AVX512DQ rather than TARGET_AVX512BW.
18761 2017-07-25 Tamar Christina <tamar.christina@arm.com>
18763 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
18765 2017-07-25 Richard Biener <rguenther@suse.de>
18767 PR tree-optimization/81455
18768 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
18769 not walk in cycles when looking for guards.
18771 2017-07-25 Richard Biener <rguenther@suse.de>
18773 PR tree-optimization/81529
18774 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
18775 when optimizing backedge uses.
18777 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
18779 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
18781 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
18782 to dl_section_ref. On AIX, append an expression to subtract
18783 the size of the section length to dl_section_ref.
18785 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
18787 * configure.ac: If any of the config.* scripts fail, exit 1.
18788 * configure: Regenerate.
18790 2017-07-25 Richard Biener <rguenther@suse.de>
18792 PR middle-end/81546
18793 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
18794 of immediate uses, be more verbose on errors.
18796 2017-07-25 Richard Biener <rguenther@suse.de>
18798 PR tree-optimization/81510
18799 * tree-vect-loop.c (vect_is_simple_reduction): When the
18800 reduction stmt is not inside the loop bail out.
18802 2017-07-25 Richard Biener <rguenther@suse.de>
18804 PR tree-optimization/81303
18805 * tree-vect-loop-manip.c (vect_loop_versioning): Build
18806 profitability check against LOOP_VINFO_NITERSM1.
18808 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18810 * domwalk.c (cmp_bb_postorder): Simplify.
18811 (sort_bbs_postorder): New function. Use it...
18812 (dom_walker::walk): ...here to optimize common cases.
18814 2017-07-25 Martin Liska <mliska@suse.cz>
18817 * ipa-visibility.c (function_and_variable_visibility): Make the
18818 redirection just on target that supports aliasing.
18819 Fix GNU coding style.
18821 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
18824 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
18826 * config/arm/rtems.h: Likewise.
18827 * config/bfin/rtems.h: Likewise.
18828 * config/i386/rtemself.h: Likewise.
18829 * config/lm32/rtems.h: Likewise.
18830 * config/m32c/rtems.h: Likewise.
18831 * config/m68k/rtemself.h: Likewise.
18832 * config/microblaze/rtems.h: Likewise.
18833 * config/mips/rtems.h: Likewise.
18834 * config/moxie/rtems.h: Likewise.
18835 * config/nios2/rtems.h: Likewise.
18836 * config/powerpcspe/rtems.h: Likewise.
18837 * config/rs6000/rtems.h: Likewise.
18838 * config/rtems.h: Likewise.
18839 * config/sh/rtems.h: Likewise.
18840 * config/sh/rtemself.h: Likewise.
18841 * config/sparc/rtemself.h: Likewise.
18843 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
18846 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
18847 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
18848 * tree-ssa-structalias.c (alias_get_name): Same.
18850 2017-07-25 Bin Cheng <bin.cheng@arm.com>
18853 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
18854 instructions if no du chain is found.
18856 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
18858 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
18860 2017-07-25 Richard Biener <rguenther@suse.de>
18862 PR middle-end/81505
18863 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
18866 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
18868 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
18869 upper-regs options.
18870 (ISA_2_7_MASKS_SERVER): Likewise.
18871 (ISA_3_0_MASKS_IEEE): Likewise.
18872 (OTHER_P8_VECTOR_MASKS): Likewise.
18873 (OTHER_VSX_VECTOR_MASKS): Likewise.
18874 (POWERPC_MASKS): Likewise.
18875 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
18876 duplicate list of options.
18877 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
18878 explicit -mupper-regs options.
18879 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
18880 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
18881 alias for -mupper-regs-df.
18882 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18883 (rs6000_init_hard_regno_mode_ok): Likewise.
18884 (rs6000_option_override_internal): Likewise.
18885 (rs6000_opt_masks): Likewise.
18886 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
18887 options in terms of whether -mvsx or -mpower8-vector was used.
18888 (TARGET_UPPER_REGS_DI): Likewise.
18889 (TARGET_UPPER_REGS_SF): Likewise.
18890 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
18891 -mupper-regs-* options.
18893 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
18895 * passes.c (emergency_dump_function): Print some empty lines and a
18896 header before the RTL dump.
18898 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
18900 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
18902 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
18905 * config/aarch64/aarch64.c (aarch64_classify_symbol):
18906 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
18908 2017-07-24 Carl Love <cel@us.ibm.com>
18910 * config/rs6000/rs6000-c.c: Add support for built-in functions
18911 vector float vec_extract_fp32_from_shorth (vector unsigned short);
18912 vector float vec_extract_fp32_from_shortl (vector unsigned short);
18913 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
18914 vec_extract_fp_from_shortl): Add defines for the two builtins.
18915 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
18916 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
18918 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
18919 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
18920 * doc/extend.texi: Update the built-in documentation file for the
18921 new built-in function.
18923 2017-07-24 Jakub Jelinek <jakub@redhat.com>
18926 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
18928 * doc/generic.texi: Likewise.
18929 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18930 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18932 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
18934 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
18935 (aarch64_mls_elt_merge<mode>): Likewise.
18937 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
18939 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18940 having __cxa_atexit.
18942 2017-07-23 Michael Collison <michael.collison@arm.com>
18944 * config/arm/arm.c (arm_option_override): Deprecate
18945 use of -mstructure-size-boundary.
18946 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
18947 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
18949 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18952 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18953 Reduce cost estimate for direct moves.
18955 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
18958 * config/i386/i386.c (ix86_option_override_internal): Disable
18959 BMI, BMI2 and TBM instructions for -m16.
18961 2017-07-21 Carl Love <cel@us.ibm.com>
18963 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18964 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18965 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18966 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18967 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18968 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18969 VMULOSW): New enum "unspec" values.
18970 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18971 altivec_vmulosw): New patterns.
18972 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18973 VMULOSW): Add definitions.
18975 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
18977 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
18978 (qdf24xx): Likewise.
18979 * config/aarch64/aarch64-options-extensions.def (rdma); New.
18980 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
18981 (AARCH64_FL_V8_1): Renumber.
18982 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
18983 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
18984 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
18985 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
18986 rdma to feature modifiers list.
18988 2017-07-21 Yury Gribov <tetra2005@gmail.com>
18990 PR middle-end/56727
18991 * ipa-visibility (function_and_variable_visibility): Convert
18992 recursive PLT call to direct call if appropriate.
18994 2017-07-21 Andrew Pinski <apinski@cavium.com>
18996 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
18997 operand 1 to see if the types precision matches.
18998 * fold-const.c (operand_equal_p): Likewise.
19000 2017-07-21 Richard Biener <rguenther@suse.de>
19002 PR tree-optimization/81303
19003 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
19004 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
19005 (vect_peeling_hash_get_lowest_cost): Adjust.
19006 (vect_enhance_data_refs_alignment): Likewise. Use
19007 vect_get_peeling_costs_all_drs to compute the penalty for no
19008 peeling to match up costs.
19010 2017-07-21 Richard Biener <rguenther@suse.de>
19012 PR tree-optimization/81500
19013 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
19014 we didn't identify a reduction path.
19016 2017-07-21 Tom de Vries <tom@codesourcery.com>
19017 Cesar Philippidis <cesar@codesourcery.com>
19019 PR gcov-profile/81442
19020 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
19023 2017-07-21 Tom de Vries <tom@codesourcery.com>
19026 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
19028 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
19029 nvptx_override_options_after_change.
19031 2017-07-21 Ulrich Drepper <drepper@redhat.com>
19033 * dwarf2out.c (output_file_names): Avoid double testing for
19034 dwarf_version >= 5.
19036 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
19038 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
19040 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19042 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
19044 (try_crossjump_to_edge): Do not punt on partitioned functions.
19046 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19048 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
19049 Put all BBs reachable only via paths crossing cold region to cold
19051 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
19053 2016-07-21 Richard Biener <rguenther@suse.de>
19055 PR tree-optimization/81303
19056 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
19057 into account prologue and epilogue iterations when raising
19058 min_profitable_iters to sth at least covering one vector iteration.
19060 2017-07-21 Tamar Christina <tamar.christina@arm.com>
19062 * config/arm/arm.c (arm_test_cpu_arch_dat):
19065 2017-07-20 Nathan Sidwell <nathan@acm.org>
19067 Remove TYPE_METHODS.
19068 * tree.h (TYPE_METHODS): Delete.
19069 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
19070 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
19071 (dbxout_type_methods): Scan TYPE_FIELDS.
19072 (dbxout_type): Don't check TYPE_METHODS here.
19073 * function.c (use_register_for_decl): Always ignore register for
19074 class types when not optimizing.
19075 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
19076 * tree.c (free_lang_data_in_type): Stitch out member functions and
19077 templates from TYPE_FIELDS.
19078 (build_distinct_type_copy, verify_type_variant,
19079 verify_type): Member fns are on TYPE_FIELDS.
19080 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
19081 * tree-pretty-print.c (dump_generic_node): Likewise.
19083 2017-07-20 Jakub Jelinek <jakub@redhat.com>
19086 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
19087 V2TImode and V4TImode.
19088 (ix86_expand_vector_extract): Likewise.
19089 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
19090 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
19091 (ssescalarmode): Handle V4TImode and V2TImode.
19092 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
19093 (*vec_extractv2ti, *vec_extractv4ti): New insns.
19094 (VEXTRACTI128_MODE): New mode iterator.
19095 (splitter for *vec_extractv?ti first element): New.
19096 (VEC_INIT_MODE): New mode iterator.
19097 (vec_init<mode>): Consolidate 3 expanders into one using
19098 VEC_INIT_MODE mode iterator.
19100 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
19102 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
19103 non_spilled_static_chain_regno_p.
19105 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
19107 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
19109 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
19111 * bb-reorder.c (connect_traces): Allow copying of blocks within
19114 2017-07-20 Richard Biener <rguenther@suse.de>
19116 * gimple.h (gimple_phi_result): Add gphi * overload.
19117 (gimple_phi_result_ptr): Likewise.
19118 (gimple_phi_arg): Likewise. Adjust index assert to only
19119 allow actual argument accesses rather than all slots available
19121 (gimple_phi_arg_def): Add gphi * overload.
19122 * tree-phinodes.c (make_phi_node): Initialize only actual
19124 (resize_phi_node): Clear memory not covered by old node,
19125 do not initialize excess argument slots.
19126 (reserve_phi_args_for_new_edge): Initialize new argument slot
19129 2017-07-20 Bin Cheng <bin.cheng@arm.com>
19131 PR tree-optimization/81388
19133 2016-07-21 Bin Cheng <bin.cheng@arm.com>
19135 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
19136 by removing computation of may_be_zero.
19138 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19139 Tom de Vries <tom@codesourcery.com>
19141 PR middle-end/81030
19142 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
19143 when gimple level profile disagrees with what RTL expander did.
19145 2017-07-20 Richard Biener <rguenther@suse.de>
19147 PR tree-optimization/61171
19148 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
19149 (vect_analyze_stmt): Add slp instance parameter.
19150 (vectorizable_reduction): Likewise.
19151 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
19152 (vect_is_simple_reduction): Deal with chains not detected
19153 as SLP reduction chain, specifically not properly associated
19154 chains containing a mix of plus/minus.
19155 (get_reduction_op): Remove.
19156 (get_initial_defs_for_reduction): Simplify, pass in whether
19157 this is a reduction chain, pass in the SLP node for the PHIs.
19158 (vect_create_epilog_for_reduction): Get the SLP instance as
19160 (vectorizable_reduction): Get the SLP instance as arg.
19161 During analysis remember the SLP node with the PHIs in the
19162 instance. Simplify getting at the vectorized reduction PHIs.
19163 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
19164 through SLP instance.
19165 (vect_slp_analyze_operations): Likewise.
19166 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
19167 (vect_transform_stmt): Likewise.
19169 2017-07-20 Tom de Vries <tom@codesourcery.com>
19171 PR tree-optimization/81489
19172 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
19173 read of phi arg location to before loop that modifies phi.
19175 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19177 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
19180 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19182 PR middle-end/81331
19183 * except.c (execute): Fix ordering issue.
19185 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
19187 PR rtl-optimization/81423
19188 * combine.c (make_compound_operation_int): Don't try to optimize
19189 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
19191 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
19193 PR rtl-optimization/81423
19194 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
19195 with a constant that is -1 in the truncated to mode.
19197 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19199 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
19200 (determine_unlikely_bbs): ... here.
19201 * predict.h (propagate_unlikely_bbs_forward): Declare.
19202 * cfgexpand.c (pass_expand::execute): Use it.
19203 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
19205 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
19206 propagate_unlikely_bbs_forward.
19208 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19210 PR middle-end/81331
19211 * except.c (maybe_add_nop_after_section_switch): New function.
19214 2017-07-19 Tom de Vries <tom@codesourcery.com>
19216 * gimple.h (gimple_phi_set_arg): Make assert more strict.
19218 2017-07-19 Tom de Vries <tom@codesourcery.com>
19220 * gimple.h (gimple_phi_arg): Make assert more strict.
19222 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
19224 * config.gcc (powerpc*-*-*): Add mmintrin.h.
19225 * config/rs6000/mmintrin.h: New file.
19226 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
19228 2017-07-19 Jakub Jelinek <jakub@redhat.com>
19230 PR tree-optimization/81346
19231 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
19233 2017-07-19 Tom de Vries <tom@codesourcery.com>
19235 * config/nvptx/nvptx.md (VECIM): Add V2DI.
19237 2017-07-19 Tom de Vries <tom@codesourcery.com>
19239 * config/nvptx/nvptx-modes.def: Add V2DImode.
19240 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
19241 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
19242 (nvptx_output_mov_insn): Handle lack of mov.b128.
19243 (nvptx_print_operand): Handle 'H' and 'L' codes.
19244 (nvptx_vector_mode_supported): Allow V2DImode.
19245 (nvptx_preferred_simd_mode): New function.
19246 (nvptx_data_alignment): New function.
19247 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
19248 nvptx_preferred_simd_mode.
19249 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
19251 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
19253 2017-07-19 Tom de Vries <tom@codesourcery.com>
19255 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
19256 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
19257 (nvptx_vector_mode_supported): New function. Allow V2SImode.
19258 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
19259 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
19260 (mov<VECIM>_insn): New define_insn.
19261 (define_expand "mov<VECIM>): New define_expand.
19263 2017-07-19 Tom de Vries <tom@codesourcery.com>
19265 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
19267 2017-07-19 Jakub Jelinek <jakub@redhat.com>
19269 PR tree-optimization/81346
19270 * fold-const.h (fold_div_compare, range_check_type): Declare.
19271 * fold-const.c (range_check_type): New function.
19272 (build_range_check): Use range_check_type.
19273 (fold_div_compare): No longer static, rewritten into
19274 a match.pd helper function.
19275 (fold_comparison): Don't call fold_div_compare here.
19276 * match.pd (X / C1 op C2): New optimization using fold_div_compare
19277 as helper function.
19279 2017-07-19 Nathan Sidwell <nathan@acm.org>
19281 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
19282 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
19283 * tree.c (find_decls_types_r, verify_type): Use
19284 TYPE_{MIN,MAX}_VALUE_RAW.
19285 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
19286 (hash_tree): Likewise.
19287 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19289 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19292 2017-07-18 Tom de Vries <tom@codesourcery.com>
19294 PR middle-end/81464
19295 * omp-expand.c (expand_omp_for_static_chunk): Handle
19296 equal-argument loop exit phi.
19298 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
19301 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
19302 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
19303 operand 2 predicate.
19304 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
19305 operand 2 predicate.
19306 (ror,rol -> rorx splitters): Use const_int_operand as
19307 operand 2 predicate.
19309 2017-06-18 Richard Biener <rguenther@suse.de>
19311 PR tree-optimization/81410
19312 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
19313 the gap in the ! slp_perm SLP case after each group.
19315 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19317 PR middle-end/81463
19318 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
19321 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19323 PR middle-end/81462
19324 * predict.c (set_even_probabilities): Cleanup; do not affect
19325 probabilities that are already known.
19326 (combine_predictions_for_bb): Call even when count is set.
19328 2017-07-18 Nathan Sidwell <nathan@acm.org>
19330 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
19333 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19336 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
19337 optimization for loop niter analysis.
19339 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
19342 * config/avr/avr.c (avr_optimize_casesi): Don't use
19343 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
19345 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
19347 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
19348 body_cost_vec from _vect_peel_extended_info.
19349 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
19350 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
19353 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19355 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
19357 2017-07-18 Richard Biener <rguenther@suse.de>
19359 PR tree-optimization/80620
19360 PR tree-optimization/81403
19361 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
19362 info when re-using a VN table entry.
19364 2017-07-18 Richard Biener <rguenther@suse.de>
19366 PR tree-optimization/81418
19367 * tree-vect-loop.c (vectorizable_reduction): Properly compute
19368 vectype_in. Verify that with lane-reducing reduction operations
19369 we have a single def-use cycle.
19371 2017-07-17 Carl Love <cel@us.ibm.com>
19373 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
19375 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19376 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19377 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19378 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19379 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19380 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19381 VMULOSW): New enum "unspec" values.
19382 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19383 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19384 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19385 altivec_vmulosw): New patterns.
19386 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19387 VMULOSW): Add definitions.
19388 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
19390 * config/alpha/alpha.c: Include predict.h.
19392 2017-07-17 Yury Gribov <tetra2005@gmail.com>
19394 * tree-vrp.c (compare_assert_loc): Fix comparison function
19395 to return predictable results.
19397 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19399 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
19401 (subdi3): Likewise.
19402 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
19403 * doc/invoke.texi (mexpand-adddi): Update text.
19405 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19407 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
19408 that also clobbers the CC register. The old expand code is moved
19410 (*arc_clzsi2): ... here.
19411 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
19412 the CC register. The old expand code is moved to ...
19413 (arc_ctzsi2): ... here.
19415 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19417 * config/arc/arc.opt (mindexed-loads): Use initial value
19418 TARGET_INDEXED_LOADS_DEFAULT.
19419 (mauto-modify-reg): Use initial value
19420 TARGET_AUTO_MODIFY_REG_DEFAULT.
19421 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19422 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19423 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19424 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19426 2017-07-17 Martin Liska <mliska@suse.cz>
19429 * opts.c (finish_options): Do not allow -fgnu-tm
19430 w/ -fsanitize={kernel-,}address. Say sorry.
19432 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19435 * tree-loop-distribution.c (classify_partition): Only assert on
19436 numer of iterations.
19437 (merge_dep_scc_partitions): Delete prameter. Update function call.
19438 (distribute_loop): Remove code handling loop with unknown niters.
19439 (pass_loop_distribution::execute): Skip loop with unknown niters.
19441 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19444 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
19445 function sort_partitions_by_post_order.
19447 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19449 PR tree-optimization/81374
19450 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
19451 the max index of basic blocks, rather than number of basic blocks.
19453 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19455 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
19457 (arc_legitimate_pic_operand_p): Likewise.
19458 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
19460 (arc_needs_pcl_p): Likewise.
19461 (arc_legitimate_pc_offset_p): Likewise.
19462 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
19463 function is also used in constrains.md.
19464 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
19465 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
19466 PLUS. Only return true/false in known cases, otherwise assert.
19467 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
19468 is already called in arc_legitimate_constant_p.
19469 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
19471 (LEGITIMATE_PIC_OPERAND_P): Use
19472 arc_raw_symbolic_reference_mentioned_p function.
19473 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
19478 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19479 Andrew Burgess <andrew.burgess@embecosm.com>
19481 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
19482 (arc_return_address_register): New function.
19483 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
19484 (arc_handle_fndecl_attribute): Add naked attribute.
19485 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19486 (TARGET_WARN_FUNC_RETURN): Likewise.
19487 (arc_allocate_stack_slots_for_args): New function.
19488 (arc_warn_func_return): Likewise.
19489 (machine_function): Change type fn_type.
19490 (arc_compute_function_type): Consider new naked function type,
19491 change function return type.
19492 (arc_must_save_register): Adapt to handle new
19493 arc_compute_function_type's return type.
19494 (arc_expand_prologue): Likewise.
19495 (arc_expand_epilogue): Likewise.
19496 (arc_return_address_regs): Delete.
19497 (arc_return_address_register): New function.
19498 (arc_epilogue_uses): Use above function.
19499 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
19500 (arc_function_type): Change encoding, add naked type.
19501 (ARC_INTERRUPT_P): Change to handle the new encoding.
19502 (ARC_FAST_INTERRUPT_P): Likewise.
19503 (ARC_NORMAL_P): Define.
19504 (ARC_NAKED_P): Likewise.
19505 (arc_compute_function_type): Delete prototype.
19506 * config/arc/arc.md (in_ret_delay_slot): Use
19507 arc_return_address_register function.
19508 (simple_return): Likewise.
19509 (p_return_i): Likewise.
19511 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19513 PR tree-optimization/81428
19514 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
19515 can't be built for those types.
19517 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19519 Remove stuff dead since r239246.
19521 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
19522 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
19523 (avr_inform_devices): Remove dead stuff.
19525 2017-07-17 Tamar Christina <tamar.christina@arm.com>
19527 * config/arm/arm_neon.h: Fix softp typo.
19529 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19531 PR tree-optimization/81365
19532 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
19533 aggregate moves onto bb predecessor edges, make sure there are no
19534 loads that could alias the lhs in between the start of bb and the
19537 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19540 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
19541 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
19542 [LSHIFTRT, outer_code = TRUNCATE]: Same.
19544 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19546 PR tree-optimization/81396
19547 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
19548 (init_symbolic_number): Initialize it to 1.
19549 (perform_symbolic_merge): Add n_ops from both operands into the new
19551 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
19552 without base_addr as useless if they need more than one operation.
19553 (bswap_replace): Handle !bswap case for NULL base_addr.
19555 2017-07-17 Tom de Vries <tom@codesourcery.com>
19558 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
19561 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
19563 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
19564 conditional builtin define __FIX_LEON3FT_B2BST.
19566 2017-07-17 Daniel Cederman <cederman@gaisler.com>
19568 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
19569 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
19572 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
19574 PR rtl-optimization/81424
19575 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
19576 to remove potential trapping from operands if -fnon-call-exceptions.
19578 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19580 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
19581 profile_proability for scalling.
19582 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
19584 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19586 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
19588 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19590 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
19591 fixpoint arithmetics.
19593 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19595 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
19596 fixpoint arithmetics.
19598 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19600 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
19601 fixpoint arithmetics.
19603 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19605 * profile-count.h (profile_probability::from_reg_br_prob_note,
19606 profile_probability::to_reg_br_prob_note): New functions.
19607 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
19608 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
19609 * predict.c (probability_reliable_p): Update.
19610 (edge_probability_reliable_p): Update.
19611 (br_prob_note_reliable_p): Update.
19612 (invert_br_probabilities): Update.
19613 (add_reg_br_prob_note): New function.
19614 (combine_predictions_for_insn): Update.
19615 * asan.c (asan_clear_shadow): Update.
19616 * cfgbuild.c (compute_outgoing_frequencies): Update.
19617 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
19618 (update_br_prob_note): Update.
19619 (rtl_verify_edges): Update.
19620 (purge_dead_edges): Update.
19621 (fixup_reorder_chain): Update.
19622 * emit-rtl.c (try_split): Update.
19623 * ifcvt.c (cond_exec_process_insns): Update.
19624 (cond_exec_process_if_block): Update.
19625 (dead_or_predicable): Update.
19626 * internal-fn.c (expand_addsub_overflow): Update.
19627 (expand_neg_overflow): Update.
19628 (expand_mul_overflow): Update.
19629 * loop-doloop.c (doloop_modify): Update.
19630 * loop-unroll.c (compare_and_jump_seq): Update.
19631 * optabs.c (emit_cmp_and_jump_insn_1): Update.
19632 * predict.h: Update.
19633 * reorg.c (mostly_true_jump): Update.
19635 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
19636 * config/alpha/alpha.c (emit_unlikely_jump): Update.
19637 * config/arc/arc.c: (emit_unlikely_jump): Update.
19638 * config/arm/arm.c: (emit_unlikely_jump): Update.
19639 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
19640 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
19641 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
19642 (ix86_print_operand): Update.
19643 (ix86_split_fp_branch): Update.
19644 (predict_jump): Update.
19645 * config/ia64/ia64.c (ia64_print_operand): Update.
19646 * config/mmix/mmix.c (mmix_print_operand): Update.
19647 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
19648 (rs6000_expand_split_stack_prologue): Update.
19649 * config/rs6000/rs6000.c: Update.
19650 * config/s390/s390.c (s390_expand_vec_strlen): Update.
19651 (s390_expand_vec_movstr): Update.
19652 (s390_expand_cs_tdsi): Update.
19653 (s390_expand_split_stack_prologue): Update.
19654 * config/sh/sh.c (sh_print_operand): Update.
19655 (expand_cbranchsi4): Update.
19656 (expand_cbranchdi4): Update.
19657 * config/sparc/sparc.c (output_v9branch): Update.
19658 * config/spu/spu.c (get_branch_target): Update.
19659 (ea_load_store_inline): Update.
19660 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
19661 * config/tilepro/tilepro.c: Update.
19663 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
19665 * gimplify.c (mostly_copy_tree_r): Revert latest change.
19666 (gimplify_save_expr): Likewise.
19668 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19670 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
19672 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19674 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
19676 * timevar.def (TV_IPA_FNSUMMARY): Define.
19678 2017-07-16 Daniel Cederman <cederman@gaisler.com>
19680 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
19681 to back store errata sensitive sequence from being generated.
19682 (sqrtdf2_fix): Likewise.
19684 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19686 * tree-ssa-threadupdate.c (compute_path_counts,
19687 update_joiner_offpath_counts): Use profile_probability.
19689 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
19692 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19694 * config/arm/arm-c.c (arm_cpu_builtins): Define
19695 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19697 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
19699 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19700 array entries to represent __ieee128 versions of the
19701 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
19702 scalar_extract_sig, and scalar_insert_exp built-in functions.
19703 (altivec_resolve_overloaded_builtin): Add special case handling
19704 for the __builtin_scalar_insert_exp function, as represented by
19705 the P9V_BUILTIN_VEC_VSIEDP constant.
19706 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
19707 exponent support for __ieee128 argument.
19708 (VSESQP): Add scalar extract signature support for __ieee128
19710 (VSTDCNQP): Add scalar test negative support for __ieee128
19712 (VSIEQP): Add scalar insert exponent support for __int128 argument
19713 with __ieee128 result.
19714 (VSIEQPF): Add scalar insert exponent support for __ieee128
19715 argument with __ieee128 result.
19716 (VSTDCQP): Add scalar test data class support for __ieee128
19718 (VSTDCNQP): Add overload support for scalar test negative with
19719 __ieee128 argument.
19720 (VSTDCQP): Add overload support for scalar test data class
19721 __ieee128 argument.
19722 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
19723 UNSPEC_VSX_SXSIGDP.
19724 (UNSPEC_VSX_SIEXPQP): New constant.
19725 (xsxexpqp): New insn for VSX scalar extract exponent quad
19727 (xsxsigqp): New insn for VSX scalar extract significand quad
19729 (xsiexpqpf): New insn for VSX scalar insert exponent quad
19730 precision with floating point argument.
19731 (xststdcqp): New expand for VSX scalar test data class quad
19733 (xststdcnegqp): New expand for VSX scalar test negative quad
19735 (xststdcqp): New insn to match expansions for VSX scalar test data
19736 class quad precision and VSX scalar test negative quad precision.
19737 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
19738 special case operand checking to enforce that second operand of
19739 VSX scalar test data class with quad precision argument is a 7-bit
19741 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19742 prototypes and descriptions of __ieee128 versions of
19743 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
19744 scalar_test_data_class, and scalar_test_neg built-in functions.
19746 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19748 PR tree-optimization/81162
19749 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
19750 replace a negate with an add.
19752 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
19754 * doc/invoke.texi (arm/-mcpu): Document +crypto.
19756 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19758 * config/arm/arm-c.c (arm_cpu_builtins): Define
19759 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19761 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19763 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
19764 (armv8-r): Set ARM Cortex-R52 as default CPU.
19765 * config/arm/arm-tables.opt: Regenerate.
19766 * config/arm/arm-tune.md: Regenerate.
19767 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19769 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
19770 extension for -mcpu=cortex-r52.
19772 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19774 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
19775 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
19776 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
19777 (fp-armv8): Define it as FP_ARMv8 only.
19778 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
19779 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
19781 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
19782 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
19783 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
19784 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
19785 than TARGET_FPU_ARMV8.
19786 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
19787 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
19788 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
19790 * config/arm/neon.md (neon_vrint): Likewise.
19791 (neon_vcvt): Likewise.
19792 (neon_<fmaxmin_op><mode>): Likewise.
19793 (<fmaxmin><mode>3): Likewise.
19794 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
19795 * config/arm/predicates.md (arm_cond_move_operator): Check against
19796 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
19798 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
19800 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
19801 to top of function.
19803 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19805 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
19806 loop in comment with memset.
19808 2017-07-14 Martin Liska <mliska@suse.cz>
19810 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
19811 * dwarf2out.c (is_java): Remove the function.
19812 (output_pubname): Remove usage of the function.
19813 (lower_bound_default): Remove usage of DW_LANG_Java.
19814 (gen_compile_unit_die): Likewise.
19815 * gcc.c: Remove compiler defaults for .java and .zip files.
19816 * gimple-expr.c (remove_suffix): Change as there's no longer
19817 extension than 4-letter one.
19818 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
19819 (gimplify_save_expr): Likewise.
19820 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
19821 as it's possible even for other languages than Java.
19822 * langhooks.h (struct lang_hooks): Remove Java from a comment.
19823 * lto-opts.c (lto_write_options): Remove reference to Java.
19824 * opts.c (strip_off_ending): Update file extension handling.
19825 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
19826 * tree-eh.c (lower_resx): Likewise.
19827 * tree.c (free_lang_data_in_type): Remove dead code.
19828 (find_decls_types_r): Likewise.
19829 (build_common_builtin_nodes): Remove Java from a comment.
19830 (verify_type): Remove dead code.
19831 * varasm.c (assemble_external): Remove Java from a comment.
19833 2017-07-14 Martin Liska <mliska@suse.cz>
19835 * opts.c (finish_options): Add quotes.
19836 (common_handle_option): Likewise.
19838 2017-07-14 Martin Liska <mliska@suse.cz>
19840 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
19841 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
19842 Remove N_SO_PASCAL.
19843 * dwarf2out.c (lower_bound_default): Do not handle
19845 (gen_compile_unit_die): Likewise.
19846 * gcc.c: Remove default extension binding for GNU Pascal.
19847 * stmt.c: Remove Pascal language from a comment.
19848 * xcoffout.c: Likewise.
19850 2017-07-13 David Malcolm <dmalcolm@redhat.com>
19853 * diagnostic-show-locus.c (fixit_cmp): New function.
19854 (layout::layout): Sort m_fixit_hints.
19855 (column_range::column_range): Assert that the values are valid.
19856 (struct char_span): New struct.
19857 (correction::overwrite): New method.
19858 (struct source_line): New struct.
19859 (line_corrections::add_hint): Add assertions. Reimplement memcpy
19860 calls in terms of classes source_line and char_span, and
19861 correction::overwrite.
19862 (selftest::test_overlapped_fixit_printing_2): New function.
19863 (selftest::diagnostic_show_locus_c_tests): Call it.
19865 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
19867 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
19868 early if there is no lhs.
19870 2017-07-13 Martin Liska <mliska@suse.cz>
19872 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
19873 (gen_reference_type_die): Likewise.
19874 * stor-layout.c: Remove Pascal-related comment.
19876 2017-07-13 Martin Liska <mliska@suse.cz>
19878 * opts.c (finish_options): Add quotes to error messages.
19879 (parse_sanitizer_options): Likewise.
19881 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
19883 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
19885 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
19887 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
19889 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
19891 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
19893 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
19895 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
19898 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
19899 provides the hardware capability bits, define the macro
19900 __BUILTIN_CPU_SUPPORTS__.
19901 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
19902 if GLIBC does not provide the hardware capability bits. Add a
19903 gcc_unreachable call if the built-in cpu function is neither
19904 __builtin_cpu_is nor __builtin_cpu_supports.
19905 (rs6000_get_function_versions_dispatcher): Change the warning
19906 that an old GLIBC is used which does not export the capability
19907 bits to be an error.
19908 * doc/extend.texi (target_clones attribute): Document the
19909 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
19910 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
19911 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
19912 the macros defined by GCC if the newer GLIBC is available.
19914 2017-07-12 Jeff Law <law@redhat.com>
19916 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
19917 remaining includes slightly.
19918 * config/riscv/riscv-builtins.c: Include profile-count.h.
19920 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
19923 * config/avr/avr.c (avr_set_current_function): In diagnostic
19924 messages: Quote keywords and (parts of) identifiers.
19925 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
19928 2017-07-12 Carl Love <cel@us.ibm.com>
19930 * config/rs6000/rs6000-c.c: Add support for built-in functions
19931 vector bool char vec_revb (vector bool char);
19932 vector bool short vec_revb (vector short char);
19933 vector bool int vec_revb (vector bool int);
19934 vector bool long long vec_revb (vector bool long long);
19935 * doc/extend.texi: Update the built-in documentation file for the
19936 new built-in functions.
19938 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19940 * config/s390/s390.md: Remove movcc splitter.
19942 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19944 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
19945 load/store on condition.
19947 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
19950 * config/avr/avr.c (avr_encode_section_info)
19951 [progmem && !TREE_READONLY]: Error if progmem object needs
19954 2017-07-11 Michael Collison <michael.collison@arm.com>
19956 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
19959 2017-07-11 Carl Love <cel@us.ibm.com>
19961 * config/rs6000/rs6000-c.c: Add support for builtins
19962 vector unsigned int vec_parity_lsbb (vector signed int);
19963 vector unsigned int vec_parity_lsbb (vector unsigned int);
19964 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
19965 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
19966 vector unsigned long long vec_parity_lsbb (vector signed long long);
19967 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
19968 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
19969 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
19970 * doc/extend.texi: Update the built-in documentation file for the
19971 new built-in functions.
19973 2017-07-11 David Malcolm <dmalcolm@redhat.com>
19975 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
19976 (layout::m_primary_loc): New field.
19977 (layout::layout): Initialize new field. Move location filtering
19978 logic from here to...
19979 (layout::maybe_add_location_range): ...this new method. Add
19980 support for filtering to just the lines already specified by other
19982 (layout::will_show_line_p): New method.
19983 (gcc_rich_location::add_location_if_nearby): New method.
19984 (selftest::test_add_location_if_nearby): New test function.
19985 (selftest::diagnostic_show_locus_c_tests): Call it.
19986 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
19989 2017-07-11 Tom de Vries <tom@codesourcery.com>
19991 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
19992 (bb_first_real_insn): New function.
19993 (nvptx_single): Add extra initialization of broadcasted condition
19996 2017-07-11 Nathan Sidwell <nathan@acm.org>
19998 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
20000 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
20002 * doc/extend.texi (AVR Function Attributes): Remove weblink to
20003 Binutils doc as TEXI will mess them up.
20004 * doc/invoke.texi (AVR Options): Same here.
20006 2017-07-11 Daniel Cederman <cederman@gaisler.com>
20008 * config/sparc/sparc.opt (mfix-ut700): New option.
20009 (mfix-gr712rc): Likewise.
20010 (sparc_fix_b2bst): New variable.
20011 * doc/invoke.texi (SPARC options): Document them.
20012 (ARM options): Fix warnings.
20013 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
20014 instructions to prevent sequences that can trigger the store-store
20015 errata for certain LEON3FT processors.
20016 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
20017 (sparc_option_override): Set sparc_fix_b2bst appropriately.
20018 * config/sparc/sparc.md (fix_b2bst): New attribute.
20019 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
20021 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
20024 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
20026 (*rsqrtsf2_sse): Ditto.
20028 (div<mode>3): Macroize insn from divdf3 and divsf3
20029 using MODEF mode iterator.
20031 2017-07-10 Martin Sebor <msebor@redhat.com>
20033 PR tree-optimization/80397
20034 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
20035 instead of testing for equality to INTEGER_TYPE.
20037 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
20039 * config.gcc: Remove uclibc from arc target spec.
20041 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
20043 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
20045 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20048 * lto-wrapper.c (remove_option): New function.
20049 (merge_and_complain): Merge PIC/PIE options more realistically.
20051 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
20053 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
20057 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
20058 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
20059 * config.in: Regenerate.
20060 * configure: Regenerate.
20061 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
20062 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
20063 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
20064 (TARGET_GASISR_PROLOGUES): ...target mask.
20065 * common/config/avr/avr-common.c
20066 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
20067 Set -mgas-isr-prologues.
20068 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
20069 INSERT_PASS_BEFORE for it.
20070 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
20071 * config/avr/avr.c (avr_option_override)
20072 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
20073 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
20074 (avr_attribute_table) <no_gccisr>: Add new function attribute.
20075 (avr_set_current_function) <is_no_gccisr>: Init machine field.
20076 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
20078 (make_avr_pass_pre_proep): New function.
20079 (emit_push_sfr) <treg>: Add argument to function and use it
20080 instead of TMP_REG.
20081 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
20082 and set machine->gasisr.yes.
20083 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
20084 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
20085 __gcc_isr.n_pushed to .L__stack_usage.
20086 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
20087 (avr_asm_final_postscan_insn): ...this new static function.
20088 * config/avr/avr.h (machine_function)
20089 <is_no_gccisr, use_L__stack_usage>: New fields.
20090 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
20091 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
20092 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
20093 (gasisr, *gasisr): New expander and insn.
20094 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20095 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
20096 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
20098 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
20100 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
20103 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
20105 Move jump-tables out of .text again.
20108 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
20109 (ASM_OUTPUT_ADDR_VEC): New function.
20110 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
20111 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
20112 INSN_ADDRESSes as asm comment.
20113 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
20114 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
20115 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
20116 * config/avr/avr.md (*tablejump): Adjust comment.
20117 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
20118 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
20120 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
20121 (avr_output_addr_vec): New proto.
20122 (avr_log_t) <insn_addresses>: New field.
20124 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
20127 * config/i386/i386.c (ix86_function_arg_advance): Set
20128 outgoing_args_on_stack to true if there are outgoing arguments
20130 (ix86_function_arg): Likewise.
20131 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
20132 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
20133 * config/i386/i386.h (machine_function): Add
20134 outgoing_args_on_stack.
20136 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
20138 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20139 supporting pthreds.
20140 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
20142 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
20144 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
20145 (REAL_H): Remove $(MACHMODE_H).
20146 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
20148 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
20149 $(MACHMODE_H) and double-int.h.
20150 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
20152 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
20155 2017-07-07 Andrew Pinski <apinski@cavium.com>
20157 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
20158 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
20160 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20162 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
20163 Add warning if GCC was not configured to link against a GLIBC that
20164 exports the hardware capability bits.
20165 (make_resolver_func): Make resolver function private and not a
20166 COMDAT function. Create the name with clone_function_name instead
20167 of make_unique_name.
20170 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
20171 correct operand in doing the split.
20173 2017-07-07 Carl Love <cel@us.ibm.com>
20175 * config/rs6000/rs6000-c: Add support for built-in function
20176 vector unsigned short vec_pack_to_short_fp32 (vector float,
20178 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
20179 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
20180 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
20181 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
20182 (convert_4f32_8i16): Add define_expand.
20183 * doc/extend.texi: Update the built-in documentation file for the
20184 new built-in function.
20186 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20188 * config/sparc/m8.md: New file.
20189 * config/sparc/sparc.md: Include m8.md.
20191 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20193 * config/sparc/sparc.opt: New option -mvis4b.
20194 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
20195 (sparc_option_override): Handle VIS4B.
20196 (enum sparc_builtins): Define
20197 SPARC_BUILTIN_DICTUNPACK{8,16,32},
20198 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
20199 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
20200 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
20201 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
20202 (check_constant_argument): New function.
20203 (sparc_vis_init_builtins): Define builtins
20204 __builtin_vis_dictunpack{8,16,32},
20205 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
20206 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
20207 __builtin_vis_fpcmpde{8,16,32}shl and
20208 __builtin_vis_fpcmpur{8,16,32}shl.
20209 (sparc_expand_builtin): Check that the constant operands to
20210 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
20211 constant and in range.
20212 * config/sparc/sparc-c.c (sparc_target_macros): Handle
20214 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
20215 (SPARC_IMM5_P): Likewise.
20216 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
20217 (enabled): Handle vis4b.
20218 (UNSPEC_DICTUNPACK): New unspec.
20219 (UNSPEC_FPCMPSHL): Likewise.
20220 (UNSPEC_FPUCMPSHL): Likewise.
20221 (UNSPEC_FPCMPDESHL): Likewise.
20222 (UNSPEC_FPCMPURSHL): Likewise.
20223 (cpu_feature): New CPU feature `vis4b'.
20224 (dictunpack{8,16,32}): New insns.
20225 (FPCSMODE): New mode iterator.
20226 (fpcscond): New code iterator.
20227 (fpcsucond): Likewise.
20228 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
20229 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
20230 (fpcmpde{8,16,32}{si,di}shl): Likewise.
20231 (fpcmpur{8,16,32}{si,di}shl): Likewise.
20232 * config/sparc/constraints.md: Define constraints `q' for unsigned
20233 2-bit integer constants and `t' for unsigned 5-bit integer
20235 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
20237 (imm5_operand_dictunpack16): Likewise.
20238 (imm5_operand_dictunpack32): Likewise.
20239 (imm2_operand): Likewise.
20240 * doc/invoke.texi (SPARC Options): Document -mvis4b.
20241 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
20242 ditunpack* and fpcmp*shl builtins.
20244 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20246 * config.gcc: Handle m8 in --with-{cpu,tune} options.
20247 * config.in: Add HAVE_AS_SPARC6 define.
20248 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
20250 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
20252 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
20253 (CPP_CPU_SPEC): Handle m8.
20254 (ASM_CPU_SPEC): Likewise.
20255 * config/sparc/sparc-opts.h (enum processor_type): Add
20257 * config/sparc/sparc.c (m8_costs): New struct.
20258 (sparc_option_override): Handle TARGET_CPU_m8.
20259 (sparc32_initialize_trampoline): Likewise.
20260 (sparc64_initialize_trampoline): Likewise.
20261 (sparc_issue_rate): Likewise.
20262 (sparc_register_move_cost): Likewise.
20263 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
20264 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
20265 (ASM_CPU64_DEFAULT_SPEC): Likewise.
20266 (CPP_CPU_SPEC): Handle M8.
20267 (ASM_CPU_SPEC): Likewise.
20268 (AS_M8_FLAG): Define.
20269 * config/sparc/sparc.md: Add m8 to the cpu attribute.
20270 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
20271 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
20273 * configure: Regenerate.
20274 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
20277 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20279 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
20281 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
20282 ("*movdi_insn_sp32"): Do not set v3pipe.
20283 ("*movsi_insn"): Likewise.
20284 ("*movdi_insn_sp64"): Likewise.
20285 ("*movsf_insn"): Likewise.
20286 ("*movdf_insn_sp32"): Likewise.
20287 ("*movdf_insn_sp64"): Likewise.
20288 ("*zero_extendsidi2_insn_sp64"): Likewise.
20289 ("*sign_extendsidi2_insn"): Likewise.
20290 ("*mov<VM32:mode>_insn"): Likewise.
20291 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20292 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20293 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
20294 ("<vlop:code><VL:mode>3"): Likewise.
20295 ("*not_<vlop:code><VL:mode>3"): Likewise.
20296 ("*nand<VL:mode>_vis"): Likewise.
20297 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
20298 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
20299 ("one_cmpl<VL:mode>2"): Likewise.
20300 ("faligndata<VM64:mode>_vis"): Likewise.
20301 ("alignaddrsi_vis"): Likewise.
20302 ("alignaddrdi_vis"): Likweise.
20303 ("alignaddrlsi_vis"): Likewise.
20304 ("alignaddrldi_vis"): Likewise.
20305 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20306 ("bmaskdi_vis"): Likewise.
20307 ("bmasksi_vis"): Likewise.
20308 ("bshuffle<VM64:mode>_vis"): Likewise.
20309 ("cmask8<P:mode>_vis"): Likewise.
20310 ("cmask16<P:mode>_vis"): Likewise.
20311 ("cmask32<P:mode>_vis"): Likewise.
20312 ("pdistn<P:mode>_vis"): Likewise.
20313 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20315 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20317 * config/sparc/sparc.md ("subtype"): New insn attribute.
20318 ("*wrgsr_sp64"): Set insn subtype.
20319 ("*rdgsr_sp64"): Likewise.
20320 ("alignaddrsi_vis"): Likewise.
20321 ("alignaddrdi_vis"): Likewise.
20322 ("alignaddrlsi_vis"): Likewise.
20323 ("alignaddrldi_vis"): Likewise.
20324 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
20325 ("fexpand_vis"): Likewise.
20326 ("fpmerge_vis"): Likewise.
20327 ("faligndata<VM64:mode>_vis"): Likewise.
20328 ("bshuffle<VM64:mode>_vis"): Likewise.
20329 ("cmask8<P:mode>_vis"): Likewise.
20330 ("cmask16<P:mode>_vis"): Likewise.
20331 ("cmask32<P:mode>_vis"): Likewise.
20332 ("fchksm16_vis"): Likewise.
20333 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
20334 ("fmean16_vis"): Likewise.
20335 ("fp<plusminus_insn>64_vis"): Likewise.
20336 ("<plusminus_insn>v8qi3"): Likewise.
20337 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20338 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
20339 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
20340 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
20341 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
20342 ("*movqi_insn"): Likewise.
20343 ("*movhi_insn"): Likewise.
20344 ("*movsi_insn"): Likewise.
20345 ("movsi_pic_gotdata_op"): Likewise.
20346 ("*movdi_insn_sp32"): Likewise.
20347 ("*movdi_insn_sp64"): Likewise.
20348 ("movdi_pic_gotdata_op"): Likewise.
20349 ("*movsf_insn"): Likewise.
20350 ("*movdf_insn_sp32"): Likewise.
20351 ("*movdf_insn_sp64"): Likewise.
20352 ("*zero_extendhisi2_insn"): Likewise.
20353 ("*zero_extendqihi2_insn"): Likewise.
20354 ("*zero_extendqisi2_insn"): Likewise.
20355 ("*zero_extendqidi2_insn"): Likewise.
20356 ("*zero_extendhidi2_insn"): Likewise.
20357 ("*zero_extendsidi2_insn_sp64"): Likewise.
20358 ("ldfsr"): Likewise.
20359 ("prefetch_64"): Likewise.
20360 ("prefetch_32"): Likewise.
20361 ("tie_ld32"): Likewise.
20362 ("tie_ld64"): Likewise.
20363 ("*tldo_ldub_sp32"): Likewise.
20364 ("*tldo_ldub1_sp32"): Likewise.
20365 ("*tldo_ldub2_sp32"): Likewise.
20366 ("*tldo_ldub_sp64"): Likewise.
20367 ("*tldo_ldub1_sp64"): Likewise.
20368 ("*tldo_ldub2_sp64"): Likewise.
20369 ("*tldo_ldub3_sp64"): Likewise.
20370 ("*tldo_lduh_sp32"): Likewise.
20371 ("*tldo_lduh1_sp32"): Likewise.
20372 ("*tldo_lduh_sp64"): Likewise.
20373 ("*tldo_lduh1_sp64"): Likewise.
20374 ("*tldo_lduh2_sp64"): Likewise.
20375 ("*tldo_lduw_sp32"): Likewise.
20376 ("*tldo_lduw_sp64"): Likewise.
20377 ("*tldo_lduw1_sp64"): Likewise.
20378 ("*tldo_ldx_sp64"): Likewise.
20379 ("*mov<VM32:mode>_insn"): Likewise.
20380 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20381 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20383 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20385 * config/sparc/sparc.md ("type"): New insn type viscmp.
20386 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
20388 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
20389 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
20390 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20391 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
20393 ("n7_vis_logical_11cycle"): Likewise.
20394 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
20395 * config/sparc/niagara2.md ("niag3_vis": Likewise.
20396 * config/sparc/niagara.md ("niag_vis"): Likewise.
20397 * config/sparc/ultra3.md ("us3_fga"): Likewise.
20398 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
20400 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20402 * config/sparc/sparc.md: New instruction type `bmask'.
20403 (bmaskdi_vis): Use the `bmask' type.
20404 (bmasksi_vis): Likewise.
20405 * config/sparc/ultra3.md (us3_array): Likewise.
20406 * config/sparc/niagara7.md (n7_array): Likewise.
20407 * config/sparc/niagara4.md (n4_array): Likewise.
20408 * config/sparc/niagara2.md (niag2_vis): Likewise.
20409 (niag3_vis): Likewise.
20410 * config/sparc/niagara.md (niag_vis): Likewise.
20412 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20414 * ipa-comdats.c: Remove optimize check from gate.
20415 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
20416 for functions not optimized.
20417 (ipa_fn_summary_read): Skip optimize check.
20418 (ipa_fn_summary_write): Likewise.
20419 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
20421 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
20423 (can_inline_edge_p): Check flag_pcc_struct_return for match.
20424 (check_callers): Give up on caller which is not optimized.
20425 (inline_small_functions): Likewise.
20426 (ipa_inline): Do not give up when not optimizing.
20427 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
20428 away unoptimizes cdtors.
20429 (whole_program_function_and_variable_visibility): Do
20430 ipa_discover_readonly_nonaddressable_vars in LTO mode.
20431 * ipa.c (process_references): Do not check optimize.
20432 (symbol_table::remove_unreachable_nodes): Update optimize check.
20433 (set_writeonly_bit): Update optimize check.
20434 (pass_ipa_cdtor_merge::gate): Do not check optimize.
20435 (pass_ipa_single_use::gate): Remove.
20437 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20439 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
20440 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
20441 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
20442 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
20443 permute_load, permute_store, adjust_extract, adjust_splat,
20444 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
20445 replace_swap_with_copy, dump_swap_insn_table,
20446 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
20447 recombine_lvx_pattern, recombine_stvx_pattern,
20448 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
20449 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
20450 to file rs6000-p8swap.c.
20451 * config/rs6000/rs6000-p8swap.c: New file.
20452 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
20453 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
20454 and rs6000*-*-* targets.
20456 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20458 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
20460 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20462 * lto-wrapper.c (merge_and_complain): Do not merge
20463 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
20464 fsigned_zeros, ftrapping_math, fwrapv.
20465 (append_compiler_options): Do not track these options.
20466 (append_linker_options): Likewie
20468 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20470 * cgraphunit.c (cgraph_node::finalize_function): When
20471 !flag_toplevel_reorde set no_reorder flag.
20472 (varpool_node::finalize_decl): Likewise.
20473 (symbol_table::compile): Drop no toplevel reorder path.
20475 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20477 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
20478 edges; zero probability is not better than uninitialized.
20480 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20482 * asan.h (asan_sanitize_allocas_p): Declare.
20483 * asan.c (asan_sanitize_allocas_p): New function.
20484 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
20485 (handle_builtin_alloca): Likewise.
20486 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
20487 if !asan_sanitize_allocas_p.
20488 * params.def (asan-instrument-allocas): Add new option.
20489 * params.h (ASAN_PROTECT_ALLOCAS): Define.
20490 * opts.c (common_handle_option): Disable allocas sanitization for
20493 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20495 * asan.c: Include gimple-fold.h.
20496 (get_last_alloca_addr): New function.
20497 (handle_builtin_stackrestore): Likewise.
20498 (handle_builtin_alloca): Likewise.
20499 (asan_emit_allocas_unpoison): Likewise.
20500 (get_mem_refs_of_builtin_call): Add new parameter, remove const
20501 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
20502 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
20503 (instrument_builtin_call): Pass gimple iterator to
20504 get_mem_refs_of_builtin_call.
20505 (last_alloca_addr): New global.
20506 * asan.h (asan_emit_allocas_unpoison): Declare.
20507 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
20508 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
20509 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
20510 if function calls alloca.
20511 * gimple-fold.c (replace_call_with_value): Remove static keyword.
20512 * gimple-fold.h (replace_call_with_value): Declare.
20513 * internal-fn.c: Include asan.h.
20514 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
20515 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
20517 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20519 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
20520 (C_SELFTEST_FLAGS): New.
20521 (CPP_SELFTEST_FLAGS): New.
20522 (SELFTEST_DEPS): New, from deps of s-selftest.
20523 (C_SELFTEST_DEPS): New, from deps of s-selftest.
20524 (CPP_SELFTEST_DEPS): New.
20525 (selftest): Add dependency on s-selftest-c++.
20526 (s-selftest): Rename to...
20527 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
20528 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
20529 than SELFTEST_FLAGS.
20530 (selftest-gdb): Rename to...
20531 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
20533 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
20534 (selftest-valgrind): Rename to...
20535 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
20537 (selftest-valgrind): Reintroduce as an alias for
20538 selftest-c-valgrind.
20539 (s-selftest-c++): New.
20540 (selftest-c++-gdb): New.
20541 (selftest-c++-valgrind): New.
20543 2017-07-06 Olivier Hainque <hainque@adacore.com>
20545 * gcc.c (process_command): When deciding if undefined variables
20546 should be ignored when processing specs, accept "gcc -v" as well.
20548 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20550 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
20551 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
20553 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20555 * config/arm/arm-cpus.in (armv8-r): Add new entry.
20556 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
20557 * config/arm/arm-tables.opt: Regenerate.
20558 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
20560 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
20562 2017-07-06 Carl Love <cel@us.ibm.com>
20564 * ChangeLog: Clean up from mid air collision
20566 2017-07-06 Carl Love <cel@us.ibm.com>
20568 * config/rs6000/rs6000-c.c: Add support for built-in functions
20569 vector signed int vec_subc (vector signed int, vector signed int);
20570 vector signed __int128 vec_subc (vector signed __int128,
20571 vector signed __int128);
20572 vector unsigned __int128 vec_subc (vector unsigned __int128,
20573 vector unsigned __int128);
20574 vector signed int vec_sube (vector signed int, vector signed int,
20575 vector signed int);
20576 vector unsigned int vec_sube (vector unsigned int,
20577 vector unsigned int,
20578 vector unsigned int);
20579 vector signed __int128 vec_sube (vector signed __int128,
20580 vector signed __int128,
20581 vector signed__int128);
20582 vector unsigned __int128 vec_sube (vector unsigned __int128,
20583 vector unsigned __int128,
20584 vector unsigned __int128);
20585 vector signed int vec_subec (vector signed int, vector signed int,
20586 vector signed int);
20587 vector unsigned int vec_subec (vector unsigned int,
20588 vector unsigned int,
20589 vector unsigned int);
20590 vector signed __int128 vec_subec (vector signed __int128,
20591 vector signed __int128,
20592 vector signed__int128);
20593 vector unsigned __int128 vec_subec (vector unsigned __int128,
20594 vector unsigned __int128,
20595 vector unsigned __int128);
20596 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
20597 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
20598 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
20599 BU_ALTIVEC_OVERLOAD_X definitions.
20600 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
20601 * doc/extend.texi: Update the built-in documentation file for the new
20602 built-in functions.
20604 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20607 * diagnostic-show-locus.c (layout::layout): Use start and finish
20608 spelling location for the start and finish of each range.
20609 * genmatch.c (linemap_client_expand_location_to_spelling_point):
20610 Add unused aspect param.
20611 * input.c (expand_location_1): Add "aspect" param, and use it
20612 to access the correct part of the location.
20613 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
20615 (expand_location_to_spelling_point): Likewise.
20616 (linemap_client_expand_location_to_spelling_point): Add "aspect"
20617 param, and pass it to expand_location_1.
20619 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
20621 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
20622 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
20623 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
20624 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
20625 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
20626 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
20627 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
20628 _mm_maskz_getmant_ss): New intrinsics.
20629 (__builtin_ia32_getexpss128_mask): Changed to ...
20630 __builtin_ia32_getexpss128_round ... this.
20631 (__builtin_ia32_getexpsd128_mask): Changed to ...
20632 __builtin_ia32_getexpsd128_round ... this.
20633 * config/i386/i386-builtin-types.def
20634 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
20635 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
20636 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
20637 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
20638 __builtin_ia32_getmantss_mask_round): New builtins.
20639 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
20640 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
20641 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
20642 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
20643 * config/i386/sse.md
20644 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
20645 avx512f_sgetexp<mode><mask_scalar_name>
20646 <round_saeonly_scalar_name> ... this.
20647 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
20648 %0, %1, %2<round_saeonly_op3>}): Changed to ...
20649 vgetexp<ssescalarmodesuffix>
20650 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20651 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
20652 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
20653 avx512f_vgetmant<mode><mask_scalar_name>
20654 <round_saeonly_scalar_name> ... this.
20655 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
20656 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
20657 vgetmant<ssescalarmodesuffix>
20658 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
20659 %0<mask_scalar_operand4>, %1, %2
20660 <round_saeonly_scalar_mask_op4>, %3} ... this.
20661 * config/i386/subst.md (mask_scalar_operand4,
20662 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
20663 round_saeonly_scalar_nimm_predicate): New subst attributes.
20665 2017-07-06 Julia Koval <julia.koval@intel.com>
20667 * config/i386/i386.c (ix86_erase_embedded_rounding):
20668 Remove code for old rounding pattern.
20670 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
20672 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
20674 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
20676 * doc/sourcebuild.texi (Test Directives, Variants of
20677 dg-require-support): Add documentation for dg-require-stack-check.
20679 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
20681 * config/i386/subst.md (mask_scalar, round_scalar,
20682 round_saeonly_scalar): New meta-templates.
20683 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
20684 round_scalar_mask_operand3, round_scalar_mask_op3,
20685 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
20686 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
20687 round_saeonly_scalar_constraint,
20688 round_saeonly_scalar_prefix): New subst attribute.
20689 * config/i386/sse.md
20690 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
20691 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
20692 <round_scalar_name> ... this.
20693 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
20694 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
20695 <round_scalar_name> ... this.
20696 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
20697 <sse>_vm<code><mode>3<mask_scalar_name>
20698 <round_saeonly_scalar_name> ... this.
20699 (v<plusminus_mnemonic><ssescalarmodesuffix>
20700 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20701 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20702 v<plusminus_mnemonic><ssescalarmodesuffix>
20703 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20704 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20705 (v<multdiv_mnemonic><ssescalarmodesuffix>
20706 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20707 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20708 v<multdiv_mnemonic><ssescalarmodesuffix>
20709 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20710 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20711 (v<maxmin_float><ssescalarmodesuffix>
20712 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
20713 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
20714 v<maxmin_float><ssescalarmodesuffix>
20715 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20716 %0<mask_scalar_operand3>, %1, %<iptr>2
20717 <round_saeonly_scalar_mask_op3>} ... this.
20719 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
20721 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
20722 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
20724 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
20725 Alan Hayward <alan.hayward@arm.com>
20726 David Sherwood <david.sherwood@arm.com>
20728 * combine.c (simplify_if_then_else): Remove "enum" before
20730 * compare-elim.c (can_eliminate_compare): Likewise.
20731 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
20733 (aarch64_lookup_simd_builtin_type): Likewise.
20734 (aarch64_simd_builtin_type): Likewise.
20735 (aarch64_init_simd_builtin_types): Likewise.
20736 (aarch64_simd_expand_args): Likewise.
20737 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
20739 (aarch64_reverse_mask): Likewise.
20740 (aarch64_simd_emit_reg_reg_move): Likewise.
20741 (aarch64_gen_adjusted_ldpstp): Likewise.
20742 (aarch64_ccmp_mode_to_code): Likewise.
20743 (aarch64_operands_ok_for_ldpstp): Likewise.
20744 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20745 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
20747 (aarch64_min_divisions_for_recip_mul): Likewise.
20748 (aarch64_reassociation_width): Likewise.
20749 (aarch64_get_condition_code_1): Likewise.
20750 (aarch64_simd_emit_reg_reg_move): Likewise.
20751 (aarch64_simd_attr_length_rglist): Likewise.
20752 (aarch64_reverse_mask): Likewise.
20753 (aarch64_operands_ok_for_ldpstp): Likewise.
20754 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20755 (aarch64_gen_adjusted_ldpstp): Likewise.
20756 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
20758 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
20759 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
20760 (arm_lookup_simd_builtin_type): Likewise.
20761 (arm_simd_builtin_type): Likewise.
20762 (arm_init_simd_builtin_types): Likewise.
20763 (arm_expand_builtin_args): Likewise.
20764 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
20765 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
20766 (ft32_setup_incoming_varargs): Likewise.
20767 (ft32_function_arg): Likewise.
20768 (ft32_function_arg_advance): Likewise.
20769 (ft32_pass_by_reference): Likewise.
20770 (ft32_arg_partial_bytes): Likewise.
20771 (ft32_valid_pointer_mode): Likewise.
20772 (ft32_addr_space_pointer_mode): Likewise.
20773 (ft32_addr_space_legitimate_address_p): Likewise.
20774 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
20776 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
20777 (ix86_emit_outlined_ms2sysv_restore): Likewise.
20778 (iamcu_alignment): Likewise.
20779 (canonicalize_vector_int_perm): Likewise.
20780 (ix86_noce_conversion_profitable_p): Likewise.
20781 (ix86_mpx_bound_mode): Likewise.
20782 (ix86_operands_ok_for_move_multiple): Likewise.
20783 * config/microblaze/microblaze-protos.h
20784 (microblaze_expand_conditional_branch_reg): Likewise.
20785 * config/microblaze/microblaze.c
20786 (microblaze_expand_conditional_branch_reg): Likewise.
20787 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20789 (rs6000_reassociation_width): Likewise.
20790 (rs6000_invalid_binary_op): Likewise.
20791 (fusion_p9_p): Likewise.
20792 (emit_fusion_p9_load): Likewise.
20793 (emit_fusion_p9_store): Likewise.
20794 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
20796 (riscv_hard_regno_mode_ok_p): Likewise.
20797 (riscv_address_insns): Likewise.
20798 (riscv_split_symbol): Likewise.
20799 (riscv_legitimize_move): Likewise.
20800 (riscv_function_value): Likewise.
20801 (riscv_hard_regno_nregs): Likewise.
20802 (riscv_expand_builtin): Likewise.
20803 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
20804 (riscv_build_integer): Likewise.
20805 (riscv_split_integer): Likewise.
20806 (riscv_legitimate_constant_p): Likewise.
20807 (riscv_cannot_force_const_mem): Likewise.
20808 (riscv_regno_mode_ok_for_base_p): Likewise.
20809 (riscv_valid_base_register_p): Likewise.
20810 (riscv_valid_offset_p): Likewise.
20811 (riscv_valid_lo_sum_p): Likewise.
20812 (riscv_classify_address): Likewise.
20813 (riscv_legitimate_address_p): Likewise.
20814 (riscv_address_insns): Likewise.
20815 (riscv_load_store_insns): Likewise.
20816 (riscv_force_binary): Likewise.
20817 (riscv_split_symbol): Likewise.
20818 (riscv_force_address): Likewise.
20819 (riscv_legitimize_address): Likewise.
20820 (riscv_move_integer): Likewise.
20821 (riscv_legitimize_const_move): Likewise.
20822 (riscv_legitimize_move): Likewise.
20823 (riscv_address_cost): Likewise.
20824 (riscv_subword): Likewise.
20825 (riscv_output_move): Likewise.
20826 (riscv_canonicalize_int_order_test): Likewise.
20827 (riscv_emit_int_order_test): Likewise.
20828 (riscv_function_arg_boundary): Likewise.
20829 (riscv_pass_mode_in_fpr_p): Likewise.
20830 (riscv_pass_fpr_single): Likewise.
20831 (riscv_pass_fpr_pair): Likewise.
20832 (riscv_get_arg_info): Likewise.
20833 (riscv_function_arg): Likewise.
20834 (riscv_function_arg_advance): Likewise.
20835 (riscv_arg_partial_bytes): Likewise.
20836 (riscv_function_value): Likewise.
20837 (riscv_pass_by_reference): Likewise.
20838 (riscv_setup_incoming_varargs): Likewise.
20839 (riscv_print_operand): Likewise.
20840 (riscv_elf_select_rtx_section): Likewise.
20841 (riscv_save_restore_reg): Likewise.
20842 (riscv_for_each_saved_reg): Likewise.
20843 (riscv_register_move_cost): Likewise.
20844 (riscv_hard_regno_mode_ok_p): Likewise.
20845 (riscv_hard_regno_nregs): Likewise.
20846 (riscv_class_max_nregs): Likewise.
20847 (riscv_memory_move_cost): Likewise.
20848 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
20849 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
20850 (rl78_addr_space_address_mode): Likewise.
20851 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20853 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20854 (rs6000_reassociation_width): Likewise.
20855 (rs6000_invalid_binary_op): Likewise.
20856 (fusion_p9_p): Likewise.
20857 (emit_fusion_p9_load): Likewise.
20858 (emit_fusion_p9_store): Likewise.
20859 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
20860 (ok_for_simple_move_operands): Likewise.
20861 (ok_for_simple_move_strict_operands): Likewise.
20862 (ok_for_simple_arith_logic_operands): Likewise.
20863 (visium_legitimize_reload_address): Likewise.
20864 (visium_select_cc_mode): Likewise.
20865 (output_cbranch): Likewise.
20866 (visium_split_double_move): Likewise.
20867 (visium_expand_copysign): Likewise.
20868 (visium_expand_int_cstore): Likewise.
20869 (visium_expand_fp_cstore): Likewise.
20870 * config/visium/visium.c (visium_pass_by_reference): Likewise.
20871 (visium_function_arg): Likewise.
20872 (visium_function_arg_advance): Likewise.
20873 (visium_libcall_value): Likewise.
20874 (visium_setup_incoming_varargs): Likewise.
20875 (visium_legitimate_constant_p): Likewise.
20876 (visium_legitimate_address_p): Likewise.
20877 (visium_legitimize_address): Likewise.
20878 (visium_secondary_reload): Likewise.
20879 (visium_register_move_cost): Likewise.
20880 (visium_memory_move_cost): Likewise.
20881 (prepare_move_operands): Likewise.
20882 (ok_for_simple_move_operands): Likewise.
20883 (ok_for_simple_move_strict_operands): Likewise.
20884 (ok_for_simple_arith_logic_operands): Likewise.
20885 (visium_function_value_1): Likewise.
20886 (rtx_ok_for_offset_p): Likewise.
20887 (visium_legitimize_reload_address): Likewise.
20888 (visium_split_double_move): Likewise.
20889 (visium_expand_copysign): Likewise.
20890 (visium_expand_int_cstore): Likewise.
20891 (visium_expand_fp_cstore): Likewise.
20892 (visium_split_cstore): Likewise.
20893 (visium_select_cc_mode): Likewise.
20894 (visium_split_cbranch): Likewise.
20895 (output_cbranch): Likewise.
20896 (visium_print_operand_address): Likewise.
20897 * expmed.c (flip_storage_order): Likewise.
20898 * expmed.h (emit_cstore): Likewise.
20899 (flip_storage_order): Likewise.
20900 * genrecog.c (validate_pattern): Likewise.
20901 * hsa-gen.c (gen_hsa_addr): Likewise.
20902 * internal-fn.c (expand_arith_overflow): Likewise.
20903 * ira-color.c (allocno_copy_cost_saving): Likewise.
20904 * lra-assigns.c (find_hard_regno_for_1): Likewise.
20905 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
20906 (process_invariant_for_inheritance): Likewise.
20907 * lra-eliminations.c (move_plus_up): Likewise.
20908 * omp-low.c (lower_oacc_reductions): Likewise.
20909 * simplify-rtx.c (simplify_subreg): Likewise.
20910 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
20911 (TARGET_CHKP_BOUND_MODE): Likewise..
20912 * targhooks.c (default_chkp_bound_mode): Likewise.
20913 (default_setup_incoming_vararg_bounds): Likewise.
20914 * targhooks.h (default_chkp_bound_mode): Likewise.
20915 (default_setup_incoming_vararg_bounds): Likewise.
20916 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
20917 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
20918 (have_whole_vector_shift): Likewise.
20919 * tree-vect-stmts.c (vectorizable_load): Likewise.
20920 * doc/tm.texi: Regenerate.
20922 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
20924 Graceful degrade if Binutils PR21472 is not available.
20927 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
20928 .rodata in flash test fails.
20929 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
20930 * confgure: Regenerate.
20931 * config.in: Regenerate.
20932 * config/avr/avr.c (avr_asm_named_section)
20933 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
20934 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
20935 (avr_asm_init_sections): Same.
20937 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20939 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
20940 (fma<VH:mode>4_intrinsic): Likewise.
20941 (*fmsub<VCVTF:mode>4): Likewise.
20942 (*fmsub<VH:mode>4_intrinsic): Likewise.
20944 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
20947 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
20948 Don't depend on "optimize > 0".
20949 (out_movhi_r_mr, out_movqi_mr_r): Same.
20950 (out_movhi_mr_r, out_movqi_r_mr): Same.
20951 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
20952 io_address_operand on "optimize > 0".
20954 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20956 * tree-loop-distribution.c: Add general explanantion on the pass.
20957 (generate_loops_for_partition): Mark distributed loop.
20958 (pg_add_dependence_edges): New parameter. Handle alias data
20959 dependence specially and record it in the parameter if asked.
20960 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
20961 (init_partition_graph_vertices, add_partition_graph_edge): New.
20962 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
20963 (free_partition_graph_vdata, build_partition_graph): New.
20964 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
20965 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
20966 (data_ref_segment_size, latch_dominated_by_data_ref): New.
20967 (compute_alias_check_pairs, version_loop_by_alias_check): New.
20968 (version_for_distribution_p, finalize_partitions): New.
20969 (distribute_loop): Handle alias data dependence specially. Factor
20970 out loop fusion code as functions and call these functions.
20972 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20974 * tree-loop-distribution.c (classify_partition): New parameter and
20975 better handle reduction statement.
20976 (rdg_build_partitions): Revise comment.
20977 (distribute_loop): Compute statements in all partitions and pass it
20978 to classify_partition.
20980 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20982 * tree-loop-distribution.c (enum partition_type): New.
20983 (struct partition): New field type.
20984 (partition_merge_into): Add parameter. Update partition type.
20985 (data_dep_in_cycle_p, update_type_for_merge): New functions.
20986 (build_rdg_partition_for_vertex): Compute partition type.
20987 (rdg_build_partitions): Dump partition type.
20988 (distribute_loop): Update calls to partition_merge_into.
20990 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20992 * tree-loop-distribution.c (struct ddr_hasher): New.
20993 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
20995 (classify_partition): Call get_data_dependence.
20996 (pg_add_dependence_edges): Ditto.
20997 (distribute_loop): Release data dependence hash table.
20999 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21001 * tree-loop-distribution.c (ref_base_address): Delete.
21002 (similar_memory_accesses): Rename ...
21003 (share_memory_accesses): ... to this. Check if partitions access
21004 the same memory reference.
21005 (distribute_loop): Call share_memory_accesses.
21007 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21009 * tree-loop-distribution.c (struct partition): New field recording
21010 its data reference.
21011 (partition_alloc, partition_free): Init and release data refs.
21012 (partition_merge_into): Merge data refs.
21013 (build_rdg_partition_for_vertex): Collect data refs for partition.
21014 (pg_add_dependence_edges): Change parameters from vector to bitmap.
21016 (distribute_loop): Remve data refs from vertice data of partition
21019 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21021 * tree-loop-distribution.c (params.h): Include header file.
21022 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
21023 (datarefs_vec): New global var.
21024 (create_rdg_vertices): Use datarefs_vec directly.
21025 (free_rdg): Don't free data references.
21026 (build_rdg): Update use. Don't free data references.
21027 (distribute_loop): Compute global variable for data references.
21028 Bail out if there are too many data references.
21030 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21032 * tree-loop-distribution.c (loop_nest): New global var.
21033 (build_rdg): Use loop directly, rather than loop nest.
21034 (pg_add_dependence_edges): Remove loop nest parameter. Use global
21036 (distribute_loop): Compute global variable loop nest. Update use.
21038 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21040 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
21041 (partition_merge_into): New parameter. Dump reason for fusion.
21042 (distribute_loop): Update use of partition_merge_into.
21044 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21046 * tree-loop-distribution.c (bb_top_order_index): New.
21047 (bb_top_order_index_size, bb_top_order_cmp): New.
21048 (stmts_from_loop): Use topological order.
21049 (pass_loop_distribution::execute): Compute and release topological
21050 order for basic blocks.
21052 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21054 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
21057 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21059 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
21060 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
21061 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
21062 * internal-fn.def (LOOP_DIST_ALIAS): New.
21063 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
21064 (fold_loop_internal_call): ... this.
21065 (vect_loop_dist_alias_call): New function.
21066 (set_uid_loop_bbs): Call fold_loop_internal_call.
21067 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
21070 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
21073 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
21074 Require dead FLAGS_REG at the beginning of a peephole.
21076 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
21079 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
21080 arguments in the call to __builtin_ia32_sbb_u32.
21081 (_subborrow_u64): Swap _X and _Y arguments in the call to
21082 __builtin_ia32_sbb_u64.
21084 2017-07-04 Jakub Jelinek <jakub@redhat.com>
21087 * tree-vrp.c (compare_assert_loc): Turn into a function template
21088 with stable template parameter. Only test if a->e is NULL,
21089 !a->e == !b->e has been verified already. Use e == NULL or
21090 e != NULL instead of e or ! e tests. If stable is true, don't use
21091 iterative_hash_expr, on the other side allow a or b or both NULL
21092 and sort the NULLs last.
21093 (process_assert_insertions): Sort using compare_assert_loc<false>
21094 instead of compare_assert_loc, later sort using
21095 compare_assert_loc<true> before calling process_assert_insertions_for
21096 in a loop. Use break instead of continue once seen NULL pointer.
21098 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21100 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21101 Cortex-R7 and Cortex-R8 processors.
21103 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21105 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
21106 uninitialized while src is not.
21108 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
21110 * common/config/arm/arm-common.c: Adjust include path for
21112 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
21113 (arm-cpu.h): Create in build directory. Adjust dependency rules.
21114 (arm-cpu-data.h): Likewise.
21115 (arm-cpu-cdata.h): Likewise.
21116 * config/arm/arm-cpu.h: Delete.
21117 * config/arm/arm-cpu-cdata.h: Delete.
21118 * config/arm/arm-cpu-data.h: Delete.
21120 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
21122 * config/arm/arm-cpus.in (cortex-a55): New.
21123 (cortex-a75): Likewise.
21124 (cortex-a75.cortex-a55): Likewise.
21125 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
21127 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
21128 * config/arm/arm-cpu-cdata.h: Regenerate.
21129 * config/arm/arm-cpu-data.h: Regenerate.
21130 * config/arm/arm-cpu.h: Regenerate.
21131 * config/arm/arm-tables.opt: Regenerate.
21132 * config/arm/arm-tune.md: Regenerate.
21134 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21136 * haifa-sched.c (sched_create_recovery_edges): Update profile.
21138 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21140 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
21143 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
21145 PR tree-optimization/81292
21146 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
21147 full_string_p, also call adjust_related_strinfos if the adjustment
21148 is simple, otherwise invalidate related strinfos.
21150 2017-07-04 Martin Liska <mliska@suse.cz>
21153 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
21154 newly created variable as DECL_IGNORED_P.
21156 2017-07-04 Martin Liska <mliska@suse.cz>
21159 * ipa-inline.c (inline_small_functions):
21160 Use xstrdup_for_dump.
21162 2017-07-04 Tom de Vries <tom@codesourcery.com>
21164 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
21166 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
21169 * config/darwin.c (darwin_function_switched_text_sections):
21172 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
21174 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
21176 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
21178 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
21180 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21182 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
21183 min_profitable_iters, and th as inclusive lower bounds.
21184 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
21185 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
21186 for min_profitable_iters and min_profitable_estimate.
21187 (vect_transform_loop): Treat th as an inclusive lower bound.
21188 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
21190 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
21193 * config/darwin.c (darwin_function_switched_text_sections):
21194 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
21195 in two pieces, and suppress the use of buf.
21197 2017-07-03 Nathan Sidwell <nathan@acm.org>
21199 * hash-table.h (hash_table_mod1): Fix indentation.
21201 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21203 PR middle-end/81290
21204 * predict.c (force_edge_cold): Be more careful about propagation
21206 * profile-count.h (profile_probability::guessed,
21207 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
21209 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
21211 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
21213 * doc/invoke.texi (rcpc architecture extension): Document it.
21215 2017-07-03 Richard Biener <rguenther@suse.de>
21217 PR tree-optimization/60510
21218 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
21219 the scalar reduction PHI and use it.
21220 (vectorizable_reduction): Properly guard the single_defuse_cycle
21221 path for non-SLP reduction chains where we cannot use it.
21222 Rework reduc_def/index and vector type deduction. Rework
21223 vector operand gathering during reduction op code-gen.
21224 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
21225 chains dissolve the chain and leave it to non-SLP reduction
21228 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21230 * tree-data-ref.h (dr_alignment): Declare.
21231 * tree-data-ref.c (dr_alignment): New function.
21232 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
21233 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
21235 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
21237 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21239 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
21240 and base_misalignment fields.
21241 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
21242 * tree-data-ref.c: Include builtins.h.
21243 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
21244 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
21245 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
21246 * tree-vect-data-refs.c: Include tree-cfg.h.
21247 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
21248 fields instead of calculating an alignment here.
21249 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
21250 innermost_loop_behavior fields.
21252 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21254 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
21256 (DR_STEP_ALIGNMENT): New macro.
21257 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
21258 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
21259 (create_data_ref): Print it.
21260 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
21261 to tell whether the step preserves vector (mis)alignment.
21262 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
21263 Move the check for an integer step and generalise to all INTEGER_CST.
21264 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
21265 Print the outer step alignment.
21267 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21269 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
21270 with offset_alignment.
21271 (DR_ALIGNED_TO): Delete.
21272 (DR_OFFSET_ALIGNMENT): New macro.
21273 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
21274 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
21275 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
21276 (create_data_ref): Likewise.
21277 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
21278 (vect_analyze_data_refs): Likewise.
21279 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
21280 creating dummy innermost behavior.
21282 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21284 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
21285 with a "innermost_loop_behavior *" and refeence tree.
21286 * tree-data-ref.c (dr_analyze_innermost): Likewise.
21287 (create_data_ref): Update call accordingly.
21288 * tree-predcom.c (find_looparound_phi): Likewise.
21290 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21292 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
21293 fields with dr_wrt_vec_loop.
21294 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
21295 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
21296 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
21297 (vect_dr_behavior): New function.
21298 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
21299 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
21300 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
21301 track whether the step preserves the misalignment.
21302 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
21303 Use vect_dr_behavior.
21304 (vect_setup_realignment): Update call accordingly.
21305 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
21306 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
21307 call to vect_create_addr_base_for_vector_ref.
21308 (vect_create_cond_for_align_checks): Likewise.
21309 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
21310 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
21311 (vect_recog_mask_conversion_pattern): Likewise.
21312 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
21313 (new_stmt_vec_info): Remove redundant zeroing.
21315 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
21317 * common/config/arm/arm-common.c (arm_be8_option): New function.
21318 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
21319 (ISA_ARMv6): Add isa_bit_be8.
21320 * config/arm/arm.h (arm_be8_option): Add prototype.
21321 (BE8_SPEC_FUNCTION): New define.
21322 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
21323 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
21324 (mlittle-endian): Similarly.
21325 (mbe8, mbe32): New options.
21326 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
21327 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
21329 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21331 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
21333 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21335 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
21336 (cleanup_tree_cfg_bb): Use it.
21337 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
21339 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
21341 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21344 * loop-doloop.c (add_test): Update profile.
21346 2017-07-03 Martin Liska <mliska@suse.cz>
21349 * sanopt.c (rewrite_usage_of_param): New function.
21350 (sanitize_rewrite_addressable_params): Likewise.
21351 (pass_sanopt::execute): Call rewrite_usage_of_param.
21353 2017-07-03 Richard Biener <rguenther@suse.de>
21355 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
21356 back to using VIEW_CONVERT_EXPR.
21358 2017-07-03 Martin Liska <mliska@suse.cz>
21361 * doc/extend.texi: Document when a resolver function is
21362 generated for target_clones.
21364 2017-07-03 Martin Liska <mliska@suse.cz>
21366 * asan.c (asan_emit_stack_protection): Unpoison just red zones
21367 and shadow memory of auto variables which are subject of
21368 use-after-scope sanitization.
21369 (asan_expand_mark_ifn): Add do set only when is_poison.
21371 2016-07-03 Richard Biener <rguenther@suse.de>
21373 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
21375 (vect_force_simple_reduction): Record reduction def -> phi mapping.
21376 (vectorizable_reduction): Perform reduction PHI creation when
21377 visiting a reduction PHI and adjust and simplify code generation
21378 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
21379 (vect_transform_loop): Visit reduction PHIs.
21380 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
21381 defs into the SLP tree.
21382 (vect_build_slp_tree): Reduction defs terminate the recursion.
21383 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
21385 (vect_get_vec_defs_for_stmt_copy): Export.
21386 (vect_get_vec_defs): Likewise.
21387 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
21389 (vect_get_vec_defs_for_stmt_copy): Declare.
21390 (vect_get_vec_defs): Likewise.
21392 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21394 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
21395 parameter with a "loop" parameter and use it instead of the
21396 loop containing DR_STMT. Don't check simple_iv when doing
21397 BB analysis. Describe the two analysis modes in the comment.
21399 2017-07-03 Tom de Vries <tom@codesourcery.com>
21401 PR tree-optimization/69468
21402 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
21403 (find_same_succ_bb): Handle ignore_edge_flags.
21405 2017-07-03 Tom de Vries <tom@codesourcery.com>
21407 PR tree-optimization/81192
21408 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
21410 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
21412 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
21414 2017-07-03 Tom de Vries <tom@codesourcery.com>
21416 PR tree-optimization/81192
21417 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
21418 BB_SAME_SUCC (bb) == NULL.
21420 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21422 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
21425 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21427 * dumpfile.c: Include profile-count.h
21428 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
21430 (insert_cond_bb): Update profile.
21431 * tree-cfg.h (insert_cond_bb): Update prototype.
21432 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
21433 * tree-dump.c: Do not include tree-cfg.
21435 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21437 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
21439 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21441 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
21444 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21446 * tree-complex.c (expand_complex_div_wide): update profile.
21448 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21449 Alan Hayward <alan.hayward@arm.com>
21450 David Sherwood <david.sherwood@arm.com>
21452 * Makefile.in (MACHMODE_H): Remove insn-modes.h
21453 (CORETYPES_H): New define.
21454 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
21455 (insn-modes-inline.h, s-modes-inline-h): New rules.
21456 (generated_files): Add insn-modes-inline.h.
21457 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
21458 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
21459 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
21460 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
21461 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
21462 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
21463 (build/gencodes.o, build/genconditions.o): Likewise.
21464 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
21465 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
21466 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
21467 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
21468 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
21469 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
21470 * coretypes.h: Include everything up to real.h for generators.
21471 Include insn-modes.h first. Include wide-int-print.h after
21472 wide-int.h. Include insn-modes-inline.h and then machmode.h.
21473 * machmode.h: Don't include insn-modes.h here.
21474 * function-tests.c: Remove includes of signop.h, machmode.h,
21475 double-int.h and wide-int.h.
21477 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
21479 * optc-save-gen.awk: Likewise.
21480 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
21481 * godump.c: Remove include of wide-int-print.h.
21482 * pretty-print.h: Likewise.
21483 * wide-int-print.cc: Likewise.
21484 * wide-int.cc: Likewise.
21485 * hash-map-tests.c: Remove include of signop.h.
21486 * hash-set-tests.c: Likewise.
21487 * rtl-tests.c: Likewise.
21488 * mkconfig.sh: Remove include of machmode.h.
21489 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
21491 (emit_insn_modes_inline_h): ...this new function. Emit the code
21492 into an insn-modes-inline.h header file, adding appropriate
21493 include guards and end comments.
21494 (emit_insn_modes_c_header): Remove include of machmode.h.
21495 (emit_min_insn_modes_c_header): Include coretypes.h rather than
21497 (main): Handle -i flag and call emit_insn_modes_inline_h when
21500 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21502 * tree-ssa-strlen.c (strinfo): Rename the length field to
21503 nonzero_chars. Add a full_string_p field.
21504 (compare_nonzero_chars, zero_length_string_p): New functions.
21505 (get_addr_stridx): Add an offset_out parameter.
21506 Use compare_nonzero_chars.
21507 (get_stridx): Update accordingly. Use compare_nonzero_chars.
21508 (new_strinfo): Update after above changes to strinfo.
21509 (set_endptr_and_length): Set full_string_p.
21510 (get_string_length): Update after above changes to strinfo.
21511 (unshare_strinfo): Update call to new_strinfo.
21512 (maybe_invalidate): Likewise.
21513 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
21514 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
21515 as a uhwi instead of an shwi. Update after above changes to
21516 strinfo and new_strinfo.
21517 (zero_length_string): Assert that chainsi contains full strings.
21518 Use zero_length_string_p. Update call to new_strinfo.
21519 (adjust_related_strinfos): Update after above changes to strinfo.
21520 Copy full_string_p from origsi.
21521 (adjust_last_stmt): Use zero_length_string_p.
21522 (handle_builtin_strlen): Update after above changes to strinfo and
21523 new_strinfo. Install the lhs as the string length if the previous
21524 entry didn't describe a full string.
21525 (handle_builtin_strchr): Update after above changes to strinfo
21527 (handle_builtin_strcpy): Likewise.
21528 (handle_builtin_strcat): Likewise.
21529 (handle_builtin_malloc): Likewise.
21530 (handle_pointer_plus): Likewise.
21531 (handle_builtin_memcpy): Likewise. Track nonzero characters
21532 that aren't necessarily followed by a nul terminator.
21533 (handle_char_store): Likewise.
21535 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21537 PR tree-optimization/80769
21538 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
21539 for malloc and calloc. Document the new invariant that all related
21540 strinfos have delayed lengths or none do.
21541 (verify_related_strinfos): Move earlier in file.
21542 (set_endptr_and_length): New function, split out from...
21543 (get_string_length): ...here. Also set the lengths of related
21545 (zero_length_string): Assert that chainsi has known (rather than
21547 (adjust_related_strinfos): Likewise.
21549 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21551 PR tree-optimization/81136
21552 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
21553 assert that two references with the same misalignment have the same
21554 compile-time misalignment if those compile-time misalignments
21557 2017-07-01 Andi Kleen <ak@linux.intel.com>
21559 * print-tree.c (print_node): Print all attributes.
21561 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21563 * cfg.c (scale_bbs_frequencies): New function.
21564 * cfg.h (scale_bbs_frequencies): Declare it.
21565 * cfgloopanal.c (single_likely_exit): Cleanup.
21566 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
21568 (scale_loop_profile): Likewise.
21569 (loop_version): Likewise.
21570 (create_empty_loop_on_edge): Update.
21571 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
21572 scale_loop_frequencies, scale_loop_profile, loopify,
21573 loop_version): Update prototypes.
21574 * modulo-sched.c (sms_schedule): Update.
21575 * predict.c (unlikely_executed_edge_p): Also check probability.
21576 (probably_never_executed_edge_p): Fix typo.
21577 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21578 * tree-parloops.c (gen_parallel_loop): Update.
21579 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
21580 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21581 * tree-ssa-loop-split.c (split_loop): Update.
21582 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21583 * tree-vect-loop-manip.c (vect_do_peeling): Update.
21584 (vect_loop_versioning): Update.
21585 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21587 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21589 * trans-mem.c (split_bb_make_tm_edge): Update profile.
21591 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21593 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
21594 to keep profile consistent.
21596 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21598 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
21599 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
21600 * profile-count.h (max_safe_multiplier): Make unsigned.
21601 (profile_count::guessed_zero): New.
21603 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21605 * bb-reorder.c (fix_up_crossing_landing_pad,
21606 fix_crossing_conditional_branches): Use make_single_succ_edge
21607 to keep profile consistent.
21609 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21611 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
21614 2017-07-01 Jakub Jelinek <jakub@redhat.com>
21617 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
21618 the right scopes, make sure cond_jump isn't preserved between multiple
21619 iterations. Search for fallthru edge whenever there are 3+ edges and
21620 use find_fallthru_edge for it.
21622 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21624 Patch by Alexander Monakov <amonakov@ispras.ru>
21625 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
21626 probabilities consistently.
21628 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21630 * pa.c (pa_expand_compare_and_swap_loop): Update call of
21631 emit_cmp_and_jump_insns.
21633 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21636 * tree-inline.c (expand_call_inline): Combine profile statuses.
21638 2017-06-30 Andrew Pinski <apinski@cavium.com>
21640 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
21641 fold_stmt returned true.
21643 2017-06-30 Nathan Sidwell <nathan@acm.org>
21645 * ggc.h (empty_string): Delete.
21646 * cfgexpand.c (expand_asm_stmt): Use plain "".
21647 * optabs.c (expand_asm_memory_barrier): Likewise.
21648 * stringpool.c (empty_string): Delete.
21649 (digit_vector, digit_string): Delete.
21650 (ggc_alloc_string): Use plain "", don't optimize single digit
21651 strings. Use ggc_alloc_atomic.
21653 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
21655 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
21656 comparison set and one other set, use the cost of the non-comparison
21659 2017-06-30 Nathan Sidwell <nathan@acm.org>
21661 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
21664 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
21666 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
21667 loops. Remove now unneeded calls to gimple_switch_set_label() that
21668 just set removed labels to NULL_TREE.
21670 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
21672 * tree-ssanames.c (set_range_info_raw): Abstract from ...
21673 (set_range_info): ...here. Only call set_range_info_raw if domain
21675 (set_nonzero_bits): Call set_range_info_raw.
21676 * tree-ssanames.h (set_range_info_raw): New.
21678 2017-06-30 Jakub Jelinek <jakub@redhat.com>
21681 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
21682 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
21683 of nonimmediate_operand and <store_mask_constraint> instead of m
21684 for the input operand. For V8FI iterator, always split if input
21685 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
21686 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
21687 <store_mask_predicate> instead of register_operand and
21688 <store_mask_constraint> instead of v for the input operand. Make
21689 sure both operands aren't MEMs for if not <mask_applied>.
21691 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
21693 * lto-wrapper.c (copy_file) Close both file descriptors before
21696 2017-06-30 Martin Liska <mliska@suse.cz>
21699 * multiple_target.c (create_dispatcher_calls): Make ifunc
21700 also for function that don't have calls or are not referenced.
21702 2017-06-30 Richard Biener <rguenther@suse.de>
21704 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
21705 analyze the first scalar stmt. Move vector type computation
21706 for the BB case here from ...
21707 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
21708 live operation processing in the SLP case properly.
21710 2017-06-30 Richard Biener <rguenther@suse.de>
21712 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
21714 2017-06-30 Martin Liska <mliska@suse.cz>
21717 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
21718 before BUILT_IN_UNWIND_RESUME when ASAN is used.
21720 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
21722 * doc/invoke.texi (AArch64): Add missing options and remove redundant
21725 2017-06-30 Richard Biener <rguenther@suse.de>
21727 PR tree-optimization/81249
21728 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
21729 condition reduction result to original scalar type.
21731 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21733 * profile-count.h (enum profile_quality): Fix typos and whitespace
21736 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21738 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
21739 type for branch probabilities.
21741 2017-06-29 Julian Brown <julian@codesourcery.com>
21742 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21744 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
21745 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
21746 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
21747 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
21749 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21751 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
21752 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
21753 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
21754 CC usage from generic code to here.
21755 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
21756 CC usage into the target macros.
21758 2017-06-29 Maya Rashish <coypu@sdf.org>
21760 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
21763 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21765 * arm/arm-builtins.c: Include profile-count.h
21766 * except.c (sjlj_emit_function_enter): Use
21767 profile_probability::unlikely.
21769 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21771 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
21772 and tocrel_offset be pointer args rather than implicitly using
21774 (legitimate_constant_pool_address_p, rs6000_emit_move,
21775 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
21776 tocrel_offset and use in toc_relative_expr_p call.
21777 (print_operand, print_operand_address): Use static tocrel_base_oac
21778 and tocrel_offset_oac.
21779 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
21782 2017-06-29 Maya Rashish <coypu@sdf.org>
21784 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
21786 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
21788 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
21789 objects, take into account only the alignment of 'op0' and 'mode1' if
21792 2017-06-29 Steve Ellcey <sellcey@cavium.com>
21794 * ccmp.c (ccmp_tree_comparison_p): New function.
21795 (ccmp_candidate_p): Update to use above function.
21796 (get_compare_parts): New function.
21797 (expand_ccmp_next): Update to use new functions.
21798 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
21800 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
21801 take mode as argument.
21802 * ccmp.h (expand_ccmp_expr): Add mode as argument.
21803 * expr.c (expand_expr_real_1): Pass mode as argument.
21805 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
21807 * combine.c (combine_instructions): Print insns to dump_file, together
21810 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21812 * asan.c (asan_emit_stack_protection): Update.
21813 (create_cond_insert_point): Update.
21814 * auto-profile.c (afdo_propagate_circuit): Update.
21815 * basic-block.h (struct edge_def): Turn probability to
21816 profile_probability.
21817 (EDGE_FREQUENCY): Update.
21818 * bb-reorder.c (find_traces_1_round): Update.
21819 (better_edge_p): Update.
21820 (sanitize_hot_paths): Update.
21821 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
21822 (make_single_succ_edge): Update.
21823 (check_bb_profile): Update.
21824 (dump_edge_info): Update.
21825 (update_bb_profile_for_threading): Update.
21826 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
21827 probabilitycount to 0.
21828 * cfgbuild.c (compute_outgoing_frequencies): Update.
21829 * cfgcleanup.c (try_forward_edges): Update.
21830 (outgoing_edges_match): Update.
21831 (try_crossjump_to_edge): Update.
21832 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
21833 (expand_gimple_tailcall): Update.
21834 (construct_init_block): Use make_single_succ_edge.
21835 (construct_exit_block): Use make_single_succ_edge.
21836 * cfghooks.c (verify_flow_info): Update.
21837 (redirect_edge_succ_nodup): Update.
21838 (split_edge): Update.
21839 (account_profile_record): Update.
21840 * cfgloopanal.c (single_likely_exit): Update.
21841 * cfgloopmanip.c (scale_loop_profile): Update.
21842 (set_zero_probability): Remove.
21843 (duplicate_loop_to_header_edge): Update.
21844 * cfgloopmanip.h (loop_version): Update prototype.
21845 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
21846 (force_nonfallthru_and_redirect): Update.
21847 (update_br_prob_note): Update.
21848 (rtl_verify_edges): Update.
21849 (purge_dead_edges): Update.
21850 (rtl_lv_add_condition_to_bb): Update.
21851 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
21852 * cgraphunit.c (init_lowered_empty_function): Update.
21853 (cgraph_node::expand_thunk): Update.
21854 * cilk-common.c: Include profile-count.h
21855 * dojump.c (inv): Remove.
21856 (jumpifnot): Update.
21857 (jumpifnot_1): Update.
21858 (do_jump_1): Update.
21860 (do_jump_by_parts_greater_rtx): Update.
21861 (do_compare_rtx_and_jump): Update.
21862 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
21863 do_jump_1. do_compare_rtx_and_jump): Update prototype.
21864 * dwarf2cfi.c: Include profile-count.h
21865 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
21866 (sjlj_emit_dispatch_table): Likewise.
21867 * explow.c: Include profile-count.h
21868 * expmed.c (emit_store_flag_force): Update.
21869 (do_cmp_and_jump): Update.
21870 * expr.c (compare_by_pieces_d::generate): Update.
21871 (compare_by_pieces_d::finish_mode): Update.
21872 (emit_block_move_via_loop): Update.
21873 (store_expr_with_bounds): Update.
21874 (store_constructor): Update.
21875 (expand_expr_real_2): Update.
21876 (expand_expr_real_1): Update.
21877 * expr.h (try_casesi, try_tablejump): Update prototypes.
21878 * gimple-pretty-print.c (dump_probability): Update.
21879 (dump_profile): New.
21880 (dump_gimple_label): Update.
21881 (dump_gimple_bb_header): Update.
21882 * graph.c (draw_cfg_node_succ_edges): Update.
21883 * hsa-gen.c (convert_switch_statements): Update.
21884 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
21885 (find_if_case_1): Update.
21886 (find_if_case_2): Update.
21887 * internal-fn.c (expand_arith_overflow_result_store): Update.
21888 (expand_addsub_overflow): Update.
21889 (expand_neg_overflow): Update.
21890 (expand_mul_overflow): Update.
21891 (expand_vector_ubsan_overflow): Update.
21892 * ipa-cp.c (good_cloning_opportunity_p): Update.
21893 * ipa-split.c (split_function): Use make_single_succ_edge.
21894 * ipa-utils.c (ipa_merge_profiles): Update.
21895 * loop-doloop.c (add_test): Update.
21896 (doloop_modify): Update.
21897 * loop-unroll.c (compare_and_jump_seq): Update.
21898 (unroll_loop_runtime_iterations): Update.
21899 * lra-constraints.c (lra_inheritance): Update.
21900 * lto-streamer-in.c (input_cfg): Update.
21901 * lto-streamer-out.c (output_cfg): Update.
21902 * mcf.c (adjust_cfg_counts): Update.
21903 * modulo-sched.c (sms_schedule): Update.
21904 * omp-expand.c (expand_omp_for_init_counts): Update.
21905 (extract_omp_for_update_vars): Update.
21906 (expand_omp_ordered_sink): Update.
21907 (expand_omp_for_ordered_loops): Update.
21908 (expand_omp_for_generic): Update.
21909 (expand_omp_for_static_nochunk): Update.
21910 (expand_omp_for_static_chunk): Update.
21911 (expand_cilk_for): Update.
21912 (expand_omp_simd): Update.
21913 (expand_omp_taskloop_for_outer): Update.
21914 (expand_omp_taskloop_for_inner): Update.
21915 * omp-simd-clone.c (simd_clone_adjust): Update.
21916 * optabs.c (expand_doubleword_shift): Update.
21917 (expand_abs): Update.
21918 (emit_cmp_and_jump_insn_1): Update.
21919 (expand_compare_and_swap_loop): Update.
21920 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
21921 * predict.c (predictable_edge_p): Update.
21922 (edge_probability_reliable_p): Update.
21923 (set_even_probabilities): Update.
21924 (combine_predictions_for_insn): Update.
21925 (combine_predictions_for_bb): Update.
21926 (propagate_freq): Update.
21927 (estimate_bb_frequencies): Update.
21928 (force_edge_cold): Update.
21929 * profile-count.c (profile_count::dump): Add missing space into dump.
21930 (profile_count::debug): Add newline.
21931 (profile_count::differs_from_p): Explicitly convert to unsigned.
21932 (profile_count::stream_in): Update.
21933 (profile_probability::dump): New member function.
21934 (profile_probability::debug): New member function.
21935 (profile_probability::differs_from_p): New member function.
21936 (profile_probability::differs_lot_from_p): New member function.
21937 (profile_probability::stream_in): New member function.
21938 (profile_probability::stream_out): New member function.
21939 * profile-count.h (profile_count_quality): Rename to ...
21940 (profile_quality): ... this one.
21941 (profile_probability): New.
21942 (profile_count): Update.
21943 * profile.c (compute_branch_probabilities): Update.
21944 * recog.c (peep2_attempt): Update.
21945 * sched-ebb.c (schedule_ebbs): Update.
21946 * sched-rgn.c (find_single_block_region): Update.
21947 (compute_dom_prob_ps): Update.
21948 (schedule_region): Update.
21949 * sel-sched-ir.c (compute_succs_info): Update.
21950 * stmt.c (struct case_node): Update.
21951 (do_jump_if_equal): Update.
21952 (get_outgoing_edge_probs): Update.
21953 (conditional_probability): Update.
21954 (emit_case_dispatch_table): Update.
21955 (expand_case): Update.
21956 (expand_sjlj_dispatch_table): Update.
21957 (emit_case_nodes): Update.
21958 * targhooks.c: Update.
21959 * tracer.c (better_p): Update.
21960 (find_best_successor): Update.
21961 * trans-mem.c (expand_transaction): Update.
21962 * tree-call-cdce.c: Update.
21963 * tree-cfg.c (gimple_split_edge): Upate.
21964 (move_sese_region_to_fn): Upate.
21965 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
21966 * tree-eh.c (lower_resx): Upate.
21967 (cleanup_empty_eh_move_lp): Upate.
21968 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21969 * tree-inline.c (copy_edges_for_bb): Update.
21970 (copy_cfg_body): Update.
21971 * tree-parloops.c (gen_parallel_loop): Update.
21972 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
21973 (gimple_gen_time_profiler): Update.
21974 * tree-ssa-dce.c (remove_dead_stmt): Update.
21975 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21976 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
21977 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
21978 (unloop_loops): Update.
21979 (try_peel_loop): Update.
21980 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21981 * tree-ssa-loop-split.c (connect_loops): Update.
21982 (split_loop): Update.
21983 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21984 (hoist_guard): Update.
21985 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
21986 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
21987 (value_replacement): Update.
21988 * tree-ssa-reassoc.c (branch_fixup): Update.
21989 * tree-ssa-tail-merge.c (replace_block_by): Update.
21990 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
21991 (create_edge_and_update_destination_phis): Update.
21992 (compute_path_counts): Update.
21993 (recompute_probabilities): Update.
21994 (update_joiner_offpath_counts): Update.
21995 (freqs_to_counts_path): Update.
21996 (duplicate_thread_path): Update.
21997 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
21998 (struct switch_conv_info): Update.
21999 (gen_inbound_check): Update.
22000 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22001 (vect_do_peeling): Update.
22002 (vect_loop_versioning): Update.
22003 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22004 (optimize_mask_stores): Update.
22005 * ubsan.c (ubsan_expand_null_ifn): Update.
22006 * value-prof.c (gimple_divmod_fixed_value): Update.
22007 (gimple_divmod_fixed_value_transform): Update.
22008 (gimple_mod_pow2): Update.
22009 (gimple_mod_pow2_value_transform): Update.
22010 (gimple_mod_subtract): Update.
22011 (gimple_mod_subtract_transform): Update.
22012 (gimple_ic): Update.
22013 (gimple_stringop_fixed_value): Update.
22014 (gimple_stringops_transform): Update.
22015 * value-prof.h: Update.
22017 2017-06-29 Carl Love <cel@us.ibm.com>
22019 * config/rs6000/rs6000-c.c: Add support for built-in functions
22020 vector signed int vec_signed (vector float);
22021 vector signed long long vec_signed (vector double);
22022 vector signed int vec_signed2 (vector double, vector double);
22023 vector signed int vec_signede (vector double);
22024 vector signed int vec_signedo (vector double);
22025 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
22026 instruction generator.
22027 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22028 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
22029 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
22031 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
22032 vunsignede_v2df): Add define_expands.
22033 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
22034 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
22035 VEC_UNSIGNEDO): Add definitions.
22036 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22037 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
22038 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
22039 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
22040 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
22041 * config/rs6000/altivec.h (vec_signed, vec_signed2,
22042 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
22043 vec_unsignede, vec_unsignedo): Add builtin defines.
22044 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
22046 * doc/extend.texi: Update the built-in documentation file for the
22047 new built-in functions.
22049 2017-06-29 Richard Biener <rguenther@suse.de>
22051 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
22052 reduction chains to LOOP_VINFO_REDUCTIONS.
22053 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
22054 SLP reductions after processing reduction chains.
22056 2017-06-29 Nathan Sidwell <nathan@acm.org>
22058 * builtins.c (fold_builtin_FUNCTION): Use
22059 lang_hooks.decl_printable_name.
22061 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
22063 PR middle-end/81194
22064 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
22065 with only one label.
22066 * stmt.c (expand_case): Assert NCASES is greater than one.
22068 2017-06-29 Richard Biener <rguenther@suse.de>
22070 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
22072 (group_case_labels): Likewise.
22073 (find_taken_edge): Push sanity checking on val to workers...
22074 (find_taken_edge_cond_expr): ... here
22075 (find_taken_edge_switch_expr): ... and here, handle cases
22076 with just a default label.
22077 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
22078 (group_case_labels): Likewise.
22079 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
22080 group_case_labels does anything cleanup the CFG again.
22082 2017-06-29 Bin Cheng <bin.cheng@arm.com>
22084 PR tree-optimization/81196
22085 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
22086 exit condition comparing two IVs.
22088 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
22090 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
22091 profile to the dummy entry at the end of the list of architectures.
22092 * config/arm/arm-cpu-cdata.h: Regenerated.
22094 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22095 Michael Collison <michael.collison@arm.com>
22098 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
22100 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
22101 (*aarch64_reg_<mode>3_minus_mask): New pattern.
22102 (*aarch64_<optab>_reg_di3_mask2): New pattern.
22103 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
22104 of shift when the shift amount is masked with constant equal to
22105 the size of the mode.
22106 * config/aarch64/predicates.md (subreg_lowpart_operator): New
22109 2017-06-29 Martin Liska <mliska@suse.cz>
22111 * config/i386/i386.opt: Change range from [1,5] to [0,5].
22113 2017-06-29 Yury Gribov <tetra2005@gmail.com>
22116 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
22119 (edge_growth_cache_entry::edge_growth_cache_entry): New
22121 (reset_edge_growth_cache): Update to use constructor.
22123 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22125 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
22126 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22127 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22129 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
22131 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
22132 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
22134 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
22136 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
22137 (*-linux-uclibc*): Add t-uclibc tmake_file.
22138 * config/t-musl: New.
22139 * config/t-uclibc: New.
22141 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
22143 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
22145 (gen_comm_data): Emit architectural setting of arch_prof.
22146 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
22148 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
22149 (armv8-m.base, armv8-m.main): Likewise.
22150 * arm-protos.h (arm_build_target): Add profile field.
22151 (arch_option): Likewise.
22152 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
22154 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
22155 arm_active_target.profile.
22157 2017-06-28 Richard Biener <rguenther@suse.de>
22159 PR middle-end/81227
22160 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
22161 TYPE_OVERFLOW_WRAPS.
22162 * match.pd (negate_expr_p): Likewise.
22163 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
22164 fold_build2, not fold_binary.
22166 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
22168 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
22169 Convert memory address to Pmode.
22170 (aarch64_print_operand): Assert MEM operands are always Pmode.
22172 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
22175 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
22176 Remove redundant if.
22177 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
22178 * config/arm/aarch-common-protos.h
22179 (aarch_forward_to_shift_is_not_shifted_re): Remove.
22180 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
22182 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
22185 * multiple_target.c (create_dispatcher_calls): Set the default
22186 clone to be static, not public.
22188 2017-06-28 Richard Biener <rguenther@suse.de>
22190 * tree-vect-loop.c (vectorizable_reduction): Move special
22191 cond reduction IV var creation ...
22192 (vect_create_epilog_for_reduction): ... here. Remove induction_index
22193 parameter. Use STMT_VINFO_VECTYPE.
22194 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
22197 2017-06-28 Martin Liska <mliska@suse.cz>
22200 * ipa-visibility.c (non_local_p): Handle visibility.
22202 2017-06-28 Martin Liska <mliska@suse.cz>
22205 * common.opt: Add IntegerRange to various options.
22206 * opt-functions.awk (integer_range_info): New function.
22207 * optc-gen.awk: Add integer_range_info to cl_options struct.
22208 * opts-common.c (decode_cmdline_option): Handle
22209 CL_ERR_INT_RANGE_ARG.
22210 (cmdline_handle_error): Likewise.
22211 * opts.c (print_filtered_help): Show valid interval in
22212 when --help is provided.
22213 * opts.h (struct cl_option): Add range_min and range_max fields.
22214 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
22216 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
22218 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
22219 (x * C EQ/NE y * C): New transformation.
22221 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
22223 * genmultilib (combination_space): Accept '+' in option names.
22225 2017-06-28 Martin Liska <mliska@suse.cz>
22228 * asan.c (instrument_derefs): Bail out inner references
22229 that are hard register variables.
22231 2017-06-28 Jakub Jelinek <jakub@redhat.com>
22234 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
22235 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
22237 2017-06-28 Richard Biener <rguenther@suse.de>
22239 * tree-vectorizer.h (vect_get_vec_defs): Remove.
22240 (vect_get_slp_defs): Adjust.
22241 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
22243 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
22245 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
22246 get_initial_defs_for_reduction instead of vect_get_vec_defs.
22247 (vectorizable_reduction): Adjust.
22248 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
22250 (vect_get_slp_defs): Likewise.
22251 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
22252 (vectorizable_bswap): Adjust.
22253 (vectorizable_call): Likewise.
22254 (vectorizable_conversion): Likewise.
22255 (vectorizable_assignment): Likewise.
22256 (vectorizable_shift): Likewise.
22257 (vectorizable_operation): Likewise.
22258 (vectorizable_store): Likewise.
22259 (vectorizable_condition): Likewise.
22260 (vectorizable_comparison): Likewise.
22262 2017-06-28 Michael Collison <michael.collison@arm.com>
22265 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
22267 (arm_gen_movmemqi): Removed unused variable 'i'.
22268 Convert 'for' loop into 'while' loop.
22269 (arm_expand_prologue): Remove last unnecessary set of insn.
22270 (thumb_pop): Remove unused variable 'pushed_words'.
22271 (thumb_exit): Remove last unnecessary set of regs_to_pop.
22273 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22275 * config/s390/predicates.md: Use s390_rel_address_ok_p.
22276 * config/s390/s390-protos.h: Add prototype of
22277 s390_rel_address_ok_p.
22278 * config/s390/s390.c (s390_got_symbol): New function.
22279 (s390_rel_address_ok_p): New function.
22280 (legitimize_pic_address): Use s390_rel_address_ok_p.
22281 (s390_load_got): Use s390_got_symbol.
22282 (s390_option_override): Issue error if
22283 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
22284 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
22286 * config/s390/s390.opt: New option mpic-data-is-text-relative.
22288 2017-06-27 Andrew Pinski <apinski@cavium.com>
22290 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
22291 (X * copysign (1.0, X)): New pattern.
22292 (X * copysign (1.0, -X)): New pattern.
22293 (copysign (-1.0, CST)): New pattern.
22295 2017-06-27 Joseph Myers <joseph@codesourcery.com>
22297 * genmultilib (combination_space): Remove variable.
22298 Validate reuse rules against regular expression for any sequence
22299 of multilib options in any order.
22301 2017-06-27 Michael Collison <michael.collison@arm.com>
22303 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
22304 call aarch64_split_simd_combine.
22305 * (aarch64_combine_internal<mode>): Delete pattern.
22306 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
22307 Allow register and subreg operands.
22309 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22311 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
22312 specific need, just fallback on defaults.
22313 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
22315 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22316 Olivier Hainque <hainque@adacore.com>
22318 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
22320 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
22321 targets. Pick a default if no particular attempt applied.
22322 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
22325 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22327 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
22328 (x86_64-wrs-vxworks7): Likewise.
22330 2017-06-27 Marek Polacek <polacek@redhat.com>
22333 * ubsan.c (instrument_null): Check get_base_address's result for null.
22335 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22337 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
22339 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22341 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
22342 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
22343 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
22344 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
22345 New function types.
22346 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
22347 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
22348 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
22349 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
22350 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
22351 BUILT_IN_FEUPDATEENV): New builtins.
22352 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
22353 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
22354 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
22355 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
22357 (builtin_structptr_types): Adjust size.
22358 * tree.c (builtin_structptr_types): Add four entries.
22360 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22361 Olivier Hainque <hainque@adacore.com>
22363 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
22364 (TLS_SYM): New local macro, forcing reference to __tls__ on
22365 link command lines for VxWorks 7 RTPs, triggering initialization
22367 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
22368 OS features TLS support, true for RTPs on VxWorks 7.
22369 * config/vxworks.c (vxworks_override_options): Setup emutls
22372 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22374 * predict.c (test_prediction_value_range): Use -1U instead of -1
22375 to avoid narrowing conversion warning.
22376 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
22377 to avoid narrowing conversion warning.
22378 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
22380 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
22382 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22384 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
22385 64bit configurations.
22386 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
22387 (SIZE_TYPE): Likewise.
22388 * config/vxworks.c (vxworks_emutls_var_fields): Use
22389 long_unsigned_type_node instead of unsigned_type_node as the offset
22390 field type, which is "pointer" mode in emutls.c.
22392 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22395 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
22397 PR middle-end/81207
22398 * gimple-fold.c (replace_call_with_call_and_fold): Handle
22399 gimple_vuse copying separately from gimple_vdef copying.
22401 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22403 * value-prof.c (free_hist): Remove call to memset and the enclosing if
22406 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22407 Olivier Hainque <hainque@adacore.com>
22409 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
22410 for all vxworks7 targets.
22411 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
22412 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
22413 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
22414 variations for VX6/VX7 and 32/64bits later on in ...
22415 (VXWORKS_LIB_SPEC): Leverage new macros.
22416 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
22417 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
22419 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22421 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
22422 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
22424 2017-06-26 Carl Love <cel@us.ibm.com>
22426 * config/rs6000/rs6000-c.c: Add support for built-in functions
22427 vector bool char vec_reve (vector bool char);
22428 vector signed char vec_reve (vector signed char);
22429 vector unsigned char vec_reve (vector unsigned char);
22430 vector bool int vec_reve (vector bool int);
22431 vector signed int vec_reve (vector signed int);
22432 vector unsigned int vec_reve (vector unsigned int);
22433 vector bool long long vec_reve (vector bool long long);
22434 vector signed long long vec_reve (vector signed long long);
22435 vector unsigned long long vec_reve (vector unsigned long long);
22436 vector bool short vec_reve (vector bool short);
22437 vector signed short vec_reve (vector signed short);
22438 vector double vec_reve (vector double);
22439 vector float vec_reve (vector float);
22440 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
22441 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
22442 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
22443 (altivec_vreve): New pattern.
22444 * config/rs6000/altivec.h (vec_reve): New define.
22445 * doc/extend.texi (vec_rev): Update the built-in documentation file
22446 for the new built-in functions.
22448 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22450 PR tree-optimization/71815
22451 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
22453 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
22455 (slsr_process_phi): Likewise.
22456 (replace_uncond_cands_and_profitable_phis): Don't replace a
22457 multiply candidate with a stride of 1 (copy or cast).
22458 (phi_incr_cost): Call uses_consumed_by_stmt rather than
22460 (lowest_cost_path): Likewise.
22461 (total_savings): Likewise.
22463 2017-06-26 Richard Biener <rguenther@suse.de>
22466 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
22467 Use def_builtin_pure for all gather builtins.
22469 2017-06-26 Richard Biener <rguenther@suse.de>
22471 PR tree-optimization/81203
22472 * tree-tailcall.c (find_tail_calls): Do not move stmts into
22473 non-dominating BBs.
22475 2017-06-26 Marek Polacek <polacek@redhat.com>
22478 * doc/invoke.texi: Document -Wmultistatement-macros.
22480 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
22482 * doc/sourcebuild.texi (ARM-specific attributes): Document new
22483 arm_neon_ok_no_float_abi effective target.
22485 2017-06-26 Richard Biener <rguenther@suse.de>
22487 PR tree-optimization/80928
22488 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
22489 (copy_bbs): Set BB_DUPLICATED flag early.
22490 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
22492 (execute_on_shrinking_pred): Likewise.
22493 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
22494 BB_DUPLICATED blocks.
22495 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
22496 iterate over all PHIs considering removal of *gsi.
22498 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
22500 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
22503 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22505 * config/rs6000/rs6000-string.c: (expand_block_clear,
22506 do_load_for_compare, select_block_compare_mode,
22507 compute_current_alignment, expand_block_compare,
22508 expand_strncmp_align_check, expand_strn_compare,
22509 expand_block_move, rs6000_output_load_multiple)
22510 Move functions related to string/block move/compare
22511 to a separate file.
22512 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
22513 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
22514 for this function which is now used in two files.
22515 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
22516 * config.gcc: Add rs6000-string.o to extra_objs for
22517 targets powerpc*-*-* and rs6000*-*-*.
22519 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
22522 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
22523 32-bit, since indexed is not valid for DImode.
22524 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
22525 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
22526 (define_peephole2 for Altivec d-form load): Add 32-bit support.
22527 (define_peephole2 for Altivec d-form store): Likewise.
22530 * multiple_target.c (create_dispatcher_calls): Only create the
22531 dispatcher call if the function is the default clone of a
22532 versioned function.
22534 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
22536 PR middle-end/80902
22537 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
22538 a call, force the call to not be a tail call.
22540 2017-06-23 Jeff Law <law@redhat.com>
22542 * doc/contrib.texi: Add entry for Steven Pemberton's work on
22545 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
22547 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
22548 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
22549 handling for early expansion of vector shifts (sl,sr,sra,rl).
22550 (builtin_function_type): Add vector shift right instructions
22551 to the unsigned argument list.
22553 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
22555 rtl-optimizatoin/79286
22556 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
22557 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
22558 trap. PIC register plus a const unspec without offset can never trap.
22560 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
22562 * tree.h (builtin_structptr_type): New type.
22563 (builtin_structptr_types): Declare new array.
22564 * tree.c (builtin_structptr_types): New array.
22565 (free_lang_data, build_common_tree_nodes): Use it.
22567 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
22570 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
22573 2017-06-22 Matt Turner <mattst88@gmail.com>
22575 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
22576 Lake models to skylake case. Assume skylake for unknown
22577 models with clflushopt.
22579 2017-06-22 Jeff Law <law@redhat.com>
22581 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
22582 frame sizes that do not satisfy aarch64_uimm12_shift.
22584 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
22586 * profile-count.h (apply_probability,
22587 apply_scale, probability_in): Fix checks for zero.
22589 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22591 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
22592 * doc/cppdiropts.texi (-I @var{dir}): Document it.
22594 2016-06-22 Richard Biener <rguenther@suse.de>
22596 * tree-vect-loop.c (vect_model_reduction_cost): Handle
22597 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
22598 REDUC_MAX_EXPR support.
22599 (vectorizable_reduction): Likewise.
22600 (vect_create_epilog_for_reduction): Likewise.
22602 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
22604 * match.pd (A / (1 << B) -> A >> B): New.
22605 * generic-match-head.c: Include optabs-tree.h.
22606 * gimple-match-head.c: Likewise.
22607 * optabs-tree.h (target_supports_op_p): New.
22608 * optabs-tree.c (target_supports_op_p): New.
22610 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22612 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
22613 $gcc_cv_ld --help output.
22614 (gcc_cv_ld_demangle): Likewise.
22615 (gcc_cv_ld_eh_frame_hdr): Likewise.
22616 (gcc_cv_ld_pie): Likewise.
22617 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
22618 (gcc_cv_ld_buildid): Likewise.
22619 (gcc_cv_ld_sysroot): Likewise.
22620 (ld_bndplt_support): Likewise.
22621 (ld_pushpopstate_support): Likewise.
22622 * configure: Regenerate.
22623 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
22625 2017-06-21 Jakub Jelinek <jakub@redhat.com>
22628 * config/i386/sse.md (round<mode>2): Renumber match_dup and
22629 operands indexes to avoid gap between operands and match_dups.
22631 2017-06-21 Andrew Pinski <apinski@cavium.com>
22633 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
22634 Increment Arith_shift and Arith_shift_reg by 1.
22635 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
22637 * config/aarch64/aarch64.c (thunderx_tunings): Enable
22638 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
22639 (aarch64_strip_extend): Add new argument and test for it.
22640 (aarch64_cheap_mult_shift_p): New function.
22641 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
22642 add a cost if it is true.
22643 Update calls to aarch64_strip_extend.
22644 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
22646 2017-06-21 Andrew Pinski <apinski@cavium.com>
22648 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
22650 (thunderxt88): Likewise.
22651 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
22652 (thunderx_prefetch_tune): New variable.
22653 (thunderx2t99_prefetch_tune): Update for the correct values.
22654 (thunderxt88_tunings): New variable.
22655 (thunderx_tunings): Use thunderx_prefetch_tune instead of
22656 generic_prefetch_tune.
22657 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
22659 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22661 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
22662 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
22663 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
22664 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
22665 (aarch64_atomic_cas<mode>, GPI): Likewise.
22667 2017-06-21 Martin Liska <mliska@suse.cz>
22669 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
22670 statements on cold and hot labels.
22671 * predict.c (tree_estimate_probability_bb): Remove the
22672 prediction from this place.
22674 2017-06-21 Martin Liska <mliska@suse.cz>
22676 PR tree-optimization/79489
22677 * gimplify.c (maybe_add_early_return_predict_stmt): New
22679 (gimplify_return_expr): Call the function.
22680 * predict.c (tree_estimate_probability_bb): Remove handling
22682 * predict.def: Update comment about early return predictor.
22683 * gimple-predict.h (is_gimple_predict): New function.
22684 * predict.def: Change default value of early return to 66.
22685 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
22687 * passes.def: Put pass_strip_predict_hints to the beginning of
22690 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
22692 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
22693 FUNCTION_DECL declarations.
22694 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
22696 (dwaf2out_decl): Likewise.
22697 * godump.c (go_early_global_decl): Skip call to the real debug hook
22698 for FUNCTION_DECL declarations.
22699 * passes.c (rest_of_decl_compilation): Skip call to the
22700 early_global_decl debug hook for FUNCTION_DECL declarations, unless
22701 -fdump-go-spec is passed.
22703 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22705 * config/i386/i386.c (struct builtin_isa): New field pure_p.
22706 Reorder for compactness.
22707 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
22708 (def_builtin_pure, def_builtin_pure2): New functions.
22709 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
22711 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22713 * match.pd (nop_convert): New predicate.
22714 ((A +- CST1) +- CST2): Allow some NOP conversions.
22716 2017-06-21 Jakub Jelinek <jakub@redhat.com>
22719 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
22720 with ctors/dtors if GOVD_SHARED is set.
22722 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22724 * config/aarch64/aarch64.md (movti_aarch64):
22725 Emit mov rather than orr.
22726 (movtf_aarch64): Likewise.
22727 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
22728 Emit mov rather than orr.
22730 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22732 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
22733 Swap alternatives, make integer dup more expensive.
22735 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22737 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
22738 Return true for non-tls symbols.
22740 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
22742 * config/aarch64/aarch64-cores.def (cortex-a55): New.
22743 (cortex-a75): Likewise.
22744 (cortex-a75.cortex-a55): Likewise.
22745 * config/aarch64/aarch64-tune.md: Regenerate.
22746 * doc/invoke.texi (-mtune): Document new values for -mtune.
22748 2017-06-21 Tom de Vries <tom@codesourcery.com>
22750 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
22751 stack_size feature.
22752 (Effective-Target Keywords, Other attributes): Suggest using
22753 dg-add-options stack_size feature to get stack limit in stack_size
22754 effective target documentation.
22756 2017-06-21 Julian Brown <julian@codesourcery.com>
22757 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22759 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
22760 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
22761 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
22763 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
22764 attribute type list for neon_multiply.
22765 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
22766 attribute type list for neon_multiply.
22767 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
22768 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
22769 attribute type list for neon_multiply.
22770 * config/arm/types.md (crypto_pmull): Add.
22771 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
22772 attribute type list.
22774 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
22776 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
22779 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22781 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
22782 to make sure not to dereference a NULL cost_classes_ptr pointer.
22784 2017-06-20 Carl Love <cel@us.ibm.com>
22786 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22787 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22788 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22789 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22790 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22791 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22792 VMULOSW): New enum "unspec" values.
22793 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22794 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22795 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22796 altivec_vmulosw): New patterns.
22797 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22798 VMULOSW): Add definitions.
22800 2017-06-20 Julia Koval <julia.koval@intel.com>
22802 * config/i386/i386.c: Fix rounding expand for new pattern.
22803 * config/i386/subst.md: Fix pattern (parallel -> unspec).
22805 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22807 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
22808 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
22810 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22812 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
22815 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22817 * config/aarch64/aarch64-cores.def: Rearrange to sort by
22818 architecture, then by implementer ID.
22819 * config/aarch64/aarch64-tune.md: Regenerate.
22821 2017-06-20 Richard Biener <rguenther@suse.de>
22823 PR middle-end/81097
22824 * fold-const.c (split_tree): Fold to type before negating.
22826 2017-06-20 David Malcolm <dmalcolm@redhat.com>
22828 * diagnostic-show-locus.c
22829 (selftest::test_fixit_deletion_affecting_newline): New function.
22830 (selftest::diagnostic_show_locus_c_tests): Call it.
22832 2017-06-20 Andreas Schwab <schwab@suse.de>
22835 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
22838 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22840 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
22841 __ARM_FEATURE_COPROC according to support.
22843 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22845 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22846 Rewritten to avoid overflow for > 32-bit pointers.
22849 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
22850 by removing enum keyword.
22851 (ubsan_type_descriptor): Likewise. Formatting fix.
22854 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
22855 splitter): Require TARGET_SSE2 in the condition.
22857 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
22860 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
22861 for doing vector set of SFmode on ISA 3.0.
22862 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
22863 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
22865 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
22866 SFmode value into a V4SF variable that was extracted from another
22867 V4SF variable without converting the element to double precision
22868 and back to single precision vector format.
22869 (vsx_insert_extract_v4sf_p9_2): Likewise.
22871 2017-06-19 Jakub Jelinek <jakub@redhat.com>
22873 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
22874 in UWHI to avoid undefined overflow.
22877 * ubsan.h (enum ubsan_encode_value_phase): New.
22878 (ubsan_encode_value): Change second argument to
22879 enum ubsan_encode_value_phase with default value of
22880 UBSAN_ENCODE_VALUE_GENERIC.
22881 * ubsan.c (ubsan_encode_value): Change second argument to
22882 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
22883 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
22884 create_tmp_var_raw instead of create_tmp_var and use a
22886 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
22887 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
22888 ubsan_encode_value callers.
22891 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
22892 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
22893 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
22895 2017-06-19 Richard Biener <rguenther@suse.de>
22897 PR middle-end/81118
22898 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
22899 estimates if we changed anything.
22901 2017-06-19 Richard Biener <rguenther@suse.de>
22903 PR tree-optimization/80887
22904 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
22905 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
22906 simplified lookups, then reset mprts_hook.
22907 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
22909 (try_to_simplify): Likewise.
22911 2017-06-19 Martin Liska <mliska@suse.cz>
22914 * gimplify.c (gimplify_switch_expr):
22915 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
22917 2017-06-19 Martin Liska <mliska@suse.cz>
22919 * doc/install.texi: Document that PGO runs in 4 stages.
22921 2017-06-19 Martin Liska <mliska@suse.cz>
22924 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
22925 to dispatcher function name.
22926 * multiple_target.c (replace_function_decl): New function.
22927 (create_dispatcher_calls): Redirect both edges and references.
22929 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
22931 * profile-count.c (profile_count::dump): Dump quality.
22932 (profile_count::differs_from_p): Update for unsigned val.
22933 * profile-count.h (profile_count_quality): New enum.
22934 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
22936 2017-06-19 Richard Biener <rguenther@suse.de>
22938 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
22939 struct function as arg.
22940 (estimate_numbers_of_iterations): Export overload with loop arg.
22941 (free_numbers_of_iterations_estimates_loop): Use an overload of
22942 free_numbers_of_iterations_estimates instead.
22943 * tree-cfg.c (remove_bb): Adjust.
22944 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
22945 * tree-parloops.c (gen_parallel_loop): Likewise.
22946 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
22948 (tree_unroll_loops_completely): Likewise.
22949 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
22950 Use an overload instead and export.
22951 (estimated_loop_iterations): Adjust.
22952 (max_loop_iterations): Likewise.
22953 (likely_max_loop_iterations): Likewise.
22954 (estimate_numbers_of_iterations): Take struct function as arg
22956 (loop_exits_before_overflow): Adjust.
22957 (free_numbers_of_iterations_estimates_loop): Use an overload.
22958 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
22959 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
22961 2017-06-19 Richard Biener <rguenther@suse.de>
22964 * ipa-prop.c (find_constructor_constant_at_offset): Handle
22965 RANGE_EXPR conservatively.
22967 2017-06-16 Carl Love <cel@us.ibm.com>
22969 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22970 definitions for vec_float, vec_float2, vec_floato,
22971 vec_floate built-ins.
22972 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
22973 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
22975 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
22976 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
22977 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
22978 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
22979 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
22980 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
22981 vec_floato): Add builtin defines.
22982 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
22983 Update the built-in documentation file for the new built-in
22986 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22988 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
22989 (mthumb): Mark as the negative of -marm.
22991 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22993 * doc/invoke.texi (ARM Options, -mcpu): Document supported
22995 (ARM Options, -mtune): Document that this accepts the same
22996 extension options as -mcpu.
22997 (ARM Options, -mfpu): Document addition of -mfpu=auto.
22999 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23001 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
23002 permitted extensions.
23004 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23006 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
23007 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
23008 (armv8-m.main): Add option +nodsp.
23009 * config/arm/arm-cpu-cdata.h: Regenerated.
23011 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23013 * config/arm/t-fuchsia: New file.
23014 * config.gcc (arm*-*-fuchsia*): Use it.
23016 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23018 * config/arm/t-symbian: Rewrite for new option infrastructure.
23020 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23022 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
23023 (MULTILIB_REQUIRED): Likewise.
23025 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23027 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
23028 (MULTILIB_RESUE): Likewise.
23029 (MULTILIB_MATCHES): Likewise.
23030 (MULTLIB_REQUIRED): Likewise.
23032 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23034 * config/arm/t-rtems: Rewrite for new option framework.
23036 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23038 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
23039 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
23040 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
23041 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
23042 * config/arm/t-multilib: ... here.
23043 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
23044 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
23045 armv7-a and armv8*-a when A-profile libraries have not been built.
23046 * config/arm/t-rmprofile: Rewrite.
23048 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23050 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
23051 with a backslash. Remove the backslash after substituting unescaped
23053 * doc/fragments.texi (MULTILIB_REUSE): Document it.
23055 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23057 * config.gcc: (arm*-*-*): When building a-profile libraries, force
23058 the driver to pass through the default setting of -mfloat-abi.
23059 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
23061 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
23062 (all_feat_combs): New rule.
23063 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
23065 * config/arm/t-aprofile: Rewrite.
23067 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23069 * config/arm/arm.h (FPUTYPE_AUTO): Define.
23070 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
23071 fpu is not specified by the user/command-line.
23072 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
23073 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
23074 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
23075 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
23076 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
23077 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
23079 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23081 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
23082 * config/arm/t-arm-elf: Rewritten.
23084 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23086 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
23087 have some floating-point instructions.
23088 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
23089 (TARGET_MAYBE_HARD_FLOAT): New macro.
23090 * config/arm/arm-builtins.c (arm_init_builtins): Use
23091 TARGET_MAYBE_HARD_FLOAT.
23092 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
23094 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23096 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
23097 (configargs.h): Include it.
23098 (arm_print_hint_for_fpu_option): New function.
23099 (arm_parse_fpu_option): New function.
23100 (candidate_extension): New class.
23101 (arm_canon_for_multilib): New function.
23102 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
23103 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
23104 (ARCH_CANONICAL_SPECS): New macro.
23105 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
23107 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23109 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
23110 are set after handling multilib fragments. Set target_cpu_default2
23113 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23115 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
23117 (arm*-*-*): Set target_cpu_default2 to a quoted string.
23118 * config/arm/parsecpu.awk (check_cpu): Validate any extension
23120 (check_arch): Likewise.
23121 * config/arm/arm.c (arm_configure_build_target): Handle
23122 TARGET_CPU_DEFAULT being a string constant. Scan any feature
23123 options in the default.
23125 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23127 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
23128 when an option is an alias of another.
23129 * config/arm/parsecpu.awk (optalias): New parser token.
23130 (gen_comm_data): Mark non-alias options as such. Emit entries
23131 for extension aliases.
23132 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
23133 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
23134 (armv6kz, armv6zk, armv6t2): Likewise.
23135 (armv7): Make vfpv3-d16 an alias.
23136 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
23138 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
23139 Sort in canonical order.
23140 (armv8-a): Sort in canonical order.
23141 (armv8.1-a, armv8.2-a): Likewise.
23142 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
23144 (cortex-a9): Sort in canonical order.
23145 * config/arm/arm.c (selftests.h): Include it.
23146 (arm_test_cpu_arch_data): New function.
23147 (arm_run_self_tests): New function.
23148 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
23149 (targetm): Move declaration to the end of the file.
23150 * arm-cpu-cdata.h: Regenerated.
23152 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23154 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
23155 call to target_mode_check describing the type of option passed.
23156 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
23157 (arm_target_thumb_only): Use arm_parse_arch_option_name or
23158 arm_parse_cpu_option_name to match parameters against list of
23160 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
23161 arm_arch_core_flags data structure.
23162 * config/arm/arm-cpu_cdata.h: Regenerated.
23164 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23166 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
23168 (arm_print_hint_for_cpu_option): Likewise.
23169 (arm_print_hint_for_arch_option): Likewise.
23170 (arm_parse_cpu_option_name): Likewise.
23171 (arm_parse_arch_option_name): Likewise.
23172 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
23173 of entries in the all_fpus list.
23174 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
23175 (arm_parse_cpu_option_name): Declare.
23176 (arm_parse_arch_option_name): Declare.
23177 (arm_parse_option_features): Declare.
23178 (arm_intialize_isa): Declare.
23179 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
23181 (gen_comm_data): ... here. Make definitions non-static.
23182 * config/arm/arm-cpu-data.h: Regenerated.
23183 * config/arm/arm-cpu-cdata.h: Regenerated.
23185 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23187 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
23188 (cpu_arch_extension): New structure.
23189 (cpu_arch_option, arch_option, cpu_option): New structures.
23190 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
23191 architecture types.
23192 (gen_data): Generate new format data tables.
23193 * config/arm/arm.c (cpu_tune): New structure.
23194 (cpu_option, processors): Delete.
23195 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
23196 (arm_print_hint_for_cpu_option): ... this and ...
23197 (arm_print_hint_for_arch_option): ... this.
23198 (arm_parse_arch_cpu_name): Delete. Replace with ...
23199 (arm_parse_cpu_option_name): ... this and ...
23200 (arm_parse_arch_option_name): ... this.
23201 (arm_unrecognized_feature): Change type of target parameter to
23203 (arm_parse_arch_cpu_features): Delete. Replace with ...
23204 (arm_parse_option_features): ... this.
23205 (arm_configure_build_target): Rework to use new configuration data
23207 (arm_print_tune_info): Rework for new configuration data tables.
23208 * config/arm/arm-cpu-data.h: Regenerated.
23209 * config/arm/arm-cpu.h: Regenerated.
23211 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23213 * Makefile.in (OBJS): Move sbitmap.o from here ...
23214 (OBJS-libcommon): ... to here.
23216 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23218 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
23219 (ISA_ALL_CRYPTO): New macro.
23220 (ISA_ALL_SIMD): New macro
23221 (ISA_ALL_FP): New macro.
23222 * config/arm/arm.c (fpu_bitlist): Update initializer.
23223 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
23225 (arm9e): Add fpu. Add option for nofp
23226 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
23227 (arm926ej-s, arm1026ej-s): Likewise.
23228 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
23229 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
23230 neon-fp16, neon-vfpv4, nofp and nosimd.
23231 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
23232 (cortex-a8): Add fpu. Add option for nofp.
23233 (cortex-a9): Add fpu. Add options for nosimd and nofp.
23234 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
23235 (cortex-r4f): Add fpu.
23236 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
23237 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
23239 (cortex-r8): Likewise.
23240 (cortex-m4): Add fpu. Add option for nofp.
23241 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
23242 (cortex-a17.cortex-a7): Likewise.
23243 (cortex-a32): Add fpu. Add options for crypto and nofp.
23244 (cortex-a35, cortex-a53): Likewise.
23245 (cortex-a57): Add fpu. Add option for crypto.
23246 (cortex-a72, cortex-a73): Likewise.
23247 (exynos-m1): Likewise.
23248 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
23249 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
23250 (cortex-m33): Add fpu. Add option for nofp.
23251 * config/arm/arm-cpu-cdata.h: Regenerated
23252 * config/arm/arm-cpu-data.h: Regenerated.
23254 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23256 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
23257 (armv5te, armv5tej): Likewise.
23258 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
23259 (armv7): Add options fp and vfpv3-d16.
23260 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
23261 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
23263 (armv7ve): Likewise.
23264 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
23265 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
23266 (armv8-a): Add nocrypto option.
23267 (armv8.1-a, armv8.2-a): Likewise.
23268 (armv8-m.main): add options fp, fp.dp and nofp.
23270 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23272 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
23274 (armv8-a+crc): Delete.
23275 (armv8.1-a): Add options simd, crypto and nofp.
23276 (armv8.2-a): Add options fp16, simd, crypto and nofp.
23277 (armv8.2-a+fp16): Delete.
23278 (armv8-m.main): Add option dsp.
23279 (armv8-m.main+dsp): Delete.
23280 (cortex-a8): Add fpu. Add nofp option.
23281 (cortex-a9): Add fpu. Add nofp and nosimd options.
23282 * config/arm/parsecpu.awk (gen_data): Generate option tables and
23283 link to main cpu and architecture data structures.
23284 (gen_comm_data): Only put isa attributes from the main architecture
23286 (option): New statement for architecture and CPU entries.
23287 * arm.c (struct cpu_option): New structure.
23288 (struct processors): Add entry for options.
23289 (arm_unrecognized_feature): New function.
23290 (arm_parse_arch_cpu_name): Ignore any characters after the first
23292 (arm_parse_arch_cpu_feature): New function.
23293 (arm_configure_build_target): Separate out any CPU and architecture
23294 features and parse separately. Don't error out if -mfpu=auto is
23295 used with only an architecture string.
23296 (arm_print_asm_arch_directives): New function.
23297 (arm_file_start): Call it.
23298 * config/arm/arm-cpu-cdata.h: Regenerated.
23299 * config/arm/arm-cpu-data.h: Likewise.
23300 * config/arm/arm-tables.opt: Likewise.
23302 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23304 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
23305 assembler when it is not -mfpu=auto.
23307 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23309 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
23310 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
23311 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
23312 (ASM_CPU_SPEC): Rewrite.
23313 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
23314 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
23315 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
23316 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
23317 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
23318 copied string is NUL-terminated. Also strip any characters prefixed
23320 (arm_rewrite_selected_arch): New function.
23321 (arm_rewrite_march): New function.
23323 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23325 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
23326 (x_arm_cpu_string, x_arm_tune_string): Likewise.
23327 (march, mcpu, mtune): Convert to string-based options.
23328 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
23329 (arm_parse_arch_cpu_name): New function.
23330 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
23331 identify selected architecture or CPU.
23332 (arm_option_save): New function.
23333 (TARGET_OPTION_SAVE): Redefine.
23334 (arm_option_restore): Restore string options.
23335 (arm_option_print): Print string options.
23337 2017-06-16 Martin Sebor <msebor@redhat.com>
23339 PR tree-optimization/80933
23340 PR tree-optimization/80934
23341 * builtins.c (fold_builtin_3): Do not handle bcmp here.
23342 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
23343 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
23344 (gimple_fold_builtin): Call them.
23346 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23348 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
23349 as unlikely; update profile.
23351 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23353 * predict.c (force_edge_cold): Handle declaring edges impossible
23356 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23358 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
23360 (try_unroll_loop_completely): Fix reporting.
23362 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23364 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
23366 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
23369 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
23370 if given a non-constant argument for an intrinsic which requires a
23373 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23375 * profile.c (compare_freqs): New function.
23376 (branch_prob): Sort edge list.
23377 (find_spanning_tree): Assume that the list is priority sorted.
23379 2017-06-16 Richard Biener <rguenther@suse.de>
23381 PR tree-optimization/81090
23382 * passes.def (pass_record_bounds): Remove.
23383 * tree-pass.h (make_pass_record_bounds): Likewise.
23384 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
23385 make_pass_record_bounds): Likewise.
23386 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
23387 not free niter estimates at the beginning but at the end.
23388 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
23390 2017-06-16 Richard Biener <rguenther@suse.de>
23392 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
23393 initializer to workaround ICE in host GCC 4.8.
23395 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23397 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
23399 (clone_inlined_nodes): Update.
23401 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23403 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
23404 prefetch settings, and enable prefetching by default at -O3.
23406 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23408 * config/aarch64/aarch64.c (aarch64_override_options_internal):
23409 Set flag_prefetch_loop_arrays according to tuning data.
23411 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23413 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
23414 New tune structure.
23415 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
23416 [Unrelated to main purpose of the patch] Place the pointer field last
23417 to enable type checking errors when tune structure are wrongly merged.
23418 * config/aarch64/aarch64.c (generic_prefetch_tune,)
23419 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
23420 (thunderx2t99_prefetch_tune): New tune constants.
23421 (tune_params *_tunings): Update all tunings (no functional change).
23422 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
23423 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
23424 from tunings structures.
23426 2017-06-16 Jakub Jelinek <jakub@redhat.com>
23429 * ubsan.c (instrument_null): Add T argument, use it instead
23430 of computing it based on IS_LHS.
23431 (instrument_object_size): Likewise.
23432 (pass_ubsan::execute): Adjust instrument_null and
23433 instrument_object_size callers to pass gimple_get_lhs or
23434 gimple_assign_rhs1 result to it. Use instrument_null instead of
23435 calling get_base_address and instrument_mem_ref. Handle
23436 aggregate call arguments for object-size sanitization.
23438 2017-06-16 Yury Gribov <tetra2005@gmail.com>
23440 PR tree-optimization/81089
23441 * tree-vrp.c (is_masked_range_test): Validate operands of
23444 2017-06-15 Martin Sebor <msebor@redhat.com>
23447 * dumpfile.c (dump_register): Avoid calling memset to initialize
23448 a class with a default ctor.
23449 * gcc.c (struct compiler): Remove const qualification.
23450 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
23451 * hash-table.h: Ditto.
23452 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
23454 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
23455 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
23457 * params.h (struct param_info): Make struct members non-const.
23458 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
23459 with default initialization.
23460 * vec.h (vec_copy_construct, vec_default_construct): New helper
23462 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
23463 with vec_copy_construct.
23464 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
23465 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
23466 * doc/invoke.texi (-Wclass-memaccess): Document.
23468 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23470 * emit-rtl.h (is_leaf): Update comment about local
23471 register allocator.
23473 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23476 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
23477 for a variable to have a section before checking if the section has a
23479 Set section to.persistent if persistent attribute is set.
23480 Warn if .persistent attribute is used on an automatic variable.
23482 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
23484 PR rtl-optimization/80474
23485 * reorg.c (update_block): Do not ignore instructions in a delay slot.
23487 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
23489 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
23492 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
23494 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
23495 (casesi): Emit bounds checking as RTL.
23496 (casesi_internal_mips16_<mode>): Remove bounds checking.
23498 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
23500 * config/xtensa/xtensa.c (xtensa_option_override): Append
23501 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
23502 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
23503 xtensa_doloop_hooks): Define unconditionally.
23504 (xtensa_reorg_loops): Only call reorg_loops in the presence of
23506 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
23507 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
23508 for it in xtensa_option_override.
23509 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
23510 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
23512 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
23514 * doc/cppopts.texi: Document '-' special value to -MF.
23516 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
23518 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
23519 (cortex_a53_fconst): Likewise.
23520 (cortex_a53_fpmul): Likewise.
23521 (cortex_a53_f_load_64): Likewise.
23522 (cortex_a53_f_load_many): Likewise.
23523 (cortex_a53_advsimd_alu): Likewise.
23524 (cortex_a53_advsimd_alu_q): Likewise.
23525 (cortex_a53_advsimd_mul): Likewise.
23526 (cortex_a53_advsimd_mul_q): Likewise.
23527 (fpmac bypass): Add new bypass for fpmac-fpmac case.
23528 Add missing fmul, r2f_cvt and fconst cases.
23530 2017-06-14 Richard Biener <rguenther@suse.de>
23532 PR middle-end/81088
23533 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
23535 (fold_binary_loc): When associating do not treat pre-existing
23536 TREE_OVERFLOW on literal constants as a reason to allow
23537 TREE_OVERFLOW on associated literal constants.
23539 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23541 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
23542 (MASK_FEATURES): New macro.
23543 * config/sparc/sparc.c (sparc_option_override): Remove the special
23544 handling of -mfpu and generalize it to all MASK_FEATURES switches.
23546 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23548 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
23549 a division of 0 if non-call exceptions are enabled.
23551 2017-06-14 Andrew Pinski <apinski@cavium.com>
23552 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23555 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
23556 Improve vector initialization code gen for only variable case.
23558 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23560 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
23562 2017-06-14 Richard Biener <rguenther@suse.de>
23564 PR tree-optimization/81083
23565 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
23568 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23570 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
23571 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
23572 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
23573 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
23574 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
23575 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
23577 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23579 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
23580 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
23582 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23584 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
23586 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23588 * config/rs6000/t-rtems: Don't handle SPE.
23590 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23592 * config/rs6000/t-linux: Don't handle SPE.
23594 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23596 * config/rs6000/eabispe.h: Delete file.
23598 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23600 * config/rs6000/t-spe: Delete file.
23602 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23604 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
23605 (rs6000_legitimate_offset_address_p): Return false for anything in
23606 V2SImode or V2SFmode.
23608 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23610 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
23611 except V2SF and V2SI. Rearrange the vector modes, and add comments.
23612 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
23614 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
23615 (rs6000_legitimate_offset_address_p): Ditto.
23616 (rs6000_emit_move): Ditto.
23617 (rs6000_init_builtins): Remove V4HI_type_node.
23619 2017-06-13 Martin Liska <mliska@suse.cz>
23622 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
23623 (gate_asan): Likewise.
23624 * asan.h (asan_no_sanitize_address_p): Remove the function.
23625 (sanitize_flags_p): New function.
23626 * builtins.def: Fix coding style.
23627 * common.opt: Use renamed enum value.
23628 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
23629 * doc/extend.texi: Document no_sanitize attribute.
23630 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
23631 to SANITIZE_UNDEFINED_NONDEFAULT.
23632 * gcc.c (sanitize_spec_function): Use the renamed enum value.
23633 * gimple-fold.c (optimize_atomic_compare_exchange_p):
23634 Use sanitize_flags_p.
23635 * gimplify.c (gimplify_function_tree): Likewise.
23636 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
23637 * opts.c (parse_no_sanitize_attribute): New function.
23638 (common_handle_option): Use renamed enum value.
23639 * opts.h (parse_no_sanitize_attribute): Declare.
23640 * tree.c (sanitize_flags_p): New function.
23641 * tree.h: Declared here.
23642 * tsan.c: Use sanitize_flags_p.
23643 * ubsan.c (ubsan_expand_null_ifn): Likewise.
23644 (instrument_mem_ref): Likewise.
23645 (instrument_bool_enum_load): Likewise.
23646 (do_ubsan_in_current_function): Remove the function.
23647 (pass_ubsan::execute): Use sanitize_flags_p.
23648 * ubsan.h: Remove do_ubsan_in_current_function
23649 * tree-cfg.c (print_no_sanitize_attr_value): New function.
23650 (dump_function_to_file): Use it here.
23652 2017-06-13 Martin Jambor <mjambor@suse.cz>
23654 PR tree-optimization/80803
23655 PR tree-optimization/81063
23656 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
23657 (propagate_subaccesses_across_link): Enqueue subtree whenever
23658 necessary instead of relying on the caller.
23660 2017-06-13 Martin Jambor <mjambor@suse.cz>
23662 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
23663 that have a first_link.
23664 (sort_and_splice_var_accesses): Do not check first_link before
23666 (subtree_mark_written_and_enqueue): Likewise.
23667 (propagate_all_subaccesses): Likewise and do not stop at first
23668 parent with a first_link.
23670 2017-06-13 Martin Jambor <mjambor@suse.cz>
23672 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
23675 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23677 * match.pd: New pattern.
23679 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23681 * tree-vrp.c (is_masked_range_test): New function.
23682 (register_edge_assert_for): Determine ranges for
23685 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23687 PR tree-optimization/67328
23688 * fold-const.c (maskable_range_p): New function.
23689 (build_range_check): Generate bittests if possible.
23691 2017-06-13 Martin Liska <mliska@suse.cz>
23693 * gimple-pretty-print.c (dump_probability): Add new argument.
23694 (dump_edge_probability): Dump both probability and count.
23695 (dump_gimple_label): Likewise.
23696 (dump_gimple_bb_header): Likewise.
23698 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
23701 * config/avr/avr-devices.c: Fix indentation.
23702 * config/avr/gen-avr-mmcu-specs.c: Dito.
23704 2017-06-13 Richard Biener <rguenther@suse.de>
23706 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
23707 instead get vector type from stmt_info.
23708 (vectorizable_reduction): Adjust. Remove dead code.
23710 2017-06-13 Richard Biener <rguenther@suse.de>
23712 PR middle-end/81065
23713 * fold-const.c (extract_muldiv_1): Remove bogus distribution
23714 case of C * (x * C2 + C3).
23715 (fold_addr_of_array_ref_difference): Properly fold index difference.
23717 2017-06-12 David S. Miller <davem@davemloft.net>
23720 * config/sparc/sparc.md (return expander): Emit frame blockage if
23721 function uses alloca.
23723 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
23725 * combine.c (make_field_assignment): Check len rather than the mode
23726 precision when calling force_to_mode.
23728 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
23730 Support multilibs and devices that see flash in RAM address range.
23733 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
23734 (avr_mcu_t) <flash_pm_offset>: New field.
23735 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
23736 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
23737 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
23738 (AVR_TINY_PM_OFFSET): Remove macro.
23739 * config/avr/avr.opt (-mshort-calls): New option.
23740 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23741 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
23742 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
23743 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
23744 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
23745 instead of avr_arch->have_jmp_call.
23746 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
23747 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
23748 avr_arch->flash_pm_offset to define.
23749 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
23750 new field flash_pm_offset. Add entry for avrxmega3.
23751 (avr_texinfo): Add entry for avrxmega3.
23752 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
23753 attiny212, attiny214,
23754 attiny412, attiny414, attiny416, attiny417,
23755 attiny814, attiny816, attiny817,
23756 attiny1614, attiny1616, attiny1617,
23757 attiny3214, attiny3216, attiny3217.
23758 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
23759 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
23760 (avr_print_operand_address) [AVR_TINY]: Same.
23761 (avr_asm_init_sections) <readonly_data_section>: Only patch
23762 callback if avr_arch->flash_pm_offset = 0.
23763 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
23764 for rodata if avr_arch->flash_pm_offset != 0.
23765 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
23766 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
23767 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
23768 (m_options): Append opt_rcall.
23769 (m_dirnames): Append dir_rcall.
23770 * config/avr/t-multilib: Regenerate.
23772 * configure.ac [target=avr]: Check whether avrxmega3 default
23773 linker description file works as needed.
23774 * configure: Regenerate.
23775 * doc/avr-mmcu.texi: Regenerate.
23776 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
23777 <__AVR_ARCH__>: Document avrxmega3 and 103.
23778 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
23779 <__AVR_SHORT_CALLS__>: Document it.
23780 <__AVR_PM_BASE_ADDRESS__>: Document it.
23781 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
23782 (AVR Variable Attributes) <progmem>: Document this is
23783 not needed for avrxmega3.
23784 (AVR Named Address Spaces) <__flash>: Dito.
23786 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
23788 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
23790 2017-06-12 Doug Rupp <rupp@adacore.com>
23792 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
23793 Append vxworks-stdint.h to the tm_file list.
23794 * config/vxworks-stdint.h: New file.
23796 2017-06-12 Martin Liska <mliska@suse.cz>
23798 PR tree-optimization/81041
23799 * tree-profile.c (gimple_gen_ic_func_profiler):
23800 Create an extra BB in profile-generate
23801 (gimple_gen_time_profiler): Likewise.
23803 2017-06-12 Jakub Jelinek <jakub@redhat.com>
23805 PR tree-optimization/81003
23806 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
23807 (update_range_test): Use it instead of force_gimple_operand_gsi.
23809 2017-06-12 Richard Biener <rguenther@suse.de>
23811 PR tree-optimization/81053
23812 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
23813 with backedge value not defined in loop. Simplify def stmt
23816 2017-06-11 Tom de Vries <tom@codesourcery.com>
23819 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
23821 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
23822 nvptx_cannot_force_const_mem.
23824 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23826 * opts.c (finish_options): Move test for flag_split_stack after
23827 it has been initialized.
23829 2017-06-11 Jason Merrill <jason@redhat.com>
23831 * tree.h (id_equal): New.
23832 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
23833 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
23834 instead of strcmp of IDENTIFIER_POINTER.
23836 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23838 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
23839 (mark_all_inlined_calls_cdtor): Fix formating.
23840 (inline_transform): Rescale profile before inlining.
23842 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23844 * cgraph.h (cgraph_edge::clone): Update prototype.
23845 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
23846 (cgraph_node::create_clone): Update.
23847 (cgraph_node::create_version_clone): Update.
23848 * tree-inline.c (copy_bb): Update.
23849 (expand_call_inline): Update.
23851 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
23853 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
23854 factored out from ...
23855 (rs6000_emit_prologue): ... here.
23857 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
23859 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
23860 factored out from ...
23861 (rs6000_emit_prologue): ... here.
23863 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23865 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
23867 (handle_missing_profiles): Fix computation of tp_first_run.
23868 (counts_to_freqs): Do not touch freqs when count is 0.
23870 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23872 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
23875 2017-06-10 Tom de Vries <tom@codesourcery.com>
23877 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23878 attributes): Document signal effective target.
23880 2017-06-10 Tom de Vries <tom@codesourcery.com>
23882 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23883 Document effective target stack_size.
23885 2017-06-09 David Malcolm <dmalcolm@redhat.com>
23887 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
23888 to the edit_context if they can be auto-applied.
23890 2017-06-9 Ian Lance Taylor <iant@golang.org>
23892 * opts.c (finish_options): If -fsplit-stack, disable implicit
23893 -forder-blocks-and-partition.
23894 * doc/invoke.texi (Optimize Options): Document that when using
23895 -fsplit-stack -forder-blocks-and-partition is not implicitly
23898 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23900 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
23901 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
23902 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
23903 * builtins.def (abort, trap, unreachable): Declare cold.
23904 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
23905 * tree-core.h (ECF_COLD): New.
23906 * tree.c (set_call_expr_flags): Handle ECF_COLD.
23907 (build_common_builtin_nodes): Mark unreachable and abort as cold.
23909 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23911 * predict.c (unlikely_executed_stmt_p): Cleanup.
23913 2017-06-09 Richard Biener <rguenther@suse.de>
23915 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
23916 model if the ref is always written to.
23918 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23920 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
23922 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23924 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
23927 2017-06-09 Tom de Vries <tom@codesourcery.com>
23930 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
23931 "target cannot support label values" when encountering LABEL_REF.
23933 2017-06-09 Martin Liska <mliska@suse.cz>
23935 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
23936 (gimple_gen_ic_func_profiler): Emit direct comparison
23937 of __gcov_indirect_call_callee with NULL.
23938 (gimple_gen_time_profiler): Change probability from
23939 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
23941 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23943 * profile.c (edge_gcov_counts): Turn to pointer.
23944 (compute_branch_probabilities, compute_branch_probabilities): Update.
23945 (branch_prob): Do not clear edge_gcov_count.
23946 * profile.h (edge_gcov_counts): Turn to pointer.
23947 (edge_gcov_count): Update.
23949 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23951 * gimple.h (gimple_check_failed): Mark cold.
23953 2017-06-09 Richard Biener <rguenther@suse.de>
23955 PR tree-optimization/66623
23956 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
23957 refactor check_reduction into two parts, properly computing
23958 whether we have to check reduction validity for outer loop
23961 2017-06-09 Richard Biener <rguenther@suse.de>
23963 PR tree-optimization/79483
23964 * graphite-scop-detection.c (order): New global.
23965 (get_order): Compute bb to order mapping that satisfies code
23966 generation constraints.
23967 (cmp_pbbs): New helper.
23968 (build_scops): Start domwalk at entry block, sort generated
23971 2017-06-09 Richard Biener <rguenther@suse.de>
23973 PR middle-end/81007
23974 * ipa-polymorphic-call.c
23975 (ipa_polymorphic_call_context::restrict_to_inner_class):
23976 Skip FIELD_DECLs with error_mark_node type.
23977 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
23980 2017-06-09 Martin Liska <mliska@suse.cz>
23982 * predict.c (struct branch_predictor): New struct.
23983 (test_prediction_value_range): New test.
23984 (predict_c_tests): New function.
23985 * selftest-run-tests.c (selftest::run_tests): Run the function.
23986 * selftest.h: Declare new tests.
23988 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
23991 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
23992 gen_add3_insn did not fail.
23993 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
23994 r0, construct that number in a temporary reg and add that reg to r0.
23995 If asked to put the result in r0 as well, fail.
23997 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
23999 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24000 for early expansion of vec_eqv.
24002 2017-06-08 Jakub Jelinek <jakub@redhat.com>
24004 PR middle-end/81005
24005 * ubsan.c (instrument_null): Avoid pointless code temporary.
24006 (pass_ubsan::execute): Instrument aggregate arguments of calls.
24008 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
24012 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
24015 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
24018 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24020 * predict.c (unlikely_executed_edge_p): Move ahead.
24021 (probably_never_executed_edge_p): Use it.
24023 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24025 PR middle-end/79988
24026 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
24027 gimple_call_builtin_p call.
24029 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24031 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
24032 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
24033 rtl_check_failed_type2, rtl_check_failed_code1,
24034 rtl_check_failed_code2, rtl_check_failed_code_mode,
24035 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
24036 rtvec_check_failed_bounds, rtl_check_failed_flag,
24037 _fatal_insn_not_found, _fatal_insn): Likewise.
24038 * tree.h (tree_contains_struct_check_failed,
24039 tree_check_failed, tree_not_check_failed,
24040 tree_class_check_failed, tree_range_check_failed,
24041 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
24042 tree_vec_elt_check_failed, phi_node_elt_check_failed,
24043 tree_operand_check_failed, omp_clause_check_failed,
24044 omp_clause_operand_check_failed, omp_clause_range_check_failed):
24047 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24049 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
24050 flag_branch_probabilities.
24051 * ipa-inline.c (edge_badness): Likewise.
24052 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
24053 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
24054 * predict.c (maybe_hot_frequency_p): Likewise.
24055 (probably_never_executed): Likewise.
24056 * sched-ebb.c (schedule_ebbs): Likewise.
24057 * sched-rgn.c (find_single_block_region): Likewise.
24058 * tracer.c (tail_duplicate): Likewise.
24060 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24062 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
24063 longer requires x_flag_profile_use.
24065 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24067 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
24068 instead of flag_reorder_blocks_and_partition.
24069 * dbxout.c (dbxout_function_end): Likewise.
24070 * dwarf2out.c (gen_subprogram_die): Likewise.
24071 * haifa-sched.c (sched_create_recovery_edges): Likewise.
24072 * hw-doloop.c (reorg_loops): Likewise.
24073 * varasm.c (assemble_start_function,
24074 assemble_end_function): Likewise.
24075 (decide_function_section): Do not check for
24076 flag_reorder_blocks_and_partition.
24078 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24080 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24082 (chkp_get_hard_register_fake_addr_expr): Ditto.
24083 (chkp_build_addr_expr): Add check for hard reg case.
24084 (chkp_parse_array_and_component_ref): Ditto.
24085 (chkp_find_bounds_1): Ditto.
24086 (chkp_process_stmt): Don't generate bounds store for
24089 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24091 * predict.c (maybe_hot_bb_p): Do not check profile status.
24092 (maybe_hot_edge_p): Likewise.
24093 (probably_never_executed): Check for zero counts even if profile
24095 (unlikely_executed_edge_p): New function.
24096 (unlikely_executed_stmt_p): New function.
24097 (unlikely_executed_bb_p): New function.
24098 (set_even_probabilities): Use unlikely predicates.
24099 (combine_predictions_for_bb): Likewise.
24100 (predict_paths_for_bb): Likewise.
24101 (predict_paths_leading_to_edge): Likewise.
24102 (determine_unlikely_bbs): New function.
24103 (estimate_bb_frequencies): Use it.
24104 (compute_function_frequency): Use zero counts even if profile is
24106 * profile-count.h: Fix typo.
24108 2017-08-08 Julia Koval <julia.koval@intel.com>
24110 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
24111 _mm512_mask_cvtsepi16_storeu_epi8,
24112 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
24113 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
24114 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
24115 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
24116 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
24117 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
24118 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
24119 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
24120 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
24121 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
24122 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
24123 __builtin_ia32_pmovuswb256mem_mask,
24124 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
24125 __builtin_ia32_pmovwb512mem_mask): New builtins.
24127 2017-08-08 Julia Koval <julia.koval@intel.com>
24129 PR target/73350,80862
24130 * config/i386/subst.md (round): Fix round pattern.
24131 * config/i386/i386.c (ix86_erase_embedded_rounding):
24132 Fix erasing rounding for the fixed pattern.
24134 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24136 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
24138 2017-06-08 Martin Liska <mliska@suse.cz>
24140 PR gcov-profile/80911
24141 * gcov.c (block_info::block_info): New constructor.
24143 2017-06-07 Carl Love <cel@us.ibm.com>
24145 * config/rs6000/rs6000-c: The return type of the following
24146 built-in functions was implemented as int not long long. Fix sign
24147 of return value for the unsigned version of vec_mulo and vec_mule.
24148 vector unsigned long long vec_bperm (vector unsigned long long,
24149 vector unsigned char)
24150 vector signed long long vec_mule (vector signed int,
24152 vector unsigned long long vec_mule (vector unsigned int,
24153 vector unsigned int)
24154 vector signed long long vec_mulo (vector signed int,
24156 vector unsigned long long vec_mulo (vector unsigned int,
24157 vector unsigned int)
24158 * doc/extend.texi: Fix the documentation for the built-in
24161 2017-06-07 Carl Love <cel@us.ibm.com>
24164 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
24167 2017-06-07 Carl Love <cel@us.ibm.com>
24169 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
24170 support, Generate doublehv for signed int/float for BE case only.
24172 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
24174 * doc/invoke.texi (mcx16): Rewrite.
24176 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24178 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
24179 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
24180 *mov<mode>_softfloat, and an anonymous splitter): Use
24181 nonimmediate_operand instead of rs6000_nonimmediate_operand.
24183 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24185 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
24187 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
24188 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
24189 (rs6000_debug_reg_global): Adjust.
24190 (rs6000_init_hard_regno_mode_ok): Adjust.
24191 (rs6000_dbx_register_number): Adjust.
24192 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
24193 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
24194 Remove SPE_ACC and SPEFSCR.
24195 (REG_ALLOC_ORDER): Ditto.
24196 (FRAME_POINTER_REGNUM): Change to 111.
24197 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
24198 (REG_CLASS_NAMES): Ditto.
24199 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
24200 (REGISTER_NAMES): Ditto.
24201 (ADDITIONAL_REG_NAMES): Ditto.
24202 (rs6000_reg_names): Ditto.
24203 * config/rs6000/rs6000.md: Renumber some register number
24206 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24208 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
24210 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
24211 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
24213 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
24214 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
24215 Delete the SPE high registers.
24216 (REG_ALLOC_ORDER): Ditto.
24217 (enum reg_class): Remove SPE_HIGH_REGS.
24218 (REG_CLASS_NAMES): Ditto.
24219 (REG_CLASS_CONTENTS): Delete the SPE high registers.
24220 (REGISTER_NAMES): Ditto.
24221 (rs6000_reg_names): Ditto.
24222 * doc/tm.texi.in: Remove SPE as example.
24223 * doc/tm.texi: Regenerate.
24225 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24227 * config/rs6000/8540.md (ppc8540_brinc): Delete.
24228 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
24229 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
24230 * config/rs6000/rs6000.md (type): Remove "brinc".
24232 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24234 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
24235 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
24236 * config/rs6000/linuxspe.h: Delete file.
24237 * config/rs6000/rs6000.md: Don't include spe.md.
24238 * config/rs6000/spe.h: Delete file.
24239 * config/rs6000/spe.md: Delete file.
24240 * config/rs6000/t-rs6000: Remove spe.md.
24242 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24244 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
24245 (reg_or_none500mem_operand): Delete.
24246 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
24247 instead of reg_or_none500mem_operand.
24249 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24251 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
24252 handling of SPE flags.
24253 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
24255 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24257 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
24259 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
24260 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
24261 paired_divv2sf3): Similar.
24262 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
24263 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
24264 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
24266 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
24267 Rename the paired_* instruction patterns.
24268 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
24270 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
24271 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
24272 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
24273 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
24274 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
24275 PAIRED_VECTOR_MODE.
24276 (struct machine_function): Delete field spe_insn_chain_scanned_p.
24277 (spe_func_has_64bit_regs_p): Delete.
24278 (spe_expand_predicate_builtin): Delete.
24279 (spe_expand_evsel_builtin): Delete.
24280 (TARGET_DWARF_REGISTER_SPAN): Do not define.
24281 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
24282 (invalid_e500_subreg): Delete.
24283 (rs6000_legitimize_address): Always force_reg op2 as well, for
24284 paired single memory accesses.
24285 (rs6000_member_type_forces_blk): Delete.
24286 (rs6000_spe_function_arg): Delete.
24287 (rs6000_expand_unop_builtin): Delete SPE handling.
24288 (rs6000_expand_binop_builtin): Ditto.
24289 (spe_expand_stv_builtin): Delete.
24290 (bdesc_2arg_spe): Delete.
24291 (spe_expand_builtin): Delete.
24292 (spe_expand_predicate_builtin): Delete.
24293 (spe_expand_evsel_builtin): Delete.
24294 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
24295 (spe_init_builtins): Delete.
24296 (spe_func_has_64bit_regs_p): Delete.
24297 (savres_routine_name): Delete "info" parameter. Adjust callers.
24298 (rs6000_emit_stack_reset): Ditto.
24299 (rs6000_dwarf_register_span): Delete.
24300 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
24301 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
24302 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
24304 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
24306 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
24307 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
24308 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
24309 mulv2sf3, divv2sf3): Delete expanders.
24311 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24313 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
24315 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24317 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
24318 * config/rs6000/rs6000.c: Ditto.
24320 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24322 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
24323 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
24324 comparison_operator.
24326 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24328 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
24329 * config/rs6000/rs6000.opt: Ditto.
24330 * config/rs6000/t-rtems: Ditto.
24332 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24334 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
24335 TARGET_E500_SINGLE by 0, simplify.
24336 * config/rs6000/rs6000.c: Ditto.
24337 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
24338 (spe_build_register_parallel): Delete.
24339 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
24340 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
24341 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
24342 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
24343 (E500_CONVERT): Delete.
24344 * config/rs6000/spe.md: Remove many patterns and all define_constants.
24346 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24348 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
24349 * config/rs6000/dfp.md: Ditto.
24350 (negdd2, *negdd2_fpr): Merge.
24351 (absdd2, *absdd2_fpr): Merge.
24352 (negtd2, *negtd2_fpr): Merge.
24353 (abstd2, *abstd2_fpr): Merge.
24354 * config/rs6000/e500.h: Delete file.
24355 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
24356 TARGET_FPRS by 1 and simplify.
24357 * config/rs6000/rs6000-c.c: Ditto.
24358 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
24359 TARGET_DF_SPE by 0.
24360 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
24362 * config/rs6000/rs6000.md: Ditto.
24363 (floatdidf2, *floatdidf2_fpr): Merge.
24364 (move_from_CR_gt_bit): Delete.
24365 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
24366 (E500_CR_IOR_COMPARE): Delete.
24367 (All patterns that require !TARGET_FPRS): Delete.
24368 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
24370 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24372 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
24374 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24376 * graphds.c (add_edge): Intitialize edge's attached data.
24377 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
24378 pointer parameter. Call pointed function on each edge during
24379 graph traversing. Skip traversing the edge when the function
24381 (graphds_dfs, graphds_scc): Ditto.
24382 (for_each_edge): New parameter. Pass the new parameter to callback
24384 * graphds.h (skip_edge_callback): New function pointer type.
24385 (graphds_dfs, graphds_scc): New function pointer parameter.
24386 (graphds_edge_callback, for_each_edge): New parameter.
24388 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24390 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
24391 out code checking if runtime alias check is possible to below ...
24392 Call the new function.
24393 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
24394 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
24396 2017-06-07 Marek Polacek <polacek@redhat.com>
24399 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
24400 TYPE_OVERFLOW_WRAPS check.
24402 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24404 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
24405 if versioning is required.
24406 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
24407 peeling with the check for versioning.
24409 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24411 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
24412 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
24413 Set true to new parameter if new ssa variable is defined.
24414 (vect_gen_vector_loop_niters): Refactor. Set range information
24415 for the new vector loop bound variable.
24416 (vect_do_peeling): Ditto.
24418 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24420 * tree-affine.c (ssa.h): Include header file.
24421 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
24422 has wrapping overflow behavior.
24424 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24426 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
24428 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24430 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
24431 (tree_to_aff_combination): ... here.
24433 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24435 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
24436 reg_pressure model function.
24437 (ivopts_global_cost_for_size): Delete.
24438 (determine_set_costs, iv_ca_recount_cost): Call new model function
24439 ivopts_estimate_reg_pressure.
24441 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24443 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
24444 expensive than udiv. Remove floating point cases from mod.
24446 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24448 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
24449 Increase idiv cost.
24451 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24453 * config/aarch64/aarch64.md
24454 (copysignsf3): Fix mask generation.
24456 2017-06-07 Jakub Jelinek <jakub@redhat.com>
24458 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
24460 (class dump_manager): Add register_dumps method.
24461 * dumpfile.c: Include langhooks.h.
24462 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
24463 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
24464 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
24465 (dump_manager::dump_register): Start with 512 entries instead of 32.
24466 (dump_manager::register_dumps): New method.
24467 * toplev.c (general_init): Instead of invoking register_dumps
24468 langhook, invoke register_dumps method on the dump manager.
24469 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
24472 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
24474 * doc/md.texi: Clarify the restrictions on a define_insn condition.
24475 Say that # requires an associated define_split to exist, and that
24476 the define_split must be suitable for use after register allocation.
24478 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24480 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
24481 (compute_outgoing_frequencies): Also initialize zero counts.
24482 (find_many_sub_basic_blocks): Do not produce uninitialized profile
24483 around loops; preserve more of profile when nothing changes.
24485 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
24487 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
24489 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
24490 * config/arm/arm-cpu-cdata.h: Regenerate.
24491 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
24492 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
24493 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
24494 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
24495 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
24497 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
24499 * config/arm/t-rmprofile: Likewise.
24500 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
24502 2017-06-06 David S. Miller <davem@davemloft.net>
24505 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
24506 blockage if function uses alloca.
24508 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24510 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
24511 New "uid" fields to hold pretty-print IDs of group and ref.
24512 Memory references are now identified as <group_id>:<ref_id>
24513 instead of using [random] addresses.
24514 (dump_mem_details): Simplify, no functional change.
24515 (dump_mem_ref): Simplify and make output more concise.
24516 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
24517 (find_or_create_group): Initialize group uid.
24518 (record_ref): Initialize ref uid. Improve debug output.
24519 (prune_group_by_reuse, should_issue_prefetch_p,)
24520 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
24521 (mark_nontemporal_store, determine_loop_nest_reuse):
24522 Improve debug output.
24524 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24526 * dbgcnt.def (prefetch): New debug counter.
24527 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
24528 (schedule_prefetches): Stop issueing prefetches if debug counter
24531 2017-06-06 Tom de Vries <tom@codesourcery.com>
24533 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
24534 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
24536 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24538 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
24539 Use aarch64_reg_or_zero predicate for operand 4.
24540 (aarch64_compare_and_swap<mode> define_insn_and_split):
24541 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
24542 (aarch64_store_exclusive<mode>): Likewise for operand 2.
24544 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24546 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
24547 (arm_compute_save_core_reg_mask): This.
24548 (thumb1_compute_save_reg_mask): Rename into ...
24549 (thumb1_compute_save_core_reg_mask): This.
24550 (arm_compute_save_reg0_reg12_mask): Adapt comment.
24551 (arm_compute_frame_layout): Likewise.
24553 2017-06-06 Richard Biener <rguenther@suse.de>
24555 PR tree-optimization/80974
24556 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
24557 keep or clear leaders SSA info.
24559 2017-06-06 Tom de Vries <tom@codesourcery.com>
24561 * config/nvptx/nvptx.c (split_mode_p): New function.
24562 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
24564 2017-06-06 Tom de Vries <tom@codesourcery.com>
24566 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
24568 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24571 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
24574 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24576 * shrink-wrap.c (handle_simple_exit): Update profile.
24577 (try_shrink_wrapping): Upate profile.
24579 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24581 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
24582 (tree_guess_outgoing_edge_probabilities): New.
24583 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
24584 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
24586 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24588 * ipa-split.c (split_function): Initialize return bb profile.
24590 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24592 * profile.c (compute_branch_probabilities): Also initialize
24593 EXIT_BLOCK profile.
24595 2017-06-06 Richard Biener <rguenther@suse.de>
24597 PR tree-optimization/80928
24598 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
24599 (vect_analyze_loop_operations): Properly guard analysis for
24601 (vect_transform_loop): Likewise.
24602 (vect_analyze_loop_2): Also reset SLP type on PHIs.
24603 (vect_model_induction_cost): Do not cost for pure SLP.
24604 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
24605 of induction in inner loop vectorization.
24606 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
24607 (vect_get_and_check_slp_defs): Handle vect_induction_def.
24608 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
24610 (vect_analyze_slp_cost_1): Cost induction.
24611 (vect_detect_hybrid_slp_stmts): Handle PHIs.
24612 (vect_get_slp_vect_defs): Likewise.
24613 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
24614 (vect_transform_stmt): Handle SLP reductions.
24615 * tree-vectorizer.h (vectorizable_induction): Adjust.
24617 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24619 * config/rs6000/rs6000.c (make_resolver_func): Update
24620 init_lowered_empty_function call.
24622 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
24624 * doc/invoke.texi: Document the -fprofile-abs-path option.
24625 * common.opt (fprofile-abs-path): New option.
24626 * gcov-io.h (gcov_write_filename): Declare.
24627 * gcov-io.c (gcov_write_filename): New function.
24628 * coverage.c (coverage_begin_function): Use gcov_write_filename.
24629 * profile.c (output_location): Likewise.
24631 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24633 * shring-wrap.c: Revert accidental commit.
24635 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
24637 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
24639 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24641 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
24643 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
24644 profile in callgraph edge.
24645 * profile-count.h (apply_probability): If THIS is 0, then result is 0
24646 (apply_scale): Likewise.
24647 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
24648 Also scale profile when inlining function with zero profile.
24649 (initialize_cfun): Update exit block profile even when it is zero.
24650 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
24651 when profile is read.
24653 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24655 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
24656 (CLONE_*): New constants to define the processors we can generate
24657 code for with the target_clone attribute.
24658 (rs6000_clone_map): New array to identify which clone processors
24659 the current program is running on.
24660 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
24661 target_clone attribute.
24662 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
24663 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
24664 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
24665 (cpu_expand_builtin): Add support for target_clone attribute.
24666 (rs6000_valid_attribute_p): Allow "default" attribute.
24667 (get_decl_name): New debug function to simplify printing the
24668 current function name in debugging statements.
24669 (rs6000_clone_priority): New functions to support the target_clone
24670 attribute, and be able to generate code to switch between ISA 2.05
24671 through ISA 3.0 (power6 through power9).
24672 (rs6000_compare_version_priority): Likewise.
24673 (rs6000_get_function_versions_dispatcher): Likewise.
24674 (make_resolver_func): Likewise.
24675 (add_condition_to_bb): Likewise.
24676 (dispatch_function_versions): Likewise.
24677 (rs6000_generate_version_dispatcher_body): Likewise.
24678 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
24679 (fusion_gpr_load_p): Fix a spacing issue.
24680 * doc/extend.texi (Common Function Attributes): Document that the
24681 PowerPC supports the target_clone attribute.
24683 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24685 * config/arm/arm.h: explain F symbol found in description of ARM
24686 register allocation in its legend.
24688 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24690 * config/mips/frame-header-opt.c: Include profile-count.h.
24691 * config/riscv/riscv.c: Include profile-count.h
24693 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24695 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
24697 (sm_set_flag_if_changed): Add bbs field.
24698 (execute_sm_if_changed_flag_set): Pass BBS.
24699 (execute_sm): Update.
24701 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24703 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
24706 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24708 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
24709 (peephole2): New peephole2 to emit the above.
24710 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
24712 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24714 * config/aarch64/aarch64.c (define_peephole2 above
24715 *sub_<shift>_<mode>): New peephole.
24717 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24719 * config/i386/i386.c (make_resolver_func): Update.
24720 * Makefile.in: Add profile-count.h and profile-count.o
24721 * auto-profile.c (afdo_indirect_call): Update to new API.
24722 (afdo_set_bb_count): Update.
24723 (afdo_propagate_edge): Update.
24724 (afdo_propagate_circuit): Update.
24725 (afdo_calculate_branch_prob): Update.
24726 (afdo_annotate_cfg): Update.
24727 * basic-block.h: Include profile-count.h
24728 (struct edge_def): Turn count to profile_count.
24729 (struct basic_block_def): Likewie.
24730 (REG_BR_PROB_BASE): Move to profile-count.h
24731 (RDIV): Move to profile-count.h
24732 * bb-reorder.c (max_entry_count): Turn to profile_count.
24733 (find_traces): Update.
24734 (rotate_loop):Update.
24735 (connect_traces):Update.
24736 (sanitize_hot_paths):Update.
24737 * bt-load.c (migrate_btr_defs): Update.
24738 * cfg.c (RDIV): Remove.
24739 (init_flow): Use alloc_block.
24740 (alloc_block): Uninitialize count.
24741 (unchecked_make_edge): Uninitialize count.
24742 (check_bb_profile): Update.
24743 (dump_edge_info): Update.
24744 (dump_bb_info): Update.
24745 (update_bb_profile_for_threading): Update.
24746 (scale_bbs_frequencies_int): Update.
24747 (scale_bbs_frequencies_gcov_type): Update.
24748 (scale_bbs_frequencies_profile_count): New.
24749 * cfg.h (update_bb_profile_for_threading): Update.
24750 (scale_bbs_frequencies_profile_count): Declare.
24751 * cfgbuild.c (compute_outgoing_frequencies): Update.
24752 (find_many_sub_basic_blocks): Update.
24753 * cfgcleanup.c (try_forward_edges): Update.
24754 (try_crossjump_to_edge): Update.
24755 * cfgexpand.c (expand_gimple_tailcall): Update.
24756 (construct_exit_block): Update.
24757 * cfghooks.c (verify_flow_info): Update.
24758 (dump_bb_for_graph): Update.
24759 (split_edge): Update.
24760 (make_forwarder_block): Update.
24761 (duplicate_block): Update.
24762 (account_profile_record): Update.
24763 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
24764 (get_estimated_loop_iterations): Update.
24765 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
24766 (single_likely_exit): Update.
24767 * cfgloopmanip.c (scale_loop_profile): Update.
24769 (set_zero_probability): Update.
24770 (lv_adjust_loop_entry_edge): Update.
24771 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24772 (purge_dead_edges): Update.
24773 (rtl_account_profile_record): Update.
24774 * cgraph.c (cgraph_node::create): Uninitialize count.
24775 (symbol_table::create_edge): Uninitialize count.
24776 (cgraph_update_edges_for_call_stmt_node): Update.
24777 (cgraph_edge::dump_edge_flags): Update.
24778 (cgraph_node::dump): Update.
24779 (cgraph_edge::maybe_hot_p): Update.
24780 * cgraph.h: Include profile-count.h
24781 (create_clone), create_edge, create_indirect_edge): Update.
24782 (cgraph_node): Turn count to profile_count.
24783 (cgraph_edge0: Likewise.
24784 (make_speculative, clone): Update.
24785 (create_edge): Update.
24786 (init_lowered_empty_function): Update.
24787 * cgraphclones.c (cgraph_edge::clone): Update.
24788 (duplicate_thunk_for_node): Update.
24789 (cgraph_node::create_clone): Update.
24790 * cgraphunit.c (cgraph_node::analyze): Update.
24791 (cgraph_node::expand_thunk): Update.
24792 * final.c (dump_basic_block_info): Update.
24793 * gimple-streamer-in.c (input_bb): Update.
24794 * gimple-streamer-out.c (output_bb): Update.
24795 * graphite.c (print_global_statistics): Update.
24796 (print_graphite_scop_statistics): Update.
24797 * hsa-brig.c: Include basic-block.h.
24798 * hsa-dump.c: Include basic-block.h.
24799 * hsa-gen.c (T sum_slice): Update.
24800 (convert_switch_statements):Update.
24801 * hsa-regalloc.c: Include basic-block.h.
24802 * ipa-chkp.c (chkp_produce_thunks): Update.
24803 * ipa-cp.c (struct caller_statistics): Update.
24804 (init_caller_stats): Update.
24805 (gather_caller_stats): Update.
24806 (ipcp_cloning_candidate_p): Update.
24807 (good_cloning_opportunity_p): Update.
24808 (get_info_about_necessary_edges): Update.
24809 (dump_profile_updates): Update.
24810 (update_profiling_info): Update.
24811 (update_specialized_profile): Update.
24812 (perhaps_add_new_callers): Update.
24813 (decide_about_value): Update.
24814 (ipa_cp_c_finalize): Update.
24815 * ipa-devirt.c (struct odr_type_warn_count): Update.
24816 (struct decl_warn_count): Update.
24817 (struct final_warning_record): Update.
24818 (possible_polymorphic_call_targets): Update.
24819 (ipa_devirt): Update.
24820 * ipa-fnsummary.c (redirect_to_unreachable): Update.
24821 * ipa-icf.c (sem_function::merge): Update.
24822 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
24823 * ipa-inline.c (compute_uninlined_call_time): Update.
24824 (compute_inlined_call_time): Update.
24825 (want_inline_small_function_p): Update.
24826 (want_inline_self_recursive_call_p): Update.
24827 (edge_badness): Update.
24828 (lookup_recursive_calls): Update.
24829 (recursive_inlining): Update.
24830 (inline_small_functions): Update.
24831 (dump_overall_stats): Update.
24832 (dump_inline_stats): Update.
24833 * ipa-profile.c (ipa_profile_generate_summary): Update.
24834 (ipa_propagate_frequency): Update.
24835 (ipa_profile): Update.
24836 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24837 * ipa-utils.c (ipa_merge_profiles): Update.
24838 * loop-doloop.c (doloop_modify): Update.
24839 * loop-unroll.c (report_unroll): Update.
24840 (unroll_loop_runtime_iterations): Update.
24841 * lto-cgraph.c (lto_output_edge): Update.
24842 (lto_output_node): Update.
24843 (input_node): Update.
24844 (input_edge): Update.
24845 (merge_profile_summaries): Update.
24846 * lto-streamer-in.c (input_cfg): Update.
24847 * lto-streamer-out.c (output_cfg): Update.
24848 * mcf.c (create_fixup_graph): Update.
24849 (adjust_cfg_counts): Update.
24850 (sum_edge_counts): Update.
24851 * modulo-sched.c (sms_schedule): Update.
24852 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
24853 * predict.c (maybe_hot_count_p): Update.
24854 (probably_never_executed): Update.
24855 (dump_prediction): Update.
24856 (combine_predictions_for_bb): Update.
24857 (propagate_freq): Update.
24858 (handle_missing_profiles): Update.
24859 (counts_to_freqs): Update.
24860 (rebuild_frequencies): Update.
24861 (force_edge_cold): Update.
24862 * predict.h: Include profile-count.h
24863 (maybe_hot_count_p, counts_to_freqs): UPdate.
24864 * print-rtl-function.c: Do not include cfg.h
24865 * print-rtl.c: Include basic-block.h
24866 * profile-count.c: New file.
24867 * profile-count.h: New file.
24868 * profile.c (is_edge_inconsistent): Update.
24869 (correct_negative_edge_counts): Update.
24870 (is_inconsistent): Update.
24871 (set_bb_counts): Update.
24872 (read_profile_edge_counts): Update.
24873 (compute_frequency_overlap): Update.
24874 (compute_branch_probabilities): Update; Initialize and deinitialize
24876 (branch_prob): Update.
24877 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
24878 (edge_gcov_count): New.
24879 (bb_gcov_count): New.
24880 * shrink-wrap.c (try_shrink_wrapping): Update.
24881 * tracer.c (better_p): Update.
24882 * trans-mem.c (expand_transaction): Update.
24883 (ipa_tm_insert_irr_call): Update.
24884 (ipa_tm_insert_gettmclone_call): Update.
24885 * tree-call-cdce.c: Update.
24886 * tree-cfg.c (gimple_duplicate_sese_region): Update.
24887 (gimple_duplicate_sese_tail): Update.
24888 (gimple_account_profile_record): Update.
24889 (execute_fixup_cfg): Update.
24890 * tree-inline.c (copy_bb): Update.
24891 (copy_edges_for_bb): Update.
24892 (initialize_cfun): Update.
24893 (freqs_to_counts): Update.
24894 (copy_cfg_body): Update.
24895 (expand_call_inline): Update.
24896 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24897 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
24898 (try_unroll_loop_completely): Update.
24899 (try_peel_loop): Update.
24900 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24901 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
24902 * tree-ssa-loop-split.c (connect_loops): Update.
24903 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
24904 * tree-ssa-reassoc.c (branch_fixup): Update.
24905 * tree-ssa-tail-merge.c (replace_block_by): Update.
24906 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
24907 (compute_path_counts): Update.
24908 (update_profile): Update.
24909 (recompute_probabilities): Update.
24910 (update_joiner_offpath_counts): Update.
24911 (estimated_freqs_path): Update.
24912 (freqs_to_counts_path): Update.
24913 (clear_counts_path): Update.
24914 (ssa_fix_duplicate_block_edges): Update.
24915 (duplicate_thread_path): Update.
24916 * tree-switch-conversion.c (case_bit_test_cmp): Update.
24917 (struct switch_conv_info): Update.
24918 * tree-tailcall.c (decrease_profile): Update.
24919 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24920 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24921 * value-prof.c (check_counter): Update.
24922 (gimple_divmod_fixed_value): Update.
24923 (gimple_mod_pow2): Update.
24924 (gimple_mod_subtract): Update.
24925 (gimple_ic_transform): Update.
24926 (gimple_stringop_fixed_value): Update.
24927 * value-prof.h (gimple_ic): Update.
24929 2017-06-02 Carl Love <cel@us.ibm.com>
24931 * config/rs6000/rs6000-c: Add support for built-in functions
24932 vector double vec_doublee (vector signed int);
24933 vector double vec_doublee (vector unsigned int);
24934 vector double vec_doublee (vector float);
24935 vector double vec_doubleh (vector signed int);
24936 vector double vec_doubleh (vector unsigned int);
24937 vector double vec_doubleh (vector float);
24938 vector double vec_doublel (vector signed int);
24939 vector double vec_doublel (vector unsigned int);
24940 vector double vec_doublel (vector float);
24941 vector double vec_doubleo (vector signed int);
24942 vector double vec_doubleo (vector unsigned int);
24943 vector double vec_doubleo (vector float);.
24944 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
24945 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
24947 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
24948 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
24949 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
24951 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
24952 vec_doublel, vec_doubleh.
24953 * doc/extend.texi: Update the built-in documentation file for the
24954 new built-in functions.
24956 2017-06-02 David Malcolm <dmalcolm@redhat.com>
24959 * ipa-inline-analysis.c (free_growth_caches): Set
24960 edge_removal_hook_holder to NULL after removing it.
24962 2017-06-02 Sudakshina Das <sudi.das@arm.com>
24964 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
24965 comparision with zero.
24967 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
24968 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24969 for early expansion of vec_min and vec_max builtins.
24970 (builtin_function_type): Add min/max unsigned variants to those
24971 identified as having unsigned arguments.
24973 2017-06-02 Olivier Hainque <hainque@adacore.com>
24975 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
24977 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24979 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
24980 Use VALL_F16 iterator rather than VALL.
24982 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24984 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
24985 Emit CBNZ inside loop when doing a strong exchange and comparing
24986 against zero. Generate the CC flags after the loop.
24988 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
24990 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
24991 (dl_section_ref): New.
24992 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
24993 On AIX, append an expression to subtract the size of the
24994 section length to dl_section_ref.
24996 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
24998 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24999 for early expansion of vector absolute builtins.
25001 2017-06-02 Richard Biener <rguenther@suse.de>
25003 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
25004 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
25006 2017-06-02 Richard Biener <rguenther@suse.de>
25008 PR tree-optimization/80948
25009 * tree-tailcall.c (find_tail_calls): Track stmts to move in
25010 stmt order as well.
25012 2017-06-02 Richard Biener <rguenther@suse.de>
25014 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
25016 * tree-vect-stmts.c (process_use): Do not mark backedge defs
25017 for inductions as relevant.
25019 2017-06-02 Richard Biener <rguenther@suse.de>
25021 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
25022 (vectorizable_induction): ... this. Remove dead code.
25024 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
25026 * builtins. (expand_builtin_alloca): Remove second parameter and
25027 infer its value from the first parameter instead.
25028 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
25030 2017-06-02 Jakub Jelinek <jakub@redhat.com>
25032 PR rtl-optimization/80903
25033 * loop-doloop.c (add_test): Unshare sequence.
25035 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
25037 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
25039 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
25041 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
25043 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
25044 xlogue_layout::get_instance, logue_layout::xlogue_layout,
25045 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
25046 (xlogue_layout::get_stub_rtx): Make static.
25047 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
25048 (xlogue_layout::compute_stub_managed_regs): Rename to...
25049 (xlogue_layout::count_stub_managed_regs): ...this.
25050 (xlogue_layout::is_stub_managed_reg): New function.
25051 (xlogue_layout::m_stub_names): Rename to...
25052 (xlogue_layout::s_stub_names): ...this, make static.
25053 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
25054 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
25055 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
25056 xlogue_layout::s_stub_names): Instantiate statics.
25057 (stub_managed_regs): Remove.
25058 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
25059 (disable_call_ms2sysv_xlogues): Rename to...
25060 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
25061 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
25063 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
25064 change after reload_completed.
25065 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
25067 (ix86_expand_prologue): Likewise.
25068 (ix86_expand_epilogue): Likewise.
25069 (ix86_expand_split_stack_prologue): Likewise.
25070 (ix86_compute_frame_layout): Remove frame parameter ...
25071 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
25072 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
25074 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
25075 (ix86_frame): Move from here ...
25076 * config/i386/i386.h (ix86_frame): ... to here.
25077 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
25078 complete ix86_frame data structure instead. Remove some_ld_name.
25080 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
25082 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
25083 symbols that hold a DECL_VALUE_EXPR.
25085 2017-06-01 Martin Jambor <mjambor@suse.cz>
25087 PR tree-optimization/80898
25088 * tree-sra.c (process_subtree_disqualification): Removed.
25089 (disqualify_candidate): Do not acll
25090 process_subtree_disqualification.
25091 (subtree_mark_written_and_enqueue): New function.
25092 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
25093 RHS has been disqualified and re-queue LHS if necessary. Apart
25094 from that, ignore disqualified RHS.
25096 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25098 * config/s390/s390.c (s390_emit_epilogue): Disable early return
25099 address fetch for z10 or later.
25101 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25103 * config/arc/arc.md (tst_movb): Add guard when splitting.
25105 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25107 * config/arc/arc.c (arc_can_eliminate): Test against
25108 arc_frame_pointer_needed.
25110 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25112 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
25113 to prevent store reordering.
25114 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
25115 (type): Add block type.
25116 (stack_tie): Define special instruction to be used in
25119 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25121 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
25122 constraint. It is not valid for the pattern.
25123 (noncommutative_binary_comparison): Likewise.
25125 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25127 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
25130 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25132 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
25133 be used by the reg-alloc.
25135 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25137 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
25138 reg-alloc when having mul64 or mul32x16 instructions.
25139 (mulsidi3): Likewise.
25140 (umulsidi3): Likewise.
25141 (mulsi32x16): New pattern.
25142 (mulsi64): Likewise.
25143 (mulsidi64): Likewise.
25144 (umulsidi64): Likewise.
25145 (MUL32x16_REG): Define.
25146 (mul64_600): Use MUL32x16_REG.
25147 (mac64_600): Likewise.
25148 (umul64_600): Likewise.
25149 (umac64_600): Likewise.
25151 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25153 * config/arc/arc.md (mulsi3_700): Make it commutative.
25155 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
25157 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
25159 (*sign_extendsidi2_insn): Likewise for movstosw.
25161 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
25163 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
25164 the type of the input discriminant value. Convert the
25165 discriminant value of signedness vary.
25167 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
25169 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
25170 Add to -Wall section.
25172 2017-06-01 Richard Biener <rguenther@suse.de>
25174 PR middle-end/66313
25175 * fold-const.c (fold_plusminus_mult_expr): If the factored
25176 factor may be zero use a wrapping type for the inner operation.
25177 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
25178 and handle moved defs.
25179 (process_assignment): Properly guard the unary op case. Return a
25180 tri-state indicating that moving the stmt before the call may allow
25181 to continue. Pass through to_move.
25182 (find_tail_calls): Handle moving unrelated defs before
25185 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
25188 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
25189 splitter result in the canonical way.
25191 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
25193 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
25194 also for 32bit target. Update insn attributes.
25195 (zero-extendsidi2 splitter): Allow all registers for operand 1.
25197 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
25199 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
25200 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
25201 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
25202 (_mm_maskz_min_ss): New intrinsics.
25204 2017-05-31 Martin Liska <mliska@suse.cz>
25206 * tree-vect-loop.c (vect_create_epilog_for_reduction):
25207 Change comment style to one we normally use.
25208 (vectorizable_reduction): Likewise.
25209 (vectorizable_induction): Likewise.
25210 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
25211 (vectorizable_call): Likewise.
25212 (vectorizable_simd_clone_call): Likewise.
25213 (vectorizable_conversion): Likewise.
25214 (vectorizable_assignment): Likewise.
25215 (vectorizable_shift): Likewise.
25216 (vectorizable_operation): Likewise.
25217 (vectorizable_store): Likewise.
25218 (vectorizable_load): Likewise.
25219 * tree-vectorizer.h: Likewise.
25221 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
25223 * passes.c (emergency_dump_function): New.
25224 * tree-pass.h (emergency_dump_function): Declare.
25225 * plugin.c (plugins_internal_error_function): Remove.
25226 * plugin.h (plugins_internal_error_function): Remove declaration.
25227 * toplev.c (internal_error_function): New static function. Use it...
25228 (general_init): ...here.
25230 2017-05-31 Graham Markall <graham.markall@embecosm.com>
25232 * config/arc/arc.c (arc_print_operand): Handle constant operands.
25233 (arc_rtx_costs): Add costs for new patterns.
25234 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
25235 * config/arc/predicates.md: Add _1_2_3_operand predicate.
25237 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
25239 * tree-ssa-strlen.c (get_next_strinfo): New function.
25240 (get_stridx_plus_constant): Use it.
25241 (zero_length_string): Likewise.
25242 (adjust_related_strinfos): Likewise.
25243 (adjust_last_stmt): Likewise.
25245 2017-05-31 Richard Biener <rguenther@suse.de>
25248 * config/i386/i386.c (ix86_expand_builtin): Remove assert
25249 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
25251 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
25253 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
25254 loop_vinfo argument and use of dependence distance vectors.
25255 Check instead whether the two references differ only in their
25256 initial value and assume that they have the same alignment if the
25257 difference is a multiple of the vector alignment.
25258 (vect_analyze_data_refs_alignment): Update call accordingly.
25260 2017-05-31 Martin Liska <mliska@suse.cz>
25263 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
25265 2017-05-31 Bin Cheng <bin.cheng@arm.com>
25267 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
25268 (create_intersect_range_checks): Move from ...
25269 * tree-data-ref.c (create_intersect_range_checks_index)
25270 (create_intersect_range_checks): ... to here.
25271 (create_runtime_alias_checks): New function factored from ...
25272 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
25273 here. Call above function.
25274 * tree-data-ref.h (create_runtime_alias_checks): New function.
25276 2017-05-31 Bin Cheng <bin.cheng@arm.com>
25278 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
25279 segment length for dr_b and compute it in wide_int.
25281 2017-05-31 Richard Biener <rguenther@suse.de>
25283 PR tree-optimization/80906
25284 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
25285 and pass through iv_map.
25286 (copy_bb_and_scalar_dependences): Adjust.
25287 (translate_pending_phi_nodes): Likewise.
25288 (copy_loop_close_phi_args): Handle code-generating IVs instead
25291 2017-05-30 David Malcolm <dmalcolm@redhat.com>
25293 * diagnostic-color.c (color_dict): Add "type-diff".
25294 (parse_gcc_colors): Update comment.
25295 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
25296 -fdiagnostics-show-template-tree and -fno-elide-type.
25297 (GCC_COLORS): Add type-diff to example.
25299 (-fdiagnostics-show-template-tree): New.
25300 (-fno-elide-type): New.
25301 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
25302 the pp_format_decoder callback. Call any m_format_postprocessor's
25304 (pretty_printer::pretty_printer): Initialize
25305 m_format_postprocessor.
25306 (pretty_printer::~pretty_printer): Delete any
25307 m_format_postprocessor.
25308 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
25309 (class format_postprocessor): New class.
25310 (struct pretty_printer::format_decoder): Document the new parameters.
25311 (struct pretty_printer::m_format_postprocessor): New field.
25312 * tree-diagnostic.c (default_tree_printer): Update for new
25313 bool and const char ** params.
25314 * tree-diagnostic.h (default_tree_printer): Likewise.
25316 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
25318 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
25319 (lwa_operand): Delete rs6000_gen_cell_microcode test.
25320 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25321 rs6000_gen_cell_microcode code.
25322 (rs6000_final_prescan_insn): Delete.
25323 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
25324 "warn-cell-microcode" entries.
25325 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
25326 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
25327 throughout. Change cc_reg_not_micro_cr0_operand to
25328 cc_reg_not_cr0_operand throughout.
25329 (*extendhi<mode>2_noload): Delete.
25330 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
25331 (mwarn-cell-microcode): Delete.
25332 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
25333 -mgen-cell-microcode and -mwarn-cell-microcode.
25335 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
25338 * config/i386/constraints.md (Yd): New constraint.
25340 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
25341 and (?Yd, r) alternatives. Update insn attributes.
25342 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
25343 and (?*Yd, r) alternatives. Update insn attributes.
25344 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
25346 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
25348 * gimplify.c (gimplify_modify_expr): Don't create a
25349 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
25352 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
25354 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
25356 2017-05-30 Richard Biener <rguenther@suse.de>
25358 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
25359 and reduc_def fields.
25360 (STMT_VINFO_REDUC_TYPE): New define.
25361 (STMT_VINFO_REDUC_DEF): Likewise.
25362 (vect_force_simple_reduction): Adjust prototype.
25363 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
25364 (vect_is_simple_reduction): Remove check_reduction argument.
25365 (vect_force_simple_reduction): Adjust and set
25366 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25367 (vectorizable_reduction): Do not re-do reduction analysis
25368 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25369 * tree-parloops.c (gather_scalar_reductions): Adjust.
25371 2017-05-30 Richard Biener <rguenther@suse.de>
25373 PR middle-end/80901
25374 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
25377 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
25379 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
25380 Introduce unknown_misalignment parameter and remove vf.
25381 (vect_peeling_hash_get_lowest_cost):
25382 Pass unknown_misalignment parameter.
25383 (vect_enhance_data_refs_alignment):
25384 Fix unsupportable data ref treatment.
25386 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25388 * tree-vect-data-refs.c (vect_get_data_access_cost):
25389 Workaround for SLP handling.
25390 (vect_enhance_data_refs_alignment):
25391 Compute costs for doing no peeling at all, compare to the best
25392 peeling costs so far and avoid peeling if cheaper.
25394 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25396 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
25397 Return peeling info and set costs to zero for unlimited cost
25399 (vect_enhance_data_refs_alignment): Also inspect all datarefs
25400 with unknown misalignment. Compute and costs for unknown
25401 misalignment, compare them to the costs for known misalignment
25402 and choose the cheapest for peeling.
25404 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25406 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
25407 (vect_get_peeling_costs_all_drs): Create function.
25408 (vect_peeling_hash_get_lowest_cost):
25409 Use vect_get_peeling_costs_all_drs.
25410 (vect_peeling_supportable): Create function.
25411 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
25413 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25415 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
25416 DR_HAS_NEGATIVE_STEP.
25417 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
25418 (vect_enhance_data_refs_alignment): Use.
25419 (vect_duplicate_ssa_name_ptr_info): Use.
25420 * tree-vectorizer.h (dr_misalignment): Use.
25421 (known_alignment_for_access_p): Use.
25423 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25426 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
25428 (has_section_name): New function.
25430 2017-05-30 Martin Liska <mliska@suse.cz>
25433 * auto-profile.c (get_function_decl_from_block): Fix
25436 2017-05-30 Richard Biener <rguenther@suse.de>
25438 PR middle-end/80876
25439 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
25441 2017-05-30 Martin Liska <mliska@suse.cz>
25443 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
25444 * dumpfile.h (struct dump_file_info): Remove ctors.
25446 2017-05-30 Martin Liska <mliska@suse.cz>
25448 * predict.def: Fix GNU coding style.
25450 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25452 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25453 Mark 'to' argument with ATTRIBUTE_UNUSED.
25455 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25457 * config/xtensa/xtensa.c (xtensa_emit_call): Use
25458 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
25459 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
25462 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
25464 * doc/install.texi (Options specification): Restore entry of
25465 --enable-sjlj-exceptions.
25467 2017-05-27 Michael Eager <eager@eagercon.com>
25470 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
25472 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
25474 * config/microblaze/microblaze.h
25475 (FIXED_REGISTERS): Update in macro.
25476 (CALL_USED_REGISTERS): Update in macro.
25478 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
25480 * doc/install.texi: Add links to macOS binary distributions.
25482 2017-05-27 Jakub Jelinek <jakub@redhat.com>
25486 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25488 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25490 2017-05-26 Martin Liska <mliska@suse.cz>
25492 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
25494 2017-05-26 Martin Liska <mliska@suse.cz>
25496 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
25497 always leading ';; '.
25498 (dump_bb_info): Likewise.
25499 (brief_dump_cfg): Likewise.
25500 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
25501 * dumpfile.c: Remove usage of TDF_VERBOSE.
25502 * dumpfile.h (enum dump_kind): Likewise.
25503 (dump_gimple_bb_header): Do not use TDF_COMMENT.
25504 * print-tree.c (debug_verbose): Remove.
25505 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
25506 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
25507 * tree-diagnostic.c (default_tree_printer): Replace
25508 TDF_DIAGNOSTIC with TDF_SLIM.
25510 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25512 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
25513 in parameter loop, rather than loop_vinfo.
25514 (create_intersect_range_checks): Ditto.
25515 (vect_create_cond_for_alias_checks): Update call to above functions.
25517 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25519 PR tree-optimization/80815
25520 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
25521 for merging runtime alias checks. Handle negative DR_STEPs.
25523 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25525 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
25527 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
25528 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
25529 out code pruning runtime alias checks.
25530 * tree-data-ref.c (prune_runtime_alias_test_list): New function
25531 factored out from above.
25532 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25534 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25536 (prune_runtime_alias_test_list): New decalaration.
25538 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25540 * tree-vect-data-refs.c (compare_tree): Rename and move ...
25541 * tree-data-ref.c (data_ref_compare_tree): ... to here.
25542 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
25543 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
25544 (operator==, comp_dr_with_seg_len_pair): Ditto.
25545 (vect_prune_runtime_alias_test_list): Ditto.
25547 2017-05-26 Martin Liska <mliska@suse.cz>
25550 * params.def: Bound partial-inlining-entry-probability param.
25552 2017-05-26 Marek Polacek <polacek@redhat.com>
25555 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
25558 2017-05-26 Richard Biener <rguenther@suse.de>
25560 PR tree-optimization/80842
25561 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
25564 2017-05-26 Richard Biener <rguenther@suse.de>
25566 PR tree-optimization/80844
25567 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
25569 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
25571 * doc/md.texi (Machine Constraints): Update x86 family
25572 machine constraints section to match 'config/i386/constraints.md'.
25574 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
25576 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
25578 2017-05-25 Nathan Sidwell <nathan@acm.org>
25580 * doc/invoke.texi (--enable-languages): Update documentation.
25582 2017-05-25 Martin Liska <mliska@suse.cz>
25584 * dumpfile.c: Add TDF_FOLDING.
25585 * dumpfile.h (enum dump_kind): Likewise.
25586 * genmatch.c (dt_simplify::gen_1): Use it.
25588 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25590 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
25592 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25594 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25595 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
25597 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25599 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
25600 * match.pd (X == C): Rewrite it here.
25601 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
25602 with_certain_nonzero_bits2): New predicates.
25603 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
25605 2017-05-24 Nathan Sidwell <nathan@acm.org>
25607 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
25610 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
25613 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
25615 * config/powerpcspe: New port. Files are copied from the rs6000
25616 port, with "rs6000" in filenames replaced by "powerpcspe".
25618 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
25620 PR rtl-optimization/80754
25621 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
25623 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
25625 * config/sparc/sparc.md (length): Return the correct value for -mflat
25626 sibcalls to match output_sibcall.
25628 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
25632 * config/rs6000/rs6000.c (struct machine_function): Add new field
25634 (rs6000_get_separate_components): Init that field, use it.
25635 (rs6000_components_for_bb): Use the field.
25637 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
25639 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
25641 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
25643 PR middle-end/80823
25644 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
25646 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25649 * config/s390/s390.c (s390_check_qrst_address): Check incoming
25650 address against address_operand predicate.
25651 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
25653 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
25655 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
25656 parameters passed indirectly.
25658 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
25660 * config/i386/i386.md (*movdi_internal): Remove SSE4
25661 alternative 18 (?r, *v). Update insn attributes.
25662 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
25663 Update insn attributes.
25664 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
25665 Update insn attributes.
25666 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
25667 alternative 1 (r, v). Remove isa attribute.
25668 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
25669 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
25670 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
25672 2017-05-23 Tom de Vries <tom@codesourcery.com>
25674 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
25677 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25679 * cgraphunit.c (symbol_table::process_new_functions): Update.
25680 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
25681 (inline_generate_summary): Rename to ...
25682 (ipa_fn_summary_generate): ... this one.
25683 (inline_read_summary): Rename to ...
25684 (ipa_fn_summary_read): ... this one.
25685 (inline_write_summary): Rename to ...
25686 (ipa_fn_summary_write): ... this one.
25687 (inline_free_summary): Rename to ...
25688 (ipa_free_fn_summary): ... this one.
25689 (pass_data_local_fn_summary, pass_local_fn_summary,
25690 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
25691 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
25692 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
25693 make_pass_ipa_fn_summary): New.
25694 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
25695 inline_write_summary, inline_free_summary): Remove.
25696 (ipa_free_fn_summary) : New.
25697 * ipa-inline.c (ipa_inline): Update.
25698 (pass_ipa_inline): Do not generate summaries.
25699 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
25701 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
25702 and add pass_ipa_fn_summary.
25703 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
25705 (make_pass_inline_parameters): Remove.
25707 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
25709 * omp-low.c (struct omp_context): Remove "default_kind" member.
25712 * omp-offload.c (execute_oacc_device_lower): Remove the
25713 parallelism dimensions function attributes for unparallelized
25714 OpenACC kernels constructs.
25716 2017-05-23 Martin Liska <mliska@suse.cz>
25718 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
25720 (cgraph_edge::make_speculative): Likewise.
25721 (cgraph_edge::resolve_speculation): Likewise.
25722 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25723 (cgraph_node::dump): Likewise.
25724 * cgraph.h: Likewise.
25725 * cgraphunit.c (analyze_functions): Likewise.
25726 (symbol_table::compile): Likewise.
25727 * ipa-cp.c (print_all_lattices): Likewise.
25728 (determine_versionability): Likewise.
25729 (initialize_node_lattices): Likewise.
25730 (ipcp_verify_propagated_values): Likewise.
25731 (estimate_local_effects): Likewise.
25732 (update_profiling_info): Likewise.
25733 (create_specialized_node): Likewise.
25734 (perhaps_add_new_callers): Likewise.
25735 (decide_about_value): Likewise.
25736 (decide_whether_version_node): Likewise.
25737 (identify_dead_nodes): Likewise.
25738 (ipcp_store_bits_results): Likewise.
25739 * ipa-devirt.c (dump_targets): Likewise.
25740 (ipa_devirt): Likewise.
25741 * ipa-icf.c (sem_item::dump): Likewise.
25742 (sem_function::equals): Likewise.
25743 (sem_variable::equals): Likewise.
25744 (sem_item_optimizer::read_section): Likewise.
25745 (sem_item_optimizer::execute): Likewise.
25746 (congruence_class::dump): Likewise.
25747 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
25748 (dump_inline_summary): Likewise.
25749 (estimate_node_size_and_time): Likewise.
25750 (inline_analyze_function): Likewise.
25751 * ipa-inline-transform.c (inline_call): Likewise.
25752 * ipa-inline.c (report_inline_failed_reason): Likewise.
25753 (want_early_inline_function_p): Likewise.
25754 (edge_badness): Likewise.
25755 (update_edge_key): Likewise.
25756 (inline_small_functions): Likewise.
25757 * ipa-profile.c (ipa_profile): Likewise.
25758 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
25759 (ipa_make_edge_direct_to_target): Likewise.
25760 (remove_described_reference): Likewise.
25761 (ipa_impossible_devirt_target): Likewise.
25762 (propagate_controlled_uses): Likewise.
25763 (ipa_print_node_params): Likewise.
25764 (ipcp_transform_function): Likewise.
25765 * ipa-pure-const.c (pure_const_read_summary): Likewise.
25766 (propagate_pure_const): Likewise.
25767 * ipa-reference.c (generate_summary): Likewise.
25768 (read_write_all_from_decl): Likewise.
25769 (propagate): Likewise.
25770 (ipa_reference_read_optimization_summary): Likewise.
25771 * ipa-utils.c (ipa_merge_profiles): Likewise.
25772 * ipa.c (walk_polymorphic_call_targets): Likewise.
25773 (symbol_table::remove_unreachable_nodes): Likewise.
25774 (ipa_single_use): Likewise.
25775 * passes.c (execute_todo): Likewise.
25776 * predict.c (drop_profile): Likewise.
25777 * symtab.c (symtab_node::get_dump_name): New function.
25778 (symtab_node::dump_name): Likewise.
25779 (symtab_node::dump_asm_name): Likewise.
25780 (symtab_node::dump_references): Likewise.
25781 (symtab_node::dump_referring): Likewise.
25782 (symtab_node::dump_base): Likewise.
25783 (symtab_node::debug_symtab): Likewise.
25784 * tree-sra.c (convert_callers_for_node): Likewise.
25785 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
25786 * value-prof.c (init_node_map): Likewise.
25788 2017-05-23 Martin Liska <mliska@suse.cz>
25790 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
25791 and symtab_node::debug_symtab to symbol_table::debug.
25792 * cgraphunit.c (analyze_functions): Use the renamed function.
25793 (symbol_table::compile): Likewise.
25794 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
25795 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
25796 * passes.c (execute_todo): Likewise.
25797 * symtab.c (symbol_table::dump): New function.
25798 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
25800 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25802 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
25803 that nonconst implies exec.
25805 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25807 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
25808 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
25809 (inline_edge_summary_vec): Turn into ...
25810 (ipa_call_summaries): ... this one.
25811 (redirect_to_unreachable, edge_set_predicate,
25812 evaluate_properties_for_edge, inline_summary_alloc,
25813 reset_ipa_call_summary, reset_inline_summary,
25814 inline_summary_t::duplicate): Update.
25815 (inline_edge_duplication_hook): Turn to ...
25816 (ipa_call_summary_t::duplicate): ... this one.
25817 (inline_edge_removal_hook): Turn to ...
25818 (ipa_call_summary_t::remove): ... this one.
25819 (dump_inline_edge_summary): Turn to ...
25820 (dump_ipa_call_summary): ... this one.
25821 (estimate_function_body_sizes): Update.
25822 (inline_update_callee_summaries): Update.
25823 (remap_edge_change_prob): Update.
25824 (remap_edge_summaries): Update.
25825 (inline_merge_summary): Update.
25826 (do_estimate_edge_time): Update.
25827 (inline_generate_summary): Update.
25828 (inline_read_section): Update.
25829 (inline_read_summary): Update.
25830 (inline_free_summary): Update.
25831 * ipa-inline.c (can_inline_edge_p): Update.
25832 (compute_inlined_call_time): Update.
25833 (want_inline_small_function_p): Update.
25834 (edge_badness): Update.
25835 (early_inliner): Update.
25836 * ipa-inline.h (inline_edge_summary): Turn to ...
25837 (ipa_call_summary): ... this one.
25838 (ipa_call_summary_t): New class.
25839 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
25840 (ipa_call_summaries): New.
25841 (inline_edge_summary): Remove.
25842 (estimate_edge_growth): Update.
25843 * ipa-profile.c (ipa_propagate_frequency_1): Update.
25844 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25845 * ipa-split.c (execute_split_functions): Update.
25846 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
25848 2017-05-23 Tom de Vries <tom@codesourcery.com>
25850 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
25851 attributes): Document rdrand effective target.
25853 2017-05-23 Tom de Vries <tom@codesourcery.com>
25855 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
25856 attributes): Sort alphabetically.
25858 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
25860 * config/avr/genmultilib.awk: Use gsub instead of gensub.
25862 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
25865 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
25866 V2DF/V2DI splat into two separate patterns, one that handles
25867 registers, and the other that only handles memory. Drop support
25868 for splatting from a GPR on ISA 2.07 and then splitting the
25869 splat into direct move and splat.
25870 (vsx_splat_<mode>_reg): Likewise.
25871 (vsx_splat_<mode>_mem): Likewise.
25873 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
25875 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
25877 2017-05-22 Jakub Jelinek <jakub@redhat.com>
25879 PR middle-end/80809
25880 * omp-low.c (finish_taskreg_remap): New function.
25881 (finish_taskreg_scan): If unit size of ctx->record_type
25882 is non-constant, unshare the size expression and replace
25883 decls in it with possible outer var refs.
25885 PR middle-end/80809
25886 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
25887 GOVD_SHARED rather than GOVD_PRIVATE with it.
25888 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
25889 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
25891 PR middle-end/80853
25892 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
25893 as last argument to build_outer_var_ref for pointer bases of array
25894 section reductions.
25896 2017-05-19 Martin Sebor <msebor@redhat.com>
25898 * print-tree.c (print_node): Print DECL_READ_P flag.
25900 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25902 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
25903 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
25904 * cgraph.c: Likewise.
25905 * cgraphunit.c: Likewise.
25906 * gengtype.c: Likewise.
25907 * ipa-cp.c: Likewise.
25908 * ipa-devirt.c: Likewise.
25909 * ipa-icf.c: Likewise.
25910 * ipa-predicate.c: Likewise.
25911 * ipa-profile.c: Likewise.
25912 * ipa-prop.c: Likewise.
25913 * ipa-split.c: Likewise.
25915 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
25916 edge_predicate_pool, dump_inline_hints,
25917 inline_summary::account_size_time, redirect_to_unreachable,
25918 edge_set_predicate, set_hint_predicate,
25919 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
25920 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
25921 inline_summary_t::remove, remap_hint_predicate_after_duplication,
25922 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
25923 ipa_call_summary_t::remove, initialize_growth_caches,
25924 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
25925 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
25926 mark_modified, unmodified_parm_1, unmodified_parm,
25927 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
25928 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
25929 compute_bb_predicates, will_be_nonconstant_expr_predicate,
25930 will_be_nonconstant_predicate, record_modified_bb_info,
25931 get_minimal_bb, record_modified, param_change_prob,
25932 phi_result_unknown_predicate, predicate_for_phi_result,
25933 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
25934 estimate_function_body_sizes, compute_inline_parameters,
25935 compute_inline_parameters_for_curren, pass_data_inline_parameters,
25936 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
25937 inline_update_callee_summaries, remap_edge_change_prob,
25938 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
25939 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
25940 inline_analyze_function, inline_summary_t::insert,
25941 inline_generate_summary, read_ipa_call_summary, inline_read_section,
25942 inline_read_summary, write_ipa_call_summary, inline_write_summary,
25943 inline_free_summary): Move to ipa-fnsummary.h
25944 (predicate_t): Remove.
25945 * ipa-fnsummary.c: New file.
25946 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
25947 (enum inline_hints_vals, inline_hints, agg_position_info,
25948 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
25949 inline_summaries, ipa_call_summary, ipa_call_summary_t,
25950 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
25951 dump_inline_summary, dump_inline_hints, inline_generate_summary,
25952 inline_read_summary, inline_write_summary, inline_free_summary,
25953 inline_analyze_function, initialize_inline_failed,
25954 inline_merge_summary, inline_update_overall_summary,
25955 compute_inline_parameters): Move to ipa-fnsummary.h
25956 * ipa-fnsummary.h: New file.
25957 * ipa-inline-transform.h: Include ipa-inline.h.
25958 * ipa-inline.c: LIkewise.
25960 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25962 * ipa-inline.c (edge_badness): Use inlined_time instead of
25963 inline_summaries->get.
25965 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25967 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
25969 2017-05-22 Nathan Sidwell <nathan@acm.org>
25971 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
25972 (fdump-lang): Document 'raw' option.
25973 * dumpfile.h (TDI_tu): Delete.
25974 * dumpfile.c (dump_files): Remove translation-unit.
25975 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
25977 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
25979 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
25980 command option from $(AWK) call.
25981 * config/avr/genmultilib.awk: Simplify and rewrite so that it
25982 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
25983 [FORMAT]: Remove handling of variable.
25984 * config/avr/t-multilib: Regenerate.
25986 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25988 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
25990 (dump_inline_summary): Do not print self_time.
25991 (estimate_function_body_sizes): Do not set self_time.
25992 (compute_inline_parameters): Likewise.
25993 (inline_read_section, inline_write_summary): Do not stream self_time.
25994 * ipa-inline.h (inline_summary): Drop self_time.
25996 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25998 * ipa-inline-analysis.c (account_size_time): Rename to ...
25999 (inline_summary::account_size_time): ... this one.
26000 (reset_ipa_call_summary): Turn to ...
26001 (ipa_call_summary::reset): ... this one.
26002 (reset_inline_summary): Turn to ...
26003 (inline_summary::reset): ... this one.
26004 (inline_summary_t::remove): Update.
26005 (inline_summary_t::duplicate): Update.
26006 (ipa_call_summary_t::remove): Update.
26007 (dump_inline_summary): Update.
26008 (estimate_function_body_sizes): Update.
26009 (compute_inline_parameters): Update.
26010 (estimate_node_size_and_time): Update.
26011 (inline_merge_summary): Update.
26012 (inline_update_overall_summary): Update.
26013 (inline_read_section): Update.
26014 (inline_write_summary): Update.
26015 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
26016 add account_size_time and reset member functions.
26017 (ipa_call_summary): Add reset function.
26018 * ipa-predicate.h (predicate::operator &): Constify.
26020 2017-05-22 Richard Biener <rguenther@suse.de>
26022 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
26024 2017-05-19 Jason Merrill <jason@redhat.com>
26026 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
26028 2017-05-19 Marek Polacek <polacek@redhat.com>
26031 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
26032 TYPE_OVERFLOW_WRAPS checks.
26034 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
26036 * tree-core.h (enum omp_clause_default_kind): Add
26037 "OMP_CLAUSE_DEFAULT_PRESENT".
26038 * tree-pretty-print.c (dump_omp_clause): Handle it.
26039 * gimplify.c (enum gimplify_omp_var_data): Add
26040 "GOVD_MAP_FORCE_PRESENT".
26041 (gimplify_adjust_omp_clauses_1): Map it to
26042 "GOMP_MAP_FORCE_PRESENT".
26043 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
26045 * gimplify.c (oacc_default_clause): Clarify.
26047 2017-05-19 Nathan Sidwell <nathan@acm.org>
26049 LANG_HOOK_REGISTER_DUMPS
26050 * toplev.c (general_init): Call register dump lang hook.
26051 * doc/invoke.texi: Document -fdump-lang option family.
26052 * dumpfile.c (dump_files): Remove class dump here.
26053 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
26054 * dumpfile.h (tree_dump_index): Remove TDI_class.
26055 * langhooks-def.h (lhd_register_dumps): Declare.
26056 (LANG_HOOKS_REGISTER_DUMPS): Define.
26057 (LANG_HOOKS_INITIALIZER): Add it.
26058 * langhooks.c (lhd_register_dumps): Define.
26059 * langhooks.h (struct lang_hooks): Add register_dumps.
26061 2017-05-19 Nathan Sidwell <nathan@acm.org>
26063 * context.h (context::set_passes): New.
26064 * context.c (context::context): Do not create pass manager.
26065 * toplev.c (general_init): Create pass manager here.
26067 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
26069 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
26070 use this splitter if two add or or instructions would also work for
26071 the constant we want to generate.
26073 2017-05-19 Richard Biener <rguenther@suse.de>
26076 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
26077 predicate evaluation.
26079 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
26081 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
26083 * ipa-inline.c (want_inline_small_function_p): Do not cast to
26086 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
26088 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26089 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26090 (inline_edge_summary_vec): Turn into ...
26091 (ipa_call_summaries): ... this one.
26092 (redirect_to_unreachable, edge_set_predicate,
26093 evaluate_properties_for_edge, inline_summary_alloc,
26094 reset_ipa_call_summary, reset_inline_summary,
26095 inline_summary_t::duplicate): Update.
26096 (inline_edge_duplication_hook): Turn to ...
26097 (ipa_call_summary_t::duplicate): ... this one.
26098 (inline_edge_removal_hook): Turn to ...
26099 (ipa_call_summary_t::remove): ... this one.
26100 (dump_inline_edge_summary): Turn to ...
26101 (dump_ipa_call_summary): ... this one.
26102 (estimate_function_body_sizes): Update.
26103 (inline_update_callee_summaries): Update.
26104 (remap_edge_change_prob): Update.
26105 (remap_edge_summaries): Update.
26106 (inline_merge_summary): Update.
26107 (do_estimate_edge_time): Update.
26108 (inline_generate_summary): Update.
26109 (inline_read_section): Update.
26110 (inline_read_summary): Update.
26111 (inline_free_summary): Update.
26112 * ipa-inline.c (can_inline_edge_p): Update.
26113 (compute_inlined_call_time): Update.
26114 (want_inline_small_function_p): Update.
26115 (edge_badness): Update.
26116 (early_inliner): Update.
26117 * ipa-inline.h (inline_edge_summary): Turn to ...
26118 (ipa_call_summary): ... this one.
26119 (ipa_call_summary_t): New class.
26120 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26121 (ipa_call_summaries): New.
26122 (inline_edge_summary): Remove.
26123 (estimate_edge_growth): Update.
26124 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26125 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26126 * ipa-split.c (execute_split_functions): Update.
26127 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26129 2017-05-19 Richard Biener <rguenther@suse.de>
26131 PR middle-end/80764
26132 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
26134 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
26136 * config/rs6000/rs6000.c (struct machine_function): Add field
26137 fpr_is_wrapped_separately.
26138 (rs6000_get_separate_components): Use 64 components. Handle the
26139 new FPR components.
26140 (rs6000_components_for_bb): Handle the FPR components.
26141 (rs6000_emit_prologue_components): Handle the FPR components.
26142 (rs6000_emit_epilogue_components): Handle the FPR components.
26143 (rs6000_set_handled_components): Handle the FPR components.
26144 (rs6000_emit_prologue): Don't output prologue code for those FPRs
26145 that are already separately shrink-wrapped.
26146 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
26147 that are already separately shrink-wrapped.
26149 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26152 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
26155 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
26156 (define_peephole2 for Altivec d-form load): Add peepholes to catch
26157 cases where the register allocator uses a move and an offsettable
26158 memory operation to/from a FPR register on ISA 2.06/2.07.
26159 (define_peephole2 for Altivec d-form store): Likewise.
26161 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
26164 * config/i386/mmx.md (*mov<mode>_internal): Enable
26165 alternatives 11, 12, 13 and 14 also for 32bit targets.
26166 Remove alternatives 15, 16, 17 and 18.
26167 * config/i386/sse.md (vec_concatv2di): Change
26168 alternative (!x, *y) to (x, ?!*Yn).
26170 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
26172 * dumpfile.h (enum dump_kind): Remove stray comma.
26174 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
26176 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
26177 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
26178 predicate::num_conditions
26179 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
26180 (CHANGED): turn into predicate::changed.
26181 (agg_position_info): Move to ipa-predicate.h
26182 (add_condition, predicate::add_clause, predicate::operator &=,
26183 predicate::or_with, predicate::evaluate, predicate::probability,
26184 dump_condition, dump_clause, predicate::dump,
26185 predicate::remap_after_duplication, predicate::remap_after_inlining,
26186 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
26187 (evaluate_conditions_for_known_args): Update.
26188 (set_cond_stmt_execution_predicate): Update.
26189 * ipa-inline.h: Include ipa-predicate.h
26190 (condition, inline_param_summary, conditions, agg_position_info,
26191 predicate): Move to ipa-predicate.h
26192 * ipa-predicate.c: New file.
26193 * ipa-predicate.h: New file.
26195 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
26197 * final.c (leaf_function_p): Check we are not in a sequence.
26199 2017-05-18 Martin Liska <mliska@suse.cz>
26201 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
26202 * dumpfile.c (dump_register): Use new enum dump_kind.
26203 (get_dump_file_name): Likewise.
26204 (dump_enable_all): Likewise.
26205 (dump_switch_p_1): Likewise.
26206 (enable_rtl_dump_file): Remove usage of TDF_RTL.
26207 * dumpfile.h (enum dump_kind): New enum type.
26208 (struct dump_file_info): Create constructor and
26209 format fields and comments.
26210 * passes.c (pass_manager::register_one_dump_file):
26212 * statistics.c (statistics_early_init): Likewise.
26213 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
26214 TDF_TREE with TDF_SLIM.
26215 (gather_memory_references_ref): Likewise.
26217 2017-05-18 Martin Liska <mliska@suse.cz>
26219 * vec.h (struct vnull): Use it.
26221 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
26223 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
26224 (true_predicate, false_predicate, true_predicate_p,
26225 false_predicate_p): Remove.
26226 (single_cond_predicate, not_inlined_predicate): Turn to member function
26228 (add_condition): Update.
26229 (add_clause): Turn to...
26230 (predicate::add_clause): ... this one; update; allow passing NULL
26232 (and_predicates): Turn to ...
26233 (predicate::operator &=): ... this one.
26234 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
26235 (or_predicates): Turn to ...
26236 (predicate::or_with): ... this one.
26237 (evaluate_predicate): Turn to ...
26238 (predicate::evaluate): ... this one.
26239 (predicate_probability): Turn to ...
26240 (predicate::probability): ... this one.
26241 (dump_condition): Update.
26242 (dump_predicate): Turn to ...
26243 (predicate::dump): ... this one.
26244 (account_size_time): Update.
26245 (edge_set_predicate): Update.
26246 (set_hint_predicate): UPdate.
26247 (evaluate_conditions_for_known_args): Update.
26248 (evaluate_properties_for_edge): Update.
26249 (remap_predicate_after_duplication): Turn to...
26250 (predicate::remap_after_duplication): ... this one.
26251 (remap_hint_predicate_after_duplication): Update.
26252 (inline_summary_t::duplicate): UPdate.
26253 (dump_inline_edge_summary): Update.
26254 (dump_inline_summary): Update.
26255 (set_cond_stmt_execution_predicate): Update.
26256 (set_switch_stmt_execution_predicate): Update.
26257 (compute_bb_predicates): Update.
26258 (will_be_nonconstant_expr_predicate): Update.
26259 (will_be_nonconstant_predicate): Update.
26260 (phi_result_unknown_predicate): Update.
26261 (predicate_for_phi_result): Update.
26262 (array_index_predicate): Update.
26263 (estimate_function_body_sizes): Update.
26264 (estimate_node_size_and_time): Update.
26265 (estimate_ipcp_clone_size_and_time): Update.
26266 (remap_predicate): Rename to ...
26267 (predicate::remap_after_inlining): ... this one.
26268 (remap_hint_predicate): Update.
26269 (inline_merge_summary): Update.
26270 (inline_update_overall_summary): Update.
26271 (estimate_size_after_inlining): Update.
26272 (read_predicate): Rename to ...
26273 (predicate::stream_in): ... this one.
26274 (read_inline_edge_summary): Update.
26275 (write_predicate): Rename to ...
26276 (predicate::stream_out): ... this one.
26277 (write_inline_edge_summary): Update.
26278 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
26279 (clause_t): Turn to uint32_t
26280 (predicate): Turn to class; implement constructor and operators
26282 (size_time_entry): Update.
26283 (inline_summary): Update.
26284 (inline_edge_summary): Update.
26286 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
26288 * fold-const.c (fold_binary_loc): Move transformation...
26289 * match.pd (C - X CMP X): ... here.
26291 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
26293 * config/sparc/sparc.c (sparc_option_override): Set function
26294 alignment for -mcpu=niagara7 to 64 to match the I$ line.
26295 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
26297 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
26299 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
26301 2017-05-18 Marek Polacek <polacek@redhat.com>
26304 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
26305 (pass_ubsan::execute): Call gimple_assign_single_p instead of
26306 gimple_assign_load_p.
26308 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
26310 PR middle-end/80692
26311 * real.c (do_compare): Give decimal_do_compare preference over
26312 comparing just the signs.
26314 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
26316 * doc/md.texi (Canonicalization of Instructions): Describe the
26317 canonical form of instructions that inherently set a condition
26320 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
26322 PR middle-end/80775
26323 * tree-cfg.c: Move deletion of unreachable case statements to after
26324 the merging of consecutive case labels.
26326 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
26328 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
26329 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
26330 restoring of callee-saved registers.
26332 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
26334 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
26335 * config/visium/visium.c (single_set_and_flags): Likewise.
26336 * config/visium/visium.md (Substitutions): Likewise.
26338 2017-05-17 Martin Liska <mliska@suse.cz>
26340 * cfg.c: Introduce dump_flags_t type and
26341 use it instead of int type.
26343 * cfghooks.c: Likewise.
26344 * cfghooks.h (struct cfg_hooks): Likewise.
26345 * cfgrtl.c: Likewise.
26346 * cfgrtl.h: Likewise.
26347 * cgraph.c (cgraph_node::get_body): Likewise.
26348 * coretypes.h: Likewise.
26349 * domwalk.c: Likewise.
26350 * domwalk.h: Likewise.
26351 * dumpfile.c (struct dump_option_value_info): Likewise.
26352 (dump_enable_all): Likewise.
26353 (dump_switch_p_1): Likewise.
26354 (opt_info_switch_p): Likewise.
26355 * dumpfile.h (enum tree_dump_index): Likewise.
26356 (struct dump_file_info): Likewise.
26357 * genemit.c: Likewise.
26358 * generic-match-head.c: Likewise.
26359 * gengtype.c (open_base_files): Likewise.
26360 * gimple-pretty-print.c: Likewise.
26361 * gimple-pretty-print.h: Likewise.
26362 * graph.c (print_graph_cfg): Likewise.
26363 * graphite-scop-detection.c (dot_all_sese): Likewise.
26364 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
26365 * loop-unroll.c (report_unroll): Likewise.
26366 * passes.c (pass_manager::register_one_dump_file): Likewise.
26367 * print-tree.c: Likewise.
26368 * statistics.c: Likewise.
26369 * tree-cfg.c: Likewise.
26370 * tree-cfg.h: Likewise.
26371 * tree-dfa.c: Likewise.
26372 * tree-dfa.h: Likewise.
26373 * tree-dump.c (dump_function): Likewise.
26374 * tree-dump.h (struct dump_info): Likewise.
26375 * tree-pretty-print.c: Likewise.
26376 * tree-pretty-print.h: Likewise.
26377 * tree-ssa-live.c: Likewise.
26378 * tree-ssa-live.h: Likewise.
26379 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
26380 * tree-vect-loop.c: Likewise.
26381 * tree-vect-slp.c: Likewise.
26383 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
26384 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26386 PR tree-optimization/80457
26387 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
26388 of all arguments to a statement as scalar_to_vec operations.
26389 (vectorizable_call): Adjust call to vect_model_simple_cost for
26391 (vectorizable_conversion): Likewise.
26392 (vectorizable_assignment): Likewise.
26393 (vectorizable_shift): Likewise.
26394 (vectorizable_operation): Likewise.
26395 (vectorizable_comparison): Likewise.
26396 (vect_is_simple_cond): Record the def types for operands.
26397 (vectorizable_condition): Likewise, call vect_model_simple_cost.
26398 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
26399 for statement argument count.
26401 2017-05-16 Carl Love <cel@us.ibm.com>
26403 * config/rs6000/rs6000-c: Add support for built-in functions
26404 vector unsigned long long vec_bperm (vector unsigned long long,
26405 vector unsigned char)
26406 vector signed long long vec_mule (vector signed int,
26408 vector unsigned long long vec_mule (vector unsigned int,
26409 vector unsigned int)
26410 vector signed long long vec_mulo (vector signed int,
26412 vector unsigned long long vec_mulo (vector unsigned int,
26413 vector unsigned int)
26414 vector signed char vec_sldw (vector signed char,
26415 vector signed char,
26417 vector unsigned char vec_sldw (vector unsigned char,
26418 vector unsigned char,
26420 vector signed short vec_sldw (vector signed short,
26421 vector signed short,
26423 vector unsigned short vec_sldw (vector unsigned short,
26424 vector unsigned short,
26426 vector signed int vec_sldw (vector signed int,
26429 vector unsigned int vec_sldw (vector unsigned int,
26430 vector unsigned int,
26432 vector signed long long vec_sldw (vector signed long long,
26433 vector signed long long,
26435 vector unsigned long long vec_sldw (vector unsigned long long,
26436 vector unsigned long long,
26438 * config/rs6000/rs6000-c: Add support for built-in functions
26439 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
26440 * config/rs6000/altivec.h: Add defintion for vec_sldw.
26441 * doc/extend.texi: Update the built-in documentation for the
26442 new built-in functions.
26444 2017-05-16 Marek Polacek <polacek@redhat.com>
26448 * tree.c (save_expr): Don't fold the expression.
26450 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
26452 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
26453 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
26454 and (?*y,m). Update insn attributes.
26456 2017-05-16 Martin Liska <mliska@suse.cz>
26458 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
26459 flags argument of print_gimple_stmt, print_gimple_expr,
26460 print_generic_stmt and print_generic_expr.
26461 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
26462 * coretypes.h: Likewise.
26463 * except.c (dump_eh_tree): Likewise.
26464 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
26465 * gimple-pretty-print.h: Likewise.
26466 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
26467 (backprop::push_to_worklist): Likewise.
26468 (backprop::pop_from_worklist): Likewise.
26469 (backprop::process_use): Likewise.
26470 (backprop::intersect_uses): Likewise.
26471 (note_replacement): Likewise.
26472 * gimple-ssa-store-merging.c
26473 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
26474 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
26475 (pass_store_merging::execute): Likewise.
26476 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
26477 (ssa_base_cand_dump_callback): Likewise.
26478 (dump_incr_vec): Likewise.
26479 (replace_refs): Likewise.
26480 (replace_mult_candidate): Likewise.
26481 (create_add_on_incoming_edge): Likewise.
26482 (create_phi_basis): Likewise.
26483 (insert_initializers): Likewise.
26484 (all_phi_incrs_profitable): Likewise.
26485 (introduce_cast_before_cand): Likewise.
26486 (replace_one_candidate): Likewise.
26487 * gimplify.c (gimplify_expr): Likewise.
26488 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
26489 (set_rename): Likewise.
26490 (rename_uses): Likewise.
26491 (copy_loop_phi_nodes): Likewise.
26492 (add_close_phis_to_merge_points): Likewise.
26493 (copy_loop_close_phi_args): Likewise.
26494 (copy_cond_phi_args): Likewise.
26495 (graphite_copy_stmts_from_block): Likewise.
26496 (translate_pending_phi_nodes): Likewise.
26497 * graphite-poly.c (print_pdr): Likewise.
26498 (dump_gbb_cases): Likewise.
26499 (dump_gbb_conditions): Likewise.
26500 (print_scop_params): Likewise.
26501 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
26502 (build_cross_bb_scalars_use): Likewise.
26503 (gather_bbs::before_dom_children): Likewise.
26504 * hsa-dump.c (dump_hsa_immed): Likewise.
26505 * ipa-cp.c (print_ipcp_constant_value): Likewise.
26506 (get_replacement_map): Likewise.
26507 * ipa-inline-analysis.c (dump_condition): Likewise.
26508 (estimate_function_body_sizes): Likewise.
26509 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
26510 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
26511 * ipa-prop.c (ipa_dump_param): Likewise.
26512 (ipa_print_node_jump_functions_for_edge): Likewise.
26513 (ipa_modify_call_arguments): Likewise.
26514 (ipa_modify_expr): Likewise.
26515 (ipa_dump_param_adjustments): Likewise.
26516 (ipa_dump_agg_replacement_values): Likewise.
26517 (ipcp_modif_dom_walker::before_dom_children): Likewise.
26518 * ipa-pure-const.c (check_stmt): Likewise.
26519 (pass_nothrow::execute): Likewise.
26520 * ipa-split.c (execute_split_functions): Likewise.
26521 * omp-offload.c (dump_oacc_loop_part): Likewise.
26522 (dump_oacc_loop): Likewise.
26523 * trans-mem.c (tm_log_emit): Likewise.
26524 (tm_memopt_accumulate_memops): Likewise.
26525 (dump_tm_memopt_set): Likewise.
26526 (dump_tm_memopt_transform): Likewise.
26527 * tree-cfg.c (gimple_verify_flow_info): Likewise.
26528 (print_loop): Likewise.
26529 * tree-chkp-opt.c (chkp_print_addr): Likewise.
26530 (chkp_gather_checks_info): Likewise.
26531 (chkp_get_check_result): Likewise.
26532 (chkp_remove_check_if_pass): Likewise.
26533 (chkp_use_outer_bounds_if_possible): Likewise.
26534 (chkp_reduce_bounds_lifetime): Likewise.
26535 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
26536 (chkp_mark_completed_bounds): Likewise.
26537 (chkp_register_incomplete_bounds): Likewise.
26538 (chkp_mark_invalid_bounds): Likewise.
26539 (chkp_maybe_copy_and_register_bounds): Likewise.
26540 (chkp_build_returned_bound): Likewise.
26541 (chkp_get_bound_for_parm): Likewise.
26542 (chkp_build_bndldx): Likewise.
26543 (chkp_get_bounds_by_definition): Likewise.
26544 (chkp_generate_extern_var_bounds): Likewise.
26545 (chkp_get_bounds_for_decl_addr): Likewise.
26546 * tree-chrec.c (chrec_apply): Likewise.
26547 * tree-data-ref.c (dump_data_reference): Likewise.
26548 (dump_subscript): Likewise.
26549 (dump_data_dependence_relation): Likewise.
26550 (analyze_overlapping_iterations): Likewise.
26551 * tree-inline.c (expand_call_inline): Likewise.
26552 (tree_function_versioning): Likewise.
26553 * tree-into-ssa.c (dump_defs_stack): Likewise.
26554 (dump_currdefs): Likewise.
26555 (dump_names_replaced_by): Likewise.
26556 (dump_update_ssa): Likewise.
26557 (update_ssa): Likewise.
26558 * tree-object-size.c (pass_object_sizes::execute): Likewise.
26559 * tree-parloops.c (build_new_reduction): Likewise.
26560 (try_create_reduction_list): Likewise.
26561 (ref_conflicts_with_region): Likewise.
26562 (oacc_entry_exit_ok_1): Likewise.
26563 (oacc_entry_exit_single_gang): Likewise.
26564 * tree-pretty-print.h: Likewise.
26565 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
26566 (get_scalar_evolution): Likewise.
26567 (add_to_evolution): Likewise.
26568 (get_loop_exit_condition): Likewise.
26569 (analyze_evolution_in_loop): Likewise.
26570 (analyze_initial_condition): Likewise.
26571 (analyze_scalar_evolution): Likewise.
26572 (instantiate_scev): Likewise.
26573 (number_of_latch_executions): Likewise.
26574 (gather_chrec_stats): Likewise.
26575 (final_value_replacement_loop): Likewise.
26576 (scev_const_prop): Likewise.
26577 * tree-sra.c (dump_access): Likewise.
26578 (disqualify_candidate): Likewise.
26579 (create_access): Likewise.
26580 (reject): Likewise.
26581 (maybe_add_sra_candidate): Likewise.
26582 (create_access_replacement): Likewise.
26583 (analyze_access_subtree): Likewise.
26584 (analyze_all_variable_accesses): Likewise.
26585 (sra_modify_assign): Likewise.
26586 (initialize_constant_pool_replacements): Likewise.
26587 (find_param_candidates): Likewise.
26588 (decide_one_param_reduction): Likewise.
26589 (replace_removed_params_ssa_names): Likewise.
26590 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
26591 * tree-ssa-copy.c (dump_copy_of): Likewise.
26592 (copy_prop_visit_cond_stmt): Likewise.
26593 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
26594 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
26595 (record_equivalences_from_stmt): Likewise.
26596 * tree-ssa-dse.c (compute_trims): Likewise.
26597 (delete_dead_call): Likewise.
26598 (delete_dead_assignment): Likewise.
26599 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
26600 (forward_propagate_into_cond): Likewise.
26601 (pass_forwprop::execute): Likewise.
26602 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
26603 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
26605 (move_computations_worker): Likewise.
26606 (execute_sm): Likewise.
26607 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
26608 (remove_exits_and_undefined_stmts): Likewise.
26609 (remove_redundant_iv_tests): Likewise.
26610 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
26611 (adjust_iv_update_pos): Likewise.
26612 * tree-ssa-math-opts.c (bswap_replace): Likewise.
26613 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
26614 (value_replacement): Likewise.
26615 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
26616 * tree-ssa-pre.c (print_pre_expr): Likewise.
26617 (get_representative_for): Likewise.
26618 (create_expression_by_pieces): Likewise.
26619 (insert_into_preds_of_block): Likewise.
26620 (eliminate_insert): Likewise.
26621 (eliminate_dom_walker::before_dom_children): Likewise.
26622 (eliminate): Likewise.
26623 (remove_dead_inserted_code): Likewise.
26624 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
26625 * tree-ssa-reassoc.c (get_rank): Likewise.
26626 (eliminate_duplicate_pair): Likewise.
26627 (eliminate_plus_minus_pair): Likewise.
26628 (eliminate_not_pairs): Likewise.
26629 (undistribute_ops_list): Likewise.
26630 (eliminate_redundant_comparison): Likewise.
26631 (update_range_test): Likewise.
26632 (optimize_range_tests_var_bound): Likewise.
26633 (optimize_vec_cond_expr): Likewise.
26634 (rewrite_expr_tree): Likewise.
26635 (rewrite_expr_tree_parallel): Likewise.
26636 (linearize_expr): Likewise.
26637 (break_up_subtract): Likewise.
26638 (linearize_expr_tree): Likewise.
26639 (attempt_builtin_powi): Likewise.
26640 (attempt_builtin_copysign): Likewise.
26641 (transform_stmt_to_copy): Likewise.
26642 (transform_stmt_to_multiply): Likewise.
26643 (dump_ops_vector): Likewise.
26644 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
26645 (print_scc): Likewise.
26646 (set_ssa_val_to): Likewise.
26647 (visit_reference_op_store): Likewise.
26648 (visit_use): Likewise.
26649 (sccvn_dom_walker::before_dom_children): Likewise.
26650 (run_scc_vn): Likewise.
26651 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
26653 (expr_hash_elt::print): Likewise.
26654 (const_and_copies::pop_to_marker): Likewise.
26655 (const_and_copies::record_const_or_copy_raw): Likewise.
26656 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
26657 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
26658 (dump_predicates): Likewise.
26659 (find_uninit_use): Likewise.
26660 (warn_uninitialized_phi): Likewise.
26661 (pass_late_warn_uninitialized::execute): Likewise.
26662 * tree-ssa.c (verify_vssa): Likewise.
26663 (verify_ssa): Likewise.
26664 (maybe_optimize_var): Likewise.
26665 * tree-vrp.c (dump_value_range): Likewise.
26666 (dump_all_value_ranges): Likewise.
26667 (dump_asserts_for): Likewise.
26668 (register_edge_assert_for_2): Likewise.
26669 (vrp_visit_cond_stmt): Likewise.
26670 (vrp_visit_switch_stmt): Likewise.
26671 (vrp_visit_stmt): Likewise.
26672 (vrp_visit_phi_node): Likewise.
26673 (simplify_cond_using_ranges_1): Likewise.
26674 (fold_predicate_in): Likewise.
26675 (evrp_dom_walker::before_dom_children): Likewise.
26676 (evrp_dom_walker::push_value_range): Likewise.
26677 (evrp_dom_walker::pop_value_range): Likewise.
26678 (execute_early_vrp): Likewise.
26680 2017-05-16 Richard Biener <rguenther@suse.de>
26682 * dwarf2out.c (loc_list_from_tree_1): Do not create
26683 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
26685 2017-05-16 Richard Biener <rguenther@suse.de>
26687 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
26689 (note_variable_value_in_expr): If we resolved the decl ref
26690 do not push to the stack.
26692 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
26694 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
26695 operations in fast-math mode.
26696 (vaddq_f16): Likewise.
26697 (vmul_f16): Likewise.
26698 (vmulq_f16): Likewise.
26699 (vsub_f16): Likewise.
26700 (vsubq_f16): Likewise.
26701 * config/arm/neon.md (add<mode>3): New.
26703 (fma:<VH:mode>3): New. Also remove outdated comment.
26706 2017-05-16 Martin Liska <mliska@suse.cz>
26710 * ipa-devirt.c (warn_types_mismatch): Fix typo.
26711 (odr_types_equivalent_p): Likewise.
26713 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
26715 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
26717 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
26720 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
26721 non-interunit SSE move alternatives with '?'.
26722 (zero-extendsidi peephole2): New peephole to skip intermediate
26723 general register in SSE zero-extend sequence.
26725 2017-05-15 Jeff Law <law@redhat.com>
26727 * reorg.c (relax_delay_slots): Create a new variable to hold
26728 the temporary target rather than clobbering TARGET_LABEL.
26730 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
26731 missing argument to extract_bit_field call.
26732 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
26734 2017-05-15 Martin Liska <mliska@suse.cz>
26737 * gcc.c (process_command): Do not allow empty argument of -o option.
26739 2017-05-15 Renlin Li <renlin.li@arm.com>
26741 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
26742 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
26743 * config/aarch64/constraints.md (Usf): Add long call check.
26744 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
26745 (call_value): Likewise.
26746 (sibcall): Likewise.
26747 (sibcall_value): Likewise.
26749 (call_value_insn): New.
26750 (sibcall_insn): Update rtx pattern.
26751 (sibcall_value_insn): Likewise.
26752 (call_internal): Remove.
26753 (call_value_internal): Likewise.
26754 (sibcall_internal): Likewise.
26755 (sibcall_value_internal): Likewise.
26756 (call_reg): Likewise.
26757 (call_symbol): Likewise.
26758 (call_value_reg): Likewise.
26759 (call_value_symbol): Likewise.
26761 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
26764 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
26766 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
26768 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
26769 compatible with CCGOCmode and with CCZmode.
26771 2017-05-14 Martin Sebor <msebor@redhat.com>
26773 PR middle-end/77671
26774 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
26775 (gimple_fold_builtin_snprintf): Same.
26776 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
26777 (gimple_fold_builtin_snprintf): Same.
26778 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
26779 of character types.
26780 (is_call_safe): New function.
26781 (try_substitute_return_value): Call it.
26782 (try_simplify_call): New function.
26783 (pass_sprintf_length::handle_gimple_call): Call it.
26785 2017-05-14 Martin Sebor <msebor@redhat.com>
26787 PR middle-end/80669
26788 * builtins.c (expand_builtin_stpncpy): Simplify.
26790 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
26792 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
26793 * config/i386/i386.h
26794 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
26795 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
26796 (struct machine_function): Add new members call_ms2sysv,
26797 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
26798 (struct machine_frame_state): New fields sp_realigned and
26799 sp_realigned_offset.
26800 * config/i386/i386.c
26801 (enum xlogue_stub): New enum.
26802 (enum xlogue_stub_sets): New enum.
26803 (class xlogue_layout): New class.
26804 (struct ix86_frame): New fields stack_realign_allocate_offset,
26805 stack_realign_offset and outlined_save_offset. Modify comments to
26806 detail stack layout when using out-of-line stubs.
26807 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
26808 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
26809 -mcall-ms2sysv-xlogues.
26810 (stub_managed_regs): New static variable.
26811 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
26812 registers managed by out-of-line stub.
26813 (disable_call_ms2sysv_xlogues): New function.
26814 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
26815 m->call_ms2sysv when appropriate and compute frame layout for
26817 (sp_valid_at, fp_valid_at): New inline functions.
26818 (choose_basereg): New function.
26819 (choose_baseaddr): Add align parameter, use choose_basereg and modify
26821 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
26822 Use align parameter of choose_baseaddr to generated aligned SSE movs
26824 (pro_epilogue_adjust_stack): Modify to track
26825 machine_frame_state::sp_realigned.
26826 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
26827 (ix86_nsaved_sseregs): Likewise.
26828 (ix86_emit_save_regs): Likewise.
26829 (ix86_emit_save_regs_using_mov): Likewise.
26830 (ix86_emit_save_sse_regs_using_mov): Likewise.
26831 (get_scratch_register_on_entry): Likewise.
26832 (gen_frame_set): New function.
26833 (gen_frame_load): Likewise.
26834 (gen_frame_store): Likewise.
26835 (emit_outlined_ms2sysv_save): Likewise.
26836 (emit_outlined_ms2sysv_restore): Likewise.
26837 (ix86_expand_prologue): Modify stack re-alignment code and call
26838 emit_outlined_ms2sysv_save when appropriate.
26839 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
26840 parameter rtx_insn *insn, which allows the function to be used to only
26841 generate the notes.
26842 (ix86_expand_epilogue): Modify validity checks of frame and stack
26843 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
26844 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
26845 * config/i386/predicates.md
26846 (save_multiple): New predicate.
26847 (restore_multiple): Likewise.
26848 * config/i386/sse.md
26849 (save_multiple<mode>): New pattern.
26850 (save_multiple_realign<mode>): Likewise.
26851 (restore_multiple<mode>): Likewise.
26852 (restore_multiple_and_return<mode>): Likewise.
26853 (restore_multiple_leave_return<mode>): Likewise.
26854 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
26856 2017-05-14 Julia Koval <julia.koval@intel.com>
26858 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
26859 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
26860 (__builtin_ia32_xsetbv): New builtins.
26861 * config/i386/i386.c (ix86_expand_special_args_builtin):
26863 (ix86_expand_builtin): Special expand for new intrinsics.
26864 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
26865 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
26866 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
26868 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26870 * cfganal.c (inverted_post_order_compute): Change argument type
26872 * cfganal.h (inverted_post_order_compute): Adjust prototype.
26873 * df-core.c (rest_of_handle_df_initialize): Adjust.
26874 (rest_of_handle_df_finish): Likewise.
26875 (df_analyze_1): Likewise.
26876 (df_analyze): Likewise.
26877 (loop_inverted_post_order_compute): Change argument to be a vec *.
26878 (df_analyze_loop): Adjust.
26879 (df_get_n_blocks): Likewise.
26880 (df_get_postorder): Likewise.
26881 * df.h (struct df_d): Change field to be a vec.
26882 * lcm.c (compute_laterin): Adjust.
26883 (compute_available): Likewise.
26884 * lra-lives.c (lra_create_live_ranges_1): Likewise.
26885 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
26886 * tree-ssa-pre.c (compute_antic): Likewise.
26888 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26890 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
26891 (depth_first_search::depth_first_search): Change structure init
26892 function to this constructor.
26893 (depth_first_search::add_bb): Rename function to this member.
26894 (depth_first_search::execute): Likewise.
26895 (flow_dfs_compute_reverse_finish): Adjust.
26897 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26899 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
26900 (longest_simple_path): Likewise.
26901 * shrink-wrap.c (spread_components): Likewise.
26902 (disqualify_problematic_components): Likewise.
26903 (emit_common_heads_for_components): Likewise.
26904 (emit_common_tails_for_components): Likewise.
26905 (insert_prologue_epilogue_for_components): Likewise.
26907 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26909 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
26912 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26914 * df-core.c (df_set_blocks): Start using auto_bitmap.
26915 (df_compact_blocks): Likewise.
26916 * df-problems.c (df_rd_confluence_n): Likewise.
26917 * df-scan.c (df_insn_rescan_all): Likewise.
26918 (df_process_deferred_rescans): Likewise.
26919 (df_update_entry_block_defs): Likewise.
26920 (df_update_exit_block_uses): Likewise.
26921 (df_entry_block_bitmap_verify): Likewise.
26922 (df_exit_block_bitmap_verify): Likewise.
26923 (df_scan_verify): Likewise.
26924 * lra-constraints.c (lra_constraints): Likewise.
26925 (undo_optional_reloads): Likewise.
26926 (lra_undo_inheritance): Likewise.
26927 * lra-remat.c (calculate_gen_cands): Likewise.
26928 (do_remat): Likewise.
26929 * lra-spills.c (assign_spill_hard_regs): Likewise.
26930 (spill_pseudos): Likewise.
26931 * tree-ssa-pre.c (bitmap_set_and): Likewise.
26932 (bitmap_set_subtract_values): Likewise.
26934 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26936 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
26937 management with auto_bitmap.
26938 (fix_inter_tick): Likewise.
26939 (fix_recovery_deps): Likewise.
26940 * ira.c (add_store_equivs): Likewise.
26941 (find_moveable_pseudos): Likewise.
26942 (split_live_ranges_for_shrink_wrap): Likewise.
26943 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
26944 (rtx_reuse_manager::seen_def_p): Likewise.
26945 (rtx_reuse_manager::set_seen_def): Likewise.
26946 * print-rtl.h (class rtx_reuse_manager): Likewise.
26948 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26950 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
26952 (migrate_btr_def): Likewise.
26953 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
26954 * df-core.c (loop_post_order_compute): Likewise.
26955 (loop_inverted_post_order_compute): Likewise.
26956 * hsa-common.h: Likewise.
26957 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
26958 * init-regs.c (initialize_uninitialized_regs): Likewise.
26959 * ipa-inline.c (resolve_noninline_speculation): Likewise.
26960 (inline_small_functions): Likewise.
26961 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
26962 * ira.c (combine_and_move_insns): Likewise.
26963 (build_insn_chain): Likewise.
26964 * loop-invariant.c (find_invariants): Likewise.
26965 * lower-subreg.c (propagate_pseudo_copies): Likewise.
26966 * predict.c (tree_predict_by_opcode): Likewise.
26967 (predict_paths_leading_to): Likewise.
26968 (predict_paths_leading_to_edge): Likewise.
26969 (estimate_loops_at_level): Likewise.
26970 (estimate_loops): Likewise.
26971 * shrink-wrap.c (try_shrink_wrapping): Likewise.
26972 (spread_components): Likewise.
26973 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
26974 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
26975 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
26976 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
26977 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
26978 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
26979 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
26980 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
26981 (mark_threaded_blocks): Likewise.
26982 (thread_through_all_blocks): Likewise.
26983 * tree-ssa.c (verify_ssa): Likewise.
26984 (execute_update_addresses_taken): Likewise.
26985 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
26987 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26989 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
26991 (post_order_compute): Likewise.
26992 (inverted_post_order_compute): Likewise.
26993 (pre_and_rev_post_order_compute_fn): Likewise.
26995 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26997 * genrecog.c (int_set::int_set): Explicitly construct our
26998 auto_vec base class.
26999 * vec.h (auto_vec::auto_vec): New constructor.
27001 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27003 * bitmap.h (class auto_bitmap): New constructor taking
27004 bitmap_obstack * argument.
27006 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27008 * bitmap.h (class auto_bitmap): Change type of m_bits to
27009 bitmap_head, and adjust ctor / dtor and member operators.
27011 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
27013 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
27014 when returned register mode doesn't match original mode.
27016 2017-05-12 Jeff Law <law@redhat.com>
27017 Jakub Jelinek <jakub@redhat.com>
27019 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
27020 we look for cc setter after the compare-elim changes.
27021 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
27022 within the vector to match what compare-elim now expects.
27023 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
27024 (xorsi3_flags, one_cmplsi2_flags): Likewise.
27026 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
27027 after the compare-elim changes.
27028 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
27029 the vector to match what compare-elim now expects.
27030 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
27031 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
27032 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
27033 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
27034 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
27036 * config/visium/visium.c (single_set_and_flags): Fix where
27037 we look for cc setter after the compare-elim changes.
27038 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
27039 with the vector to match what compare-elim now expects.
27040 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
27041 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
27042 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
27043 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
27044 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
27045 (neg<mode>2_insn_set_overflow): Likewise.
27047 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
27049 PR middle-end/79794
27050 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
27051 maybe_expand_insn call, set ops[0].target. If still set after call,
27052 set alt_rtl. Add extra arg to recursive calls.
27053 (extract_bit_field): Add alt_rtl argument. Pass to
27055 * expmed.h (extract_bit_field): Fix prototype.
27056 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
27057 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
27058 to extract_bit_field_calls.
27059 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
27060 Pass alt_rtl to extract_bit_field calls.
27061 * calls.c (store_unaligned_arguments_into_psuedos)
27062 load_register_parameters): Pass extra NULL to extract_bit_field calls.
27063 * optabs.c (maybe_legitimize_operand): Clear op->target when call
27065 * optabs.h (struct expand_operand): Add target bitfield.
27067 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
27069 * compare-elim.c (try_eliminate_compare): Canonicalize
27070 operation with embedded compare to
27071 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
27072 (set (reg) (operation)].
27074 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
27076 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
27079 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
27080 cost of adding a carry flag for ADC instruction.
27081 [case MINUS]: Ignore the cost of subtracting a carry flag
27082 for SBB instruction.
27084 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
27086 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
27088 * config/rs6000/bmiintrin.h: New file.
27089 * config/rs6000/bmi2intrin.h: New file.
27090 * config/rs6000/x86intrin.h: New file.
27092 2017-05-12 Jeff Law <law@redhat.com>
27094 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
27097 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
27099 PR middle-end/80707
27100 * tree-cfg.c: Remove cfg edges of unreachable case statements.
27102 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27104 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27105 early expansion of vector divide builtins.
27106 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
27107 builtins identified as having unsigned arguments.
27109 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27111 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
27112 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
27113 expansion of vector logical operations (and, andc, or, xor,
27116 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27118 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
27119 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
27121 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27123 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27124 early GIMPLE expansion of vector multiplies.
27126 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27128 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
27129 TARGET_HAVE_MOVT conditional.
27130 (movt splitter): Likewise.
27132 2017-05-12 Richard Biener <rguenther@suse.de>
27134 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
27135 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27136 Fold all stmts not inplace.
27138 2017-05-12 Richard Biener <rguenther@suse.de>
27140 PR tree-optimization/80713
27141 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
27142 inserted_exprs bit for not removed stmts.
27144 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
27146 PR middle-end/69921
27147 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
27148 parallelized" attribute for parallelized OpenACC kernels.
27149 * omp-offload.c (execute_oacc_device_lower): Use it.
27151 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
27152 Set "oacc kernels" attribute.
27153 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
27154 parameter. Adjust all users.
27155 (oacc_fn_attrib_kernels_p): Remove function.
27156 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
27157 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
27158 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
27159 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
27160 assert "oacc kernels" attribute is set.
27162 2017-05-11 Carl Love <cel@us.ibm.com>
27164 * config/rs6000/rs6000-c: Add support for built-in functions
27165 vector unsigned char vec_popcnt (vector signed char)
27166 vector unsigned char vec_popcnt (vector unsigned char)
27167 vector unsigned short vec_popcnt (vector signed short)
27168 vector unsigned short vec_popcnt (vector unsigned short)
27169 vector unsigned int vec_popcnt (vector signed int)
27170 vector unsigned int vec_popcnt (vector unsigned int)
27171 vector unsigned long long vec_popcnt (vector signed long long)
27172 vector unsigned long long vec_popcnt (vector unsigned long long)
27173 vector signed long long vec_slo (vector signed long long,
27174 vector signed char)
27175 vector signed long long vec_slo (vector signed long long,
27176 vector unsigned char)
27177 vector unsigned long long vec_slo (vector unsigned long long,
27178 vector signed char)
27179 vector unsigned long long vec_slo (vector unsigned long long,
27180 vector unsigned char)
27181 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
27182 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
27183 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
27184 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
27185 * doc/extend.texi: Update the built-in documentation file for the
27186 new built-in functions.
27188 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
27190 * attribs.h (sorted_attr_string): Move machine independent
27191 functions for target clone support from the i386 port to common
27192 code. Rename ix86_function_versions to common_function_versions.
27193 Rename make_name to make_unique_name.
27194 (common_function_versions): Likewise.
27195 (make_unique_name): Likewise.
27196 (make_dispatcher_decl): Likewise.
27197 (is_function_default_version): Likewise.
27198 * attribs.c (attr_strcmp): Likewise.
27199 (sorted_attr_string): Likewise.
27200 (common_function_versions): Likewise.
27201 (make_unique_name): Likewise.
27202 (make_dispatcher_decl): Likewise.
27203 (is_function_default_version): Likewise.
27204 * config/i386/i386.c (attr_strcmp): Likewise.
27205 (sorted_attr_string): Likewise.
27206 (ix86_function_versions): Likewise.
27207 (make_name): Likewise.
27208 (make_dispatcher_decl): Likewise.
27209 (is_function_default_version): Likewise.
27210 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
27212 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27215 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
27216 Account for direct move costs for vec_construct of integer
27219 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
27222 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
27223 (UNSPEC_STX_ATOMIC): Ditto.
27224 (loaddi_via_sse): New insn.
27225 (storedi_via_sse): Ditto.
27226 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
27227 Update corresponding peephole2 patterns.
27228 (atomic_storedi_fpu): Ditto.
27230 2017-05-11 Julia Koval <julia.koval@intel.com>
27232 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
27233 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
27235 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
27236 (__builtin_ia32_rsqrt14ss_mask): New builtins.
27237 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
27239 2017-05-11 Nathan Sidwell <nathan@acm.org>
27241 * graphite-poly.c: Include dumpfile.h.
27243 * dumpfle.h (dump_function): Declare here ...
27244 * tree-dump.h (dump_function): ... not here.
27245 * dumpfile.c: #include tree-cfg.h.
27246 (dump_function): Move here from ...
27247 * tree-dump.c (dump_function): ... here.
27248 * gimplify.c: #include splay-tree.h, not tree-dump.h.
27249 * graphite-poly.c: Don't include tree-dump.h.
27250 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
27251 * print-tree.c: Likewise.
27252 * stor-layout.c: Likewise.
27253 * tree-nested.c: Likewise.
27255 * dumpfile.c (dump_start): Use TDF_FLAGS.
27256 (dump_enable_all): Fix TDF_KIND check thinko.
27258 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
27260 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27261 array entries to represent two legal parameterizations of the
27262 overloaded __builtin_cmpb function, as represented by the
27263 P6_OV_BUILTIN_CMPB constant.
27264 (altivec_resolve_overloaded_builtin): Add special case handling
27265 for the __builtin_cmpb function, as represented by the
27266 P6_OV_BUILTIN_CMPB constant.
27267 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
27268 (BU_P6_64BIT_2): New macro.
27269 (BU_P6_OVERLOAD_2): New macro
27270 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
27271 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
27272 (CMPB): Add overload support to represent both 32-bit and 64-bit
27273 compare-bytes function.
27274 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
27275 support for TARGET_CMPB.
27276 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
27277 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27278 documentation of the __builtin_cmpb overloaded built-in function.
27280 2017-05-11 Richard Biener <rguenther@suse.de>
27282 PR tree-optimization/80705
27283 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
27284 bases are not vectorizable.
27286 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27288 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
27289 when counting register pressure.
27291 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27293 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
27294 (iv_ca_more_deps): Renamed to ...
27295 (iv_ca_compare_deps): ... this.
27296 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
27298 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27300 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
27302 (determine_group_iv_costs): ... here.
27303 (find_inv_vars_cb): Record inv var if it's not recorded before.
27305 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27307 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
27308 (get_shiftadd_cost): Ditto.
27310 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27312 * tree-ssa-address.c: Include header file.
27313 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
27315 (add_to_parts): Refactor.
27316 (addr_to_parts): New parameter. Update use of move_hint_to_base.
27317 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
27320 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27322 PR tree-optimization/53090
27323 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
27325 (extract_cond_operands): Detect condition with IV on both sides
27326 and return COMP_IV_EXPR_2.
27327 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
27328 (rewrite_use_compare): Simplify by removing call to function
27329 extract_cond_operands.
27331 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27333 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
27334 (extract_cond_operands): Detect condition comparing against non-
27335 invariant bound and return appropriate enum value.
27336 (find_interesting_uses_cond): Update use of extract_cond_operands.
27337 Handle its return value accordingly.
27338 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
27340 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27342 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
27343 nonlinear iv_use computation in loop invariant sensitive way.
27345 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27347 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
27348 (find_iv_candidates): Call relate_compare_use_with_all_cands.
27350 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27352 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
27353 (dump_cand): Support iv_cand.inv_exprs.
27354 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
27356 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
27359 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27361 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
27363 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
27364 as local function. Include necessary header files.
27365 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
27367 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27369 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
27371 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27373 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
27374 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
27375 RSHIFT_EXPR and BIT_NOT_EXPR.
27377 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27379 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
27380 (adjust_setup_cost): New parameter supporting round up adjustment.
27381 (struct address_cost_data): Delete.
27382 (force_expr_to_var_cost): Don't bound cost with spill_cost.
27383 (split_address_cost, ptr_difference_cost): Delete.
27384 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
27385 (struct ainc_cost_data): New struct.
27386 (get_address_cost_ainc): New function.
27387 (get_address_cost, get_computation_cost): Reimplement.
27388 (determine_group_iv_cost_address): Record inv_expr for all uses of
27390 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
27391 (iv_ca_has_deps): Reimplemented to ...
27392 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
27394 (iv_ca_extend): Call iv_ca_more_deps.
27396 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27398 * tree-ssa-address.c (struct mem_address): Move to header file.
27399 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
27400 * tree-ssa-address.h (struct mem_address): Move from C file.
27401 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
27403 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27405 * tree-affine.h (aff_combination_type): New interface.
27406 (aff_combination_zero_p): Remove static.
27407 (aff_combination_const_p): New interface.
27408 (aff_combination_singleton_var_p): New interfaces.
27410 2017-05-11 Richard Biener <rguenther@suse.de>
27412 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27413 Skip unreachable blocks and destinations.
27414 (eliminate): Move stmt removal and fixup ...
27415 (fini_eliminate): ... here. Skip inserted exprs.
27416 (pass_pre::execute): Move fini_pre after fini_eliminate.
27417 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
27418 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
27419 PRE to get rid of dead code that has invalid SSA form and
27420 split critical edges again.
27422 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27424 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27426 2017-05-11 Richard Biener <rguenther@suse.de>
27428 * passes.c (execute_function_todo): Verify loops if they are
27429 said to be up-to-date.
27430 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
27431 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
27433 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
27436 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
27437 handle calling assemble_external ourself.
27440 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
27441 modes with zero size. Enhance comment.
27443 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27445 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
27446 built-ins for vec_xl and vec_xst with short and char pointer
27449 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27451 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
27452 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
27453 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
27454 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
27455 (_mm_maskz_min_round_ss): New intrinsics.
27456 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
27457 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
27458 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
27459 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
27460 (__builtin_ia32_minss_mask_round): New builtins.
27461 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27462 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27463 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
27465 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
27466 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
27468 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
27471 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27473 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
27474 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
27475 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
27476 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
27477 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
27478 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
27479 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
27480 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27481 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27482 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
27483 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
27484 (__builtin_ia32_mulss_mask_round): New builtins.
27485 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27486 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27487 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
27489 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
27490 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27492 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27495 2017-05-10 Julia Koval <julia.koval@intel.com>
27497 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
27498 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
27499 (_mm256_setr_m128i): New intrinsics.
27501 2017-05-10 Julia Koval <julia.koval@intel.com>
27503 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
27504 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
27505 (_mm_maskz_rcp14_ss): New intrinsics.
27506 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
27507 (__builtin_ia32_rcp14ss_mask): New builtins.
27508 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
27510 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
27512 PR tree-optimization/51513
27513 * tree-cfg.c (gimple_seq_unreachable_p): New function.
27514 (assert_unreachable_fallthru_edge_p): Use it.
27515 (group_case_labels_stmt): Likewise.
27516 * tree-cfg.h: Prototype it.
27517 * stmt.c: Include cfghooks.h and tree-cfg.h.
27518 (emit_case_dispatch_table) <gap_label>: New local variable.
27519 Use it to fill dispatch table gaps.
27520 Test for default_label before updating probabilities.
27521 (expand_case) <default_label>: Remove unneeded initialization.
27522 Test for unreachable default case statement and remove its edge.
27523 Set default_label accordingly.
27524 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
27526 2017-05-10 Carl Love <cel@us.ibm.com>
27528 * config/rs6000/rs6000-c: Add support for built-in functions
27529 vector signed char vec_neg (vector signed char)
27530 vector signed short int vec_neg (vector short int)
27531 vector signed int vec_neg (vector signed int)
27532 vector signed long long vec_neg (vector signed long long)
27533 vector float vec_neg (vector float)
27534 vector double vec_neg (vector double)
27535 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
27537 * config/rs6000/altivec.h: Add define for vec_neg
27538 * doc/extend.texi: Update the built-in documentation for the
27539 new built-in functions.
27541 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27543 PR tree-optimization/77644
27544 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
27546 2017-05-10 Nathan Sidwell <nathan@acm.org>
27548 * dumpfile.h (TDI_lang_all): New.
27549 (TDF_KIND): New. Renumber others
27550 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
27552 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
27554 (get_dump_file_name): Adjust suffix generation.
27555 (dump_enable_all): Use TDF_KIND.
27556 * doc/invoke.texi (-fdump-lang-all): Document.
27558 * dumpfile.h: Tabify.
27560 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
27563 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
27564 Move member access before delete.
27566 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
27568 * tree-inline.c (expand_call_inline): Split block at stmt
27571 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
27574 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
27575 are now unused after splitting mov{sf,sd}_hardfloat.
27576 (f32_lr2): Likewise.
27577 (f32_lm): Likewise.
27578 (f32_lm2): Likewise.
27579 (f32_li): Likewise.
27580 (f32_li2): Likewise.
27581 (f32_lv): Likewise.
27582 (f32_sr): Likewise.
27583 (f32_sr2): Likewise.
27584 (f32_sm): Likewise.
27585 (f32_sm2): Likewise.
27586 (f32_si): Likewise.
27587 (f32_si2): Likewise.
27588 (f32_sv): Likewise.
27589 (f32_dm): Likewise.
27590 (f32_vsx): Likewise.
27591 (f32_av): Likewise.
27592 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
27593 For movsf, order stores so the VSX stores occur before the GPR
27594 store which encourages the register allocator to use a traditional
27595 FPR instead of a GPR. For movsd, order the stores so that the GPR
27596 store comes before the VSX stores to allow the power6 to work.
27597 This is due to the power6 not having a 32-bit integer store
27598 instruction from a FPR.
27599 (movsf_hardfloat): Likewise.
27600 (movsd_hardfloat): Likewise.
27602 2017-05-09 Martin Sebor <msebor@redhat.com>
27604 PR translation/80280
27605 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
27608 PR translation/80280
27609 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
27610 data member added in r247778.
27611 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
27613 2017-05-09 Nathan Sidwell <nathan@acm.org>
27615 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
27617 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
27620 2017-05-09 Marek Polacek <polacek@redhat.com>
27622 * doc/invoke.texi: Fix typo.
27624 2017-05-09 Richard Biener <rguenther@suse.de>
27626 * tree-vrp.c (vrp_val_is_max): Adjust comment.
27627 (vrp_val_is_min): Likewise.
27628 (set_value_range_to_value): Likewise.
27629 (set_value_range_to_nonnegative): Likewise.
27630 (gimple_assign_nonzero_p): Likewise.
27631 (gimple_stmt_nonzero_p): Likewise.
27632 (vrp_int_const_binop): Likewise. Remove unreachable case.
27633 (adjust_range_with_scev): Adjust comments.
27634 (compare_range_with_value): Likewise.
27635 (extract_range_from_phi_node): Likewise.
27636 (test_for_singularity): Likewise.
27638 2017-05-09 Richard Biener <rguenther@suse.de>
27640 * tree-vrp.c (get_single_symbol): Add assert that we don't
27641 get overflowed constants as invariant part.
27642 (compare_values_warnv): Add comment before the TREE_NO_WARNING
27643 checks. Use wi::cmp instead of recursing for integer constants.
27644 (compare_values): Just ignore whether we assumed undefined
27645 overflow instead of failing the compare.
27646 (extract_range_for_var_from_comparison_expr): Add comment before the
27647 TREE_NO_WARNING sets.
27648 (test_for_singularity): Likewise.
27649 (extract_range_from_comparison): Do not disable optimization
27650 when we assumed undefined overflow.
27651 (extract_range_basic): Remove init of unused var.
27653 2017-05-09 Richard Biener <rguenther@suse.de>
27655 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
27656 (extract_range_from_multiplicative_op_1): Adjust.
27657 (extract_range_from_binary_expr_1): Use int_const_binop.
27659 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
27662 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
27663 rs6000_store_data_bypass_p in seven define_bypass directives and
27664 in several comments.
27665 * config/rs6000/rs6000-protos.h: Add prototype for
27666 rs6000_store_data_bypass_p function.
27667 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
27668 function implements slightly different (rs6000-specific) semantics
27669 than store_data_bypass_p, returning false rather than aborting
27670 with assertion error when arguments do not satisfy the
27671 requirements of store data bypass.
27672 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
27673 rs6000_store_data_bypass_p.
27675 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
27677 * config/xtensa/xtensa-protos.h
27678 (xtensa_initial_elimination_offset): New declaration.
27679 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
27680 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
27681 macro definition, add case for FRAME_POINTER_REGNUM when
27682 FRAME_GROWS_DOWNWARD.
27683 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
27684 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
27685 xtensa_initial_elimination_offset.
27687 2017-05-08 Nathan Sidwell <nathan@acm.org>
27689 * doc/invoke.texi: Alphabetize -fdump options.
27691 2017-05-08 Martin Sebor <msebor@redhat.com>
27693 PR translation/80280
27694 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
27696 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
27698 * target.def (compute_frame_layout): New optional target hook.
27699 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
27700 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
27701 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
27703 * reload1.c (verify_initial_elim_offsets): Likewise.
27704 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
27705 (use_simple_return_p): Call arm_compute_frame_layout if needed.
27706 (arm_get_frame_offsets): Split up into this ...
27707 (arm_compute_frame_layout): ... and this function.
27709 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
27711 * config/aarch64/constraints.md (Usa): New constraint.
27712 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
27714 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
27716 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
27717 with_multilib_list after it has been checked.
27719 2017-05-08 Richard Biener <rguenther@suse.de>
27721 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
27722 (bitmap_set_subtract_values): Likewise.
27724 2017-05-08 Richard Biener <rguenther@suse.de>
27726 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
27727 (gimple_assign_nonzero): ... this and remove strict_overflow_p
27729 (gimple_stmt_nonzero_warnv_p): Rename to ...
27730 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
27732 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
27733 (extract_range_basic): Adjust, do not disable propagation on
27734 strict overflow sensitive simplification.
27735 (vrp_visit_cond_stmt): Likewise.
27737 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
27739 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
27740 body size unconditionally.
27742 2017-05-07 Jeff Law <law@redhat.com>
27745 2017-05-06 Jeff Law <law@redhat.com>
27746 PR tree-optimization/78496
27747 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27750 PR tree-optimization/78496
27751 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27752 (simplify_stmt_using_ranges): Call it.
27753 (vrp_dom_walker::before_dom_children): Extract equivalences
27754 from an ASSERT_EXPR with an equality comparison against a
27757 2017-05-06 Jeff Law <law@redhat.com>
27759 PR tree-optimization/78496
27760 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27763 PR tree-optimization/78496
27764 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27765 (simplify_stmt_using_ranges): Call it.
27766 (vrp_dom_walker::before_dom_children): Extract equivalences
27767 from an ASSERT_EXPR with an equality comparison against a
27770 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27772 * lra-constraints.c (lra_copy_reg_equiv): New function.
27773 (split_reg): Use it to copy equivalence information from the
27774 original register to the spill register.
27776 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27778 PR rtl-optimization/75964
27779 * simplify-rtx.c (simplify_const_relational_operation): Remove
27780 invalid handling of comparisons of integer ABS.
27782 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
27784 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
27785 initialize to zero.
27786 (init_regs): Remove declaration.
27787 (function_arg_advance_32): Initialize error_p as boolean variable.
27789 2017-05-05 Nathan Sidwell <nathan@acm.org>
27791 * store-motion.c (remove_reachable_equiv_notes): Reformat long
27792 lines. Use for (;;).
27794 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27796 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
27797 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
27798 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
27799 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
27800 VF=2 that require versioning.
27802 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27804 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
27807 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27809 * diagnostic.h (diagnostic_override_option_index): Convert from
27810 macro to inline function.
27812 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27814 * diagnostic.c (last_module_changed_p): New function.
27815 (set_last_module): New function.
27816 (diagnostic_report_current_module): Convert macro usage to
27817 the above functions.
27818 * diagnostic.h (diagnostic_context::last_module): Strengthen
27819 from const line_map * to const line_map_ordinary *.
27820 (diagnostic_last_module_changed): Delete macro.
27821 (diagnostic_set_last_module): Delete macro.
27823 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27825 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
27826 with diagnostic_report_diagnostic.
27827 (diagnostic_n_impl_richloc): Likewise.
27828 * diagnostic.h (report_diagnostic): Delete macro.
27829 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
27830 with diagnostic_report_diagnostic.
27831 * substring-locations.c (format_warning_va): Likewise.
27833 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27835 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
27836 save/restor of format_spec. Move option-printing code to...
27837 (print_option_information): ...this new function, and
27838 reimplement by simply printing to the pretty_printer,
27839 rather than appending to the format string.
27841 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27843 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
27844 handling logic into...
27845 (update_effective_level_from_pragmas): ...this new function.
27847 2017-05-04 Andrew Waterman <andrew@sifive.com>
27849 * config/riscv/riscv.opt (mstrict-align): New option.
27850 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
27851 (SLOW_UNALIGNED_ACCESS): Define.
27852 (riscv_slow_unaligned_access): Declare.
27853 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
27855 (riscv_slow_unaligned_access): New variable.
27856 (rocket_tune_info): Set slow_unaligned_access to true.
27857 (optimize_size_tune_info): Set slow_unaligned_access to false.
27858 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
27859 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
27860 (riscv_option_override): Set riscv_slow_unaligned_access.
27861 * doc/invoke.texi: Add -mstrict-align to RISC-V.
27863 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
27865 * config/riscv/riscv.md: Unify indentation.
27867 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27872 * config/rs6000/rs6000.md (u code attribute): Add FIX and
27874 (extendsi<mode>2): Add support for doing sign extension via
27875 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
27876 don't have ISA 3.0 instructions.
27877 (extendsi<mode>2 splitter): Likewise.
27878 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
27879 generate the normal insns since SImode can now go in vector
27880 registers. Disallow the special UNSPECs needed for previous
27881 machines to hide SImode being used. Add new insns
27882 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
27883 (fix_trunc<mode>si2_stfiwx): Likewise.
27884 (fix_trunc<mode>si2_internal): Likewise.
27885 (fixuns_trunc<mode>si2): Likewise.
27886 (fixuns_trunc<mode>si2_stfiwx): Likewise.
27887 (fctiw<u>z_<mode>_smallint): Likewise.
27888 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
27889 of floating point to 32-bit integer from doing a direct move to
27890 the GPR registers to do a store.
27891 (fctiwz_<mode>): Break long line.
27893 2017-05-05 Bin Cheng <bin.cheng@arm.com>
27895 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
27896 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
27897 (addr_list, addr_offset_valid_p): New.
27898 (split_address_groups): Check offset validity with above function.
27899 (gt-tree-ssa-loop-ivopts.h): Include header file.
27901 2017-05-05 Nathan Sidwell <nathan@acm.org>
27903 * config.gcc (arm*-*-*): Add missing 'fi'.
27905 2017-05-05 Steve Ellcey <sellcey@cavium.com>
27907 * doc/invoke.texi (-fopt-info): Explicitly say order of options
27908 included in -fopt-info does not matter.
27909 * doc/optinfo.texi (-fopt-info): Fix description of default
27910 behavour. Explicitly say order of options included in -fopt-info
27913 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27915 * config.gcc: Allow combinations of aprofile and rmprofile values for
27916 --with-multilib-list.
27917 * config/arm/t-multilib: New file.
27918 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
27919 variables. Remove setting of ISA and floating-point ABI in
27920 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
27921 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
27922 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
27923 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
27925 * config/arm/t-rmprofile: Likewise except for the matches changes.
27926 * doc/install.texi (--with-multilib-list): Document the combination of
27927 aprofile and rmprofile values and warn about pitfalls in doing that.
27929 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
27931 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
27932 (movdi_aarch64): Likewise.
27934 2017-05-05 Jakub Jelinek <jakub@redhat.com>
27936 PR tree-optimization/80632
27937 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
27939 (build_arrays): Initialize it for virtual phis.
27940 (fix_phi_nodes): Use it for virtual phis.
27942 PR tree-optimization/80558
27943 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
27944 [x, y] op z into [x op, y op z] for op & or | if conditions
27947 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
27948 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27951 * config/arm/arm.md (use_literal_pool): Remove.
27952 (64-bit immediate split): No longer takes cost into consideration
27953 if arm_disable_literal_pool is enabled.
27954 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
27955 used when arm_disable_literal_pool is enabled.
27956 (arm_max_const_double_inline_cost): Remove use of
27957 arm_disable_literal_pool.
27958 (push_minipool_fix): Add assert.
27959 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
27960 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
27961 (no_literal_pool_sf_immediate): New.
27963 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27965 PR tree-optimization/80613
27966 * tree-ssa-dce.c (propagate_necessity): Remove cases for
27967 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
27969 2017-05-05 Richard Biener <rguenther@suse.de>
27971 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
27973 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
27975 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
27976 of this flag from insn conditions due to removal from r247495.
27978 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
27980 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
27982 (arm_early_store_addr_dep_ptr): Likewise.
27983 * config/arm/aarch-common-protos.h
27984 (arm_early_load_addr_dep_ptr): Add prototype.
27985 (arm_early_store_addr_dep_ptr): Likewise.
27986 * config/arm/cortex-a53.md: Add new bypasses.
27988 2017-05-05 Jakub Jelinek <jakub@redhat.com>
27990 * tree.c (next_type_uid): Change type to unsigned.
27991 (type_hash_canon): Decrement back next_type_uid if
27992 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
27993 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
27996 2017-05-04 Martin Sebor <msebor@redhat.com>
27998 * builtins.c: Fix a trivial typo in a comment.
28000 PR middle-end/79234
28001 * builtins.c (check_sizes): Adjust to handle reading past the end.
28002 Avoid printing excessive upper bound of ranges. Use %E to print
28003 tree nodes instead of converting them to %wu.
28004 (expand_builtin_memchr): New function.
28005 (compute_dest_size): Rename...
28006 (compute_objsize): ...to this.
28007 (expand_builtin_memcpy): Adjust.
28008 (expand_builtin_mempcpy): Adjust.
28009 (expand_builtin_strcat): Adjust.
28010 (expand_builtin_strcpy): Adjust.
28011 (check_strncat_sizes): Adjust.
28012 (expand_builtin_strncat): Adjust.
28013 (expand_builtin_strncpy): Adjust and simplify.
28014 (expand_builtin_memset): Adjust.
28015 (expand_builtin_bzero): Adjust.
28016 (expand_builtin_memcmp): Adjust.
28017 (expand_builtin): Handle memcmp.
28018 (maybe_emit_chk_warning): Check strncat just once.
28020 2017-05-04 Martin Sebor <msebor@redhat.com>
28022 PR preprocessor/79214
28023 PR middle-end/79222
28024 PR middle-end/79223
28025 * builtins.c (check_sizes): Add inlining context and issue
28026 warnings even when -Wno-system-headers is set.
28027 (check_strncat_sizes): Same.
28028 (expand_builtin_strncat): Same.
28029 (expand_builtin_memmove): New function.
28030 (expand_builtin_stpncpy): Same.
28031 (expand_builtin): Handle memmove and stpncpy.
28033 2017-05-04 Bin Cheng <bin.cheng@arm.com>
28035 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
28036 which is not used any more.
28038 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28040 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
28042 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28044 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
28045 (cortexa53_tunings): Likewise.
28046 (cortexa57_tunings): Likewise.
28047 (cortexa72_tunings): Likewise.
28048 (cortexa73_tunings): Likewise.
28050 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28052 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
28053 Set loop alignment to 8.
28055 2017-05-04 Martin Sebor <msebor@redhat.com>
28057 PR translation/80280
28058 * builtins.c (expand_builtin_object_size): Add missing quoting to
28059 %D and like directives.
28060 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
28061 (hsa_type_for_tree_type): Same.
28062 (verify_function_arguments): Same.
28063 * symtab.c (symbol_table::change_decl_assembler_name): Same.
28064 * varasm.c (get_section): Same.
28067 2017-05-04 Martin Sebor <msebor@redhat.com>
28069 PR translation/80280
28070 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
28072 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28074 * config/aarch64/aarch64.c (generic_addrcost_table):
28075 Change HI/TI mode setting.
28077 2017-05-04 Martin Jambor <mjambor@suse.cz>
28079 PR tree-optimization/80622
28080 * tree-sra.c (comes_initialized_p): New function.
28081 (build_accesses_from_assign): Only set write lazily when
28082 comes_initialized_p is false.
28083 (analyze_access_subtree): Use comes_initialized_p.
28084 (propagate_subaccesses_across_link): Assert !comes_initialized_p
28085 instead of testing for PARM_DECL.
28087 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28089 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
28090 constraint on operand 0 to allow more general addressing modes.
28091 Adjust output template.
28092 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
28094 * config/aarch64/aarch64-protos.h
28095 (aarch64_address_valid_for_prefetch_p): Declare prototype.
28096 * config/aarch64/constraints.md (Dp): New address constraint.
28097 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
28100 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
28102 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
28103 update use of estimate_ipcp_clone_size_and_time.
28104 (estimate_local_effects): Update use of
28105 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
28106 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
28107 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
28108 Return nonspecialized time.
28110 2017-05-04 Richard Biener <rguenther@suse.de>
28112 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
28113 for the last VUSE which def dominates the PHI. Directly call
28115 (get_continuation_for_phi_1): Remove.
28117 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
28119 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
28120 to explain the use of truncating division. Cap the number of
28121 iterations to the maximum given by nb_iterations_upper_bound,
28124 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28126 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
28127 * configure: Regenerate.
28128 * config.in: Regenerate.
28129 * config/i386/driver-mingw32.c: new file.
28130 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
28131 * config.host: Link driver-mingw32.o on MinGW host.
28132 * doc/install.texi: Document new --enable-mingw-wildcard configure
28135 2017-05-04 Marek Polacek <polacek@redhat.com>
28137 PR tree-optimization/80612
28138 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
28140 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28141 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
28143 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
28144 (movt splitter): Likewise.
28145 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
28146 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
28147 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
28148 block for Thumb-1 with MOVT.
28149 (thumb2_legitimate_address_p): Move code block ...
28150 (can_avoid_literal_pool_for_label_p): ... into this new function.
28151 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
28153 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
28154 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
28155 "M-profile targets with the MOVT instruction".
28157 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28159 * config/arm/arm-builtins.c (arm_init_builtins): Rename
28160 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
28161 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
28163 2017-05-04 Martin Liska <mliska@suse.cz>
28165 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
28166 variable cond_code.
28168 2017-05-04 Richard Biener <rguenther@suse.de>
28170 * tree.c (array_at_struct_end_p): Handle arrays at struct
28171 end with flexarrays more conservatively. Refactor and treat
28172 arrays of arrays or aggregates more strict. Fix
28173 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
28174 * tree.c (array_at_struct_end_p): Adjust prototype.
28175 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
28176 * gimple-fold.c (get_range_strlen): Likewise.
28177 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
28179 2017-05-04 Richard Biener <rguenther@suse.de>
28181 PR tree-optimization/31130
28182 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
28184 (supports_overflow_infinity): Likewise.
28185 (is_negative_overflow_infinity): Likewise.
28186 (is_positive_overflow_infinity): Likewise.
28187 (is_overflow_infinity): Likewise.
28188 (stmt_overflow_infinity): Likewise.
28189 (overflow_infinity_range_p): Likewise.
28190 (usable_range_p): Remove as always returning true.
28191 (make_overflow_infinity): Remove.
28192 (negative_overflow_infinity): Likewise.
28193 (positive_overflow_infinity): Likewise.
28194 (avoid_overflow_infinity): Likewise.
28195 (set_value_range): Adjust accordingly.
28196 (set_value_range_to_nonnegative): Likewise, remove now unused
28197 overflow_infinity arg.
28198 (vrp_operand_equal_p): Adjust.
28199 (update_value_range): Likewise.
28200 (range_int_cst_singleton_p): Likewise.
28201 (operand_less_p): Likewise.
28202 (compare_values_warnv): Likewise.
28203 (extract_range_for_var_from_comparison_expr): Likewise.
28204 (vrp_int_const_binop): Likewise.
28205 (zero_nonzero_bits_from_vr): Likewise.
28206 (extract_range_from_multiplicative_op_1): Likewise.
28207 (extract_range_from_binary_expr_1): Likewise.
28208 (extract_range_from_unary_expr): Likewise.
28209 (extract_range_from_comparison): Likewise.
28210 (extract_range_basic): Likewise.
28211 (adjust_range_with_scev): Likewise.
28212 (compare_ranges): Likewise.
28213 (compare_range_with_value): Likewise.
28214 (dump_value_range): Likewise.
28215 (test_for_singularity): Likewise, remove strict_overflow_p parameter
28217 (simplify_cond_using_ranges): Adjust.
28219 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28221 * brig-builtins.def: Added a builtin for class_f64.
28222 * builtin-types.def: Added a builtin type needed by class_f64.
28224 2017-05-03 Jason Merrill <jason@redhat.com>
28226 * timevar.def: Add TV_CONSTEXPR.
28228 2017-05-03 David Malcolm <dmalcolm@redhat.com>
28230 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
28232 2017-05-03 Martin Jambor <mjambor@suse.cz>
28234 * ipa-prop.c (ipa_update_after_lto_read): Removed.
28235 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
28236 * ipa-cp.c (ipcp_propagate_stage): Do not call
28237 ipa_update_after_lto_read.
28238 * ipa-inline.c (ipa_inline): Likewise.
28240 2017-05-03 Martin Jambor <mjambor@suse.cz>
28242 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
28243 tag. Added a default constructor and a destructor.
28244 (ipa_edge_args_sum_t): New class;
28245 (ipa_edge_args_sum): Declare.
28246 (ipa_edge_args_vector): Remove declaration.
28247 (IPA_EDGE_REF): Use ipa_edge_args_sum.
28248 (ipa_free_edge_args_substructures): Remove declaration.
28249 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
28250 (ipa_edge_args_info_available_for_edge_p): Likewise.
28251 * ipa-prop.c (ipa_edge_args_vector): Removed.
28252 (edge_removal_hook_holder): Likewise.
28253 (edge_duplication_hook_holder): Likewise.
28254 (ipa_edge_args_sum): New variable.
28255 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
28256 ipa_edge_args_vector.
28257 (ipa_free_edge_args_substructures): Likewise.
28258 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
28259 ipa_edge_args_vector.
28260 (ipa_edge_removal_hook): Turned into method
28261 ipa_edge_args_sum_t::remove.
28262 (ipa_edge_duplication_hook): Turned into method
28263 ipa_edge_args_sum_t::duplicate.
28264 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
28265 registering edge hooks.
28266 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
28267 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
28268 ipa_edge_args_sum instead of ipa_edge_args_vector.
28269 * ipa-profile.c (ipa_profile): Likewise.
28271 2017-05-03 Martin Jambor <mjambor@suse.cz>
28273 * symbol-summary.h (function_summary): New method exists.
28274 (function_summary::symtab_removal): Deallocate through release.
28275 (call_summary): New class.
28276 (gt_ggc_mx): New overload.
28277 (gt_pch_nx): Likewise.
28278 (gt_pch_nx): Likewise.
28280 2017-05-03 Jeff Law <law@redhat.com>
28282 PR tree-optimization/78496
28283 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
28284 from simplify_cond_using_ranges. Split off code to walk
28285 backwards through casts into ...
28286 (simplify_cond_using_ranges_2): New function.
28287 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
28288 (execute_vrp): After identifying jump threads, call
28289 simplify_cond_using_ranges_2.
28291 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
28294 * ipa-inline.h (inline_summary): Add ctor.
28295 (create_ggc): Do not use ggc_cleared_alloc.
28297 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
28298 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28300 * gcc.c (handle_braces): Support escaping in switch matching
28302 * doc/invoke.texi (Spec Files): Document it.
28303 Remove superfluous @code markup in items.
28305 2017-05-03 David Malcolm <dmalcolm@redhat.com>
28307 * diagnostic-show-locus.c (struct column_range): New struct.
28308 (get_affected_columns): New function.
28309 (get_printed_columns): New function.
28310 (struct correction): New struct.
28311 (correction::ensure_capacity): New function.
28312 (correction::ensure_terminated): New function.
28313 (struct line_corrections): New struct.
28314 (line_corrections::~line_corrections): New dtor.
28315 (line_corrections::add_hint): New function.
28316 (layout::print_trailing_fixits): Reimplement in terms of the new
28318 (selftest::test_overlapped_fixit_printing): New function.
28319 (selftest::diagnostic_show_locus_c_tests): Call it.
28321 2017-05-03 Nathan Sidwell <nathan@acm.org>
28323 Canonicalize canonical type hashing
28324 * tree.h (type_hash_canon_hash): Declare.
28325 * tree.c (type_hash_list, attribute_hash_list): Move into
28326 type_hash_canon_hash.
28327 (build_type_attribute_qual_variant): Break out hash code calc into
28328 type_hash_canon_hash.
28329 (type_hash_canon_hash): New. Generic type hash computation.
28330 (build_range_type_1, build_array_type_1, build_function_type,
28331 build_method_type_directly, build_offset_type, build_complex_type,
28332 make_vector_type): Call it.
28334 2017-05-03 Richard Biener <rguenther@suse.de>
28336 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
28337 When all DRs have unknown misaligned do not always peel
28338 when there is a store but apply the same costing model as if
28339 there were only loads.
28341 2017-05-03 Richard Biener <rguenther@suse.de>
28344 PR tree-optimization/80492
28345 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
28346 compare_base_decls returning dont-know properly.
28348 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
28350 * config/arm/iterators.md (CCSI): New mode iterator.
28351 (arch): New mode attribute.
28352 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
28353 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
28354 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
28355 code iterator for success result mode.
28356 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
28357 the corresponding new insn generators.
28359 2017-05-03 Bin Cheng <bin.cheng@arm.com>
28362 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28363 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28365 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
28367 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
28368 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
28369 (DDR_A): Wrap DDR argument in brackets.
28370 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
28371 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
28372 (DDR_REVERSED_P): Likewise.
28374 2017-05-03 Jakub Jelinek <jakub@redhat.com>
28376 PR tree-optimization/79472
28377 * tree-switch-conversion.c (struct switch_conv_info): Add
28378 contiguous_range and default_case_nonstandard fields.
28379 (collect_switch_conv_info): Compute contiguous_range and
28380 default_case_nonstandard fields, don't clear final_bb if
28381 contiguous_range and only the default case doesn't have the required
28383 (check_all_empty_except_final): Set default_case_nonstandard instead
28384 of failing if contiguous_range and the default case doesn't have empty
28386 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
28387 and only the default case doesn't have the required constants. Skip
28389 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
28390 if default_case_nonstandard.
28391 (build_constructors): Build constant 1 just once. Assert that default
28392 values aren't inserted in between cases if contiguous_range. Skip
28394 (build_arrays): Skip virtual phis.
28395 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
28396 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
28397 Handle virtual phis.
28398 (gen_inbound_check): Handle default_case_nonstandard case.
28399 (process_switch): Adjust check_final_bb caller. Call
28400 gather_default_values with the first non-default case instead of
28401 default case if default_case_nonstandard.
28403 2017-05-02 Nathan Sidwell <nathan@acm.org>
28405 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
28406 check. Fix formatting.
28408 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
28410 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
28411 errors when comparing specialized and unspecialized times.
28413 2017-05-02 David Malcolm <dmalcolm@redhat.com>
28415 * diagnostic-show-locus.c
28416 (layout::should_print_annotation_line_p): Make private.
28417 (layout::print_annotation_line): Make private.
28418 (layout::annotation_line_showed_range_p): Make private.
28419 (layout::show_ruler): Make private.
28420 (layout::print_source_line): Make private. Pass in line and
28421 line_width, rather than calling location_get_source_line. Drop
28423 (layout::print_leading_fixits): New method.
28424 (layout::print_any_fixits): Rename to...
28425 (layout::print_trailing_fixits): ...this, and make private.
28426 Don't print newline fixits.
28427 (diagnostic_show_locus): Move logic for printing one row into...
28428 (layout::print_line): ...this new function. Move the
28429 location_get_source_line call and error-handling from
28430 print_source_line to here. Call print_leading_fixits, and rename
28431 print_any_fixits to print_trailing_fixits.
28432 (selftest::test_fixit_insert_containing_newline): Update now that
28433 newlines are partially supported.
28434 (selftest::test_fixit_insert_containing_newline_2): New test.
28435 (selftest::test_fixit_replace_containing_newline): Update comments.
28436 (selftest::diagnostic_show_locus_c_tests): Call the new test.
28437 * edit-context.c (class added_line): New class.
28438 (class edited_line): Describe newline handling in comment.
28439 (edited_line::actually_edited_p): New method.
28440 (edited_line::print_content): Delete redundant decl.
28441 (edited_line::m_predecessors): New field.
28442 (edited_file::print_content): Call edited_line::print_content.
28443 (edited_file::print_diff): Update to support newlines.
28444 (edited_file::print_diff_hunk): Likewise.
28445 (edited_file::print_run_of_changed_lines): New function.
28446 (edited_file::print_diff_line): Convert to...
28447 (print_diff_line): ...this.
28448 (edited_file::get_effective_line_count): New function.
28449 (edited_line::edited_line): Initialize new field m_predecessors.
28450 (edited_line::~edited_line): Clean up m_predecessors.
28451 (edited_line::apply_fixit): Handle newlines.
28452 (edited_line::get_effective_line_count): New function.
28453 (edited_line::print_content): New function.
28454 (edited_line::print_diff_lines): New function.
28455 (selftest::test_applying_fixits_insert_containing_newline): New
28457 (selftest::test_applying_fixits_replace_containing_newline): New
28459 (selftest::insert_line): New function.
28460 (selftest::test_applying_fixits_multiple_lines): Add example of
28462 (selftest::edit_context_c_tests): Call the new tests.
28464 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28466 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
28467 parameter cand. Update dump information.
28468 (get_computation_cost): Update uses.
28470 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28472 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
28473 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
28474 (get_computation_at, rewrite_use_address): Update use of
28475 get_computation_aff.
28477 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28479 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
28480 (get_computation): Delete.
28481 (get_computation_cost): Implement like get_computation_cost_at.
28482 Use get_computation_at.
28483 (get_computation_cost_at): Delete.
28484 (rewrite_use_nonlinear_expr): Use get_computation_at.
28485 (rewrite_use_compare, remove_unused_ivs): Ditto.
28487 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28489 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
28491 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28493 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
28494 (ivopts_global_cost_for_size): Rename parameter and update uses.
28495 (iv_ca_recount_cost): Update uses.
28496 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
28497 candidates seperately in n_invs and n_cands.
28498 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
28500 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28502 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
28503 (find_inv_vars_cb): New.
28504 (find_depends): Renamed to ...
28505 (find_inv_vars): ... this.
28506 (add_candidate_1, force_var_cost): Call find_inv_vars.
28507 (split_address_cost, determine_group_iv_cost_cond): Ditto.
28509 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28511 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
28512 inv_vars. Add inv_exprs.
28513 (struct iv_cand): Rename depends_on to inv_vars.
28514 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
28515 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
28516 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
28517 (dump_cand): Dump inv_vars.
28518 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
28519 (record_invariant, find_depends, add_candidate_1): Ditto.
28520 (set_group_iv_cost, force_var_cost): Ditto.
28521 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
28522 (get_computation_cost_at, get_computation_cost): Ditto.
28523 (determine_group_iv_cost_generic): Ditto.
28524 (determine_group_iv_cost_address): Ditto.
28525 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
28526 (determine_group_iv_costs): Ditto.
28527 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
28528 (iv_ca_set_remove_invariants): Renamed to ...
28529 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
28530 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
28531 (iv_ca_set_add_invariants): Renamed to ...
28532 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
28533 (iv_ca_set_cp): Use iv_ca_set_add_invs.
28534 (iv_ca_has_deps): Support inv_vars and inv_exprs.
28535 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
28536 (create_new_ivs): Remove useless dump.
28538 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28540 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
28542 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
28543 (iv_ca_set_no_cp, create_new_iv): Ditto.
28545 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28547 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28549 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28551 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
28552 function tree_check2.
28554 2017-05-02 Martin Liska <mliska@suse.cz>
28556 * doc/gcov.texi: Add missing preposition.
28557 * gcov.c (function_info::function_info): Properly fill up
28558 all member variables.
28560 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28562 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
28564 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28566 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
28568 2017-05-02 Martin Liska <mliska@suse.cz>
28571 * lto-streamer-in.c (lto_read_tree_1): Remove
28572 LTO_STREAMER_DEBUG.
28573 * lto-streamer.c (struct tree_hash_entry): Likewise.
28574 (struct tree_entry_hasher): Likewise.
28575 (tree_entry_hasher::hash): Likewise.
28576 (tree_entry_hasher::equal): Likewise.
28577 (lto_streamer_init): Likewise.
28578 (lto_orig_address_map): Likewise.
28579 (lto_orig_address_get): Likewise.
28580 (lto_orig_address_remove): Likewise.
28581 * lto-streamer.h: Likewise.
28582 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
28583 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
28585 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
28587 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
28588 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
28589 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
28590 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
28591 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
28592 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
28593 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
28594 (mm_maskz_sub_ss): New intrinsics.
28595 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28596 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28597 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
28598 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
28599 (__builtin_ia32_subss_mask_round): New builtins.
28600 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28601 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28602 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
28604 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
28605 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28607 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28610 2017-05-02 Martin Jambor <mjambor@suse.cz>
28612 PR tree-optimization/78687
28613 * tree-sra.c (access): New field parent.
28614 (process_subtree_disqualification): New function.
28615 (disqualify_candidate): Call it.
28616 (build_accesses_from_assign): Reset write flag if creating an
28618 (build_access_subtree): Fill in parent field and also prpagate
28619 down grp_write flag.
28620 (create_artificial_child_access): New parameter set_grp_write, set
28621 grp_write to its value.
28622 (propagate_subaccesses_across_link): Also propagate grp_write flag
28624 (propagate_all_subaccesses): Push the closest parent back to work
28625 queue if add_access_to_work_queue returned true.
28627 2017-05-02 Richard Biener <rguenther@suse.de>
28629 * common.opt (fstrict-overflow): Alias negative to fwrapv.
28630 * doc/invoke.texi (fstrict-overflow): Remove all traces of
28631 -fstrict-overflow documentation.
28632 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
28633 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
28634 flag_strict_overflow.
28635 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
28636 * lto-opts.c (lto_write_options): Do not stream it.
28637 * lto-wrapper.c (merge_and_complain): Do not handle it.
28638 * opts.c (default_options_table): Do not set -fstrict-overflow.
28639 (finish_options): Likewise do not clear it when sanitizing.
28640 * simplify-rtx.c (simplify_const_relational_operation): Do not
28641 test flag_strict_overflow.
28643 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
28645 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
28646 using enabled attribute.
28647 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
28648 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
28649 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
28650 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
28651 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
28652 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
28653 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
28654 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
28655 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
28656 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
28658 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
28660 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
28662 2017-05-02 Richard Biener <rguenther@suse.de>
28664 PR tree-optimization/80591
28666 2017-04-10 Richard Biener <rguenther@suse.de>
28668 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28671 2017-05-02 Richard Biener <rguenther@suse.de>
28673 PR tree-optimization/80549
28674 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
28675 (cleanup_tree_cfg_noloop): Create forwarders to known loop
28676 headers if they do not have a preheader.
28678 2017-05-02 Martin Liska <mliska@suse.cz>
28681 * common.opt: Fix typo.
28682 * doc/invoke.texi: Likewise.
28684 2017-05-01 Jan Beulich <jbeulich@suse.com>
28686 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
28687 swapping, add (x,x,m,x,n) alternative.
28689 2017-05-01 Nathan Sidwell <nathan@acm.org>
28691 * calls.c (combine_pending_stack_adjustment_and_call): Remove
28692 unnecessary unadjusted_alignment check.
28694 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28697 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
28699 * gimplify.c (gimplify_cilk_detach): New function.
28700 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
28701 * tree-core.h: Document EXPR_CILK_SPAWN.
28702 * tree.h (EXPR_CILK_SPAWN): Define.
28704 2017-05-01 David Malcolm <dmalcolm@redhat.com>
28706 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
28707 to use new fixit_hint representation, using the "replace" logic.
28708 (get_line_span_for_fixit_hint): Likewise.
28709 (layout::print_any_fixits): Likewise.
28710 (selftest::test_one_liner_many_fixits): Rename to...
28711 (selftest::test_one_liner_many_fixits_1): ...this, and update
28712 comment and expected output to reflect that the multiple fix-it
28713 hints are now consolidated into one insertion.
28714 (selftest::test_one_liner_many_fixits_2): New test.
28715 (selftest::test_diagnostic_show_locus_one_liner): Update for
28717 (selftest::test_fixit_consolidation): Update for fix-it API
28719 * diagnostic.c (print_parseable_fixits): Likewise.
28720 * edit-context.c (edited_line::m_line_events): Convert from
28721 auto_vec <line_event *> to auto_vec <line_event>.
28722 (class line_event): Convert from abstract base class to a concrete
28723 class, taking over the role of replace_event.
28724 (class insert_event): Delete.
28725 (class replace_event): Rename to class line_event. Convert to
28727 (edit_context::add_fixits): Reimplement.
28728 (edit_context::apply_insert): Delete.
28729 (edit_context::apply_replace): Rename to...
28730 (edit_context::apply_fixit): ...this. Convert to half-open range.
28731 (edited_file::apply_insert): Delete.
28732 (edited_file::apply_replace): Rename to...
28733 (edited_file::apply_fixit): ...this.
28734 (edited_line::~edited_line): Drop deletion of events.
28735 (edited_line::apply_insert): Delete.
28736 (edited_line::apply_replace): Rename to...
28737 (edited_line::apply_fixit): ...this. Convert to half-open range.
28738 Update for change to type of m_line_events.
28739 * edit-context.h (edit_context::apply_insert): Delete.
28740 (edit_context::apply_replace): Rename to...
28741 (edit_context::apply_fixit): ...this.
28743 2017-05-01 Martin Sebor <msebor@redhat.com>
28745 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
28748 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
28751 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
28752 __get_cpuid_max returns 0.
28753 (__get_cpuid_count): Ditto.
28755 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
28757 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
28758 replacement expression is another instance of one of its arguments.
28760 2017-05-01 Jakub Jelinek <jakub@redhat.com>
28763 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
28764 check for stack push/pop autoinc.
28765 * config/i386/i386.c (ix86_agi_dependent): Return false
28766 if the only reason why modified_in_p returned true is that
28767 addr is SP based and set_insn is a push or pop.
28769 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28771 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
28774 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28777 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
28778 (account_size_time): Use two predicates - exec_pred and
28780 (evaluate_conditions_for_known_args): Compute both clause and
28782 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
28783 (inline_summary_t::duplicate): Update.
28784 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
28786 (compute_inline_parameters): Likewise.
28787 (estimate_edge_size_and_time): Update caluclation of time.
28788 (estimate_node_size_and_time): Compute both time and nonspecialized
28790 (estimate_ipcp_clone_size_and_time): Update.
28791 (inline_merge_summary): Update.
28792 (do_estimate_edge_time): Update.
28793 (do_estimate_edge_size): Update.
28794 (do_estimate_edge_hints): Update.
28795 (inline_read_section, inline_write_summary): Stream both new predicates.
28796 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
28798 (compute_inlined_call_time): Cleanup.
28799 (big_speedup_p): Update.
28800 (edge_badness): Update.
28801 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
28802 (size_time_entry): Replace predicate by exec_predicate and
28803 nonconst_predicate.
28804 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
28805 (estimate_edge_time): Return also nonspec_time.
28806 (reset_edge_growth_cache): Update.
28808 2017-04-29 Jakub Jelinek <jakub@redhat.com>
28810 PR rtl-optimization/80491
28811 * ifcvt.c (noce_process_if_block): When looking for x setter
28812 with missing else_bb, don't check only the insn right before
28813 cond_earliest, but look for the last insn that x is modified in
28814 within the same bb.
28816 PR rtl-optimization/80491
28817 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
28819 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
28821 PR tree-optimization/80487
28822 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
28824 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28826 PR tree-optimization/79697
28827 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
28828 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
28829 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
28831 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
28832 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
28834 2017-04-28 Martin Sebor <msebor@redhat.com>
28836 PR tree-optimization/80523
28837 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
28838 (init_target_to_host_charmap, target_to_host, target_strtol10): New
28840 (maybe_warn, format_directive, parse_directive): Use new functions.
28841 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
28843 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
28845 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
28847 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
28849 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
28850 target_header_dir): Set correctly.
28851 * configure: Regenerated.
28852 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28853 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28854 instead of SYSTEM_HEADER_DIR.
28856 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
28858 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
28859 (estimate_local_effects): Likewise.
28860 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
28861 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
28862 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
28863 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
28864 do_estimate_edge_time, estimate_edge_time): Likewise.
28865 * ipa-inline-analysis.c (estimate_node_size_and_time,
28866 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
28867 (estimate_time_after_inlining): Remove.
28869 2017-04-28 Martin Liska <mliska@suse.cz>
28871 * doc/gcov.texi: Enhance documentation of gcov.
28873 2017-04-28 Martin Liska <mliska@suse.cz>
28875 * doc/gcov.texi: Sort options in alphabetic order.
28876 * doc/gcov-dump.texi: Likewise.
28877 * doc/gcov-tool.texi: Likewise.
28878 * gcov.c (print_usage): Likewise.
28879 * gcov-dump.c (print_usage): Likewise.
28880 * gcov-tool.c (print_merge_usage_message): Likewise.
28881 (print_rewrite_usage_message): Likewise.
28882 (print_overlap_usage_message): Likewise.
28884 2017-04-28 Martin Liska <mliska@suse.cz>
28886 PR gcov-profile/53915
28887 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
28889 2017-04-28 Martin Liska <mliska@suse.cz>
28891 PR gcov-profile/79891
28892 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
28893 is marked by compiler as living on a line.
28894 (get_cycles_count): Remove usage of the union.
28895 (output_intermediate_file): Likewise.
28896 (find_source): Fix GNU coding style.
28897 (accumulate_line_counts): Remove old non-all block mode.
28898 (output_lines): Remove usage of the union.
28899 * profile.c (output_location): Include all BBs, even if
28900 belonging to a same line (and file) as a previous BB.
28902 2017-04-28 Martin Liska <mliska@suse.cz>
28904 * gcov.c (process_args): Handle new argument 'w'.
28905 (read_graph_file): Assign ID to BBs.
28906 (output_branch_count): Display BB # if verbose flag is set.
28907 (output_lines): Likewise for arcs.
28908 (print_usage): Add '--verbose' option help.
28909 * doc/gcov.texi: Document --verbose (-w) option.
28911 2017-04-28 Martin Liska <mliska@suse.cz>
28913 * gcov.c (struct block_location_info): New struct.
28914 (process_file): Fill up the new structure.
28915 (read_graph_file): Replace usage of encoding by the newly added
28917 (add_line_counts): Likewise.
28918 (accumulate_line_counts): Remove usage of the union.
28919 (function_info::function_info): New function.
28920 (function_info::~function_info): Likewise.
28921 (process_file): Call delete instead of release_function.
28922 (release_function): Release the function.
28923 (release_structures): Call delete instead of release_function.
28924 (solve_flow_graph): Replace usage of num_blocks.
28925 (find_exception_blocks): Likewise.
28926 (output_lines): Fix GNU coding style.
28928 2017-04-28 Martin Liska <mliska@suse.cz>
28931 * coverage.c (coverage_remove_note_file): New function.
28932 * coverage.h: Declare the function.
28933 * toplev.c (finalize): Clean if an error has been seen.
28935 2017-04-28 Martin Liska <mliska@suse.cz>
28937 PR gcov-profile/80031
28938 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
28939 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
28940 * gcov.c (read_graph_file): Read just number of blocks.
28941 * profile.c (branch_prob): Do not stream 0 flags per a basic
28944 2017-04-28 Martin Liska <mliska@suse.cz>
28946 * gcov-dump.c (tag_*): Add new argument to declarations.
28947 (dump_gcov_file): Likewise.
28948 (tag_blocks): Add and use new argument depth.
28949 (tag_arcs): Likewise.
28950 (tag_lines): Likewise.
28951 (tag_counters): Likewise.
28952 (tag_summary): Likewise.
28953 (dump_working_sets): Use depth to do a proper indentation.
28955 2017-04-28 Jakub Jelinek <jakub@redhat.com>
28958 * cgraph.h (symtab_node::debug_symtab): No longer inline.
28959 * symtab.c (symtab_node::debug_symtab): Move definition here.
28961 2017-04-28 Richard Biener <rguenther@suse.de>
28963 * lto-streamer.h (LTO_major_version): Bump to 7.
28965 2017-04-28 Richard Biener <rguenther@suse.de>
28967 * tree-vrp.c (assert_info): New struct.
28968 (add_assert_info): New helper.
28969 (register_edge_assert_for_2): Refactor to add asserts to a vector
28971 (register_edge_assert_for_1): Likewise.
28972 (register_edge_assert_for): Likewise.
28973 (finish_register_edge_assert_for): New helper actually registering
28974 asserts where live on edge.
28975 (find_conditional_asserts): Adjust.
28976 (find_switch_asserts): Likewise.
28977 (evrp_dom_walker::try_find_new_range): Generalize.
28978 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
28980 2017-04-27 Marek Polacek <polacek@redhat.com>
28983 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
28984 arg10 and arg11 to itype.
28986 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
28988 * doc/extend.texi (Object Size Checking): Improve grammar.
28990 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
28993 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
28994 that the logic for permitting reciprocal estimates matches that
28997 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29000 * tree.c (type_cache_hasher::equal): Only compare
29001 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
29002 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
29003 non-aggregate element types.
29004 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
29005 about the flag on ARRAY_TYPEs in the comment, formatting fix.
29007 2017-04-27 Richard Biener <rguenther@suse.de>
29009 PR middle-end/80533
29010 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
29011 stripping ARRAY_REFs from MEM_EXPR make sure we're not
29012 keeping a reference to a trailing array.
29014 2017-04-27 Richard Biener <rguenther@suse.de>
29016 PR middle-end/80539
29017 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
29018 being in loop-closed SSA form conservatively.
29019 (chrec_fold_multiply_poly_poly): Likewise.
29021 2017-04-27 Tamar Christina <tamar.christina@arm.com>
29023 PR middle-end/79665
29024 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
29025 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
29027 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29030 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
29031 (aarch64_function_arg_alignment): Return unsigned int again, but still
29032 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
29033 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29034 Don't emit -Wpsabi note.
29035 (aarch64_function_arg_boundary): Likewise.
29036 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29039 2017-04-26 Nathan Sidwell <nathan@acm.org>
29041 * tree.h (crc32_unsigned_n): Declare.
29042 (crc32_unsigned, crc32_unsigned): Make inline.
29043 * tree.c (crc32_unsigned_bits): Replace with ...
29044 (crc32_unsigned_n): ... this.
29045 (crc32_unsigned, crc32_byte): Remove.
29046 (crc32_string): Remove unnecessary braces.
29048 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
29050 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
29051 * ipa-inline-analysis.c (MAX_TIME): Remove.
29052 (account_size_time): Use sreal for time.
29053 (dump_inline_summary): Update.
29054 (estimate_function_body_sizes): Update.
29055 (estimate_edge_size_and_time): Update.
29056 (estimate_calls_size_and_time): Update.
29057 (estimate_node_size_and_time): Update.
29058 (inline_merge_summary): Update.
29059 (inline_update_overall_summary): Update.
29060 (estimate_time_after_inlining): Update.
29061 (inline_read_section): Update.
29062 (inline_write_summary): Update.
29063 * ipa-inline.c (compute_uninlined_call_time): Update.
29064 (compute_inlined_call_time): Update.
29065 (recursive_inlining): Update.
29066 (inline_small_functions): Update.
29067 (dump_overall_stats): Update.
29068 * ipa-inline.h: Include sreal.h.
29069 (size_time_entry): Turn time to sreal.
29070 (inline_summary): Turn self_time nad time to sreal.
29072 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
29074 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
29076 (sreal::stream_out, sreal::stream_in): New.
29077 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
29079 2017-04-25 Jakub Jelinek <jakub@redhat.com>
29081 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
29084 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
29087 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
29089 (dimode_scalar_chain::compute_convert_gain): Ditto.
29090 (dimode_scalar_chain::make_vector_copies): Ditto.
29091 (dimode_scalar_chain::convert_reg): Ditto.
29092 (dimode_scalar_chain::convert_insn): Ditto.
29093 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
29094 (VI248_AVX512BW_1): New mode iterator.
29095 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
29096 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
29099 2017-04-25 Martin Sebor <msebor@redhat.com>
29101 PR tree-optimization/80497
29102 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
29103 constants are representable in HOST_WIDE_INT.
29104 (parse_directive): Ditto.
29106 2017-04-25 Martin Sebor <msebor@redhat.com>
29109 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
29110 (new_zero_array): Adjust signature.
29111 (dom_info::dom_init): Used unsigned rather that size_t.
29112 (dom_info::dom_info): Same.
29114 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29115 Jakub Jelinek <jakub@redhat.com>
29118 * config/arm/arm.c: Include gimple.h.
29119 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29120 returns negative, increment ncrn only if it returned positive.
29121 (arm_needs_doubleword_align): Return int instead of bool,
29122 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
29123 members, but if there is any such non-FIELD_DECL
29124 > PARM_BOUNDARY aligned decl, return -1 instead of false.
29125 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29126 returns negative, increment nregs only if it returned positive.
29127 (arm_setup_incoming_varargs): Likewise.
29128 (arm_function_arg_boundary): Emit -Wpsabi note if
29129 arm_needs_doubleword_align returns negative, return
29130 DOUBLEWORD_ALIGNMENT only if it returned positive.
29132 2017-04-25 Marek Polacek <polacek@redhat.com>
29135 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29136 first argument to type.
29138 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
29141 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
29142 type checks to test for compatibility instead of equality.
29144 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29145 Jakub Jelinek <jakub@redhat.com>
29148 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
29150 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
29151 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
29152 the alignment computation, but return their maximum in warn_alignment.
29153 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29154 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
29156 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
29157 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29160 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29162 * config/arc/simdext.md (dmpyh): Fix typo.
29164 2017-04-25 Richard Biener <rguenther@suse.de>
29166 PR tree-optimization/80492
29167 * alias.c (compare_base_decls): Handle registers with asm
29168 specification conservatively.
29169 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29170 compare_base_decls returning dont-know properly.
29172 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29174 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
29175 (legitimate_offset_address_p): New function.
29176 (arc_legitimate_address_p): Use above function.
29178 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29180 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
29182 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29184 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
29185 ACCH registers whenever they are available.
29187 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29189 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
29190 double regs fix when not used.
29192 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29194 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
29196 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
29197 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
29199 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29201 * config/arc/arc.c (arc_output_addsi): Check for h-register class
29202 when emitting short ADD instructions.
29204 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29206 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
29208 (cmpsi_cc_c_insn): Likewise.
29209 (cbranchsi4_scratch): Compute proper instruction length using
29210 compact_hreg_operand.
29211 * config/arc/predicates.md (compact_hreg_operand): New predicate.
29213 2017-04-25 Richard Biener <rguenther@suse.de>
29215 PR middle-end/80509
29216 * passes.c (pass_manager::pass_manager): Initialize
29217 m_name_to_pass_map.
29219 2017-04-25 Richard Biener <rguenther@suse.de>
29221 PR tree-optimization/79201
29222 * tree-ssa-sink.c (statement_sink_location): Handle calls.
29224 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29227 * config/s390/vector.md: Split MEM->GPR vector moves for
29228 non-s_operand addresses.
29230 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29233 * config/s390/predicates.md (reload_const_wide_int_operand): New
29235 * config/s390/s390.md ("movti"): Remove d/P alternative.
29236 ("movti_bigconst"): New pattern definition.
29238 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29241 * s390-protos.h (s390_expand_cs_hqi): Removed.
29242 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
29243 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
29244 modes as well as CCZ1mode and CCZmode.
29245 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
29246 signature of s390_emit_compare_and_swap.
29247 (s390_expand_cs_hqi): Likewise, make static.
29248 (s390_expand_cs_tdsi): Generate an explicit compare before trying
29249 compare-and-swap, in some cases.
29250 (s390_expand_cs): Wrapper function.
29251 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
29253 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
29254 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
29255 patterns for small and large integers. Forbid symref memory operands.
29256 Move expander to s390.c. Require cc register.
29257 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
29258 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
29259 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
29260 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
29261 symref memory operands. Remove CC mode and call s390_match_ccmode
29263 ("atomic_exchange<mode>"): Allow and implement all integer modes.
29265 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29267 * config/s390/s390.md (define_peephole2): New peephole to help
29268 combining the load-and-test pattern with volatile memory.
29270 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29272 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
29273 with CCZmode for TARGET_Z196.
29275 2017-04-25 Jakub Jelinek <jakub@redhat.com>
29277 PR rtl-optimization/80501
29278 * combine.c (make_compound_operation_int): Set subreg_code to SET
29279 even for AND with mask of the sign bit of mode.
29281 PR rtl-optimization/80500
29282 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
29283 sum's initial value.
29285 2017-04-25 Julian Brown <julian@codesourcery.com>
29286 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29288 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
29290 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
29292 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
29294 2017-04-25 Julian Brown <julian@codesourcery.com>
29295 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29297 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
29298 (thunderx2t99_sha): New Reservation.
29300 2017-04-25 Julian Brown <julian@codesourcery.com>
29301 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29303 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
29304 type for 1-element load.
29306 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
29308 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
29310 2017-04-24 Martin Jambor <mjambor@suse.cz>
29312 PR tree-optimization/80293
29313 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
29314 char arrays not totally scalarizable if it is false.
29315 (analyze_all_variable_accesses): Pass correct value in the new
29316 parameter. Add a statistics counter.
29318 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
29320 PR middle-end/79931
29321 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
29323 2017-04-24 Richard Biener <rguenther@suse.de>
29325 PR tree-optimization/80494
29326 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
29327 out for complex types.
29329 2017-04-24 Richard Biener <rguenther@suse.de>
29331 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
29332 * tree-ssa-sccvn.c (print_scc): Print SCC size.
29333 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
29334 (DFS): Adjust and never fail.
29335 (sccvn_dom_walker::fail): Remove.
29336 (sccvn_dom_walker::before_dom_children): Adjust.
29337 (run_scc_vn): Likewise and never fail.
29338 * tree-ssa-pre.c (pass_pre::execute): Adjust.
29339 (pass_fre::execute): Likewise.
29341 2017-04-24 Richard Biener <rguenther@suse.de>
29343 PR tree-optimization/79725
29344 * tree-ssa-sink.c (statement_sink_location): Return whether
29345 failure reason was zero uses. Move that check later.
29346 (sink_code_in_bb): Deal with zero uses by removing the stmt
29349 2017-04-24 Richard Biener <rguenther@suse.de>
29352 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
29353 pointer-based references.
29355 2017-04-24 Richard Biener <rguenther@suse.de>
29358 * pass_manager.h (pass_manager::operator new): Remove.
29359 (pass_manager::operator delete): Likewise.
29360 * passes.c (pass_manager::operator new): Remove.
29361 (pass_manager::operator delete): Likewise.
29362 (pass_manager::pass_manager): Zero individual pass members.
29364 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
29367 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
29368 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
29369 Check "XEXP (src, 1)" operand here.
29370 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
29371 Check "XEXP (src, 1)" operand here.
29372 (dimode_scalar_chain::make_vector_copies): Detect count register
29373 of a shift instruction. Zero extend count register from QImode
29374 to DImode to satisfy vector shift pattern count operand predicate.
29375 Substitute vector shift count operand with a DImode copy.
29376 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
29379 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
29381 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
29382 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29383 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
29384 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29385 (UNSPEC_NOREX_MEM): Remove definition.
29387 2017-04-21 Richard Biener <rguenther@suse.de>
29389 PR tree-optimization/79547
29390 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
29391 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
29392 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
29393 without any constraints.
29395 2017-04-21 Richard Biener <rguenther@suse.de>
29397 PR tree-optimization/78847
29398 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
29400 2017-04-21 Richard Biener <rguenther@suse.de>
29402 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
29403 (build_distinct_type_copy): Likewise.
29404 (build_variant_type_copy): Likewise.
29405 * tree.c (build_qualified_type): Pass down mem-stat info.
29406 (build_distinct_type_copy): Likewise.
29407 (build_variant_type_copy): Likewise.
29409 2017-04-21 Richard Biener <rguenther@suse.de>
29411 PR tree-optimization/80237
29412 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
29414 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
29415 for a simplified result.
29417 2016-04-21 Richard Biener <rguenther@suse.de>
29419 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
29420 sth as strict as a simple_iv but a chrec without symbols and an
29421 operand defined in the loop we are peeling (and not some subloop).
29422 (propagate_constants_for_unrolling): Propagate all constants.
29424 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29427 * config/i386/i386.c (print_reg): Remove assert for disalowed
29428 regno values, call output_operand_lossage instead.
29430 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29433 * config/i386/constraints.md (Yc): New register constraint.
29434 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
29435 Use Yc constraint for alternative 2 of operand 0. Remove
29436 preferred_for_speed attribute.
29438 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29440 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
29441 lastprivate clauses in SIMT case.
29443 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
29445 * doc/invoke.texi (-Wextra-semi): Document new warning option.
29447 2017-04-20 Richard Biener <rguenther@suse.de>
29449 PR tree-optimization/57796
29450 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
29451 as N scalar stores.
29452 (vect_model_load_cost): Cost gathers as N scalar loads.
29454 2017-04-20 Richard Biener <rguenther@suse.de>
29456 * ggc-page.c (ggc_allocated_p): Rename to ...
29457 (safe_lookup_page_table_entry): ... this and return the lookup
29459 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
29461 2017-04-20 Richard Biener <rguenther@suse.de>
29463 PR tree-optimization/80453
29464 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
29465 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
29466 from the conditions.
29467 (vn_phi_eq): Pass them down.
29468 (vn_phi_lookup): Record them.
29469 (vn_phi_insert): Likewise.
29471 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
29473 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
29474 uninitialized variable warning to avoid buffer overrun.
29476 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29479 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
29480 is suppressed for '{ 0 }' in C.
29482 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29484 * BASE-VER: Set to 8.0.0.
29486 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
29488 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
29489 priority .init_array and .fini_array section with SECTION_NOTYPE
29492 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29494 PR middle-end/80423
29495 * tree.h (build_array_type): Add typeless_storage default argument.
29496 * tree.c (type_cache_hasher::equal): Also compare
29497 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
29498 (build_array_type): Add typeless_storage argument, set
29499 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
29501 (build_nonshared_array_type): Adjust build_array_type_1 caller.
29502 (build_array_type): Likewise. Add typeless_storage argument.
29504 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29505 Jakub Jelinek <jakub@redhat.com>
29507 PR tree-optimization/80426
29508 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
29509 operation on symbolic operands, also compute the overflow for the
29510 invariant part when the operation degenerates into a negation.
29512 2017-04-19 Jakub Jelinek <jakub@redhat.com>
29515 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
29516 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
29519 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
29521 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
29524 * config/avr/avr.c (tree.h): Include it.
29525 (cgraph.h): Include it.
29526 (avr_encode_section_info): Don't warn for uninitialized progmem
29527 variable if it's just an alias.
29529 2017-04-19 Richard Biener <rguenther@suse.de>
29532 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
29533 when needed by AutoPGO.
29535 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
29538 * doc/lto.texi: Remove an extra 'that'.
29540 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
29542 PR rtl-optimization/80429
29543 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
29544 are only used in debug insns.
29546 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29547 Vladimir Makarov <vmakarov@redhat.com>
29549 * config/sparc/predicates.md (input_operand): Add comment. Return
29550 true for any memory operand when LRA is in progress.
29551 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
29553 2017-04-18 Jeff Law <law@redhat.com>
29556 * mips.md ({return,simple_return}_internal): Do not overwrite
29559 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29561 PR tree-optimization/80443
29562 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
29563 instead of adding 1, subtract -1 and similarly instead of subtracting
29566 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
29568 PR rtl-optimization/80357
29569 * haifa-sched.c (tmp_bitmap): New variable.
29570 (model_recompute): Handle duplicate use records.
29571 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
29572 (free_global_sched_pressure_data): Free it.
29574 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
29577 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29578 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29579 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29580 instead of SYSTEM_HEADER_DIR.
29582 2017-04-18 Jeff Law <law@redhat.com>
29584 PR middle-end/80422
29585 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
29586 predecessors after walking up the insn chain.
29588 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29591 * dwarf2out.c (modified_type_die): Try harder not to emit internal
29592 sizetype type into debug info.
29594 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29597 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
29598 unneeded test for TARGET_UPPER_REGS_SF.
29599 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
29601 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29604 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
29605 instead of gsi_after_labels.
29607 2017-04-18 Jeff Law <law@redhat.com>
29609 * regcprop.c (maybe_mode_change): Avoid creating copies of the
29613 2017-04-13 Jeff Law <law@redhat.com>
29614 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
29615 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
29617 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
29620 * config/avr/avr.c (intl.h): Include it.
29621 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
29623 2017-04-18 Martin Liska <mliska@suse.cz>
29625 PR gcov-profile/78783
29626 * gcov-tool.c (gcov_output_files): Validate that destination
29627 file is either removed by the tool or by a user.
29629 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
29630 Guy Benyei <guybe@mellanox.com>
29632 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
29633 block, and do not negate it, the stored id is already negative.
29635 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
29637 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
29639 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29642 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
29643 masks of options that should be turned off if the VSX vector
29644 options are turned off.
29645 (OTHER_P8_VECTOR_MASKS): Likewise.
29646 (OTHER_VSX_VECTOR_MASKS): Likewise.
29647 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
29648 rs6000_disable_incompatible_switches to validate no type switches
29650 (rs6000_incompatible_switch): New function to disallow turning on
29651 other vector options if -mno-vsx, -mno-power8-vector, or
29652 -mno-power9-vector are specified.
29654 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29656 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
29658 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29660 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
29661 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
29662 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
29663 (ARG_POINTER_CFA_OFFSET): Likewise.
29665 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29667 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
29668 conditions to take advantage of various optimizations.
29670 2017-04-13 Jeff Law <law@redhat.com>
29672 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
29673 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
29674 (zero_extendsidi2_dext): Likewise.
29676 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29679 * fold-const.c (fold_ternary_loc): Revert
29680 use op0 instead of fold_convert_loc (loc, type, arg0) part of
29683 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
29685 PR rtl-optimization/80343
29686 * lra-remat.c (update_scratch_ops): Assign original hard reg to
29687 new scratch pseudo.
29689 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
29692 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
29693 to ubsan_encode_value.
29695 2017-04-13 Jeff Law <law@redhat.com>
29697 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
29698 appearing in DEBUG_INSNs.
29700 2017-04-13 Martin Liska <mliska@suse.cz>
29702 PR gcov-profile/80413
29703 * gcov-io.c (gcov_write_string): Copy to buffer just when
29704 allocated size is greater than zero.
29706 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29709 * dwarf2out.c (decls_for_scope): Ignore declarations of
29710 current_function_decl in BLOCK_NONLOCALIZED_VARS.
29712 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
29715 * ipa-visibility.c (non_local_p): Fix typos.
29716 (localize_node): When localizing symbol in same comdat group,
29717 dissolve the group only when we know external symbols are going
29719 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
29721 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29723 PR tree-optimization/79390
29724 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
29725 order does not result in usable sequence, retry with reversed operand
29731 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
29732 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
29733 op0 instead of fold_convert_loc (loc, type, arg0).
29735 2017-04-12 Jeff Law <law@redhat.com>
29737 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
29738 has a delay slot in the generated code.
29740 * config/cris/cris.md (cris_preferred_reload_class): Return
29741 GENNONACR_REGS rather than GENERAL_REGS.
29743 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29746 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
29747 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
29748 signedness of the result type.
29750 2017-04-12 Richard Biener <rguenther@suse.de>
29751 Jeff Law <law@redhat.com>
29753 PR tree-optimization/80359
29754 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
29755 trim stores to TARGET_MEM_REFs.
29757 2017-04-12 Richard Biener <rguenther@suse.de>
29759 PR tree-optimization/79390
29760 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
29761 threading case even more.
29763 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
29766 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
29767 for quad_address_p for TImode, instead of just not indexed_address.
29769 2017-04-12 Richard Biener <rguenther@suse.de>
29770 Bernd Edlinger <bernd.edlinger@hotmail.de>
29772 PR middle-end/79671
29773 * alias.c (component_uses_parent_alias_set_from): Handle
29774 TYPE_TYPELESS_STORAGE.
29775 (get_alias_set): Likewise.
29776 * tree-core.h (tree_type_common): Add typeless_storage flag.
29777 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
29778 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
29779 for types containing members with TYPE_TYPELESS_STORAGE.
29780 (place_field): Likewise.
29781 (layout_type): Likewise for ARRAY_TYPE.
29782 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
29783 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
29784 TYPE_TYPELESS_STORAGE.
29785 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
29787 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29790 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29791 first argument to type.
29793 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29797 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
29798 CONST0_RTX (mode) rather than const0_rtx where appropriate.
29799 (rs6000_expand_binop_builtin): Likewise.
29800 (rs6000_expand_ternop_builtin): Likewise; also add missing
29801 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
29802 vshasigma built-ins.
29803 * doc/extend.texi: Document that vec_xxpermdi's third argument
29804 must be a constant.
29806 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
29808 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
29809 Use shift_const cost parameter when calculating gain of STV shifts.
29811 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
29813 PR rtl-optimization/70478
29814 * lra-constraints.c (process_alt_operands): Check memory for
29815 disfavoring memory insn operand.
29817 2017-04-11 Jakub Jelinek <jakub@redhat.com>
29819 PR middle-end/80100
29820 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
29821 left shift in unsigned HOST_WIDE_INT type.
29823 PR rtl-optimization/80385
29824 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
29825 (not (neg X)) into (plus X -1) for complex or non-integral modes.
29828 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
29829 if they have any depend clauses.
29831 2017-04-11 Martin Liska <mliska@suse.cz>
29834 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
29835 * ipa-split.c (split_function): Create a local comdat symbol
29836 if caller is in a comdat group.
29838 2017-04-11 Martin Liska <mliska@suse.cz>
29841 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
29844 2017-04-11 Martin Sebor <msebor@redhat.com>
29846 PR middle-end/80364
29847 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
29848 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
29850 (directive::set_width, directive::set_precision, format_character):
29852 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
29855 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
29858 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
29859 conflict, set target->arch_name instead of target->cpu_name.
29861 2017-04-11 Richard Biener <rguenther@suse.de>
29863 PR tree-optimization/80374
29864 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
29865 build_zero_cst, remove fold_convertible_p check again.
29867 2017-04-11 Martin Liska <mliska@suse.cz>
29870 * ubsan.c (instrument_object_size): Do not instrument register
29873 2017-04-11 Jakub Jelinek <jakub@redhat.com>
29876 * config/i386/i386-builtin-types.def
29877 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
29878 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
29879 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
29880 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
29881 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
29882 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
29883 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
29884 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
29885 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
29886 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
29887 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
29888 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
29889 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
29890 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
29891 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
29892 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
29893 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
29894 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
29895 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
29896 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
29897 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
29898 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
29899 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
29900 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
29901 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
29902 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
29903 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
29904 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
29905 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
29906 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
29907 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
29908 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
29909 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
29910 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
29911 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
29912 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
29913 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
29914 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
29915 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
29916 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
29917 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
29918 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
29919 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
29920 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
29921 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
29922 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
29924 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
29925 flag to second_arg_count, handle 4 argument function type _COUNT
29926 aliases, handle second_arg_count on second argument rather than last.
29928 2017-04-10 Jeff Law <law@redhat.com>
29930 PR tree-optimization/80374
29931 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
29932 record anything if we can not convert integer_zero_node to the
29935 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29938 * config/rs6000/rs6000.c (rs6000_option_override_internal):
29939 Enhance special handling given to the TARGET_P9_MINMAX option in
29940 relation to certain other options.
29942 2017-04-10 Bin Cheng <bin.cheng@arm.com>
29944 PR tree-optimization/80153
29945 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
29946 remove POINTER_PLUS_EXPR's base part directly, rather than through
29949 2017-04-10 Richard Biener <rguenther@suse.de>
29950 Bin Cheng <bin.cheng@arm.com>
29952 PR tree-optimization/80153
29953 * tree-affine.c (aff_combination_to_tree): Get base pointer from
29954 the first element of pointer type aff_tree. Build result expr in
29956 (add_elt_to_tree): Convert to type unconditionally. Remove other
29957 fold_convert calls.
29958 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
29959 (rewrite_use_nonlinear_expr): Check invariant using iv information.
29961 2017-04-10 Richard Biener <rguenther@suse.de>
29963 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29966 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
29968 PR rtl-optimization/70478
29969 * lra-constraints.c (curr_small_class_check): New.
29970 (update_and_check_small_class_inputs): New.
29971 (process_alt_operands): Update curr_small_class_check. Disfavor
29972 alternative insn memory operands. Check available regs for small
29975 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
29978 * config/mips/mips.opt (-mvirt): Update description.
29979 * doc/invoke.texi (-mvirt): Likewise.
29981 2017-04-10 Richard Biener <rguenther@suse.de>
29983 PR middle-end/80362
29984 * fold-const.c (fold_binary_loc): Look at unstripped ops when
29985 looking for NEGATE_EXPR in -A / -B to A / B folding.
29987 2017-04-10 Martin Liska <mliska@suse.cz>
29989 PR gcov-profile/80224
29990 * gcov.c (print_usage): Fix usage string.
29991 (get_gcov_intermediate_filename): Remove.
29992 (output_gcov_file): Use both for normal and intermediate format.
29993 (generate_results): Do not initialize special file for
29994 intermediate format.
29996 2017-04-10 Richard Biener <rguenther@suse.de>
29998 PR tree-optimization/80304
29999 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
30002 2017-04-10 Nathan Sidwell <nathan@acm.org>
30005 * config/rs6000/rs6000.c (rs6000_vector_type): New.
30006 (rs6000_init_builtins): Use it.
30008 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30010 * config/arm/arm.md (<mrc>): Add mode to SET source.
30011 (<mrrc>): Likewise.
30013 2017-04-10 Richard Biener <rguenther@suse.de>
30015 PR middle-end/80344
30016 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
30018 2017-04-10 Jakub Jelinek <jakub@redhat.com>
30021 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
30022 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
30023 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
30024 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
30025 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
30026 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
30027 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
30028 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
30029 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
30030 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
30031 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
30032 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
30033 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
30034 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
30035 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
30036 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
30037 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
30038 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
30039 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
30040 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
30041 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
30042 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
30043 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
30045 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
30047 PR rtl-optimization/70478
30048 * lra-constraints.c: Reverse the last patch.
30050 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
30052 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
30053 Add comment for WCHAR_T.
30055 2017-04-08 Martin Liska <mliska@suse.cz>
30058 2017-04-07 Martin Liska <mliska@suse.cz>
30061 * ipa-split.c (split_function): Add function part to a same comdat
30064 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30067 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
30069 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
30071 * rs6000/rs6000.c (vec_load_pendulum): Rename...
30072 (vec_pairing): ...to this.
30073 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
30074 (rs6000_sched_init): Adjust for name change.
30075 (struct rs6000_sched_context): Likewise.
30076 (rs6000_init_sched_context): Likewise.
30077 (rs6000_set_sched_context): Likewise.
30079 2017-04-07 Jakub Jelinek <jakub@redhat.com>
30085 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
30087 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
30088 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
30089 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
30091 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
30093 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
30095 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
30097 PR rtl-optimization/70703
30098 * ira-color.c (update_conflict_hard_regno_costs): Use
30099 int64_t instead of HOST_WIDE_INT.
30101 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
30103 PR rtl-optimization/70478
30104 * lra-constraints.c (process_alt_operands): Disfavor alternative
30105 insn memory operands.
30107 2017-04-07 Jeff Law <law@redhat.com>
30109 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
30110 CALL and NOTE_INSN_CALL_ARG_LOCATION.
30112 2017-04-07 Martin Liska <mliska@suse.cz>
30115 * config/aarch64/aarch64.c (aarch64_process_target_attr):
30116 Show error message instead of an ICE.
30118 2017-04-07 Martin Liska <mliska@suse.cz>
30121 * ipa-split.c (split_function): Add function part to a same comdat
30124 2017-04-07 Richard Biener <rguenther@suse.de>
30126 PR middle-end/80341
30127 * tree.c (get_unwidened): Also handle ! for_type case for
30129 * convert.c (do_narrow): Split out from ...
30130 (convert_to_integer_1): ... here. Do not pass final truncation
30131 type to get_unwidened for TRUNC_DIV_EXPR.
30133 2017-04-07 Richard Biener <rguenther@suse.de>
30135 * tree-affine.c (wide_int_ext_for_comb): Take type rather
30137 (aff_combination_const): Adjust.
30138 (aff_combination_scale): Likewise.
30139 (aff_combination_add_elt): Likewise.
30140 (aff_combination_add_cst): Likewise.
30141 (aff_combination_convert): Likewise.
30142 (add_elt_to_tree): Likewise. Remove unused argument.
30143 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
30145 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
30147 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
30149 * config/arm/arm.c (arm_default_short_enums): Use
30150 ARM_DEFAULT_SHORT_ENUMS.
30151 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
30153 2017-04-06 Jakub Jelinek <jakub@redhat.com>
30156 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
30157 members with redundant out-of-class redeclaration.
30159 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
30162 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
30163 * config/i386/i386.md (*zero_extendsidi2):
30164 Add (?*x,*x) and (?*v,*v) alternatives.
30166 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
30169 * config/i386/i386.c (ix86_expand_builtin)
30170 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
30171 mode from insn data. Convert operands to insn operand mode.
30172 Copy operands that don't satisfy insn predicate to a register.
30174 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
30176 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
30179 2017-04-06 Richard Biener <rguenther@suse.de>
30181 PR tree-optimization/80334
30182 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
30183 preserve alignment of accesses.
30185 2017-04-06 Richard Biener <rguenther@suse.de>
30187 PR tree-optimization/80262
30188 * tree-sra.c (build_ref_for_offset): Preserve address-space
30190 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
30191 Drop useless address-space information on MEM_REF offsets.
30193 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
30195 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
30197 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
30199 PR rtl-optimization/70703
30200 * ira-color.c (update_conflict_hard_regno_costs): Use
30201 HOST_WIDE_INT instead of long.
30203 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
30206 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
30207 not defined for x86_64 target. Add -mmmx target option when __SSE2__
30209 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
30210 for x86_64 target. Handle -m3dnowa option.
30212 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
30214 PR rtl-optimization/70703
30215 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
30216 (update_conflict_hard_regno_costs): Use long instead of unsigned
30217 arithmetic for cost calculation.
30219 2017-04-05 Jakub Jelinek <jakub@redhat.com>
30220 Bernd Edlinger <bernd.edlinger@hotmail.de>
30223 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
30226 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
30229 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
30230 ptr_mode with Pmode throughout.
30231 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
30232 into probe_stack_range and use DImode.
30234 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
30237 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
30238 call_eh_return is true.
30240 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30242 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
30243 Initialize last_match_fntype_index.
30245 2017-04-05 Jakub Jelinek <jakub@redhat.com>
30248 * tree-nvr.c: Include internal-fn.h.
30249 (pass_return_slot::execute): Ignore internal calls without
30252 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30253 Richard Biener <rguenther@suse.de>
30256 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
30257 captures used multiple times, except for the last use.
30258 * generic-match-head.c: Include gimplify.h.
30260 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30262 PR tree-optimization/79390
30263 * target.h (struct noce_if_info): Declare.
30264 * targhooks.h (default_noce_conversion_profitable_p): Declare.
30265 * target.def (noce_conversion_profitable_p): New target hook.
30266 * ifcvt.h (struct noce_if_info): New type, moved from ...
30267 * ifcvt.c (struct noce_if_info): ... here.
30268 (noce_conversion_profitable_p): Renamed to ...
30269 (default_noce_conversion_profitable_p): ... this. No longer
30271 (noce_try_store_flag_constants, noce_try_addcc,
30272 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
30273 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
30274 instead of noce_conversion_profitable_p.
30275 * config/i386/i386.c: Include ifcvt.h.
30276 (ix86_option_override_internal): Don't override
30277 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
30278 (ix86_noce_conversion_profitable_p): New function.
30279 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
30280 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
30281 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
30282 * doc/tm.texi: Regenerated.
30284 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30286 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
30289 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
30292 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
30293 instructions for small multiply cores.
30295 2017-04-04 Jeff Law <law@redhat.com>
30297 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
30299 (mips_expand_vec_perm_const): Initialize elements in orig_perm
30300 that are not set by the loop over the elements.
30302 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30305 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
30306 int mode, convert_modes it to mode as unsigned, otherwise use
30307 lowpart_subreg to mode rather than SImode.
30308 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
30309 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
30310 Use DImode instead of SImode for the shift count operand.
30311 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
30314 2017-04-04 Richard Biener <rguenther@suse.de>
30316 PR middle-end/80281
30317 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
30318 arithmetic done for the negate or the plus. Simplify.
30319 (A - (-B) -> A + B): Likewise.
30320 * fold-const.c (split_tree): Make sure to not negate pointers.
30322 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
30324 PR rtl-optimization/60818
30325 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
30326 a compare of comparisons with the thing compared if this results
30327 in a different machine mode.
30329 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
30331 * alias.c (base_alias_check): Fix typo in comment.
30332 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
30333 * cgraphunit.c (symbol_table::compile): Likewise.
30334 * collect2.c (maybe_run_lto_and_relink): Likewise.
30335 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
30336 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
30337 * config/avr/avr.c (avr_map_op_t): Likewise.
30338 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
30339 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
30340 * config/epiphany/epiphany.md (movcc): Likewise.
30341 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
30342 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
30344 * config/mips/mips.c (mips_save_restore_reg): Likewise.
30345 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
30346 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
30347 * config/sh/sh.c (sh_rtx_costs): Likewise.
30348 * fold-const.c (fold_truth_andor): Likewise.
30349 * genautomata.c (collapse_flag): Likewise.
30350 * gengtype.h (struct type::u::s): Likewise.
30351 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
30352 * input.c (FORMAT_AMOUNT): Likewise.
30353 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
30354 (known_aggs_to_agg_replacement_list): Likewise.
30355 * ipa-inline-analysis.c: Likewise.
30356 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
30357 * ipa-polymorphic-call.c
30358 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
30359 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
30360 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
30362 * modulo-sched.c (apply_reg_moves): Likewise.
30363 * omp-expand.c (build_omp_regions_1): Likewise.
30364 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
30365 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
30366 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
30367 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
30368 * value-prof.c: Likewise.
30369 * var-tracking.c (val_reset): Likewise.
30371 2017-04-03 Richard Biener <rguenther@suse.de>
30373 PR tree-optimization/80275
30374 * fold-const.c (split_address_to_core_and_offset): Handle
30377 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
30379 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
30380 descriptors is at least equal to that of functions.
30382 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30384 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
30386 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30389 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
30390 (mov<IMOD4:mode>): New expander.
30391 (*mov<IMOD4:mode>_internal): New insn and split pattern.
30393 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
30395 PR rtl-optimization/79405
30396 * fwprop.c (propagations_left): New variable.
30397 (forward_propagate_into): Decrement it.
30398 (fwprop_init): Initialize it.
30399 (fw_prop): If the variable has reached zero, stop propagating.
30400 (fwprop_addr): Ditto.
30402 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30405 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
30406 a FUNCTION_DECL, pass it as decl instead of origin to
30409 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
30411 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
30414 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
30417 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
30418 TARGET_VSX_SMALL_INTEGER.
30420 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30422 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30423 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
30425 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30427 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
30428 extraction from odd-numbered MSA register.
30430 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30432 PR middle-end/80173
30433 * expmed.c (store_bit_field_1): Don't attempt to create
30434 a word subreg out of hard registers wider than word if they
30435 have HARD_REGNO_NREGS of 1 for their mode.
30437 PR middle-end/80163
30438 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
30439 conversions to integer types wider than word and pointer.
30442 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
30443 (rtx_equal_for_cselib_p): Pass 0 to it.
30444 * cselib.c (cselib_hasher::equal): Likewise.
30445 (rtx_equal_for_cselib_1): Add depth argument. If depth
30446 is 128, don't look up VALUE locs and punt. Increment
30447 depth in recursive calls when walking VALUE locs.
30449 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
30451 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
30452 (make_gcov_file_name): Use the canonical path name for generating
30454 (read_line): Fix handling of files with ascii null bytes.
30456 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
30458 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
30459 to initialise a vector register instead
30460 of using a const_int.
30462 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30464 PR translation/80189
30465 * gimplify.c (omp_default_clause): Use %qs instead of %s in
30466 diagnostic messages.
30468 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
30471 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
30472 (dfp_diex_<mode>): Update mode of operand 1.
30473 * doc/extend.texi (dxex, dxexq): Document change to return type.
30474 (diex, diexq): Document change to argument type.
30476 2017-03-30 Martin Jambor <mjambor@suse.cz>
30479 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
30480 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
30481 it reflects the signature changes performed at the callee side.
30482 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
30483 to cgraph_build_function_type_skip_args.
30484 (build_function_decl_skip_args): Adjust call to the above function.
30486 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30489 * config/i386/sse.md
30490 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
30491 register as dest whenever it is a MEM not rtx_equal_p to the
30492 corresponding dup operand, and when forcing into reg move the
30493 reg into the memory afterwards.
30494 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
30495 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
30496 for the force_reg mode.
30497 (avx512vl_vextractf128<mode>): Use register as dest either
30498 always when a MEM, or when it is a MEM not rtx_equal_p to the
30499 corresponding dup operand, or even not when it is a CONST_VECTOR
30500 depending on the mode and lo vs. hi.
30501 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
30503 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
30504 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
30505 Likewise. Require that operands[2] is even.
30506 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
30507 Remove extraneous parens. Require that operands[2] is a multiple
30509 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
30510 operands[0] is a MEM if <mask_applied>, the predicates/constraints
30511 disallow memory then.
30513 2017-03-30 Richard Biener <rguenther@suse.de>
30515 PR tree-optimization/77498
30516 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
30517 to non-constants over backedges.
30519 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
30521 PR rtl-optimization/80233
30522 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
30523 as last_combined_insn. Do not test for BARRIER_P separately.
30525 2017-03-29 Andreas Schwab <schwab@suse.de>
30528 * calls.c (prepare_call_address): Convert funexp to Pmode before
30529 copying to temp reg.
30531 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30533 PR tree-optimization/80158
30534 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
30535 Handle possible future case of more than one alternate
30537 (replace_rhs_if_not_dup): Likewise.
30538 (replace_one_candidate): Likewise.
30540 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
30542 PR rtl-optimization/80193
30543 * ira.c (ira): Do not check allocation for LRA.
30545 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
30547 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
30548 (nvptx_output_simt_exit): Declare.
30549 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
30550 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
30551 (init_softstack_frame): Move initialization of crtl->is_leaf to...
30552 (nvptx_declare_function_name): ...here. Emit declaration of local
30553 memory space buffer for omp_simt_enter insn.
30554 (nvptx_output_unisimt_switch): New.
30555 (nvptx_output_softstack_switch): New.
30556 (nvptx_output_simt_enter): New.
30557 (nvptx_output_simt_exit): New.
30558 * config/nvptx/nvptx.h (struct machine_function): New fields
30559 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
30560 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
30561 (UNSPECV_SIMT_EXIT): Ditto.
30562 (omp_simt_enter_insn): New insn.
30563 (omp_simt_enter): New expansion.
30564 (omp_simt_exit): New insn.
30565 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
30567 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
30568 (expand_GOMP_SIMT_ENTER_ALLOC): New.
30569 (expand_GOMP_SIMT_EXIT): New.
30570 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
30571 (GOMP_SIMT_ENTER_ALLOC): Ditto.
30572 (GOMP_SIMT_EXIT): Ditto.
30573 * target-insns.def (omp_simt_enter): New insn.
30574 (omp_simt_exit): Ditto.
30575 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
30577 (lower_rec_simd_input_clauses): Implement SIMT privatization.
30578 (lower_rec_input_clauses): Likewise.
30579 (lower_lastprivate_clauses): Handle SIMT privatization.
30581 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
30582 (ompdevlow_adjust_simt_enter): New.
30583 (find_simtpriv_var_op): New.
30584 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
30585 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
30587 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
30588 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
30589 (copy_decl_for_dup_finish): Ditto.
30591 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
30593 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
30596 * config/i386/i386.c (ix86_option_override_internal): Always
30597 allow -mpreferred-stack-boundary=3 for 64-bit targets.
30599 2017-03-28 Bin Cheng <bin.cheng@arm.com>
30601 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
30603 2017-03-28 Bin Cheng <bin.cheng@arm.com>
30605 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
30606 mark new edge's irreducible flag accordign to it.
30607 (vect_do_peeling): Check loop preheader edge's irreducible flag
30608 and pass it to function slpeel_add_loop_guard.
30610 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
30612 PR tree-optimization/80218
30613 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30614 Update block frequencies and counts.
30616 2017-03-28 Richard Biener <rguenther@suse.de>
30618 PR tree-optimization/78644
30619 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
30620 of a simplification result we may not use it at all.
30622 2017-03-28 Richard Biener <rguenther@suse.de>
30625 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
30626 without arguments, generate default definition of a SSA name.
30628 2017-03-28 Richard Biener <rguenther@suse.de>
30630 PR middle-end/80222
30631 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
30632 TYPE_REF_CAN_ALIAS_ALL references.
30633 * fold-const.c (fold_indirect_ref_1): Likewise.
30635 2017-03-28 Martin Liska <mliska@suse.cz>
30638 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
30639 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
30641 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
30642 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
30644 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
30645 (EXTRA_SPECS): Define.
30646 (SUBTARGET_EXTRA_SPECS): Likewise.
30647 (SUBTARGET_CPP_SPEC): Likewise.
30648 * config/arc/elf.h (EXTRA_SPECS): Renamed to
30649 SUBTARGET_EXTRA_SPECS.
30650 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
30652 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
30654 * config/arc/simdext.md (vst64_insn): Update pattern.
30655 (vld32wh_insn): Likewise.
30656 (vld32wl_insn): Likewise.
30657 (vld64_insn): Likewise.
30658 (vld32_insn): Likewise.
30660 2017-03-28 Marek Polacek <polacek@redhat.com>
30663 * fold-const.c (fold_comparison): Use protected_set_expr_location
30664 instead of SET_EXPR_LOCATION.
30666 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
30668 * tree.c (add_expr): Avoid name lookup warning.
30670 2017-03-27 Jeff Law <law@redhat.com>
30672 PR tree-optimization/80216
30673 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
30674 function name. Limit recursion depth.
30675 (record_temporary_equivalences): Corresponding changes.
30677 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
30679 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
30682 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30685 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
30687 * cfgcleanup.c (reg_note_cfa_p): New array.
30688 (insns_have_identical_cfa_notes): New function.
30689 (old_insns_match_p): Don't cross-jump in between /f
30690 and non-/f instructions. If both i1 and i2 are frame related,
30691 verify all CFA notes, their order and content.
30693 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
30696 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
30697 HImode and SImode with zero extend to DImode to one insn.
30698 (bswap<mode>2_extenddi): Likewise.
30699 (bswapsi2_extenddi): Likewise.
30700 (bswaphi2_extendsi): Likewise.
30701 (bswaphi2): Combine bswap HImode and SImode into one insn.
30702 Separate memory insns from swapping register.
30703 (bswapsi2): Likewise.
30704 (bswap<mode>2): Likewise.
30705 (bswaphi2_internal): Delete, no longer used.
30706 (bswapsi2_internal): Likewise.
30707 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
30708 store, and gpr<-gpr swap insns.
30709 (bswap<mode>2_store): Likewise.
30710 (bswaphi2_reg): Register only splitter, combine with the splitter.
30711 (bswaphi2 splitter): Likewise.
30712 (bswapsi2_reg): Likewise.
30713 (bswapsi2 splitter): Likewise.
30714 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
30715 the insns into load, store, and register/register insns.
30716 (bswapdi2_ldbrx): Likewise.
30717 (bswapdi2_load): Likewise.
30718 (bswapdi2_store): Likewise.
30719 (bswapdi2_reg): Likewise.
30721 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
30723 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
30724 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
30726 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
30729 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
30731 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30732 special handling for target option conflicts between dform
30733 options (-mpower9-dform, -mpower9-dform-vector,
30734 -mpower9-dform-scalar) and -mno-direct-move.
30736 2017-03-27 Richard Biener <rguenther@suse.de>
30738 PR tree-optimization/80181
30739 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
30741 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30743 * config/arc/predicates.md (move_double_src_operand): Replace the
30744 call to move_double_src_operand with a call to address_operand.
30746 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30748 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
30749 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
30750 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
30752 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30754 * config/arc/predicates.md (long_immediate_loadstore_operand):
30755 Consider scaled addresses cases.
30757 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30759 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
30760 restored when in interrupt.
30761 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
30762 doesn't have delay slot.
30764 2017-03-27 Richard Biener <rguenther@suse.de>
30767 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
30768 inlined thunk clones.
30770 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30773 * asan.c (instrument_derefs): Copy over last operand from
30774 original COMPONENT_REF to the new COMPONENT_REF with
30775 DECL_BIT_FIELD_REPRESENTATIVE.
30776 * ubsan.c (instrument_object_size): Likewise.
30778 2017-03-27 Richard Biener <rguenther@suse.de>
30780 PR tree-optimization/80170
30781 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
30782 sure DR/SCEV didnt fold in constants we do not see when looking
30783 at the reference base alignment.
30785 2017-03-27 Richard Biener <rguenther@suse.de>
30787 PR middle-end/80171
30788 * gimple-fold.c (fold_ctor_reference): Properly guard against
30789 NULL return value from canonicalize_constructor_val.
30791 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
30794 * config/i386/i386.c (ix86_expand_builtin)
30795 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
30796 flags reg setting and flags reg using instructions.
30797 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
30798 clobbering instructions to zero extend op2.
30800 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
30802 * doc/install.texi (Configuration) <--with-aix-soname>:
30803 Update link to AIX ld.
30805 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
30807 PR rtl-optimization/80160
30808 PR rtl-optimization/80159
30809 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
30810 reg_alternate_class into account.
30812 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
30815 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
30816 to consider in curr_insn_transform.
30818 2017-03-24 Jakub Jelinek <jakub@redhat.com>
30820 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
30821 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
30822 and emit_mode_inner.
30824 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30826 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
30827 argument to the overloaded builtin variants. Use the new flag to
30828 deprecate certain builtin variants.
30829 * config/s390/s390-builtin-types.def: Add new builtin types.
30830 * config/s390/s390-builtins.h: Support new flags field for
30831 overloaded builtins.
30832 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
30833 (s390_macro_to_expand): Enable vector float data type.
30834 (s390_cpu_cpp_builtins_internal): Indicate support of the new
30835 builtins by incrementing the __VEC__ version number.
30836 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
30838 (s390_resolve_overloaded_builtin): Emit error messages depending
30839 on the builtin flags.
30840 * config/s390/s390.c (s390_expand_builtin): Support additional
30841 flags argument. Change error message to match the messages
30842 emitted in s390-c.c.
30843 * config/s390/s390.md: New UNSPEC_* constants.
30844 (op_type): Add new instruction types.
30845 * config/s390/vecintrin.h: Add new builtins and test data class
30847 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
30848 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
30849 (VEC_INEXACT, VEC_NOINEXACT): New constants.
30850 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
30851 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
30852 ("vec_mergel<mode>"): V_HW -> VEC_HW.
30854 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
30855 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
30856 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
30857 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
30859 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
30860 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
30861 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
30862 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
30864 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
30865 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
30866 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
30867 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
30868 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
30869 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
30870 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
30872 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
30873 ("vec_scatter_element<V_HW_4:mode>_DI")
30874 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
30875 ("vec_fpint<mode>", "vflls")
30876 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
30877 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
30878 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
30879 ("*vec_cmphe<mode>_cc"): ... these.
30881 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
30882 mode constant instead of magic value.
30884 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30886 * config/s390/s390.c (s390_expand_vec_compare): Support other
30887 vector floating point modes than just V2DF.
30888 (s390_expand_vcond): Likewise.
30889 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
30890 (s390_cannot_change_mode_class): Prevent mode changes between TF
30891 and V1TF in vector registers.
30892 * config/s390/s390.md (DF, SF): New mode attributes.
30893 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
30894 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
30895 SFmode support for VRs.
30896 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
30898 (VFT, VF_HW): New mode iterators.
30899 (vw, sdx): New mode attributes.
30900 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
30901 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
30902 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
30903 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
30904 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
30905 also the new vector floating point modes. Renaming to ...
30907 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
30908 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
30909 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
30910 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
30911 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
30912 ("vec_unordered<mode>"): ... these.
30914 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
30915 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
30916 ("*vec_extendv2df"): New insn definitions.
30918 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30920 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
30921 ("mulditi3_2", "*muldi3_sign"): New patterns.
30922 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
30923 rename the pattern definition.
30925 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30927 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
30929 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
30931 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30933 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
30934 instruction if possible.
30935 * config/s390/vector.md (vec_halfnumelts): New mode
30937 ("*vec_vllezlf<mode>"): New pattern.
30939 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30941 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
30942 ("popcountv4si2", "popcountv2di2"): Rename to ...
30943 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
30944 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
30946 ("popcount<mode>2_vxe"): New pattern.
30948 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30950 * common/config/s390/s390-common.c (processor_flags_table): Add
30952 * config.gcc: Add arch12.
30953 * config/s390/driver-native.c (s390_host_detect_local_cpu):
30954 Default to arch12 for unknown CPU model numbers.
30955 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
30956 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
30957 PROCESSOR_max sanity check.
30958 * config/s390/s390-opts.h (enum processor_type): Add
30960 * config/s390/s390.c (processor_table): Add arch12.
30961 (s390_expand_builtin): Add check for B_VXE flag.
30962 (s390_issue_rate): Add PROCESSOR_ARCH12.
30963 (s390_get_sched_attrmask): Likewise.
30964 (s390_get_unit_mask): Likewise.
30965 (s390_sched_score): Enable z13 scheduling for arch12.
30966 (s390_sched_reorder): Likewise.
30967 (s390_sched_variable_issue): Likewise.
30968 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
30970 (s390_tune_attr): Use z13 scheduling also for arch12.
30971 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
30972 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
30973 (TARGET_VXE_P): New macros.
30974 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
30975 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
30976 * config/s390/s390.opt: Add arch12 as processor_type.
30978 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30980 * config/s390/s390.md
30981 ("fixuns_truncdddi2", "fixuns_trunctddi2")
30982 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
30983 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
30985 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
30986 Rename expanders to ...
30988 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
30989 ("fixuns_truncdddi2_emu"): ... these.
30991 ("fixuns_trunc<mode>si2_emu"): New expander.
30993 ("*fixuns_truncdfdi2_z13"): Rename to ...
30994 ("*fixuns_truncdfdi2_vx"): ... this.
30996 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30998 * config/s390/2964.md: Remove the single element vector compare
30999 instructions which are no longer used.
31000 * config/s390/s390.c (s390_select_ccmode): Remove handling of
31002 (s390_canonicalize_comparison): Remove handling of DFmode
31004 (s390_expand_vec_compare_scalar): Remove function.
31005 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
31006 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
31008 ("*cmp<mode>_ccs"): Add wfcdb instruction.
31010 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31012 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
31014 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
31015 will anyway by matched by mov<mode>_64dfp.
31017 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31019 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
31020 vlef/vstef. Add missing operand to vleif.
31022 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31024 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
31025 pair for all vector types with 64 bit elements.
31026 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
31027 * config/s390/vector.md (V_HW_64): ... here.
31028 (V_128_NOSINGLE): New mode iterator.
31029 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
31030 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
31031 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
31032 ("*vec_load_pairv2di"): Change to ...
31033 ("*vec_load_pair<mode>"): ... this one.
31035 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31037 * config/s390/constraints.md: Add comments.
31038 (jKK): Reject element sizes > 8 bytes.
31039 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
31041 * config/s390/s390.md: Add the s_operand checks formerly in
31042 s390_split_ok_p to various splitters where they are still
31044 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
31045 for 128 bit vectors. Plus two splitters.
31047 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31049 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
31052 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31055 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
31056 error if the boundary argument is not constant.
31058 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31060 PR rtl-optimization/80112
31061 * loop-doloop.c (doloop_condition_get): Don't check condition
31062 if cmp isn't SET with IF_THEN_ELSE src.
31064 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31066 PR tree-optimization/80158
31067 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
31068 replacing a candidate statement, also replace it for the
31069 candidate's alternate interpretation.
31070 (replace_rhs_if_not_dup): Likewise.
31071 (replace_one_candidate): Likewise.
31073 2017-03-24 Richard Biener <rguenther@suse.de>
31075 PR tree-optimization/80167
31076 * graphite-isl-ast-to-gimple.c
31077 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
31079 (translate_isl_ast_to_gimple::get_rename): Likewise.
31081 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31083 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
31084 handling of certain combinations of target options, including the
31085 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
31086 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
31088 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31091 * config/arm/arm.md (*load_multiple): Add reload_completed to
31092 matching condition.
31094 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31095 Richard Biener <rguenth@suse.de>
31097 PR tree-optimization/79908
31098 PR tree-optimization/80136
31099 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
31100 been cast away, gimplify_and_add suffices.
31102 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
31104 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
31106 2017-03-23 Richard Biener <rguenther@suse.de>
31108 PR tree-optimization/80032
31109 * gimplify.c (gimple_push_cleanup): Forced unconditional
31110 cleanups still have to go to the conditional_cleanups
31113 2017-03-22 Jakub Jelinek <jakub@redhat.com>
31115 PR tree-optimization/80072
31116 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
31118 (next_operand_entry_id): Change type to unsigned int.
31119 (sort_by_operand_rank): Make sure to return the right return value
31120 even if unsigned fields are bigger than INT_MAX.
31121 (struct oecount): Change cnt and id type to unsigned int.
31122 (oecount_hasher::equal): Formatting fix.
31123 (oecount_cmp): Make sure to return the right return value
31124 even if unsigned fields are bigger than INT_MAX.
31125 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
31128 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
31129 TREE_READONLY on result if writing it more than once.
31132 * doc/invoke.texi (-fsanitize=thread): Document that with
31133 -fnon-call-exceptions atomics are not able to throw
31137 * tsan.c: Include tree-eh.h.
31138 (instrument_builtin_call): Call maybe_clean_eh_stmt or
31139 maybe_clean_or_replace_eh_stmt where needed.
31140 (instrument_memory_accesses): Add cfg_changed argument.
31141 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
31142 if it returned true.
31143 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
31145 PR rtl-optimization/63191
31146 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
31147 wrapper function, moved the whole old content into ...
31148 (ix86_delegitimize_address_1): ... this. New inline function.
31149 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
31150 true as last argument instead of ix86_delegitimize_address.
31152 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
31154 * config/aarch64/aarch64.c (generic_branch_cost): Copy
31155 cortexa57_branch_cost.
31157 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
31159 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
31161 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31164 * doc/md.texi (Constraints): Document wA constraint.
31165 * config/rs6000/constraints.md (wA): New.
31166 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
31167 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
31168 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
31169 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
31171 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
31174 * gimplify.c (is_oacc_declared): New function.
31175 (oacc_default_clause): Use it to set default flags for acc declared
31176 variables inside parallel regions.
31177 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
31178 declared variables.
31179 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
31180 declare attribute to any decl as necessary.
31182 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
31185 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
31186 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
31187 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
31188 (arm_arch_lpae): This.
31189 * config/arm/arm.c (arm_arch7ve): Rename into ...
31190 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
31191 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
31194 2017-03-22 Martin Liska <mliska@suse.cz>
31197 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
31198 error message instead of an ICE.
31200 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31202 * doc/extend.texi (6.11 Additional Floating Types): Revise.
31204 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
31206 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
31208 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31211 2017-03-21 Martin Sebor <msebor@redhat.com>
31213 * doc/extend.texi: Use "cannot" instead of "can't."
31214 * doc/hostconfig.texi: Same.
31215 * doc/install.texi: Same.
31216 * doc/invoke.texi: Same.
31217 * doc/loop.texi: Same.
31218 * doc/md.texi: Same.
31219 * doc/objc.texi: Same.
31220 * doc/rtl.texi: Same.
31221 * doc/tm.texi: Same.
31222 * doc/tm.texi.in: Same.
31223 * doc/trouble.texi: Same.
31225 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
31228 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
31229 (collect_checksum_attributes): Set it.
31230 (die_checksum_ordered): Use it.
31232 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31234 PR tree-optimization/79908
31235 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
31236 change: For a VA_ARG whose LHS has been cast away, use
31237 force_gimple_operand to construct the side effects.
31239 2017-03-21 David Malcolm <dmalcolm@redhat.com>
31241 PR translation/80001
31242 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
31243 more amenable to translation.
31244 (oacc_loop_auto_partitions): Likewise.
31246 2017-03-21 Marek Polacek <polacek@redhat.com>
31247 Martin Sebor <msebor@redhat.com>
31249 PR tree-optimization/80109
31250 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
31251 on INTEGRAL_TYPE_P.
31253 2017-03-21 Jakub Jelinek <jakub@redhat.com>
31254 Segher Boessenkool <segher@kernel.crashing.org>
31257 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
31258 check reg_used_between_p between insn and one of succ or succ2
31259 depending on if succ is artificial insn not inserted into insn
31262 2017-03-21 Martin Liska <mliska@suse.cz>
31264 PR gcov-profile/80081
31265 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
31266 * doc/gcc.texi: Include gcov-dump stuff.
31267 * doc/gcov-dump.texi: New file.
31269 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
31271 PR rtl-optimization/79150
31272 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
31273 conditional jump, if the jump is the last insn of the loop.
31275 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31276 Richard Biener <rguenth@suse.de>
31278 PR tree-optimization/79908
31279 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
31280 been cast away, use force_gimple_operand to construct the side
31283 2017-03-21 Martin Liska <mliska@suse.cz>
31285 PR libfortran/79956
31286 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
31289 2017-03-21 Brad Spengler <spender@grsecurity.net>
31292 * plugin.c (htab_hash_plugin): New function.
31293 (add_new_plugin): Use it and adjust.
31294 (parse_plugin_arg_opt): Adjust.
31295 (init_one_plugin): Likewise.
31297 2017-03-21 Richard Biener <rguenther@suse.de>
31299 PR tree-optimization/80032
31300 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
31301 if set force the cleanup to happen unconditionally.
31302 (gimplify_target_expr): Push inserted clobbers with force_uncond
31303 to avoid them being removed by control-dependent DCE.
31305 2017-03-21 Richard Biener <rguenther@suse.de>
31307 PR tree-optimization/80122
31308 * tree-inline.c (copy_bb): Do not expans va-arg packs or
31309 va_arg_pack_len when the inlined call stmt requires pack
31311 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
31313 2017-03-21 Jakub Jelinek <jakub@redhat.com>
31316 * tsan.c (instrument_builtin_call): If the memory model argument
31317 is not a constant, assume it is valid.
31320 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
31323 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
31325 PR rtl-optimization/79910
31326 * combine.c (can_combine_p): Do not allow combining an I0 or I1
31327 if its dest is used by an insn before I2 (other than the combined
31328 insns themselves, which are properly handled already).
31330 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
31333 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31335 * combine.c (record_used_regs): New static function.
31336 (try_combine): Handle situations where there is an additional
31337 instruction between I2 and I3 which needs to have a LOG_LINK
31341 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31343 * combine.c (try_combine): Delete redundant i1 test. Call
31344 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31346 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31349 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
31350 alternatives 13/14.
31352 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31354 PR tree-optimization/80054
31355 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
31356 the optimization if a PHI or any of its arguments is not dominated
31357 by the candidate's basis. Use gphi* rather than gimple* as
31359 (replace_profitable_candidates): Clean up a gimple* variable that
31360 should be a gphi* variable.
31362 2017-03-20 Martin Sebor <msebor@redhat.com>
31365 * doc/extend.texi (attribute constructor): Document present limitation.
31367 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
31370 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
31371 __POWER9_VECTOR__ #ifdef control, change template definition to
31372 use Power9-specific built-in function.
31373 (vec_any_eq): Likewise.
31374 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
31375 to control outcomes from this test.
31376 (vector_ae_<mode>p): For VEC_F modes, likewise.
31378 2017-03-20 Ian Lance Taylor <iant@google.com>
31380 * config/i386/i386.c (ix86_function_regparm): Save an extra
31381 register for -fsplit-stack with DECL_STATIC_CHAIN.
31383 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31386 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
31387 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
31389 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31391 * config/riscv/riscv.c (riscv_print_operand): Use "fence
31393 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
31396 2017-03-20 Marek Polacek <polacek@redhat.com>
31399 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
31401 2017-03-20 Richard Biener <rguenther@suse.de>
31403 PR tree-optimization/80113
31404 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
31405 allocate extra SSA name for PHI def.
31406 (add_close_phis_to_outer_loops): Likewise.
31407 (add_close_phis_to_merge_points): Likewise.
31408 (copy_loop_close_phi_args): Likewise.
31409 (copy_cond_phi_nodes): Likewise.
31411 2017-03-20 Martin Liska <mliska@suse.cz>
31413 PR middle-end/79753
31414 * tree-chkp.c (chkp_build_returned_bound): Do not build
31415 returned bounds for a LHS that's not a BOUNDED_P type.
31417 2017-03-20 Martin Liska <mliska@suse.cz>
31421 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
31422 COMPLEX_CST and VECTOR_CST.
31424 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31427 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
31428 target operand. A new splitter adds the clobber statement in case
31429 the target operand is dead anyway.
31431 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
31433 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
31434 to age-old versions of binutils and glibc.
31436 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
31438 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
31440 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31442 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
31444 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31446 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
31447 requirement for binutils 2.13.
31449 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31451 * combine.c (try_combine): Delete redundant i1 test. Call
31452 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31454 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
31456 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
31457 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
31459 <riscv64-*-elf>: Re-arrange section
31460 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
31461 <riscv32-*-linux>: Likewise.
31462 <riscv64-*-elf>: Likewise
31463 <riscv64-*-linux>: Likewise.
31465 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
31468 * aarch64.opt(verbose-cost-dump): Fix typo.
31470 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
31473 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
31474 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
31476 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31478 * reload.c (find_reloads): When reloading a nonoffsettable address,
31479 use RELOAD_OTHER for it and its address reloads.
31481 PR rtl-optimization/79910
31482 * combine.c (record_used_regs): New static function.
31483 (try_combine): Handle situations where there is an additional
31484 instruction between I2 and I3 which needs to have a LOG_LINK
31487 2017-03-17 Jeff Law <law@redhat.com>
31489 PR tree-optimization/71437
31490 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
31491 conditional in the hash table first.
31492 (vrp_dom_walker::before_dom_children): Extract condition from
31493 ASSERT_EXPR. Record condition, its inverion and any implied
31494 conditions as well.
31496 2017-03-17 Marek Polacek <polacek@redhat.com>
31497 Markus Trippelsdorf <markus@trippelsdorf.de>
31499 PR tree-optimization/80079
31500 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
31503 2017-03-17 Richard Biener <rguenther@suse.de>
31505 PR middle-end/80075
31506 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
31507 Properly verify the LHS before the RHS possibly claims to be
31509 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
31512 2017-03-17 Martin Jambor <mjambor@suse.cz>
31514 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
31515 (List of -O2 options): Likewise.
31516 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
31519 2017-03-17 Tom de Vries <tom@codesourcery.com>
31521 * gcov-dump.c (print_usage): Print bug_report_url.
31523 2017-03-17 Richard Biener <rguenther@suse.de>
31525 PR middle-end/80050
31526 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
31527 (parser::peek): Likewise.
31529 2017-03-17 Richard Biener <rguenther@suse.de>
31531 PR tree-optimization/80048
31532 * sese.c (free_sese_info): Properly release rename_map and
31533 copied_bb_map elements.
31535 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
31537 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
31538 Add linked-list forward and backlinks. Insert on
31539 construction, remove on destruction.
31540 (class pass_store_merging): Add m_stores_head field.
31541 (pass_store_merging::terminate_and_process_all_chains):
31542 Iterate over m_stores_head list.
31543 (pass_store_merging::terminate_all_aliasing_chains):
31545 (pass_store_merging::execute): Check for debug stmts first.
31546 Push new chains onto the m_stores_head stack.
31548 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
31551 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
31552 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
31553 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
31555 2017-03-16 Jeff Law <law@redhat.com>
31557 PR tree-optimization/71437
31558 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
31559 member function. Implementation moved into after_dom_children
31560 member function and into the threader's thread_outgoing_edges
31562 (dom_opt_dom_walker::after_dom_children): Simplify by moving
31563 some code into new thread_outgoing_edges.
31564 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
31565 definition. Simplify marker handling (do it here). Assume we always
31566 have the available expression and the const/copies tables.
31567 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
31569 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
31570 * tree-vrp.c (equiv_stack): No longer file scoped.
31571 (vrp_dom_walker): New class.
31572 (vrp_dom_walker::before_dom_children): New member function.
31573 (vrp_dom_walker::after_dom_children): Likewise.
31574 (identify_jump_threads): Setup domwalker. Use it rather than
31575 walking edges in a random order by hand. Simplify setup/finalization.
31576 (finalize_jump_threads): Remove.
31577 (vrp_finalize): Do not call identify_jump_threads here.
31578 (execute_vrp): Do it here instead and call thread_through_all_blocks
31581 PR tree-optimization/71437
31582 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
31584 (simplify_stmt_for_jump_threading): Add basic_block argument. All
31586 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
31587 (dom_opt_dom_walker::thread_across_edge): Remove
31588 handle_dominating_asserts argument. All callers changed.
31589 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
31590 changes. Remove calls to lhs_of_dominating_assert. Other
31591 uses of handle_dominating_asserts turn into unconditional code
31592 (simplify_control_stmt_condition_1): Likewise.
31593 (simplify_control_stmt_condition): Likewise.
31594 (thread_through_normal_block, thread_across_edge): Likewise.
31595 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
31596 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
31597 object if it is not an SSA_NAME.
31598 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
31599 before calling into the VRP specific simplifiers.
31600 (identify_jump_threads): Remove handle_dominating_asserts
31603 2017-03-16 Jakub Jelinek <jakub@redhat.com>
31606 * tree-diagnostic.c (default_tree_printer): No longer static.
31607 * tree-diagnostic.h (default_tree_printer): New prototype.
31609 2017-03-16 Tamar Christina <tamar.christina@arm.com>
31611 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
31612 Change ins into fmov.
31614 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31616 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
31617 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
31618 Use h_con constraint for operand 1.
31619 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
31620 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
31622 2017-03-15 Jeff Law <law@redhat.com>
31624 PR tree-optimization/71437
31625 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
31626 (record_temporary_equivalences): Use it.
31628 PR tree-optimization/71437
31629 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
31630 tree-ssa-scopedtables.
31631 (lookup_avail_expr, build_and_record_new_cond): Likewise.
31632 (record_conditions, record_cond, vuse_eq): Likewise.
31633 (record_edge_info): Adjust to API tweak of record_conditions.
31634 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
31635 (record_temporary_equivalences, optimize_stmt): Likewise.
31636 (eliminate_redundant_computations): Likewise.
31637 (record_equivalences_from_stmt): Likewise.
31638 * tree-ssa-scopedtables.c: Include options.h and params.h.
31639 (vuse_eq): New function, moved from tree-ssa-dom.c
31640 (build_and_record_new_cond): Likewise.
31641 (record_conditions): Likewise. Accept vector of conditions rather
31642 than edge_equivalence structure for first argument.
31643 for the first argument.
31644 (avail_exprs_stack::lookup_avail_expr): New member function, moved
31645 from tree-ssa-dom.c.
31646 (avail_exprs_stack::record_cond): Likewise.
31647 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
31648 from tree-ssa-dom.c.
31649 (avail_exprs_stack): Add new member functions lookup_avail_expr
31651 (record_conditions): Declare.
31653 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
31656 * lra-constraints.c (process_alt_operands): Increase reject for
31657 reloading an input/output operand.
31659 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
31662 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
31663 insns to convert from signed/unsigned char/short to IEEE 128-bit
31665 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
31667 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
31670 * config/i386/i386.c (ix86_vector_duplicate_value): Create
31671 subreg of inner mode for values already in registers.
31673 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
31675 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
31676 iteration reg is used after the loop.
31678 2017-03-14 Martin Sebor <msebor@redhat.com>
31680 PR tree-optimization/79800
31681 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
31682 precision in negative-positive range.
31683 (format_floating): Call non-const overload with adjusted precision.
31685 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
31688 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
31691 2017-03-14 Martin Sebor <msebor@redhat.com>
31693 PR middle-end/80020
31694 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
31695 * builtins.def (aligned_alloc): Use it.
31698 * Makefile.in (GTFILES): Add calls.c.
31699 * calls.c: Include "gt-calls.h".
31701 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
31703 PR rtl-optimization/79728
31704 * regs.h (struct target_regs): New field
31705 x_contains_allocatable_regs_of_mode.
31706 (contains_allocatable_regs_of_mode): New macro.
31707 * reginfo.c (init_reg_sets_1): Initialize it, and change
31708 contains_reg_of_mode so it includes global regs as well.
31709 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
31710 rather than contains_regs_of_mode.
31712 2017-03-14 Martin Liska <mliska@suse.cz>
31714 * doc/invoke.texi: Document options that can't be combined with
31715 -fcheck-pointer-bounds.
31717 2017-03-14 Martin Liska <mliska@suse.cz>
31719 PR middle-end/79831
31720 * doc/invoke.texi (-Wchkp): Document the option.
31722 2017-03-14 Martin Liska <mliska@suse.cz>
31724 * Makefile.in: Install gcov-dump.
31726 2017-03-14 Martin Liska <mliska@suse.cz>
31728 * multiple_target.c (expand_target_clones): Bail out for
31729 an invalid attribute.
31731 2017-03-14 Richard Biener <rguenther@suse.de>
31733 * alias.c (struct alias_set_entry): Pack properly.
31734 * cfgloop.h (struct loop): Likewise.
31735 * cse.c (struct set): Likewise.
31736 * ipa-utils.c (struct searchc_env): Likewise.
31737 * loop-invariant.c (struct invariant): Likewise.
31738 * lra-remat.c (struct cand): Likewise.
31739 * recog.c (struct change_t): Likewise.
31740 * rtl.h (struct address_info): Likewise.
31741 * symbol-summary.h (function_summary): Likewise.
31742 * tree-loop-distribution.c (struct partition): Likewise.
31743 * tree-object-size.c (struct object_size_info): Likewise.
31744 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
31745 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
31746 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
31747 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
31748 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
31749 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
31750 (struct _stmt_vec_info): Likewise.
31752 2017-03-14 Martin Liska <mliska@suse.cz>
31755 * multiple_target.c (create_dispatcher_calls): Check that
31756 a target can create a function dispatcher.
31758 2017-03-14 Martin Liska <mliska@suse.cz>
31761 * multiple_target.c (expand_target_clones): Drop local.local
31762 flag for default implementation.
31764 2017-03-14 Richard Biener <rguenther@suse.de>
31766 PR tree-optimization/80030
31767 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
31769 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
31771 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
31772 gcc_fallthrough() instead of __attribute__((fallthrough));
31774 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31776 * doc/gcc.texi: Remove "up" link to (DIR).
31777 * doc/gccint.texi: Ditto.
31779 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31781 * doc/install.texi (Specific) <avr>: Remove reference to
31784 2017-03-13 Jeff Law <law@redhat.com>
31786 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
31787 attribute rather than comments.
31789 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
31790 match_scratch operand is highest.
31792 2017-03-13 Martin Liska <mliska@suse.cz>
31794 PR middle-end/78339
31795 * ipa-pure-const.c (warn_function_noreturn): If the declarations
31796 is a CHKP clone, use original declaration.
31798 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31800 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
31801 (arc_conditional_register_usage): Use a different allocation order
31802 when optimizing for size.
31803 * common/config/arc/arc-common.c (arc_option_optimization_table):
31804 Section anchors default on when optimizing for size.
31806 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31808 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
31810 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31812 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
31813 * config/arc/arc.md (cpu_facility): Add cd variant.
31814 (*movqi_insn): Add code density variant.
31815 (*movhi_insn): Likewise.
31816 (*movqi_insn): Likewise.
31817 (*addsi3_mixed): Likewise.
31818 (subsi3_insn): Likewise.
31820 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31822 * config/arc/arc.md (movsi_cond_exec): Update constraint.
31824 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31826 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
31827 expressions with MINUS and UNARY ops.
31829 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31832 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
31834 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
31835 between vec_select and vector argument.
31836 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
31837 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
31838 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
31839 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
31840 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
31841 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
31843 2017-03-13 Richard Biener <rguenther@suse.de>
31846 * params.def (vect-max-peeling-for-alignment): Fix typo.
31848 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31850 * doc/install.texi (Specific) <mips-*-*>: Remove description of
31851 issue that only occurred with binutils below 2.18.
31853 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31855 * doc/install.texi (Specific) <cris-axis-elf>: No longer
31856 refer to binutils 2.11/2.12 minimum.
31858 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31860 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
31861 ftp.kernel.org and simplify binutils requirement.
31863 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
31865 * doc/invoke.texi (Warning Options): Fix spelling of link-time
31867 (Optimize Options): Ditto. Also remove redundancy.
31869 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31871 PR translation/79848
31872 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
31874 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
31875 to G_ to avoid double translation.
31877 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31879 PR translation/79923
31880 * auto-profile.c (get_combined_location): Convert leading
31881 character of diagnostics to lower case and remove trailing period.
31882 (read_profile): Likewise for various diagnostics.
31883 * config/arm/arm.c (arm_option_override): Remove trailing period
31884 from various diagnostics.
31885 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
31886 (msp430_expand_delay_cycles): Likewise.
31888 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31891 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
31892 full command-line argument, rather than just "str".
31893 (aarch64_validate_march): Likewise.
31894 (aarch64_validate_mtune): Likewise.
31896 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
31898 PR rtl-optimization/78911
31899 * lra-assigns.c (must_not_spill_p): New function.
31900 (spill_for): Use it.
31902 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31904 PR tree-optimization/79981
31905 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
31906 ATOMIC_COMPARE_EXCHANGE ifn result.
31907 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
31908 IFN_ATOMIC_COMPARE_EXCHANGE.
31910 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31913 * opts.c (parse_sanitizer_options): Add missing question mark to
31914 "did you mean" message.
31916 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31918 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
31920 (VMULEUH_UNS): Likewise.
31921 (VMULOUB_UNS): Likewise.
31922 (VMULOUH_UNS): Likewise.
31923 * config/rs6000/rs6000.c (builtin_function_type): Remove
31924 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
31926 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31929 * read-rtl-function.c (function_reader::read_rtx_operand): Update
31930 x with result of extra_parsing_for_operand_code_0.
31931 (function_reader::extra_parsing_for_operand_code_0): Convert
31932 return type from void to rtx, returning x. When reading
31933 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
31934 larger size containing struct block_symbol.
31936 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
31938 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
31939 -mfloat128-hardware without -m64.
31941 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
31944 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
31945 entries to the case statement that marks unsigned arguments to
31946 overloaded functions.
31948 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31950 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
31951 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
31953 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
31956 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
31957 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
31959 2017-03-10 Martin Liska <mliska@suse.cz>
31963 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
31964 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
31967 2017-03-10 Olivier Hainque <hainque@adacore.com>
31969 * tree-switch-conversion (array_value_type): Start by resetting
31970 candidate type to it's main variant.
31972 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31974 PR rtl-optimization/79909
31975 * combine.c (try_combine): Use simplify_replace_rtx on individual
31976 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
31977 of the whole CALL_INSN_FUNCTION_USAGE.
31979 PR tree-optimization/79972
31980 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
31981 get_range_info on SSA_NAMEs. Formatting fixes.
31983 2017-03-10 Richard Biener <rguenther@suse.de>
31984 Jakub Jelinek <jakub@redhat.com>
31986 PR tree-optimization/77975
31987 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
31988 edge to be constant.
31989 (get_val_for): For constant x return it. Formatting fix.
31990 (loop_niter_by_eval): Avoid pointless looping if the next iteration
31991 would use the same bases as the current one.
31993 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31995 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
31996 instead of vec_select for V1TImode.
31997 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
31999 (VSX_LE_128): Add V1TI to this mode iterator.
32000 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
32001 (*vsx_le_perm_store_<mode>): Likewise.
32002 (pre-reload splitter for VSX stores): Likewise.
32003 (post-reload splitter for VSX stores): Likewise.
32004 (*vsx_xxpermdi2_le_<mode>): Likewise.
32005 (*vsx_lxvd2x2_le_<mode>): Likewise.
32006 (*vsx_stxvd2x2_le_<mode>): Likewise.
32008 2017-03-09 Michael Eager <eager@eagercon.com>
32010 Correct failures with --enable-checking=yes,rtl.
32012 * config/microblaze/microblaze.c (microblaze_expand_shift):
32013 Replace GET_CODE test with CONST_INT_P and INTVAL test with
32014 test for const0_rtx.
32015 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
32016 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
32018 2017-03-09 Richard Biener <rguenther@suse.de>
32020 PR tree-optimization/79977
32021 * graphite-scop-detection.c (scop_detection::merge_sese):
32022 Handle the case of extra exits to blocks dominating the entry.
32024 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
32026 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
32029 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
32031 PR rtl-optimization/79949
32032 * lra-constraints.c (process_alt_operands): Check memory when
32033 trying to predict a cycle. Print about the overall increase.
32035 2017-03-09 Richard Biener <rguenther@suse.de>
32037 PR middle-end/79971
32038 * gimple-expr.c (useless_type_conversion_p): Preserve
32039 TYPE_SATURATING for fixed-point types.
32041 2017-03-09 Richard Biener <rguenther@suse.de>
32044 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
32045 alignment of BLKmode params.
32047 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32050 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
32051 (VALL_NO_V2Q): Likewise.
32053 * config/aarch64/aarch64-simd.md
32054 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
32056 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
32057 VALL_NO_V2Q mode iterator.
32058 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
32060 2017-03-09 Martin Liska <mliska@suse.cz>
32062 PR tree-optimization/79631
32063 * tree-chkp-opt.c (chkp_is_constant_addr): Call
32064 tree_int_cst_sign_bit just for INTEGER constants.
32066 2017-03-09 Martin Liska <mliska@suse.cz>
32070 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
32073 2017-03-09 Marek Polacek <polacek@redhat.com>
32076 * tree.c (inchash::add_expr): Handle TREE_VEC.
32078 2017-03-09 Martin Liska <mliska@suse.cz>
32081 (chkp_narrow_size_and_offset): New function.
32082 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
32083 (void chkp_parse_bit_field_ref): New function.
32084 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
32085 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
32087 2017-03-09 Martin Liska <mliska@suse.cz>
32090 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
32091 (chkp_find_bounds_1): Remove gcc_unreachable.
32093 2017-03-09 Jakub Jelinek <jakub@redhat.com>
32096 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
32097 BUILT_IN_SYNC*, determine the access type from the size suffix and
32098 always build a MEM_REF with that type. Handle forgotten
32099 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
32102 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
32103 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
32104 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
32105 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
32106 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
32107 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
32108 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
32109 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
32110 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
32111 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
32112 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
32113 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
32114 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
32115 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
32116 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
32117 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
32118 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
32119 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
32120 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
32121 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
32122 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
32123 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
32124 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
32125 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
32126 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
32127 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
32128 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
32129 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
32130 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
32131 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
32132 definitions outside of __OPTIMIZE__ guarded section.
32135 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
32136 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
32137 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
32138 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
32141 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32143 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
32144 ("vfenez<mode>"): Add missing constraints.
32146 2017-03-08 Martin Sebor <msebor@redhat.com>
32149 * config/nds32/nds32.c (nds32_option_override):
32150 Fix misspelled diagnostic.
32152 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32155 * gimplify.c (gimplify_omp_for): Replace index var in outer
32156 taskloop statement with an artificial variable and add
32157 OMP_CLAUSE_PRIVATE clause for it.
32159 2017-03-08 Richard Biener <rguenther@suse.de>
32161 PR tree-optimization/79955
32162 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
32163 for accesses that are completely outside of the variable.
32165 2017-03-08 Andrew Haley <aph@redhat.com>
32167 PR tree-optimization/79943
32168 * tree-ssa-loop-split.c (compute_new_first_bound): When
32169 calculating the new upper bound, (END-BEG) should be added, not
32172 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32174 * config/avr/avr.md (setmemhi): Make sure match_dup
32175 operand number comes before match_scratch.
32177 2017-03-08 Richard Biener <rguenther@suse.de>
32179 PR tree-optimization/79920
32180 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
32181 with ncopies == 1 to ...
32182 (vect_transform_slp_perm_load): ... here. Properly compute
32183 all element loads by iterating VF times over the group. Do
32184 not handle ncopies (computed in a broken way) in
32185 vect_create_mask_and_perm.
32187 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32190 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
32191 is a uniform vector, use uniform_vector_p return value instead of
32192 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
32194 2017-03-07 Marek Polacek <polacek@redhat.com>
32196 PR middle-end/79809
32197 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
32198 (alloca_call_type): Likewise.
32200 2017-03-07 Martin Liska <mliska@suse.cz>
32202 * gcov.c (process_args): Put comment to correct location.
32204 2017-03-07 Martin Liska <mliska@suse.cz>
32206 PR middle-end/68270
32207 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
32208 Use array_at_struct_end_p instead of DECL_CHAIN (field).
32209 (chkp_narrow_bounds_for_field): Likewise.
32210 (chkp_parse_array_and_component_ref): Pass one more argument to
32213 2017-03-07 Richard Biener <rguenther@suse.de>
32215 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
32218 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
32220 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
32221 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
32223 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32226 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
32227 to end of description.
32228 (PARAM_MAX_STORES_TO_MERGE): Likewise.
32230 2017-03-07 Jakub Jelinek <jakub@redhat.com>
32232 PR rtl-optimization/79901
32233 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
32235 (*avx512f_<code><mode>3<mask_name>): ... this.
32236 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
32237 iterator instead of VI8_AVX2_AVX512BW.
32239 PR rtl-optimization/79901
32240 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
32241 min/max expander, expand it using expand_vec_cond_expr.
32244 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
32247 2017-03-06 Jakub Jelinek <jakub@redhat.com>
32250 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
32251 to void * for PCH reasons.
32252 * dwarf2out.c (output_loc_operands, output_die): Cast
32253 v.val_vec.array to unsigned char *.
32255 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
32258 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
32261 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
32263 PR rtl-optimization/79571
32264 * lra-constraints.c (process_alt_operands): Calculate static
32265 reject and subtract it from overall when only addresses will be
32268 2017-03-06 Julia Koval <julia.koval@intel.com>
32271 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
32272 incoming stack boundary to 128 for 64-bit targets.
32274 2017-03-06 Richard Biener <rguenther@suse.de>
32276 PR tree-optimization/79894
32277 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
32278 to NULL after folding it.
32280 2017-03-06 Richard Biener <rguenther@suse.de>
32282 PR tree-optimization/79824
32283 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
32284 check disabling peeling for gaps.
32286 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
32288 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
32289 attributes): Document gettimeofday.
32291 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
32293 * config/s390/s390.c (s390_option_override_internal): Set
32294 PARAM_MIN_VECT_LOOP_BOUND
32296 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
32298 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
32299 * config/s390/s390.md: Likewise.
32301 2017-03-06 Jakub Jelinek <jakub@redhat.com>
32304 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
32305 (<avx2_avx512>_perm<mode>): Rename to ...
32306 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
32308 (<avx512>_perm<mode>_mask): Rename to ...
32309 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
32311 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
32312 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
32313 instead of VI8F_256_512.
32314 (avx512f_perm<mode>): New define_expand.
32315 (avx512f_perm<mode>_mask): Likewise.
32316 (avx512f_perm<mode>_1<mask_name>): New define_insn.
32317 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
32319 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32321 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
32322 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
32324 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
32326 2017-03-06 Martin Liska <mliska@suse.cz>
32329 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
32330 when having a SSA NAME w/o VAR_DECL assigned to it.
32332 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32334 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
32335 msa_dpsub_<su>_d): Fix MODE for vec_select.
32337 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32339 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
32341 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
32343 2017-03-06 Richard Biener <rguenther@suse.de>
32345 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
32346 * plugin.c (register_plugin_info): Likewise.
32347 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
32349 2017-03-05 Jakub Jelinek <jakub@redhat.com>
32351 * config/i386/sse.md (sse_storehps, sse_storelps,
32352 avx_<castmode><avxsizesuffix>_<castmode>,
32353 avx512f_<castmode><avxsizesuffix>_<castmode>,
32354 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
32355 in condition that at least one operand is not a MEM.
32357 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32359 PR middle-end/79805
32360 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
32361 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
32363 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
32364 gimple_call_nothrow_p flag based on whether original builtin can throw.
32365 If it can, emit following stmts on the fallthrough edge.
32366 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
32367 don't create new bb if inserting just debug stmts on the edge, try to
32368 insert them on the fallthru bb or just reset debug stmts.
32370 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
32373 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
32374 restore recog_data (including the operand rtxes inside it) around
32375 the call to get_insn_template.
32377 2017-03-03 Martin Sebor <msebor@redhat.com>
32379 PR tree-optimization/79699
32380 * context.c (context::~context): Free MPFR caches to avoid
32381 a memory leak on program exit.
32383 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32385 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
32386 Use wide_int::ulow () instead of .elt (0).
32388 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32390 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
32391 (*pushxf): Limit oF constraint to 32bit targets and add oC
32392 constraint for 64bit targets.
32393 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
32394 (*pushdf): Change rmF constraint to rmC.
32396 2017-03-03 Martin Liska <mliska@suse.cz>
32398 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
32399 Remove unused variable.
32401 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32404 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
32405 is a memory operand, increase num_memory.
32406 (ix86_expand_args_builtin): Likewise.
32408 2017-03-03 Jan Hubicka <jh@suse.cz>
32411 * ipa-devirt.c (maybe_record_node): Properly handle
32412 __cxa_pure_virtual visibility.
32414 2017-03-03 Martin Liska <mliska@suse.cz>
32416 PR tree-optimization/79803
32417 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
32419 (pass_loop_prefetch::execute): Disabled optimization if an
32420 assumption about L1 cache size is not met.
32422 2017-03-03 Martin Liska <mliska@suse.cz>
32424 PR rtl-optimization/79574
32425 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
32426 (hash_scan_set): Likewise.
32427 (dump_hash_table): Likewise.
32428 (hoist_code): Likewise.
32430 2017-03-03 Richard Biener <rguenther@suse.de>
32432 * fixed-value.c (fixed_from_string): Restore use of elt (1)
32433 in place of uhigh ().
32434 (fixed_convert_from_real): Likewise.
32436 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32439 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
32441 2017-03-03 Richard Biener <rguenther@suse.de>
32443 PR middle-end/79818
32444 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
32445 TYPE_OVERFLOW_UNDEFINED check.
32447 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32449 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
32451 (vector_ae_<mode>_p): Likewise.
32452 (vector_nez_<mode>_p): Likewise.
32453 (vector_ne_v2di_p): Likewise.
32454 (vector_ae_v2di_p): Likewise.
32455 (vector_ne_<mode>_p): Likewise.
32456 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
32458 (vsx_tsqrt<mode>2_fe): Likewise.
32460 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
32463 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
32465 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32467 PR rtl-optimization/79780
32468 * cprop.c (one_cprop_pass): When second and further conditional trap
32469 in a single basic block is turned into an unconditional trap, turn it
32470 into a deleted note to avoid RTL verification failures.
32472 2017-03-02 Richard Biener <rguenther@suse.de>
32474 * fold-const.c (const_binop): Use ulow () instead of elt (0).
32476 2017-03-02 Richard Biener <rguenther@suse.de>
32478 PR tree-optimization/79345
32480 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
32481 param and abort the walk, returning -1 if it is hit.
32482 (walk_aliased_vdefs): Take a limit param and pass it on.
32483 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
32484 defaulting to 0 and return a signed int.
32485 * tree-ssa-uninit.c (struct check_defs_data): New struct.
32486 (check_defs): New helper.
32487 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
32488 about uninitialized memory.
32489 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
32490 bogus uninitialized warning.
32491 (fixed_convert_from_real): Likewise.
32493 2017-03-02 Bin Cheng <bin.cheng@arm.com>
32495 PR tree-optimization/66768
32496 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
32497 iv_use if base object can't be determined.
32499 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32501 PR tree-optimization/79345
32502 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
32503 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
32504 (get_pattern_stats): Initialize it.
32505 * genemit.c (gen_expand): Verify match_scratch numbers come after
32506 match_operand/match_dup numbers.
32507 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
32508 match_scratch numbers.
32509 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
32511 * config/s390/s390.md (trunctdsd2): Likewise.
32513 2017-03-02 Richard Biener <rguenther@suse.de>
32515 * wide-int.h (wide_int_storage::operator=): Implement in terms
32518 2017-03-02 Richard Biener <rguenther@suse.de>
32520 PR tree-optimization/79777
32521 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
32522 the to insert expression to sth existing.
32524 2017-03-01 Martin Sebor <msebor@redhat.com>
32526 PR middle-end/79692
32527 * gimple-ssa-sprintf.c
32528 (directive::known_width_and_precision): New function.
32529 (format_integer): Use it.
32530 (get_mpfr_format_length): Consider the full range of precision
32531 when computing %g output with the # flag. Set the likely byte
32532 count to 3 rather than 1 when precision is indeterminate.
32533 (format_floating): Correct the lower bound of precision.
32535 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32537 * doc/invoke.texi: Document default code model for 64-bit Linux.
32539 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32542 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
32543 udiv rather than div since input pattern is unsigned.
32545 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
32547 * config/i386/i386.c (print_reg): Warn for values of
32548 unsupported size in integer register.
32550 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
32553 * config/rs6000/predicates.md (current_file_function_operand): Do
32554 not allow self calls to be local if the function is replaceable.
32556 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
32559 * config/rs6000/altivec.h (vec_ctz and others): Change the
32560 preprocessor macro that controls conditional compilation from
32561 _ARCH_PWR9 to __POWER9_VECTOR__.
32562 (vec_all_ne): Change parameterization of __altivec_scalar_pred
32563 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
32564 control (instead of _ARCH_PWR9 control) so that template
32565 definition uses power9-specific function.
32566 (vec_any_eq): Likewise.
32567 (vec_all_ne): Change macro definition to use a power9-specific
32568 expansion under #ifdef __POWER9_VECTOR__ control (instead of
32569 _ARCH_PWR9 control).
32570 (vec_any_eq) Likewise.
32571 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
32572 expansion for CMPNEF to remove support for xvcmpnesp instruction.
32573 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
32574 support for xvcmpnedp instruction.
32575 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
32576 macro expansion so that Power9 implementation of vec_all_ne does
32577 not use the AltiVec predicate framework.
32578 (VCMPNEH_P): Likewise.
32579 (VCMPNEW_P): Likewise.
32580 (VCMPNED_P): Likewise.
32581 (VCMPNEFP_P): Likewise.
32582 (VCMPNEDP_P): Likewise.
32583 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
32584 implementation of vec_any_eq to not use AltiVec predicate
32586 (VCMPAEH_P): Likewise.
32587 (VCMPAEW_P): Likewise.
32588 (VCMPAED_P): Likewise.
32589 (VCMPAEFP_P): Likewise.
32590 (VCMPAEDP_P): Likewise.
32591 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
32592 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
32593 not use the AltiVec predicate framework.
32594 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
32595 of vec_any_eq to not use AltiVec predicate framework.
32596 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32597 support for predefined __POWER9_VECTOR__ macro to indicate that
32598 Power9 instruction selection is enabled.
32599 (altivec_overloaded_builtins): Remove extraneous
32600 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
32601 function argument types RS6000_BTI_bool_V16QI and
32602 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
32603 entry for overloaded function argument types RS6000_BTI_bool_V4SI
32604 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
32605 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
32606 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
32607 Power9 for implementations of vec_cmpne. Change the signature for
32608 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
32609 (representing vec_all_ne) to remove the previously described first
32610 argument of type RS6000_BTI_INTSI, as this was an artifact of
32611 reliance on the AltiVec predicate framework, which is no longer
32612 used in the implementation of these functions. Add
32613 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
32614 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
32615 since, unlike the AltiVec predicate framework implementation, we
32616 do not share function descriptors between vec_alle and vec_anyeq.
32617 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
32618 set of modes that receive special treatment even when
32619 TARGET_P9_VECTOR is true. The special treatment emits code that
32620 does not depend on Power9 instructions.
32621 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
32622 define_expand to not rely on AltiVec predicate framework.
32623 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
32625 (vector_ne_v2di_p): Change this define_expand to not rely on
32626 AltiVec predicate framework.
32627 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
32629 (vector_ne_<mode>_p): Change this define_expand to not rely on
32630 AltiVec predicate framework.
32631 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
32633 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
32634 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
32635 define_insn pattern.
32636 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
32637 define_insn pattern because the xvcmpne<VSs>. instruction is not
32639 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
32640 instruction is not supported.
32642 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32644 * config/nvptx/nvptx.c: Include intl.h.
32646 2017-03-01 Martin Jambor <mjambor@suse.cz>
32649 * ipa-prop.h (ipa_bits): Removed field known.
32650 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
32651 to pointers. Adjusted their comments to warn about their sharing.
32652 (ipcp_transformation_summary): Change bits to a vector of pointers.
32653 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
32654 (ipa_get_ipa_bits_for_value): Declare.
32655 * tree-vrp.h (value_range): Mark as GTY((for_user)).
32656 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
32657 (ipa_bits_hash_table): Likewise.
32658 (ipa_vr_ggc_hash_traits): Likewise.
32659 (ipa_vr_hash_table): Likewise.
32660 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
32661 being pointers and vr_known being removed.
32662 (ipa_set_jf_unknown): Likewise.
32663 (ipa_get_ipa_bits_for_value): New function.
32664 (ipa_set_jfunc_bits): Likewise.
32665 (ipa_get_value_range): New overloaded functions.
32666 (ipa_set_jfunc_vr): Likewise.
32667 (ipa_compute_jump_functions_for_edge): Use the above functions to
32668 construct bits and vr parts of jump functions.
32669 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
32670 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
32672 (ipcp_grow_transformations_if_necessary): Also allocate
32673 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
32675 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
32676 them. Fix too long lines.
32677 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
32678 vr_known being removed.
32679 (ipa_read_jump_function): Use new setter functions to construct bits
32680 and vr parts of jump functions or set them to NULL.
32681 (write_ipcp_transformation_info): Adjust for bits being pointers.
32682 (read_ipcp_transformation_info): Likewise.
32683 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
32685 Include gt-ipa-prop.h.
32686 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
32688 (ipcp_store_bits_results): Likewise.
32689 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
32690 Do not write to existing jump functions but use a temporary instead.
32692 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32695 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
32696 attempt to use its first operand as BIT_FIELD_REF base.
32698 2017-03-01 Richard Biener <rguenther@suse.de>
32700 PR middle-end/79721
32701 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
32702 interpolating formula in wrapping arithmetic.
32703 (chrec_apply): Convert chrec_evaluate return value to wanted type.
32705 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32707 PR tree-optimization/79734
32708 * tree-vect-generic.c (expand_vector_condition): Optimize
32709 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
32710 Handle VEC_COND_EXPR where comparison has different inner width from
32711 type's inner width.
32713 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
32715 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
32716 markup, and similar issues. Remove @opindex entries for things
32717 that aren't options. Add missing -mmpy-option entries.
32719 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32721 PR tree-optimization/79737
32722 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
32723 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
32724 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
32725 instead of byte_size. Formatting fix.
32726 (shift_bytes_in_array_right): Formatting fix.
32728 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
32731 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
32732 condition on optimize for the leaf function test.
32734 2017-02-28 Martin Liska <mliska@suse.cz>
32737 * read-rtl-function.c (function_reader::handle_unknown_directive):
32738 Bail out when one uses -flto.
32740 2017-02-28 Martin Liska <mliska@suse.cz>
32742 * common.opt: Replace space with tabular for options of <number>
32744 * config/i386/i386.opt: Show <number> value for
32745 -mlarge-data-threshold.
32746 * opts.c (print_filtered_help): Do not display number in hexadecimal
32749 2017-02-28 Martin Liska <mliska@suse.cz>
32751 * common.opt: Fix --help=option -Q for options which are of
32754 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
32756 * config/i386/i386.c (print_reg): Error out for values
32757 of 8-bit size in invalid integer register.
32759 2017-02-28 Martin Sebor <msebor@redhat.com>
32761 PR tree-optimization/79691
32762 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
32764 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32767 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
32768 gcc_unreachable with output_operand_lossage.
32770 2017-02-28 Richard Biener <rguenther@suse.de>
32772 PR tree-optimization/79740
32773 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
32775 (visit_nary_op): Insert the nary into the hashtable if we
32776 pattern-matched sth.
32777 * tree-ssa-pre.c (eliminate_insert): Robustify.
32779 2017-02-28 Richard Biener <rguenther@suse.de>
32781 PR middle-end/79731
32782 * fold-const.c (decode_field_reference): Reject out-of-bound
32785 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32787 * config/i386/i386.c: Include intl.h.
32788 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
32789 instead of just cond ? "..." : "...".
32790 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
32791 * coverage.c (read_counts_file): Likewise.
32792 * omp-offload.c: Include intl.h.
32793 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
32794 of just cond ? "..." : "...".
32795 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
32796 of just cond ? "..." : "...".
32798 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
32801 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
32803 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
32804 'tune for' CPU name.
32805 * config/arm/arm-cpu-data.h: Regenerated.
32807 2017-02-28 Richard Biener <rguenther@suse.de>
32809 PR tree-optimization/79732
32810 * tree-inline.c (expand_call_inline): Do not shadow var.
32812 2017-02-28 Richard Biener <rguenther@suse.de>
32814 PR tree-optimization/79723
32815 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
32816 address-space properly.
32818 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
32820 * doc/optinfo.texi (Optimization groups): Fix option used for
32822 * doc/invoke.texi (-fopt-info): Document "omp".
32823 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
32824 (OPTGROUP_ALL): Add OPTGROUP_OMP.
32825 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
32826 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
32827 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
32829 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
32831 * dumpfile.c (optgroup_options): Instead of "openmp", associate
32832 OPTGROUP_OMP with "omp".
32834 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
32837 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
32838 for arithmetic shift of unsigned V2DI.
32840 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
32842 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
32843 arc/linux.h headers.
32844 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
32845 (LINK_SPEC): Likewise.
32846 (ARC_TLS_EXTRA_START_SPEC): Likewise.
32847 (EXTRA_SPECS): Likewise.
32848 (STARTFILE_SPEC): Likewise.
32849 (ENDFILE_SPEC): Likewise.
32850 (LIB_SPEC): Likewise.
32851 (TARGET_SDATA_DEFAULT): Likewise.
32852 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
32853 (MULTILIB_DEFAULTS): Likewise.
32854 (DWARF2_UNWIND_INFO): Likewise.
32855 * config/arc/big.h: New file.
32856 * config/arc/elf.h: Likewise.
32857 * config/arc/linux.h: Likewise.
32858 * config/arc/t-uClibc: Remove.
32860 2017-02-27 Bin Cheng <bin.cheng@arm.com>
32862 PR tree-optimization/77536
32863 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
32864 (tree_transform_and_unroll_loop): Use above function to compute the
32865 estimated niter of unrolled loop and use it when scaling profile.
32866 Also use count info rather than frequency if it's non-zero.
32867 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
32868 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
32869 (vect_transform_loop): Call above function.
32871 2017-02-27 Richard Biener <rguenther@suse.de>
32873 PR tree-optimization/45397
32874 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
32875 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
32876 (visit_nary_op): Add pattern matching for CSEing sign-changed
32877 or truncated operations with wider ones.
32879 2017-02-27 Richard Biener <rguenther@suse.de>
32881 PR tree-optimization/79690
32882 * tree-vect-stmts.c (vectorizable_store): Use vector type
32883 built from the DR with address-space.
32885 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
32887 * doc/invoke.texi (Optimize Options): Refine the description
32888 of asan-use-after-return.
32890 2017-02-25 Alan Modra <amodra@gmail.com>
32892 PR rtl-optimization/79584
32893 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
32894 base, not ad->base_term, the reg within base. Remove assertion
32895 that ad->base == ad->base_term. Replace gen_int_mode using
32896 bogus mode with const0_rtx.
32898 2017-02-25 Jakub Jelinek <jakub@redhat.com>
32900 PR middle-end/79396
32901 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
32902 FMA_EXPR like tcc_binary or tcc_unary.
32904 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
32907 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
32909 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
32910 (output_loc_operands): Handle DW_OP_call_ref and
32911 DW_OP_GNU_variable_value.
32912 (struct variable_value_struct): New type.
32913 (struct variable_value_hasher): Likewise.
32914 (variable_value_hash): New variable.
32915 (string_types): Remove.
32916 (copy_loc_descr): New function.
32917 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
32918 (prepend_loc_descr_to_each): New function.
32919 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
32920 instead of add_loc_descr_to_each if the first argument is single
32921 location list and the second has multiple.
32922 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
32923 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
32924 when looking for variable value which doesn't have other location info.
32925 (loc_list_from_tree): Formatting fix.
32926 (gen_array_type_die): Simplify DW_AT_string_length handling.
32927 (adjust_string_types): Remove.
32928 (gen_subprogram_die): Don't call adjust_string_types nor test/set
32929 string_types. Call resolve_variable_values.
32930 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
32931 (resolve_addr_in_expr): Likewise. Add A argument.
32932 (copy_deref_exprloc): Remove deref argument. Adjust for the
32933 original expression being DW_OP_GNU_variable_value with optionally
32934 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
32935 optionally after it.
32936 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
32937 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
32938 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
32939 (variable_value_hasher::hash, variable_value_hasher::equal): New
32941 (resolve_variable_value_in_expr, resolve_variable_value,
32942 resolve_variable_values, note_variable_value_in_expr,
32943 note_variable_value): New functions.
32944 (dwarf2out_early_finish): Call note_variable_value on all toplevel
32947 2017-02-24 Jakub Jelinek <jakub@redhat.com>
32950 * opts.h (handle_generated_option): Add GENERATED_P argument.
32951 * opts-common.c (handle_option): Adjust function comment.
32952 (handle_generated_option): Add GENERATED_P argument, pass it to
32954 (control_warning_option): Pass false to handle_generated_option
32956 * opts.c (maybe_default_option): Pass true to handle_generated_option
32958 * optc-gen.awk: Likewise.
32960 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32962 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
32963 a REG, look at the REG it is a SUBREG of.
32964 (splitter for cmpeqsi_t): Ditto.
32966 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32968 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
32969 the special USEs with the pattern of the insn, not the insn itself.
32971 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
32974 * doc/invoke.texi: Document -mload-store-pairs.
32976 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32977 Sandra Loosemore <sandra@codesourcery.com>
32979 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
32980 argument isn't a CONST_INT.
32981 (nios2_alternate_compare_const): Assert op is a CONST_INT.
32982 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
32983 (nios2_validate_compare): Bypass alternate compare logic if *op2
32984 is not a CONST_INT.
32985 (ldstwm_operation_p): Return false if first_base is not a REG or
32986 if first_offset is not a CONST_INT.
32988 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32990 * config/cris/cris.md: Use correct operand in a define_peephole2.
32992 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32994 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
32996 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32998 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
32999 this_insn if it is an INSN or JUMP_INSN.
33000 (force_offsettable): Look at base, not at addr.
33001 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
33002 on things that aren't necessarily CONST_INTs.
33004 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
33006 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
33007 -mfpmath=sse is the default also for x86-32 targets with SSE2
33008 instruction set when @option{-ffast-math} is enabled
33010 2017-02-24 Jeff Law <law@redhat.com>
33012 PR rtl-optimizatoin/79286
33013 * ira.c (update_equiv_regs): Drop may_trap_p exception to
33016 2017-02-24 Richard Biener <rguenther@suse.de>
33018 PR tree-optimization/79389
33019 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
33022 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
33024 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
33025 function comment to reflect reality.
33026 (loop_exits_before_overflow): Fix typo in function description.
33028 2017-02-24 Richard Biener <rguenther@suse.de>
33030 PR tree-optimization/79389
33031 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
33032 properly that a threading opportunity exists. Detect conditional
33033 copy/constant propagation opportunities.
33035 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
33037 * config/visium/visium.md (type): Add trap.
33038 (b): New mode attribute.
33039 (*btst): Rename into...
33040 (*btst<mode>): ...this and adjust.
33041 (*cbranchsi4_btst_insn): Rename into...
33042 (*cbranch<mode>4_btst_insn): ...this and adjust.
33043 (trap): New define_insn.
33045 2017-02-23 Jakub Jelinek <jakub@redhat.com>
33047 PR tree-optimization/79389
33048 * ifcvt.c (struct noce_if_info): Add rev_cond field.
33049 (noce_reversed_cond_code): New function.
33050 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
33051 reversed_comparison_code. Formatting fix.
33052 (noce_try_store_flag): Test rev_cond != NULL in addition to
33053 reversed_comparison_code.
33054 (noce_try_store_flag_constants): Likewise.
33055 (noce_try_store_flag_mask): Likewise.
33056 (noce_try_addcc): Use rev_cond if non-NULL instead of
33057 reversed_comparison_code.
33058 (noce_try_cmove_arith): Likewise. Formatting fixes.
33059 (noce_try_minmax, noce_try_abs): Clear rev_cond.
33060 (noce_find_if_block): Initialize rev_cond.
33061 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
33062 instead of false as last argument never attempt to reverse it
33065 2017-02-23 Bin Cheng <bin.cheng@arm.com>
33067 PR tree-optimization/79663
33068 * tree-predcom.c (combine_chains): Process refs in reverse order
33069 only for ZERO length chains, and add explaining comment.
33071 2017-02-23 Jeff Law <law@redhat.com>
33073 PR tree-optimization/79578
33074 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
33075 in call to operand_equal_p.
33077 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
33080 * config/i386/cpuid.h: Fix another undefined behavior.
33082 2017-02-23 Richard Biener <rguenther@suse.de>
33084 PR tree-optimization/79683
33085 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
33086 vector types for data-refs.
33088 2017-02-23 Martin Liska <mliska@suse.cz>
33090 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
33092 2017-02-23 Jakub Jelinek <jakub@redhat.com>
33094 PR middle-end/79665
33095 * internal-fn.c (get_range_pos_neg): Moved to ...
33096 * tree.c (get_range_pos_neg): ... here. No longer static.
33097 * tree.h (get_range_pos_neg): New prototype.
33098 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
33099 are known to be in between 0 and signed maximum inclusive, try to
33100 expand both unsigned and signed divmod and use the cheaper one from
33103 2017-02-22 Jeff Law <law@redhat.com>
33105 PR tree-optimization/79578
33106 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
33107 to compare base operands.
33109 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
33112 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
33113 gpc_reg_operand instead of fpr_reg_operand.
33115 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
33117 * config/mips/mips.c (mips_return_in_memory): Force FP
33118 vector types to be returned in memory for o32 ABI.
33120 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33122 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
33123 instead of DW_TAG_member for static data member declarations and don't
33124 set no_linkage_name for static inline data members.
33125 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
33128 2017-02-22 Martin Liska <mliska@suse.cz>
33130 * doc/invoke.texi: Replace inequality signs with square brackets
33133 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
33136 * lra-constraints.c (simplify_operand_subreg): Handle
33137 WORD_REGISTER_OPERATIONS targets.
33139 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33142 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
33143 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
33144 elimination by swapping fld*.
33146 2017-02-22 Richard Biener <rguenther@suse.de>
33148 PR tree-optimization/79673
33149 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
33150 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
33151 irrelevant address-space qualifiers and avoiding a
33152 ADDR_SPACE_CONVERT_EXPR from fold_convert.
33154 2017-02-22 Richard Biener <rguenther@suse.de>
33156 PR tree-optimization/79666
33157 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
33158 to not symbolically negate if that may introduce undefined
33161 2017-02-22 Martin Liska <mliska@suse.cz>
33164 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
33165 * data-streamer-out.c (streamer_write_gcov_count_stream):
33167 * value-prof.c (stream_out_histogram_value): Make assert more
33168 precise based on type of counter.
33170 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
33173 * config/i386/i386.md (standard_x87sse_constant_load splitter):
33174 Use nonimmediate_operand instead of memory_operand for operand 1.
33175 (float-extend standard_x87sse_constant_load splitter): Ditto.
33177 2017-02-21 Jeff Law <law@redhat.com>
33179 PR tree-optimization/79621
33180 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
33181 blocks with edges to themselves.
33183 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33186 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
33187 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
33188 Use gimple_call_builtin_p.
33191 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
33192 on temporarily removed DEBUG_INSNs.
33194 PR tree-optimization/79649
33195 * tree-loop-distribution.c (classify_partition): Give up on
33196 non-generic address space loads/stores.
33198 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
33200 * doc/loop.texi (Loop manipulation): Remove nonexistent
33201 tree_ssa_loop_version from the documentation.
33202 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
33204 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33207 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
33208 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
33209 * config/rs6000/rs6000.c: Include except.h.
33210 (rs6000_expand_split_stack_prologue): Call
33211 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
33213 2017-02-21 Martin Jambor <mjambor@suse.cz>
33216 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
33217 have been analyzed.
33219 2017-02-21 Martin Jambor <mjambor@suse.cz>
33221 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
33222 for backward compatibility only.
33223 * doc/invoke.texi (Option Summary): Remove all references to
33224 -fipa-cp-alignment.
33226 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
33230 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33232 * lra-constraints.c (curr_insn_transform): Handle
33233 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
33235 2017-02-21 Martin Liska <mliska@suse.cz>
33237 * config/i386/i386.opt: Replace -masm-dialect with -masm.
33239 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
33241 PR translation/79638
33242 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
33244 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
33247 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
33248 (arm_function_ok_for_sibcall): Return false for an indirect call by
33249 descriptor if all the argument registers are used.
33250 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
33251 alignment of the function.
33253 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33255 PR tree-optimization/61441
33256 * simplify-rtx.c (simplify_const_unary_operation): For
33257 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
33258 the sNaN unmodified.
33260 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33262 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33263 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33264 instead of SYSTEM_HEADER_DIR.
33266 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
33267 Martin Liška <mliska@suse.cz>
33269 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
33270 Fix typos and grammar, use active voice, and clarify.
33272 2017-02-20 Marek Polacek <polacek@redhat.com>
33274 PR middle-end/79537
33275 * gimplify.c (gimplify_expr): Handle unused *&&L;.
33278 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
33280 2017-02-20 Jakub Jelinek <jakub@redhat.com>
33283 * config/i386/i386.c (ix86_expand_builtin): Handle
33284 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
33285 ix86_builtins_isa[fcode].isa as a requirement of those
33286 flags and any other flag in the bitmask.
33287 (ix86_init_mmx_sse_builtins): Use 0 instead of
33288 ~OPTION_MASK_ISA_64BIT as mask.
33289 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
33290 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
33291 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
33292 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
33294 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33297 * lra-constraints.c (split_reg): Check requested split mode
33298 is supported by the register.
33300 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33302 * lra-constraints.c (simplify_operand_subreg): Remove early
33305 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33308 * lra-constraints.c (curr_insn_transform): Tighten condition
33309 for converting SUBREG reloads from OP_OUT to OP_INOUT.
33311 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33314 * lra-constraints.c (curr_insn_transform): Handle
33315 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
33317 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
33320 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
33322 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
33324 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
33327 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
33330 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33332 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
33335 2017-02-18 Jakub Jelinek <jakub@redhat.com>
33337 * final.c (last_columnnum, override_columnnum): New variables.
33338 (final_start_function): Set last_columnnum, pass it to begin_prologue
33339 hook and pass 0 to dwarf2out_begin_prologue.
33340 (final_scan_insn): Update override_columnnum. Pass last_columnnum
33341 to source_line debug hook.
33342 (notice_source_line): Compute last_columnnum and for debug_column_info
33343 return true on column changes.
33344 * debug.h (struct gcc_debug_hooks): Add column argument to
33345 source_line and begin_prologue hooks.
33346 (debug_nothing_int_charstar_int_bool): Remove prototype.
33347 (debug_nothing_int_int_charstar,
33348 debug_nothing_int_int_charstar_int_bool): New prototypes.
33349 (dwarf2out_begin_prologue): Add column argument.
33350 * debug.c (do_nothing_debug_hooks): Adjust source_line and
33351 begin_prologue hooks.
33352 (debug_nothing_int_charstar_int_bool): Remove.
33353 (debug_nothing_int_int_charstar,
33354 debug_nothing_int_int_charstar_int_bool): New functions.
33355 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
33356 through to dwarf2out_source_line.
33357 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
33358 (dwarf2out_source_line): Add column argument, emit it if requested.
33359 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
33361 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33362 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33363 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
33364 through to dwarf2out_begin_prologue.
33365 (vmsdbgout_source_line): Add column argument, pass it through to
33366 dwarf2out_source_line.
33367 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
33368 dbxout_source_line caller.
33369 (dbxout_source_line): Add column argument.
33371 * common.opt (gno-column-info, gcolumn-info): New options.
33372 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
33373 (check_die): Also test for multiple DW_AT_decl_column attributes.
33374 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
33375 DW_AT_decl_column if requested.
33376 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
33378 (gen_variable_die): Likewise.
33379 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
33380 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
33383 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
33384 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
33385 (ix86_handle_option): Handle OPT_m3dnowa.
33386 * doc/invoke.texi (-m3dnowa): Document.
33387 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
33388 -m3dnowa instead of -m3dnow -march=athlon.
33391 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
33392 instead of gcc_assert for K, r and R code checks. Formatting fixes.
33394 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33397 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
33398 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
33399 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
33400 generator for vsx_xxpermdi_<mode>_be.
33401 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
33402 force big-endian semantics.
33403 (vsx_xxpermdi_<mode>_be): New define_expand with same
33404 implementation as previous version of vsx_xxpermdi_<mode>.
33406 2017-02-17 Jakub Jelinek <jakub@redhat.com>
33408 PR tree-optimization/79327
33409 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
33410 variable, its initialization and use.
33412 2017-02-17 Julia Koval <julia.koval@intel.com>
33414 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
33415 (OPTION_MASK_ISA_PKU_UNSET): New.
33416 (ix86_handle_option): Handle -mrdpid.
33417 * config/i386/cpuid.h (bit_RDPID): New.
33418 * config/i386/driver-i386.c (host_detect_local_cpu):
33419 Detect RDPID feature.
33420 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
33421 * config/i386/i386-c.c (ix86_target_macros_internal):
33423 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
33424 (ix86_valid_target_attribute_inner_p): Add "rdpid".
33425 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
33426 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
33427 * config/i386/i386.md (define_insn "rdpid"): New.
33428 * config/i386/i386.opt Add -mrdpid.
33429 * config/i386/immintrin.h (_rdpid_u32): New.
33431 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
33433 PR rtl-optimization/79541
33434 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
33435 instead of transforming it into USE.
33437 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
33439 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
33440 If HONOR_SNANS (SFmode) force the input to a register.
33441 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
33442 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
33443 an frsp or similar insn.
33445 2017-02-17 Martin Liska <mliska@suse.cz>
33447 PR rtl-optimization/79577
33448 * params.def (selsched-max-sched-times): Increase minimum to 1.
33450 2017-02-17 Martin Liska <mliska@suse.cz>
33452 PR rtl-optimization/79574
33453 * gcse.c (want_to_gcse_p): Prevent integer overflow.
33455 2017-02-17 Martin Liska <mliska@suse.cz>
33457 PR tree-optimization/79529
33458 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
33459 ssa_defined_default_def_p to handle cases which are implicitly
33461 * tree-ssa.c (ssa_defined_default_def_p): New function.
33462 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
33463 which are implicitly defined.
33464 * tree-ssa.h (ssa_defined_default_def_p): Declare.
33466 2017-02-17 Richard Biener <rguenther@suse.de>
33468 PR middle-end/79576
33469 * params.def (max-ssa-name-query-depth): Limit to 10.
33471 2017-02-17 Richard Biener <rguenther@suse.de>
33473 PR tree-optimization/79552
33474 * tree-ssa-structalias.c (visit_loadstore): Properly verify
33477 2017-02-17 Richard Biener <rguenther@suse.de>
33480 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
33482 2017-02-17 Marek Polacek <polacek@redhat.com>
33484 PR middle-end/79536
33485 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
33486 (fold_negate_expr): New wrapper.
33488 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
33490 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
33491 Correct terminology and de-emphasize pre-standard behavior.
33493 2017-02-16 Alan Modra <amodra@gmail.com>
33495 PR rtl-optimization/79286
33496 * ira.c (def_dominates_uses): New function.
33497 (update_equiv_regs): Don't create an equivalence for insns that
33498 may trap where the register def does not dominate the use.
33500 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
33502 PR rtl-optimization/78127
33503 * lra.c (lra): Call lra_eliminate before finish the loop after
33506 2017-02-16 Richard Biener <rguenther@suse.de>
33508 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
33510 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
33511 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
33512 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
33513 (isl_val_int_from_wi): New function.
33514 (extract_affine_gmp): Rename to ...
33515 (extract_affine_wi): ... this, take a widest_int.
33516 (extract_affine_int): Just wrap extract_affine_wi.
33517 (add_param_constraints): Use isl_val_int_from_wi.
33518 (add_loop_constraints): Likewise, and extract_affine_wi.
33520 2017-02-15 Jeff Law <law@redhat.com>
33522 PR middle-end/79521
33523 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
33524 ira_init_register_move_cost_if_necessary.
33526 2017-02-15 Martin Sebor <msebor@redhat.com>
33528 PR middle-end/32003
33529 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
33530 removed in a prior commit.
33532 2017-02-15 Bin Cheng <bin.cheng@arm.com>
33534 PR tree-optimization/79347
33535 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
33536 counters during peeling.
33538 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
33540 * Makefile.in (site.exp): Remove "set ISLVER".
33542 2017-02-15 Jakub Jelinek <jakub@redhat.com>
33545 * real.c (real_from_integer): Call real_convert even for decimal.
33547 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
33550 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
33552 2017-02-14 Andrew Pinski <apinski@cavium.com>
33554 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
33555 cores and change the partno/implementer to be correct.
33556 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
33557 the 'B" as the implementer.
33558 * config/aarch64/aarch64-tune.md: Regenerate.
33560 2017-02-14 Carl Love <cel@us.ibm.com>
33562 * config/rs6000/rs6000.c: Add case statement entry to make the
33563 xvcvuxdsp built-in argument unsigned.
33564 * config/rs6000/vsx.md: Fix the source and return operand types so they
33565 match the instruction definitions from the ISA document. Fix typo
33566 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
33569 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
33572 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
33573 member early_clobber_alts.
33574 * lra-lives.c (reg_early_clobber_p): New.
33575 (process_bb_lives): Use it.
33576 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
33577 (debug_operand_data): Initialize early_clobber_alts.
33578 (setup_operand_alternative): Set up early_clobber_alts.
33579 (collect_non_operand_hard_regs): Ditto. Pass early clobber
33580 alternatives to new_insn_reg.
33581 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
33583 (lra_update_insn_regno_info): Pass the new arg.
33585 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33587 PR middle-end/79505
33588 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
33589 (new_oacc_loop_raw): Don't clear already cleared fields.
33592 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
33593 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
33594 _mm512_prefetch_i64gather_ps): New inline functions and macros.
33596 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
33599 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
33601 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
33604 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
33605 the extra instruction to the right place to store 128-bit constant
33608 2017-02-14 Martin Sebor <msebor@redhat.com>
33610 PR middle-end/79448
33611 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
33612 warning for strings of unknown length.
33614 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
33616 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
33618 2017-02-14 Jeff Law <law@redhat.com>
33621 * ira-costs.c (scan_one_insn): Initialize register move costs
33622 for pseudos seen in USE/CLOBBER insns.
33624 PR tree-optimization/79095
33625 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
33626 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
33627 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
33628 if the operands are known to be not equal, then the resulting range
33630 (intersect_ranges): If the new range is ~[0,0] and the old range is
33631 wide, then prefer ~[0,0].
33632 * tree-vrp.c (overflow_comparison_p_1): New function.
33633 (overflow_comparison_p): New function.
33634 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
33635 if NAME is used in an overflow test.
33636 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
33637 overflow check that can be expressed as an equality test, then adjust
33638 ops to be that equality test.
33640 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33642 * config/s390/s390-builtin-types.def: Remove flags argument.
33643 * config/s390/s390.c (s390_init_builtins): Likewise.
33645 2017-02-14 Martin Liska <mliska@suse.cz>
33647 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
33648 vector. Fix trailing white spaces.
33650 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
33652 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
33655 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33657 PR rtl-optimization/68664
33658 * config/arm/arm.c (arm_sched_can_speculate_insn):
33659 New function. Declare prototype.
33660 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
33662 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33664 PR rtl-optimization/68664
33665 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
33667 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
33669 2017-02-14 Amit Pawar <amit.pawar@amd.com>
33671 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
33672 max skip bytes for function, loop and jump.
33674 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33676 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
33677 ABS_EXPR for gimple dump.
33679 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33682 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
33684 PR tree-optimization/79408
33685 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
33686 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
33687 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
33688 also if rhs1 is INTEGER_CST.
33690 2017-02-14 Richard Biener <rguenther@suse.de>
33692 PR middle-end/79432
33693 * tree-into-ssa.c (insert_phi_nodes): When the function can
33694 have abnormal edges rewrite SSA names with broken use-def
33695 dominance out of SSA and register them for PHI insertion.
33697 2017-02-13 Martin Sebor <msebor@redhat.com>
33699 PR middle-end/79496
33700 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
33701 clearing info.nowrite flag when snprintf size argument is a range.
33703 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33705 * cprop.c (cprop_jump): Add missing space in string literal.
33706 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
33707 (get_constraint_for_component_ref): Likewise.
33708 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
33709 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
33710 * lra-constraints.c (process_alt_operands): Likewise.
33711 * ipa-inline.c (inline_small_functions): Likewise.
33712 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
33713 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
33714 * trans-mem.c (diagnose_tm_1_op): Likewise.
33715 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
33716 (grid_parallel_clauses_gridifiable): Likewise.
33718 * config/nvptx/mkoffload.c (process): Add space in between
33721 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
33722 "MOD4_SSE_REGS" and "ALL_REGS".
33724 * spellcheck.c (test_data): Add , in between "foo" and "food".
33726 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33729 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
33730 boundary crossing check and subsequent code generation agree.
33732 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33734 * config/aarch64/aarch64.c (has_memory_op): Delete.
33735 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
33738 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33740 PR rtl-optimization/79388
33741 PR rtl-optimization/79450
33742 * combine.c (distribute_notes): When removing TEM_INSN for which
33743 corresponding dest has last value recorded, invalidate that last
33746 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33748 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
33749 of explicit '@'. Add missing assembly comment marker on branch costs
33752 2017-02-13 Nathan Sidwell <nathan@acm.org>
33754 * gengtype-lex.l (<in_struct>): Add '/'.
33756 2017-02-13 Martin Liska <mliska@suse.cz>
33759 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
33761 2017-02-13 Richard Biener <rguenther@suse.de>
33763 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
33765 * configure: Re-generate.
33766 * config.in: Likewise.
33767 * graphite-dependences.c: Simplify as if
33768 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
33769 * graphite-isl-ast-to-gimple.c: Likewise.
33770 * graphite-optimize-isl.c: Likewise.
33771 * graphite-poly.c: Likewise.
33772 * graphite-sese-to-poly.c: Likewise.
33773 * graphite.h: Likewise.
33774 * toplev.c: Include isl/version.h and use isl_version () for
33775 printing the ISL version.
33776 * doc/install.texi: Update ISL requirement.
33778 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33780 * doc/standards.texi (Standards): Update reference to
33783 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33785 * doc/extend.texi (Named Address Spaces): sourceware.org now
33787 * doc/install.texi (Binaries): Ditto.
33790 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33792 * doc/cpp.texi: Replace "stringify"/"stringification" with C
33793 standard terminology "stringize"/"stringizing" throughout.
33794 * doc/cppinternals.texi: Likewise.
33796 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33798 * doc/extend.texi: Fix some spelling mistakes and typos.
33799 * doc/invoke.texi: Likewise.
33801 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
33804 * params.def (inline-min-speedup) Change from 10 to 8.
33806 2017-02-11 Jakub Jelinek <jakub@redhat.com>
33808 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
33811 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
33814 * ipa-inline-analysis.c (get_minimal_bb): New function.
33815 (record_modified): Use it.
33816 (remap_edge_change_prob): Handle also ancestor functions.
33818 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
33820 * doc/contrib.texi (Contributors): Remove broken link into
33821 the Mauve CVS repository.
33823 2017-02-11 Jakub Jelinek <jakub@redhat.com>
33825 PR middle-end/79454
33826 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
33827 result computation whenever lhs doesn't have vector mode, not
33828 just when it has BLKmode.
33830 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
33832 * doc/makefile.texi (profiledbootstrap): Refer to the
33833 installation instructions only in textual form.
33835 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33838 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
33840 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
33842 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
33843 (Specific): Update mingw-w64 reference.
33845 (Specific): Remove broken link to Renesas RX processor.
33847 2017-02-10 Richard Biener <rguenther@suse.de>
33849 * toplev.c (process_options): Do not mention obsolete graphite
33850 options when printing sorry message about missing graphite support.
33851 Mention -floop-nest-optimize.
33853 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
33855 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
33856 (vtst_p16): Likewise.
33857 (vtstq_p8): Likewise.
33858 (vtstq_p16): Likewise.
33860 (vtstq_p64): Likewise.
33861 * config/arm/arm_neon.h (vgetq_lane_p64): New.
33862 (vset_lane_p64): New.
33863 (vsetq_lane_p64): New.
33865 2017-02-10 Jakub Jelinek <jakub@redhat.com>
33867 PR tree-optimization/79411
33868 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
33869 stmt operands are SSA_NAMEs used in abnormal phis.
33870 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
33873 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
33876 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
33879 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
33881 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
33883 2017-02-09 Jakub Jelinek <jakub@redhat.com>
33885 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
33889 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
33890 not arbitrary TREE_CONSTANT.
33893 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
33894 "omp declare target link" attribute unless is_global_var.
33895 * omp-offload.c (find_link_var_op): Likewise.
33897 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
33898 Chung-Lin Tang <cltang@codesourcery.com>
33900 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
33902 (gimplify_adjust_omp_clauses): Don't delete TILE.
33903 (gimplify_omp_for): Deal with TILE.
33904 * internal-fn.c (expand_GOACC_TILE): New function.
33905 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
33907 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
33908 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
33910 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
33911 avoid DIV for outermost collapse var.
33912 (expand_oacc_for): Insert tile element loop as needed. Adjust.
33913 Remove out of date comments, fix whitespace.
33914 * omp-general.c (omp_extract_for_data): Deal with tiling.
33915 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
33916 adjust OLF_DIM_BASE value.
33917 (struct omp_for_data): Add tiling field.
33918 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
33919 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
33920 for auto loops. Remove default auto determining, moved to
33921 oacc_loop_fixed_partitions.
33922 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
33923 stmts, add e_mask field.
33924 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
33925 (oacc_thread_numbers): Use oacc_dim_call.
33926 (oacc_xform_tile): New.
33927 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
33928 (finish_oacc_loop): Adjust for ifns vector.
33929 (oacc_loop_discover_walk): Append loop abstraction sites to list,
33930 add case for GOACC_TILE fns.
33931 (oacc_loop_xform_loop): Delete.
33932 (oacc_loop_process): Iterate over call list directly, and add
33933 handling for GOACC_TILE fns.
33934 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
33936 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
33937 vector partitioning to outer loops. Assign 2 partitions to loops
33938 when available. Add TILE handling.
33939 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
33940 (execite_oacc_device_lower): Process GOACC_TILE fns,
33941 ignore unknown specs.
33942 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
33943 * tree.c (omp_clause_num_ops): Adjust TILE ops.
33944 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
33946 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
33948 * configure.ac (ACX_BUGURL): Update.
33949 * configure: Regenerate.
33951 2017-02-09 Richard Biener <rguenther@suse.de>
33953 PR tree-optimization/69823
33954 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
33955 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
33957 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
33959 * config/arc/arc-c.def: Add __NPS400__ definition.
33960 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
33961 (TARGET_NPS400): Define.
33963 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
33965 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
33967 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
33968 pointer, arch_info.
33969 (arc_cpu_types): Fill the arch_info field with a pointer into the
33970 arc_arch_types table.
33971 (arc_selected_cpu): Declare.
33972 * config/arc/arc.c (arc_selected_cpu): Make global.
33973 (arc_selected_arch): Delete.
33974 (arc_base_cpu): Delete.
33975 (arc_override_options): Remove references to deleted variables,
33976 update access to arch information.
33977 (ARC_OPT): Update access to arch information.
33978 (ARC_OPTX): Likewise.
33979 * config/arc/arc.h (arc_base_cpu): Remove declaration.
33980 (TARGET_ARC600): Update access to arch information.
33981 (TARGET_ARC601): Likewise.
33982 (TARGET_ARC700): Likewise.
33983 (TARGET_EM): Likewise.
33984 (TARGET_HS): Likewise.
33985 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
33988 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
33991 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
33992 condition/operands for integer GE/LE/GEU/LEU operations.
33994 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
33996 PR translation/79397
33997 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
34000 2017-02-08 Martin Jambor <mjambor@suse.cz>
34003 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
34004 whether allocation happened.
34005 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
34006 nothing was allocated.
34008 2017-02-08 Jakub Jelinek <jakub@redhat.com>
34010 PR tree-optimization/79408
34011 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
34012 constant, but SSA_NAME with a known integer range, use the minimum
34013 of that range instead of op1 to determine if modulo can be replaced
34014 with its first operand.
34016 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34018 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
34020 2017-02-08 Richard Biener <rguenther@suse.de>
34022 PR tree-optimization/71824
34023 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34024 Check all loops contained in the merged region.
34026 2017-02-07 Andrew Pinski <apinski@cavium.com>
34028 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
34030 2017-02-07 Andrew Pinski <apinski@cavium.com>
34032 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
34033 (thunderxt88): Likewise.
34034 (thunderxt81): Disable LSE and change v8.1 to v8.
34035 (thunderxt83): Likewise.
34037 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34038 Richard Biener <rguenther@suse.de>
34040 PR middle-end/79399
34041 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
34042 type from int to size_t.
34043 * ira-costs.c (struct_costs_size): Change type from int to size_t.
34045 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34047 PR rtl-optimization/79386
34048 * cprop.c (bypass_conditional_jumps): Initialize
34049 bypass_last_basic_block already before splitting bbs after
34050 unconditional traps...
34051 (bypass_conditional_jumps): ... rather than here.
34054 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
34055 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
34056 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
34057 fix -masm=intel patterns.
34059 2017-02-07 Richard Biener <rguenther@suse.de>
34061 PR tree-optimization/79256
34062 PR middle-end/79278
34063 * builtins.c (get_object_alignment_2): Use min_align_of_type
34064 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
34065 and ADJUST_FIELD_ALIGN.
34067 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
34069 * doc/tm.texi: Regenerate.
34070 * stor-layout.c (layout_decl): Adjust.
34071 (update_alignment_for_field): Likewise.
34072 (place_field): Likewise.
34073 (min_align_of_type): Likewise.
34074 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
34075 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
34076 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
34077 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
34078 * config/frv/frv.c (frv_adjust_field_align): Likewise.
34079 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
34080 * config/i386/i386.c (x86_field_alignment): Likewise.
34081 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
34082 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
34083 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
34084 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
34085 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
34086 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
34090 2017-01-30 Richard Biener <rguenther@suse.de>
34092 PR tree-optimization/79256
34093 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34094 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34097 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
34099 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
34100 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
34101 builtins to SImode and emit a zero-extend, if necessary.
34103 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34105 * docs/invoke.texi (RISC-V Options): Alphabetize.
34107 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34109 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
34112 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34114 * config/riscv/riscv.c: New file.
34115 * common/config/riscv/riscv-common.c: Likewise.
34116 * config.gcc: Likewise.
34117 * config/riscv/constraints.md: Likewise.
34118 * config/riscv/elf.h: Likewise.
34119 * config/riscv/generic.md: Likewise.
34120 * config/riscv/linux.h: Likewise.
34121 * config/riscv/multilib-generator: Likewise.
34122 * config/riscv/peephole.md: Likewise.
34123 * config/riscv/pic.md: Likewise.
34124 * config/riscv/predicates.md: Likewise.
34125 * config/riscv/riscv-builtins.c: Likewise.
34126 * config/riscv/riscv-c.c: Likewise.
34127 * config/riscv/riscv-ftypes.def: Likewise.
34128 * config/riscv/riscv-modes.def: Likewise.
34129 * config/riscv/riscv-opts.h: Likewise.
34130 * config/riscv/riscv-protos.h: Likewise.
34131 * config/riscv/riscv.h: Likewise.
34132 * config/riscv/riscv.md: Likewise.
34133 * config/riscv/riscv.opt: Likewise.
34134 * config/riscv/sync.md: Likewise.
34135 * config/riscv/t-elf-multilib: Likewise.
34136 * config/riscv/t-linux: Likewise.
34137 * config/riscv/t-linux-multilib: Likewise.
34138 * config/riscv/t-riscv: Likewise.
34139 * configure.ac: Likewise.
34140 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
34141 Waterman as RISC-V maintainers.
34142 * doc/install.texi: Add RISC-V entries.
34143 * doc/invoke.texi: Add RISC-V options section.
34144 * doc/md.texi: Add RISC-V constraints section.
34145 * configure: Regenerated.
34147 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
34150 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
34151 false values to be constant vectors with all 0 or all 1 bits set.
34152 (vcondu<mode><mode>): Likewise.
34153 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
34155 (fpmask_comparison_operator): Update comment.
34156 (vecint_comparison_operator): New predicate.
34157 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
34158 vector conditionals when the true and false values are constant
34159 vectors with all 0 bits or all 1 bits set.
34161 2017-02-06 Martin Sebor <msebor@redhat.com>
34163 PR tree-optimization/79376
34164 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
34166 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
34168 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
34169 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
34170 to simplify split condition.
34172 2017-02-06 Jakub Jelinek <jakub@redhat.com>
34174 * omp-expand.c (oxpand_omp_atomic_fetch_op,
34175 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
34178 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
34180 PR rtl-optimization/68664
34181 * target.def (can_speculate_insn): New hook.
34182 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
34183 * doc/tm.texi: Regenerate.
34184 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
34185 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
34186 (rs6000_sched_can_speculate_insn): New function.
34188 2017-02-06 Jakub Jelinek <jakub@redhat.com>
34190 PR tree-optimization/79284
34191 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
34192 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
34193 vectorizable_mask_load_store, vectorizable_operation,
34194 vect_is_simple_cond, get_same_sized_vectype): Use it instead
34195 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
34196 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
34197 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
34198 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
34199 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
34200 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
34201 is_gimple_assign (stmt). Replace another such test with
34202 is_gimple_assign (stmt).
34204 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
34207 * config/avr/avr.c (rtl-iter.h): Include it.
34208 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
34209 (avr_legitimate_combined_insn): ...and implementation.
34211 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
34213 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
34214 * config/s390/s390.c (s390_const_operand_ok)
34215 (s390_canonicalize_comparison, s390_extract_part)
34216 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
34217 (s390_contiguous_bitmask_p, s390_rtx_costs)
34218 (legitimize_pic_address): Likewise.
34219 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
34220 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
34221 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
34222 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
34223 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
34225 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
34227 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
34228 REGNO($0) == REGNO($1).
34230 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34232 * config/s390/linux.h(SIZE_TYPE): Add comment.
34234 2017-02-06 Julian Brown <julian@codesourcery.com>
34235 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34236 Virendra Pathak <virendra.pathak@broadcom.com>
34238 * config/aarch64/aarch64-cores.def: Change the scheduler
34240 * config/aarch64/aarch64.md: Include thunderx2t99.md.
34241 * config/aarch64/thunderx2t99.md: New file.
34243 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34245 * doc/standards.texi (Go Language): Update link to language
34248 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
34250 * tree-eh.c (lower_resx): Sanitize profile.
34251 (cleanup_empty_eh_move_lp): Likewise.
34253 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
34256 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
34258 * cfgloopmanip.h (loop_version): Update prototype.
34259 * modulo-sched.c (sms_schedule): Update call of loop_version.
34260 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
34261 * tree-parloops.c (gen_parallel_loop): Likewise.
34262 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
34263 * tree-ssa-loop-split.c (split_loop): Likewise.
34264 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
34265 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
34267 2017-02-05 Martin Liska <mliska@suse.cz>
34270 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
34272 (print_operand_address): Initialize a struct to zero.
34274 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34276 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
34277 garbage collector only in textual form.
34279 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34281 * doc/extend.texi (x86 specific memory model extensions for
34282 transactional memory): Simplify a phrase.
34284 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
34287 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
34288 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
34289 (atomic_storedi_1): Likewise.
34291 2017-02-04 Jakub Jelinek <jakub@redhat.com>
34293 PR tree-optimization/79338
34294 * tree-parloops.c (gather_scalar_reductions): Don't call
34295 vect_analyze_loop_form for loop->inner before destroying loop's
34298 2017-02-03 Martin Sebor <msebor@redhat.com>
34300 PR tree-optimization/79327
34301 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
34302 when precision has resulted in leading zeros.
34303 (format_integer): Adjust the likely counter to assume an unknown
34304 argument that may be zero is non-zero.
34306 2017-02-03 Jason Merrill <jason@redhat.com>
34309 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
34310 avoid copying non-taken branch.
34312 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34314 PR tree-optimization/79340
34315 * tree-vect-loop.c (vectorizable_reduction): Release
34316 vec_defs elements after safe_splicing them into other vectors.
34319 PR tree-optimization/79327
34320 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
34321 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
34323 (format_integer): Use wide_int_to_tree instead of build_int_cst
34324 + to_?hwi. If argmin is NULL, just set argmin and argmax to
34325 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
34326 of shortest and longest sequence.
34328 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
34330 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
34331 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
34333 2017-02-03 Walter Lee <walt@tilera.com>
34336 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
34337 after initial stackframe link reg save.
34338 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
34340 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34343 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
34344 wu for stxssp alternative.
34346 2017-02-03 Martin Sebor <msebor@redhat.com>
34348 PR tree-optimization/79352
34349 * gimple-fold.c (get_range_strlen): Add argument.
34350 (get_range_strlen): Change return type to bool.
34351 (get_maxval_strlen): Pass in a dummy argument.
34352 * gimple-fold.h (get_range_strlen): Change return type to bool.
34353 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
34354 * tree.h (array_at_struct_end_p): Add argument.
34355 * tree.c (array_at_struct_end_p): Handle it.
34357 2017-02-03 Martin Liska <mliska@suse.cz>
34360 * multiple_target.c (create_dispatcher_calls): Redirect edge
34361 from a caller of a dispatcher.
34362 (expand_target_clones): Make the clones local.
34363 (ipa_target_clone): Do both target clones and resolvers.
34364 (ipa_dispatcher_calls): Remove the pass.
34365 (pass_dispatcher_calls::gate): Likewise.
34366 (make_pass_dispatcher_calls): Likewise.
34367 * passes.def (pass_target_clone): Put as very first IPA early
34370 2017-02-03 Martin Liska <mliska@suse.cz>
34372 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
34373 in case of a function with ifunc attribute.
34375 2017-02-03 Martin Liska <mliska@suse.cz>
34377 * cgraph.c (cgraph_node::dump): Dump function version info.
34378 * symtab.c (symtab_node::dump_base): Add missing new line.
34380 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34382 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
34383 (ifcombine_ifandif): Use it.
34385 2017-02-03 Martin Liska <mliska@suse.cz>
34387 * doc/invoke.texi: Document default value for
34388 use-after-scope-direct-emission-threshold.
34390 2017-02-03 Martin Liska <mliska@suse.cz>
34392 PR tree-optimization/79339
34393 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
34394 (format_floating): Likewise.
34396 2017-02-03 Martin Liska <mliska@suse.cz>
34399 * ipa-prop.c (ipa_node_params_t::insert): Remove current
34401 (ipa_node_params_t::remove): Likewise.
34402 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
34403 initialization from removed ipa_node_params_t::insert.
34404 (ipa_node_params::~ipa_node_params): Move from removed
34405 ipa_node_params_t::release.
34406 * symbol-summary.h (symbol_summary::m_released): New member.
34407 Do not release a summary twice. Do not allow to call finalizer
34408 for types of a summary that live in GGC memory.
34410 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34412 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
34415 2017-02-02 Martin Sebor <msebor@redhat.com>
34417 PR middle-end/79275
34418 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
34419 (format_string): Tighten up the range of output for non-constant
34420 strings and correct the expected range for wide non-constant strings.
34422 2017-02-02 Martin Sebor <msebor@redhat.com>
34424 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
34426 PR middle-end/32003
34427 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
34429 (-fdump-tree-@var): Add to index and document how to come up
34430 with pass-specific option and dump file names.
34431 (-fdump-passes): Clarify where to look for output.
34433 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34435 PR middle-end/77445
34436 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
34437 statistics of the analyzed path; allow threading for speed when
34438 any of BBs along the path are optimized for speed.
34440 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
34442 PR middle-end/78468
34443 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
34444 settings of the virtual registers.
34447 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
34449 * explow.c (get_dynamic_stack_size): Take known alignment of stack
34450 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
34453 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34455 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
34456 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
34458 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34460 * config/s390/s390.md: Add missing comments with the expanded
34462 * config/s390/vector.md: Likewise.
34463 * config/s390/vx-builtins.md: Likewise.
34465 2017-02-02 Jakub Jelinek <jakub@redhat.com>
34468 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
34469 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
34470 conditions on a single line.
34472 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34474 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
34475 __S390_VX__ to __VX__.
34477 2017-02-01 Andrew Pinski <apinski@cavium.com>
34479 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
34480 stmt_info to record_stmt_cost.
34481 (vect_get_known_peeling_cost): Pass stmt_info if known to
34483 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
34484 cpu_vector_cost field into
34485 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
34486 field into vec_int_stmt_cost and vec_fp_stmt_cost.
34487 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
34488 splitting of scalar_stmt_cost and vec_stmt_cost.
34489 (thunderx_vector_cost): Likewise.
34490 (cortexa57_vector_cost): LIkewise.
34491 (exynosm1_vector_cost): Likewise.
34492 (xgene1_vector_cost): Likewise.
34493 (thunderx2t99_vector_cost): Improve after the splitting of the two
34495 (aarch64_builtin_vectorization_cost): Update for the splitting of
34496 scalar_stmt_cost and vec_stmt_cost.
34498 2017-02-01 Torvald Riegel <triegel@redhat.com>
34499 Richard Henderson <rth@redhat.com>
34501 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
34502 conditional on existance of a fast atomic load.
34503 * optabs-query.c (can_atomic_load_p): New function.
34504 * optabs-query.h (can_atomic_load_p): Declare it.
34505 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
34506 no fast atomic load is available for the particular size of access.
34507 (expand_atomic_compare_and_swap): Likewise.
34508 (expand_atomic_load): Likewise.
34509 (expand_atomic_store): Likewise.
34510 (expand_atomic_fetch_op): Likewise.
34511 * testsuite/lib/target-supports.exp
34512 (check_effective_target_sync_int_128): Remove x86 because it provides
34513 no fast atomic load.
34514 (check_effective_target_sync_int_128_runtime): Likewise.
34516 2017-02-01 Richard Biener <rguenther@suse.de>
34518 * graphite.c: Include tree-vectorizer.h for find_loop_location.
34519 (graphite_transform_loops): Provide opt-info for optimized nests.
34520 * tree-parloop.c (parallelize_loops): Provide opt-info for
34521 parallelized loops.
34523 2017-02-01 Richard Biener <rguenther@suse.de>
34525 PR middle-end/79315
34526 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
34527 was not set before.
34529 2017-02-01 Richard Biener <rguenther@suse.de>
34531 PR tree-optimization/71824
34532 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34533 Verify the loops are valid in the merged SESE region.
34534 (scop_detection::can_represent_loop_1): Check analyzing the
34535 evolution of the number of iterations in the region succeeds.
34537 2017-01-31 Ian Lance Taylor <iant@golang.org>
34539 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
34540 REG_ARGS_SIZE note to 32-bit push insns and call insn.
34542 2017-01-31 David Malcolm <dmalcolm@redhat.com>
34544 PR preprocessor/79210
34545 * input.c (get_substring_ranges_for_loc): Replace line_width
34546 assertion with error-handling.
34548 2017-01-31 Richard Biener <rguenther@suse.de>
34550 PR tree-optimization/77318
34551 * graphite-sese-to-poly.c (extract_affine): Fix assert.
34552 (create_pw_aff_from_tree): Take loop parameter.
34553 (add_condition_to_pbb): Pass loop of the condition to
34554 create_pw_aff_from_tree.
34556 2017-01-31 Jakub Jelinek <jakub@redhat.com>
34558 * config/s390/s390.c (s390_asan_shadow_offset): New function.
34559 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
34561 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
34565 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
34567 (convert_int_to_float128): Likewise.
34568 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
34569 (convert_int_to_float128): Likewise.
34570 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
34571 (UNSPEC_IEEE128_CONVERT): Likewise.
34572 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
34573 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
34574 Use local variables for IBM extended format.
34575 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
34576 (fix_trunc<mode>si2_fprs): Likewise.
34577 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
34578 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
34579 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
34580 to know that we can now have integers of all sizes in vector
34582 (fix<uns>_<mode>di2_hw): Likewise.
34583 (float<uns>_<mode>si2_hw): Likewise.
34584 (fix_<mode>si2_hw): Likewise.
34585 (fixuns_<mode>si2_hw): Likewise.
34586 (float<uns>_<mode>di2_hw): Likewise.
34587 (float_<mode>di2_hw): Likewise.
34588 (float_<mode>si2_hw): Likewise.
34589 (floatuns_<mode>di2_hw): Likewise.
34590 (floatuns_<mode>si2_hw): Likewise.
34591 (xscvqp<su>wz_<mode>): Delete, no longer used.
34592 (xscvqp<su>dz_<mode>): Likewise.
34593 (xscv<su>dqp_<mode>): Likewise.
34594 (ieee128_mfvsrd_64bit): Likewise.
34595 (ieee128_mfvsrd_32bit): Likewise.
34596 (ieee128_mfvsrwz): Likewise.
34597 (ieee128_mtvsrw): Likewise.
34598 (ieee128_mtvsrd_64bit): Likewise.
34599 (ieee128_mtvsrd_32bit): Likewise.
34601 2017-01-31 Martin Liska <mliska@suse.cz>
34604 * ipa-prop.c (ipa_free_all_node_params): Call release method
34605 instead of ~sumbol_summary to not to trigger double times
34608 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
34610 PR tree-optimization/71691
34611 * bitmap.h (class auto_bitmap): New.
34612 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
34613 is_maybe_undefined instead of ssa_undefined_value_p.
34615 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34617 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
34618 __S390_ARCH_LEVEL__ to __ARCH__.
34620 2017-01-31 Jakub Jelinek <jakub@redhat.com>
34622 PR tree-optimization/79267
34623 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
34624 if should_remove_lhs_p is true.
34626 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
34629 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
34630 (add_alignment_attribute): New.
34631 (base_type_die): Add alignment attribute.
34632 (subrange_type_die): Likewise.
34633 (modified_type_die): Likewise.
34634 (gen_array_type_die): Likewise.
34635 (gen_descr_array_type_die: Likewise.
34636 (gen_enumeration_type_die): Likewise.
34637 (gen_subprogram_die): Likewise.
34638 (gen_variable_die): Likewise.
34639 (gen_field_die): Likewise.
34640 (gen_ptr_to_mbr_type_die): Likewise.
34641 (gen_struct_or_union_type_die): Likewise.
34642 (gen_subroutine_type_die): Likewise.
34643 (gen_typedef_die): Likewise.
34644 (base_type_cmp): Compare alignment attribute.
34646 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34649 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
34650 (setb_unsigned) New pattern for setb with CCUNS.
34651 * config/rs6000/rs6000.c (expand_block_compare): Use a different
34652 subfc./subfe sequence to avoid overflow problems. Generate a
34653 shorter sequence with cmpld/setb for power9.
34654 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
34655 for generating subfc. instruction.
34656 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
34657 now uses this instruction.
34659 2017-01-30 Ian Lance Taylor <iant@google.com>
34662 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
34663 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
34665 2017-01-30 Martin Sebor <msebor@redhat.com>
34667 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
34668 Move constant to the right of a relational operator.
34669 (get_mpfr_format_length, format_character, format_string): Ditto.
34670 (should_warn_p, maybe_warn): Same.
34672 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
34674 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
34677 * asan.c (get_translation_unit_decl): Remove function.
34678 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
34680 2017-01-30 Martin Liska <mliska@suse.cz>
34682 PR gcov-profile/79259
34683 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
34684 -fprofile-generate.
34686 2017-01-30 Martin Liska <mliska@suse.cz>
34689 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
34690 Initialize variables with NULL value.
34692 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
34695 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
34697 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
34699 2017-01-30 Richard Biener <rguenther@suse.de>
34701 PR tree-optimization/79276
34702 * tree-vrp.c (process_assert_insertions): Properly adjust common
34703 when removing a duplicate.
34705 2017-01-30 Richard Biener <rguenther@suse.de>
34707 PR tree-optimization/79256
34708 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34709 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34711 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
34713 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
34716 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
34717 ("*r<noxa>sbg_<mode>_sll_bitmask")
34718 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
34719 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
34720 Use contiguous_bitmask_nowrap_operand.
34722 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34725 * config/rs6000/altivec.h (vec_xl): Revise #define.
34726 (vec_xst): Likewise.
34728 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
34730 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
34732 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
34734 PR rtl-optimization/79194
34735 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
34736 traps before call to bypass_conditional_jumps.
34738 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34740 PR tree-optimization/71374
34741 * lra-constraints.c (check_conflict_input_operands): New.
34742 (match_reload): Use it.
34744 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34747 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
34748 account to calculate conflict_set.
34750 2017-01-27 Bin Cheng <bin.cheng@arm.com>
34752 PR rtl-optimization/78559
34753 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
34754 other_insn in combine.
34756 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
34758 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
34759 uint16_type_node for BT_UINT16.
34761 2017-01-27 David Malcolm <dmalcolm@redhat.com>
34763 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
34764 "RTL Tests" to menu.
34765 (GIMPLE Tests): New node.
34766 (RTL Tests): New node.
34768 2017-01-27 Richard Biener <rguenther@suse.de>
34770 PR tree-optimization/79245
34771 * tree-loop-distribution.c (distribute_loop): Apply cost
34772 modeling also to detected patterns.
34774 2017-01-27 Richard Biener <rguenther@suse.de>
34776 PR tree-optimization/71433
34777 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
34778 (compare_assert_loc): New function.
34779 (process_assert_insertions): Sort and optimize assert locations
34780 to remove duplicates and push down identical assertions on
34781 edges to their destination block.
34783 2017-01-27 Richard Biener <rguenther@suse.de>
34785 PR tree-optimization/79244
34786 * tree-vrp.c (remove_range_assertions): Forcefully propagate
34787 out SSA names even if abnormal.
34789 2017-01-27 Jakub Jelinek <jakub@redhat.com>
34791 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
34792 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
34793 instead of MPFR_RNDN.
34795 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
34798 * arm.c (arm_option_override): Don't call build_target_option_node
34799 until after doing all option overrides.
34800 (arm_valid_target_attribute_tree): Likewise.
34802 2017-01-27 Martin Liska <mliska@suse.cz>
34804 * doc/invoke.texi (-fprofile-arcs): Document profiling support
34805 for {cd}tors and C++ {cd}tors.
34807 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
34809 * config/s390/s390.md ("*setmem_long_and")
34810 ("*setmem_long_and_31z"): Use zero_extend instead of and.
34812 2017-01-26 Martin Sebor <msebor@redhat.com>
34814 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
34817 2017-01-26 Martin Sebor <msebor@redhat.com>
34819 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
34820 HAVE_DFmode before using XFmode or DFmode.
34821 (parse_directive): Avoid using the z length modifier to avoid
34822 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
34824 PR middle-end/78703
34825 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
34826 to accept adjustment as an array.
34827 (get_int_range): New function.
34828 (struct directive): Make width and prec arrays.
34829 (directive::set_width, directive::set_precision): Call get_int_range.
34830 (format_integer, format_floating): Handle width and precision ranges.
34831 (format_string, parse_directive): Same.
34833 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34836 * dwarf2out.c (generate_skeleton_bottom_up): For children with
34837 comdat_type_p set, just clone them, but keep the children in the
34841 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
34842 which have direct callers with -fvar-tracking-assignments enabled
34844 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
34845 inside of type units.
34847 2017-01-26 Martin Sebor <msebor@redhat.com>
34849 PR middle-end/78703
34850 * gimple-ssa-sprintf.c (struct result_range): Add likely and
34852 (struct format_result): Replace number_chars, number_chars_min,
34853 and number_chars_max with a single member of struct result_range.
34855 (format_result::operator+=): Adjust.
34856 (struct fmtresult): Remove bounded. Handle likely and unlikely
34858 (fmtresult::adjust_for_width_or_precision): New function.
34859 (fmtresult:type_max_digits): New function.
34860 (bytes_remaining): Handle likely and unlikely counters.
34861 (min_bytes_remaining): Remove.
34862 (format_percent): Simplify.
34863 (format_integer, format_floating): Set likely and unlikely counters.
34864 (get_string_length, format_character, format_string): Same.
34865 (format_plain, should_warn_p): New function.
34866 (maybe_warn): Call should_warn_p. Update diagnostic messages
34867 and handle those for all directives, including plain strings.
34868 (format_directive): Handle likely and unlikely counters.
34869 Remove unnecessary quoting from diagnostics. Add an informational
34871 (add_bytes): Remove.
34872 (pass_sprintf_length::compute_format_length): Simplify.
34873 (try_substitute_return_value): Handle likely and unlikely counters.
34875 2017-01-26 Carl Love <cel@us.ibm.com>
34877 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
34878 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
34880 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
34883 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
34884 endianess for subregs into account.
34885 * lra-constraints.c (lra_constraints): Do risky transformations
34886 always on the first iteration.
34887 * lra-lives.c (check_pseudos_live_through_calls): Add arg
34888 last_call_used_reg_set.
34889 (process_bb_lives): Define and use last_call_used_reg_set.
34890 * lra.c (lra): Always continue after lra_constraints on the first
34893 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
34895 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
34897 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
34899 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34901 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
34902 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
34903 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
34904 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
34905 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
34906 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
34907 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
34908 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
34909 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
34911 2017-01-26 Marek Polacek <polacek@redhat.com>
34914 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
34915 for the third operand.
34917 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34919 PR middle-end/79236
34920 * omp-low.c (struct omp_context): Add simt_stmt field.
34921 (scan_omp_for): Return omp_context *.
34922 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
34923 context to the _simt_ SIMD stmt.
34924 (lower_omp_for): For combined SIMD with sibling _simt_
34925 SIMD, make sure to use the same decls in _looptemp_
34926 clauses as in the sibling.
34928 2017-01-26 David Sherwood <david.sherwood@arm.com>
34930 PR middle-end/79212
34931 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
34934 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34937 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
34938 emit fld b; fld a; if possible.
34940 * brig-builtins.def: Update copyright years.
34941 * config/arm/arm_acle_builtins.def: Update copyright years.
34943 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
34946 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
34947 constraint instead of o for the stxsd instruction.
34949 2017-01-25 Carl Love <cel@us.ibm.com>
34951 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
34952 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
34954 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
34956 * doc/invoke.texi (C++ Dialect Options): Fix typo.
34958 2017-01-25 Richard Biener <rguenther@suse.de>
34960 PR tree-optimization/69264
34961 * target.def (vector_alignment_reachable): Improve documentation.
34962 * doc/tm.texi: Regenerate.
34963 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
34965 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
34966 earlier changes with respect to TYPE_USER_ALIGN.
34967 (vector_alignment_reachable_p): Likewise. Improve dumping.
34969 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34972 * config/arm/arm.md (xordi3): Force constant operand into a register
34975 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34977 * doc/invoke.texi (-fstore-merging): Correct default optimization
34978 levels at which it is enabled.
34979 (-O): Move -fstore-merging from list to...
34982 2017-01-25 Richard Biener <rguenther@suse.de>
34985 * omp-expand.c: Include debug.h.
34986 (expand_omp_taskreg): Make sure to generate early debug before
34987 outlining anything from a function.
34988 (expand_omp_target): Likewise.
34989 (grid_expand_target_grid_body): Likewise.
34991 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
34994 * asan.c (get_translation_unit_decl): New function.
34995 (asan_add_global): Extract modules file name from globals
34996 TRANSLATION_UNIT_DECL name.
34998 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
35001 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
35002 for long calls with APCS frame and VFP.
35004 2017-01-24 David Malcolm <dmalcolm@redhat.com>
35006 * cfg.c (original_copy_tables_initialized_p): New function.
35007 * cfg.h (original_copy_tables_initialized_p): New decl.
35008 * cfgrtl.c (relink_block_chain): Guard the call to
35009 free_original_copy_tables with a call to
35010 original_copy_tables_initialized_p.
35011 * cgraph.h (symtab_node::native_rtl_p): New decl.
35012 * cgraphunit.c (symtab_node::native_rtl_p): New function.
35013 (symtab_node::needed_p): Don't assert for early assembly output
35014 for __RTL functions.
35015 (cgraph_node::finalize_function): Set "force_output" for __RTL
35017 (cgraph_node::analyze): Bail out early for __RTL functions.
35018 (analyze_functions): Update assertion to support __RTL functions.
35019 (cgraph_node::expand): Bail out early for __RTL functions.
35020 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
35022 * function.h (struct function): Update comment for field
35024 * gimple-expr.c: Include "tree-pass.h".
35025 (gimple_has_body_p): Return false for __RTL functions.
35026 * Makefile.in (OBJS): Add run-rtl-passes.o.
35027 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
35029 (gcc::pass_manager::get_clean_slate): New accessor.
35030 * passes.c: Include "insn-addr.h".
35031 (should_skip_pass_p): Add logging. Update logic for running
35032 "expand" to be compatible with both __GIMPLE and __RTL. Guard
35033 property-provider override so it is only done for gimple passes.
35035 (skip_pass): New function.
35036 (execute_one_pass): Call skip_pass when skipping passes.
35037 * read-md.c (md_reader::read_char): Support filtering
35038 the input to a subset of line numbers.
35039 (md_reader::md_reader): Initialize fields
35040 m_first_line and m_last_line.
35041 (md_reader::read_file_fragment): New function.
35042 * read-md.h (md_reader::read_file_fragment): New decl.
35043 (md_reader::m_first_line): New field.
35044 (md_reader::m_last_line): New field.
35045 * read-rtl-function.c (function_reader::create_function): Only
35046 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
35047 curr_properties. Set DECL_INITIAL to a dummy block.
35048 (read_rtl_function_body_from_file_range): New function.
35049 * read-rtl-function.h (read_rtl_function_body_from_file_range):
35051 * run-rtl-passes.c: New file.
35052 * run-rtl-passes.h: New file.
35054 2017-01-24 Jeff Law <law@redhat.com>
35056 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
35059 2017-01-24 Bin Cheng <bin.cheng@arm.com>
35061 PR tree-optimization/79159
35062 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
35063 (record_nonwrapping_iv): Improve boundary using above function if no
35064 value range information.
35066 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
35067 Martin Jambor <mjambor@suse.cz>
35069 * brig-builtins.def: New file.
35070 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
35071 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
35072 (DEF_HSAIL_SAT_BUILTIN): Likewise.
35073 (DEF_HSAIL_INTR_BUILTIN): Likewise.
35074 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
35075 * builtin-types.def (BT_INT8): New.
35076 (BT_INT16): Likewise.
35077 (BT_UINT8): Likewise.
35078 (BT_UINT16): Likewise.
35079 (BT_FN_ULONG): Likewise.
35080 (BT_FN_UINT_INT): Likewise.
35081 (BT_FN_UINT_ULONG): Likewise.
35082 (BT_FN_UINT_LONG): Likewise.
35083 (BT_FN_UINT_PTR): Likewise.
35084 (BT_FN_ULONG_PTR): Likewise.
35085 (BT_FN_INT8_FLOAT): Likewise.
35086 (BT_FN_INT16_FLOAT): Likewise.
35087 (BT_FN_UINT32_FLOAT): Likewise.
35088 (BT_FN_UINT16_FLOAT): Likewise.
35089 (BT_FN_UINT8_FLOAT): Likewise.
35090 (BT_FN_UINT64_FLOAT): Likewise.
35091 (BT_FN_UINT16_UINT32): Likewise.
35092 (BT_FN_UINT32_UINT16): Likewise.
35093 (BT_FN_UINT16_UINT16_UINT16): Likewise.
35094 (BT_FN_INT_PTR_INT): Likewise.
35095 (BT_FN_UINT_PTR_UINT): Likewise.
35096 (BT_FN_LONG_PTR_LONG): Likewise.
35097 (BT_FN_ULONG_PTR_ULONG): Likewise.
35098 (BT_FN_VOID_UINT64_UINT64): Likewise.
35099 (BT_FN_UINT8_UINT8_UINT8): Likewise.
35100 (BT_FN_INT8_INT8_INT8): Likewise.
35101 (BT_FN_INT16_INT16_INT16): Likewise.
35102 (BT_FN_INT_INT_INT): Likewise.
35103 (BT_FN_UINT_FLOAT_UINT): Likewise.
35104 (BT_FN_FLOAT_UINT_UINT): Likewise.
35105 (BT_FN_ULONG_UINT_UINT): Likewise.
35106 (BT_FN_ULONG_UINT_PTR): Likewise.
35107 (BT_FN_ULONG_ULONG_ULONG): Likewise.
35108 (BT_FN_UINT_UINT_UINT): Likewise.
35109 (BT_FN_VOID_UINT_PTR): Likewise.
35110 (BT_FN_UINT_UINT_PTR: Likewise.
35111 (BT_FN_UINT32_UINT64_PTR): Likewise.
35112 (BT_FN_INT_INT_UINT_UINT): Likewise.
35113 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
35114 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
35115 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
35116 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
35117 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
35118 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
35119 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
35120 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
35121 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
35122 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
35123 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
35124 * doc/frontends.texi: List BRIG FE.
35125 * doc/install.texi (Testing): Add BRIG tesring requirements.
35126 * doc/invoke.texi (Overall Options): Mention BRIG.
35127 * doc/standards.texi (Standards): Doucment BRIG HSA version.
35129 2017-01-24 Richard Biener <rguenther@suse.de>
35131 PR translation/79208
35132 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
35134 2017-01-24 Martin Jambor <mjambor@suse.cz>
35137 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
35138 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
35139 and known_contexts.
35141 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
35143 PR middle-end/79123
35144 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
35145 casts from signed to unsigned really don't have a range.
35147 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
35149 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
35150 GMP_RNDx for compatiblity.
35152 2017-01-24 Martin Liska <mliska@suse.cz>
35155 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
35156 that would prevent us to call alloca with -1 as argument.
35158 2017-01-24 Jakub Jelinek <jakub@redhat.com>
35160 * dwarf2out.c (output_compilation_unit_header, output_file_names):
35161 Avoid -Wformat-security warning.
35163 2017-01-23 Andrew Pinski <apinski@cavium.com>
35165 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
35168 2017-01-23 Martin Sebor <msebor@redhat.com>
35170 PR middle-end/78703
35171 * gimple-ssa-sprintf.c (warn_level): New global.
35172 (format_integer): Use it here and throughout the rest of the file.
35173 Use the same switch to compute sign as base.
35174 (maybe_warn): New function.
35175 (format_directive): Factor out warnings into maybe_warn.
35176 Add debugging output. Use warn_level.
35177 (add_bytes): Use warn_level.
35178 (pass_sprintf_length::compute_format_length): Add debugging output.
35179 (try_substitute_return_value): Same.
35180 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
35182 PR middle-end/78703
35183 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
35184 (struct fmtresult, format_integer, format_floating): Adjust.
35185 (fmtresult::fmtresult): Set max correctly in two argument ctor.
35186 (get_string_length, format_string,format_directive): Same.
35187 (pass_sprintf_length::compute_format_length): Same.
35188 (try_substitute_return_value): Simplify slightly.
35190 PR middle-end/78703
35191 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
35192 (fmtresult::operator+=): Outlined.
35193 (struct fmtresult): Add ctors.
35194 (struct conversion_spec): Rename...
35195 (struct directive): ...to this. Add and remove data members.
35196 (directive::set_width, directive::set_precision): New functions.
35197 (format_percent): Use fmtresult ctor.
35198 (get_width_and_precision): Remove.
35199 (format_integer): Make naming changes. Avoid computing width and
35201 (format_floating): Same. Adjust indentation.
35202 (format_character, format_none): New functions.
35203 (format_string): Moved character handling to format_character.
35204 (format_directive): Remove arguments, change return type.
35205 (parse_directive): New function.
35206 (pass_sprintf_length::compute_format_length): Move directive
35207 parsing to parse_directive.
35209 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35211 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
35212 (assign_assembler_name_if_needed): ... this.
35213 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
35214 (assign_assembler_name_if_needed): ... this.
35215 (free_lang_data_in_cgraph): Adjust callers.
35216 * cgraphunit.c (cgraph_node::analyze): Likewise.
35217 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
35220 2017-01-23 Richard Biener <rguenther@suse.de>
35222 PR tree-optimization/79088
35223 PR tree-optimization/79188
35224 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
35225 resetting loop bounds after last path deletion. Reset loop
35226 bounds of the target loop, make code match the comments.
35227 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
35228 Make sure loops need no fixups.
35230 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35232 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
35233 exponent support with double type for first argument.
35234 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
35235 type returned by __builtin_vec_extract_sig,
35236 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
35237 functions from "vector int" to "vector unsigned int" or from
35238 "vector long long int" to "vector unsigned long long int".
35239 Changed type returned by __builtin_vec_extract_exp,
35240 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
35241 functions from "vector int" to "vector unsigned int" or from
35242 "vector long long int" to "vector unsigned long long int".
35243 Changed return type of __builtin_vec_test_data_class,
35244 __builtin_vec_test_data_class_sp, and
35245 __builtin_vec_test_data_class_dp from "vector int" to
35246 "vector bool int" or from "vector long long int" to "vector bool
35247 long long int" and changed second argument type from "unsigned
35248 int" to "int". Added new overloaded function forms "vector float
35249 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
35250 "vector float __builtin_vec_insert_exp_sp (vector float, vector
35251 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
35252 double, vector unsigned long long int)" and "vector double
35253 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
35254 long int)". Changed return type of
35255 __builtin_scalar_test_data_class and
35256 __builtin_scalar_test_data_class_sp and
35257 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
35258 int" and changed second argument from "unsigned int" to "int".
35259 Changed type returned by __builtin_scalar_test_neg,
35260 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
35261 from "int" to "bool int". Added new overloaded function form
35262 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
35263 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
35264 exponent double-precision with floating point first argument.
35265 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
35266 documentation of scalar_test_data_class, scalar_test_neg,
35267 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
35268 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
35269 vec_test_data_class built-in functions to reflect refinements in
35270 their type signatures.
35272 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
35274 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
35276 (aarch64_elf_asm_destructor): Likewise.
35278 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
35280 PR rtl-optimization/78634
35281 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
35282 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
35283 * ifcvt.c (noce_try_cmove): Add missing cost check.
35285 PR rtl-optimization/71724
35286 * combine.c (if_then_else_cond): Look for situations where it is
35287 beneficial to undo the work of one of the recursive calls.
35289 2017-01-23 Bin Cheng <bin.cheng@arm.com>
35291 PR tree-optimization/70754
35292 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
35293 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
35294 combined stmt before it if not NULL.
35295 (combine_chains): Process refs reversely and compute dominance point
35298 2017-01-23 Martin Liska <mliska@suse.cz>
35300 PR tree-optimization/79196
35301 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
35302 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
35304 (strlen_optimize_stmt): Call the renamed function.
35306 2017-01-23 Michael Matz <matz@suse.de>
35308 PR tree-optimization/78384
35309 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
35311 2017-01-23 Richard Biener <rguenther@suse.de>
35313 PR tree-optimization/79186
35314 * tree-vrp.c (register_new_assert_for): Make sure we've seen
35315 both incoming edges before moving an assert.
35317 2017-01-23 Martin Jambor <mjambor@suse.cz>
35319 * ipa-prop.c (load_from_param_1): Removed.
35320 (load_from_unmodified_param): Bits from load_from_param_1 put back
35322 (load_from_param): Removed.
35323 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
35324 with stmt. Reverted back to use of load_from_unmodified_param.
35326 2017-01-23 Martin Jambor <mjambor@suse.cz>
35329 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
35330 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
35331 field a pointer to garbage collected vector, mark lattices and
35332 ipcp_orig_node with GTY((skip)).
35333 (ipa_get_param_count): Adjust to descriptors being a pointer.
35334 (ipa_get_param): Likewise.
35335 (ipa_get_type): Likewise.
35336 (ipa_get_param_move_cost): Likewise.
35337 (ipa_set_param_used): Likewise.
35338 (ipa_get_controlled_uses): Likewise.
35339 (ipa_set_controlled_uses): Likewise.
35340 (ipa_is_param_used): Likewise.
35341 (ipa_node_params_t): Move into garbage collector. New methods insert
35343 (ipa_node_params_sum): Annotate wth GTY(()).
35344 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
35346 (ipa_load_from_parm_agg): Adjust declaration.
35347 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
35348 * ipa-profile.c (ipa_profile): Likewise.
35349 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
35350 (ipa_populate_param_decls): Make descriptors parameter garbage
35352 (ipa_dump_param): Adjust to descriptors being a pointer.
35353 (ipa_alloc_node_params): Likewise.
35354 (ipa_initialize_node_params): Likewise.
35355 (load_from_param_1): Make descriptors parameter garbage collected.
35356 (load_from_unmodified_param): Likewise.
35357 (load_from_param): Likewise.
35358 (ipa_load_from_parm_agg): Likewise.
35359 (ipa_node_params::~ipa_node_params): Removed.
35360 (ipa_free_all_node_params): Remove call to delete operator.
35361 (ipa_node_params_t::insert): New.
35362 (ipa_node_params_t::remove): Likewise.
35363 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
35364 copy known_csts and known_contexts vectors.
35365 (ipa_read_node_info): Adjust to descriptors being a pointer.
35366 (ipcp_modif_dom_walker): Make m_descriptors field garbage
35368 (ipcp_transform_function): Make descriptors variable garbage
35371 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
35373 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
35374 * config/i386/avx512dqintrin.h: Ditto.
35375 * config/i386/avx512fintrin.h: Ditto.
35376 * config/i386/i386.c: Handle new builtins.
35377 * config/i386/i386-builtin.def: Add new builtins.
35378 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
35379 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
35381 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35382 Martin Liska <mliska@suse.cz>
35384 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
35385 * asan.c (asan_expand_poison_ifn): Support stores and use
35386 appropriate ASAN report function.
35387 * internal-fn.c (expand_ASAN_POISON_USE): New function.
35388 * internal-fn.def (ASAN_POISON_USE): Declare.
35389 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
35390 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
35391 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
35392 ASAN_POISON calls w/o LHS.
35393 * tree-ssa.c (execute_update_addresses_taken): Create clobber
35394 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
35395 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
35396 * gimplify.c (asan_poison_variables): Add attribute
35397 use_after_scope_memory to variables that really needs to live
35399 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
35400 having the attribute.
35402 2017-01-23 Martin Liska <mliska@suse.cz>
35404 * asan.c (create_asan_shadow_var): New function.
35405 (asan_expand_poison_ifn): Likewise.
35406 * asan.h (asan_expand_poison_ifn): New declaration.
35407 * internal-fn.c (expand_ASAN_POISON): Likewise.
35408 * internal-fn.def (ASAN_POISON): New builtin.
35409 * sanopt.c (pass_sanopt::execute): Expand
35410 asan_expand_poison_ifn.
35411 * tree-inline.c (copy_decl_for_dup_finish): Make function
35413 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
35414 * tree-ssa.c (is_asan_mark_p): New function.
35415 (execute_update_addresses_taken): Rewrite local variables
35416 (identified just by use-after-scope as addressable) into SSA.
35418 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35420 * doc/install.texi (Specific): opensource.apple.com uses https
35421 now. Remove trailing slash.
35423 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35425 * README.Portability: Remove note on an Irix compatibility issue.
35427 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
35429 * gcov.c (INCLUDE_ALGORITHM): Define.
35430 (INCLUDE_VECTOR): Define.
35431 No longer include <vector> and <algorithm> directly.
35433 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
35435 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
35437 * doc/invoke.texi (Code Gen Options): Ditto.
35439 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
35442 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
35444 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
35446 rtl-optimization/79125
35447 * cprop.c (local_cprop_pass): Handle cases where we make an
35448 unconditional trap.
35450 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
35454 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
35455 read from, for big endian.
35457 2017-01-20 Jiong Wang <jiong.wang@arm.com>
35459 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
35460 register pauth builtins for LP64 only.
35462 2017-01-20 Marek Polacek <polacek@redhat.com>
35465 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
35468 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35470 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
35472 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
35473 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
35474 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
35476 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35479 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
35480 in big-endian mode when they are not a single duplicated value.
35482 2017-01-20 Richard Biener <rguenther@suse.de>
35484 * BASE-VER: Bump to 7.0.1.
35486 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35488 * omp-low.c (omplow_simd_context): New struct. Use it...
35489 (lower_rec_simd_input_clauses): ...here and...
35490 (lower_rec_input_clauses): ...here to hold common data. Adjust all
35491 references to idx, lane, max_vf, is_simt.
35493 2017-01-20 Graham Markall <graham.markall@embecosm.com>
35495 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
35498 2017-01-20 Martin Jambor <mjambor@suse.cz>
35500 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
35501 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
35503 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
35504 (GTFILES): Rename hsa.c to hsa-common.c.
35505 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
35506 * hsa-dump.c: Likewise.
35507 * hsa-gen.c: Likewise.
35508 * hsa-regalloc.c: Likewise.
35509 * ipa-hsa.c: Likewise.
35510 * omp-expand.c: Likewise.
35511 * omp-low.c: Likewise.
35512 * toplev.c: Likewise.
35514 2017-01-20 Marek Polacek <polacek@redhat.com>
35517 * doc/invoke.texi: Document -Wduplicated-branches.
35518 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
35519 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
35520 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
35521 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
35522 return 0 only when not OEP_LEXICOGRAPHIC.
35523 (fold_build_cleanup_point_expr): Use the expression
35524 location when building CLEANUP_POINT_EXPR.
35525 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
35526 * tree.c (add_expr): Handle error_mark_node.
35528 2017-01-20 Martin Liska <mliska@suse.cz>
35531 * tree-profile.c (init_ic_make_global_vars): Do not call
35533 (gimple_init_gcov_profiler): Likewise.
35535 2017-01-20 Martin Liska <mliska@suse.cz>
35538 * cgraph.h (maybe_create_reference): Remove argument and
35540 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
35542 * ipa-cp.c (create_specialized_node): Likewise.
35543 * symtab.c (symtab_node::maybe_create_reference): Handle
35544 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
35546 2017-01-20 Martin Liska <mliska@suse.cz>
35548 * read-rtl-function.c (function_reader::create_function): Use
35549 build_decl instread of build_decl_stat.
35551 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
35553 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
35554 * config/i386/avx512dqintrin.h: Ditto.
35555 * config/i386/avx512fintrin.h: Ditto.
35556 * config/i386/i386-builtin-types.def: Add new types.
35557 * config/i386/i386.c: Handle new types.
35558 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
35559 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
35560 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
35561 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
35562 (__builtin_ia32_kshiftridi): New.
35563 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
35565 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
35569 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
35570 define to rs6000_init_stack_protect_guard.
35571 (rs6000_init_stack_protect_guard): New function.
35573 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
35574 Yunqiang Su <yunqiang.su@imgtec.com>
35576 * config.gcc (supported_defaults): Add madd4.
35577 (with_madd4): Add validation.
35578 (all_defaults): Add madd4.
35579 * config/mips/mips.opt (mmadd4): New option.
35580 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35582 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35584 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
35585 (ISA_HAS_FUSED_MADD4): Likewise.
35586 * doc/invoke.texi (-mmadd4): Document the new option.
35587 * doc/install.texi (--with-madd4): Document the new option.
35589 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35591 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
35592 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
35593 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
35594 (aarch64_init_pauth_hint_builtins): New.
35595 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
35596 (aarch64_expand_builtin): Expand new builtins.
35598 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35600 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
35601 * combine-stack-adj.c (no_unhandled_cfa): Handle
35602 REG_CFA_TOGGLE_RA_MANGLE.
35603 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
35604 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
35605 info for return address signing.
35606 (aarch64_expand_epilogue): Likewise.
35608 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35610 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
35611 * config/aarch64/aarch64-protos.h
35612 (aarch64_return_address_signing_enabled): New declaration.
35613 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
35615 (aarch64_expand_prologue): Sign return address before it's pushed onto
35617 (aarch64_expand_epilogue): Authenticate return address fetched from
35619 (aarch64_override_options): Sanity check for ILP32 and ISA level.
35620 (aarch64_attributes): New function attributes for "sign-return-address".
35621 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
35622 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
35623 ("*do_return"): Generate combined instructions according to key index.
35624 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
35625 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
35627 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
35628 * config/aarch64/aarch64.opt (msign-return-address=): New.
35629 * doc/extend.texi (AArch64 Function Attributes): Documents
35630 "sign-return-address=".
35631 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
35633 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
35635 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
35636 overall option summary.
35638 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35640 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
35641 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
35642 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
35643 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
35645 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
35647 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
35648 -mpower9-minmax by default for -mcpu=power9.
35649 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
35650 128-bit floating point.
35652 2017-01-20 Alan Modra <amodra@gmail.com>
35654 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
35655 optimizing for size.
35657 2017-01-20 Alan Modra <amodra@gmail.com>
35660 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
35661 for strcmp and strncmp from corresponding builtin decl.
35663 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
35665 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
35666 instead of i386/rtems-64.h.
35667 * config/i386/rtems-64.h: Remove.
35669 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
35673 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
35675 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
35677 2017-01-19 Tamar Christina <tamar.christina@arm.com>
35679 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
35680 Change int to HOST_WIDE_INT.
35681 * config/aarch64/aarch64-protos.h
35682 (aarch64_simd_gen_const_vector_dup): Likewise.
35683 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
35685 2017-01-19 David Malcolm <dmalcolm@redhat.com>
35687 * langhooks-def.h (lhd_type_for_size): New decl.
35688 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
35689 * langhooks.c (lhd_type_for_size): New function, taken from
35692 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
35694 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
35695 define_bypass for CR latency.
35696 (power9-cracked-alu): Update bypass latency and remove power9-branch.
35697 (power9-alu2): Add define_bypass for CR latency.
35699 (power9-mul): Update insn latency.
35700 (power9-mul-compare): Update insn latency, bypass latency and remove
35703 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35705 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
35707 * config/aarch64/aarch64.md
35708 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
35709 aarch64_nopcrelative_literal_loads.
35710 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
35712 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
35714 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
35715 TARGET_LOONGSON_3A.
35716 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
35718 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
35721 * config.gcc (supported_defaults): Add lxc1-sxc1.
35722 (with_lxc1_sxc1): Add validation.
35723 (all_defaults): Add lxc1-sxc1.
35724 * config/mips/mips.opt (mlxc1-sxc1): New option.
35725 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35727 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35728 __mips_no_lxc1_sxc1.
35729 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
35730 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
35731 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
35733 2017-01-19 Richard Biener <rguenther@suse.de>
35735 PR tree-optimization/72488
35736 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
35737 sure to restore SSA info.
35738 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
35740 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
35742 PR rtl-optimization/79121
35743 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
35744 of the inner type when shifting an extended value.
35746 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
35749 * symtab.c (symtab_node::equal_address_to): Fix comparing of
35750 interposable aliases.
35752 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
35755 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
35756 Use the evmergelohi instruction.
35757 (mov_si<mode>_e500_subreg4_2_le): Likewise.
35758 (mov_sitf_e500_subreg8_2_be): Likewise.
35759 (mov_sitf_e500_subreg12_2_le): Likewise.
35760 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
35761 (mov_si<mode>_e500_subreg4_2_be): Likewise.
35762 (mov_sitf_e500_subreg8_2_le): Likewise.
35763 (mov_sitf_e500_subreg12_2_be): Likewise.
35765 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35767 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
35768 attribute from vecsimple to vecperm.
35769 (altivec_vbpermq2): Likewise.
35771 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35774 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
35776 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35777 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
35778 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
35779 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
35780 case where N arg is SIZE_MAX.
35781 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
35782 (cmpstrsi): Add pattern.
35784 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
35786 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
35787 __builtin_vec_revb builtins.
35788 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
35789 built-in functions to support generation of the ISA 3.0 XXBR<x>
35790 vector byte reverse instructions.
35791 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
35792 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
35793 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
35794 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
35795 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
35796 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
35797 (P9V_BUILTIN_VEC_REVB): Likewise.
35798 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
35799 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
35800 (p9_xxbrq_v16qi): Likewise.
35801 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
35802 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
35803 (p9_xxbrh_v8hi): Likewise.
35804 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
35805 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
35806 vec_revb built-in functions.
35808 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
35810 PR rtl-optimization/78952
35811 * config/i386/i386.md (any_extract): New code iterator.
35812 (*insvqi_2): Use any_extract for source operand.
35813 (*insvqi_3): Use any_shiftrt for source operand.
35815 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
35817 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
35819 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
35821 2017-01-18 Matthias Klose <doko@ubuntu.com>
35823 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
35825 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35827 * config/rs6000/altivec.h (vec_bperm): Change #define.
35828 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
35829 (altivec_vbpermq2): New define_insn.
35830 (altivec_vbpermd): Likewise.
35831 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
35832 function interface.
35833 (VBPERMD): Likewise.
35834 (VBPERM): New polymorphic function interface.
35835 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
35836 Add entries for P9V_BUILTIN_VEC_VBPERM.
35837 * doc/extend.texi: Add interfaces for vec_bperm.
35839 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35841 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
35842 first letter of error messages.
35843 (s390_resolve_overloaded_builtin): Likewise.
35844 * config/s390/s390.c (s390_expand_builtin): Likewise.
35845 (s390_invalid_arg_for_unprototyped_fn): Likewise.
35846 (s390_valid_target_attribute_inner_p): Likewise.
35847 * config/s390/s390.md ("tabort"): Likewise.
35849 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
35851 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
35852 (ISA_AVOID_DIV_HILO): New macro.
35853 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
35854 (ISA_HAS_DDIV): Likewise.
35856 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35858 * doc/invoke.texi (fabi-version): Correct number of occurrences.
35860 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35862 * doc/invoke.texi (fabi-version): Spelling fix.
35864 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35867 * doc/invoke.texi (fabi-version): Mention mangling fix for
35870 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35873 * doc/invoke.texi (fabi-version): Document discriminator mangling.
35875 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
35878 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
35879 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
35881 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
35883 (stack_protect_test): Ditto.
35884 * config/rs6000/rs6000.opt (mstack-protector-guard=,
35885 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
35887 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
35888 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
35889 -mstack-protector-guard-offset=.
35890 (RS/6000 and PowerPC Options): Ditto.
35892 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
35894 * config/i386/i386.h (MASK_CLASS_P): New define.
35895 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
35896 there are no registers from different register sets also when
35897 mask registers are used. Update function comment.
35898 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
35899 to (*k/*r) and (*k/*km) alternatives.
35901 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
35903 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
35904 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
35905 (EH_RETURN_HANDLER_RTX): New define.
35906 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
35907 Force frame pointer in EH return functions.
35908 (aarch64_expand_epilogue): Add barrier for eh_return.
35909 (aarch64_final_eh_return_addr): Remove.
35910 (aarch64_eh_return_handler_rtx): New function.
35911 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
35913 (aarch64_eh_return_handler_rtx): New prototype.
35915 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35917 * config/rs6000/altivec.h (vec_rlmi): New #define.
35918 (vec_vrlnm): Likewise.
35919 (vec_rlnm): Likewise.
35920 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
35921 (UNSPEC_VRLNM): Likewise.
35922 (VIlong): New mode iterator.
35923 (altivec_vrl<VI_char>mi): New define_insn.
35924 (altivec_vrl<VI_char>nm): Likewise.
35925 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
35927 (VRLDNM): Likewise.
35928 (RLNM): New polymorphic function entry.
35929 (VRLWMI): New monomorphic function entry.
35930 (VRLDMI): Likewise.
35931 (RLMI): New polymorphic function entry.
35932 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
35933 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
35934 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
35937 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35940 * dwarf2out.c (field_byte_offset): Restore the
35941 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
35942 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
35943 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
35946 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
35949 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
35951 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35954 * dwarf2out.c (add_data_member_location_attribute): For constant
35955 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
35956 instead of DW_AT_data_member_location, DW_AT_bit_offset and
35957 DW_AT_byte_size attributes.
35959 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
35961 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
35962 after forcing to constant memory when the code model is medium.
35964 2017-01-17 Julia Koval <julia.koval@intel.com>
35967 * config/i386/avx512fintrin.h
35968 (_mm512_i32gather_ps): Change __addr type to void const*.
35969 (_mm512_mask_i32gather_ps): Ditto.
35970 (_mm512_i32gather_pd): Ditto.
35971 (_mm512_mask_i32gather_pd): Ditto.
35972 (_mm512_i64gather_ps): Ditto.
35973 (_mm512_mask_i64gather_ps): Ditto.
35974 (_mm512_i64gather_pd): Ditto.
35975 (_mm512_mask_i64gather_pd): Ditto.
35976 (_mm512_i32gather_epi32): Ditto.
35977 (_mm512_mask_i32gather_epi32): Ditto.
35978 (_mm512_i32gather_epi64): Ditto.
35979 (_mm512_mask_i32gather_epi64): Ditto.
35980 (_mm512_i64gather_epi32): Ditto.
35981 (_mm512_mask_i64gather_epi32): Ditto.
35982 (_mm512_i64gather_epi64): Ditto.
35983 (_mm512_mask_i64gather_epi64): Ditto.
35984 (_mm512_i32scatter_ps): Change __addr type to void*.
35985 (_mm512_mask_i32scatter_ps): Ditto.
35986 (_mm512_i32scatter_pd): Ditto.
35987 (_mm512_mask_i32scatter_pd): Ditto.
35988 (_mm512_i64scatter_ps): Ditto.
35989 (_mm512_mask_i64scatter_ps): Ditto.
35990 (_mm512_i64scatter_pd): Ditto.
35991 (_mm512_mask_i64scatter_pd): Ditto.
35992 (_mm512_i32scatter_epi32): Ditto.
35993 (_mm512_mask_i32scatter_epi32): Ditto.
35994 (_mm512_i32scatter_epi64): Ditto.
35995 (_mm512_mask_i32scatter_epi64): Ditto.
35996 (_mm512_i64scatter_epi32): Ditto.
35997 (_mm512_mask_i64scatter_epi32): Ditto.
35998 (_mm512_i64scatter_epi64): Ditto.
35999 (_mm512_mask_i64scatter_epi64): Ditto.
36000 * config/i386/avx512pfintrin.h
36001 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
36002 (_mm512_mask_prefetch_i32gather_ps): Ditto.
36003 (_mm512_mask_prefetch_i64gather_pd): Ditto.
36004 (_mm512_mask_prefetch_i64gather_ps): Ditto.
36005 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
36006 (_mm512_prefetch_i32scatter_ps): Ditto.
36007 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
36008 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
36009 (_mm512_prefetch_i64scatter_pd): Ditto.
36010 (_mm512_prefetch_i64scatter_ps): Ditto.
36011 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
36012 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
36013 * config/i386/avx512vlintrin.h
36014 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
36015 (_mm_mmask_i32gather_ps): Ditto.
36016 (_mm256_mmask_i32gather_pd): Ditto.
36017 (_mm_mmask_i32gather_pd): Ditto.
36018 (_mm256_mmask_i64gather_ps): Ditto.
36019 (_mm_mmask_i64gather_ps): Ditto.
36020 (_mm256_mmask_i64gather_pd): Ditto.
36021 (_mm_mmask_i64gather_pd): Ditto.
36022 (_mm256_mmask_i32gather_epi32): Ditto.
36023 (_mm_mmask_i32gather_epi32): Ditto.
36024 (_mm256_mmask_i32gather_epi64): Ditto.
36025 (_mm_mmask_i32gather_epi64): Ditto.
36026 (_mm256_mmask_i64gather_epi32): Ditto.
36027 (_mm_mmask_i64gather_epi32): Ditto.
36028 (_mm256_mmask_i64gather_epi64): Ditto.
36029 (_mm_mmask_i64gather_epi64): Ditto.
36030 (_mm256_i32scatter_ps): Change __addr type to void*.
36031 (_mm256_mask_i32scatter_ps): Ditto.
36032 (_mm_i32scatter_ps): Ditto.
36033 (_mm_mask_i32scatter_ps): Ditto.
36034 (_mm256_i32scatter_pd): Ditto.
36035 (_mm256_mask_i32scatter_pd): Ditto.
36036 (_mm_i32scatter_pd): Ditto.
36037 (_mm_mask_i32scatter_pd): Ditto.
36038 (_mm256_i64scatter_ps): Ditto.
36039 (_mm256_mask_i64scatter_ps): Ditto.
36040 (_mm_i64scatter_ps): Ditto.
36041 (_mm_mask_i64scatter_ps): Ditto.
36042 (_mm256_i64scatter_pd): Ditto.
36043 (_mm256_mask_i64scatter_pd): Ditto.
36044 (_mm_i64scatter_pd): Ditto.
36045 (_mm_mask_i64scatter_pd): Ditto.
36046 (_mm256_i32scatter_epi32): Ditto.
36047 (_mm256_mask_i32scatter_epi32): Ditto.
36048 (_mm_i32scatter_epi32): Ditto.
36049 (_mm_mask_i32scatter_epi32): Ditto.
36050 (_mm256_i32scatter_epi64): Ditto.
36051 (_mm256_mask_i32scatter_epi64): Ditto.
36052 (_mm_i32scatter_epi64): Ditto.
36053 (_mm_mask_i32scatter_epi64): Ditto.
36054 (_mm256_i64scatter_epi32): Ditto.
36055 (_mm256_mask_i64scatter_epi32): Ditto.
36056 (_mm_i64scatter_epi32): Ditto.
36057 (_mm_mask_i64scatter_epi32): Ditto.
36058 (_mm256_i64scatter_epi64): Ditto.
36059 (_mm256_mask_i64scatter_epi64): Ditto.
36060 (_mm_i64scatter_epi64): Ditto.
36061 (_mm_mask_i64scatter_epi64): Ditto.
36062 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
36063 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
36064 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
36065 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
36066 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
36067 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
36068 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
36069 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
36070 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
36071 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
36072 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
36073 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
36074 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
36075 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
36076 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
36077 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
36078 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
36079 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
36080 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
36081 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
36082 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
36083 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
36084 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
36085 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
36086 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
36087 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
36088 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
36089 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
36090 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
36091 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
36092 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
36093 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
36094 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
36095 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
36096 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
36097 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
36098 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
36099 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
36100 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
36101 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
36102 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
36103 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
36104 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
36105 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
36106 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
36107 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
36108 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
36109 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
36110 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
36111 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
36112 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
36113 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
36114 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
36115 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
36116 definitions accordingly.
36118 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
36119 Kuan-Lin Chen <kuanlinchentw@gmail.com>
36122 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
36125 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
36128 * ira-conflicts.c (ira_build_conflicts): Update total conflict
36129 hard regs for inner regno.
36131 2017-01-17 Martin Liska <mliska@suse.cz>
36134 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
36135 assumption and add comment.
36137 2017-01-17 Nathan Sidwell <nathan@acm.org>
36139 * ipa-visibility.c (localize_node): New function, broken out of ...
36140 (function_and_variable_visibility): ... here. Call it.
36142 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36144 PR middle-end/77445
36145 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
36146 correctly set frequency of oudgoing edge.
36147 (duplicate_thread_path): Fix profile updating.
36149 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36152 * configure.ac: Add GCC_BASE_VER.
36153 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
36154 version from BASE-VER file.
36155 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
36156 (gcc.o): Depend on $(BASEVER).
36157 * common.opt (dumpfullversion): New option.
36158 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
36159 * doc/invoke.texi: Document -dumpfullversion.
36160 * doc/install.texi: Document --with-gcc-major-version-only.
36161 * configure: Regenerated.
36163 2017-01-17 Richard Biener <rguenther@suse.de>
36165 PR tree-optimization/71433
36166 * tree-vrp.c (register_new_assert_for): Merge same asserts
36167 on all incoming edges.
36168 (process_assert_insertions_for): Handle insertions at the
36171 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
36173 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
36174 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
36176 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
36179 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
36182 2017-01-17 Alan Modra <amodra@gmail.com>
36185 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
36186 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
36187 symbolic stack limit when pic.
36189 2017-01-16 Martin Sebor <msebor@redhat.com>
36191 PR tree-optimization/78608
36192 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
36194 2017-01-16 Jeff Law <law@redhat.com>
36197 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
36198 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
36199 for several include directories that may be relative to sysroot.
36200 * config/i386/x-mingw32 (gplus_includedir): Define.
36201 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
36202 (native_system_includedir): Likewise.
36203 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
36204 override if TARGET_SYSTEM_ROOT is defined.
36205 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
36207 PR tree-optimization/79090
36208 PR tree-optimization/33562
36209 PR tree-optimization/61912
36210 PR tree-optimization/77485
36211 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
36212 and computed trims into the dump file.
36214 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36216 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
36218 2017-01-16 Jakub Jelinek <jakub@redhat.com>
36221 * gimplify.c (gimplify_init_constructor): If want_value and
36222 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
36226 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
36227 sequence. Formatting fixes.
36228 (doloop_optimize): Formatting fixes.
36231 * gcc.c (debug_level_greater_than_spec_func): New function.
36232 (static_spec_functions): Add debug-level-gt spec function.
36233 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
36235 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
36236 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
36237 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
36238 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
36239 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
36240 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
36242 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
36244 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
36245 QImode fixups to general and mask registers only.
36247 2017-01-16 Carl Love <cel@us.ibm.com>
36249 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
36250 for built-in functions
36251 vector signed char vec_nabs (vector signed char)
36252 vector signed short vec_nabs (vector signed short)
36253 vector signed int vec_nabs (vector signed int)
36254 vector signed long long vec_nabs (vector signed long long)
36255 vector float vec_nabs (vector float)
36256 vector double vec_nabs (vector double)
36257 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
36259 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
36260 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
36261 * doc/extend.texi: Update the documentation file for the new built-in
36264 2017-01-16 Martin Sebor <msebor@redhat.com>
36266 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
36269 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36271 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
36272 UNSPEC_VSX__XXSPLTD to require special splat handling.
36274 2017-01-16 David Malcolm <dmalcolm@redhat.com>
36277 * system.h: Poison strndup.
36279 2017-01-16 Alan Modra <amodra@gmail.com>
36282 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
36285 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
36287 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
36289 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
36291 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
36292 call recog here. Assert that INSN_CODE (insn) is non-negative.
36294 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
36297 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
36299 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
36300 in the currently scheduled RTL fragment.
36302 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
36304 PR rtl-optimization/78751
36305 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
36308 2017-01-14 Jeff Law <law@redhat.com>
36310 PR tree-optimization/79090
36311 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
36312 variable length stores.
36313 (compute_trims): Delete dead assignment to *trim_tail.
36314 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
36317 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
36319 PR rtl-optimization/78626
36320 PR rtl-optimization/78727
36321 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
36322 of a block, and split such blocks after everything else is finished.
36324 2017-01-14 Alan Modra <amodra@gmail.com>
36327 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
36328 target legitimate_combined_insn.
36329 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
36330 (rs6000_legitimate_combined_insn): New function.
36331 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
36333 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
36334 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
36335 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
36337 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
36339 * doc/frontends.texi (G++ and GCC): Remove references to Java.
36341 2017-01-13 Jeff Law <law@redhat.com>
36343 PR tree-optimization/33562
36344 PR tree-optimization/61912
36345 PR tree-optimization/77485
36346 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
36348 (delete_dead_assignment): Likewise.
36349 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
36350 statement to delete_dead_call and delete_dead_assignment.
36352 2017-01-13 David Malcolm <dmalcolm@redhat.com>
36355 * substring-locations.c (format_warning_va): Strengthen case 1 so
36356 that both endpoints of the substring must be within the format
36357 range for just the substring to be printed.
36359 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
36361 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
36362 * config/i386/i386.c (ix86_target_string): Add missing options
36363 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
36364 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
36365 flags_other and ix86_target_other to flags2_other. Display unknown
36367 (ix86_valid_target_attribute_inner_p): Add missing options and
36368 reorder options by implied ISAs, as in ix86_target_string.
36370 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36372 * hash-table.h (hash_table::too_empty_p): New function.
36373 (hash_table::expand): Use it.
36374 (hash_table::traverse): Likewise.
36375 (hash_table::empty_slot): Use sizeof (value_type) instead of
36376 sizeof (PTR) to convert bytes to elements. Shrink the table
36377 if the current size is excessive for the current number of
36380 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36382 * ira-costs.c (record_reg_classes): Break from the inner loop
36383 early once alt_fail is known to be true. Update outer loop
36384 handling accordingly.
36386 2017-01-13 Jeff Law <law@redhat.com>
36388 * tree-ssa-dse.c (decrement_count): New function.
36389 (increment_start_addr, maybe_trim_memstar_call): Likewise.
36390 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
36391 when we know the partially dead statement is a mem* function.
36393 PR tree-optimization/61912
36394 PR tree-optimization/77485
36395 * tree-ssa-dse.c: Include expr.h.
36396 (maybe_trim_constructor_store): New function.
36397 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
36399 PR tree-optimization/33562
36400 PR tree-optimization/61912
36401 PR tree-optimization/77485
36402 * doc/invoke.texi: Document new dse-max-object-size param.
36403 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
36404 * tree-ssa-dse.c: Include params.h.
36405 (dse_store_status): New enum.
36406 (initialize_ao_ref_for_dse): New, partially extracted from
36408 (valid_ao_ref_for_dse, normalize_ref): New.
36409 (setup_live_bytes_from_ref, compute_trims): Likewise.
36410 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
36411 (maybe_trim_partially_dead_store): Likewise.
36412 (maybe_trim_complex_store): Likewise.
36413 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
36414 Track what bytes live from the original store. Return tri-state
36415 for dead, partially dead or live.
36416 (dse_dom_walker): Add constructor, destructor and new private members.
36417 (delete_dead_call, delete_dead_assignment): New extracted from
36419 (dse_optimize_stmt): Make a member of dse_dom_walker.
36420 Use initialize_ao_ref_for_dse.
36422 PR tree-optimization/33562
36423 PR tree-optimization/61912
36424 PR tree-optimization/77485
36425 * sbitmap.h (bitmap_count_bits): Prototype.
36426 (bitmap_clear_range, bitmap_set_range): Likewise.
36427 * sbitmap.c (bitmap_clear_range): New function.
36428 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
36430 2017-01-13 Martin Liska <mliska@suse.cz>
36433 * function.c (set_cfun): Add new argument force.
36434 * function.h (set_cfun): Likewise.
36435 * ipa-inline-transform.c (inline_call): Use the function when
36436 strict alising from is dropped for function we inline to.
36438 2017-01-13 Richard Biener <rguenther@suse.de>
36440 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
36441 for dumping GIMPLE INTEGER_CSTs.
36443 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
36445 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
36446 to 201112L since C++17.
36448 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
36451 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
36452 if -fsanitize=kernel-address is present.
36454 2017-01-13 Richard Biener <rguenther@suse.de>
36456 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
36457 as _Literal ( type ) number in case usual suffixes do not
36458 preserve all information.
36460 2017-01-13 Richard Biener <rguenther@suse.de>
36462 PR tree-optimization/77283
36463 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
36464 and ssa-iterators.h.
36465 (is_feasible_trace): Implement a cost model based on joiner
36468 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
36471 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
36472 char or short to __float128/_Float128 directly.
36474 2017-01-12 Martin Sebor <msebor@redhat.com>
36476 to -Wformat-overflow.
36477 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
36478 (min_bytes_remaining): Same.
36479 (get_string_length): Same.
36480 (format_string): Same.
36481 (format_directive): Same.
36483 (pass_sprintf_length::handle_gimple_call): Same.
36485 2017-01-12 Jakub Jelinek <jakub@redhat.com>
36487 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
36488 info.nowrite calls with no lhs that can't throw. Return bool
36489 whether gsi_remove has been called or not.
36490 (pass_sprintf_length::handle_gimple_call): Return bool whether
36491 try_substitute_return_value called gsi_remove. Formatting fix.
36492 (pass_sprintf_length::execute): Don't use gsi_remove if
36493 handle_gimple_call returned true.
36496 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
36497 be removed due to side-effects, don't remove following barrier nor
36498 turn the successor edge into fallthru edge.
36500 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36503 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
36504 element-reversing loads and stores as not swappable.
36506 2017-01-12 Nathan Sidwell <nathan@acm.org>
36507 Nicolai Stange <nicstange@gmail.com>
36509 * combine.c (try_combine): Don't ignore result of overlap checking
36510 loop. Combine overlap & asm check into single loop.
36512 2017-01-12 Richard Biener <rguenther@suse.de>
36514 * tree-pretty-print.c (dump_generic_node): Provide -gimple
36515 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
36517 2017-01-12 Richard Biener <rguenther@suse.de>
36519 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
36520 and TS_TARGET_OPTION directly derive from TS_BASE.
36521 * tree-core.h (tree_optimization_option): Derive from tree_base.
36522 (tree_target_option): Likewise.
36524 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
36526 * config/i386/i386.c (memory_address_length): Increase len
36527 only when rip_relative_addr_p returns false.
36529 2017-01-11 Julia Koval <julia.koval@intel.com>
36531 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
36532 (OPTION_MASK_ISA_SGX_SET): New.
36533 (ix86_handle_option): Handle OPT_msgx.
36534 * config.gcc: Added sgxintrin.h.
36535 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
36536 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
36537 * config/i386/i386.c (ix86_target_string): Add -msgx.
36539 (ix86_option_override_internal): Handle new options.
36540 (ix86_valid_target_attribute_inner_p): Add sgx.
36541 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
36542 * config/i386/i386.opt: Add msgx.
36543 * config/i386/sgxintrin.h: New file.
36544 * config/i386/x86intrin.h: Add sgxintrin.h.
36546 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36549 * fold-const.c (maybe_nonzero_address): Return 1 for function
36551 (tree_single_nonzero_warnv_p): Don't handle function local objects
36555 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
36558 2017-01-11 David Malcolm <dmalcolm@redhat.com>
36561 * opts.c: Include "spellcheck.h"
36562 (struct string_fragment): New struct.
36563 (struct edit_distance_traits<const string_fragment &>): New
36565 (get_closest_sanitizer_option): New function.
36566 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
36568 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36570 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
36572 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
36573 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
36574 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
36575 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
36576 for initial die_offset if dwarf_split_debug_info.
36577 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
36578 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
36580 (output_skeleton_debug_sections): Formatting fix. Use
36581 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
36582 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
36584 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
36586 * config/arm/cortex-a53.md: Add bypasses for
36587 cortex_a53_r2f_cvt.
36588 (cortex_a53_r2f): Only use for transfers.
36589 (cortex_a53_f2r): Likewise.
36590 (cortex_a53_r2f_cvt): Add reservation for conversions.
36591 (cortex_a53_f2r_cvt): Likewise.
36593 2017-01-11 Tamar Christina <tamar.christina@arm.com>
36595 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
36596 to all inlined functions, change static to extern.
36598 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
36601 * config/arm/arm.c (legitimize_pic_address): Handle reference to
36603 (arm_assemble_integer): Likewise.
36605 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
36607 * config.gcc: Use new awk script to check CPU, FPU and architecture
36608 parameters for --with-... options.
36609 * config/arm/parsecpu.awk: New file
36610 * config/arm/arm-cpus.in: New file.
36611 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
36613 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
36615 * config/arm/t-arm: Update dependency rules.
36616 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
36617 of processing .def files.
36618 * config/arm/genopt.sh: Deleted.
36619 * config/arm/gentune.sh: Deleted.
36620 * config/arm/arm-cores.def: Deleted.
36621 * config/arm/arm-arches.def: Deleted.
36622 * config/arm/arm-fpus.def: Deleted.
36623 * config/arm/arm-tune.md: Regenerated.
36624 * config/arm/arm-tables.opt: Regenerated.
36625 * config/arm/arm-cpu.h: New generated file.
36626 * config/arm/arm-cpu-data.h: New generated file.
36627 * config/arm/arm-cpu-cdata.h: New generated file.
36629 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
36632 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
36634 (input_varpool_node): Unpack dynamically_initialized bit.
36636 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
36638 PR rtl-optimization/79032
36639 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
36640 the alignment of the adjusted memory reference against that of MODE,
36641 instead of the alignment of the original memory reference.
36643 2017-01-11 Martin Jambor <mjambor@suse.cz>
36645 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
36647 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
36648 decorated functions.
36650 2017-01-11 Richard Biener <rguenther@suse.de>
36652 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
36653 set range/nonnull info for PHI results. Do not set it on
36654 stmts marked for removal.
36656 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
36658 * expr.c (store_field): In the bitfield case, fetch the return value
36659 from the registers before applying a single big-endian adjustment.
36660 Always do a final load for a BLKmode value not larger than a word.
36662 2017-01-10 David Malcolm <dmalcolm@redhat.com>
36665 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
36666 that we correctly handle column numbers greater than
36667 LINE_MAP_MAX_COLUMN_NUMBER.
36669 2017-01-10 Martin Sebor <msebor@redhat.com>
36671 PR middle-end/78245
36672 * gimple-ssa-sprintf.c (get_destination_size): Call
36673 {init,fini}object_sizes.
36674 * tree-object-size.c (addr_object_size): Adjust.
36675 (pass_through_call): Adjust.
36676 (pass_object_sizes::execute): Adjust.
36677 * tree-object-size.h (fini_object_sizes): Declare.
36679 2017-01-10 Martin Sebor <msebor@redhat.com>
36681 PR tree-optimization/78775
36682 * builtins.c (get_size_range): Move...
36683 * calls.c: ...to here.
36684 (alloc_max_size): Accept zero argument.
36685 (operand_signed_p): Remove.
36686 (maybe_warn_alloc_args_overflow): Call get_size_range.
36687 * calls.h (get_size_range): Declare.
36689 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
36691 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
36692 from TI's devices.csv file as of September 2016.
36693 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
36695 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
36697 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
36698 * doc/invoke.texi: Likewise.
36699 * doc/md.texi: Likewise.
36700 * doc/objc.texi: Likewise.
36702 2017-01-10 Joshua Conner <joshconner@google.com>
36704 * config/arm/fuchsia-elf.h: New file.
36705 * config/fuchsia.h: New file.
36706 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
36707 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
36709 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
36711 2016-01-10 Richard Biener <rguenther@suse.de>
36713 PR tree-optimization/79034
36714 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
36715 Propagate out degenerate PHIs in the joiner.
36717 2017-01-10 Martin Liska <mliska@suse.cz>
36719 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
36720 (sort_congruence_classes_by_decl_uid): Likewise.
36721 (sort_congruence_class_groups_by_decl_uid): Likewise.
36722 (sem_item_optimizer::merge_classes): Sort class, groups in these
36723 classes and members in the groups by DECL_UID of declarations.
36724 This would make merge operations stable.
36726 2017-01-10 Martin Liska <mliska@suse.cz>
36728 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
36729 usage of m_classes_vec.
36730 (sem_item_optimizer::~sem_item_optimizer): Likewise.
36731 (sem_item_optimizer::get_group_by_hash): Likewise.
36732 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
36733 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
36734 (sem_item_optimizer::verify_classes): Likewise.
36735 (sem_item_optimizer::process_cong_reduction): Likewise.
36736 (sem_item_optimizer::dump_cong_classes): Likewise.
36737 (sem_item_optimizer::merge_classes): Likewise.
36738 * ipa-icf.h (congruence_class_hash): Rename from
36739 congruence_class_group_hash. Remove declaration of m_classes_vec.
36741 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
36743 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
36744 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
36745 * config.gcc: Add avx512vpopcntdqintrin.h.
36746 * config/i386/avx512vpopcntdqintrin.h: New.
36747 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
36748 * config/i386/i386-builtin-types.def: Add new types.
36749 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
36750 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
36751 __builtin_ia32_vpopcountq_v8di_mask): New.
36752 * config/i386/i386-c.c (ix86_target_macros_internal): Define
36753 __AVX512VPOPCNTDQ__.
36754 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
36755 (PTA_AVX512VPOPCNTDQ): Define.
36756 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
36757 TARGET_AVX512VPOPCNTDQ_P): Define.
36758 * config/i386/i386.opt: Add mavx512vpopcntdq.
36759 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
36760 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
36762 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36764 PR middle-end/77484
36765 * predict.def (PRED_CALL): Set to 67.
36767 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
36769 * expr.c (store_field): In the bitfield case, if the value comes from
36770 a function call and is of an aggregate type returned in registers, do
36771 not modify the field mode; extract the value in all cases if the mode
36772 is BLKmode and the size is not larger than a word.
36774 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
36777 * config/i386/cpuid.h: Fix undefined behavior.
36779 2017-01-04 Jeff Law <law@redhat.com>
36781 PR tree-optimization/79007
36782 PR tree-optimization/67955
36783 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
36784 conservative for pt.null when flag_non_call_exceptions is on.
36786 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36788 PR translation/79019
36789 PR translation/79020
36790 * params.def (PARAM_INLINE_MIN_SPEEDUP,
36791 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
36792 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
36794 * config/avr/avr.opt (maccumulate-args): Likewise.
36795 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
36796 * common.opt (freport-bug): Likewise.
36797 * cif-code.def (CIF_FINAL_ERROR): Likewise.
36798 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
36799 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
36800 translatable string.
36801 * config/i386/i386.c (function_value_32): Likewise.
36802 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
36803 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
36805 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
36806 * common/config/msp430/msp430-common.c (msp430_handle_option):
36808 * symtab.c (symtab_node::verify_base): Likewise.
36809 * opts.c (set_debug_level): Likewise.
36810 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
36811 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
36812 missing whitespace to translatable strings.
36813 * config/avr/avr.md (bswapsi2): Fix typo in comment.
36814 * config/sh/superh.h: Likewise.
36815 * config/i386/xopintrin.h: Likewise.
36816 * config/i386/znver1.md: Likewise.
36817 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
36818 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
36819 * double-int.h (struct double_int): Likewise.
36820 * double-int.c (div_and_round_double): Likewise.
36821 * wide-int.cc: Likewise.
36822 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
36823 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
36824 * cfgcleanup.c (crossjumps_occured): Renamed to ...
36825 (crossjumps_occurred): ... this.
36826 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
36829 PR tree-optimization/78899
36830 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
36831 returning bool return struct loop *, NULL for failure and the new
36833 (versionable_outer_loop_p): Don't version outer loop if it has
36834 dont_vectorized bit set.
36835 (tree_if_conversion): When versioning outer loop, ensure
36836 tree_if_conversion is performed also on the inner loop of the
36837 non-vectorizable outer loop copy.
36838 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
36839 LOOP_VECTORIZED in inner loop of the scalar outer loop and
36840 prevent vectorization of it.
36841 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
36842 the outer loop vectorization of the non-scalar version is attempted
36843 before vectorization of the inner loop in scalar version. If
36844 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
36845 vectorization of its inner loop.
36846 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
36847 has 2 inner loops, rename also on edges from bb whose single pred
36848 is outer_loop->header. Fix typo in function comment.
36850 2017-01-09 Martin Sebor <msebor@redhat.com>
36853 * asan.c (asan_emit_stack_protection): Increase local buffer size
36854 to avoid snprintf truncation warning.
36856 2017-01-09 Andrew Pinski <apinski@cavium.com>
36858 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
36859 to reference thunderx2t99 for the tuning structure
36860 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
36862 (thunderx2t99_extra_costs): This.
36863 * config/aarch64/aarch64-tune.md: Regenerate.
36864 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
36865 (vulcan_addrcost_table): This.
36866 (vulcan_regmove_cost): Rename to ...
36867 (thunderx2t99_regmove_cost): This.
36868 (vulcan_vector_cost): Rename to ...
36869 (thunderx2t99_vector_cost): this.
36870 (vulcan_branch_cost): Rename to ...
36871 (thunderx2t99_branch_cost): This.
36872 (vulcan_tunings): Rename to ...
36873 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
36874 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
36876 2017-01-09 Martin Jambor <mjambor@suse.cz>
36880 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
36881 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
36882 (propagate_vr_accross_jump_function): Use the above function for all
36883 value range computations for pass-through jump functions and type
36884 converasion from explicit value range values.
36885 (ipcp_propagate_stage): Do not attempt to deduce types of formal
36886 parameters from TYPE_ARG_TYPES.
36887 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
36888 (ipa_write_node_info): Stream type of the actual argument.
36889 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
36891 2017-01-09 Martin Liska <mliska@suse.cz>
36894 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
36895 (lookup_compiler): Do not show error message with have_E.
36897 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36899 PR tree-optimization/78938
36900 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
36901 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
36902 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
36903 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
36906 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36908 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
36911 2017-01-09 Richard Biener <rguenther@suse.de>
36913 PR tree-optimization/78997
36914 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
36915 name condition properly.
36917 2017-01-09 Richard Biener <rguenther@suse.de>
36920 * dwarf2out.c (is_cxx): New overload with context.
36921 (is_naming_typedef_decl): Use it.
36923 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36925 * invoke.texi (Option Summary): Correct spacing in option lists
36926 and add line breaks to fix over-long lines.
36928 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36930 PR middle-end/17660
36932 * extend.texi (Common Variable Attributes): Add xref to GCC
36933 Internals manual to explain mode attribute keywords.
36935 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36938 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
36939 and Preprocessor Options.
36940 (Options for Linking): Document -pthread here....
36941 (RS/6000 and PowerPC Options): ...not here.
36942 (Solaris 2 Options): ...or here.
36943 * doc/cppopts.texi: Document -pthread.
36945 2017-01-08 Martin Sebor <msebor@redhat.com>
36947 PR middle-end/77708
36948 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
36949 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
36950 New member functions.
36951 (format_directive): Used them.
36953 (pass_sprintf_length::handle_gimple_call): Same.
36954 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
36955 to avoid truncation for any argument.
36956 (extract_affine_mul): Same.
36957 * tree.c (get_file_function_name): Same.
36959 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36961 PR middle-end/77484
36962 * predict.def (PRED_INDIR_CALL): Set to 86.
36964 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36966 PR preprocessor/54124
36967 * doc/cppopts.texi: Reformat -d subtable to list the full name
36968 of the options. Add cross-reference to the docs for the general
36969 compiler -d options.
36970 * doc/invoke.texi (Developer Options): Add cross-reference to the
36971 preprocessor-specific -d option documentation.
36973 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36975 PR preprocessor/13498
36976 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
36977 redudant material, and reflect new command-line options.
36978 (System Headers): Likewise.
36980 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36982 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
36983 -isystem, and -idirafter. Copy-edit.
36984 * doc/cppopts.texi: Copy-edit. Remove contradiction about
36985 default for -ftrack-macro-expansion. Delete obsolete and
36986 badly-formatted implementation details about -fdebug-cpp output.
36987 * doc/cppwarnopts.texi: Copy-edit.
36989 2017-01-07 David Malcolm <dmalcolm@redhat.com>
36992 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
36993 that the transition from a max line width >= 1<<10 to narrower
36994 lines works correctly.
36996 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
36998 * doc/options.texi (PerFunction): New.
36999 * opt-functions.awk (switch_flags): Map both Optimization and
37000 PerFunction to CL_OPTIMIZATION.
37001 * opth-gen.awk: Test for PerFunction flag along with
37003 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
37004 it only when the latter is present. Skip those that don't in
37005 the hash function generator.
37006 * common.opt (fvar-tracking): Mark as PerFunction instead of
37008 (fvar-tracking-assignments): Likewise.
37009 (fvar-tracking-assignments-toggle): Likewise.
37010 (fvar-tracking-uninit): Likewise.
37012 2017-01-07 Jakub Jelinek <jakub@redhat.com>
37014 PR translation/79018
37015 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
37018 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
37021 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
37022 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
37023 TARGET_LONG_JUMP_TABLE_OFFSETS.
37024 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
37025 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
37026 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
37027 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
37028 * config/m68k/m68k.md (tablejump expander): Likewise.
37029 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
37030 TARGET_LONG_JUMP_TABLE_OFFSETS.
37031 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
37032 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
37034 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
37035 David Holsgrove <david.holsgrove@xilinx.com>
37037 * common/config/microblaze/microblaze-common.c
37038 (TARGET_EXCEPT_UNWIND_INFO): Remove.
37039 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
37041 * config/microblaze/microblaze.c (microblaze_must_save_register)
37042 (microblaze_expand_epilogue, microblaze_return_addr): Handle
37044 (microblaze_eh_return): New function.
37045 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
37046 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
37047 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
37048 * config/microblaze/microblaze.md (eh_return): New pattern.
37050 2017-01-06 Jakub Jelinek <jakub@redhat.com>
37052 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
37053 GCC_DIAGNOSTIC_STRINGIFY): Define.
37055 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
37057 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37059 * config/arm/arm.md (<mcrr>): New.
37061 * config/arm/arm.c (arm_arch5te): New.
37062 (arm_option_override): Set arm_arch5te.
37063 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
37065 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
37066 (arm_mcrr_qualifiers): ... this. New.
37067 (MRRC_QUALIFIERS): Define to...
37068 (arm_mrrc_qualifiers): ... this. New.
37069 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
37071 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
37072 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
37073 (MRRCI, mrrc, MRRC): New.
37074 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
37075 VUNSPEC_MRRC2): New.
37077 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37079 * config/arm/arm.md (<mcr>): New.
37081 * config/arm/arm.c (arm_coproc_builtin_available): Add
37082 support for mcr, mrc, mcr2 and mrc2.
37083 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
37084 (arm_mcr_qualifiers): ... this. New.
37085 (MRC_QUALIFIERS): Define to ...
37086 (arm_mrc_qualifiers): ... this. New.
37087 (MCR_QUALIFIERS): Define to ...
37088 (arm_mcr_qualifiers): ... this. New.
37089 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
37091 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
37092 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
37093 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
37094 VUNSPEC_MRC2): New.
37096 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37098 * config/arm/arm.md (*ldc): New.
37102 * config/arm/arm.c (arm_coproc_builtin_available): Add
37103 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
37104 (arm_coproc_ldc_stc_legitimate_address): New.
37105 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
37106 'qualifier_const_pointer'.
37107 (LDC_QUALIFIERS): Define to...
37108 (arm_ldc_qualifiers): ... this. New.
37109 (STC_QUALIFIERS): Define to...
37110 (arm_stc_qualifiers): ... this. New.
37111 * config/arm/arm-protos.h
37112 (arm_coproc_ldc_stc_legitimate_address): New.
37113 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
37114 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
37115 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
37116 stc2, stcl, stc2l): New.
37117 * config/arm/constraints.md (Uz): New.
37118 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
37119 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
37120 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
37121 VUNSPEC_STC2L): New.
37123 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37125 * config/arm/arm.md (<cdp>): New.
37126 * config/arm/arm.c (neon_const_bounds): Rename this ...
37127 (arm_const_bounds): ... this.
37128 (arm_coproc_builtin_available): New.
37129 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
37130 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
37131 (CDP_QUALIFIERS): Define to...
37132 (arm_cdp_qualifiers): ... this. New.
37134 (arm_expand_builtin_args): Add case for 6 arguments.
37135 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
37136 (arm_const_bounds): ... this.
37137 (arm_coproc_builtin_available): New.
37138 * config/arm/arm_acle.h (__arm_cdp): New.
37140 * config/arm/arm_acle_builtins.def (cdp): New.
37142 * config/arm/iterators.md (CDPI,CDP,cdp): New.
37143 * config/arm/neon.md: Rename all 'neon_const_bounds' to
37144 'arm_const_bounds'.
37145 * config/arm/types.md (coproc): New.
37146 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
37147 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
37148 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
37149 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
37151 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37153 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
37154 (UBINOP_QUALIFIERS): New.
37156 (acle_builtin_data): New. Change comment.
37157 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
37158 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
37159 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
37160 arm_acle_builtins.def.
37161 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
37162 (arm_init_acle_builtins): New.
37163 (CRC32_BUILTIN): Remove.
37164 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
37165 crc32cb, crc32ch and crc32cw.
37166 (arm_init_crc32_builtins): Remove.
37167 (arm_init_builtins): Use arm_init_acle_builtins rather
37168 than arm_init_crc32_builtins.
37169 (arm_expand_acle_builtin): New.
37170 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
37171 * config/arm/arm_acle_builtins.def: New.
37173 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37175 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
37176 (arm_builtin_datum): ... this.
37177 (arm_init_neon_builtin): Rename to ...
37178 (arm_init_builtin): ... this. Add a new parameters PREFIX
37179 and USE_SIG_IN_NAME.
37180 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
37181 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
37182 'arm_builtin_datum'.
37183 (arm_init_vfp_builtins): Likewise.
37184 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
37185 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
37186 (arm_expand_neon_args): Rename to ...
37187 (arm_expand_builtin_args): ... this. Rename builtin_arg
37188 enum values and differentiate between ARG_BUILTIN_MEMORY
37189 and ARG_BUILTIN_NEON_MEMORY.
37190 (arm_expand_neon_builtin_1): Rename to ...
37191 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
37192 values, arm_expand_builtin_args and add bool parameter NEON.
37193 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
37194 (arm_expand_vfp_builtin): Likewise.
37195 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
37197 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37199 PR middle-end/77484
37200 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
37201 * predict.c (tree_estimate_probability_bb): Reverse direction of
37202 polymorphic call predictor.
37204 2017-01-06 David Malcolm <dmalcolm@redhat.com>
37206 * passes.c (execute_one_pass): Split out pass-skipping logic into...
37207 (determine_pass_name_match): ...this new function and...
37208 (should_skip_pass_p): ...this new function.
37210 2017-01-06 Nathan Sidwell <nathan@acm.org>
37212 * ipa-visibility.c (function_and_variable_visibility): Reformat
37213 comments and long lines. Remove extrneous if.
37214 * symtab.c (symtab_node::make_decl_local): Fix code format.
37215 (symtab_node::set_section_for_node): Fix comment typo.
37217 2017-01-06 Martin Liska <mliska@suse.cz>
37220 * lra-constraints.c: Rename invariant to lra_invariant.
37221 * predict.c (set_even_probabilities): Initialize e to NULL.
37223 2017-01-05 Martin Sebor <msebor@redhat.com>
37225 PR tree-optimization/78910
37226 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
37227 (format_integer): Correct off-by-one error in the handling
37228 of precision with negative numbers in signed conversions..
37230 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
37232 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
37234 2017-01-05 Jakub Jelinek <jakub@redhat.com>
37236 PR tree-optimization/71016
37237 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
37238 factor_out_conditional_conversion. Formatting fix.
37239 (factor_out_conditional_conversion): Add cond_stmt argument.
37240 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
37241 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
37244 2017-01-05 David Malcolm <dmalcolm@redhat.com>
37246 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
37247 read-rtl-function.o, and selftest-rtl.o.
37248 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
37249 (selftest::aarch64_test_loading_full_dump): New function.
37250 (selftest::aarch64_run_selftests): New function.
37251 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
37252 selftest::aarch64_run_selftests.
37253 * config/i386/i386.c
37254 (selftest::ix86_test_loading_dump_fragment_1): New function.
37255 (selftest::ix86_test_loading_call_insn): New function.
37256 (selftest::ix86_test_loading_full_dump): New function.
37257 (selftest::ix86_test_loading_unspec): New function.
37258 (selftest::ix86_run_selftests): Call the new functions.
37259 * emit-rtl.c (maybe_set_max_label_num): New function.
37260 * emit-rtl.h (maybe_set_max_label_num): New decl.
37261 * function.c (instantiate_decls): Guard call to
37262 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
37263 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
37265 * gensupport.c (gen_reader::gen_reader): Pass "false"
37266 for new "compact" param of rtx_reader.
37267 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
37268 rather than an empty string for NULL strings.
37269 * read-md.c: Potentially include config.h rather than bconfig.h.
37270 Wrap include of errors.h with #ifdef GENERATOR_FILE.
37271 (have_error): New global, copied from errors.c.
37272 (md_reader::read_name): Rename to...
37273 (md_reader::read_name_1): ...this, adding "out_loc" param,
37274 and converting "missing name or number" to returning false, rather
37276 (md_reader::read_name): Reimplement in terms of read_name_1.
37277 (md_reader::read_name_or_nil): New function.
37278 (md_reader::read_string): Handle "(nil)" by returning NULL.
37279 (md_reader::md_reader): Add new param "compact".
37280 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
37281 (md_reader::read_file): New method.
37282 * read-md.h (md_reader::md_reader): Add new param "compact".
37283 (md_reader::read_file): New method.
37284 (md_reader::is_compact): New accessor.
37285 (md_reader::read_name): Convert return type from void to file_location.
37286 (md_reader::read_name_or_nil): New decl.
37287 (md_reader::read_name_1): New decl.
37288 (md_reader::m_compact): New field.
37289 (noop_reader::noop_reader): Pass "false" for new "compact" param
37291 (rtx_reader::rtx_reader): Add new "compact" param.
37292 (rtx_reader::read_rtx_operand): Make virtual and convert return
37293 type from void to rtx.
37294 (rtx_reader::read_until): New decl.
37295 (rtx_reader::handle_any_trailing_information): New virtual function.
37296 (rtx_reader::postprocess): New virtual function.
37297 (rtx_reader::finalize_string): New virtual function.
37298 (rtx_reader::m_in_call_function_usage): New field.
37299 (rtx_reader::m_reuse_rtx_by_id): New field.
37300 * read-rtl-function.c: New file.
37301 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
37302 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
37303 (selftest::verify_three_block_rtl_cfg): New decl.
37304 * read-rtl-function.h: New file.
37305 * read-rtl.c: Potentially include config.h rather than bconfig.h.
37306 For host, include function.h, memmodel.h, and emit-rtl.h.
37307 (one_time_initialization): New function.
37308 (struct compact_insn_name): New struct.
37309 (compact_insn_names): New array.
37310 (find_code): Handle insn codes in compact dumps.
37311 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
37312 (bind_subst_iter_and_attr): Likewise.
37313 (add_condition_to_string): Likewise.
37314 (add_condition_to_rtx): Likewise.
37315 (apply_attribute_uses): Likewise.
37316 (add_current_iterators): Likewise.
37317 (apply_iterators): Likewise.
37318 (initialize_iterators): Guard usage of apply_subst_iterator with
37319 #ifdef GENERATOR_FILE.
37320 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
37321 (md_reader::read_mapping): Likewise.
37322 (add_define_attr_for_define_subst): Likewise.
37323 (add_define_subst_attr): Likewise.
37324 (read_subst_mapping): Likewise.
37325 (check_code_iterator): Likewise.
37326 (rtx_reader::read_rtx): Likewise. Move one-time initialization
37328 (one_time_initialization): New function.
37329 (rtx_reader::read_until): New method.
37330 (read_flags): New function.
37331 (parse_reg_note_name): New function.
37332 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
37333 Handle reuse_rtx ids.
37334 Wrap iterator lookup within #ifdef GENERATOR_FILE.
37335 Add parsing support for RTL dumps, mirroring the special-cases in
37336 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
37337 values, and calling handle_any_trailing_information.
37338 (rtx_reader::read_rtx_operand): Convert return type from void
37339 to rtx, returning return_rtx. Handle case 'e'. Call
37340 finalize_string on XSTR and XTMPL fields.
37341 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
37342 "(nil)" values were omitted. Call the postprocess vfunc on the
37344 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
37345 class ctor. Initialize m_in_call_function_usage. Call
37346 one_time_initialization.
37347 * rtl-tests.c (selftest::test_uncond_jump): Call
37348 set_new_first_and_last_insn.
37349 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
37350 * selftest-rtl.c: New file.
37351 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
37352 (selftest::get_insn_by_uid): New decl.
37353 * selftest-run-tests.c (selftest::run_tests): Call
37354 read_rtl_function_c_tests.
37355 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
37356 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
37359 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
37361 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
37362 operands in a special way. Assert that pos+len <= mode precision.
37364 2017-01-05 Jakub Jelinek <jakub@redhat.com>
37366 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
37367 3 argument Alias with unlimited for the negative form.
37368 (fno-vect-cost-model): Removed.
37370 2017-01-05 Martin Liska <mliska@suse.cz>
37372 * hsa-gen.c (gen_hsa_divmod): New function.
37373 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
37375 2017-01-05 Martin Liska <mliska@suse.cz>
37378 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
37381 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37383 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
37384 small constant length operands.
37386 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37388 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
37389 between loop iterations.
37391 2017-01-05 Martin Liska <mliska@suse.cz>
37394 * gimplify.c (gimplify_decl_expr): Compare to
37395 asan_poisoned_variables instread of checking flags.
37396 (gimplify_target_expr): Likewise.
37397 (gimplify_expr): Likewise.
37398 (gimplify_function_tree): Conditionally initialize
37399 asan_poisoned_variables.
37401 2017-01-04 Jeff Law <law@redhat.com>
37403 PR tree-optimizatin/78812
37404 * rtl.h (contains_mem_rtx_p): Prototype.
37405 * ifcvt.c (containts_mem_rtx_p): Move from here to...
37406 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
37407 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
37408 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
37409 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
37411 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37413 * input.c (assert_char_at_range): Default-initialize actual_range.
37415 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37417 * df-scan.c (df_ref_create_structure): Make regno unsigned,
37418 to match the caller.
37420 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37422 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
37423 insns after final jump in test to emit dummy move.
37425 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37427 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
37428 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
37430 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37432 * multiple_target.c (create_dispatcher_calls): Init e_next.
37433 * tree-ssa-loop-split.c (split_loop): Init border.
37434 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
37437 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
37442 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
37443 (altivec_register_operand): Do not return true if the operand
37444 contains a SUBREG mixing SImode and SFmode.
37445 (vsx_register_operand): Likewise.
37446 (vsx_reg_sfsubreg_ok): New predicate.
37447 (vfloat_operand): Do not return true if the operand contains a
37448 SUBREG mixing SImode and SFmode.
37449 (vint_operand): Likewise.
37450 (vlogical_operand): Likewise.
37451 (gpc_reg_operand): Likewise.
37452 (int_reg_operand): Likewise.
37453 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
37454 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
37455 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
37457 (rs6000_emit_move_si_sf_subreg): New helper function.
37458 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
37459 fixup SUBREGs involving SImode and SFmode.
37460 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
37461 numbers for the new peephole2 optimization.
37462 (peephole2 for SFmode unions): New peephole2 to optimize cases in
37463 the GLIBC math library that do AND/IOR/XOR operations on single
37464 precision floating point.
37465 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
37466 target macros to say whether we need to avoid SUBREGs mixing
37468 (TARGET_ALLOW_SF_SUBREG): Likewise.
37469 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
37470 (UNSPEC_SI_FROM_SF): Likewise.
37471 (iorxor): Change spacing.
37472 (and_ior_xor): New iterator for AND, IOR, and XOR.
37473 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
37474 (movdi_from_sf_zero_ext): Likewise.
37475 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
37476 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
37477 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
37478 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
37479 (fms<mode>4): Likewise.
37480 (fnma<mode>4): Likewise.
37481 (fnms<mode>4): Likewise.
37482 (nfma<mode>4): Likewise.
37483 (nfms<mode>4): Likewise.
37485 2017-01-04 Marek Polacek <polacek@redhat.com>
37488 * doc/invoke.texi: Document -Wpointer-compare.
37490 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37492 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
37495 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
37496 descriptions for -gdwarf-5 and emit them as uleb128 instead of
37499 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
37502 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
37503 documentation of the powerpc_popcntb_ok attribute.
37504 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
37505 code to issue warning messages if a requested CPU configuration is
37506 not supported by the binary (assembler and loader) toolchain.
37507 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
37508 made to define a built-in function that has been disabled.
37509 (paired_init_builtins): Add assertion to prevent ICE if attempt is
37510 made to define a built-in function that has been disabled.
37511 (altivec_init_builtins): Add comment explaining why definition
37512 of the DST built-in functions is not preceded by an assertion
37513 check. Add assertions to prevent ICE if attempts are made to
37514 define an altivec predicate or an abs* built-in function that has
37516 (htm_init_builtins): Add comment explaining why definition of the
37517 htm built-in functions is not preceded by an assertion check.
37519 2017-01-04 Jeff Law <law@redhat.com>
37521 PR tree-optimizatin/67955
37522 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
37523 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
37524 the points-to solution does not include pt_null. Use DECL_PT_UID
37527 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
37529 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
37530 Use gen_int_mode instead of gen_lopwart for const_int operands.
37532 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37534 PR tree-optimization/71563
37535 * match.pd: Simplify X << Y into X if Y is known to be 0 or
37536 out of range value - has low bits known to be zero.
37538 2017-01-04 Alan Modra <amodra@gmail.com>
37540 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
37541 * configure: Regenerate.
37542 * config.in: Regenerate.
37544 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37547 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
37548 a substring of the message, but strcmp with the whole message. Ifdef
37549 ENABLE_NLS, translate the message first using dgettext.
37551 2017-01-03 Jeff Law <law@redhat.com>
37553 PR tree-optimizatin/78856
37554 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
37555 (mark_threaded_blocks): Remove code to truncate thread paths that
37556 cross multiple loop headers. Instead invalidate the cached loop
37557 iteration information and handle case of a thread path walking
37558 into an irreducible region.
37560 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
37563 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
37564 assertions. Add support for doing the signbit if the IEEE 128-bit
37565 floating point value is in a GPR.
37566 * config/rs6000/rs6000.md (Fsignbit): Delete.
37567 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
37568 Update the length attribute if the value is in a GPR.
37569 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
37570 the sign or zero extension instruction, since the value is always 0/1.
37571 (signbit<mode>2_dm2): Delete using <Fsignbit>.
37574 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
37575 extracting SImode to a GPR register so that we can generate a
37576 store, limit the vector to be in a traditional Altivec register
37577 for the vextuwrx instruction.
37579 2017-01-03 Ian Lance Taylor <iant@google.com>
37581 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
37583 2017-01-03 Martin Sebor <msebor@redhat.com>
37585 PR tree-optimization/78696
37586 * gimple-ssa-sprintf.c (format_floating): Correct handling of
37587 precision. Use MPFR for %f for greater fidelity. Correct handling
37589 (pass_sprintf_length::compute_format_length): Set width and precision
37590 specified by asrerisk to void_node for vararg functions.
37591 (try_substitute_return_value): Adjust dump output.
37593 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
37595 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
37597 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
37599 * doc/invoke.texi (SPARC options): Document -mlra as the default.
37600 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
37601 -mlra/-mno-lra was passed to the compiler.
37603 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
37605 PR rtl-optimization/65618
37606 * emit-rtl.c (try_split): Move initialization of "before" and
37607 "after" to just before the call to emit_insn_after_setloc.
37609 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
37611 * doc/md.texi (Standard Names): Remove reference to Java frontend.
37613 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
37615 * dwarf2out.c (gen_enumeration_type_die): When
37616 -gno-strict-dwarf, add a DW_AT_encoding attribute.
37618 2017-01-03 Jakub Jelinek <jakub@redhat.com>
37620 PR tree-optimization/78965
37621 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
37622 Change first argument from const call_info & to call_info &. For %n
37623 set info.nowrite to false.
37625 PR middle-end/78901
37626 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
37627 possibly throwing calls.
37629 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
37630 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
37631 and fns handling, rather than in a separate case SSA_NAME.
37633 2017-01-02 Jeff Law <law@redhat.com>
37635 * config/darwin-driver.c (darwin_driver_init): Const-correctness
37636 fixes for first_period and second_period variables.
37638 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
37641 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
37642 (*insvqi_1): New insn pattern.
37643 (*insvqi_1_mem_rex64): Ditto.
37644 (*insvqi_2): Ditto.
37645 (*insvqi_3): Rename from *insvqi.
37647 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
37649 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
37651 * doc/cfg.texi (Edges): Remove reference to Java.
37652 (Maintaining the CFG): Ditto.
37654 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37656 PR middle-end/77674
37657 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
37658 transparent aliases.
37660 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37662 PR middle-end/77484
37663 * predict.def (PRED_CALL): Update hitrate.
37664 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
37665 * predict.c (tree_estimate_probability_bb): Split CALL predictor
37666 into direct/indirect/polymorphic variants.
37668 2017-01-01 Jakub Jelinek <jakub@redhat.com>
37670 Update copyright years.
37672 * gcc.c (process_command): Update copyright notice dates.
37673 * gcov-dump.c (print_version): Ditto.
37674 * gcov.c (print_version): Ditto.
37675 * gcov-tool.c (print_version): Ditto.
37676 * gengtype.c (create_file): Ditto.
37677 * doc/cpp.texi: Bump @copying's copyright year.
37678 * doc/cppinternals.texi: Ditto.
37679 * doc/gcc.texi: Ditto.
37680 * doc/gccint.texi: Ditto.
37681 * doc/gcov.texi: Ditto.
37682 * doc/install.texi: Ditto.
37683 * doc/invoke.texi: Ditto.
37685 Copyright (C) 2017 Free Software Foundation, Inc.
37687 Copying and distribution of this file, with or without modification,
37688 are permitted in any medium without royalty provided the copyright
37689 notice and this notice are preserved.