1 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4 instead of MEM_P in a couple more places. Fix formatting issues.
6 2017-12-04 Jim Wilson <jimw@sifive.com>
8 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
9 instead of GP_REG_LAST-1.
10 (riscv_adjust_libcall_cfi_prologue): Likewise.
11 (riscv_adjust_libcall_cri_epilogue): Likewise.
12 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
15 2017-12-04 Luis Machado <luis.machado@linaro.org>
17 * ipa-pure-const.c (check_decl): Add missing newline.
18 (state_from_flags): Likewise.
20 2017-12-04 Jeff Law <law@redhat.com>
22 PR tree-optimizatin/78496
23 * gimple-ssa-evrp-analyze.h
24 (evrp_range_analyzer::get_vr_values): Simplify.
25 * gimple-ssa-evrp-analyze.c: Corresponding changes.
26 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
27 and gimple-ssa-evrp-analyze.h.
28 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
29 (simplify_stmt_for_jump_threading): Copy a blob of code from
30 tree-vrp.c to use ranges to simplify statements.
31 (dom_opt_dom_walker::before_dom_children): Call
32 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
33 (dom_opt_dom_walker::after_dom_children): Similarly for
34 evrp_range_analyzer::leave.
35 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
38 * gimple-ssa-evrp-analyze.c
39 (evrp_range_analyzer::extract_range_from_stmt): Always use
40 vr_values::update_value_range so preexisting range info is
41 medged with new range info, even if the new range is VR_VARYING.
43 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
45 * combine.c: Adjust comment.
46 (use_crosses_set_p): Delete.
47 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
50 2017-12-04 Richard Biener <rguenther@suse.de>
52 PR tree-optimization/83255
53 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
54 Re-add zero-iteration check.
56 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
58 PR rtl-optimization/83245
59 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
60 hard registers as earlyclobber, also if not in an asm.
62 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
66 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
69 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
70 rs6000_cpu based on cpu_index, not tune_index.
72 2017-12-04 Richard Biener <rguenther@suse.de>
74 PR tree-optimization/83238
75 * graphite-scop-detection.c (scop_detection::merge_sese): Make
76 code match comment, rejecting invalid SESE regions.
78 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
80 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
81 require base operand is a REG_POINTER prior to reload on targets
82 with non-equivalent space registers.
84 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
86 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
87 (update_specialized_profile): Fix updating of counts.
88 (perhaps_add_new_callers): Likewise.
90 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
93 * x86-tune.def: Remove obsolette FIXMEs.
94 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
95 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
96 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
98 (X86_TUNE_PAD_RETURNS): Disable for generic.
100 2017-12-02 Jakub Jelinek <jakub@redhat.com>
102 PR tree-optimization/83170
103 PR tree-optimization/83241
104 * gimple-ssa-store-merging.c
105 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
106 gimple_vuse (ins_stmt) in case it has changed.
107 (imm_store_chain_info::output_merged_store): Likewise.
109 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
113 * tree-cfg.c (pass_warn_function_return::execute): Handle
114 __builtin_ubsan_handle_missing_return like __builtin_unreachable
115 with BUILTINS_LOCATION.
119 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
120 is BLKmode for vector field with vector raw mode, use TYPE_MODE
121 instead of DECL_MODE.
123 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
124 last argument to rtx pointer.
125 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
126 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
127 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
128 depending on the chosen ISAs.
129 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
130 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
131 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
133 * config/i386/sse.md (mov<mode>_internal): Likewise.
134 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
136 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
138 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
140 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
142 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
143 parameters from prototype.
144 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
145 parameters. Don't print a comment.
146 (emit_fusion_gpr_load): Adjust.
147 (emit_fusion_load_store): Adjust.
148 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
149 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
150 comment on the second line.
152 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
155 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
156 rs6000_cpu based on cpu_index, not tune_index.
158 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
160 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
161 which_alternative instead of which_alternative + 1.
162 (output_asm_insn): Print an extra tab if the template is short.
164 2017-12-01 Jim Wilson <jimw@sifive.com>
166 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
168 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
169 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
171 * doc/tm.texi: Regenerate.
173 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
176 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
177 whether we can allocate pseudos before trying to fix an address.
178 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
179 memory address is indexed or indirect.
180 (floatuns_<mode>si2_hw2): Likewise.
182 2017-12-01 Jason Merrill <jason@redhat.com>
184 * Makefile.in (TAGS): Add c-family/*.cc.
186 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
188 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
189 (call_value_insn): Likewise.
190 (sibcall_insn): Likewise.
191 (sibcall_value_insn): Likewise.
192 (movsi_aarch64): Likewise.
193 (movdi_aarch64): Likewise.
194 (add_losym_): Likewise.
195 (ldr_got_small_): Likewise.
196 (ldr_got_small_sidi): Likewise.
197 (ldr_got_small_28k_): Likewise.
198 (ldr_got_small_28k_sidi): Likewise.
199 * config/aarch64/aarch64.c (aarch64_print_address_internal):
200 Move output_addr_const to symbolic case. Add error check.
202 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
204 * config/s390/predicates.md (plus16_Q_operand): New predicate.
205 * config/s390/s390.md: Disable MVC merging peephole if it would
206 disable operand forwarding.
207 (new peephole2): Split MVCs if it would turn them into up to 2
210 2017-12-01 Richard Biener <rguenther@suse.de>
212 PR tree-optimization/83232
213 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
214 detection of same access. Instead of breaking the group here
215 do not consider the duplicate. Add comment explaining real fix.
217 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
219 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
221 2017-12-01 Sudakshina Das <sudi.das@arm.com>
223 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
225 2017-12-01 Jakub Jelinek <jakub@redhat.com>
227 * function.h (struct function): Remove cilk_frame_decl,
228 is_cilk_function and calls_cilk_spawn fields.
229 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
231 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
232 cilk_elemental field.
233 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
234 * target.def: Adjust comment.
235 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
236 Don't test cilk_elemental.
238 PR tree-optimization/83233
239 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
240 bswap_stat name for the struct.
243 * tree.h (SWITCH_BREAK_LABEL_P): Define.
244 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
245 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
246 SWITCH_BREAK_LABEL_P set on the label.
247 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
248 added for default case if it was missing and not all cases covered.
249 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
250 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
252 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
253 to avoid -Wimplicit-fallthrough warning.
254 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
255 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
257 PR tree-optimization/83221
258 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
260 (init_reassoc): Formatting fix.
263 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
264 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
268 * tree-cfg.c: Include asan.h.
269 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
271 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
273 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
275 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
277 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
278 vpcmpeqd instruction.
280 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
282 * config/i386/i386.c (standard_sse_constant_opcode): Fix
283 registers type for 128bit mode.
285 2017-11-30 Jakub Jelinek <jakub@redhat.com>
287 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
289 * gengtype-state.c (read_state_pair): Likewise.
290 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
291 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
292 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
293 * ifcvt.c (noce_try_store_flag_constants): Likewise.
294 * tree-ssa-ccp.c (ccp_finalize): Likewise.
295 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
296 * builtins.c (fold_builtin_3): Likewise.
297 * graphite-scop-detection.c
298 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
299 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
302 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
305 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
306 parameter. Return true if ignore_decl_rtl_set_p is true and other
307 conditions are satisfied.
308 * asan.h (asan_protect_global): Add new parameter.
309 * varasm.c (categorize_decl_for_section): Pass true as second parameter
310 to asan_protect_global calls.
312 2017-11-30 Jim Wilson <jimw@sifive.com>
314 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
315 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
316 -mstrict-align, add info on default value. Delete redundant lines for
317 -mabi. Add missing -mexplicit-relocs docs.
319 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
321 * config/arc/arc.md (trap): New pattern.
323 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
325 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
326 instruction to end into a delay slot.
327 * config/arc/arc.md (cond_delay_insn): Check if the instruction
328 can be placed into a delay slot against reg_note.
330 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
332 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
333 labels number of usages.
335 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
337 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
339 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
341 2017-11-30 Jakub Jelinek <jakub@redhat.com>
344 * internal-fn.c (expand_mul_overflow): Optimize unsigned
345 multiplication by power of 2 constant into two shifts + comparison.
347 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
350 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
352 2017-11-30 Richard Biener <rguenther@suse.de>
354 PR tree-optimization/83202
355 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
356 allow_peel argument and guard peeling.
357 (canonicalize_loop_induction_variables): Likewise.
358 (canonicalize_induction_variables): Pass false.
359 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
360 peeling from cunrolli.
362 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
364 * combine.c (try_combine): Print a message to dump file whenever
365 I0, I1, or I2 cannot be combined into I3.
367 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
369 PR rtl-optimization/83156
370 PR rtl-optimization/82621
371 * combine.c (try_combine): Don't split an I2 if one of the dests is
372 set again before I3. Allow unused dests.
374 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
376 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
378 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
380 PR rtl-optimization/80818
381 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
382 recursively. Use insn code for clobber.
383 (lra_set_insn_recog_data): Pass the new arg to
384 collect_non_operand_hard_regs.
385 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
387 (lra_update_insn_regno_info): Pass insn to
388 add_regs_to_insn_regno_info.
390 2017-11-29 Jim Wilson <jimw@sifive.com>
391 Andrew Waterman <andrew@sifive.com>
393 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
394 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
395 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
396 use SINGLE_SHIFT_COST.
397 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
398 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
400 2017-11-29 Julia Koval <julia.koval@intel.com>
402 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
403 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
404 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
405 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
406 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
407 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
408 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
409 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
410 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
411 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
412 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
413 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
414 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
415 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
416 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
417 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
418 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
419 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
420 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
421 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
422 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
423 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
424 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
425 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
426 __builtin_ia32_vpshldv_v2di_mask,
427 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
428 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
429 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
431 2017-11-29 Julia Koval <julia.koval@intel.com>
433 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
434 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
435 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
436 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
437 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
438 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
439 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
440 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
441 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
442 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
443 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
444 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
445 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
446 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
447 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
448 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
449 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
450 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
451 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
452 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
453 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
455 2017-11-29 Daniel Cederman <cederman@gaisler.com>
457 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
458 movsi_pic_gotdata_op instruction as a load for the UT699 errata
461 2017-11-29 Martin Aberg <maberg@gaisler.com>
463 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
464 to prevent b2bst errata sequence.
465 (sqrtdf2_fix): Likewise.
467 2017-11-29 Daniel Cederman <cederman@gaisler.com>
469 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
470 (div_sqrt_insn_p): New function.
471 (sparc_do_work_around_errata): Insert NOP instructions to
472 prevent sequences that could trigger the TN-0013 errata for
473 certain LEON3 processors.
474 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
475 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
476 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
477 (in_branch_delay): Prevent div and sqrt in delay slot if
479 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
481 2017-11-29 Daniel Cederman <cederman@gaisler.com>
483 * config/sparc/sparc.c (atomic_insn_p): New function.
484 (sparc_do_work_around_errata): Insert NOP instructions to
485 prevent sequences that could trigger the TN-0010 errata for
487 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
488 instruction referable in atomic_insns_p.
490 2017-11-29 Daniel Cederman <cederman@gaisler.com>
492 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
493 (atomic_compare_and_swap_leon3_1): Likewise.
496 2017-11-29 Daniel Cederman <cederman@gaisler.com>
498 * config/sparc/sparc.c (fpop_insn_p): New function.
499 (sparc_do_work_around_errata): Insert NOP instructions to
500 prevent sequences that could trigger the TN-0012 errata for
502 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
503 * config/sparc/sparc.md (fix_gr712rc): New attribute.
504 (in_branch_annul_delay): Prevent floating-point instructions
505 in delay slot of annulled integer branch.
507 2017-11-29 Richard Biener <rguenther@suse.de>
509 PR tree-optimization/83202
510 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
512 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
513 nodes vectorized to the same stmts multiple times.
514 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
515 (vect_analyze_slp_instance): Adjust.
516 (scalar_stmts_to_slp_tree_map_t): New typedef.
517 (vect_schedule_slp_instance): Add a map recording the SLP node
518 representing the vectorized stmts for a set of scalar stmts.
519 Avoid code-generating redundancies.
520 (vect_schedule_slp): Allocate map and pass it down.
522 2017-11-29 Nathan Sidwell <nathan@acm.org>
525 * tree.c (build_complex_type): Fix canonicalization. Only fill in
528 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
530 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
531 cases for printing LDP/STP memory addresses.
532 (aarch64_print_address_internal): Renamed from
533 aarch64_print_operand_address, added parameter, add Pmode check.
534 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
535 (aarch64_print_operand_address): Indirect to
536 aarch64_print_address_internal.
537 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
538 'y' operand output specifier.
540 2017-11-29 Jakub Jelinek <jakub@redhat.com>
543 * tree.c (build_simple_mem_ref_loc): Handle
544 get_addr_base_and_unit_offset returning a MEM_REF.
547 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
550 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
553 2017-11-28 David Malcolm <dmalcolm@redhat.com>
555 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
557 (layout::move_to_column): Likewise.
559 2017-11-28 Jakub Jelinek <jakub@redhat.com>
562 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
563 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
565 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
566 Martin Jambor <mjambor@suse.cz>
569 * tree.h (expr_type_first_operand_type_p): Declare
570 * tree.c (expr_type_first_operand_type_p): New function.
571 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
572 (ipa_value_from_jfunc): Adjust declaration.
573 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
574 Use it as result type for arithmetics, unless it is NULL in which case
575 be more conservative.
576 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
577 ipa_get_jf_pass_through_result.
578 (propagate_vals_across_pass_through): Likewise.
579 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
580 is to propagate_vals_across_pass_through.
581 (propagate_constants_across_call): Pass PARM_TYPE to
582 propagate_scalar_across_jump_function.
583 (find_more_scalar_values_for_callers_subset): Pass parameter type to
584 ipa_value_from_jfunc.
585 (cgraph_edge_brings_all_scalars_for_node): Likewise.
586 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
587 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
588 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
589 target_type, pass it to ipa_value_from_jfunc.
590 (update_indirect_edges_after_inlining): Pass parameter type to
591 try_make_edge_direct_simple_call.
593 2017-11-28 Jeff Law <law@redhat.com>
595 * gimple-ssa-evrp-analyze.c
596 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
597 refine ranges if scev_initialized_p returns true.
598 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
600 2017-11-28 Julia Koval <julia.koval@intel.com>
602 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
603 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
604 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
605 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
606 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
607 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
608 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
609 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
610 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
611 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
612 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
613 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
614 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
615 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
616 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
617 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
618 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
619 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
620 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
621 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
622 __builtin_ia32_vpshrd_v2di_mask): New builtins.
623 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
625 2017-11-28 Julia Koval <julia.koval@intel.com>
627 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
628 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
629 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
630 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
631 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
632 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
633 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
634 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
635 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
636 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
637 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
638 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
639 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
640 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
641 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
642 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
643 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
644 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
645 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
646 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
647 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
648 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
649 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
650 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
651 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
652 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
653 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
654 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
655 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
656 __builtin_ia32_vpshld_v2di_mask): New builtins.
657 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
658 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
660 2017-11-28 Richard Biener <rguenther@suse.de>
662 PR tree-optimization/80776
663 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
665 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
667 (evrp_range_analyzer::record_ranges_from_incoming_edges):
668 If the incoming edge is an effective fallthru because the other
669 edge only reaches a __builtin_unreachable () then record ranges
670 derived from the controlling condition in SSA info.
671 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
672 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
674 2017-11-28 Olivier Hainque <hainque@adacore.com>
676 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
677 on mingw build hosts.
679 2017-11-28 Jakub Jelinek <jakub@redhat.com>
682 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
684 2017-11-28 Richard Biener <rguenther@suse.de>
687 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
688 copies generated from memcpy use a character array as reference
691 2017-11-28 Julia Koval <julia.koval@intel.com>
692 Sebastian Peryt <sebastian.peryt@intel.com>
694 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
695 c-family/c-cilkplus.o, c-family/array-notation-common.o,
696 cilk-common.o, cilk.h, cilk-common.c): Remove.
698 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
699 * builtins.c (is_builtin_name): Remove cilkplus condition.
700 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
701 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
702 cilk-builtins.def, cilkplus.def): Remove.
703 * cif-code.def (CILK_SPAWN): Remove.
704 * cilk-builtins.def: Delete.
705 * cilk-common.c: Ditto.
707 * cilkplus.def: Ditto.
708 * config/darwin.h (fcilkplus): Delete.
709 * cppbuiltin.c: Ditto.
710 * doc/extend.texi: Remove cilkplus doc.
711 * doc/generic.texi: Ditto.
712 * doc/invoke.texi: Ditto.
713 * doc/passes.texi: Ditto.
714 * gcc.c (fcilkplus): Remove.
715 * gengtype.c (cilk.h): Remove.
716 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
718 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
720 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
722 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
723 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
725 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
726 inline_read_section): Ditto.
727 * ipa-inline-analysis.c (cilk.h): Remove.
728 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
729 * lto-wrapper.c (merge_and_complain, append_compiler_options,
730 append_linker_options): Remove condition for fcilkplus.
731 * lto/lto-lang.c (cilk.h): Remove.
732 (lto_init): Remove condition for fcilkplus.
733 * omp-expand.c (expand_cilk_for_call): Delete.
734 (expand_omp_taskreg, expand_omp_for_static_chunk,
735 expand_omp_for): Remove cilkplus
737 (expand_cilk_for): Delete.
738 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
739 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
740 execute_lower_omp, diagnose_sb_0): Ditto.
741 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
742 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
743 * tree-nested.c: Ditto.
744 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
745 (dump_generic_node): Ditto.
746 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
747 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
749 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
751 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
753 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
754 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
755 for vector mode and !TARGET_SIMD.
757 2017-11-28 Jakub Jelinek <jakub@redhat.com>
759 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
761 * tree.h (SWITCH_LABELS): Remove.
762 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
763 assert SWITCH_BODY is non-NULL.
764 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
766 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
768 PR tree-optimization/80788
769 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
770 has TREE_OVERFLOW set, call drop_tree_overflow.
772 2017-11-28 Richard Biener <rguenther@suse.de>
774 PR tree-optimization/83158
775 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more
778 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
781 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
782 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
784 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
786 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
787 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
789 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
792 * varasm.c (compare_constant): Take the mode of the constants into
793 account when comparing floating point constants.
795 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
797 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
798 from explicit instantiation of debug_helper.
799 * vec.h (DEFINE_DEBUG_VEC): Ditto.
801 2017-11-27 Richard Biener <rguenther@suse.de>
803 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
806 2017-11-27 Richard Biener <rguenther@suse.de>
808 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
809 (type_hash_canon): Also clear int_cst_hash_table entry for
811 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
813 2017-11-27 Tamar Christina <tamar.christina@arm.com>
815 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
816 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
817 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
819 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
821 * hash-map.h (gt_cleare_cache): Avoid UB.
823 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
825 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
826 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
827 (decide_unroll_runtime_iterations): Likewise.
828 (decide_unroll_stupid): Likewise.
830 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
833 * config/i386/i386.md: Add a loop with incssp.
835 2017-11-27 Martin Jambor <mjambor@suse.cz>
837 PR tree-optimization/81248
838 * tree-sra.c (splice_param_accesses): Remove size check.
839 (decide_one_param_reduction): Fix size check.
840 * gimple-pretty-print.c (dump_profile): Silence warning.
841 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
843 2017-11-27 Jakub Jelinek <jakub@redhat.com>
846 * dbxout.c (lastlineno): New variable.
847 (dbx_debug_hooks): Use dbxout_switch_text_section as
848 switch_text_section debug hook.
849 (dbxout_function_end): Switch to current_function_section
850 rather than function_section. If crtl->has_bb_partition,
851 output just one N_FUN, depending on in_cold_section_p.
852 (dbxout_source_line): Remember last lineno in lastlineno.
853 (dbxout_switch_text_section): New function.
854 (dbxout_function_decl): Adjust dbxout_block caller.
855 (dbx_block_with_cold_children): New function.
856 (dbxout_block): Return true if any LBRAC/RBRAC have been
857 emitted. Use dbx_block_with_cold_children at depth == 0
858 in second partition. Add PARENT_BLOCKNUM argument, pass
859 it optionally adjusted to children. Output LBRAC/RBRAC
860 around recursive call only if the block is in the current
861 partition, if not and anything was output, emit empty
863 * final.c (final_scan_insn): Compute cold_function_name
864 before calling switch_text_section debug hook. Call
865 that hook even if dwarf2out_do_frame if not emitting
869 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
872 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
874 PR rtl-optimization/82488
875 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
877 2017-11-26 Julia Koval <julia.koval@intel.com>
879 * config/i386/i386.c (processor_target_table): Add skylake_cost for
881 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
884 2017-11-26 Julia Koval <julia.koval@intel.com>
886 * config/i386/driver-i386.c (host_detect_local_cpu):
887 Detect skylake-avx512.
889 2017-11-26 Julia Koval <julia.koval@intel.com>
891 * config.gcc: Add -march=cannonlake.
892 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
893 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
894 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
895 (PTA_CANNONLAKE): New.
896 (processor_target_table): Add cannonlake.
897 (ix86_option_override_internal): Ditto.
898 (fold_builtin_cpu): Ditto.
899 (get_builtin_code_for_version): Handle cannonlake.
900 (M_INTEL_COREI7_CANNONLAKE): New.
901 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
902 * doc/invoke.texi: Add -march=cannonlake.
904 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
906 * plugin.c (add_new_plugin): Use platform-specific library extensions.
907 (try_init_one_plugin): Alternative implementation for MinGW.
908 * Makefile.in (plugin_implib): New.
909 (gengtype-lex.c): Fix broken AIX workaround.
910 * configure: Regenerate.
911 * doc/plugins.texi: Document support for MinGW.
913 2017-11-25 Jakub Jelinek <jakub@redhat.com>
915 PR rtl-optimization/81553
916 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
917 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
918 is a shift where C1 has different mode than the whole shift, use C1's
919 mode for MULT rather than the shift's mode.
922 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
923 builtins not enabled in the currently selected ISA.
925 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
927 PR tree-optimization/71026
928 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
929 (insert_reciprocals): Change to insert reciprocals before a division
930 by a square and to insert the square of a reciprocal.
931 (execute_cse_reciprocals_1): Change to consider division by a square.
932 (register_division_in): Add importance parameter.
934 2017-11-24 Richard Biener <rguenther@suse.de>
936 PR tree-optimization/82402
937 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
938 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
940 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
942 * match.pd (0-ptr): New transformation.
944 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
947 * ipa-inline.c (inline_small_functions): Set current badnes correctly
948 when skipping checking.
950 2017-11-24 Richard Biener <rguenther@suse.de>
952 PR tree-optimization/83128
953 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
954 (vn_reference_lookup_3): Likewise.
956 2017-11-24 Jakub Jelinek <jakub@redhat.com>
959 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
960 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
963 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
964 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
965 check for the latter with gimple_call_builtin_p. Do not handle
966 BUILT_IN_STPNCPY_CHK which is not a pass through call.
968 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
970 * config/arm/arm_neon.h: Fix pragma GCC push_options before
973 2017-11-23 Julia Koval <julia.koval@intel.com>
975 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
976 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
977 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
978 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
979 _mm512_maskz_expandloadu_epi16): New intrinsics.
980 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
981 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
982 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
983 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
984 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
985 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
986 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
987 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
988 _mm256_maskz_expandloadu_epi8): New intrinsics.
989 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
990 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
991 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
992 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
993 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
994 * config/i386/sse.md (VI248_VLBW): New iterator.
995 (expand<mode>_mask, expand<mode>_maskz): New patterns.
997 2017-11-23 Julia Koval <julia.koval@intel.com>
999 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
1000 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
1001 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
1002 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
1003 _mm512_mask_compressstoreu_epi16): New.
1004 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
1005 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
1006 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
1007 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
1008 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
1009 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
1010 _mm256_mask_compressstoreu_epi8): New.
1011 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
1012 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
1013 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
1014 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
1015 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
1016 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
1017 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
1018 __builtin_ia32_compresshi128_mask,
1019 __builtin_ia32_compressstoreuqi512_mask,
1020 __builtin_ia32_compressstoreuhi512_mask,
1021 __builtin_ia32_compressstoreuqi256_mask,
1022 __builtin_ia32_compressstoreuqi128_mask,
1023 __builtin_ia32_compressstoreuhi256_mask,
1024 __builtin_ia32_compressstoreuhi128_mask): New builtins.
1025 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
1027 (ix86_expand_special_args_builtin): Handle new types.
1028 (s4fma_expand): Handle new builtin array.
1029 * config/i386/immintrin.h: Include new headers.
1030 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
1031 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
1033 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1035 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
1036 fixes. Declare temp and g variables at the top in order to avoid
1037 {} in most of the cases.
1039 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
1041 * match.pd (ptr-0): New transformation.
1043 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
1045 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
1046 (struct addr_mode_cost_table): New.
1047 (struct tune_params): Add field addr_mode_costs.
1048 * config/arm/arm.c (generic_addr_mode_costs): New.
1049 (arm_slowmul_tune): Initialise addr_mode_costs field.
1050 (arm_fastmul_tune): Likewise.
1051 (arm_strongarm_tune): Likewise.
1052 (arm_xscale_tune): Likewise.
1053 (arm_9e_tune): Likewise.
1054 (arm_marvell_pj4_tune): Likewise.
1055 (arm_v6t2_tune): Likewise.
1056 (arm_cortex_tune): Likewise.
1057 (arm_cortex_a8_tune): Likewise.
1058 (arm_cortex_a7_tune): Likewise.
1059 (arm_cortex_a15_tune): Likewise.
1060 (arm_cortex_a35_tune): Likewise.
1061 (arm_cortex_a53_tune): Likewise.
1062 (arm_cortex_a57_tune): Likewise.
1063 (arm_exynosm1_tune): Likewise.
1064 (arm_xgene1_tune): Likewise.
1065 (arm_cortex_a5_tune): Likewise.
1066 (arm_cortex_a9_tune): Likewise.
1067 (arm_cortex_a12_tune): Likewise.
1068 (arm_cortex_a73_tune): Likewise.
1069 (arm_v7m_tune): Likewise.
1070 (arm_cortex_m7_tune): Likewise.
1071 (arm_v6m_tune): Likewise.
1072 (arm_fa726te_tune): Likewise.
1073 (arm_mem_costs): Use table lookup to calculate cost of addressing
1076 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
1078 * config/arm/arm.c (arm_mem_costs): New function.
1079 (arm_rtx_costs_internal): Use arm_mem_costs.
1081 2017-11-23 Mark Wielaard <mark@klomp.org>
1083 * dwarf2out.c (init_sections_and_labels): Use generation to create
1084 unique ranges_section_label and ranges_base_label. Return generation.
1085 (output_rnglists): Add generation argument. Use generation to create
1086 unique ranges labels.
1087 (dwarf2out_finish): Get generation from init_sections_and_labels
1088 and pass generation to output_rnglists.
1090 2017-11-23 Mike Stump <mikestump@comcast.net>
1091 Eric Botcazou <ebotcazou@adacore.com>
1093 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
1094 * cfgloop.h (struct loop): Add unroll field.
1095 * function.h (struct function): Add has_unroll bitfield.
1096 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
1097 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
1098 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
1099 (pass_rtl_unroll_loops::gate): Likewise.
1100 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
1101 for which loop->unroll==1.
1102 (decide_unroll_constant_iterations): Use note for consistency and
1103 take loop->unroll into account. Return early if loop->unroll is set.
1104 Fix thinko in existing test.
1105 (decide_unroll_runtime_iterations): Use note for consistency and
1106 take loop->unroll into account.
1107 (decide_unroll_stupid): Likewise.
1108 * lto-streamer-in.c (input_cfg): Read loop->unroll.
1109 * lto-streamer-out.c (output_cfg): Write loop->unroll.
1110 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
1112 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
1113 (print_loop): Print loop->unroll if set.
1114 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
1115 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
1116 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
1118 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
1119 loop->unroll is set and smaller than the trip count. Otherwise bypass
1120 entirely the heuristics if loop->unroll is set. Remove dead note.
1121 Fix off-by-one bug in other note.
1122 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
1123 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
1125 (tree_unroll_loops_completely): Make static.
1126 (pass_complete_unroll::execute): Use correct type for variable.
1127 (pass_complete_unrolli::execute): Fix formatting.
1128 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
1130 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
1132 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
1133 enable when TARGET_PREFER_AVX128 is set.
1135 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1137 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
1139 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
1140 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
1142 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1144 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
1145 frequencies. Fix estimation of aggregate parameters.
1147 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1149 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
1152 2017-11-23 Tom de Vries <tom@codesourcery.com>
1154 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
1156 2017-11-23 Tom de Vries <tom@codesourcery.com>
1158 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
1161 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
1164 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
1165 sibcall_value_pcrel_fdpic): Use local variable instead of
1167 (calli_tbr_rel): Add missing operand 2.
1168 (call_valuei_tbr_rel): Add missing operand 3.
1170 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1173 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
1174 bitpos/bitsize covering the whole destination, use store_expr only if
1175 the complex mode is the same. Otherwise, use expand_normal and if
1176 it returns CONCAT, subreg each part separately instead of trying to
1177 subreg the whole result.
1179 2017-11-23 Richard Biener <rguenther@suse.de>
1181 PR tree-optimization/23094
1182 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
1183 come here from walking over backedges in the first iteration.
1184 (vn_reference_lookup_3): Skip clobbers that store the same value.
1186 2017-11-23 Richard Biener <rguenther@suse.de>
1188 PR tree-optimization/81403
1189 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
1190 a block we need a leader relative to.
1191 (phi_translate_1): For nary processing require a leader from
1192 get_representative_for given we run expression simplification
1193 using match-and-simplify. Remove previous fix.
1195 2017-11-22 Jeff Law <law@redhat.com>
1197 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
1198 Use new method allocate_value_range rather than accessing the
1199 vrp_value_range_pool data member directly.
1200 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
1201 to use extract_range_from_stmt method to avoid need for
1202 extract_range_from_assignment method.
1203 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
1204 method rather than setting values_propgated data member directly.
1205 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
1206 and values propagated data members and extract_range_from_assignment
1207 method. Reorder private data members to conform to standards.
1208 Add new methods set_lattice_propagation_complete and
1209 allocate_value_range.
1211 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
1213 PR rtl-optimization/83030
1214 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
1215 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
1216 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
1217 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
1218 CROSSING_JUMP_P flag is already set before setting it.
1219 * cfgrtl.c (fixup_partition_crossing): Likewise.
1220 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
1223 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1225 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
1226 Handle the case where both arguments are using gen_const_vec_series.
1228 2017-11-22 David Malcolm <dmalcolm@redhat.com>
1231 * pretty-print.c (pp_format): Move quoting implementation to
1232 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
1233 to pass address of "quote" local.
1234 (pp_begin_quote): New function.
1235 (pp_end_quote): New function.
1236 * pretty-print.h (printer_fn): Convert penultimate param from bool
1238 (pp_begin_quote): New decl.
1239 (pp_end_quote): New decl.
1240 * tree-diagnostic.c (default_tree_printer): Convert penultimate
1241 param from bool to bool *.
1242 * tree-diagnostic.h (default_tree_printer): Likewise.
1244 2017-11-22 Jeff Law <law@redhat.com>
1246 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
1247 Initialize vr_values.
1248 (evrp_range_analyzer::try_find_new_range): Call methods attached to
1249 vr_values via vr_values class instance rather than delegators.
1250 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
1251 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1252 (evrp_range_analyzer::push_value_range): Likewise.
1253 (evrp_range_analyzer::pop_value_range): Likewise.
1254 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
1255 most delegators. Those remaining are exposed as public interfaces.
1256 Make vr_values a pointer and private.
1257 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
1259 (evrp_range_analyzer::get_vr_value): New method.
1260 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
1261 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
1262 (class evrp_dom_walker): Attach evrp_folder class, initialize
1263 it in the ctor. Remove temporary delegators.
1264 (evrp_dom_walker::before_dom_children): Call methods in attached
1265 evrp_range_analyzer class via class instance pointer. Use
1266 free value_range_constant_singleton to remove need for
1267 op_with_constant_singleton_value delegator method. Do not
1268 create a vrp_prop class instance for every call! Narrow
1269 scope of a couple variables.
1270 (evrp_dom_walker::cleanup): Call methods in attached
1271 evrp_range_analyzer class via class instance pointer.
1272 * vr-values.h (class vr_values): Privatize many methods and
1275 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1277 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
1278 * cfgexpand.c (expand_debug_expr): Remove handling for them.
1279 * expr.c (expand_expr_real_2): Likewise.
1280 * fold-const.c (const_unop): Likewise.
1281 * optabs-tree.c (optab_for_tree_code): Likewise.
1282 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1283 * tree-inline.c (estimate_operator_cost): Likewise.
1284 * tree-pretty-print.c (dump_generic_node): Likewise.
1285 (op_code_prio): Likewise.
1286 (op_symbol_code): Likewise.
1287 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
1288 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
1289 * internal-fn.c (direct_internal_fn_optab): New function.
1290 (direct_internal_fn_array, direct_internal_fn_supported_p
1291 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
1292 * fold-const-call.c (fold_const_reduction): New function.
1293 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
1295 * tree-vect-loop.c: Include internal-fn.h.
1296 (reduction_code_for_scalar_code): Rename to...
1297 (reduction_fn_for_scalar_code): ...this and return an internal
1299 (vect_model_reduction_cost): Take an internal_fn rather than
1301 (vect_create_epilog_for_reduction): Likewise. Build calls rather
1303 (vectorizable_reduction): Use internal functions rather than tree
1304 codes for the reduction operation. Update calls to the functions
1306 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
1307 Use calls to internal functions rather than REDUC tree codes.
1308 * config/aarch64/aarch64-simd.md: Update comment accordingly.
1310 2017-11-22 Olivier Hainque <hainque@adacore.com>
1312 * config/vxworks.c (vxworks_override_options): Pick default
1313 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
1314 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
1315 DWARF_GNAT_ENCODINGS_DEFAULT.
1316 * config/vxworksae.h: Likewise.
1318 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
1320 PR tree-optimization/83104
1321 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
1324 2017-11-22 Marek Polacek <polacek@redhat.com>
1325 H.J. Lu <hongjiu.lu@intel.com>
1326 Jason Merrill <jason@redhat.com>
1331 * calls.c (initialize_argument_information): Call
1332 warn_parameter_passing_abi target hook.
1333 (store_one_arg): Use 0 for empty record size. Don't push 0 size
1334 argument onto stack.
1335 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
1336 * common.opt: Update -fabi-version description.
1337 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
1338 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
1340 (ix86_is_empty_record): New function.
1341 (ix86_warn_parameter_passing_abi): New function.
1342 (TARGET_EMPTY_RECORD_P): Redefine.
1343 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
1344 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
1345 * doc/tm.texi: Regenerated.
1346 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
1347 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
1348 * dwarf2out.c (get_ultimate_context): Move to tree.c.
1349 * explow.c (hard_function_value): Call arg_int_size_in_bytes
1350 instead of int_size_in_bytes.
1351 * expr.c (copy_blkmode_to_reg): Likewise.
1352 * function.c (aggregate_value_p): Return 0 for empty types.
1353 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi target hook.
1354 (locate_and_pad_parm): Call arg size_in_bytes instead
1356 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
1357 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
1358 * target.def (empty_record_p, warn_parameter_passing_abi): New target
1360 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
1361 (std_gimplify_va_arg_expr): Skip empty records. Call
1362 arg_size_in_bytes instead size_in_bytes.
1363 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
1364 * tree-core.h (tree_type_common): Add empty_flag.
1365 (tree_decl_common): Update comments.
1366 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
1368 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1369 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
1371 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1372 * tree.c (default_is_empty_type): New function.
1373 (default_is_empty_record): New function.
1374 (arg_int_size_in_bytes): New function.
1375 (arg_size_in_bytes): New function.
1376 (get_ultimate_context): New function.
1377 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
1378 TRANSLATION_UNIT_WARN_EMPTY_P.
1379 (default_is_empty_record, arg_int_size_in_bytes,
1380 arg_size_in_bytes, get_ultimate_context): Declare.
1382 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1384 * config/arm/arm.c (cmse_clear_registers): New function.
1385 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
1386 code by call to cmse_clear_registers.
1387 (cmse_nonsecure_entry_clear_before_return): Likewise.
1389 2017-11-22 Tamar Christina <tamar.christina@arm.com>
1391 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
1392 (vdot_s32, vdotq_s32): New.
1393 (vdot_lane_u32, vdotq_lane_u32): New.
1394 (vdot_lane_s32, vdotq_lane_s32): New.
1397 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1400 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
1401 for unsigned values with fewer HWIs than the precision.
1402 (test_overflow): New function.
1403 (wide_int_cc_tests): Call it.
1405 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1407 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
1408 has the same mode class as Pmode.
1410 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1412 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
1413 padding_bits_to_clear_ptr.
1414 (cmse_nonsecure_entry_clear_before_return): Likewise.
1416 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1418 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
1419 auto_sbitap instead of integer bitfield to control register needing
1422 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1424 PR tree-optimization/83044
1425 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
1426 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
1427 that need the upper bound. Subtract offset from
1428 get_addr_base_and_unit_offset only if positive and subtract it
1429 before division by eltsize rather than after it.
1432 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
1433 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
1434 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
1438 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
1439 expand_binop with *mul_widen_optab, make sure at least one of the
1440 operands doesn't have VOIDmode.
1443 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
1445 PR rtl-optimization/82044
1446 PR tree-optimization/82042
1447 * dse.c (record_store): Check for overflow.
1448 (check_mem_read_rtx): Properly check for overflow if width == -1, call
1449 add_wild_read instead of clear_rhs_from_active_local_stores on
1450 overflow and log it into dump_file.
1452 2017-11-22 Richard Biener <rguenther@suse.de>
1454 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
1455 fake edges to exit when looking for a place to insert.
1456 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
1458 (insert_into_preds_of_block): Commit edge insertion immediately,
1459 assert that doesn't require new BBs.
1460 (fini_pre): Release expressions.
1461 (pass_pre::execute): Shuffle things around a bit, if the fn
1462 is too large do not compute AVAIL either as this is really the
1465 2017-11-22 Richard Biener <rguenther@suse.de>
1467 PR tree-optimization/83089
1468 * tree-if-conv.c (pass_if_conversion::execute): If anything
1469 changed reset SCEV and free the number of iteration estimates.
1471 2017-11-21 Martin Sebor <msebor@redhat.com>
1473 PR tree-optimization/82945
1474 * calls.h (warn_nonstring_bound): Remove unused function.
1476 2017-11-21 Martin Sebor <msebor@redhat.com>
1478 PR tree-optimization/82945
1479 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
1480 * calls.h (maybe_warn_nonstring_arg): Declare new function.
1481 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
1483 (initialize_argument_information): Call maybe_warn_nonstring_arg.
1484 * calls.h (get_attr_nonstring_decl): Declare new function.
1485 * doc/extend.texi (attribute nonstring): Update.
1486 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
1487 get_attr_nonstring_decl and handle it.
1488 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
1489 detection of nul-termination.
1490 (strlen_to_stridx): Change to a pointer.
1491 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
1492 (pass_strlen::execute): Same.
1494 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
1496 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
1497 for the new option -mprefer-vector-width=[none|128|256|512].
1498 * config/i386/i386.c (ix86_target_string): remove old style options
1499 -mprefer-avx256 and make -mprefer-avx128 as alias.
1500 (ix86_option_override_internal): Apply defaults for the
1501 -mprefer-vector-width=[128|256] option.
1502 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
1503 Implement macros to work with -mprefer-vector-width=.
1504 * config/i386/i386.opt: Implemented option
1505 -mprefer-vector-width=[none|128|256|512].
1506 * doc/invoke.texi: Documentation for
1507 -mprefer-vector-width=[none|128|256|512].
1509 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
1511 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
1513 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
1515 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
1517 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
1519 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
1521 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
1522 * expr.c (expand_expr_real_2): Likewise.
1523 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
1524 fold_binary_loc): Likewise.
1525 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
1526 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
1527 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
1528 MINUS_EXPR transformations.
1529 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
1530 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
1531 * tree-inline.c (estimate_operator_cost): Likewise.
1532 * tree-pretty-print.c (dump_generic_node, op_code_prio,
1533 op_symbol_code): Likewise.
1534 * tree-vect-stmts.c (vectorizable_operation): Likewise.
1535 * vr-values.c (extract_range_from_binary_expr): Likewise.
1536 * varasm.c (initializer_constant_valid_p_1): Likewise.
1537 * tree.def: New tree code POINTER_DIFF_EXPR.
1539 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
1541 * config/i386/i386.md (*bswap<mode>2_movbe): Add
1542 integer suffix to movbe mnemonic.
1543 (*bswaphi2_movbe): Ditto.
1544 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
1546 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1549 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
1550 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
1551 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
1553 2017-11-21 Martin Liska <mliska@suse.cz>
1555 * tree-inline.c (expand_call_inline): Remove not needed
1558 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
1559 Jakub Jelinek <jakub@redhat.com>
1562 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
1563 Remove static keyword from f variable.
1565 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1567 PR tree-optimization/83086
1568 * gimple-ssa-store-merging.c
1569 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
1570 rather than n.base_addr.
1572 2017-11-21 Martin Liska <mliska@suse.cz>
1574 PR rtl-optimization/82044
1575 PR tree-optimization/82042
1576 * dse.c (check_mem_read_rtx): Check for overflow.
1578 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1580 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
1583 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
1585 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
1586 and its corresponding call arg location note.
1588 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
1589 Andrew Burgess <andrew.burgess@embecosm.com>
1591 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
1593 (arc_return_slot_offset): Likewise.
1594 (arc_eh_return_address_location): New declaration.
1595 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
1596 (MUST_SAVE_REGISTER): Add exception handler case.
1597 (MUST_SAVE_RETURN_ADDR): Likewise.
1598 (arc_frame_pointer_required): Likewise.
1599 (arc_frame_pointer_needed): New function.
1600 (arc_compute_frame_size): Changed.
1601 (arc_expand_prologue): Likewise.
1602 (arc_expand_epilogue): Likewise.
1603 (arc_initial_elimination_offset): Likewise.
1604 (arc_return_slot_offset): Delete.
1605 (arc_eh_return_address_location): New function.
1606 (arc_builtin_setjmp_frame_value): Likewise.
1607 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
1608 (EH_RETURN_STACKADJ_RTX): Define.
1609 (EH_RETURN_HANDLER_RTX): Likewise.
1610 * config/arc/arc.md (eh_return): Delete.
1612 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1614 * print-tree.h (debug_vec_tree): Remove prototype.
1615 * gdbinit.in (pvt): Remove macro.
1617 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1619 PR tree-optimization/83047
1620 * gimple-ssa-store-merging.c
1621 (imm_store_chain_info::output_merged_store): If the loads with the
1622 same vuse are in different basic blocks, for load_gsi pick a load
1623 location that is dominated by the other loads.
1626 * config/i386/i386.c (ix86_memmodel_check): Start
1627 -Winvalid-memory-model diagnostics with lowercase letter.
1630 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
1631 set in_cold_section_p to the partition containing loc_list->first.
1632 When seeing loc_list->last_before_switch node, update secname and
1633 perform range_across_switch second partition handling only after that.
1636 * run-rtl-passes.c: Include debug.h.
1637 (run_rtl_passes): Call debug_hooks->assembly_start.
1638 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
1642 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
1643 OPTAB_DIRECT in calls to expand_simple_binop.
1645 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1648 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
1650 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1653 * spellcheck.h (best_match::blithely_get_best_candidate): New
1656 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1658 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
1659 with lower case letter.
1661 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
1663 * config/i386/i386.md (bswaphi2): New expander.
1664 (*bswaphi2_movbe): New insn pattern.
1665 (bswaphi -> rorhi pepehole2): New peephole pattern.
1667 2017-11-20 Jeff Law <law@redhat.com>
1669 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
1670 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
1671 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
1672 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
1673 gimple-ssa-evrp-analyze.h.
1675 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
1676 set BB_VISITED here.
1677 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
1679 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1681 * tree-predcom.c: Add general comment on Store-Store chains.
1682 (split_data_refs_to_components): Postpone clearing eliminate_store_p
1684 (get_chain_last_ref_at): Rename into...
1685 (get_chain_last_write_at): ...this.
1686 (get_chain_last_write_before_load): New function.
1687 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
1688 CT_STORE_STORE when write reference is added.
1689 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
1690 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
1691 (initialize_root_vars_store_elim_1): Ditto.
1692 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
1693 definition is created.
1694 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
1695 chain by replacing it with dominant stored value.
1697 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1699 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
1701 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
1703 * vr-values.c (extract_range_from_binary_expr): Use a full range
1706 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
1708 * config/arm/arm.md (R4_REGNUM): Define constant.
1709 (nonsecure_call_internal): Remove r4 clobber.
1710 (nonsecure_call_value_internal): Likewise.
1711 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
1712 clobber and resequence match_operands.
1713 (nonsecure_call_value_reg_thumb1_v5): Likewise.
1714 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
1715 (nonsecure_call_value_reg_thumb2): Likewise.
1717 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1719 PR tree-optimization/78821
1720 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
1721 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
1722 to the address of the base rather than the base itself.
1723 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
1724 (find_bswap_or_nop_finalize): New function.
1725 (find_bswap_or_nop): Use it.
1726 (bswap_replace): Return a tree rather than bool, change first
1727 argument from gimple * to gimple_stmt_iterator, allow inserting
1728 into an empty sequence, allow ins_stmt to be NULL - then emit
1729 all stmts into gsi. Fix up MEM_REF address gimplification.
1730 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
1731 (struct store_immediate_info): Add N and INS_STMT non-static
1733 (store_immediate_info::store_immediate_info): Initialize them
1734 from newly added ctor args.
1735 (merged_store_group::apply_stores): Formatting fixes. Sort by
1737 (stmts_may_clobber_ref_p): For stores call also
1738 refs_anti_dependent_p.
1739 (gather_bswap_load_refs): New function.
1740 (imm_store_chain_info::try_coalesce_bswap): New method.
1741 (imm_store_chain_info::coalesce_immediate_stores): Use it.
1742 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
1743 (imm_store_chain_info::output_merged_store): Fail if number of
1744 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
1745 and NOP_EXPR rhs_code.
1746 (pass_store_merging::process_store): Compute n and ins_stmt, if
1747 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
1748 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
1750 (pass_store_merging::execute): Calculate dominators.
1752 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
1753 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
1754 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
1755 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
1756 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
1757 class pass_optimize_bswap, bswap_replace,
1758 pass_optimize_bswap::execute): Moved to ...
1759 * gimple-ssa-store-merging.c: ... this file.
1760 Include optabs-tree.h.
1761 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
1762 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
1763 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
1764 anonymous namespace, remove static keywords.
1765 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
1766 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
1768 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
1771 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
1773 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
1775 * vec.h (debug_helper): New function.
1776 (DEFINE_DEBUG_VEC): New macro.
1777 * hash-set.h (debug_helper): New function.
1778 (DEFINE_DEBUG_HASH_SET): New macro.
1779 * cfg.c (debug_slim (edge)): New function.
1780 Call DEFINE_DEBUG_VEC for edges.
1781 Call DEFINE_DEBUG_HASH_SET for edges.
1782 * cfghooks.c (debug_slim (basic_block)): New function.
1783 Call DEFINE_DEBUG_VEC for basic blocks.
1784 Call DEFINE_DEBUG_HASH_SET for basic blocks.
1785 * print-tree.c (debug_slim): New function to handle trees.
1786 Call DEFINE_DEBUG_VEC for trees.
1787 Call DEFINE_DEBUG_HASH_SET for trees.
1788 (debug (vec<tree, va_gc>) &): Remove.
1789 (debug (<vec<tree, va_gc>) *): Remove.
1790 * print-rtl.c (debug_slim): New function to handle const_rtx.
1791 Call DEFINE_DEBUG_VEC for rtx_def.
1792 Call DEFINE_DEBUG_VEC for rtx_insn.
1793 Call DEFINE_DEBUG_HASH_SET for rtx_def.
1794 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
1795 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
1796 (debug (vec<rtx_insn *> *ptr): Remove.
1797 (debug_insn_vector): Remove.
1798 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
1800 2017-11-20 Tom de Vries <tom@codesourcery.com>
1802 PR rtl-optimization/82020
1803 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
1804 IF_THEN_ELSE condition.
1806 2017-11-19 Jeff Law <law@redhat.com>
1808 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1809 of degenerates resulting from ignoring an edge.
1811 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1814 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
1816 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1819 * profile-count.c (profile_count::to_sreal_scale): Fix return value
1820 for uninitialied counts.
1822 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1825 * tree-inline.c (estimate_num_insns): Set to 1 at least.
1827 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1830 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
1832 2017-11-19 Tom de Vries <tom@codesourcery.com>
1834 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
1837 2017-11-19 Tom de Vries <tom@codesourcery.com>
1839 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
1842 2017-11-19 Tom de Vries <tom@codesourcery.com>
1844 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1847 2017-11-19 Tom de Vries <tom@codesourcery.com>
1849 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
1851 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
1853 2017-11-19 Tom de Vries <tom@codesourcery.com>
1855 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
1857 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
1858 semicolon after MCORE_EXPORT_NAME call.
1860 2017-11-19 Tom de Vries <tom@codesourcery.com>
1863 * vmsdbgout.c (vmsdbgout_early_finish): New function.
1864 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
1866 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
1868 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
1869 (cgraph_edge::clone): Cleanup updating of profile.
1870 * ipa-cp.c (update_profiling_info): Likewise.
1871 * ipa-inline-transform.c (inline_transform): Likewise.
1872 * ipa-inline.c (inline_small_functions): Add missing space to dump.
1873 * ipa-split.c (execute_split_functions): Do not split when function
1875 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
1876 * profile-count.c (profile_count::dump): Add global0.
1877 (profile_count::to_cgraph_frequency): Do not ICE when entry is
1879 (profile_count::to_sreal_scale): Likewise.
1880 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
1881 (profile_count::combine_with_ipa_count): New function.
1882 * profile-count.h (profile_guessed_global0adjusted): New.
1883 (profile_count::adjusted_zero): New.
1884 (profile_count::global0adjusted): New.
1885 (profile_count::combine_with_ipa_count): New.
1886 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
1887 correct profile of return block of split functions.
1888 (copy_cfg_body): Remove unused profile_count.
1889 (copy_body): Likewise.
1890 (expand_call_inline): Update.
1891 (tree_function_versioning): Update.
1893 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
1895 * hash-set.h (hash_set::empty): New.
1896 * tree-ssa-threadbackward.h: Delete.
1897 * tree-ssa-threadbackward.c (class thread_jumps): New.
1898 Move max_threaded_paths into class.
1899 (fsm_find_thread_path): Remove arguments that are now in class.
1900 (profitable_jump_thread_path): Rename to...
1901 (thread_jumps::profitable_jump_thread_path): ...this.
1902 (convert_and_register_jump_thread_path): Rename to...
1903 (thread_jumps::convert_and_register_current_path): ...this.
1904 (check_subpath_and_update_thread_path): Rename to...
1905 (thread_jumps::check_subpath_and_update_thread_path): ...this.
1906 (register_jump_thread_path_if_profitable): Rename to...
1907 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
1908 (handle_phi): Rename to...
1909 (thread_jumps::handle_phi): ...this.
1910 (handle_assignment): Rename to...
1911 (thread_jumps::handle_assignment): ...this.
1912 (fsm_find_control_statement_thread_paths): Rename to...
1913 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
1914 (find_jump_threads_backwards): Rename to...
1915 (thread_jumps::find_jump_threads_backwards): ...this.
1916 Initialize path local data.
1917 (pass_thread_jumps::execute): Call find_jump_threads_backwards
1918 from within thread_jumps class.
1919 (pass_early_thread_jumps::execute): Same.
1921 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1923 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
1925 2017-11-17 Steve Ellcey <sellcey@cavium.com>
1927 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
1928 to canonical location.
1930 2017-11-17 Steve Ellcey <sellcey@cavium.com>
1933 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
1935 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
1937 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
1939 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
1940 rcrt1.o%s/grcrt1.o%s for -static-pie.
1942 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1944 * i386.c (ix86_multiplication_cost, ix86_division_cost,
1945 ix86_shift_rotate_cost): Break out from ...
1946 (ix86_rtx_costs): ... here.
1947 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
1950 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1952 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
1953 when entry block was promoted unlikely.
1954 (estimate_bb_frequencies): Increase frequency scale.
1955 * profile-count.h (profile_count): Export precision info.
1957 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1959 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
1960 disturbin profile of entry block.
1962 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1964 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
1967 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1969 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
1970 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
1972 2017-11-17 Jeff Law <law@redhat.com>
1974 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
1975 from evrp_dom_walker class. Various methods moved into new class.
1976 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
1977 (evrp_range_analyzer::enter): New method.
1978 (evrp_range_analyzer::leave): New method.
1979 (evrp_dom_walker): Remove delegators no longer needed by this class.
1980 Replace vr_values data member with evrp_range_analyzer
1982 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
1983 method extracted from evrp_dom_walker::before_dom_children.
1984 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
1985 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
1987 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
1988 Add private copy constructor and move assignment operators.
1989 Privatize methods and class data where trivially possible.
1990 (evrp_dom_walker::cleanup): New function, extracted from
1991 execute_early_vrp. Simplify access to class data.
1993 * vr-values.h (get_output_for_vrp): Prototype.
1994 * vr-values.c (get_output_for_vrp): New function extracted from
1995 vrp_visit_assignment_or_call and extract_range_from_stmt.
1996 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
1998 2017-11-17 Luis Machado <luis.machado@linaro.org>
2000 * config/aarch64/aarch64.c
2001 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
2002 (qdf24xx_tunings) <autoprefetcher_model>: Set to
2003 tune_params::AUTOPREFETCHER_WEAK.
2005 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2008 * config/arm/arm.c (arm_valid_target_attribute_rec):
2009 Parse "arch=" and "+<ext>".
2010 (arm_valid_target_attribute_tree): Re-init global options.
2011 (arm_option_override): Make non-static.
2012 (arm_options_perform_arch_sanity_checks): Make errors fatal.
2013 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
2014 (__ARM_FEATURE_CRC32): Support undef.
2015 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
2016 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
2018 2017-11-17 David Malcolm <dmalcolm@redhat.com>
2020 * gdbinit.in (break-on-diagnostic): New command.
2022 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2024 * config/i386/i386.c (ix86_expand_epilogue): Change simple
2025 return to indirect jump for EH return if control-flow
2026 protection is enabled. Change explicit 'false' argument in
2027 pro_epilogue_adjust_stack with a value of
2029 * config/i386/i386.md (simple_return_indirect_internal):
2030 Remove SImode restriction to support 64-bit.
2032 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2034 * combine.c (added_notes_insn): New.
2035 (try_combine): Handle added_notes_insn like added_links_insn.
2036 Rewrite return value code.
2037 (distribute_notes): Set added_notes_insn to the earliest insn we added
2040 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2042 PR rtl-optimization/82621
2043 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
2044 dest of one of those SETs is unused.
2046 2017-11-17 Richard Biener <rguenther@suse.de>
2049 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
2050 * tree-pretty-print.c (dump_generic_node): Handle
2051 annot_expr_parallel_kind.
2052 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
2053 * gimplify.c (gimple_boolify): Likewise.
2055 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2057 * config.gcc (extra_headers): Add cet.h for x86 targets.
2058 * config/i386/cet.h: New file.
2059 * doc/install.texi: Add --enable-cet/--disable-cet.
2061 2017-11-17 Richard Biener <rguenther@suse.de>
2063 PR tree-optimization/83017
2064 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
2065 (gen_parallel_loop): Properly count iterations.
2066 (parallelize_loops): Handle loop->can_be_parallel independent
2067 of flag_loop_parallelize_all. Make static profitability test match
2069 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
2070 * invoke.texi (parloops-min-per-thread): Document.
2072 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
2074 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
2075 upstreaming review comments.
2077 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2079 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
2080 with fast unaligned access.
2081 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
2083 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
2085 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
2086 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
2088 2017-11-17 Richard Biener <rguenther@suse.de>
2090 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
2091 folding of references.
2093 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
2096 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
2097 of replacing call to strncmp with corresponding call to strcmp when
2100 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
2102 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
2103 option prefer-avx256 for skylake-avx512 configuration.
2104 * config/i386/i386.c (ix86_option_override_internal): Ditto.
2105 (get_builtin_code_for_version): Ditto.
2107 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
2108 Monk Chiang <sh.chiang04@gmail.com>
2110 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
2111 (FIXED_REGISTERS): Reserve more register numbers.
2112 (CALL_USED_REGISTERS): Likewise.
2113 (REG_ALLOC_ORDER): Likewise.
2114 (REG_CLASS_CONTENTS): Likewise.
2115 (REGISTER_NAMES): Likewise.
2117 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
2118 Kito Cheng <kito.cheng@gmail.com>
2120 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
2122 * config/nds32/iterators.md: Add vector mode iterators and attributes.
2124 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
2126 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
2127 parameter list for vec_splats.
2129 2017-11-16 Joseph Myers <joseph@codesourcery.com>
2131 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
2133 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
2135 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
2138 * doc/install.texi: Document incompatibility of Perl >=5.6.26
2139 with the required version of automake 1.11.6.
2141 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
2143 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
2145 (DU_C2_3_power9): Correct reservation combinations.
2146 (FP_DIV_power9, VEC_DIV_power9): New.
2147 (power9-alu): Split out rotate/shift...
2148 (power9-rot): ...to here, correct dispatch resource.
2149 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
2151 (power9-fp): Correct latency.
2152 (power9-sdiv): Add div/sqrt resource.
2153 (power9-ddiv): Correct latency, add div/sqrt resource.
2154 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
2155 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
2157 (power9-qpdiv, power9-qpmul): Adjust resource usage.
2159 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2161 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
2162 switch statement mapping KF built-ins to TF built-ins if we don't
2163 have the proper ISA 3.0 assembler support.
2165 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2167 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
2168 (lower_emutls_function_body): Do not compute it.
2170 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2172 * ipa-split.c (split_bb_info): Turn time to sreal.
2173 (split_point): Likewise.
2174 (dump_split_point): Likewise.
2175 (fine_split_points): Likewise.
2176 (execute_split_functions): Only zero split_bbs; turn time to sreals.
2178 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2180 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
2183 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2185 * predict.c (combine_predictions_for_bb): Preserve zero predicted
2187 (expensive_function_p): Remove useless assert.
2188 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
2190 2017-11-16 Martin Sebor <msebor@redhat.com>
2192 PR tree-optimization/82588
2193 PR tree-optimization/82583
2194 * tree-vrp.c (check_array_ref): Handle flexible array members,
2195 string literals, and inner indices.
2196 (search_for_addr_array): Add detail to diagnostics.
2198 2017-11-16 Nathan Sidwell <nathan@acm.org>
2202 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
2203 (SET_DECL_ASSEMBLER_NAME): Forward to
2204 overwrite_decl_assembler_name.
2205 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
2206 (overwrite_decl_assembler_name): Declare.
2207 * tree.c (overwrite_decl_assembler_name): New.
2208 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
2209 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
2210 (LANG_HOOKS_INITIALIZER): Add it.
2211 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
2212 * langhooks.c (lhd_set_decl_assembler_name): Use
2213 SET_DECL_ASSEMBLER_NAME.
2214 (lhd_overwrite_decl_assembler_name): Default implementation.
2216 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2217 Jackson Woodruff <jackson.woodruff@arm.com>
2219 PR tree-optimization/71026
2220 * match.pd: Canonicalize constant multiplies in division.
2222 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2224 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
2225 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
2226 Remove OPT_fomit_frame_pointer entry.
2227 * common/config/arc/arc-common.c: Likewise.
2228 * common/config/arm/arm-common.c: Likewise.
2229 * common/config/avr/avr-common.c: Likewise.
2230 * common/config/c6x/c6x-common.c: Likewise.
2231 * common/config/cr16/cr16-common.c: Likewise.
2232 * common/config/cris/cris-common.c: Likewise.
2233 * common/config/epiphany/epiphany-common.c: Likewise.
2234 * common/config/fr30/fr30-common.c: Likewise.
2235 * common/config/frv/frv-common.c: Likewise.
2236 * common/config/ia64/ia64-common.c: Likewise.
2237 * common/config/iq2000/iq2000-common.c: Likewise.
2238 * common/config/lm32/lm32-common.c: Likewise.
2239 * common/config/m32r/m32r-common.c: Likewise.
2240 * common/config/mcore/mcore-common.c: Likewise.
2241 * common/config/microblaze/microblaze-common.c: Likewise.
2242 * common/config/mips/mips-common.c: Likewise.
2243 * common/config/mmix/mmix-common.c: Likewise.
2244 * common/config/mn10300/mn10300-common.c: Likewise.
2245 * common/config/nios2/nios2-common.c: Likewise.
2246 * common/config/pa/pa-common.c: Likewise.
2247 * common/config/pdp11/pdp11-common.c: Likewise.
2248 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
2249 * common/config/riscv/riscv-common.c: Likewise.
2250 * common/config/rs6000/rs6000-common.c: Likewise.
2251 * common/config/rx/rx-common.c: Likewise.
2252 * common/config/s390/s390-common.c: Likewise.
2253 * common/config/sh/sh-common.c: Likewise.
2254 * common/config/sparc/sparc-common.c: Likewise.
2255 * common/config/tilegx/tilegx-common.c: Likewise.
2256 * common/config/tilepro/tilepro-common.c: Likewise.
2257 * common/config/v850/v850-common.c: Likewise.
2258 * common/config/visium/visium-common.c: Likewise.
2259 * common/config/xstormy16/xstormy16-common.c: Likewise.
2260 * common/config/xtensa/xtensa-common.c: Likewise.
2261 * invoke.texi (-fomit-frame-pointer): Update documentation.
2263 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2265 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
2267 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2269 * cfg.c (scale_bbs_frequencies_int,
2270 cale_bbs_frequencies_gcov_type): Remove.
2271 * cfg.h (scale_bbs_frequencies_int,
2272 cale_bbs_frequencies_gcov_type): Remove.
2274 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2276 * tree-ssa-loop-manip.c
2277 (scale_dominated_blocks_in_loop): Update to profile counts.
2278 (tree_transform_and_unroll_loop): Likewise.
2280 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2282 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
2283 scale_bbs_frequencies_int.
2285 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2287 * final.c (compute_alignments): Use counts rather than frequencies.
2289 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2291 * cfgloopanal.c: Include sreal.h
2292 (average_num_loop_insns): Use counts and sreal for accounting.
2294 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2296 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
2299 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2300 Kito Cheng <kito.cheng@gmail.com>
2302 * config/nds32/constraints.md: Provide more constraints.
2303 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
2304 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
2305 support constraints usage.
2307 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2309 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
2311 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2312 Kito Cheng <kito.cheng@gmail.com>
2314 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
2315 * config/nds32/nds32.opt: Refine the layout.
2316 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
2317 TARGET_EXT_STRING): Support new options.
2318 * config/nds32/nds32.h: Likewise.
2319 * config/nds32/nds32.md: Likewise.
2320 * config/nds32/nds32-predicates.c: Likewise.
2321 * config/nds32/constraints.md: Likewise.
2322 * common/config/nds32/nds32-common.c: Likewise.
2324 2017-11-16 Julia Koval <julia.koval@intel.com>
2327 * config/i386/gfniintrin.h: Add sse check.
2328 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
2330 2017-11-16 Julia Koval <julia.koval@intel.com>
2332 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
2333 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
2334 (ix86_handle_option): Handle -mavx512vbmi2.
2335 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
2336 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
2337 * config/i386/i386-c.c (__AVX512VBMI2__): New.
2338 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
2339 (ix86_valid_target_attribute_inner_p): Ditto.
2340 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
2341 * config/i386/i386.opt (mavx512vbmi2): New option.
2342 * doc/invoke.texi: Add new option.
2344 2017-11-16 Julia Koval <julia.koval@intel.com>
2346 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
2347 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
2348 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
2349 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
2350 _mm512_gf2p8mul_epi8): New intrinsics.
2351 * config/i386/i386-builtin-types.def
2352 (V64QI_FTYPE_V64QI_V64QI): New type.
2353 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
2354 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
2355 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
2356 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
2357 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
2358 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2360 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
2362 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
2363 explicitly as a stream of bytes.
2365 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2367 * config/rs6000/altivec.h (vec_xst_be): New #define.
2368 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
2369 and externalize from *altivec_vperm_<mode>_internal.
2370 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
2372 (XL_BE_V8HI): Likewise.
2373 (XL_BE_V4SI): Likewise.
2374 (XL_BE_V4SI): Likewise.
2375 (XL_BE_V2DI): Likewise.
2376 (XL_BE_V4SF): Likewise.
2377 (XL_BE_V2DF): Likewise.
2378 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
2379 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
2380 all array entries with these keys: VSX_BUILTIN_VEC_XL,
2381 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
2382 VSX_BUILTIN_VEC_XST_BE.
2383 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
2384 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
2386 (altivec_init_builtins): Replace conditional calls to def_builtin
2387 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
2388 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
2389 with unconditional calls. Remove calls to def_builtin for
2390 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
2391 __builtin_vec_xst_be.
2392 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
2393 to define_expand, and add alternate RTL generation for P8.
2394 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
2395 vsx_ld_elemrev_v8hi.
2396 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
2397 add alternate RTL generation for P8.
2398 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
2399 vsx_ld_elemrev_v16qi.
2400 (vsx_st_elemrev_v8hi): Convert define_insn
2401 to define_expand, and add alternate RTL generation for P8.
2402 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
2403 vsx_st_elemrev_v8hi.
2404 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
2405 add alternate RTL generation for P8.
2406 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
2407 vsx_st_elemrev_v16qi.
2409 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
2412 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
2413 TARGET_AVX512ER check.
2414 (ix86_option_override_internal): Set MASK_VZEROUPPER if
2415 neither -mzeroupper nor -mno-zeroupper is used and
2416 TARGET_EMIT_VZEROUPPER is set.
2417 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
2418 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
2420 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
2422 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2423 folding of vector compares.
2424 (fold_build_vec_cmp): New helper function.
2425 (fold_compare_helper): New helper function.
2426 (builtin_function_type): Add compare builtins to the list of functions
2427 having unsigned arguments. Cosmetic updates to comment indentation.
2428 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
2429 the not+eq combination.
2431 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2433 PR tree-optimization/82726
2434 PR tree-optimization/70754
2435 * tree-predcom.c (order_drefs_by_pos): New function.
2436 (combine_chains): Move code setting has_max_use_after to...
2437 (try_combine_chains): ...here. New parameter. Sort combined chains
2438 according to position information.
2439 (tree_predictive_commoning_loop): Update call to above function.
2440 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
2442 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2444 PR tree-optimization/82726
2446 2017-01-23 Bin Cheng <bin.cheng@arm.com>
2448 PR tree-optimization/70754
2449 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
2450 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
2451 combined stmt before it if not NULL.
2452 (combine_chains): Process refs reversely and compute dominance point
2456 2017-02-23 Bin Cheng <bin.cheng@arm.com>
2458 PR tree-optimization/79663
2459 * tree-predcom.c (combine_chains): Process refs in reverse order
2460 only for ZERO length chains, and add explaining comment.
2462 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2464 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
2465 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
2466 * doc/invoke.texi (ARM Options): Add armv8.3-a.
2468 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2470 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
2472 2017-11-15 Martin Liska <mliska@suse.cz>
2474 * tree-cfg.c (pass_warn_function_return::execute):
2475 Compare warn_return_type for greater than zero.
2477 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
2481 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
2482 to return true on Xeon and not on Xeon Phi.
2483 (ix86_check_avx256_register): Changed to ...
2484 (ix86_check_avx_upper_register): ... this. Add extra check for
2485 VALID_AVX512F_REG_OR_XI_MODE.
2486 (ix86_avx_u128_mode_needed): Changed
2487 ix86_check_avx256_register to ix86_check_avx_upper_register.
2488 (ix86_check_avx256_stores): Changed to ...
2489 (ix86_check_avx_upper_stores): ... this. Changed
2490 ix86_check_avx256_register to ix86_check_avx_upper_register.
2491 (ix86_avx_u128_mode_after): Changed
2492 avx_reg256_found to avx_upper_reg_found. Changed
2493 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
2494 (ix86_avx_u128_mode_entry): Changed
2495 ix86_check_avx256_register to ix86_check_avx_upper_register.
2496 (ix86_avx_u128_mode_exit): Ditto.
2497 * config/i386/i386.h: (host_detect_local_cpu): New define.
2499 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2501 * config/arm/xgene1.md (xgene1): Split into automatons
2502 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
2503 (xgene1_f_load): Adjust reservations and/or types.
2504 (xgene1_f_store): Likewise.
2505 (xgene1_load_pair): Likewise.
2506 (xgene1_store_pair): Likewise.
2507 (xgene1_fp_load1): Likewise.
2508 (xgene1_load1): Likewise.
2509 (xgene1_store1): Likewise.
2510 (xgene1_move): Likewise.
2511 (xgene1_alu): Likewise.
2512 (xgene1_simd): Likewise.
2513 (xgene1_bfm): Likewise.
2514 (xgene1_neon_load1): Likewise.
2515 (xgene1_neon_store1): Likewise.
2516 (xgene1_neon_logic): Likewise.
2517 (xgene1_neon_st1): Likewise.
2518 (xgene1_neon_ld1r): Likewise.
2519 (xgene1_alu_cond): Added.
2520 (xgene1_shift_reg): Likwise.
2521 (xgene1_bfx): Likewise.
2522 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
2524 2017-11-15 Jakub Jelinek <jakub@redhat.com>
2527 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
2529 (can_widen_mult_without_libcall): New function.
2530 (expand_mul_overflow): If only checking unsigned mul overflow,
2531 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
2532 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
2533 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
2534 (expand_DIVMOD): Formatting fix.
2535 * expmed.h (expand_mult): Add NO_LIBCALL argument.
2536 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
2537 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
2539 PR tree-optimization/82977
2540 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
2541 constructed temporary to strlen_to_stridx.put.
2543 2017-11-15 Martin Liska <mliska@suse.cz>
2545 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
2546 * configure: Regenerate.
2548 2017-11-15 Martin Liska <mliska@suse.cz>
2551 * config/sh/sh-mem.cc: Use proper probability for
2554 2017-11-14 Jeff Law <law@redhat.com>
2556 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
2557 the red zone for stack_clash_protection_final_dynamic_probe targets
2558 when the total dynamic stack size is zero bytes.
2560 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
2561 blocks is post order.
2563 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
2565 * dumpfile.h (TDF_COMPARE_DEBUG): New.
2566 * final.c (rest_of_clean_state): Set it for the
2567 -fcompare-debug dump.
2568 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
2569 class when TDF_COMPARE_DEBUG is set.
2571 * dwarf2out.c (gen_producer_string): Discard
2574 2017-11-15 Joseph Myers <joseph@codesourcery.com>
2577 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
2578 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
2579 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
2580 (__TGMATH_REAL_2_3): Remove macros.
2581 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
2582 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
2584 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
2586 (remquo): Define using __TGMATH_REAL_3.
2588 2017-11-14 Jeff Law <law@redhat.com>
2590 * vr-values.c: New file with contents extracted from tree-vrp.c.
2591 * Makefile.in (OBJS): Add vr-values.o
2592 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
2593 (set_value_range, set_and_canonicalize_value_range): Likewise.
2594 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
2595 (value_range_constant_singleton, symbolic_range_p): Likewise.
2596 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
2597 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
2598 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
2599 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
2600 (find_case_label_range, find_case_label_index): Likewise.
2601 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
2602 (range_int_cst_singleton_p, value_inside_range): Likewise.
2603 (get_single_symbol): Likewise.
2604 (switch_update): Move structure definition here.
2605 (to_remove_edges, to_update_switch_stmts): Provide externs.
2606 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
2607 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
2608 (vrp_val_is_min, set_value_range): Likewise.
2609 (set_and_canonicalize_value_range, copy_value_range): Likewise.
2610 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
2611 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
2612 (range_is_nonnull, range_int_cst_p): Likewwise.
2613 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
2614 (get_single_symbol, operand_less_p): Likewise
2615 (compare_values_warnv, compare_values): Likewise.
2616 (value_inside_range, value_range_constant_singleton): Likewise.
2617 (zero_nonzero_bitgs_from_vr): Likewise.
2618 (extract_range_from_binary_expr_1): Likewise.
2619 (overflow_comparison_p): Likewise.
2620 (to_remove_edges, to_update_switch_stmts): Likewise.
2621 (find_case_label-index, find_case_label_range): Likewise.
2622 (switch_update, set_value_range_to_nonnegative): Remove.
2623 (set_value_range_to_truthvalue): Likewise.
2624 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
2625 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
2626 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
2627 (find_case_label_ranges, test_for_singularity): Likewise.
2628 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
2629 (x_vr_values): Move to its remaining use site.
2631 2017-11-10 Jeff Law <law@redhat.com>
2633 * vr-values.h (VR_INITIALIZER): Move #define here.
2634 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
2635 * Makefile.in (OBJS): Add tree-evrp.o
2636 * tree-vrp.h (assert_info): Move structure definition here.
2637 (set_value_range_to_varying): Prototype.
2638 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2639 (infer_value_range, register_edge_assert_for): Likewise.
2640 (stmt_interesting_for_vrp): Likewise.
2641 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
2642 (set_value_range_to_varying): No longer static.
2643 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2644 (infer_value_range, register_edge_assert_for): Likewise.
2646 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2648 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
2649 generate the XXBRD instruction.
2651 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
2652 (rs6000_builtin_type_compatible): Treat _Float128 and long double
2653 as being compatible if -mabi=ieeelongdouble.
2654 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
2655 to setup float128 built-ins with hardware support.
2656 (BU_FLOAT128_HW_2): Likewise.
2657 (BU_FLOAT128_HW_3): Likewise.
2658 (BU_FLOAT128_HW_VSX_1): Likewise.
2659 (BU_FLOAT128_HW_VSX_2): Likewise.
2660 (scalar_extract_expq): Change float128 built-in functions to
2661 accommodate having both KFmode and TFmode functions. Use the
2662 KFmode variant as the default.
2663 (scalar_extract_sigq): Likewise.
2664 (scalar_test_neg_qp): Likewise.
2665 (scalar_insert_exp_q): Likewise.
2666 (scalar_insert_exp_qp): Likewise.
2667 (scalar_test_data_class_qp): Likewise.
2668 (sqrtf128_round_to_odd): Delete processing the round to odd
2669 built-in functions as special built-in functions, and define them
2670 as float128 built-ins. Use the KFmode variant as the default.
2671 (truncf128_round_to_odd): Likewise.
2672 (addf128_round_to_odd): Likewise.
2673 (subf128_round_to_odd): Likewise.
2674 (mulf128_round_to_odd): Likewise.
2675 (divf128_round_to_odd): Likewise.
2676 (fmaf128_round_to_odd): Likewise.
2677 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
2678 support for KFmode and TFmode xststdcqp calls.
2679 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
2680 point, switch the built-in handlers for the get/set float128
2681 exponent, get float128 mantissa, float128 test built-ins, and the
2682 float128 round to odd built-in functions. Eliminate creating the
2683 float128 round to odd built-in functions as special built-ins.
2684 (rs6000_init_builtins): Eliminate special creation of the float128
2685 round to odd built-in functions.
2686 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
2687 function insns to support both TFmode and KFmode variants.
2688 (xsxsigqp_<mode>): Likewise.
2689 (xsiexpqpf_<mode>): Likewise.
2690 (xsiexpqp_<mode>): Likewise.
2691 (xststdcqp_<mode>): Likewise.
2692 (xststdcnegqp_<mode>): Likewise.
2693 (xststdcqp_<mode>): Likewise.
2695 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2697 * tree-ssa-threadupdate.c (compute_path_counts): Remove
2698 unused path_in_freq_ptr parameter.
2699 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
2701 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2703 * ipa-inline.c (edge_badness): Dump sreal frequency.
2704 (compute_inlined_call_time): Match natural implementaiton ...
2705 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
2706 forgotten division by CGRAPH_FREQ_BASE.
2708 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2710 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
2711 Solaris 11. Update comment.
2712 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
2714 * config/sol2.h (STARTFILE_SPEC): Likewise.
2715 * configure: Regenerate.
2717 2017-11-14 Carl Love <cel@us.ibm.com>
2719 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
2720 le_ and be_ prefixes to swap* variables. Remove
2721 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
2724 2017-11-14 Jason Merrill <jason@redhat.com>
2726 Support GTY((cache)) on hash_map.
2727 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
2728 (ggc_cache_remove): Override it instead of ggc_mx.
2729 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
2730 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
2731 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
2732 (simple_cache_map_traits): Override maybe_mx.
2733 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
2734 (hash_map): Friend gt_cleare_cache.
2735 (gt_cleare_cache): New.
2736 * tree.h (tree_cache_traits): New hash_map traits class.
2737 (tree_cache_map): New typedef.
2739 2017-11-14 Richard Biener <rguenther@suse.de>
2741 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
2742 paramter and handling.
2743 (cleanup_control_flow_bb): Likewise.
2744 (cleanup_control_flow_pre): New helper performing a DFS walk
2745 to call cleanup_control_flow_bb in PRE order.
2746 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
2747 via cleanup_control_flow_pre.
2749 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
2751 * config/aarch64/aarch64-simd.md
2752 (aarch64_simd_bsl<mode>_internal): Remove DImode.
2753 (*aarch64_simd_bsl<mode>_alt): Likewise.
2754 (aarch64_simd_bsldi_internal): New.
2755 (aarch64_simd_bsldi_alt): Likewise.
2757 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2759 * tracer.c (better_p): Do not compare frequencies.
2760 * reg-stack.c (better_edge): Likewise.
2761 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
2764 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2766 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
2767 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
2768 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
2769 * lto-streamer-in.c (input_function): Use update_max_bb_count.
2770 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
2771 * predict.c (maybe_hot_frequency_p): Inline to ...
2772 (maybe_hot_count_p): ... here; rewrite to counts.
2773 (counts_to_freqs): Rename to ...
2774 (update_max_bb_count): ... this one.
2775 (expensive_function_p): Use counts.
2776 (estimate_bb_frequencies): Update.
2777 (rebuild_frequencies): Update.
2778 * predict.h (counts_to_freqs): Rename to ...
2779 (update_max_bb_count): ... this one.
2780 * profile.c (compute_branch_probabilities): Add debug info
2781 * tree-inline.c (expand_call_inline): Update debug info.
2782 (optimize_inline_calls): Use update_max_bb_count..
2783 (tree_function_versioning): Use update_max_bb_count..
2784 * value-prof.c (gimple_value_profile_transformations):
2785 Do not use update_max_bb_count.
2787 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2789 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
2790 always use frequencies.
2792 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2794 * bb-reorder.c: Remove frequencies from comments.
2795 (better_edge_p): Use profile counts.
2796 (find_traces): Dump profile counts.
2797 (rotate_loop): Use profile counts.
2798 (find_traces_1_round): Likewise.
2799 (connect_better_edge_p): Use counts instead of probabilities for
2801 (copy_bb_p): Drop early check for non-0 frequency.
2802 (sanitize_hot_paths): Update comments.
2804 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2806 * ipa-split.c (struct split_point): Add count.
2807 (consider_split): Do not compute incoming frequency; compute incoming
2808 count and store it to split_point.
2809 (split_function): Set count of the call to split part correctly.
2811 2017-11-13 Carl Love <cel@us.ibm.com>
2813 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
2815 2017-11-13 Tom Tromey <tom@tromey.com>
2817 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
2819 2017-11-13 Carl Love <cel@us.ibm.com>
2821 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
2822 Add support for builtins:
2823 unsigned int vec_first_{,miss}_match_{,or_eos}index,
2824 vector {un,}signed {char,int,short},
2825 vector {un,}signed {char,int,short}) arguments.
2826 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
2827 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
2828 Add BU_P9V_AV_2 expansions for the builtins.
2829 * config/rs6000/altivec.h (vec_first_match_index,
2830 vec_first_mismatch_index, vec_first_match_or_eos_index,
2831 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
2832 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
2833 new extern declaration.
2834 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
2835 * config/rs6000/vsx.md (first_match_index_<mode>,
2836 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
2837 first_mismatch_or_eos_index_<mode>): Add define expand.
2838 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
2839 * doc/extend.texi: Update the built-in documenation file for the new
2842 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
2844 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
2845 into the min/max operations for _Float<N> and _Float<N>X types.
2847 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
2850 * dwarf2out.c (do_eh_frame): New static variable.
2851 (dwarf2out_begin_prologue): Set it.
2852 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
2854 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2856 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
2858 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
2860 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
2863 * debug.h (dwarf2out_do_eh_frame): Declare.
2864 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
2865 (dwarf2out_do_frame): Use it.
2866 (dwarf2out_do_cfi_asm): Likewise.
2867 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
2868 (dwarf2out_assembly_start): Likewise.
2869 (dwarf2out_begin_prologue): Fix comment.
2870 * toplev.c (compile_file): Always call dwarf2out_frame_finish
2871 if the target needs either debug or unwind DWARF2 info.
2872 * lto-opts.c (lto_write_options): Do not save -fexceptions,
2873 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
2874 -ftrapping-math, -ftrapv and -fwrapv.
2876 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2878 * cgraph.c (cgraph_edge::sreal_frequency): New function.
2879 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
2880 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
2881 (estimate_edge_size_and_time): Likewise.
2882 (ipa_merge_fn_summary_after_inlining): Likewise.
2883 * ipa-inline.c (cgraph_freq_base_rec): Remove.
2884 (compute_uninlined_call_time): Use sreal_frequency.
2885 (compute_inlined_call_time): Likewise.
2886 (ipa_inline): Do not initialize cgraph_freq_base_rec.
2887 * profile-count.c: Include sreal.h.
2888 (profile_count::to_sreal_scale): New.
2889 * profile-count.h: Forward declare sreal.
2890 (profile_count::to_sreal_scale): Declare.
2892 2017-11-13 Nathan Sidwell <nathan@acm.org>
2894 * diagnostic.c (maybe_line_and_column): New.
2895 (diagnostic_get_location_text): Use it.
2896 (diagnostic_report_current_module): Likewise.
2897 (test_diagnostic_get_location_text): Add tests.
2899 2017-11-13 Luis Machado <luis.machado@linaro.org>
2901 * doc/md.texi (Specifying processor pipeline description): Fix
2902 incorrect latency for the div instruction example.
2904 2017-11-13 Jakub Jelinek <jakub@redhat.com>
2906 PR tree-optimization/78821
2907 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
2908 that bit_not_p is the same.
2909 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
2910 (split_group): Count precisely bit_not_p bits in each statement.
2911 (invert_op): New function.
2912 (imm_store_chain_info::output_merged_store): Use invert_op to
2913 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
2914 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
2916 2017-11-13 Martin Liska <mliska@suse.cz>
2918 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
2919 (struct source_info): Likewise.
2920 (add_branch_counts): Likewise.
2921 (add_line_counts): Likewise.
2922 (function_summary): Likewise.
2923 (output_intermediate_line): Likewise.
2924 (generate_results): Likewise.
2926 2017-11-13 Martin Liska <mliska@suse.cz>
2928 * gcov.c (struct block_info): Remove typedef for block_t.
2929 (struct line_info): Likewise.
2930 (line_info::has_block): Likewise.
2931 (EXIT_BLOCK): Likewise.
2932 (unblock): Likewise.
2933 (circuit): Likewise.
2934 (get_cycles_count): Likewise.
2935 (process_file): Likewise.
2936 (read_graph_file): Likewise.
2937 (solve_flow_graph): Likewise.
2938 (find_exception_blocks): Likewise.
2939 (add_line_counts): Likewise.
2940 (accumulate_line_info): Likewise.
2941 (output_line_details): Likewise.
2943 2017-11-13 Martin Liska <mliska@suse.cz>
2945 * gcov.c (struct arc_info): Remove typedef for arc_t.
2946 (struct line_info): Likewise.
2947 (add_branch_counts): Likewise.
2948 (output_branch_count): Likewise.
2949 (function_info::~function_info): Likewise.
2950 (circuit): Likewise.
2951 (output_intermediate_line): Likewise.
2952 (read_graph_file): Likewise.
2953 (solve_flow_graph): Likewise.
2954 (find_exception_blocks): Likewise.
2955 (add_line_counts): Likewise.
2956 (accumulate_line_info): Likewise.
2957 (output_line_details): Likewise.
2958 (output_function_details): Likewise.
2960 2017-11-13 Martin Liska <mliska@suse.cz>
2962 * gcov.c (struct function_info): Remove typedef for function_t.
2963 (struct source_info): Likewise.
2964 (source_info::get_functions_at_location): Likewise.
2965 (solve_flow_graph): Likewise.
2966 (find_exception_blocks): Likewise.
2967 (add_line_counts): Likewise.
2968 (output_intermediate_file): Likewise.
2969 (process_file): Likewise.
2970 (generate_results): Likewise.
2971 (release_structures): Likewise.
2972 (read_graph_file): Likewise.
2973 (read_count_file): Likewise.
2974 (accumulate_line_counts): Likewise.
2975 (output_lines): Likewise.
2977 2017-11-13 Martin Liska <mliska@suse.cz>
2979 * gcov.c (function_info::function_info): Remove num_counts
2980 and add vector<gcov_type>.
2981 (function_info::~function_info): Use the vector.
2982 (process_file): Likewise.
2983 (read_graph_file): Likewise.
2984 (read_count_file): Likewise.
2985 (solve_flow_graph): Likewise.
2987 2017-11-13 Martin Liska <mliska@suse.cz>
2989 * gcov.c (function_info::is_artificial): New function.
2990 (process_file): Erase all artificial early.
2991 (generate_results): Skip as all artificial are already
2994 2017-11-13 Martin Liska <mliska@suse.cz>
2996 * gcov.c (read_graph_file): Store to global vector of functions.
2997 (read_count_file): Iterate the vector.
2998 (process_file): Likewise.
2999 (generate_results): Likewise.
3000 (release_structures): Likewise.
3002 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3004 PR tree-optimization/82954
3005 * gimple-ssa-store-merging.c
3006 (imm_store_chain_info::coalesce_immediate_stores): If
3007 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
3009 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
3011 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
3012 Upddate call to ENDIAN_LANE_N.
3013 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
3014 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
3015 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
3016 and use aarch64_endian_lane_rtx.
3017 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
3019 2017-11-12 Tom de Vries <tom@codesourcery.com>
3021 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
3023 2017-11-12 Tom de Vries <tom@codesourcery.com>
3025 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
3027 2017-11-12 Tom de Vries <tom@codesourcery.com>
3029 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
3031 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3032 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3033 * defaults.h (ASM_OUTPUT_LABELREF): Same.
3035 2017-11-11 Martin Sebor <msebor@redhat.com>
3038 * doc/extend.texi (attribute nonstring): Remove spurious argument.
3041 * prefic.c (translate_name): Replace strncpy with memcpy to
3042 avoid -Wstringop-truncation.
3044 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3046 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
3048 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3050 * predict.c (maybe_hot_frequency_p): Do not use cfun.
3052 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3054 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
3057 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
3059 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
3060 enable generating XXBRH if the value is in a vector register.
3061 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
3062 value is in a vector register.
3063 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
3064 register to register bswap64's instead of doing the GPR sequence
3065 used on previous machines.
3066 (bswapdi2_xxbrd): New insn.
3067 (bswapdi2_reg): Disallow on ISA 3.0.
3068 (register to register bswap64 splitter): Do not split the insn on
3069 ISA 3.0 systems that use XXBRD.
3071 2017-11-10 Martin Sebor <msebor@redhat.com>
3074 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
3076 (find_subframework_file): Same.
3078 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3080 * auto-profile.c (afdo_indirect_call): Drop frequency.
3081 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
3082 (cgraph_node::create_edge): Drop frequency argument.
3083 (cgraph_node::create_indirect_edge): Drop frequency argument.
3084 (cgraph_edge::make_speculative): Drop frequency arguments.
3085 (cgraph_edge::resolve_speculation): Do not update frequencies
3086 (cgraph_edge::dump_edge_flags): Do not dump frequency.
3087 (cgraph_node::dump): Check consistency in IPA mode.
3088 (cgraph_edge::maybe_hot_p): Use IPA counter.
3089 (cgraph_edge::verify_count_and_frequency): Rename to ...
3090 (cgraph_edge::verify_count): ... this one; drop frequency checking.
3091 (cgraph_node::verify_node): Update.
3092 * cgraph.h (struct cgraph_edge): Drop frequency.
3093 (cgraph_edge::frequency): New function.
3094 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
3096 (cgraph_edge::rebuild_edges): Likewise.
3097 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
3098 (duplicate_thunk_for_node): Do not pass frequency.
3099 (cgraph_node::create_clone): Scale only counts.
3100 (cgraph_node::create_virtual_clone): Do not pass frequency.
3101 (cgraph_node::create_edge_including_clones): Do not pass frequency.
3102 (cgraph_node::create_version_clone): Do not pass frequency.
3103 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
3104 (cgraph_node::expand_thunk): Do not pass frequency.
3105 (cgraph_node::create_wrapper): Do not pass frequency.
3106 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
3108 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
3109 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
3110 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
3111 (gather_caller_stats): Use frequency function.
3112 (ipcp_cloning_candidate_p): Use frequency function.
3113 (ipcp_propagate_stage): Use frequency function.
3114 (get_info_about_necessary_edges): Use frequency function.
3115 (update_profiling_info): Update only IPA profile.
3116 (update_specialized_profile): Use frequency functoin.
3117 (perhaps_add_new_callers): Update only IPA profile.
3118 * ipa-devirt.c (ipa_devirt): Use IPA profile.
3119 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
3120 (dump_ipa_call_summary): Use frequency function.
3121 (estimate_edge_size_and_time): Use frequency function.
3122 (ipa_merge_fn_summary_after_inlining): Use frequency function.
3123 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
3124 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
3125 (update_noncloned_counts): ... ths one; scale counts only.
3126 (clone_inlined_nodes): Do not scale frequency.
3127 (inline_call): Do not pass frequency.
3128 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
3129 (compute_inlined_call_time): Use IPA profile.
3130 (want_inline_small_function_p): Use IPA profile.
3131 (want_inline_self_recursive_call_p): Use IPA profile.
3132 (edge_badness): Use IPA profile.
3133 (lookup_recursive_calls): Use IPA profile.
3134 (recursive_inlining): Do not pass frequency.
3135 (resolve_noninline_speculation): Do not update frequency.
3136 (inline_small_functions): Collect max of IPA profile.
3137 (dump_overall_stats): Dump IPA porfile.
3138 (dump_inline_stats): Dump IPA porfile.
3139 (ipa_inline): Collect IPA stats.
3140 * ipa-inline.h (clone_inlined_nodes): Update prototype.
3141 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
3142 (ipa_propagate_frequency): Use frequency function.
3143 (ipa_profile): Cleanup.
3144 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
3145 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
3146 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
3147 (input_node): Do not stream frequency.
3148 (input_edge): Do not stream frequency.
3149 (merge_profile_summaries): Scale only IPA profiles.
3150 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
3151 * predict.c (drop_profile): Do not recompute frequency.
3152 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
3153 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
3154 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
3155 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
3156 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
3157 * tree-inline.c (copy_bb): Do not scale frequency.
3158 (expand_call_inline): Do not scale frequency.
3159 (tree_function_versioning): Do not scale frequency.
3160 * ubsan.c (ubsan_create_edge): Do not pass frequency.
3162 2017-11-10 Julia Koval <julia.koval@intel.com>
3164 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
3165 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
3166 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
3167 (_mm256_maskz_gf2p8affine_epi64_epi8)
3168 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
3169 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
3170 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
3171 (__builtin_ia32_vgf2p8affineqb_v32qi)
3172 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
3173 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
3175 2017-11-10 Tamar Christina <tamar.christina@arm.com>
3179 (arm_option_override): Refactor.
3180 (arm_option_reconfigure_globals): New.
3181 (arm_options_perform_arch_sanity_checks): New.
3182 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
3184 (arm_options_perform_arch_sanity_checks): Likewise
3186 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
3188 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
3189 (power9-qpmul): New.
3190 * rs6000/rs6000.md ("type" attr): Add qmul.
3191 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
3192 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
3193 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
3195 2017-11-10 Martin Sebor <msebor@redhat.com>
3198 * builtins.c (compute_objsize): Handle arrays that
3199 compute_builtin_object_size likes to fail for. Make extern.
3200 * builtins.h (compute_objsize): Declare.
3201 (check_strncpy_sizes): New function.
3202 (expand_builtin_strncpy): Call check_strncpy_sizes.
3203 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
3204 -Wstringop-truncation.
3205 (gimple_fold_builtin_strncat): Same.
3206 * gimple.c (gimple_build_call_from_tree): Set call location.
3207 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
3208 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
3209 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
3210 (handle_builtin_strlen): Use strlen_to_stridx.
3211 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
3213 Use strlen_to_stridx.
3214 (pass_strlen::execute): Release strlen_to_stridx.
3215 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
3216 (-Wstringop-truncation): Document new option.
3218 2017-11-10 Martin Liska <mliska@suse.cz>
3220 PR gcov-profile/82702
3221 * gcov.c (main): Handle intermediate files in a different
3223 (get_gcov_intermediate_filename): New function.
3224 (output_gcov_file): Remove support of intermediate files.
3225 (generate_results): Allocate intermediate file.
3226 (release_structures): Clean-up properly fn_end.
3227 (output_intermediate_file): Start iterating with line 1.
3229 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3231 PR tree-optimization/82929
3232 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3233 ops_swapped_p non-static data member.
3234 (store_immediate_info::store_immediate_info): Clear it.
3235 (imm_store_chain_info::coalesce_immediate_stores): If swapping
3236 ops set ops_swapped_p.
3237 (count_multiple_uses): Handle ops_swapped_p.
3239 2017-11-10 Martin Liska <mliska@suse.cz>
3241 * coverage.c (coverage_init): Stream information about
3242 support of has_unexecuted_blocks.
3243 * doc/gcov.texi: Document that.
3244 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
3245 * gcov.c (read_graph_file): Likewise.
3246 (output_line_beginning): Fix a small issue with
3249 2017-11-10 Bin Cheng <bin.cheng@arm.com>
3251 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
3252 reference of trivial component.
3254 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3257 * gimple-ssa-store-merging.c
3258 (pass_store_merging::terminate_all_aliasing_chains): For
3259 gimple_store_p stmts also call refs_output_dependent_p.
3261 PR rtl-optimization/82913
3262 * compare-elim.c (try_merge_compare): Punt if def_insn is not
3265 2017-11-09 Jeff Law <law@redhat.com>
3267 * vr-values.h: New file with vr_values class.
3268 * tree-vrp.c: Include vr-values.h
3269 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
3270 data objects into the vr_values class.
3271 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
3272 (get_value_range): Make it a member function within vr_values class.
3273 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
3274 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
3275 (op_with_constant_singleton_value_range): Likewise.
3276 (extract_range_for_var_from_comparison_expr): Likewise.
3277 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
3278 (extract_range_from_binary_expr): Likewise.
3279 (extract_range_from_unary_expr): Likewise.
3280 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
3281 (check_for_binary_op_overflow, extract_range_basic): Likewise.
3282 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
3283 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
3284 (compare_name_with_value, compare_names): Likewise.
3285 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
3286 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
3287 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
3288 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
3289 (extract_range_from_phi_node): Likewise.
3290 (simplify_truth_ops_using_ranges): Likewise.
3291 (simplify_div_or_mod_using_ranges): Likewise.
3292 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
3293 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
3294 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
3295 (simplify_float_conversion_using_ranges): Likewise.
3296 (simplify_internal_call_using_ranges): Likewise.
3297 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
3298 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
3299 poitner via x_vr_values for calls into gimple folder.
3300 (vrp_initialize_lattice): Make this the vr_values ctor.
3301 (vrp_free_lattice): Make this the vr_values dtor.
3302 (set_vr_value): New function.
3303 (class vrp_prop): Add vr_values data member. Add various member
3304 functions as well as member functions that delegate to vr_values.
3305 (check_array_ref): Make a member function within vrp_prop class.
3306 (search_for_addr_array, vrp_initialize): Likewise.
3307 (vrp_finalize): Likewise. Revamp to avoid direct access to
3308 vr_value, values_propagated, etc.
3309 (check_array_bounds): Extract vrp_prop class instance pointer from
3310 walk info structure. Use it to call member functions.
3311 (check_all_array_refs): Make a member function within vrp_prop class.
3312 Smuggle class instance pointer via walk info structure.
3313 (x_vr_values): New local static.
3314 (vrp_valueize): Use x_vr_values to get class instance.
3315 (vr_valueize_1): Likewise.
3316 (class vrp_folder): Add vr_values data member. Add various member
3317 functions as well as member functions that delegate to vr_values.
3318 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
3319 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
3320 class instance from vr_values. Use it to call member functions.
3321 (vrp_dom_walker): Add vr_values data member.
3322 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
3323 instance via x_vr_values.
3324 (identify_jump_threads): Accept vr_values as argument. Store
3325 it into the walker structure.
3326 (evrp_dom_walker): Add vr_values class data member. Add various
3328 (evrp_dom_walker::try_find_new_range): Use vr_values data
3329 member to access the memory allocator.
3330 (evrp_dom_walker::before_dom_children): Store vr_values class
3331 instance into the vrp_folder class.
3332 (evrp_dom_walker::push_value_range): Rework to avoid direct
3333 access to num_vr_values and vr_value.
3334 (evrp_dom_walker::pop_value_range): Likewise.
3335 (execute_early_vrp): Remove call to vrp_initialize_lattice.
3336 Use vr_values to get to dump_all_value_ranges member function.
3337 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
3338 and simplify_cond_using_ranges_2 via vrp_prop class instance.
3339 Pass vr_values class instance down to identify_jump_threads.
3340 Remove call to vrp_free_lattice.
3341 (debug_all_value_ranges): Remove.
3343 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
3344 (vrp_folder): Likewise.
3346 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
3347 Get it from the existing bitmap instead.
3348 (vrp_intersect_ranges_1): Likewise.
3350 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3352 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3354 (store_immediate_info::store_immediate_info): Add bitnotp argument,
3355 set bit_not_p to it.
3356 (imm_store_chain_info::coalesce_immediate_stores): Break group
3357 if bit_not_p is different.
3358 (count_multiple_uses, split_group,
3359 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
3360 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
3361 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
3362 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
3363 to store_immediate_info ctor.
3365 2017-11-09 Jim Wilson <jimw@sifive.com>
3367 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
3368 (scan_prog_file): Likewise.
3370 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3372 * bb-reorder.c (max_entry_frequency): Remove.
3373 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
3374 connect_traces, push_to_next_round_p): Remove prototypes.
3375 (find_traces_1_round): Use counts only.
3376 (push_to_next_round_p): Likewise.
3377 (find_traces): Likewise.
3378 (rotate_loop): Likewise.
3379 (find_traces_1_round): Likewise.
3380 (connect_traces): Likewise.
3381 (edge_order): Likewise.
3383 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
3385 * config/arm/arm.c (output_return_instruction): Add comments to
3386 indicate requirement for cmse_nonsecure_entry return to account
3387 for the size of clearing instruction output here.
3388 (thumb_exit): Likewise.
3389 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
3390 return in hardfloat mode.
3392 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
3394 * config/rs6000/rs6000.c (machine_function): Add a bool,
3395 "toc_is_wrapped_separately".
3396 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
3397 if it wasn't explicitly set or unset, we are optimizing for speed, and
3398 doing separate shrink-wrapping.
3399 (rs6000_get_separate_components): Enable the TOC component if
3400 saving the TOC register in the prologue.
3401 (rs6000_components_for_bb): Handle the TOC component.
3402 (rs6000_emit_prologue_components): Store the TOC register where needed.
3403 (rs6000_set_handled_components): Mark TOC as handled, if handled.
3404 (rs6000_emit_prologue): Don't save the TOC if that is already done.
3406 2017-11-09 Martin Jambor <mjambor@suse.cz>
3408 * ipa-param-manipulation.c: New file.
3409 * ipa-param-manipulation.h: Likewise.
3410 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
3411 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
3412 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
3413 (ipa_parm_adjustment): Likewise.
3414 (ipa_parm_adjustment_vec): Likewise.
3415 (ipa_get_vector_of_formal_parms): Moved declaration to
3416 ipa-param-manipulation.h.
3417 (ipa_get_vector_of_formal_parm_types): Likewise.
3418 (ipa_modify_formal_parameters): Likewise.
3419 (ipa_modify_call_arguments): Likewise.
3420 (ipa_combine_adjustments): Likewise.
3421 (ipa_dump_param_adjustments): Likewise.
3422 (ipa_modify_expr): Likewise.
3423 (ipa_get_adjustment_candidate): Likewise.
3424 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
3425 ipa-param-manipulation.c.
3426 (ipa_get_vector_of_formal_parm_types): Likewise.
3427 (ipa_modify_formal_parameters): Likewise.
3428 (ipa_modify_call_arguments): Likewise.
3429 (ipa_modify_expr): Likewise.
3430 (get_ssa_base_param): Likewise.
3431 (ipa_get_adjustment_candidate): Likewise.
3432 (index_in_adjustments_multiple_times_p): Likewise.
3433 (ipa_combine_adjustments): Likewise.
3434 (ipa_dump_param_adjustments): Likewise.
3435 * tree-sra.c: Also include ipa-param-manipulation.h
3436 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
3439 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3440 Alan Hayward <alan.hayward@arm.com>
3441 David Sherwood <david.sherwood@arm.com>
3443 * doc/sourcebuild.texi (vect_masked_store): Document.
3445 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3446 Alan Hayward <alan.hayward@arm.com>
3447 David Sherwood <david.sherwood@arm.com>
3449 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
3451 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3452 Alan Hayward <alan.hayward@arm.com>
3453 David Sherwood <david.sherwood@arm.com>
3455 * doc/sourcebuild.texi (vect_variable_length): Document.
3457 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3458 Alan Hayward <alan.hayward@arm.com>
3459 David Sherwood <david.sherwood@arm.com>
3461 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
3463 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3464 Alan Hayward <alan.hayward@arm.com>
3465 David Sherwood <david.sherwood@arm.com>
3467 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
3469 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3470 Alan Hayward <alan.hayward@arm.com>
3471 David Sherwood <david.sherwood@arm.com>
3473 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
3474 previously undocumented selectors.
3475 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
3477 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3479 * doc/rtl.texi (const_vector): Say that elements can be
3480 const_wide_ints too.
3481 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
3482 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
3483 (gen_vec_duplicate): Use it instead of CONSTANT_P.
3484 * optabs.c (expand_vector_broadcast): Likewise.
3486 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3487 Alan Hayward <alan.hayward@arm.com>
3488 David Sherwood <david.sherwood@arm.com>
3490 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
3491 scaled index even if the unscaled address was invalid.
3492 Don't increase the complexity of using a scale in that case.
3494 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3495 Alan Hayward <alan.hayward@arm.com>
3496 David Sherwood <david.sherwood@arm.com>
3498 * doc/rtl.texi: Rewrite the subreg rules so that they partition
3499 the inner register into REGMODE_NATURAL_SIZE bytes rather than
3500 UNITS_PER_WORD bytes.
3501 * emit-rtl.c (validate_subreg): Divide subregs into blocks
3502 based on REGMODE_NATURAL_SIZE of the inner mode.
3503 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
3504 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
3505 * expmed.c (lowpart_bit_field_p): Divide the value up into
3506 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
3507 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
3508 whether something is likely to occupy more than one register.
3510 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3513 * ipa-inline-transform.c (update_noncloned_frequencies): Use
3514 profile_count::adjust_for_ipa_scaling.
3515 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
3516 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
3518 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
3520 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3522 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
3523 (split_group): Add total_orig and total_new arguments, estimate the
3524 number of statements related to the store group without store merging
3525 and with store merging.
3526 (imm_store_chain_info::output_merged_store): Adjust split_group
3527 callers, punt if estimated number of statements with store merging
3528 is not smaller than estimated number of statements without it.
3530 (handled_load): Remove has_single_use checks.
3531 (pass_store_merging::process_store): Likewise.
3533 2017-11-09 Richard Biener <rguenther@suse.de>
3535 PR tree-optimization/82902
3536 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
3538 2017-11-09 Martin Liska <mliska@suse.cz>
3541 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
3544 2017-11-09 Martin Liska <mliska@suse.cz>
3546 PR tree-optimization/82669
3547 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
3549 2017-11-09 Martin Liska <mliska@suse.cz>
3551 PR gcov-profile/48463
3552 * coverage.c (coverage_begin_function): Output also end locus
3553 of a function and information whether the function is
3555 * gcov-dump.c (tag_function): Parse and print the information.
3556 * gcov.c (INCLUDE_MAP): Add include.
3557 (INCLUDE_SET): Likewise.
3558 (struct line_info): Move earlier in the source file because
3559 of vector<line_info> in function_info structure.
3560 (line_info::line_info): Likewise.
3561 (line_info::has_block): Likewise.
3562 (struct source_info): Add new member index.
3563 (source_info::get_functions_at_location): New function.
3564 (function_info::group_line_p): New function.
3565 (output_intermediate_line): New function.
3566 (output_intermediate_file): Use the mentioned function.
3567 (struct function_start): New.
3568 (struct function_start_pair_hash): Likewise.
3569 (process_file): Add code that identifies group functions.
3570 Assign lines either to global or function scope.
3571 (generate_results): Skip artificial functions.
3572 (find_source): Assign index for each source file.
3573 (read_graph_file): Read new flag artificial and end_line.
3574 (add_line_counts): Assign it either to global of function scope.
3575 (accumulate_line_counts): Isolate core of the function to
3576 accumulate_line_info and call it for both function and global
3578 (accumulate_line_info): New function.
3579 (output_line_beginning): Fix GNU coding style.
3580 (print_source_line): New function.
3581 (output_line_details): Likewise.
3582 (output_function_details): Likewise.
3583 (output_lines): Iterate both source (global) scope and function
3585 (struct function_line_start_cmp): New class.
3586 * doc/gcov.texi: Reflect changes in documentation.
3588 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3591 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
3592 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
3593 and similarly for not instead of neg.
3595 2017-11-08 Andi Kleen <ak@linux.intel.com>
3597 * config/i386/i386.opt: Add -mforce-indirect-call.
3598 * config/i386/predicates.md: Check for flag_force_indirect_call.
3599 * doc/invoke.texi: Document -mforce-indirect-call
3601 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
3603 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
3605 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
3606 config/riscv/riscv.c (predict.h): New include.
3607 (riscv_slow_unaligned_access_p): No longer static.
3608 (riscv_block_move_straight): Add require.
3609 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3611 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3614 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
3615 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
3616 nonimmediate_operand predicate for operand 1 instead of
3619 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3621 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
3623 * config/aarch64/constraints.md (Uml): New constraint.
3624 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
3627 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3629 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
3630 of two vec_duplicates into a vec_concat.
3632 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3634 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3635 Simplify vec_merge of vec_duplicate and vec_concat.
3636 * config/aarch64/constraints.md (Utq): New constraint.
3637 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
3640 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3642 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3643 Simplify vec_merge of vec_duplicate and const_vector.
3644 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
3646 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
3647 mode iterator. Update predicate on operand 1 to
3648 handle non-const_vec constants. Delete constraints.
3649 (*aarch64_combinez_be<mode>): Likewise for operand 2.
3651 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3653 PR tree-optimization/78821
3654 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
3656 (store_operand_info::store_operand_info): Initialize it to false.
3657 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
3658 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
3659 store in the group, and if chain_info is non-NULL, to ignore altogether
3661 (compatible_load_p): Fail if bit_not_p does not match.
3662 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
3663 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
3664 (pass_store_merging::process_store): Adjust
3665 terminate_all_aliasing_chains calls to pass NULL in all current spots,
3666 call terminate_all_aliasing_chains newly when adding a store into
3667 a chain with non-NULL chain_info.
3669 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3671 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
3673 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3675 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3677 (aarch64_layout_frame): Initialise emit_frame_chain.
3678 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
3679 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
3681 2017-11-08 Martin Liska <mliska@suse.cz>
3683 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
3684 of gimple_call_internal_p.
3686 2017-11-07 Tom de Vries <tom@codesourcery.com>
3688 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
3690 2017-11-07 Tom de Vries <tom@codesourcery.com>
3692 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3695 2017-11-08 Martin Liska <mliska@suse.cz>
3698 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
3700 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3702 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
3704 2017-11-07 Jakub Jelinek <jakub@redhat.com>
3707 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
3708 (*cmp<mode>_ccz_1): New insn with $k alternative.
3711 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
3712 RTX_COMM_COMPARE as commutative as well.
3713 (ix86_binary_operator_ok): Formatting fix.
3714 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
3715 *<code><mode>3<mask_name><round_saeonly_name>,
3716 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
3717 *<s>mul<mode>3_highpart<mask_name>,
3718 *vec_widen_umult_even_v16si<mask_name>,
3719 *vec_widen_umult_even_v8si<mask_name>,
3720 *vec_widen_umult_even_v4si<mask_name>,
3721 *vec_widen_smult_even_v16si<mask_name>,
3722 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
3723 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
3724 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
3725 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
3726 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
3727 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
3728 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
3729 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
3730 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
3731 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
3732 ix86_binary_operator_ok. Formatting fixes.
3733 (*<plusminus_insn><mode>3<mask_name><round_name>,
3734 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
3737 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
3739 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
3740 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
3741 eq and ne if TARGET_ISEL.
3742 (cmp): New code_iterator.
3743 (UNS, UNSU_, UNSIK): New code_attrs.
3744 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
3745 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
3747 ("eq<mode>3"): ... this.
3748 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
3750 ("ne<mode>3"): ... this.
3752 2017-11-07 Julia Koval <julia.koval@intel.com>
3755 * common/config/i386/i386-common.c
3756 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
3757 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
3758 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
3759 * config/i386/i386.opt: Ditto.
3760 * config/i386/i386.c (ix86_target_string): Ditto.
3761 (ix86_option_override_internal): Ditto.
3762 (ix86_init_mpx_builtins): Move MPX to args2.
3763 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
3764 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3765 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3766 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3767 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3768 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3769 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
3771 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
3774 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
3775 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
3776 (zero-extendsidi peephole2): Remove peephole.
3778 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3781 * stor-layout.c: Include attribs.h.
3782 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
3783 explicit lookup of "aligned" attribute.
3785 2017-11-07 Andrew Waterman <andrew@sifive.com>
3787 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
3788 (riscv_expand_block_move): Likewise.
3789 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
3791 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
3792 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
3793 * config/riscv/riscv.c (riscv_block_move_straight): New function.
3794 (riscv_adjust_block_mem): Likewise.
3795 (riscv_block_move_loop): Likewise.
3796 (riscv_expand_block_move): Likewise.
3797 * config/riscv/riscv.md (movmemsi): New pattern.
3799 2017-11-07 Michael Clark <michaeljclark@mac.com>
3801 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
3802 (MUSL_DYNAMIC_LINKER): Likewise.
3804 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
3806 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
3807 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
3810 2017-11-07 Richard Biener <rguenther@suse.de>
3812 * match.pd: Fix build.
3814 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
3815 Jackson Woodruff <jackson.woodruff@arm.com>
3817 PR tree-optimization/71026
3818 * match.pd: Canonicalize negate in division.
3820 2017-11-07 Sudakshina Das <sudi.das@arm.com>
3823 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
3825 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
3827 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
3828 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
3830 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
3832 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
3833 non-scalar integral types.
3834 * match.pd (negate_expr_p): Handle MINUS_EXPR.
3835 (-(A-B), -(~A)): New transformations.
3837 2017-11-07 Tom de Vries <tom@codesourcery.com>
3839 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
3840 semicolon after "do {} while (0)".
3841 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3842 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3843 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3844 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3845 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3847 2017-11-07 Tom de Vries <tom@codesourcery.com>
3849 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
3850 after "do {} while (0)".
3851 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3852 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3853 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3854 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3855 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3857 2017-11-07 Tom de Vries <tom@codesourcery.com>
3860 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
3862 (arm_rtx_costs_internal): Add missing semicolon after
3863 HANDLE_NARROW_SHIFT_ARITH call.
3865 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
3867 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3868 disable isel if it was not set explicitly.
3870 2017-11-06 James Bowman <james.bowman@ftdichip.com>
3872 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
3873 (add_type_attribute) likewise.
3875 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
3877 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
3878 of struct ix86_frame.
3879 (ix86_initial_elimination_offset): Likewise.
3880 (ix86_expand_split_stack_prologue): Likewise.
3882 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
3884 * tree-vrp.h (enum value_range_type): Update stale comment.
3886 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3887 Alan Hayward <alan.hayward@arm.com>
3888 David Sherwood <david.sherwood@arm.com>
3890 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
3891 (aarch64_expand_vec_perm_const): Take the number of units too.
3892 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
3893 (aarch64_expand_vec_perm_const): Likewise.
3894 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
3895 (vec_perm<mode>): Update accordingly.
3897 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3898 Alan Hayward <alan.hayward@arm.com>
3899 David Sherwood <david.sherwood@arm.com>
3901 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
3902 Take the number of units too.
3903 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
3904 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
3905 but check for a vector mode before rather than after the call.
3906 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
3907 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
3908 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
3909 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
3910 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
3911 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
3912 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
3913 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
3914 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
3915 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
3916 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
3917 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
3918 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
3919 (aarch64_sqdmull2_n<mode>): Update accordingly.
3921 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3922 Alan Hayward <alan.hayward@arm.com>
3923 David Sherwood <david.sherwood@arm.com>
3925 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
3926 the number of units too.
3927 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
3928 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
3929 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
3930 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
3931 (vec_store_lanesxi<mode>): Update accordingly.
3933 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3934 Alan Hayward <alan.hayward@arm.com>
3935 David Sherwood <david.sherwood@arm.com>
3937 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
3938 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
3939 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
3940 of units rather than the mode.
3941 * config/aarch64/iterators.md (nunits): New mode attribute.
3942 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
3943 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
3944 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
3945 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
3946 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
3947 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
3948 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
3949 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
3950 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
3951 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
3952 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3953 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
3954 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
3955 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
3956 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
3957 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
3958 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
3959 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
3960 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
3961 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
3962 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
3963 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
3964 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
3965 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3966 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
3967 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
3968 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3969 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
3970 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
3971 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
3972 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
3973 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
3974 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
3975 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
3976 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
3977 (aarch64_simd_vec_setv2di): Likewise.
3979 2017-11-06 Carl Love <cel@us.ibm.com>
3981 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
3983 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
3984 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
3986 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
3988 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
3989 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
3990 Add power 8 macro expansions.
3991 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
3992 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
3993 power 8 instructions. (VSX_XXBR): Add iterator.
3995 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
3997 * config/arm/arm.md (predicable_short_it): Change default to "no",
3998 improve documentation, remove uses that are identical to the default.
3999 (enabled_for_depr_it): Rename to enabled_for_short_it.
4000 * gcc/config/arm/arm-fixed.md (predicable_short_it):
4001 Remove default uses.
4002 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
4003 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
4004 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
4005 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
4007 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
4010 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
4011 float128 helper macros, which are no longer used after deleting
4012 the old 'q' built-in functions, and moving the round to odd
4013 built-in functions to being special built-in functions.
4014 (BU_FLOAT128_2): Likewise.
4015 (BU_FLOAT128_1_HW): Likewise.
4016 (BU_FLOAT128_2_HW): Likewise.
4017 (BU_FLOAT128_3_HW): Likewise.
4018 (FABSQ): Delete old 'q' built-in functions.
4019 (COPYSIGNQ): Likewise.
4020 (SQRTF128_ODD): Move round to odd built-in functions to be
4021 special built-in functions, so that we can handle
4022 -mabi=ieeelongdouble.
4023 (TRUNCF128_ODD): Likewise.
4024 (ADDF128_ODD): Likewise.
4025 (SUBF128_ODD): Likewise.
4026 (MULF128_ODD): Likewise.
4027 (DIVF128_ODD): Likewise.
4028 (FMAF128_ODD): Likewise.
4029 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
4030 built-in names to 'f128'.
4031 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
4032 old 'q' built-in functions, as the machine independent code for
4033 'f128' built-in functions handles this.
4034 (rs6000_expand_builtin): Add expansion for float128 round to odd
4035 functions, keying off on -mabi=ieeelongdouble of whether to use
4036 the KFmode or TFmode variant.
4037 (rs6000_init_builtins): Initialize the _Float128 round to odd
4039 * doc/extend.texi (PowerPC Built-in Functions): Document the old
4040 _Float128 'q' built-in functions are now mapped into the new
4041 'f128' built-in functions.
4043 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
4045 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
4046 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
4048 2017-11-06 David Malcolm <dmalcolm@redhat.com>
4051 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
4052 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
4053 * toplev.c: Include "ipa-fnsummary.h".
4054 (toplev::finalize): Call ipa_fnsummary_c_finalize.
4056 2017-11-06 Jakub Jelinek <jakub@redhat.com>
4058 PR tree-optimization/82838
4059 * gimple-ssa-store-merging.c
4060 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
4061 on a separate gimple_seq which is then appended to seq.
4063 2017-11-06 Jeff Law <law@redhat.com>
4066 * config/i386/i386.c (PROBE_INTERVAL): Remove.
4067 (get_probe_interval): New functions.
4068 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
4069 (ix86_adjust_stack_and_probe): Likewise.
4070 (output_adjust_stack_and_probe): Likewise.
4071 (ix86_emit_probe_stack_range): Likewise.
4072 (ix86_expand_prologue): Likewise.
4074 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4076 PR tree-optimization/82816
4077 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
4078 if the modes of the two types are the same.
4079 (convert_plusminus_to_widen): Likewise.
4081 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4083 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
4085 (usadv16qi): New define_expand.
4086 (usadv8hi): New define_expand.
4088 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
4091 * ipa-inline-transform.c (update_noncloned_frequencies): Always
4093 (inline_transform): Likewise.
4094 * predict.c (counts_to_freqs): Remove useless conditional.
4095 * profile-count.h (profile_count::apply_scale): Move sanity check.
4096 * tree-inline.c (copy_bb): Always scale.
4097 (copy_cfg_body): Likewise.
4099 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
4102 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
4104 (*cmp_ite0): Add enabled_for_depr_it attribute.
4105 (*cmp_ite1): Likewise.
4107 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
4109 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
4112 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4114 * tree-vrp.c (vrp_int_const_binop): Return true on success and
4115 return the value by pointer.
4116 (extract_range_from_multiplicative_op_1): Update accordingly.
4117 Return as soon as an operation fails.
4119 2017-11-05 Tom de Vries <tom@codesourcery.com>
4122 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
4123 (DEF_SANITIZER_BUILTIN): ... here.
4124 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
4125 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
4127 2017-11-05 Tom de Vries <tom@codesourcery.com>
4130 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
4132 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
4133 ASM_OUTPUT_BEFORE_CASE_LABEL call.
4134 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
4136 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4137 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4138 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4140 2017-11-05 Tom de Vries <tom@codesourcery.com>
4143 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
4146 2017-11-04 Michael Clark <michaeljclark@mac.com>
4148 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
4149 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
4151 (*addsi3_extended): Likewise.
4152 (*addsi3_extended2): Likewise.
4153 (<optab>si3): Likewise.
4154 (<optab>di3): Likewise.
4155 (<optab><mode>3): Likewise.
4156 (<*optabe>si3_internal): Likewise.
4157 (zero_extendqi<SUPERQI:mode>2): Likewise.
4158 (*add<mode>hi3): Likewise.
4159 (*xor<mode>hi3): Likewise.
4160 (<optab>di3): Likewise.
4161 (*<optab>si3_extend): Likewise.
4162 (*sge<u>_<X:mode><GPR:mode>): Likewise.
4163 (*slt<u>_<X:mode><GPR:mode>): Likewise.
4164 (*sle<u>_<X:mode><GPR:mode>): Likewise.
4166 2017-11-04 Andrew Waterman <andrew@sifive.com>
4168 * config/riscv/riscv.c (riscv_option_override): Conditionally set
4169 TARGET_STRICT_ALIGN based upon -mtune argument.
4171 2017-11-04 Andrew Waterman <andrew@sifive.com>
4173 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
4175 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
4177 * config/i386/i386.c (choose_basereg): Use optional scratch
4178 register and add assertion.
4179 (x86_emit_outlined_ms2sysv_save): Use scratch register when
4180 needed, and don't allocate stack.
4181 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
4182 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
4183 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
4185 2017-11-03 Jeff Law <law@redhat.com>
4187 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
4188 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
4189 to probe at the start of a noreturn function.
4191 2017-11-03 Jakub Jelinek <jakub@redhat.com>
4193 PR tree-optimization/78821
4194 * gimple-ssa-store-merging.c: Update the file comment.
4195 (MAX_STORE_ALIAS_CHECKS): Define.
4196 (struct store_operand_info): New type.
4197 (store_operand_info::store_operand_info): New constructor.
4198 (struct store_immediate_info): Add rhs_code and ops data members.
4199 (store_immediate_info::store_immediate_info): Add rhscode, op0r
4200 and op1r arguments to the ctor, initialize corresponding data members.
4201 (struct merged_store_group): Add load_align_base and load_align
4203 (merged_store_group::merged_store_group): Initialize them.
4204 (merged_store_group::do_merge): Update them.
4205 (merged_store_group::apply_stores): Pick the constant for
4206 encode_tree_to_bitpos from one of the two operands, or skip
4207 encode_tree_to_bitpos if neither operand is a constant.
4208 (class pass_store_merging): Add process_store method decl. Remove
4209 bool argument from terminate_all_aliasing_chains method decl.
4210 (pass_store_merging::terminate_all_aliasing_chains): Remove
4211 var_offset_p argument and corresponding handling.
4212 (stmts_may_clobber_ref_p): New function.
4213 (compatible_load_p): New function.
4214 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
4215 if there is overlap and rhs_code is not INTEGER_CST. For
4216 non-overlapping stores terminate group if rhs is not mergeable.
4217 (get_alias_type_for_stmts): Change first argument from
4218 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
4219 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
4220 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
4222 (get_location_for_stmts): Change first argument from
4223 auto_vec<gimple *> & to vec<gimple *> &.
4224 (struct split_store): Remove orig_stmts data member, add orig_stores.
4225 (split_store::split_store): Create orig_stores rather than orig_stmts.
4226 (find_constituent_stmts): Renamed to ...
4227 (find_constituent_stores): ... this. Change second argument from
4228 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
4229 to info structures rather than the statements.
4230 (split_group): Rename ALLOW_UNALIGNED argument to
4231 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
4232 it. Adjust find_constituent_stores caller.
4233 (imm_store_chain_info::output_merged_store): Handle rhs_code other
4234 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
4235 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
4236 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
4237 (mem_valid_for_store_merging): New function.
4238 (handled_load): New function.
4239 (pass_store_merging::process_store): New method.
4240 (pass_store_merging::execute): Use process_store method. Adjust
4241 terminate_all_aliasing_chains caller.
4243 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4245 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4246 Return true for more constants, symbols and label references.
4247 (aarch64_valid_floating_const): Remove unused function.
4249 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4252 * config/aarch64/aarch64.c (aarch64_layout_frame):
4253 Undo forcing of LR at bottom of frame.
4255 2017-11-03 Jeff Law <law@redhat.com>
4258 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
4259 for int_registers_saved.
4261 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
4262 extracted from tree-ssa-dom.c.
4263 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
4264 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
4265 (record_equivalences_from_incoming_edge): Add additional argument
4266 to single_pred_edge_ignoring_loop_edges call.
4267 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
4268 (uncprop_dom_walker::before_dom_children): Add additional argument
4269 to single_pred_edge_ignoring_loop_edges call.
4270 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
4271 single_pred_edge_ignoring_loop_edges rather than open coding.
4272 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
4274 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
4276 * match.pd (-(-A)): Rewrite.
4278 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4280 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
4281 (rs6000_emit_int_cmove): New declaration.
4282 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
4283 (rs6000_emit_sISEL): Delete.
4284 (rs6000_emit_int_cmove): Make non-static.
4285 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
4286 instead of rs6000_emit_sISEL.
4288 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
4290 * asan.c (create_cond_insert_point): Maintain profile.
4291 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
4293 * basic-block.h (struct basic_block_def): Remove frequency.
4294 (EDGE_FREQUENCY): Use to_frequency
4295 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
4297 (find_traces): Update to use to_frequency.
4298 (find_traces_1_round): Likewise; use only IPA counts.
4299 (bb_to_key): Likewise.
4300 (connect_traces): Use IPA counts only.
4301 (copy_bb_p): Update to use to_frequency.
4302 (fix_up_crossing_landing_pad): Likewise.
4303 (sanitize_hot_paths): Likewise.
4304 * bt-load.c (basic_block_freq): Likewise.
4305 * cfg.c (init_flow): Set count_max to uninitialized.
4306 (check_bb_profile): Remove frequencies; check counts.
4307 (dump_bb_info): Do not dump frequencies.
4308 (update_bb_profile_for_threading): Update counts only.
4309 (scale_bbs_frequencies_int): Likewise.
4310 (MAX_SAFE_MULTIPLIER): Remove.
4311 (scale_bbs_frequencies_gcov_type): Update counts only.
4312 (scale_bbs_frequencies_profile_count): Update counts only.
4313 (scale_bbs_frequencies): Update counts only.
4314 * cfg.h (struct control_flow_graph): Add count-max.
4315 (update_bb_profile_for_threading): Update prototype.
4316 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
4317 (find_many_sub_basic_blocks): Likewise.
4318 * cfgcleanup.c (try_forward_edges): Likewise.
4319 (try_crossjump_to_edge): Likewise.
4320 * cfgexpand.c (expand_gimple_cond): Likewise.
4321 (expand_gimple_tailcall): Likewise.
4322 (construct_init_block): Likewise.
4323 (construct_exit_block): Likewise.
4324 * cfghooks.c (verify_flow_info): Check consistency of counts.
4325 (dump_bb_for_graph): Do not dump frequencies.
4326 (split_block_1): Do not update frequencies.
4327 (split_edge): Do not update frequencies.
4328 (make_forwarder_block): Do not update frequencies.
4329 (duplicate_block): Do not update frequencies.
4330 (account_profile_record): Do not update frequencies.
4331 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
4332 for global heuristics.
4333 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
4334 (expected_loop_iterations_unbounded): Use counts only.
4335 * cfgloopmanip.c (scale_loop_profile): Simplify.
4336 (create_empty_loop_on_edge): Simplify
4338 (duplicate_loop_to_header_edge): Simplify
4339 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
4340 (update_br_prob_note): Take care of removing note when profile
4342 (relink_block_chain): Do not dump frequency.
4343 (rtl_account_profile_record): Use to_frequency.
4344 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
4345 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
4346 (cgraph_update_edges_for_call_stmt_node): Likewise.
4347 (cgraph_edge::verify_count_and_frequency): Update.
4348 (cgraph_node::verify_node): Temporarily disable frequency verification.
4349 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
4350 to_cgraph_frequency.
4351 (cgraph_edge::rebuild_edges): Convert to ipa counts.
4352 * cgraphunit.c (init_lowered_empty_function): Do not initialize
4354 (cgraph_node::expand_thunk): Update profile.
4355 * except.c (dw2_build_landing_pads): Do not update frequency.
4356 * final.c (compute_alignments): Use to_frequency.
4357 (dump_basic_block_info): Do not dump frequency.
4358 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
4359 (dump_gimple_bb_header): Do not dump frequency.
4360 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
4362 * gimple-streamer-in.c (input_bb): Do not stream frequency.
4363 * gimple-streamer-out.c (output_bb): Do not stream frequency.
4364 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
4365 (init_before_recovery): Do not update frequency.
4366 (sched_create_recovery_edges): Do not update frequency.
4367 * hsa-gen.c (convert_switch_statements): Do not update frequency.
4368 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
4369 (ipa_cp_c_finalize): Set max_count to uninitialized.
4370 * ipa-fnsummary.c (get_minimal_bb): Use counts.
4371 (param_change_prob): Use counts.
4372 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
4374 * ipa-split.c (consider_split): Use to_frequency.
4375 (split_function): Use to_frequency.
4376 * ira-build.c (loop_compare_func): Likewise.
4377 (mark_loops_for_removal): Likewise.
4378 (mark_all_loops_for_removal): Likewise.
4379 * loop-doloop.c (doloop_modify): Do not update frequency.
4380 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
4382 * lto-streamer-in.c (input_function): Update count_max.
4383 * omp-expand.c (expand_omp_taskreg): Update count_max.
4384 * omp-simd-clone.c (simd_clone_adjust): Update profile.
4385 * predict.c (maybe_hot_frequency_p): Use to_frequency.
4386 (maybe_hot_count_p): Use ipa counts only.
4387 (maybe_hot_bb_p): Simplify.
4388 (maybe_hot_edge_p): Simplify.
4389 (probably_never_executed): Do not take frequency argument.
4390 (probably_never_executed_bb_p): Do not pass frequency.
4391 (probably_never_executed_edge_p): Likewise.
4392 (combine_predictions_for_bb): Check that profile is nonzero.
4393 (propagate_freq): Do not set frequency.
4394 (drop_profile): Simplify.
4395 (counts_to_freqs): Simplify.
4396 (expensive_function_p): Use to_frequency.
4397 (propagate_unlikely_bbs_forward): Simplify.
4398 (determine_unlikely_bbs): Simplify.
4399 (estimate_bb_frequencies): Add hack to silence graphite issues.
4400 (compute_function_frequency): Use ipa counts.
4401 (pass_profile::execute): Update.
4402 (rebuild_frequencies): Use counts only.
4403 (force_edge_cold): Use counts only.
4404 * profile-count.c (profile_count::dump): Dump new count types.
4405 (profile_count::differs_from_p): Check compatiblity.
4406 (profile_count::to_frequency): New function.
4407 (profile_count::to_cgraph_frequency): New function.
4408 * profile-count.h (struct function): Declare.
4409 (enum profile_quality): Add profile_guessed_local and
4410 profile_guessed_global0.
4411 (class profile_proability): Decrease number of bits to 29;
4412 update from_reg_br_prob_note and to_reg_br_prob_note.
4413 (class profile_count: Update comment; decrease number of bits
4414 to 61. Check compatibility.
4415 (profile_count::compatible_p): New private member function.
4416 (profile_count::ipa_p): New member function.
4417 (profile_count::operator<): Handle global zero correctly.
4418 (profile_count::operator>): Handle global zero correctly.
4419 (profile_count::operator<=): Handle global zero correctly.
4420 (profile_count::operator>=): Handle global zero correctly.
4421 (profile_count::nonzero_p): New member function.
4422 (profile_count::force_nonzero): New member function.
4423 (profile_count::max): New member function.
4424 (profile_count::apply_scale): Handle IPA scalling.
4425 (profile_count::guessed_local): New member function.
4426 (profile_count::global0): New member function.
4427 (profile_count::ipa): New member function.
4428 (profile_count::to_frequency): Declare.
4429 (profile_count::to_cgraph_frequency): Declare.
4430 * profile.c (OVERLAP_BASE): Delete.
4431 (compute_frequency_overlap): Delete.
4432 (compute_branch_probabilities): Do not use compute_frequency_overlap.
4433 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
4434 * sched-ebb.c (rank): Use counts only.
4435 * shrink-wrap.c (handle_simple_exit): Use counts only.
4436 (try_shrink_wrapping): Use counts only.
4437 (place_prologue_for_one_component): Use counts only.
4438 * tracer.c (find_best_predecessor): Use to_frequency.
4439 (find_trace): Use to_frequency.
4440 (tail_duplicate): Use to_frequency.
4441 * trans-mem.c (expand_transaction): Do not update frequency.
4442 * tree-call-cdce.c: Do not update frequency.
4443 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
4444 (gimple_merge_blocks): Likewise.
4445 (gimple_split_edge): Likewise.
4446 (gimple_duplicate_sese_region): Likewise.
4447 (gimple_duplicate_sese_tail): Likewise.
4448 (move_sese_region_to_fn): Likewise.
4449 (gimple_account_profile_record): Likewise.
4450 (insert_cond_bb): Likewise.
4451 * tree-complex.c (expand_complex_div_wide): Likewise.
4452 * tree-eh.c (lower_resx): Update profile.
4453 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
4455 (initialize_cfun): Do not initialize frequencies
4456 (freqs_to_counts): Delete.
4457 (copy_cfg_body): Ignore count parameter.
4458 (copy_body): Update.
4459 (expand_call_inline): Update count_max.
4460 (optimize_inline_calls): Update count_max.
4461 (tree_function_versioning): Update count_max.
4462 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
4463 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
4465 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
4466 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
4467 (try_peel_loop): Likewise.
4468 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
4470 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
4471 (tree_transform_and_unroll_loop): Do not use frequencies
4472 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
4473 Use reliable prediction only.
4474 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
4475 * tree-ssa-sink.c (select_best_block): Use to_frequency.
4476 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
4477 probability scaling.
4478 * tree-ssa-threadupdate.c (create_block_for_threading): Do
4479 not update frequency
4480 (any_remaining_duplicated_blocks): Likewise.
4481 (update_profile): Likewise.
4482 (estimated_freqs_path): Delete.
4483 (freqs_to_counts_path): Delete.
4484 (clear_counts_path): Delete.
4485 (ssa_fix_duplicate_block_edges): Likewise.
4486 (duplicate_thread_path): Likewise.
4487 * tree-switch-conversion.c (gen_inbound_check): Use counts.
4488 * tree-tailcall.c (decrease_profile): Do not update frequency.
4489 (eliminate_tail_call): Likewise.
4490 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
4491 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
4492 (optimize_mask_stores): Likewise.
4493 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4494 * ubsan.c (ubsan_expand_null_ifn): Update profile.
4495 (ubsan_expand_ptr_ifn): Update profile.
4496 * value-prof.c (gimple_ic): Simplify.
4497 * value-prof.h (gimple_ic): Update prototype.
4498 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
4499 * ipa-inline.c (compute_uninlined_call_time): Be sure that
4501 (want_inline_self_recursive_call_p): Likewise.
4502 (resolve_noninline_speculation): Only cummulate defined counts.
4503 (inline_small_functions): Use nonzero_p.
4504 (ipa_inline): Do not access freed node.
4506 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4508 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4509 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
4511 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
4513 * config/riscv/riscv.c (riscv_legitimize_move): Handle
4514 non-legitimate address.
4516 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4518 * config/rs6000/rs6000.md (*lt0_disi): Delete.
4519 (*lt0_<mode>di, *lt0_<mode>si): New.
4521 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4523 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
4524 TARGET_PAIRED_FLOAT.
4526 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
4527 Jim Wilson <jim.wilson@linaro.org>
4529 * config/aarch64/aarch64-cores.def (saphira): New CPU.
4530 * config/aarch64/aarch64-tune.md: Regenerated.
4531 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
4532 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
4534 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
4536 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
4537 (arc_expand_prologue): Restore blink for millicode.
4538 * config/arc/linux.h (LINK_EH_SPEC): Defined.
4540 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
4543 * config/i386/i386.c (ix86_vector_duplicate_value): Use
4544 gen_vec_duplicate after forcing the scalar into a register.
4546 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
4548 * combine (try_combine): Print the insns input to try_combine to the
4551 2017-11-02 Steve Ellcey <sellcey@cavium.com>
4554 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
4555 Remove second argument from aarch64_process_target_attr call.
4556 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
4558 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
4560 (aarch64_handle_attr_arch): Remove second argument.
4561 (aarch64_handle_attr_cpu): Ditto.
4562 (aarch64_handle_attr_tune): Ditto.
4563 (aarch64_handle_attr_isa_flags): Ditto.
4564 (aarch64_process_one_target_attr): Ditto.
4565 (aarch64_process_target_attr): Ditto.
4566 (aarch64_option_valid_attribute_p): Remove second argument.
4567 on aarch64_process_target_attr call.
4569 2017-11-02 David Malcolm <dmalcolm@redhat.com>
4571 * diagnostic.c: Include "selftest-diagnostic.h".
4572 (selftest::assert_location_text): New function.
4573 (selftest::test_diagnostic_get_location_text): New function.
4574 (selftest::diagnostic_c_tests): Call it.
4576 2017-11-02 David Malcolm <dmalcolm@redhat.com>
4578 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
4579 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
4580 (class selftest::test_diagnostic_context): Move to...
4581 * selftest-diagnostic.c: New file.
4582 * selftest-diagnostic.h: New file.
4584 2017-11-02 James Bowman <james.bowman@ftdichip.com>
4586 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
4587 offset range for FT32B.
4588 * config/ft32/ft32.h: option "mcompress" enables relaxation.
4589 * config/ft32/ft32.md: Add TARGET_NOPM.
4590 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
4591 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
4593 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
4595 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
4597 2017-11-02 Jeff Law <law@redhat.com>
4599 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
4600 virtual keyword on FINAL OVERRIDE members.
4602 * tree-ssa-propagate.h (ssa_propagation_engine): Group
4603 virtuals together. Add virtual destructor.
4604 (substitute_and_fold_engine): Similarly.
4606 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
4608 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
4610 2017-11-02 Richard Biener <rguenther@suse.de>
4612 PR tree-optimization/82795
4613 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
4615 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4617 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
4619 (gcc_GAS_CHECK_FEATURE): Remove.
4620 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
4621 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
4622 * configure: Regenerate.
4624 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
4626 * config/arc/arc.c (hwloop_optimize): Account for empty
4629 2017-11-02 Richard Biener <rguenther@suse.de>
4632 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
4633 Truncate ARRAY_REF index and element size.
4635 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4637 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
4639 2017-11-01 Jeff Law <law@redhat.com>
4641 * tree-ssa-ccp.c (ccp_folder): New class derived from
4642 substitute_and_fold_engine.
4643 (ccp_folder::get_value): New member function.
4644 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
4645 (ccp_fold_stmt): Remove prototype.
4646 (ccp_finalize): Call substitute_and_fold from the ccp_class.
4647 * tree-ssa-copy.c (copy_folder): New class derived from
4648 substitute_and_fold_engine.
4649 (copy_folder::get_value): Renamed from get_value.
4650 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
4651 * tree-vrp.c (vrp_folder): New class derived from
4652 substitute_and_fold_engine.
4653 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
4654 (vrp_folder::get_value): New member function.
4655 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
4656 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
4657 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
4658 provide a class interface to folder/substitute routines.
4659 (ssa_prop_fold_stmt_fn): Remove typedef.
4660 (ssa_prop_get_value_fn): Likewise.
4661 (subsitute_and_fold): Remove prototype.
4662 (replace_uses_in): Likewise.
4663 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
4664 Renamed from replace_uses_in. Call the virtual member function
4665 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
4666 (substitute_and_fold_dom_walker): Remove initialization of
4667 data member entries for calbacks. Add substitute_and_fold_engine
4668 member and initialize it.
4669 (substitute_and_fold_dom_walker::before_dom_children0: Use the
4670 member functions for get_value, replace_phi_args_in c
4671 replace_uses_in, and fold_stmt calls.
4672 (substitute_and_fold_engine::substitute_and_fold): Renamed from
4673 substitute_and_fold. Remove assert. Update ctor call.
4675 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
4676 (ssa_prop_visit_phi_fn): Likewise.
4677 (class ssa_propagation_engine): New class to provide an interface
4679 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
4681 (ssa_prop_visit_phi): Likewise.
4682 (ssa_propagation_engine::simulate_stmt): Moved into class.
4683 Call visit_phi/visit_stmt from the class rather than via
4684 file scoped static variables.
4685 (ssa_propagation_engine::simulate_block): Moved into class.
4686 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
4687 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
4688 set file scoped statics for the visit_stmt/visit_phi callbacks.
4689 * tree-complex.c (complex_propagate): New class derived from
4690 ssa_propagation_engine.
4691 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
4692 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
4693 (tree_lower_complex): Call ssa_propagate via the complex_propagate
4695 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
4696 ssa_propagation_engine.
4697 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
4698 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
4699 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
4700 * tree-ssa-copy.c (copy_prop): New class derived from
4701 ssa_propagation_engine.
4702 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
4703 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
4704 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
4705 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
4706 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
4707 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
4708 (execute_vrp): Call ssa_propagate from the vrp_prop class.
4710 2017-11-01 Jakub Jelinek <jakub@redhat.com>
4712 PR rtl-optimization/82778
4713 PR rtl-optimization/82597
4714 * compare-elim.c (struct comparison): Add in_a_setter field.
4715 (find_comparison_dom_walker::before_dom_children): Remove killed
4716 bitmap and df_simulate_find_defs call, instead walk the defs.
4717 Compute last_setter and initialize in_a_setter. Merge definitions
4718 with first initialization for a few variables.
4719 (try_validate_parallel): Use insn_invalid_p instead of
4720 recog_memoized. Return insn rather than just the pattern.
4721 (try_merge_compare): Fix up comment. Don't uselessly test if
4722 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
4724 (execute_compare_elim_after_reload): Remove df_chain_add_problem
4727 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4728 Alan Hayward <alan.hayward@arm.com>
4729 David Sherwood <david.sherwood@arm.com>
4731 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
4732 aarch64_hard_regno_nregs to get the number of registers
4735 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4736 Alan Hayward <alan.hayward@arm.com>
4737 David Sherwood <david.sherwood@arm.com>
4739 * config/aarch64/constraints.md (Upl): Rename to...
4741 * config/aarch64/aarch64.md
4742 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
4745 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4746 Alan Hayward <alan.hayward@arm.com>
4747 David Sherwood <david.sherwood@arm.com>
4749 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
4750 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
4753 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4754 Alan Hayward <alan.hayward@arm.com>
4755 David Sherwood <david.sherwood@arm.com>
4757 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
4758 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
4759 (aarch64_evpc_dup): Generate rtl direcly, rather than using
4761 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
4762 of a single element.
4763 * config/aarch64/iterators.md: Add a comment above the permute
4764 unspecs to say that they are generated directly by
4765 aarch64_expand_vec_perm_const.
4766 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
4768 2017-11-01 Nathan Sidwell <nathan@acm.org>
4770 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
4772 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4774 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
4775 and medany code models, and describe what they do.
4777 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4779 Revert accidental duplicate:
4781 * combine.c (can_change_dest_mode): Reject changes in
4782 REGMODE_NATURAL_SIZE.
4784 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
4786 PR rtl-optimization/64682
4787 PR rtl-optimization/69567
4788 PR rtl-optimization/69737
4789 PR rtl-optimization/82683
4790 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
4791 register mentioned in the note, drop the note, unless it came from I3,
4792 in which case it should go to I3 again.
4794 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4796 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
4797 and return false if not.
4798 (clear_bytes_written_by, live_bytes_read): Update accordingly.
4800 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4802 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
4803 range is known to be empty.
4805 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4806 Alan Hayward <alan.hayward@arm.com>
4807 David Sherwood <david.sherwood@arm.com>
4809 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
4810 and CONST_VECTOR_NUNITS instead of computing the number of units from
4811 the byte sizes of the vector and element.
4812 (simplify_binary_operation_1): Likewise.
4813 (simplify_const_binary_operation): Likewise.
4814 (simplify_ternary_operation): Likewise.
4816 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4817 Alan Hayward <alan.hayward@arm.com>
4818 David Sherwood <david.sherwood@arm.com>
4820 * var-tracking.c (INT_MEM_OFFSET): Replace with...
4821 (int_mem_offset): ...this new function.
4822 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
4823 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
4824 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
4827 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4828 Alan Hayward <alan.hayward@arm.com>
4829 David Sherwood <david.sherwood@arm.com>
4831 * lower-subreg.c (interesting_mode_p): New function.
4832 (compute_costs, find_decomposable_subregs, decompose_register)
4833 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
4834 (resolve_clobber, dump_choices): Use it.
4836 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4837 Alan Hayward <alan.hayward@arm.com>
4838 David Sherwood <david.sherwood@arm.com>
4840 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
4842 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4843 Alan Hayward <alan.hayward@arm.com>
4844 David Sherwood <david.sherwood@arm.com>
4846 * alias.c (find_base_value, find_base_term): Only process integer
4847 truncations. Check the precision rather than the size.
4849 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4850 Alan Hayward <alan.hayward@arm.com>
4851 David Sherwood <david.sherwood@arm.com>
4853 * machmode.h (is_narrower_int_mode): New function
4854 * optabs.c (expand_float, expand_fix): Use it.
4855 * dwarf2out.c (rotate_loc_descriptor): Likewise.
4857 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4858 Alan Hayward <alan.hayward@arm.com>
4859 David Sherwood <david.sherwood@arm.com>
4861 * rtl.h (narrower_subreg_mode): New function.
4862 * ira-color.c (update_costs_from_allocno): Use it.
4864 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4865 Alan Hayward <alan.hayward@arm.com>
4866 David Sherwood <david.sherwood@arm.com>
4868 * optabs-query.h (convert_optab_p): New function, split out from...
4869 (convert_optab_handler): ...here.
4870 (widening_optab_handler): Delete.
4871 (find_widening_optab_handler): Remove permit_non_widening parameter.
4872 (find_widening_optab_handler_and_mode): Likewise. Provide an
4873 override that operates on mode class wrappers.
4874 * optabs-query.c (widening_optab_handler): Delete.
4875 (find_widening_optab_handler_and_mode): Remove permit_non_widening
4876 parameter. Assert that the two modes are the same class and that
4877 the "from" mode is narrower than the "to" mode. Use
4878 convert_optab_handler instead of widening_optab_handler.
4879 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
4880 instead of widening_optab_handler.
4881 * expr.c (expand_expr_real_2): Update calls to
4882 find_widening_optab_handler.
4883 * optabs.c (expand_widen_pattern_expr): Likewise.
4884 (expand_binop_directly): Take the insn_code as a parameter.
4885 (expand_binop): Only call find_widening_optab_handler for
4886 conversion optabs; use optab_handler otherwise. Update calls
4887 to find_widening_optab_handler and expand_binop_directly.
4888 Use convert_optab_handler instead of widening_optab_handler.
4889 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
4890 find_widening_optab_handler and use scalar_mode rather than
4892 (convert_plusminus_to_widen): Likewise.
4894 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4895 Alan Hayward <alan.hayward@arm.com>
4896 David Sherwood <david.sherwood@arm.com>
4898 * machmode.h (fixed_size_mode): New class.
4899 * rtl.h (get_pool_mode): Return fixed_size_mode.
4900 * gengtype.c (main): Add fixed_size_mode.
4901 * target.def (get_raw_result_mode): Return a fixed_size_mode.
4902 (get_raw_arg_mode): Likewise.
4903 * doc/tm.texi: Regenerate.
4904 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
4905 * targhooks.c (default_get_reg_raw_mode): Likewise.
4906 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
4907 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
4908 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
4909 (msp430_get_raw_result_mode): Likewise.
4910 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
4911 * dbxout.c (dbxout_parms): Require fixed-size modes.
4912 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
4913 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
4914 * omp-low.c (lower_oacc_reductions): Likewise.
4915 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
4916 (simplify_subreg): Update accordingly.
4917 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
4918 (force_const_mem): Update accordingly. Return NULL_RTX for modes
4919 that aren't fixed-size.
4920 (get_pool_mode): Return a fixed_size_mode.
4921 (output_constant_pool_2): Take a fixed_size_mode.
4923 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4924 Alan Hayward <alan.hayward@arm.com>
4925 David Sherwood <david.sherwood@arm.com>
4927 * doc/rtl.texi (vec_series): Document.
4928 (const): Say that the operand can be a vec_series.
4929 * rtl.def (VEC_SERIES): New rtx code.
4930 * rtl.h (const_vec_series_p_1): Declare.
4931 (const_vec_series_p): New function.
4932 * emit-rtl.h (gen_const_vec_series): Declare.
4933 (gen_vec_series): Likewise.
4934 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
4935 (gen_vec_series): Likewise.
4936 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
4937 * simplify-rtx.c (simplify_unary_operation): Handle negations
4939 (simplify_binary_operation_series): New function.
4940 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
4941 (test_vector_ops_series): New function.
4942 (test_vector_ops): Call it.
4943 * config/powerpcspe/altivec.md (altivec_lvsl): Use
4944 gen_const_vec_series.
4945 (altivec_lvsr): Likewise.
4946 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
4948 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4949 Alan Hayward <alan.hayward@arm.com>
4950 David Sherwood <david.sherwood@arm.com>
4952 * doc/rtl.texi (const): Update description of address constants.
4953 Say that vector constants are allowed too.
4954 * common.md (E, F): Use CONSTANT_P instead of checking for
4956 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
4957 checking for CONST_VECTOR.
4958 * expmed.c (make_tree): Use build_vector_from_val for a CONST
4960 * expr.c (expand_expr_real_2): Check for vector modes instead
4961 of checking for CONST_VECTOR.
4962 * rtl.h (const_vec_p): New function.
4963 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
4964 (unwrap_const_vec_duplicate): Handle them here too.
4966 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4967 David Malcolm <dmalcolm@redhat.com>
4968 Alan Hayward <alan.hayward@arm.com>
4969 David Sherwood <david.sherwood@arm.com>
4971 * rtl.h (vec_duplicate_p): New function.
4972 * selftest-rtl.c (assert_rtx_eq_at): New function.
4973 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
4974 (assert_rtx_eq_at): Declare.
4975 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
4976 * selftest-run-tests.c (selftest::run_tests): Call it.
4977 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
4978 (simplify_unary_operation_1): Recursively handle vector duplicates.
4979 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
4981 (simplify_subreg): Handle subregs of vector duplicates.
4982 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
4983 (selftest::simplify_rtx_c_tests): New functions.
4985 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4986 Alan Hayward <alan.hayward@arm.com>
4987 David Sherwood <david.sherwood@arm.com>
4989 * emit-rtl.h (gen_const_vec_duplicate): Declare.
4990 (gen_vec_duplicate): Likewise.
4991 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
4993 (gen_const_vector): ...here.
4994 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
4995 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
4996 whose elements are all equal.
4997 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
4998 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
4999 (simplify_relational_operation): Likewise.
5000 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
5002 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
5003 (aarch64_expand_vector_init): Likewise.
5004 * config/arm/arm.c (neon_vdup_constant): Likewise.
5005 (neon_expand_vector_init): Likewise.
5006 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
5007 (arm_block_set_unaligned_vect): Likewise.
5008 (arm_block_set_aligned_vect): Likewise.
5009 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
5010 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
5011 (expand_vec_perm_even_odd_pack): Likewise.
5012 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
5013 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
5014 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
5015 gen_const_vec_duplicate.
5016 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
5017 * config/mips/mips.c (mips_gen_const_int_vector): Use
5018 gen_const_vec_duplicate.
5019 (mips_expand_vector_init): Use CONST0_RTX.
5020 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
5021 (define_split): Use gen_const_vec_duplicate.
5022 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
5023 (define_split): Use gen_const_vec_duplicate.
5024 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
5025 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
5026 * config/spu/spu.c (spu_const): Likewise.
5028 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5029 Alan Hayward <alan.hayward@arm.com>
5030 David Sherwood <david.sherwood@arm.com>
5032 * combine.c (can_change_dest_mode): Reject changes in
5033 REGMODE_NATURAL_SIZE.
5035 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
5037 * configure.ac (--enable-libssp): New.
5038 (gcc_cv_libc_provides_ssp): Check for explicit setting before
5039 trying to determine target-specific default. Adjust indentation.
5040 * configure: Regenerated.
5041 * doc/install.texi (Configuration): Expand --disable-libssp
5044 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
5046 config/i386/i386.c (ix86_expand_epilogue): Correct stack
5049 2017-10-31 Martin Jambor <mjambor@suse.cz>
5052 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
5054 2017-10-31 David Malcolm <dmalcolm@redhat.com>
5056 * auto-profile.c (autofdo_source_profile::read): Use
5057 UNKNOWN_LOCATION rather than 0.
5058 * diagnostic-core.h (warning_at_rich_loc): Rename to...
5059 (warning_at): ...this overload.
5060 (warning_at_rich_loc_n): Rename to...
5061 (warning_n): ...this overload.
5062 (error_at_rich_loc): Rename to...
5063 (error_at): ...this overload.
5064 (pedwarn_at_rich_loc): Rename to...
5065 (pedwarn): ...this overload.
5066 (permerror_at_rich_loc): Rename to...
5067 (permerror): ...this overload.
5068 (inform_at_rich_loc): Rename to...
5069 (inform): ...this overload.
5070 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
5071 (diagnostic_n_impl_richloc): Rename to...
5072 (diagnostic_n_impl): ...this rich_location *-based decl.
5073 (inform_at_rich_loc): Rename to...
5074 (inform): ...this, and add an assertion.
5075 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
5076 (warning_at_rich_loc): Rename to...
5077 (warning_at): ...this, and add an assertion.
5078 (warning_at_rich_loc_n): Rename to...
5079 (warning_n): ...this, and add an assertion.
5080 (warning_n): Update location_t-based implementation for removal of
5081 location_t-based diagnostic_n_impl.
5082 (pedwarn_at_rich_loc): Rename to...
5083 (pedwarn): ...this, and add an assertion.
5084 (permerror_at_rich_loc): Rename to...
5085 (permerror): ...this, and add an assertion.
5086 (error_n): Update for removal of location_t-based diagnostic_n_impl.
5087 (error_at_rich_loc): Rename to...
5088 (error_at): ...this, and add an assertion.
5089 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
5090 (driver::do_spec_on_infiles): Likewise.
5091 * substring-locations.c (format_warning_va): Update for renaming
5092 of inform_at_rich_loc.
5094 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
5096 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
5097 _Float<N>X built-in functions so that the variant without the
5098 "__builtin_" prefix is only enabled for the GNU C and Objective C
5099 languages when they are in non-strict ANSI/ISO mode.
5100 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
5101 * target.def (floatn_builtin_p): Add a target hook to control
5102 whether _Float<N> and _Float<N>X built-in functions without the
5103 "__builtin_" prefix are enabled, and return true for C and
5104 Objective C in the default hook. Include langhooks.h in
5106 * targhooks.h (default_floatn_builtin_p): Likewise.
5107 * targhooks.c (default_floatn_builtin_p): Likewise.
5108 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
5109 floatn_builtin_p target hook.
5110 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
5112 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
5113 Eric Botcazou <ebotcazou@adacore.com>
5115 PR rtl-optimization/81803
5116 * lra-constraints.c (curr_insn_transform): Also reload the whole
5117 register for a strict subreg no wider than a word if this is for
5118 a WORD_REGISTER_OPERATIONS target.
5120 2017-10-31 Jason Merrill <jason@redhat.com>
5122 * gdbinit.in: Skip over inlines from timevar.h.
5124 2017-10-31 Martin Liska <mliska@suse.cz>
5126 * doc/gcov.texi: Document new option.
5127 * gcov.c (print_usage): Likewise print it.
5128 (process_args): Support the argument.
5129 (format_count): New function.
5130 (format_gcov): Use the function.
5132 2017-10-31 Martin Liska <mliska@suse.cz>
5134 * gcov.c (struct name_map): do not use typedef.
5135 Define operator== and operator<.
5136 (name_search): Remove.
5137 (name_sort): Remove.
5138 (main): Do not allocate names.
5139 (process_file): Add vertical space.
5140 (generate_results): Use std::find.
5141 (release_structures): Do not release memory.
5142 (find_source): Use std::find.
5144 2017-10-31 Martin Liska <mliska@suse.cz>
5146 * gcov.c (struct line_info): Remove it's typedef.
5147 (line_info::line_info): Add proper ctor.
5148 (line_info::has_block): Do not use a typedef.
5149 (struct source_info): Do not use typedef.
5150 (circuit): Likewise.
5151 (get_cycles_count): Likewise.
5152 (output_intermediate_file): Iterate via vector iterator.
5153 (add_line_counts): Use std::vector methods.
5154 (accumulate_line_counts): Likewise.
5155 (output_lines): Likewise.
5157 2017-10-31 Martin Liska <mliska@suse.cz>
5159 * gcov.c (struct source_info): Remove typedef.
5160 (source_info::source_info): Add proper ctor.
5161 (accumulate_line_counts): Use struct, not it's typedef.
5162 (output_gcov_file): Likewise.
5163 (output_lines): Likewise.
5164 (main): Do not allocate an array.
5165 (output_intermediate_file): Use size of vector container.
5166 (process_file): Resize the vector.
5167 (generate_results): Do not preallocate, use newly added vector
5169 (release_structures): Do not release sources.
5170 (find_source): Use vector methods.
5171 (add_line_counts): Do not use typedef.
5173 2017-10-31 Martin Liska <mliska@suse.cz>
5175 * doc/gcov.texi: Document that.
5176 * gcov.c (add_line_counts): Mark lines with a non-executed
5178 (output_line_beginning): Handle such lines.
5179 (output_lines): Pass new argument.
5180 (output_intermediate_file): Print it in intermediate format.
5182 2017-10-31 Martin Liska <mliska@suse.cz>
5184 * color-macros.h: New file.
5185 * diagnostic-color.c: Factor out color related to macros to
5187 * doc/gcov.texi: Document -k option.
5188 * gcov.c (INCLUDE_STRING): Include string.h.
5189 (print_usage): Add -k option.
5190 (process_args): Parse it.
5191 (pad_count_string): New function.
5192 (output_line_beginning): Likewise.
5193 (DEFAULT_LINE_START): New macro.
5194 (output_lines): Support color output.
5196 2017-10-31 Martin Liska <mliska@suse.cz>
5198 PR gcov-profile/82633
5199 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
5200 their interaction with GCOV infrastructure.
5201 * configure.ac: Add -fkeep-{inline,static}-functions to
5203 * configure: Regenerate.
5205 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
5208 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
5210 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
5213 * config/rs6000/rs6000.md (allocate_stack): Force update interval
5214 into a register if it does not fit into an immediate offset field.
5216 2017-10-31 Olivier Hainque <hainque@adacore.com>
5218 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
5220 2017-10-31 Julia Koval <julia.koval@intel.com>
5222 * config.gcc: Add gfniintrin.h.
5223 * config/i386/gfniintrin.h: New.
5224 * config/i386/i386-builtin-types.def
5225 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5226 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5227 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5228 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5229 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5230 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
5231 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
5232 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
5233 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
5234 V64QI_FTYPE_V64QI_V64QI_INT): New types.
5235 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5236 * config/i386/immintrin.h: Include gfniintrin.h.
5237 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
5239 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
5241 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
5243 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5245 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
5246 (arm_ashldi3_1bit): Remove pattern.
5247 (ashrdi3): Remove shift by 1 expansion.
5248 (arm_ashrdi3_1bit): Remove pattern.
5249 (lshrdi3): Remove shift by 1 expansion.
5250 (arm_lshrdi3_1bit): Remove pattern.
5251 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
5252 cost of ashldi3 by 1.
5253 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
5254 (<shift>di3_neon): Likewise.
5256 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5258 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
5259 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
5260 and (*aarch64_simd_mov<VQ:mode>).
5261 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
5262 pattern alternative.
5263 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
5264 attributes to match pattern alternative.
5266 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
5268 * config.gcc (powerpc*-*-*): Add emmintrin.h.
5269 * config/rs6000/emmintrin.h: New file.
5270 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
5272 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5274 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
5275 * (movdi_vfp_cortexa8): Remove pattern.
5277 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5279 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
5281 (Specific, alpha*-dec-osf5.1): Remove.
5282 (Specific, mips-sgi-irix5): Remove.
5283 (Specific, mips-sgi-irix6): Remove.
5285 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5288 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
5289 arguments to clear_bit_region_be.
5291 2017-10-30 Jim Wilson <wilson@tuliptree.org>
5293 * gimplify.c: Include memmodel.h.
5295 2017-10-30 Martin Jambor <mjambor@suse.cz>
5297 * omp-grid.c (grid_attempt_target_gridification): Also insert a
5298 condition whether loop should be executed at all.
5300 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
5302 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5303 gimple folding of vec_madd() intrinsics.
5304 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
5305 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
5306 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
5308 2017-10-30 Richard Biener <rguenther@suse.de>
5310 PR tree-optimization/82762
5312 2017-10-23 Richard Biener <rguenther@suse.de>
5314 PR tree-optimization/82129
5316 2017-08-01 Richard Biener <rguenther@suse.de>
5318 PR tree-optimization/81181
5319 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5320 (compute_antic): ... end of iteration here.
5322 2017-10-30 Joseph Myers <joseph@codesourcery.com>
5324 * doc/invoke.texi (C Dialect Options): Document -std=c17,
5325 -std=iso9899:2017 and -std=gnu17.
5326 * doc/standards.texi (C Language): Document C17 support.
5327 * doc/cpp.texi (Overview): Mention -std=c17.
5328 (Standard Predefined Macros): Document C11 and C17 values of
5329 __STDC_VERSION__. Do not refer to C99 support as incomplete.
5330 * doc/extend.texi (Inline): Do not list individual options for
5331 standards newer than C99.
5332 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
5334 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
5337 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
5339 * asan.c (asan_finish_file): Align asan globals array by shadow
5342 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5345 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
5346 (struct store_immediate_info): Add bitregion_start and bitregion_end
5348 (store_immediate_info::store_immediate_info): Add brs and bre
5349 arguments and initialize bitregion_{start,end} from those.
5350 (struct merged_store_group): Add bitregion_start, bitregion_end,
5351 align_base and mask fields. Drop unnecessary struct keyword from
5352 struct store_immediate_info. Add do_merge method.
5353 (clear_bit_region_be): Use memset instead of loop storing zeros.
5354 (merged_store_group::do_merge): New method.
5355 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
5356 stores as long as the surrounding bitregions have no gaps.
5357 (merged_store_group::merge_overlapping): Use do_merge.
5358 (merged_store_group::apply_stores): Test that bitregion_{start,end}
5359 is byte aligned, rather than requiring that start and width are
5360 byte aligned. Drop unnecessary struct keyword from
5361 struct store_immediate_info. Allocate and populate also mask array.
5362 Make start of the arrays relative to bitregion_start rather than
5363 start and size them according to bitregion_{end,start} difference.
5364 (struct imm_store_chain_info): Drop unnecessary struct keyword from
5365 struct store_immediate_info.
5366 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
5367 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
5368 struct keyword from struct store_immediate_info.
5369 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
5370 between stores as long as the surrounding bitregions have no gaps.
5372 (struct split_store): Add orig non-static data member.
5373 (split_store::split_store): Initialize orig to false.
5374 (find_constituent_stmts): Return store_immediate_info *, non-NULL
5375 if there is exactly a single original stmt. Change stmts argument
5376 to pointer from reference, if NULL, don't push anything to it. Add
5377 first argument, use it to optimize skipping over orig stmts that
5378 are known to be before bitpos already. Simplify.
5379 (split_group): Return unsigned int count how many stores are or
5380 would be needed rather than a bool. Add allow_unaligned argument.
5381 Change split_stores argument from reference to pointer, if NULL,
5382 only do a dry run computing how many stores would be produced.
5383 Rewritten algorithm to use both alignment and misalign if
5384 !allow_unaligned and handle bitfield stores with gaps.
5385 (imm_store_chain_info::output_merged_store): Set start_byte_pos
5386 from bitregion_start instead of start. Compute allow_unaligned
5387 here, if true, do 2 split_group dry runs to compute which one
5388 produces fewer stores and prefer aligned if equal. Punt if
5389 new count is bigger or equal than original before emitting any
5390 statements, rather than during that. Remove no longer needed
5391 new_ssa_names tracking. Replace num_stmts with
5392 split_stores.length (). Use 32-bit stack allocated entries
5393 in split_stores auto_vec. Try to reuse original store lhs/rhs1
5394 if possible. Handle bitfields with gaps.
5395 (pass_store_merging::execute): Ignore bitsize == 0 stores.
5396 Compute bitregion_{start,end} for the stores and construct
5397 store_immediate_info with that. Formatting fixes.
5399 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
5402 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
5403 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
5405 2017-10-29 Jim Wilson <wilson@tuliptree.org>
5407 * gimplify.c: Include tm_p.h.
5409 * common.opt (gcoff): Re-add as ignored option.
5410 (gcoff1, gcoff2, gcoff3): Likewise.
5412 * Makefile.in (OBJS): Delete sdbout.o.
5413 (GTFILES): Delete $(srcdir)/sdbout.c.
5414 * debug.h: Delete sdb_debug_hooks.
5415 * final.c: Delete sdbout.h include.
5416 (final_scan_insn): Delete SDB_DEBUG check.
5417 (rest_of_clean_state): Likewise.
5418 * output.h: Delete sdb_begin_function_line.
5421 * toplev.c: Delete sdbout.h include.
5422 (process_options): Delete SDB_DEBUG check.
5423 * tree-core.h (tree_type_common): Delete pointer field of
5425 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
5426 TYPE_SYMTAB_POINTER.
5427 * tree.h (TYPE_SYMTAB_POINTER): Delete.
5428 (TYPE_SYMTAB_IS_POINTER): Delete.
5429 (TYPE_SYMTAB_IS_DIE): Renumber.
5430 * xcoffout.c: Refer to former sdbout.c file.
5431 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
5433 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
5434 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
5435 (Debugging Options): Delete -gcoff.
5436 (-gstabs): Delete SDB reference.
5438 (-gcoff@var{level}): Delete.
5439 * doc/passes.texi (Debugging information output): Delete SDB and
5440 sdbout.c references.
5441 * doc/tm.texi: Regenerate.
5442 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
5443 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
5445 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
5446 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
5447 (SDB_DEBUGGING_INFO): Delete.
5448 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
5449 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
5450 * target.def (output_source_filename): Delete COFF reference.
5452 * common.opt (gcoff): Delete.
5453 (gxcoff+): Update Negative chain.
5454 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
5456 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
5457 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
5458 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
5460 (expand_function_start): Change sdb reference to past tense.
5461 (expand_function_end): Change sdb reference to past tense.
5462 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
5463 * opts.c (debug_type_names): Delete coff entry.
5464 (common_handle_option): Delete OPT_gcoff case.
5465 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
5467 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
5468 * config/cris/cris.h: Delete SDB reference in comment.
5469 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
5470 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
5471 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
5472 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
5474 (ix86_expand_prologue): Likewise.
5475 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
5476 * config/ia64/ia64.h: Likewise.
5477 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
5478 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
5480 * config/mmix/mmix.h: Likewise.
5481 * config/nds32/nds32.c: Likewise.
5482 * config/stormy/storym16.h: Likewise.
5483 * config/visium/visium.h: Likewise.
5484 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
5486 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
5488 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
5489 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
5490 FRAME_POINTER_REGNUM point at high end of local var area.
5492 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
5494 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
5495 Move comment around. Do not reset best_edge for a copiable
5496 destination if the copy would cause a partition change.
5497 (better_edge_p): Remove redundant check.
5499 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
5501 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
5503 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5505 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
5506 for math functions that have _Float<N> and _Float<N>X variants.
5507 (mathfn_built_in_2): Add support for math functions that have
5508 _Float<N> and _Float<N>X variants.
5509 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
5510 (expand_builtin_mathfn_ternary): Add support for fma with
5511 _Float<N> and _Float<N>X variants.
5512 (expand_builtin): Likewise.
5513 (fold_builtin_3): Likewise.
5514 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
5515 create math function _Float<N> and _Float<N>X variants as external
5517 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
5518 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
5519 the __builtin_ prefix and if not strict ansi, without the prefix.
5520 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
5521 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
5522 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
5523 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
5524 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
5525 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
5526 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
5527 function signatures for fma _Float<N> and _Float<N>X variants.
5528 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
5529 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
5530 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
5531 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
5532 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
5533 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
5534 * gencfn-macros.c (print_case_cfn): Add support for math functions
5535 that have _Float<N> and _Float<N>X variants.
5536 (print_define_operator_list): Likewise.
5537 (fltfn_suffixes): Likewise.
5539 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
5540 for math functions that have _Float<N> and _Float<N>X variants.
5541 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
5542 and _Float<N>X variants.
5543 (COPYSIGN): Likewise.
5546 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
5547 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
5549 (integer_valued_read_call_p): Likewise.
5550 * fold-const-call.c (fold_const_call_ss): Likewise.
5551 (fold_const_call_sss): Add support for copysign, fmin, and fmax
5552 _Float<N> and _Float<N>X variants.
5553 (fold_const_call_ssss): Add support for fma _Float<N> and
5554 _Float<N>X variants.
5555 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
5556 support for copysign and fma _Float<N> and _Float<N>X variants.
5557 (backprop::process_builtin_call_use): Likewise.
5558 * tree-call-cdce.c (can_test_argument_range); Add support for
5559 sqrt _Float<N> and _Float<N>X variants.
5560 (edom_only_function): Likewise.
5561 (get_no_error_domain): Likewise.
5562 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
5563 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
5564 copysign _Float<N> and _Float<N>X variants.
5565 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
5566 handled by machine independent code.
5567 (FMAF128): Likewise.
5568 * doc/cpp.texi (Common Predefined Macros): Document defining
5569 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
5570 fma _Float<N> and _Float<N>X variants.
5572 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
5575 * config/i386/i386-modes.def (CCFPU): Remove definition.
5576 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
5577 (ix86_cc_modes_compatible): Ditto.
5578 (ix86_expand_carry_flag_compare): Ditto.
5579 (ix86_expand_int_movcc): Ditto.
5580 (ix86_expand_int_addcc): Ditto.
5581 (ix86_reverse_condition): Ditto.
5582 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
5583 Return true/false for unordered/ordered fp comparisons.
5584 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
5585 (ix86_prepare_fp_compare_args): Update for rename.
5586 (ix86_expand_fp_compare): Update for rename. Generate unordered
5587 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
5588 (ix86_expand_sse_compare_and_jump): Ditto.
5589 * config/i386/predicates.md (fcmov_comparison_operator):
5590 Remove CCFPU mode handling.
5591 (ix86_comparison_operator): Ditto.
5592 (ix86_carry_flag_operator): Ditto.
5593 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
5594 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
5595 (*cmpu<mode>_cc_i387): Ditto.
5596 (FPCMP): Remove mode iterator.
5597 (unord): Remove mode attribute.
5598 (unord_subst): New define_subst transformation
5599 (unord): New define_subst attribute.
5601 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
5602 (*cmpi<unord>xf_i387): Ditto.
5603 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
5604 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
5605 using unord_subst transformation.
5606 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
5607 (round_saeonly): Also handle CCFP mode.
5608 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
5609 Remove UNSPEC_SAHF unspec handling.
5611 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5613 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
5615 2017-10-27 Jeff Law <law@redhat.com>
5617 * gimple-ssa-sprintf.c: Include domwalk.h.
5618 (class sprintf_dom_walker): New class, derived from dom_walker.
5619 (sprintf_dom_walker::before_dom_children): New function.
5620 (struct call_info): Moved into sprintf_dom_walker class
5621 (compute_formath_length, handle_gimple_call): Likewise.
5622 (sprintf_length::execute): Call the dominator walker rather
5623 than walking the statements.
5625 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
5626 gimple statement locations.
5627 (check_array_bounds): Corresponding changes. Get the statement's
5628 location directly from wi->stmt.
5630 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
5633 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
5635 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5637 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
5638 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
5640 2017-10-27 Jakub Jelinek <jakub@redhat.com>
5643 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
5644 * config/i386/i386.c (maybe_get_pool_constant): Removed.
5645 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
5646 maybe_get_pool_constant.
5647 * config/i386/predicates.md (zero_extended_scalar_load_operand):
5650 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5652 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
5653 2.26 caveat. Update gas and gld versions.
5654 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
5657 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5659 * cgraph.h (set_malloc_flag): Declare.
5660 * cgraph.c (set_malloc_flag_1): New function.
5661 (set_malloc_flag): Likewise.
5662 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
5663 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
5665 (read_ipa_call_summary): Add support for reading is_return_callee.
5666 (write_ipa_call_summary): Stream is_return_callee.
5667 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
5668 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
5669 ipa-prop.h, ipa-fnsummary.h.
5670 (pure_const_names): Change to static.
5671 (malloc_state_e): Define.
5672 (malloc_state_names): Define.
5673 (funct_state_d): Add field malloc_state.
5674 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
5675 (check_retval_uses): New function.
5676 (malloc_candidate_p): Likewise.
5677 (analyze_function): Add support for malloc attribute.
5678 (pure_const_write_summary): Stream malloc_state.
5679 (pure_const_read_summary): Add support for reading malloc_state.
5680 (dump_malloc_lattice): New function.
5681 (propagate_malloc): New function.
5682 (warn_function_malloc): New function.
5683 (ipa_pure_const::execute): Call propagate_malloc and
5684 ipa_free_fn_summary.
5685 (pass_local_pure_const::execute): Add support for malloc attribute.
5686 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
5687 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
5689 2017-10-27 Martin Liska <mliska@suse.cz>
5691 PR gcov-profile/82457
5692 * doc/invoke.texi: Document that one needs a non-strict ISO mode
5693 for fork-like functions to be properly instrumented.
5695 2017-10-27 Richard Biener <rguenther@suse.de>
5698 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
5699 info when we redirected EH.
5701 2017-10-26 Michael Collison <michael.collison@arm.com>
5703 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
5705 (<optab>_trunchf<GPI:mode>2: New pattern.
5706 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
5707 * config/aarch64/iterators.md (wv): New mode attribute.
5708 (vf, VF): New mode attributes.
5709 (vgp, VGP): New mode attributes.
5710 (s): Update attribute with SImode and DImode prefixes.
5712 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
5714 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
5715 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
5716 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
5717 (nios2_option_overide): Initialize it. Don't allow R0-relative
5718 addressing with PIC.
5719 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
5720 (nios2_symbolic_constant_p): Likewise.
5721 (nios2_legitimate_address_p): Likewise.
5722 (nios2_r0rel_section_name_p): New.
5723 (nios2_symbol_ref_in_r0rel_data_p): New.
5724 (nios2_emit_move_sequence): Handle r0rel_constant_p.
5725 (r0rel_constant_p): New.
5726 (nios2_print_operand_address): Handle r0rel_constant_p.
5727 (nios2_cdx_narrow_form_p): Likewise.
5728 * config/nios2/nios2.opt (mr0rel-sec=): New option.
5729 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
5730 (Nios II Options): Document -mr0rel-sec.
5732 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
5734 * config/nios2/nios2.c: Include xregex.h.
5735 (nios2_gprel_sec_regex): New.
5736 (nios2_option_overide): Initialize it. Don't allow GP-relative
5737 addressing with PIC.
5738 (nios2_small_section_name_p): Check for regex match.
5739 * config/nios2/nios2.opt (mgprel-sec=): New option.
5740 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
5741 (Nios II Options): Document -mgprel-sec.
5743 2017-10-26 Jim Wilson <wilson@tuliptree.org>
5745 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
5747 2017-10-26 Tom de Vries <tom@codesourcery.com>
5749 PR tree-optimization/82707
5750 * gimple.c (gimple_copy): Fix unsharing of
5751 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
5753 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
5755 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
5756 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
5757 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
5758 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
5759 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
5760 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
5761 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
5762 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
5763 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
5764 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
5765 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
5766 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
5767 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
5768 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
5769 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
5770 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
5771 _mm512_mask_cmpunord_ps_mask): New intrinsics.
5773 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5775 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
5777 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
5778 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
5779 warning to rs6000.c. Remove the Undocumented flag, since it has
5781 (-mabi=ibmlongdouble): Likewise.
5782 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
5783 already set, set the default format for long double.
5784 (rs6000_debug_reg_global): Print whether long double is IBM or
5786 (rs6000_option_override_internal): Rework setting long double
5787 format. Only warn if the user is changing the long double default
5788 and they did not use -Wno-psabi.
5789 * doc/invoke.texi (PowerPC options): Update the documentation for
5790 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
5792 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5793 Alan Hayward <alan.hayward@arm.com>
5794 David Sherwood <david.sherwood@arm.com>
5796 * rtl.h (wider_subreg_mode): New function.
5797 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
5798 rather than an unsigned int *.
5799 * ira-color.c (regno_max_ref_width): Replace with...
5800 (regno_max_ref_mode): ...this new variable.
5801 (coalesced_pseudo_reg_slot_compare): Update accordingly.
5802 Use wider_subreg_mode.
5803 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
5804 rather than an unsigned int *.
5805 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
5806 (process_alt_operands): Likewise.
5807 (invariant_p): Likewise.
5808 * lra-spills.c (assign_mem_slot): Likewise.
5809 (add_pseudo_to_slot): Likewise.
5810 * lra.c (collect_non_operand_hard_regs): Likewise.
5811 (add_regs_to_insn_regno_info): Likewise.
5812 * reload1.c (regno_max_ref_width): Replace with...
5813 (regno_max_ref_mode): ...this new variable.
5814 (reload): Update accordingly. Update call to
5815 ira_sort_regnos_for_alter_reg.
5816 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
5817 (init_eliminable_invariants): Update to use regno_max_ref_mode.
5818 (scan_paradoxical_subregs): Likewise.
5820 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5822 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
5823 (aarch64_frame): Add emit_frame_chain boolean.
5824 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5825 Move eh_return case to aarch64_layout_frame.
5826 (aarch64_layout_frame): Initialize emit_frame_chain.
5827 (aarch64_expand_prologue): Use emit_frame_chain.
5829 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5831 * config/aarch64/aarch64.c (aarch64_layout_frame):
5832 Ensure LR is always stored at the bottom of the callee-saves.
5833 Remove rarely used frame layout which saves callee-saves at top of
5834 frame, so the store of LR can be used as a valid probe in all cases.
5836 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5838 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
5839 Improve unaligned TImode/TFmode base/offset split.
5841 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5842 Alan Hayward <alan.hayward@arm.com>
5843 David Sherwood <david.sherwood@arm.com>
5845 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
5846 * combine.c (find_single_use_1): Likewise.
5847 (expand_field_assignment): Likewise.
5848 (move_deaths): Likewise.
5849 * lra-constraints.c (simplify_operand_subreg): Likewise.
5850 (curr_insn_transform): Likewise.
5851 * lra.c (collect_non_operand_hard_regs): Likewise.
5852 (add_regs_to_insn_regno_info): Likewise.
5853 * rtlanal.c (reg_referenced_p): Likewise.
5854 (covers_regno_no_parallel_p): Likewise.
5856 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5858 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
5859 Don't print any bits outside the precision of the value.
5860 * wide-int.cc (test_printing): Add some new tests.
5862 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5864 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
5865 supports -xbrace_comment option.
5866 * configure: Regenerate.
5867 * config.in: Regenerate.
5868 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
5869 (ASM_CPU_SPEC): Use it.
5871 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5873 * target.def (static_rtx_alignment): New hook.
5874 * targhooks.h (default_static_rtx_alignment): Declare.
5875 * targhooks.c (default_static_rtx_alignment): New function.
5876 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
5877 * doc/tm.texi: Regenerate.
5878 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
5879 instead of targetm.constant_alignment. Remove call to
5881 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5882 (cris_preferred_mininum_alignment): New function, split out from...
5883 (cris_constant_alignment): ...here.
5884 (cris_static_rtx_alignment): New function.
5885 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
5887 (ix86_constant_alignment): ...here.
5888 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5889 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5890 (mmix_static_rtx_alignment): New function.
5891 * config/spu/spu.c (spu_static_rtx_alignment): New function.
5892 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5894 2017-10-26 Tamar Christina <tamar.christina@arm.com>
5897 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
5898 Add flag_trapping_math and flag_fp_int_builtin_inexact.
5900 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
5902 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
5903 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
5904 mark as a sign-extending load.
5905 (local_pic_load_u): Define.
5907 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
5910 * fold-const.c (operand_equal_for_comparison_p): Also return true
5911 if ARG0 is a simple variant of ARG1 with narrower precision.
5912 (fold_ternary_loc): Always pass unstripped operands to the predicate.
5914 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
5916 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
5918 * i386.h (processor_costs): Add gather_static, gather_per_elt,
5919 scatter_static, scatter_per_elt.
5920 * x86-tune-costs.h: Add new cost entries.
5922 2017-10-25 Richard Biener <rguenther@suse.de>
5924 * tree-ssa-sccvn.h (vn_eliminate): Declare.
5925 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
5926 class pass_fre): Move to ...
5927 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
5928 class pass_fre): ... here and adjust for statistics.
5930 2017-10-25 Jakub Jelinek <jakub@redhat.com>
5933 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
5934 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
5935 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
5936 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
5939 2017-10-25 Richard Biener <rguenther@suse.de>
5941 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
5942 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
5943 eliminate_push_avail, eliminate_insert): Move inside...
5944 (class eliminate_dom_walker): ... this class in preparation
5946 (fini_eliminate): Remove by merging with ...
5947 (eliminate): ... this function. Adjust for class changes.
5948 (pass_pre::execute): Remove fini_eliminate call.
5949 (pass_fre::execute): Likewise.
5951 2017-10-24 Jakub Jelinek <jakub@redhat.com>
5954 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
5955 (VPERMI2, VPERMI2I): New mode iterators.
5956 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
5957 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
5959 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
5960 mode iterator. Remove 3 old define_insn patterns.
5961 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
5962 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
5963 VPERMI2 mode iterator, remove the other two expanders.
5964 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
5965 to use VPERMI2 mode iterator, add another alternative for vpermi2*
5966 instructions, remove the other two patterns.
5967 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
5968 mode iterator, remove the other two patterns.
5969 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
5970 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
5971 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
5972 and adjust argument order accordingly.
5973 (ix86_expand_vec_perm): Adjust caller.
5974 (expand_vec_perm_1): Likewise.
5975 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
5976 (expand_vec_perm_vpermt2_vpshub2): ... this.
5977 (ix86_expand_vec_perm_const_1): Adjust caller.
5978 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
5981 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
5982 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
5983 (vec_shl_<mode>): Remove unused expander.
5984 (avx512bw_<shift_insn><mode>3): New define_insn.
5985 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
5986 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
5988 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
5991 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
5994 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
5996 PR rtl-optimization/82396
5997 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
5998 (autopref_multipass_init): Simplify initialization.
5999 (autopref_rank_data): Simplify sort order.
6000 * gcc/sched-int.h (autopref_multipass_data_): Remove
6001 multi_mem_insn_p, min_offset and max_offset.
6003 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6006 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6007 Check special value of flag_omit_frame_pointer.
6008 (aarch64_can_eliminate): Likewise.
6009 (aarch64_override_options_after_change_1): Simplify handling of
6010 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
6012 2017-10-24 Richard Biener <rguenther@suse.de>
6014 PR tree-optimization/82697
6015 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
6016 zero for conditional load and unconditional store.
6018 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6020 * doc/install.texi: Document bootstrap-cet.
6022 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6025 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
6026 ENDBR instruction at function entrance if function is only
6029 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6032 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
6033 patterns to better describe from which operation the CF is computed.
6034 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
6035 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
6036 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
6037 is 0, use _0 suffixed expanders instead of emitting a comparison
6040 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
6042 * config/i386/i386.md(*movsf_internal, *movdf_internal):
6043 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
6045 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
6048 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
6049 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
6050 * loop-iv.c (iv_get_reaching_def): Likewise.
6051 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
6052 variable is promoted and the partition contains undefined values.
6054 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6056 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
6058 (nios2_address_cost): Define.
6059 (nios2_legitimize_address): Recognize (exp + constant) directly.
6060 (TARGET_ADDRESS_COST): Define.
6062 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6064 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
6065 (nios2_symbolic_memory_operand_p): Declare.
6066 (nios2_split_large_constant): Declare.
6067 (nios2_split_symbolic_memory_operand): Declare.
6068 * config/nios2/nios2.c: Adjust includes.
6069 (nios2_symbolic_constant_allowed): New.
6070 (nios2_symbolic_constant_p): New.
6071 (nios2_plus_symbolic_constant_p): New.
6072 (nios2_valid_addr_expr_p): Recognize addresses involving
6074 (nios2_legitimate_address_p): Likewise, also LO_SUM.
6075 (nios2_symbolic_memory_operand_p): New.
6076 (nios2_large_constant_p): New.
6077 (nios2_split_large_constant): New.
6078 (nios2_split_plus_large_constant): New.
6079 (nios2_split_symbolic_memory_operand): New.
6080 (nios2_legitimize_address): Code refactoring. Handle addresses
6081 involving symbolic constants.
6082 (nios2_emit_move_sequence): Likewise.
6083 (nios2_print_operand): Improve error output.
6084 (nios2_print_operand_address): Handle LO_SUM.
6085 (nios2_cdx_narrow_form_p): Likewise.
6086 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
6087 operands involving symbolic constants.
6088 (movhi_internal, movsi_internal): Likewise.
6089 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
6090 (extendhisi2, extendqi<mode>2): Likewise.
6092 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6094 * tree-pass.h (PROP_rtl_split_insns): Define.
6095 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
6097 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6099 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
6101 2017-10-23 Jakub Jelinek <jakub@redhat.com>
6104 * target.def (const_not_ok_for_debug_p): Default to
6105 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
6106 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
6107 * targhooks.c (default_const_not_ok_for_debug_p): New function.
6108 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
6109 which targetm.const_not_ok_for_debug_p returned true.
6110 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
6112 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
6114 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
6115 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
6116 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
6118 (ix86_const_not_ok_for_debug_p): New function.
6119 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
6120 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
6122 2017-10-23 David Malcolm <dmalcolm@redhat.com>
6125 * system.h: Conditionally include "unique-ptr.h" if
6126 INCLUDE_UNIQUE_PTR is defined.
6127 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
6128 of defining INCLUDE_UNIQUE_PTR before including "system.h".
6130 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
6132 * config/rl78/rl78.md: New define_expand "subdi3".
6134 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
6137 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
6138 DF_REF_INSN if DF_REF_INSN_INFO is false.
6140 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
6142 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
6143 xmm_move instead of sse_move.
6144 (sse_store_index): New function.
6145 (ix86_register_move_cost): Be more sensible about mismatch stall;
6146 model AVX moves correctly; make difference between sse->integer and
6148 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
6149 moves; make difference between SSE and AVX.
6150 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
6151 and zmm_move. Increase size of sse load and store tables;
6152 add unaligned load and store tables; add ssemmx_to_integer.
6153 * x86-tune-costs.h: Update all entries according to real
6154 move latencies from Agner Fog's manual and chip documentation.
6156 2017-10-23 Jakub Jelinek <jakub@redhat.com>
6159 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
6160 * config/i386/constraints.md (Wf): New constraint.
6161 * config/i386/i386.md (UNSPEC_SBB): New unspec.
6162 (cmp<dwi>_doubleword): Removed.
6163 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
6164 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
6165 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
6166 expand with cmp<dwi>_doubleword. For LTU and GEU use
6167 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
6169 * common.opt (gcolumn-info): Enable by default.
6170 * doc/invoke.texi (gcolumn-info): Document new default.
6172 2017-10-23 Richard Biener <rguenther@suse.de>
6174 PR tree-optimization/82672
6175 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
6176 Fold the stmt if we propagated into it.
6178 2017-10-23 Richard Biener <rguenther@suse.de>
6180 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
6181 (bitmap_remove_expr_from_set): ... this. All callers call this
6182 for non-constant values.
6183 (bitmap_set_subtract): Rename to...
6184 (bitmap_set_subtract_expressions): ... this. Adjust and
6186 (bitmap_set_contains_value): Remove superfluous check.
6187 (bitmap_set_replace_value): Inline into single caller ...
6188 (bitmap_value_replace_in_set): ... here and simplify.
6189 (dependent_clean): Merge into ...
6190 (clean): ... this using an overload. Adjust.
6191 (prune_clobbered_mems): Adjust.
6192 (compute_antic_aux): Likewise.
6193 (compute_partial_antic_aux): Likewise.
6195 2017-10-23 Richard Biener <rguenther@suse.de>
6197 PR tree-optimization/82129
6199 2017-08-01 Richard Biener <rguenther@suse.de>
6201 PR tree-optimization/81181
6202 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6203 (compute_antic): ... end of iteration here.
6205 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6207 * target.def (starting_frame_offset): New hook.
6208 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
6209 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
6210 * doc/tm.texi.in: Regenerate.
6211 * hooks.h (hook_hwi_void_0): Declare.
6212 * hooks.c (hook_hwi_void_0): New function.
6213 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
6214 STARTING_FRAME_OFFSET.
6215 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
6216 * reload1.c (reload): Likewise.
6217 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
6218 instead of STARTING_FRAME_OFFSET.
6219 * function.c (try_fit_stack_local): Likewise.
6220 (assign_stack_local_1): Likewise
6221 (instantiate_virtual_regs): Likewise.
6222 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
6223 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
6224 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
6225 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
6226 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
6227 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
6228 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
6229 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
6230 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
6231 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
6232 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
6233 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
6234 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
6235 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
6236 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
6237 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
6238 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
6239 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
6240 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
6241 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
6242 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
6243 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
6244 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
6245 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
6246 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
6247 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
6248 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
6249 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
6250 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
6251 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
6252 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
6253 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
6254 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
6255 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
6256 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
6257 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
6258 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
6259 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
6260 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
6261 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
6262 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
6263 * config/avr/avr.c (avr_starting_frame_offset): Make static and
6264 return a HOST_WIDE_INT.
6265 (avr_builtin_setjmp_frame_value): Use it instead of
6266 STARTING_FRAME_OFFSET.
6267 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6268 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
6269 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
6271 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6272 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
6273 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
6274 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6275 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
6276 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
6277 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6278 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
6279 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
6280 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6281 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
6282 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
6284 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6285 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
6286 * config/mips/mips.c (mips_compute_frame_info): Refer to
6287 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6288 (mips_starting_frame_offset): New function.
6289 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6290 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
6291 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
6292 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
6293 and return a HOST_WIDE_INT.
6294 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6295 (mmix_initial_elimination_offset): Refer to
6296 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6297 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
6298 * config/pa/pa.c (pa_starting_frame_offset): New function.
6299 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
6300 (pa_expand_prologue): Likewise.
6301 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6302 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
6303 !FRAME_GROWS_DOWNWARD handling to...
6304 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6305 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
6306 !FRAME_GROWS_DOWNWARD handling to...
6307 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6308 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
6309 !FRAME_GROWS_DOWNWARD handling to...
6310 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6311 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
6313 (rs6000_starting_frame_offset): New function.
6314 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
6315 !FRAME_GROWS_DOWNWARD handling to...
6316 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6317 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
6318 !FRAME_GROWS_DOWNWARD handling to...
6319 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6320 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
6321 !FRAME_GROWS_DOWNWARD handling to...
6322 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6323 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
6324 (rs6000_starting_frame_offset): New function.
6325 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
6326 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
6327 * config/vax/vax.c (vax_starting_frame_offset): New function.
6328 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
6329 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6330 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
6331 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
6332 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6333 * system.h (STARTING_FRAME_OFFSET): Poison.
6335 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6337 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
6338 SCALAR_TYPE_MODE instead of TYPE_MODE.
6340 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6341 Alan Hayward <alan.hayward@arm.com>
6342 David Sherwood <david.sherwood@arm.com>
6344 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
6346 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6347 Alan Hayward <alan.hayward@arm.com>
6348 David Sherwood <david.sherwood@arm.com>
6350 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
6352 2017-10-23 Richard Biener <rguenther@suse.de>
6354 PR tree-optimization/82129
6355 * tree-ssa-pre.c (bitmap_set_and): Remove.
6356 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
6357 canonicalizing expressions in the set to those with lowest
6358 ID rather than taking that from the first edge.
6360 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6362 * combine.c (rtx_equal_for_field_assignment_p): Use
6363 byte_lowpart_offset.
6365 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6366 Alan Hayward <alan.hayward@arm.com>
6367 David Sherwood <david.sherwood@arm.com>
6369 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
6370 to a SUBREG_PROMOTED_VAR.
6372 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6373 Alan Hayward <alan.hayward@arm.com>
6374 David Sherwood <david.sherwood@arm.com>
6376 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
6377 (expand_debug_source_expr): Likewise.
6378 * combine.c (combine_simplify_rtx): Likewise.
6379 * cse.c (fold_rtx): Likewise.
6380 * optabs.c (expand_float): Likewise.
6381 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6382 (simplify_binary_operation_1): Likewise.
6384 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6385 Alan Hayward <alan.hayward@arm.com>
6386 David Sherwood <david.sherwood@arm.com>
6388 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
6389 (record_promoted_value): Likewise.
6390 * expr.c (expand_expr_real_2): Likewise.
6391 * ree.c (update_reg_equal_equiv_notes): Likewise.
6392 (combine_set_extension): Likewise.
6393 * rtlanal.c (low_bitmask_len): Likewise.
6394 * simplify-rtx.c (neg_const_int): Likewise.
6395 (simplify_binary_operation_1): Likewise.
6397 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6398 Alan Hayward <alan.hayward@arm.com>
6399 David Sherwood <david.sherwood@arm.com>
6401 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
6402 * regcprop.c (maybe_mode_change): Likewise.
6403 * reload1.c (alter_reg): Likewise.
6405 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6407 * inchash.h (inchash::hash::add_wide_int): New function.
6408 * lto-streamer-out.c (hash_tree): Use it.
6410 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6412 * inchash.h (inchash::hash::add_wide_int): Rename to...
6413 (inchash::hash::add_hwi): ...this.
6414 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
6415 (polymorphic_call_target_hasher::hash): Likewise.
6416 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
6417 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
6418 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
6419 * lto-streamer-out.c (hash_tree): Likewise.
6420 * optc-save-gen.awk: Likewise.
6421 * tree.c (add_expr): Likewise.
6423 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6426 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
6427 for ordered inequality comparisons even with TARGET_IEEE_FP.
6429 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6432 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
6433 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
6434 Expand with cmp<dwi>_doubleword.
6436 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6438 * extend.texi: Add x86 specific to 'nocf_check' attribute.
6439 List CET intrinsics.
6440 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
6441 specific to -fcf-protection option.
6443 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6445 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
6446 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
6447 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
6448 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
6449 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
6450 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
6451 (extra_objs): Add cet.o for Linux/x86 targets.
6452 (tmake_file): Add i386/t-cet for Linux/x86 targets.
6453 * config/i386/cet.c: New file.
6454 * config/i386/cetintrin.h: Likewise.
6455 * config/i386/t-cet: Likewise.
6456 * config/i386/cpuid.h (bit_SHSTK): New.
6457 (bit_IBT): Likewise.
6458 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
6459 pass IBT and SHSTK bits.
6460 * config/i386/i386-builtin-types.def
6461 (VOID_FTYPE_UNSIGNED_PVOID): New.
6462 (VOID_FTYPE_UINT64_PVOID): Likewise.
6463 * config/i386/i386-builtin.def: Add CET intrinsics.
6464 * config/i386/i386-c.c (ix86_target_macros_internal): Add
6465 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
6466 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
6467 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
6469 * config/i386/i386.c (rest_of_insert_endbranch): New.
6470 (pass_data_insert_endbranch): Likewise.
6471 (pass_insert_endbranch): Likewise.
6472 (make_pass_insert_endbranch): Likewise.
6473 (ix86_notrack_prefixed_insn_p): Likewise.
6474 (ix86_target_string): Add -mibt, -mshstk flags.
6475 (ix86_option_override_internal): Add flag_cf_protection
6477 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
6478 (ix86_print_operand): Add 'notrack' prefix output.
6479 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
6480 (ix86_expand_builtin): Expand CET intrinsics.
6481 (x86_output_mi_thunk): Add 'endbranch' instruction.
6482 * config/i386/i386.h (TARGET_IBT): New.
6483 (TARGET_IBT_P): Likewise.
6484 (TARGET_SHSTK): Likewise.
6485 (TARGET_SHSTK_P): Likewise.
6486 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
6487 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
6488 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
6489 (builtin_setjmp_setup): New pattern.
6490 (builtin_longjmp): Likewise.
6491 (rdssp<mode>): Likewise.
6492 (incssp<mode>): Likewise.
6493 (saveprevssp): Likewise.
6494 (rstorssp): Likewise.
6495 (wrss<mode>): Likewise.
6496 (wruss<mode>): Likewise.
6497 (setssbsy): Likewise.
6498 (clrssbsy): Likewise.
6499 (nop_endbr): Likewise.
6500 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
6502 * config/i386/immintrin.h: Include <cetintrin.h>.
6503 * config/i386/linux-common.h
6504 (file_end_indicate_exec_stack_and_cet): New prototype.
6505 (TARGET_ASM_FILE_END): New.
6507 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
6509 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
6510 latencies instead of having separate table; make difference between
6511 integer and float costs.
6512 * i386.h (processor_costs): Remove scalar_stmt_cost,
6513 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
6514 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
6516 * x86-tune-costs.h: Remove entries which has been removed in
6517 procesor_costs from all tables; make cond_taken_branch_cost
6518 and cond_not_taken_branch_cost COST_N_INSNS based.
6520 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
6522 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
6524 2017-10-20 Jakub Jelinek <jakub@redhat.com>
6526 * config/i386/i386.md (isa): Remove fma_avx512f.
6527 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
6528 <avx512>_fmadd_<mode>_mask3<round_name>,
6529 <avx512>_fmsub_<mode>_mask<round_name>,
6530 <avx512>_fmsub_<mode>_mask3<round_name>,
6531 <avx512>_fnmadd_<mode>_mask<round_name>,
6532 <avx512>_fnmadd_<mode>_mask3<round_name>,
6533 <avx512>_fnmsub_<mode>_mask<round_name>,
6534 <avx512>_fnmsub_<mode>_mask3<round_name>,
6535 <avx512>_fmaddsub_<mode>_mask<round_name>,
6536 <avx512>_fmaddsub_<mode>_mask3<round_name>,
6537 <avx512>_fmsubadd_<mode>_mask<round_name>,
6538 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
6539 (*vec_widen_umult_even_v16si<mask_name>,
6540 *vec_widen_smult_even_v16si<mask_name>): Likewise.
6541 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
6543 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6545 * extend.texi: Add 'nocf_check' documentation.
6546 * gimple.texi: Add second parameter to
6547 gimple_build_call_from_tree.
6548 * invoke.texi: Add -fcf-protection documentation.
6549 * rtl.texi: Add REG_CALL_NOTRACK documenation.
6551 2017-10-20 Richard Biener <rguenther@suse.de>
6553 PR tree-optimization/82473
6554 * tree-vect-loop.c (vectorizable_reduction): Properly get at
6555 the largest input type.
6557 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6559 * c-attribs.c (handle_nocf_check_attribute): New function.
6560 (c_common_attribute_table): Add 'nocf_check' handling.
6561 * gimple-parser.c: Add second argument NULL to
6562 gimple_build_call_from_tree.
6563 * attrib.c (comp_type_attributes): Check nocf_check attribute.
6564 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
6566 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
6567 * common.opt: Add fcf-protection flag.
6568 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
6569 * flag-types.h: Add enum cf_protection_level.
6570 * gimple.c (gimple_build_call_from_tree): Add second parameter.
6571 Add 'nocf_check' attribute propagation to gimple call.
6572 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
6573 (gimple_build_call_from_tree): Update prototype.
6574 (gimple_call_nocf_check_p): New function.
6575 (gimple_call_set_nocf_check): Likewise.
6576 * gimplify.c: Add second argument to gimple_build_call_from_tree.
6577 * ipa-icf.c: Add nocf_check attribute in statement hash.
6578 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
6579 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
6580 * toplev.c (process_options): Add flag_cf_protection handling.
6582 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6584 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
6586 2017-10-20 Richard Biener <rguenther@suse.de>
6588 PR tree-optimization/82603
6589 * tree-if-conv.c (predicate_mem_writes): Make sure to only
6590 remove false predicated stores.
6592 2017-10-20 Richard Biener <rguenther@suse.de>
6594 * graphite-isl-ast-to-gimple.c
6595 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
6596 Remove return value and simplify, dump copied stmt after lhs
6598 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
6599 Reduce dump verbosity.
6600 (gsi_insert_earliest): Likewise.
6601 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
6602 * graphite.c (print_global_statistics): Adjust dumping.
6603 (print_graphite_scop_statistics): Likewise.
6604 (print_graphite_statistics): Do not dump loops here.
6605 (graphite_transform_loops): But here.
6607 2017-10-20 Nicolas Roche <roche@adacore.com>
6609 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
6610 * configure: Regenerate.
6612 2017-10-20 Jakub Jelinek <jakub@redhat.com>
6615 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
6616 functions when optimizing replace GIMPLE_RETURN stmts with
6617 calls to __builtin_unreachable ().
6620 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
6621 for -fsanitize=thread link of executables.
6622 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
6623 link of executables.
6626 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
6628 (<shift_insn><mode>3<mask_name>): Change the last of the 3
6629 define_insns for logical vector shifts to use VI248_AVX512BW
6630 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
6631 condition, useless isa and prefix attributes. Change the first
6632 2 of these define_insns to ...
6633 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
6634 define_insn for avx512vl.
6635 (<shift_insn><mode>3): ... and this, new define_insn without
6636 masking for non-avx512vl.
6639 * config/i386/sse.md (*andnot<mode>3,
6640 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
6641 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
6642 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
6643 not applied use empty suffix even for TARGET_AVX512VL.
6644 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
6645 is applied, supply evex,evex or evex,evex,evex instead of just
6648 2017-10-20 Julia Koval <julia.koval@intel.com>
6650 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
6651 (OPTION_MASK_ISA_GFNI_UNSET): New.
6652 (ix86_handle_option): Handle OPT_mgfni.
6653 * config/i386/cpuid.h (bit_GFNI): New.
6654 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
6655 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
6656 * config/i386/i386.c (ix86_target_string): Add -mgfni.
6657 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
6658 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
6659 * config/i386/i386.opt: Add mgfni.
6661 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
6663 * config/msp430/msp430.c (msp430_option_override): Disable
6664 -fdelete-null-pointer-checks.
6665 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
6667 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6669 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
6670 of x87 and SSE instructions.
6672 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6674 * asan.c (create_cond_insert_point): Do not update edge count.
6675 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
6676 (afdo_propagate_circuit): Likewise.
6677 (afdo_calculate_branch_prob): Likewise.
6678 (afdo_annotate_cfg): Likewise.
6679 * basic-block.h (struct edge_def): Remove count.
6680 (edge_def::count): New accessor.
6681 * bb-reorder.c (rotate_loop): Update.
6682 (find_traces_1_round): Update.
6683 (connect_traces): Update.
6684 (sanitize_hot_paths): Update.
6685 * cfg.c (unchecked_make_edge): Update.
6686 (make_single_succ_edge): Update.
6687 (check_bb_profile): Update.
6688 (dump_edge_info): Update.
6689 (update_bb_profile_for_threading): Update.
6690 (scale_bbs_frequencies_int): Update.
6691 (scale_bbs_frequencies_gcov_type): Update.
6692 (scale_bbs_frequencies_profile_count): Update.
6693 (scale_bbs_frequencies): Update.
6694 * cfganal.c (connect_infinite_loops_to_exit): Update.
6695 * cfgbuild.c (compute_outgoing_frequencies): Update.
6696 (find_many_sub_basic_blocks): Update.
6697 * cfgcleanup.c (try_forward_edges): Update.
6698 (try_crossjump_to_edge): Update
6699 * cfgexpand.c (expand_gimple_cond): Update
6700 (expand_gimple_tailcall): Update
6701 (construct_exit_block): Update
6702 * cfghooks.c (verify_flow_info): Update
6703 (redirect_edge_succ_nodup): Update
6704 (split_edge): Update
6705 (make_forwarder_block): Update
6706 (duplicate_block): Update
6707 (account_profile_record): Update
6708 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6709 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6710 * cfgloopmanip.c (scale_loop_profile): Update.
6712 (lv_adjust_loop_entry_edge): Update.
6713 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
6714 (force_nonfallthru_and_redirect): Update.
6715 (purge_dead_edges): Update.
6716 (rtl_flow_call_edges_add): Update.
6717 * cgraphunit.c (init_lowered_empty_function): Update.
6718 (cgraph_node::expand_thunk): Update.
6719 * gimple-pretty-print.c (dump_probability): Update.
6720 (dump_edge_probability): Update.
6721 * gimple-ssa-isolate-paths.c (isolate_path): Update.
6722 * haifa-sched.c (sched_create_recovery_edges): Update.
6723 * hsa-gen.c (convert_switch_statements): Update.
6724 * ifcvt.c (dead_or_predicable): Update.
6725 * ipa-inline-transform.c (inline_transform): Update.
6726 * ipa-split.c (split_function): Update.
6727 * ipa-utils.c (ipa_merge_profiles): Update.
6728 * loop-doloop.c (add_test): Update.
6729 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
6730 * lto-streamer-in.c (input_cfg): Update.
6731 (input_function): Update.
6732 * lto-streamer-out.c (output_cfg): Update.
6733 * modulo-sched.c (sms_schedule): Update.
6734 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6735 * predict.c (maybe_hot_edge_p): Update.
6736 (unlikely_executed_edge_p): Update.
6737 (probably_never_executed_edge_p): Update.
6738 (dump_prediction): Update.
6739 (drop_profile): Update.
6740 (propagate_unlikely_bbs_forward): Update.
6741 (determine_unlikely_bbs): Update.
6742 (force_edge_cold): Update.
6743 * profile.c (compute_branch_probabilities): Update.
6744 * reg-stack.c (better_edge): Update.
6745 * shrink-wrap.c (handle_simple_exit): Update.
6746 * tracer.c (better_p): Update.
6747 * trans-mem.c (expand_transaction): Update.
6748 (split_bb_make_tm_edge): Update.
6749 * tree-call-cdce.c: Update.
6750 * tree-cfg.c (gimple_find_sub_bbs): Update.
6751 (gimple_split_edge): Update.
6752 (gimple_duplicate_sese_region): Update.
6753 (gimple_duplicate_sese_tail): Update.
6754 (gimple_flow_call_edges_add): Update.
6755 (insert_cond_bb): Update.
6756 (execute_fixup_cfg): Update.
6757 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
6758 * tree-complex.c (expand_complex_div_wide): Update.
6759 * tree-eh.c (lower_resx): Update.
6760 (unsplit_eh): Update.
6761 (cleanup_empty_eh_move_lp): Update.
6762 * tree-inline.c (copy_edges_for_bb): Update.
6763 (freqs_to_counts): Update.
6764 (copy_cfg_body): Update.
6765 * tree-ssa-dce.c (remove_dead_stmt): Update.
6766 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6767 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
6768 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
6769 (unloop_loops): Update.
6770 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6771 * tree-ssa-loop-split.c (connect_loops): Update.
6772 (split_loop): Update.
6773 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6774 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
6775 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
6776 * tree-ssa-reassoc.c (branch_fixup): Update.
6777 * tree-ssa-tail-merge.c (replace_block_by): Update.
6778 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
6779 (compute_path_counts): Update.
6780 (update_profile): Update.
6781 (recompute_probabilities): Update.
6782 (update_joiner_offpath_counts): Update.
6783 (estimated_freqs_path): Update.
6784 (freqs_to_counts_path): Update.
6785 (clear_counts_path): Update.
6786 (ssa_fix_duplicate_block_edges): Update.
6787 (duplicate_thread_path): Update.
6788 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
6789 (case_bit_test_cmp): Update.
6790 (collect_switch_conv_info): Update.
6791 (gen_inbound_check): Update.
6792 (do_jump_if_equal): Update.
6793 (emit_cmp_and_jump_insns): Update.
6794 * tree-tailcall.c (decrease_profile): Update.
6795 (eliminate_tail_call): Update.
6796 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6797 (vect_do_peeling): Update.
6798 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6799 * ubsan.c (ubsan_expand_null_ifn): Update.
6800 (ubsan_expand_ptr_ifn): Update.
6801 * value-prof.c (gimple_divmod_fixed_value): Update.
6802 (gimple_mod_pow2): Update.
6803 (gimple_mod_subtract): Update.
6804 (gimple_ic): Update.
6805 (gimple_stringop_fixed_value): Update.
6807 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6810 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
6812 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
6814 PR rtl-optimization/82395
6815 * ira-color.c (allocno_priority_compare_func): Fix comparison step
6816 based on non_spilled_static_chain_regno_p.
6818 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6820 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
6821 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
6822 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
6824 2017-10-19 Martin Sebor <msebor@redhat.com>
6826 PR tree-optimization/82596
6827 * tree.c (array_at_struct_end_p): Handle STRING_CST.
6829 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
6831 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
6832 (get_mem_refs_of_builtin_call): Likewise.
6833 * builtins.c (expand_builtin_apply): Adjust call to
6834 allocate_dynamic_stack_space.
6835 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
6836 the third argument to allocate_dynamic_stack_space, otherwise -1.
6837 (expand_builtin): Deal with all alloca variants.
6838 (is_inexpensive_builtin): Likewise.
6839 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
6840 * calls.c (special_function_p): Deal with all alloca variants.
6841 (initialize_argument_information): Adjust call to
6842 allocate_dynamic_stack_space.
6843 (expand_call): Likewise.
6844 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
6845 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
6846 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
6847 use it for the stack usage computation.
6848 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
6849 * function.c (gimplify_parameters): Call build_alloca_call_expr.
6850 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
6851 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
6852 (in_loop_p): Remove first argument and useless check.
6853 (pass_walloca::execute): Remove useless test and adjust call to above.
6854 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
6855 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
6856 (gimplify_call_expr): Deal with all alloca variants.
6857 * hsa-gen.c (gen_hsa_alloca): Likewise.
6858 (gen_hsa_insns_for_call): Likewise.
6859 * ipa-pure-const.c (special_builtin_state): Likewise.
6860 * tree-chkp.c (chkp_build_returned_bound): Likewise.
6861 * tree-object-size.c (alloc_object_size): Likewise.
6862 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
6863 (call_may_clobber_ref_p_1): Likewise.
6864 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
6865 (ccp_fold_stmt): Likewise.
6866 (optimize_stack_restore): Likewise.
6867 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
6868 (mark_all_reaching_defs_necessary_1): Likewise.
6869 (propagate_necessity): Likewise.
6870 (eliminate_unnecessary_stmts): Likewise.
6871 * tree.c (build_common_builtin_nodes): Build
6872 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
6873 (build_alloca_call_expr): New function.
6874 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
6875 (CASE_BUILT_IN_ALLOCA): Likewise.
6876 (build_alloca_call_expr): Declare.
6877 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
6879 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
6882 * dwarf2out.c (new_die_raw): New static inline function.
6883 (new_die): Use it to create the DIE.
6884 (add_AT_external_die_ref): Likewise.
6885 (clone_die): Likewise.
6886 (clone_as_declaration): Likewise.
6887 (dwarf2out_vms_debug_main_pointer): Likewise.
6888 (base_type_die): Likewise. Remove early return for corner cases.
6889 Do not call add_pubtype on the DIE here.
6890 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
6891 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
6892 typedefs for base types with DW_AT_endianity. Make sure a DIE with
6893 native order exists for base types, attach the DIE manually and call
6894 add_pubtype on it. Do not equate a reverse order DIE to the type.
6896 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
6898 * config/arm/arm.c (align_ok_ldrd_strd): New function.
6899 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
6901 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
6903 2017-10-19 Jakub Jelinek <jakub@redhat.com>
6905 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
6906 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
6907 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
6908 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
6909 * opts.c (sanitizer_opts): Add builtin.
6910 * ubsan.c (instrument_builtin): New function.
6911 (pass_ubsan::execute): Call it.
6912 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
6913 * doc/invoke.texi: Document -fsanitize=builtin.
6915 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
6916 builtins, store max (log2 (align), 0) into uchar field instead of
6917 align into uptr field.
6918 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
6919 store uchar 0 field instead of uptr 0 field.
6920 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
6921 instead of passing one address of struct with 2 locations pass
6922 two addresses of structs with 1 location each.
6923 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
6924 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
6925 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
6926 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
6927 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
6928 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
6929 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
6930 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
6932 2017-10-19 Martin Liska <mliska@suse.cz>
6935 * file-find.c (remove_prefix): Remove.
6936 * file-find.h (remove_prefix): Likewise.
6937 * gcc-ar.c: Remove smartness of lookup.
6939 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
6941 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
6942 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
6943 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
6945 2017-10-19 Jakub Jelinek <jakub@redhat.com>
6948 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
6949 (setcc + and to xor + setcc): New peephole2.
6951 2017-10-19 Tom de Vries <tom@codesourcery.com>
6953 * doc/sourcebuild.texi (Test Directives, Variants of
6954 dg-require-support): Add dg-require-stack-size.
6956 2017-10-19 Martin Liska <mliska@suse.cz>
6959 * gimplify.c (gimplify_decl_expr): Do not instrument variables
6960 that have a large alignment.
6961 (gimplify_target_expr): Likewise.
6963 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
6965 PR rtl-optimization/82602
6966 * ira.c (rtx_moveable_p): Return false for volatile asm.
6968 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
6971 * config/i386/i386-modes.def (CCGZ): New CC mode.
6972 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
6973 * config/i386/predicates.md (ix86_comparison_operator):
6975 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
6976 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
6977 with double-word subtraction.
6978 (put_condition_code): Handle CCGZmode.
6980 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
6982 * wide-int.cc (debug (const wide_int &)): New.
6983 (debug (const wide_int *)): New.
6984 (debug (const widest_int &)): New.
6985 (debug (const widest_int *)): New.
6987 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
6990 * lra-constraints.c (curr_insn_transform): Use non-input operand
6991 instead of output one for matched reload.
6993 2017-10-18 Bin Cheng <bin.cheng@arm.com>
6995 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
6996 (tree-ssa-loop-ivopts.h): New header file.
6997 (struct builtin_info): New fields.
6998 (classify_builtin_1): Compute and record base and offset parts for
6999 memset builtin partition by calling strip_offset.
7000 (offset_cmp, fuse_memset_builtins): New functions.
7001 (finalize_partitions): Fuse adjacent memset partitions by calling
7003 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
7004 Expose the interface.
7005 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
7007 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7009 PR tree-optimization/82574
7010 * tree-loop-distribution.c (find_single_drs): New parameter. Check
7011 that data reference must be executed exactly once per iteration
7012 against the outermost loop in nest.
7013 (classify_partition): Update call to above function.
7015 2017-10-18 Richard Biener <rguenther@suse.de>
7017 PR tree-optimization/82591
7018 * graphite.c (graphite_transform_loops): Move code gen message
7020 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7021 Here. Handle scop_to_isl_ast failing.
7022 (scop_to_isl_ast): Limit the number of ISL operations.
7024 2017-10-18 Richard Biener <rguenther@suse.de>
7026 * graphite-isl-ast-to-gimple.c
7027 (translate_isl_ast_to_gimple::set_rename): Simplify.
7028 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
7029 (graphite_copy_stmts_from_block): ... here.
7030 (copy_bb_and_scalar_dependences): Simplify.
7031 (add_parameters_to_ivs_params): Canonicalize.
7032 (generate_entry_out_of_ssa_copies): Simplify.
7033 * graphite-sese-to-poly.c (extract_affine_name): Simplify
7034 by passing in ISL dimension.
7035 (parameter_index_in_region_1): Rename to ...
7036 (parameter_index_in_region): ... this.
7037 (extract_affine): Adjust assert, pass down parameter index.
7038 (add_param_constraints): Use range-info when available.
7039 (build_scop_context): Adjust.
7040 * sese.c (new_sese_info): Adjust.
7041 (free_sese_info): Likewise.
7042 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
7043 Remove unused typedefs.
7044 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
7046 2017-10-18 Martin Liska <mliska@suse.cz>
7048 * combine.c (simplify_compare_const): Add gcc_fallthrough.
7050 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7052 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
7053 (s390_sched_init): Do not reset s390_sched_state if we entered the
7054 current basic block via a fallthru edge and all others are unlikely.
7056 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7058 * config/s390/s390.c (NUM_SIDES): New variable.
7059 (LONGRUNNING_THRESHOLD): New variable.
7060 (LATENCY_FACTOR): New variable.
7061 (s390_sched_score): Decrease score for long-running instructions on
7063 (s390_sched_variable_issue): Perform bookkeeping for long-running
7066 2017-10-18 Richard Biener <rguenther@suse.de>
7068 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
7069 Simplify with removal of the parameter rename map.
7070 (set_rename): Likewise.
7071 (should_copy_to_new_region): Likewise.
7072 (graphite_copy_stmts_from_block): Likewise.
7073 (copy_bb_and_scalar_dependences): Remove initialization of
7074 unused copied_bb_map.
7076 (copy_internal_parameters): Likewise.
7077 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
7078 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
7079 Use INTEGRAL_TYPE_P.
7080 (parameter_index_in_region_1): Rename to ...
7081 (assign_parameter_index_in_region): ... this. Assert we have
7082 a parameter we handle.
7083 (scan_tree_for_params): Adjust.
7084 * sese.h (parameter_rename_map_t): Remove.
7085 (struct sese_info_t): Remove unused parameter_rename_map and
7086 copied_bb_map members.
7087 * sese.c (new_sese_info): Adjust.
7088 (free_sese_info): Likewise.
7090 2017-10-18 Martin Liska <mliska@suse.cz>
7093 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
7094 on an abnormal edge.
7096 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
7098 * doc/invoke.texi (ffunction-sections and fdata-sections):
7101 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
7103 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
7104 the use statement can throw internally.
7106 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
7108 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
7109 any RTX present on the RHS of a SET.
7110 * compare-elim.c (try_eliminate_compare): Restore comment.
7112 2017-10-17 Jakub Jelinek <jakub@redhat.com>
7114 * langhooks.h (struct lang_hooks): Document that tree_size langhook
7115 may be also called on tcc_type nodes.
7116 * langhooks.c (lhd_tree_size): Likewise.
7118 2017-10-17 David Malcolm <dmalcolm@redhat.com>
7120 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
7121 format_warning_at_substring.
7122 (maybe_warn): Convert source_range * param to a location_t. Pass
7123 UNKNOWN_LOCATION rather than NULL to fmtwarn.
7124 (format_directive): Remove code to extract source_ranges and
7125 source_range * in favor of just a location_t.
7126 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
7128 * substring-locations.c (format_warning_va): Convert
7129 source_range * param to a location_t.
7130 (format_warning_at_substring): Likewise.
7131 * substring-locations.h (format_warning_va): Likewise.
7132 (format_warning_at_substring): Likewise.
7134 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
7136 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
7138 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
7139 and scatter/gather ops.
7141 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
7142 vec_gather_load and vec_scatter_store.
7143 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
7144 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
7145 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
7146 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
7147 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
7148 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
7150 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
7152 * reg-stack.c (compare_for_stack_reg): Add bool argument.
7153 Detect FTST instruction and handle its register pops. Only pop
7154 second operand if can_pop_second_op is true.
7155 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
7156 set can_pop_second_op to false in the compare_for_stack_reg call.
7158 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
7159 output_fp_compare for stack register operands.
7160 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
7161 instructions here. Do not emit stack register pops here. Assert
7162 that FCOMPP pops next to top stack register. Rewrite function.
7164 2017-10-17 Nathan Sidwell <nathan@acm.org>
7167 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
7168 use DECL_ASSEMBLER_NAME_RAW.
7171 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
7174 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
7175 Wilco Dijkstra <wilco.dijkstra@arm.com>
7177 * builtins.c (expand_builtin_update_setjmp_buf): Add a
7178 converstion to Pmode from the buf_addr.
7180 2017-10-17 Richard Biener <rguenther@suse.de>
7182 * graphite-dependences.c (scop_get_reads_and_writes): Change
7183 output parameters to references.
7185 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
7187 PR 71026/tree-optimization
7188 * fold-const.c (distribute_real_division): Removed.
7189 (fold_binary_loc): Remove calls to distribute_real_divison.
7191 2017-10-17 Richard Biener <rguenther@suse.de>
7193 * graphite-scop-detection.c
7194 (scop_detection::stmt_has_simple_data_refs_p): Always use
7195 the full nest as region.
7196 (try_generate_gimple_bb): Likewise.
7197 * sese.c (scalar_evolution_in_region): Simplify now that
7198 SCEV can handle instantiation in regions.
7199 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
7200 in the non-loop part of a function if requested.
7202 2017-10-17 Richard Biener <rguenther@suse.de>
7204 PR tree-optimization/82563
7205 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
7207 (graphite_regenerate_ast_isl): Call it.
7208 * graphite-scop-detection.c (build_scops): Remove entry edge split.
7210 2017-10-17 Jakub Jelinek <jakub@redhat.com>
7212 PR tree-optimization/82549
7213 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
7214 Formatting fixes. Instead of calling make_bit_field_ref with negative
7217 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
7219 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
7220 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
7221 _mm_maskz_reduce_ss): New.
7222 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
7223 __builtin_ia32_reducess_mask): Ditto..
7224 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
7225 * config/i386/sse.md (reduces<mode>): Renamed to ...
7226 (reduces<mode><mask_scalar_name>): ... this.
7227 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
7229 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
7230 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
7232 2017-10-16 David Malcolm <dmalcolm@redhat.com>
7234 * Makefile.in (OBJS): Add unique-ptr-tests.o.
7235 * selftest-run-tests.c (selftest::run_tests): Call
7236 selftest::unique_ptr_tests_cc_tests.
7237 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
7238 * unique-ptr-tests.cc: New file.
7240 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
7243 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7245 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7246 (make_hard_regno_born, make_hard_regno_dead): Update
7247 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
7249 2017-10-16 Jeff Law <law@redhat.com>
7251 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7253 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
7255 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
7257 2017-10-16 Olivier Hainque <hainque@adacore.com>
7259 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
7260 with_cpu if we were configured for an e500v2 target cpu name.
7262 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
7264 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
7265 * doc/invoke.texi: Document +nodsp as a valid extension for
7268 2017-10-16 Martin Liska <mliska@suse.cz>
7270 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
7271 (test_set_range): Likewise.
7272 (test_range_functions): Rename to ...
7273 (test_bit_in_range): ... this.
7274 (sbitmap_c_tests): Add new test.
7276 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7278 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
7280 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
7281 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
7283 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7285 * config/aarch64/aarch64-builtins.c
7286 (aarch64_types_quadopu_lane_qualifiers): New.
7287 (TYPES_QUADOPU_LANE): New.
7288 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
7289 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
7290 (aarch64_<sur>dot_laneq<vsi2qi>): New.
7291 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
7292 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
7293 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
7294 (Vdottype, DOTPROD): New.
7295 (sur): Add SDOT and UDOT.
7297 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7299 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
7300 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
7301 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
7303 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
7304 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
7305 Enable TARGET_DOTPROD.
7306 (cortex-a75.cortex-a55): Likewise.
7307 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
7309 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7311 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
7312 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
7314 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
7316 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
7317 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
7318 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
7319 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
7320 * config/arm/types.md (neon_dot, neon_dot_q): New.
7321 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
7323 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7325 * config/arm/arm.h (TARGET_DOTPROD): New.
7326 * config/arm/arm.c (arm_arch_dotprod): New.
7327 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
7328 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
7329 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
7330 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
7331 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
7332 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
7333 * doc/invoke.texi (armv8.2-a): Document dotprod
7335 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7337 * i386.c (ix86_vec_cost): New function.
7338 (ix86_rtx_costs): Handle vector operations better.
7339 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
7340 * x86-tune-costs.h: Add new costs to all tables.
7342 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7344 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
7346 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
7347 divsd, sqrtss and sqrtsd
7348 * x86-tune-costs.h: Add new entries to all costs.
7349 (znver1_cost): Fix to match real instruction latencies.
7351 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7352 Michael Collison <michael.collison@arm.com>
7354 * compare-elim.c: Include emit-rtl.h.
7355 (can_merge_compare_into_arith): New function.
7356 (try_validate_parallel): Likewise.
7357 (try_merge_compare): Likewise.
7358 (try_eliminate_compare): Call the above when no previous clobber
7360 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
7363 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7367 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
7368 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
7372 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
7373 to be any operand_equal_p operands. For & (B - 1) require
7374 B to be power of 2. Recognize
7375 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
7377 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7380 * optabs.c (expand_memory_blockage): Fix call of
7381 targetm.have_memory_blockage.
7383 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7386 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
7387 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
7388 objects to extra_objs instead of overwriting it.
7390 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7392 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
7393 Use any_fp_register_operand as operand[3] predicate. Simplify
7394 equality test for operands[2] and operands[4] memory location.
7395 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
7396 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
7397 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
7398 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
7399 any_fp_register_operand as operand[1] predicate. Simplify
7400 equality test for operands[0] and operands[3] memory location.
7401 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
7402 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
7403 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
7405 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7407 * target-insns.def: Add memory_blockage.
7408 * optabs.c (expand_memory_blockage): New function.
7409 (expand_asm_memory_barrier): Rename ...
7410 (expand_asm_memory_blockage): ... to this.
7411 (expand_mem_thread_fence): Call expand_memory_blockage
7412 instead of expand_asm_memory_barrier.
7413 (expand_mem_singnal_fence): Ditto.
7414 (expand_atomic_load): Ditto.
7415 (expand_atomic_store): Ditto.
7416 * doc/md.texi (Standard Pattern Names For Generation):
7417 Document memory_blockage instruction pattern.
7419 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
7421 * config/rl78/rl78.c (rl78_emit_libcall): New function.
7422 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
7423 * config/rl78/rl78.md: New define_expand "adddi3".
7425 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7427 * cfghooks.c (verify_flow_info): Disable check that all probabilities
7430 2017-10-13 Jeff Law <law@redhat.com>
7432 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
7434 2017-10-13 Jakub Jelinek <jakub@redhat.com>
7437 * internal-fn.c (expand_mul_overflow): If both operands have
7438 the same highpart of -1 or 0 and the topmost bit of lowpart
7439 is different, overflow is if res <= 0 rather than res < 0.
7441 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
7443 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
7444 TARGET_P9_VECTOR code for unaligned_load case.
7446 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7448 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
7450 2017-10-13 Nathan Sidwell <nathan@acm.org>
7452 * tree-core.h (tree_contains_struct): Make bool.
7453 * tree.c (tree_contains_struct): Likewise.
7454 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
7455 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
7456 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
7457 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
7458 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
7460 2017-10-13 Richard Biener <rguenther@suse.de>
7462 * graphite-isl-ast-to-gimple.c
7463 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
7464 parameters and dominance check.
7465 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
7466 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
7467 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
7468 Do not update SSA form here or do intermediate IL verification.
7469 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
7470 (graphite_initialize): Remove check on the number of loops in
7471 the function and inline into graphite_transform_loops.
7472 (graphite_finalize): Inline into graphite_transform_loops.
7473 (graphite_transform_loops): Perform SSA update and IL verification
7475 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
7477 2017-10-13 Richard Biener <rguenther@suse.de>
7479 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
7480 graphite_expression_type_precision): Avoid global constructor
7482 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
7483 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
7484 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
7485 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
7486 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
7487 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
7489 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
7492 * config/i386/i386.h (ix86_red_zone_size): New.
7493 * config/i386/i386.md (push peephole2s): Replace
7494 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
7496 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7497 Alan Hayward <alan.hayward@arm.com>
7498 David Sherwood <david.sherwood@arm.com>
7500 * combine.c (can_change_dest_mode): Reject changes in
7501 REGMODE_NATURAL_SIZE.
7503 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7504 Alan Hayward <alan.hayward@arm.com>
7505 David Sherwood <david.sherwood@arm.com>
7507 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
7508 (expand_debug_source_expr): Likewise.
7509 * combine.c (combine_simplify_rtx): Likewise.
7510 * cse.c (fold_rtx): Likewise.
7511 * fwprop.c (canonicalize_address): Likewise.
7512 * targhooks.c (default_shift_truncation_mask): Likewise.
7514 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7515 Alan Hayward <alan.hayward@arm.com>
7516 David Sherwood <david.sherwood@arm.com>
7518 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
7519 (widened_mode): Likewise.
7520 (expand_unop): Likewise.
7521 * ree.c (transform_ifelse): Likewise.
7522 (merge_def_and_ext): Likewise.
7523 (combine_reaching_defs): Likewise.
7524 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7526 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7527 Alan Hayward <alan.hayward@arm.com>
7528 David Sherwood <david.sherwood@arm.com>
7530 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
7531 * combine.c (gen_lowpart_for_combine): Likewise.
7532 * dwarf2out.c (rtl_for_decl_location): Likewise.
7533 * final.c (alter_subreg): Likewise.
7534 * rtlhooks.c (gen_lowpart_general): Likewise.
7535 (gen_lowpart_if_possible): Likewise.
7537 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7538 Alan Hayward <alan.hayward@arm.com>
7539 David Sherwood <david.sherwood@arm.com>
7541 * calls.c (expand_call): Use subreg_lowpart_offset.
7542 * cse.c (cse_insn): Likewise.
7543 * regcprop.c (copy_value): Likewise.
7544 (copyprop_hardreg_forward_1): Likewise.
7546 2017-10-13 Jakub Jelinek <jakub@redhat.com>
7549 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
7550 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
7551 =Q constraints to +Q and into insn condition add check
7552 that operands[0] and operands[1] are equal.
7553 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
7554 =Q constraints to +Q and into insn condition add check
7555 that operands[0] is equal to either operands[1] or operands[2].
7558 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
7559 instead of handling MINUS_EXPR twice (once for each argument),
7560 canonicalize operand order and handle just once, use rtype where
7561 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
7564 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
7565 any values of __C while still being pattern recognizable as a simple
7568 2017-10-13 Richard Biener <rguenther@suse.de>
7570 PR tree-optimization/82451
7572 2017-10-02 Richard Biener <rguenther@suse.de>
7574 PR tree-optimization/82355
7575 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7576 a mapping for the enclosing loop but avoid generating one for
7578 (copy_bb_and_scalar_dependences): Remove premature codegen
7579 error on PHIs in blocks duplicated into multiple places.
7580 * graphite-scop-detection.c
7581 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7582 in the region use it as loop and nest to analyze the DR in.
7583 (try_generate_gimple_bb): Likewise.
7584 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7585 (add_loop_constraints): For blocks in a loop not in the region
7586 create a dimension with a single iteration.
7587 * sese.h (gbb_loop_at_index): Remove assert.
7589 * cfgloop.c (loop_preheader_edge): For the loop tree root
7590 return the single successor of the entry block.
7591 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7592 Reset the SCEV hashtable and niters.
7593 * graphite-scop-detection.c
7594 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
7595 assert that we only have POLYNOMIAL_CHREC that vary in loops
7596 contained in the region.
7597 (scop_detection::graphite_can_represent_expr): Adjust.
7598 (scop_detection::stmt_has_simple_data_refs_p): For loops
7599 not in the region set loop to NULL. The nest is now the
7600 entry edge to the region.
7601 (try_generate_gimple_bb): Likewise.
7602 * sese.c (scalar_evolution_in_region): Adjust for
7603 instantiate_scev change.
7604 * tree-data-ref.h (graphite_find_data_references_in_stmt):
7605 Make nest parameter the edge into the region.
7606 (create_data_ref): Likewise.
7607 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
7608 entry edge into a region and adjust instantiate_scev calls.
7609 (create_data_ref): Likewise.
7610 (graphite_find_data_references_in_stmt): Likewise.
7611 (find_data_references_in_stmt): Pass the loop preheader edge
7612 from the nest argument.
7613 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
7614 parameter the edge into the region.
7615 (instantiate_parameters): Use the loop preheader edge as entry.
7616 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
7618 (get_instantiated_value_entry): Make instantiate_below parameter
7619 the edge into the region.
7620 (instantiate_scev_name): Likewise. Adjust dominance checks,
7621 when we cannot use loop-based instantiation instantiate by
7622 walking use-def chains.
7623 (instantiate_scev_poly): Adjust.
7624 (instantiate_scev_binary): Likewise.
7625 (instantiate_scev_convert): Likewise.
7626 (instantiate_scev_not): Likewise.
7627 (instantiate_array_ref): Remove.
7628 (instantiate_scev_3): Likewise.
7629 (instantiate_scev_2): Likewise.
7630 (instantiate_scev_1): Likewise.
7631 (instantiate_scev_r): Do not blindly handle N-operand trees.
7632 Do not instantiate array-refs. Handle all constants and invariants.
7633 (instantiate_scev): Make instantiate_below parameter
7634 the edge into the region.
7635 (resolve_mixers): Use the loop preheader edge for the region
7636 parameter to instantiate_scev_r.
7637 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
7639 2017-10-13 Richard Biener <rguenther@suse.de>
7641 PR tree-optimization/82525
7642 * graphite-isl-ast-to-gimple.c
7643 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
7645 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
7646 Fail code generation when we cannot represent the isl integer.
7647 (binary_op_to_tree): Elide modulo operations that are no-ops
7648 in the type we code generate. Remove now superfluous code
7651 2017-10-13 Richard Biener <rguenther@suse.de>
7653 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
7654 (scop_detection::harmful_loop_in_region): Remove premature
7655 IV type restriction.
7656 (scop_detection::graphite_can_represent_scev): We can handle
7657 pointer IVs just fine.
7659 2017-10-13 Alan Modra <amodra@gmail.com>
7661 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
7662 "Clobbers and Scratch Registers". Add paragraph on
7663 alternative to clobbers for scratch registers and OpenBLAS
7666 2017-10-13 Alan Modra <amodra@gmail.com>
7668 * doc/extend.texi (Clobbers): Correct vax example. Delete old
7669 example of a memory input for a string of known length. Move
7670 commentary out of table. Add a number of new examples
7671 covering array memory inputs.
7673 2017-10-12 Martin Liska <mliska@suse.cz>
7675 PR tree-optimization/82493
7676 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
7677 (test_range_functions): New function.
7678 (sbitmap_c_tests): Likewise.
7679 * selftest-run-tests.c (selftest::run_tests): Run new tests.
7680 * selftest.h (sbitmap_c_tests): New function.
7682 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7684 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7686 * config/rs6000/amo.h: Fix spacing issue.
7688 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7691 * config/i386/i386.md (*ashl<mode>3_mask_1,
7692 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
7693 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
7696 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
7698 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
7699 (profile_probability): Set max_probability
7700 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
7702 * profile-count.c (profile_probability::differs_from_p): Do not
7703 rely on max_probaiblity == 10000
7705 2017-10-12 Jeff Law <law@redhat.com>
7707 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
7710 2017-10-12 Martin Sebor <msebor@redhat.com>
7714 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
7715 (handle_alias_pairs): Call it.
7716 * common.opt (-Wattribute-alias): New option.
7717 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
7718 * doc/invoke.texi (-Wattribute-alias): Document.
7720 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
7723 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
7725 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7727 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7728 (make_hard_regno_born, make_hard_regno_dead): Update
7729 bb_killed_pseudos and bb_gen_pseudos.
7731 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
7733 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
7735 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
7737 * config/alpha/alpha.c (alpha_split_conditional_move):
7738 Use std::swap instead of manually swapping.
7739 (alpha_stdarg_optimize_hook): Ditto.
7740 (alpha_canonicalize_comparison): Ditto.
7742 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7744 * tree-loop-distribution.c (struct builtin_info): New struct.
7745 (struct partition): Refactor fields into struct builtin_info.
7746 (partition_free): Free struct builtin_info.
7747 (build_size_arg_loc, build_addr_arg_loc): Delete.
7748 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
7749 information from struct builtin_info.
7750 (find_single_drs): New function refactored from classify_partition.
7751 Also moved builtin validity checks to this function.
7752 (compute_access_range, alloc_builtin): New functions.
7753 (classify_builtin_st, classify_builtin_ldst): New functions.
7754 (classify_partition): Refactor code into functions find_single_drs,
7755 classify_builtin_st and classify_builtin_ldst.
7756 (distribute_loop): Don't do runtime alias check when distributing
7758 (find_seed_stmts_for_distribution): New function.
7759 (pass_loop_distribution::execute): Refactor code finding seed
7760 stmts into above function. Support distribution for the innermost
7761 two-level loop nest. Adjust dump information.
7763 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7765 * tree-loop-distribution.c: Adjust the general comment.
7766 (NUM_PARTITION_THRESHOLD): New macro.
7767 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
7768 (classify_partition): Skip builtin pattern of loop nest's inner loop.
7769 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
7770 in call to build_partition_graph.
7771 (finalize_partitions): New parameter. Make loop distribution more
7772 conservative by fusing more partitions.
7773 (distribute_loop): Don't do runtime alias check in case of loop nest
7775 (find_seed_stmts_for_distribution): New function.
7776 (prepare_perfect_loop_nest): New function.
7777 (pass_loop_distribution::execute): Refactor code finding seed stmts
7778 and loop nest into above functions. Support loop nest distribution.
7779 Adjust dump information accordingly.
7781 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7783 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
7784 and set PTYPE_SEQUENTIAL for merged partition.
7786 2017-10-12 Richard Biener <rguenther@suse.de>
7788 PR tree-optimization/69728
7790 2017-09-19 Richard Biener <rguenther@suse.de>
7792 PR tree-optimization/69728
7793 * graphite-sese-to-poly.c (schedule_error): New global.
7794 (add_loop_schedule): Handle empty domain by failing the
7796 (build_original_schedule): Handle schedule_error.
7798 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
7799 domain by returning an unchanged schedule.
7801 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7803 * genrecog.c (validate_pattern): For VEC_SELECT verify that
7804 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
7806 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
7808 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
7811 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7814 * expr.c (store_field): Don't optimize away bitsize == 0 store
7815 from CALL_EXPR with addressable return type.
7817 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
7819 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
7820 * config/rs6000/rs6000.md (sel): Delete mode attribute.
7821 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
7822 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
7823 TARGET_ISEL instead of TARGET_ISEL<sel>.
7825 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
7827 * config/rs6000/rs6000.c
7828 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
7830 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
7832 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
7834 (reg_or_cint_operand): Fix comment.
7835 (reg_or_zero_operand): New predicate.
7836 * config/rs6000/rs6000-protos.h (output_isel): Delete.
7837 * config/rs6000/rs6000.c (output_isel): Delete.
7838 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
7839 instead of reg_or_cint_operand. Output instruction directly (not via
7841 (isel_unsigned_<mode>): Ditto.
7842 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
7843 gpc_reg_operand. Add an instruction alternative for this. Output
7844 instruction directly.
7845 (*isel_reversed_unsigned_<mode>): Ditto.
7847 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7849 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
7850 (TARGET_CANONICALIZE_COMPARISON): Define.
7852 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
7855 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
7856 Check whether the dest is REG before adding REG_EQUIV note.
7858 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
7861 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7863 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7864 (make_hard_regno_born, make_hard_regno_dead): Update
7865 bb_killed_pseudos and bb_gen_pseudos.
7867 2017-10-11 Nathan Sidwell <nathan@acm.org>
7869 * incpath.h (enum incpath_kind): Name enum, prefix values.
7870 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
7871 * incpath.c (heads, tails): Use INC_MAX.
7872 (add_env_var_paths, add_standard_paths): Use incpath_kind.
7873 (merge_include_chains, split_quote_chain,
7874 register_include_chains): Update incpath_kind names.
7875 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
7876 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
7878 (add_framework_path, darwin_register_objc_includes): Likewise.
7879 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
7881 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7883 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
7884 Do not use float_operator operator predicate.
7885 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
7886 * config/i386/predicates.md (float_operator): Remove predicate.
7888 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7890 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
7891 (*jccxf_i387): Ditto.
7892 (*jcc<mode>_i387): Ditto.
7893 (*jccu<mode>_i387): Ditto.
7894 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
7895 (*jcc_*_i387 splitters): Remove.
7896 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
7897 * config/i386/i386.c (ix86_split_fp_branch): Remove.
7898 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
7901 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
7903 * profile-count.h (slow_safe_scale_64bit): New function.
7904 (safe_scale_64bit): New inline.
7905 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
7906 * profile-count.c: Include wide-int.h
7907 (slow_safe_scale_64bit): New.
7909 2017-10-11 Nathan Sidwell <nathan@acm.org>
7911 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
7912 HAS_DECL_ASSEMBLER_NAME_P.
7913 * gimple-expr.c (gimple_decl_printable_name: Check
7914 HAS_DECL_ASSEMBLER_NAME_P too.
7915 * ipa-utils.h (type_in_anonymous_namespace_p): Check
7916 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
7917 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
7918 * passes.c (rest_of_decl_compilation): Check
7919 HAS_DECL_ASSEMBLER_NAME_P too.
7920 * recog.c (verify_changes): Likewise.
7921 * tree-pretty-print.c (dump_decl_name): Likewise.
7922 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
7924 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
7925 (SET_DECL_ASSEMBLER_NAME): Use it.
7926 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
7927 (COPY_DECL_ASSEMBLER_NAME): Likewise.
7928 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
7930 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
7932 * config.gcc (i386, x86_64): Add extra objects.
7933 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
7934 (ix86_min_insn_size): Declare.
7935 (ix86_issue_rate): Declare.
7936 (ix86_adjust_cost): Declare.
7937 (ia32_multipass_dfa_lookahead): Declare.
7938 (ix86_macro_fusion_p): Declare.
7939 (ix86_macro_fusion_pair_p): Declare.
7940 (ix86_bd_has_dispatch): Declare.
7941 (ix86_bd_do_dispatch): Declare.
7942 (ix86_core2i7_init_hooks): Declare.
7943 (ix86_atom_sched_reorder): Declare.
7944 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
7945 (COSTS_N_BYTES): Move to x86-tune-costs.h.
7946 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
7947 (rip_relative_addr_p): Rename to ...
7948 (ix86_rip_relative_addr_p): ... this one; export.
7949 (memory_address_length): Update.
7950 (ix86_issue_rate): Move to x86-tune-sched.c.
7951 (ix86_flags_dependent): Move to x86-tune-sched.c.
7952 (ix86_agi_dependent): Move to x86-tune-sched.c.
7953 (exact_dependency_1): Move to x86-tune-sched.c.
7954 (exact_store_load_dependency): Move to x86-tune-sched.c.
7955 (ix86_adjust_cost): Move to x86-tune-sched.c.
7956 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
7957 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
7958 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
7959 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
7960 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
7961 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
7962 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
7963 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
7964 (min_insn_size): Rename to ...
7965 (ix86_min_insn_size): ... this one; export.
7966 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
7967 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
7968 (core2i7_first_cycle_multipass_backtrack): Move to
7969 x86-tune-sched-core.c.
7970 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
7971 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
7972 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
7973 (ix86_avoid_jump_mispredicts): Update.
7974 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
7975 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
7976 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
7977 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
7978 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
7979 (MAX_INSN): Move to ix86-tune-sched-bd.c.
7980 (MAX_IMM): Move to ix86-tune-sched-bd.c.
7981 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
7982 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
7983 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
7984 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
7985 (MAX_STORE): Move to ix86-tune-sched-bd.c.
7986 (BIG): Move to ix86-tune-sched-bd.c.
7987 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
7988 (enum insn_path): Move to ix86-tune-sched-bd.c.
7989 (get_mem_group): Move to ix86-tune-sched-bd.c.
7990 (is_cmp): Move to ix86-tune-sched-bd.c.
7991 (dispatch_violation): Move to ix86-tune-sched-bd.c.
7992 (is_branch): Move to ix86-tune-sched-bd.c.
7993 (is_prefetch): Move to ix86-tune-sched-bd.c.
7994 (init_window): Move to ix86-tune-sched-bd.c.
7995 (allocate_window): Move to ix86-tune-sched-bd.c.
7996 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
7997 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
7998 (process_end_window): Move to ix86-tune-sched-bd.c.
7999 (allocate_next_window): Move to ix86-tune-sched-bd.c.
8000 (find_constant): Move to ix86-tune-sched-bd.c.
8001 (get_num_immediates): Move to ix86-tune-sched-bd.c.
8002 (has_immediate): Move to ix86-tune-sched-bd.c.
8003 (get_insn_path): Move to ix86-tune-sched-bd.c.
8004 (get_insn_group): Move to ix86-tune-sched-bd.c.
8005 (count_num_restricted): Move to ix86-tune-sched-bd.c.
8006 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
8007 (add_insn_window): Move to ix86-tune-sched-bd.c.
8008 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
8009 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
8010 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
8011 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
8012 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
8013 (do_dispatch): Move to ix86-tune-sched-bd.c.
8014 (has_dispatch): Move to ix86-tune-sched-bd.c.
8015 * config/i386/t-i386: Add new object files.
8016 * config/i386/x86-tune-costs.h: New file.
8017 * config/i386/x86-tune-sched-atom.c: New file.
8018 * config/i386/x86-tune-sched-bd.c: New file.
8019 * config/i386/x86-tune-sched-core.c: New file.
8020 * config/i386/x86-tune-sched.c: New file.
8022 2017-10-11 Liu Hao <lh_mouse@126.com>
8024 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
8025 the generic version below instead.
8026 (should_colorize): Recognize Windows consoles as terminals
8028 * pretty-print.c [__MINGW32__] (write_all): New function.
8029 [__MINGW32__] (find_esc_head): Likewise.
8030 [__MINGW32__] (find_esc_terminator): Likewise.
8031 [__MINGW32__] (eat_esc_sequence): Likewise.
8032 [__MINGW32__] (mingw_ansi_fputs): New function that handles
8034 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
8037 2017-10-11 Richard Biener <rguenther@suse.de>
8039 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
8040 Properly call analyze_scalar_evolution with the loop of the stmt.
8042 2017-10-11 Richard Biener <rguenther@suse.de>
8044 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
8045 * tree-core.h (tree_base): Add chrec_var union member.
8046 * tree.h (CHREC_VAR): Remove.
8047 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
8048 * tree-chrec.h (build_polynomial_chrec): Adjust.
8049 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
8050 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
8052 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
8054 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
8055 * match.pd: ... here.
8056 ((T) X == (T) Y): Relax condition.
8058 2017-10-11 Bin Cheng <bin.cheng@arm.com>
8060 PR tree-optimization/82472
8061 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
8063 (break_alias_scc_partitions): Update postorder number.
8065 2017-10-11 Martin Liska <mliska@suse.cz>
8068 * opts.c (parse_no_sanitize_attribute): Do not use error_value
8070 * opts.h (parse_no_sanitize_attribute): Remove last argument.
8072 2017-10-11 Martin Liska <mliska@suse.cz>
8074 * print-rtl.c (print_insn): Move declaration of idbuf
8075 to same scope as name.
8077 2017-10-11 Martin Liska <mliska@suse.cz>
8082 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8085 2017-10-11 Martin Liska <mliska@suse.cz>
8088 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8091 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
8093 * config.gcc (default_gnu_indirect_function): Default to yes for
8094 arm*-*-linux* with glibc.
8096 2017-10-11 Richard Biener <rguenther@suse.de>
8098 * tree-scalar-evolution.c (get_scalar_evolution): Handle
8099 default-defs and types we do not want to analyze.
8100 (interpret_loop_phi): Replace unreachable code with an assert.
8101 (compute_scalar_evolution_in_loop): Remove and inline ...
8102 (analyze_scalar_evolution_1): ... here, replacing condition with
8103 what makes the intent clearer. Remove handling of cases
8104 get_scalar_evolution now handles.
8106 2017-10-10 Jim Wilson <wilson@tuliptree.org>
8108 PR rtl-optimization/81434
8109 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
8110 comment for main loop. In sched_group_found if, also add checks for
8111 pass and min_cost_group.
8113 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
8115 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
8116 (rs6000_insn_cost): New function.
8117 * config/rs6000/rs6000.md (cost): New attribute.
8119 2017-10-10 Jakub Jelinek <jakub@redhat.com>
8120 H.J. Lu <hongjiu.lu@intel.com>
8124 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
8125 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
8126 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
8127 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
8128 to OPTION_MASK_ISA_AVX512VL - builtins that have both
8129 OPTION_MASK_ISA_MMX and some other bit set require both
8130 mmx and the ISAs without the mmx bit.
8131 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
8132 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
8133 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
8134 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
8135 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
8136 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
8137 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
8138 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
8139 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
8140 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
8141 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
8142 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
8143 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
8144 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
8145 Add OPTION_MASK_ISA_MMX.
8147 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
8149 * config.gcc (armv7*-*-freebsd*): New target.
8150 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
8152 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
8154 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
8155 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
8158 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
8160 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
8161 a template parameter.
8162 (WIDE_INT_REF_FOR): Update accordingly.
8163 * tree.h (wi::int_traits <const_tree>): Delete.
8164 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
8165 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
8166 (wi::tree_to_wide_ref): New typedef.
8167 (wi::to_wide): New function.
8168 * calls.c (get_size_range): Use wi::to_wide when operating on
8170 * cgraph.c (cgraph_node::create_thunk): Likewise.
8171 * config/i386/i386.c (ix86_data_alignment): Likewise.
8172 (ix86_local_alignment): Likewise.
8173 * dbxout.c (stabstr_O): Likewise.
8174 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
8175 * expr.c (const_vector_from_tree): Likewise.
8176 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
8177 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
8178 (fold_negate_expr_1, int_const_binop_1, const_binop)
8179 (fold_convert_const_int_from_real, optimize_bit_field_compare)
8180 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
8181 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
8182 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
8183 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
8184 (fold_not_const, round_up_loc): Likewise.
8185 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
8186 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
8187 (alloca_call_type): Likewise.
8188 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
8189 * godump.c (go_output_typedef): Likewise.
8190 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
8191 * internal-fn.c (get_min_precision): Likewise.
8192 * ipa-cp.c (ipcp_store_vr_results): Likewise.
8193 * ipa-polymorphic-call.c
8194 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
8195 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
8196 (ipa_modify_call_arguments): Likewise.
8197 * match.pd: Likewise.
8198 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
8199 * print-tree.c (print_node_brief, print_node): Likewise.
8200 * stmt.c (expand_case): Likewise.
8201 * stor-layout.c (layout_type): Likewise.
8202 * tree-affine.c (tree_to_aff_combination): Likewise.
8203 * tree-cfg.c (group_case_labels_stmt): Likewise.
8204 * tree-data-ref.c (dr_analyze_indices): Likewise.
8205 (prune_runtime_alias_test_list): Likewise.
8206 * tree-dump.c (dequeue_and_dump): Likewise.
8207 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
8208 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
8209 * tree-pretty-print.c (dump_generic_node): Likewise.
8210 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
8211 (simple_iv_with_niters): Likewise.
8212 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
8213 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
8214 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
8215 * tree-ssa-loop-niter.c (split_to_var_and_offset)
8216 (refine_value_range_using_guard, number_of_iterations_ne_max)
8217 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
8218 (get_cst_init_from_scev, record_nonwrapping_iv)
8219 (scev_var_range_cant_overflow): Likewise.
8220 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
8221 * tree-ssa-pre.c (compute_avail): Likewise.
8222 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
8223 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
8224 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
8225 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
8226 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
8227 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
8228 (dump_case_nodes, try_switch_expansion): Likewise.
8229 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
8230 (vect_do_peeling): Likewise.
8231 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
8232 * tree-vect-stmts.c (vectorizable_load): Likewise.
8233 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
8234 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
8235 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
8236 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
8237 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
8238 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
8239 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
8240 (evrp_dom_walker::before_dom_children): Likewise.
8241 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
8242 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
8243 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
8244 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
8245 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
8246 (get_range_pos_neg): Likewise.
8247 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
8248 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
8249 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
8250 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
8251 * config/avr/avr.c (avr_fold_builtin): Likewise.
8252 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
8253 * config/msp430/msp430.c (msp430_attr): Likewise.
8254 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
8255 * config/powerpcspe/powerpcspe-c.c
8256 (altivec_resolve_overloaded_builtin): Likewise.
8257 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
8258 (rs6000_expand_ternop_builtin): Likewise.
8259 * config/rs6000/rs6000-c.c
8260 (altivec_resolve_overloaded_builtin): Likewise.
8261 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
8262 (rs6000_expand_ternop_builtin): Likewise.
8263 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
8265 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8267 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
8268 when copying loop nest with only one inner loop.
8270 2017-10-10 Richard Biener <rguenther@suse.de>
8272 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
8273 blocks if SCEV is active.
8274 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
8276 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
8277 (scev_initialize): Assert we are not yet initialized.
8279 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8281 * tree-loop-distribution.c (generate_loops_for_partition): Remove
8282 inner loop's exit stmt by making it always exit the loop, otherwise
8283 we would generate an infinite empty loop.
8285 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8287 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
8288 renaming variables in new preheader if it's deleted.
8290 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8292 * tree-loop-distribution.c (struct partition): Remove unused field
8293 loops of the structure.
8294 (partition_alloc, partition_free): Ditto.
8295 (build_rdg_partition_for_vertex): Ditto.
8297 2017-10-09 Jeff Law <law@redhat.com>
8299 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
8300 return type to match prototype and documentation.
8302 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8304 * config/rs6000/rs6000.c (processor_costs): Move to ...
8305 * config/rs6000/rs6000.h: ... here.
8306 (rs6000_cost): Declare.
8308 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
8310 * except.c (setjmp_fn): New global variable.
8311 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
8312 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
8313 if DONT_USE_BUILTIN_SETJMP is defined.
8315 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8317 * target.def (insn_cost): New hook.
8318 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
8319 * doc/tm.texi: Regenerate.
8320 * rtlanal.c (insn_cost): Use the new hook.
8322 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8324 * combine.c (combine_validate_cost): Compute the new insn_cost,
8325 not just pattern_cost.
8326 (try_combine): Adjust comment.
8328 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8330 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
8332 * combine.c (uid_insn_cost): Adjust comment.
8333 (combine_validate_cost): Adjust comment. Use pattern_cost instead
8335 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
8336 * dse.c (find_shift_sequence): Ditto.
8337 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
8338 (bb_valid_for_noce_process_p): Use pattern_cost.
8339 * rtl.h (insn_rtx_cost): Delete.
8340 (pattern_cost): New prototype.
8341 (insn_cost): New prototype.
8342 * rtlanal.c (insn_rtx_cost): Rename to...
8343 (pattern_cost): ... this.
8346 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
8348 * config/i386/i386.md (*jcc_2): Remove insn pattern.
8349 (*jcc<mode>_0_r_i387): Ditto.
8350 (*jccxf_r_i387): Ditto.
8351 (*jcc<mode>_r_i387): Ditto.
8352 (*jccu<mode>_r_i387): Ditto.
8353 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
8354 (*jcc): Rename from *jcc_1.
8356 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8358 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
8359 deferred rescans after the lvx/stvx recombination pre-pass.
8361 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
8363 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
8364 memory operation instruction support.
8365 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
8366 (rs6000-ibm-aix[789]*): Likewise.
8367 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
8368 Document new functions.
8370 2017-10-09 Richard Biener <rguenther@suse.de>
8372 PR tree-optimization/82397
8373 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
8374 equality only for semantically equal trees.
8376 2017-10-09 Richard Biener <rguenther@suse.de>
8378 PR tree-optimization/82449
8379 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
8380 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
8381 allow constant addresses.
8382 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
8385 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8387 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
8390 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8393 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
8396 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8399 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
8401 2017-10-09 Jakub Jelinek <jakub@redhat.com>
8404 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
8405 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
8407 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
8409 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
8410 (WI_BINARY_PREDICATE_RESULT): Likewise.
8411 (wi::binary_traits::operator_result): New type.
8412 (wi::binary_traits::predicate_result): Likewise.
8413 (generic_wide_int::operator~, unary generic_wide_int::operator-)
8414 (generic_wide_int::operator==, generic_wide_int::operator!=)
8415 (generic_wide_int::operator&, generic_wide_int::and_not)
8416 (generic_wide_int::operator|, generic_wide_int::or_not)
8417 (generic_wide_int::operator^, generic_wide_int::operator+
8418 (binary generic_wide_int::operator-, generic_wide_int::operator*):
8420 (operator~, unary operator-, operator==, operator!=, operator&)
8421 (operator|, operator^, operator+, binary operator-, operator*): New
8423 * expr.c (get_inner_reference): Use wi::bit_and_not.
8424 * fold-const.c (fold_binary_loc): Likewise.
8425 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
8426 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
8427 (bit_value_binop): Likewise.
8428 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
8429 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
8430 (extract_range_from_binary_expr_1): Likewise.
8431 (masked_increment): Likewise.
8432 (simplify_bit_ops_using_ranges): Likewise.
8434 2017-10-09 Martin Jambor <mjambor@suse.cz>
8437 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
8438 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
8439 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
8441 (hsa_fixup_mov_insn_type): New function.
8442 (hsa_op_with_type::get_in_type): Use it.
8443 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
8444 immediates in an assert.
8445 (hsa_op_with_type::extend_int_to_32bit): New method.
8446 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
8447 types. Convert to dest type if necessary.
8448 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
8449 (reg_for_gimple_ssa): Pass false as min32int to
8450 hsa_type_for_scalar_tree_type.
8451 (gen_hsa_addr): Fixup type when creating addresable temporary.
8452 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
8453 (gen_hsa_unary_operation): Extend operands and convert to dest type if
8454 necessary. Call hsa_fixup_mov_insn_type.
8455 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
8456 extend operands and convert to dest type if necessary.
8457 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
8458 to dest type if necessary.
8459 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
8460 if conversion nt necessary and size matches.
8461 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
8462 to dest type if necessary.
8463 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
8464 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
8466 (gen_hsa_clrsb): Likewise.
8467 (gen_hsa_ffs): Likewise.
8468 (gen_hsa_divmod): Extend operands and convert to dest type if
8470 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
8472 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8474 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
8475 Remove empty default arguments. Use a brace block as output
8477 (conditional return): Ditto.
8479 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
8480 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
8482 (group_ending_nop): Ditto.
8483 (doloop_end): Ditto.
8484 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
8485 (splitters for those): Ditto.
8487 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8489 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
8490 a conditional jump (and the compare for it) so that pc_rtx is the
8492 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
8493 for the deleted and renamed ctr<mode>_internal[234] patterns.
8494 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
8495 Delete second conditional return pattern.
8496 (ctr<mode>_internal2): Delete this second bdnz pattern.
8497 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
8498 (ctr<mode>_internal4): Delete this second bdz pattern.
8500 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
8502 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
8503 (always_initialized_rtx_for_ssa_name_p): New predicate.
8504 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
8505 (finish_out_of_ssa): Free new field of SA.
8506 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
8507 * tree-ssa-coalesce.c: Include tree-ssa.h.
8508 (get_parm_default_def_partitions): Remove extern keyword.
8509 (get_undefined_value_partitions): New function.
8510 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
8511 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
8513 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
8515 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
8517 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
8519 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
8521 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
8522 for targets that preffer 128bit.
8524 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
8526 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
8528 2017-10-08 Olivier Hainque <hainque@adacore.com>
8530 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
8531 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
8533 (thumb_set_return_address): Likewise.
8535 2017-10-08 Olivier Hainque <hainque@adacore.com>
8537 * common/config/arm/arm-common.c (arm_except_unwind_info):
8538 Handle DWARF2_UNWIND_INFO.
8540 2017-10-07 Michael Collison <michael.collison@arm.com>
8542 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
8545 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
8547 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
8548 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
8549 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
8550 defined, force the creation of a new block for a dispatch label.
8552 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
8554 * invoke.texi (Wsuggest-attribute=cold): Document.
8555 * common.opt (Wsuggest-attribute=cold): New
8556 * ipa-pure-const.c (warn_function_cold): New function.
8557 * predict.c (compute_function_frequency): Use it.
8558 * predict.h (warn_function_cold): Declare.
8560 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
8562 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
8565 2017-10-06 Martin Liska <mliska@suse.cz>
8567 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
8568 keyword for member functions.
8569 (struct sanopt_tree_couple): New struct.
8570 (struct sanopt_tree_couple_hash): New function.
8571 (struct sanopt_ctx): Add new hash_map.
8572 (has_dominating_ubsan_ptr_check): New function.
8573 (record_ubsan_ptr_check_stmt): Likewise.
8574 (maybe_optimize_ubsan_ptr_ifn): Likewise.
8575 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
8576 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
8578 2017-10-06 Sudakshina Das <sudi.das@arm.com>
8581 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
8582 aarch64_simd_valid_immediate on CONST_VECTORs.
8583 (aarch64_reg_or_bic_imm): Likewise.
8585 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
8587 PR rtl-optimization/82396
8588 * haifa-sched.c (ready_sort_real): Disable qsort checking.
8590 2017-10-06 Sebastian Pop <sebpop@gmail.com>
8592 * graphite-dependences.c (scop_get_reads): Move code to...
8593 (scop_get_must_writes): Move code to...
8594 (scop_get_may_writes): Move code to...
8595 (scop_get_reads_and_writes): ... here.
8596 (scop_get_dependences): Call scop_get_reads_and_writes.
8598 2017-10-06 Jakub Jelinek <jakub@redhat.com>
8600 PR tree-optimization/82434
8601 * fold-const.h (can_native_encode_type_p,
8602 can_native_encode_string_p): Remove.
8603 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
8604 don't encode anything, just return what would be otherwise returned.
8605 (native_encode_fixed, native_encode_complex, native_encode_vector):
8607 (native_encode_string): Likewise. Inline by hand
8608 can_native_encode_string_p.
8609 (can_native_encode_type_p): Remove.
8610 (can_native_encode_string_p): Remove.
8611 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
8612 STRING_CSTs using can_native_encode_string_p, test all
8613 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
8614 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
8615 argument from native_encode_expr.
8616 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
8617 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
8620 2017-10-06 Richard Biener <rguenther@suse.de>
8622 PR tree-optimization/82397
8623 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
8624 operand_equal_p but rely on data_ref_compare_tree for detecting
8626 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
8627 to match up with dr_group_sort_cmp.
8629 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8632 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
8634 * config/s390/s390-builtin-types.def: Regenerate.
8636 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8639 * config/s390/s390-builtin-types.def: Regenerate.
8640 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
8641 Change flag from B_VXE to B_VX.
8642 (s390_vec_min_dbl): Remove B_VXE flag.
8644 2017-10-06 Richard Biener <rguenther@suse.de>
8646 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
8647 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
8648 translate_isl_ast_to_gimple::is_valid_rename,
8649 translate_isl_ast_to_gimple::get_rename,
8650 translate_isl_ast_to_gimple::get_def_bb_for_const,
8651 translate_isl_ast_to_gimple::get_new_name,
8652 translate_isl_ast_to_gimple::collect_all_ssa_names,
8653 translate_isl_ast_to_gimple::copy_loop_phi_args,
8654 translate_isl_ast_to_gimple::collect_all_ssa_names,
8655 translate_isl_ast_to_gimple::copy_loop_phi_args,
8656 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
8657 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
8658 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
8659 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
8660 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
8661 translate_isl_ast_to_gimple::copy_cond_phi_args,
8662 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
8663 translate_isl_ast_to_gimple::edge_for_new_close_phis,
8664 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
8665 translate_isl_ast_to_gimple::rename_uses,
8666 translate_isl_ast_to_gimple::rename_all_uses): Remove.
8667 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
8668 (set_rename_for_each_def): Likewise.
8669 (graphite_copy_stmts_from_block): Handle debug stmt resetting
8670 here. Handle rewriting SCEV analyzable uses here.
8671 (copy_bb_and_scalar_dependences): Generate code for PHI
8673 (graphite_regenerate_ast_isl): Adjust.
8674 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
8675 (add_write, add_read): New functions.
8676 (build_cross_bb_scalars_def): Use it and simplify.
8677 (build_cross_bb_scalars_use): Likewise.
8678 (graphite_find_cross_bb_scalar_vars): Inline into...
8679 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
8680 simulating out-of-SSA. Compute liveout and add dependencies.
8681 (build_scops): Force an empty entry block.
8682 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
8684 (sese_build_liveouts): Declare.
8685 (sese_trivially_empty_bb_p): Likewise.
8686 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
8687 compute liveout and debug_liveout.
8688 (sese_bad_liveouts_use): Remove.
8689 (sese_reset_debug_liveouts_bb): Likewise.
8690 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
8691 (sese_build_liveouts): Build liveout and debug_liveout and store
8693 (new_sese_info): Adjust.
8694 (free_sese_info): Likewise.
8695 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
8696 do not build liveout here.
8697 (move_sese_in_condition): Adjust region entry.
8698 (scev_analyzable_p): Match up with chrec_apply requirements.
8699 (sese_trivially_empty_bb_p): New.
8700 * tree-into-ssa.c (get_reaching_def): Properly support generating
8701 default-defs for incremental rewrite of anonymous names.
8703 2017-10-06 Richard Biener <rguenther@suse.de>
8705 * graphite-sese-to-poly.c (extract_affine): For casts increasing
8706 precision do not perform modulo reduction.
8708 2017-10-06 Richard Biener <rguenther@suse.de>
8710 PR tree-optimization/82436
8711 * tree-vect-slp.c (vect_supported_load_permutation_p): More
8712 conservatively choose the vectorization factor when checking
8713 whether we can perform the required load permutation.
8714 (vect_transform_slp_perm_load): Assert when we may not fail.
8716 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
8718 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
8719 message for incompatible -msdata=* and -mcall-* options.
8721 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8723 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
8724 rate for post-reload scheduling.
8726 2017-10-05 Tamar Christina <tamar.christina@arm.com>
8728 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
8730 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8732 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
8733 to improve monte carlo in scimark.
8735 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8737 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
8738 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
8739 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
8740 pentium4_cost, nocona_cost): Set reassociation width to 1.
8741 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
8742 width to 2 for fp operations and 1 otherwise.
8743 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
8745 (atom_cost): Set reassociation width to 2.
8746 (slm_cost, generic_cost): Set fp reassociation width
8747 to 2 and 1 otherwise.
8748 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
8749 (core_cost): Set fp reassociation width to 4 and vector to 2.
8750 (ix86_reassociation_width): Rewrite using cost table; special case
8751 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
8752 and TARGET_AVX128_OPTIMAL.
8753 * config/i386/i386.h (processor_costs): Add
8754 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
8755 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
8756 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
8757 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
8758 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
8759 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
8761 2017-10-05 Nathan Sidwell <nathan@acm.org>
8763 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
8765 2017-10-05 Tamar Christina <tamar.christina@arm.com>
8767 * config/arm/arm.c (arm_test_fpu_data): New.
8768 (arm_run_selftests): Call arm_test_fpu_data.
8770 2017-10-04 Nathan Sidwell <nathan@acm.org>
8772 * toplev.c (toplev::main): Remove excess parens on pretty_printer
8774 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
8776 2017-10-04 Sudakshina Das <sudi.das@arm.com>
8778 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
8779 check type for aarch64_simd_valid_immediate.
8780 (aarch64_output_simd_mov_immediate): Update prototype.
8781 (aarch64_simd_valid_immediate): Update prototype.
8782 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
8783 support for ORR-immediate.
8784 (and<mode>3): modified pattern to add support for BIC-immediate.
8785 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
8786 now checks for valid immediate for BIC and ORR based on new enum
8788 (aarch64_output_simd_mov_immediate): Function now used to output
8789 BIC/ORR imm as well based on new enum argument.
8790 * config/aarch64/constraints.md (Do): New vector immediate constraint.
8792 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
8793 (aarch64_reg_or_bic_imm): Likewise.
8795 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8797 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
8798 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
8800 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
8804 PR rtl-optimization/82396
8805 * haifa-sched.c (autopref_multipass_init): Simplify
8807 (autopref_rank_data): Simplify sort order.
8808 * sched-int.h (autopref_multipass_data_): Remove
8809 multi_mem_insn_p, min_offset and max_offset.
8811 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8813 * doc/sourcebuild.texi: Document vect_peeling_profitable.
8815 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8817 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
8820 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8822 * doc/sourcebuild.texi: Document vect_long_mult.
8824 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
8826 PR tree-optimization/82413
8827 * fold-const.c (build_range_check): Use widest_int when comparing
8828 the maximum ETYPE value with HIGH.
8830 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
8832 PR rtl-optimization/82396
8833 * haifa-sched.c (autopref_multipass_init): Simplify
8835 (autopref_rank_data): Simplify sort order.
8836 * sched-int.h (autopref_multipass_data_): Remove
8837 multi_mem_insn_p, min_offset and max_offset.
8839 2017-10-04 Jakub Jelinek <jakub@redhat.com>
8841 PR tree-optimization/82381
8842 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
8843 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
8846 PR tree-optimization/82374
8847 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
8848 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
8849 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
8850 current_function_decl to the new decl.
8852 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
8854 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
8855 helper macro for IEEE float128 hardware built-in functions.
8856 (SQRTF128_ODD): Add built-in functions with the round-to-odd
8858 (TRUNCF128_ODD): Likewise.
8859 (ADDF128_ODD): Likewise.
8860 (SUBF128_ODD): Likewise.
8861 (MULF128_ODD): Likewise.
8862 (DIVF128_ODD): Likewise.
8863 (FMAF128_ODD): Likewise.
8864 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
8865 UNSPEC_TRUNC_ROUND_TO_ODD.
8866 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
8867 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
8868 floating point round to odd instructions.
8869 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
8870 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
8871 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
8872 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
8873 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
8874 (trunc<mode>sf2_hw): Change the truncate with round to odd
8875 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
8876 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
8877 to odd hardware instructions.
8878 (sub<mode>3_odd): Likewise.
8879 (mul<mode>3_odd): Likewise.
8880 (div<mode>3_odd): Likewise.
8881 (sqrt<mode>2_odd): Likewise.
8882 (fma<mode>4_odd): Likewise.
8883 (fms<mode>4_odd): Likewise.
8884 (nfma<mode>4_odd): Likewise.
8885 (nfms<mode>4_odd): Likewise.
8886 (trunc<mode>df2_odd): Change the truncate with round to odd
8887 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
8889 * doc/extend.texi (PowerPC built-in functions): Update documentation
8890 for existing IEEE float128-bit built-in functions. Add built-in
8891 functions that generate the IEEE 128-bit floating point round to
8894 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
8896 PR rtl-optimization/77729
8897 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
8898 to (X&(C1&~C2))|C2 transformations.
8900 2017-10-03 Martin Jambor <mjambor@suse.cz>
8902 PR tree-optimization/82363
8903 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
8904 mismatch, mark lacc written regardless of racc.
8906 2017-10-03 Jakub Jelinek <jakub@redhat.com>
8908 PR tree-optimization/82381
8909 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
8910 stmt_to_insert nor wheather SSA_NAMEs are default defs.
8911 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
8912 fallthrough into reassoc_stmt_dominates_stmt_p.
8915 * combine.c (combine_instructions): Don't combine in unreachable
8918 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
8921 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
8922 function to not use the have_cpu variable. Do not set cpu_index,
8923 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
8925 (rs6000_valid_attribute_p): Remove duplicate initializations of
8926 old_optimize and func_optimize.
8927 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
8928 (rs6000_activate_target_options): Make global.
8929 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
8932 2017-10-02 Jakub Jelinek <jakub@redhat.com>
8934 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
8935 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
8936 Set *poffset to 0 and *psize and *pmax_size to -1 if
8937 *poffset + *psize overflows in HOST_WIDE_INT.
8939 PR tree-optimization/82387
8940 PR tree-optimization/82388
8941 PR tree-optimization/82389
8942 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
8943 instead of live_bytes non-NULL.
8945 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
8948 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
8951 2017-10-02 Richard Biener <rguenther@suse.de>
8953 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
8954 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
8955 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
8957 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
8959 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
8960 requested precision matches the type's.
8961 * calls.c (alloc_max_size): Calculate the new candidate size as
8962 a widest_int and use wi::to_widest when comparing it with the
8963 current candidate size.
8964 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
8965 zero rather than integer_zero_node.
8966 * match.pd: Check for a no-op conversion before using wi::add
8967 rather than after. Use tree_to_uhwi when summing small shift
8968 counts into an unsigned int.
8970 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
8971 Alan Hayward <alan.hayward@arm.com>
8972 David Sherwood <david.sherwood@arm.com>
8975 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
8976 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
8977 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
8978 POINTER_AND_FP_REGS.
8980 2017-10-02 Richard Biener <rguenther@suse.de>
8982 PR tree-optimization/82355
8983 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8984 a mapping for the enclosing loop but avoid generating one for
8986 (copy_bb_and_scalar_dependences): Remove premature codegen
8987 error on PHIs in blocks duplicated into multiple places.
8988 * graphite-scop-detection.c
8989 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8990 in the region use it as loop and nest to analyze the DR in.
8991 (try_generate_gimple_bb): Likewise.
8992 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8993 (add_loop_constraints): For blocks in a loop not in the region
8994 create a dimension with a single iteration.
8995 * sese.h (gbb_loop_at_index): Remove assert.
8997 2017-10-01 Kevin Buettner <kevinb@redhat.com>
8999 * omp-expand.c (adjust_context_scope): New function.
9000 (expand_parallel_call): Call adjust_context_scope.
9002 2017-10-01 Jeff Law <law@redhat.com>
9004 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
9005 dom_opt_dom_walker class with direct access to private members.
9006 Add comments. Call test_for_singularity.
9007 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
9008 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
9009 m_dummy_cond anymore.
9010 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
9012 (pass_dominator:execute): Build the dummy_cond here and pass it
9013 to the dom_opt_dom_walker ctor.
9014 (test_for_singularity): New function.
9016 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
9017 Maya Rashish <coypu@sdf.org>
9019 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
9020 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
9021 (alpha*-*-netbsd*) Use nbsd_tm_file.
9022 (arm*-*-netbsdelf*) Likewise.
9023 (i[34567]86-*-netbsdelf*) Likewise.
9024 (x86_64-*-netbsd*) Likewise.
9025 (mips*-*-netbsd*) Likewise.
9026 (powerpc-*-netbsd*) Likewise.
9027 (sh*-*-netbsd*) Likewise.
9028 (sparc-*-netbsdelf*) Likewise.
9029 (sparc64-*-netbsd*) Likewise.
9030 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
9032 (vax-*-netbsdelf*) Likewise.
9033 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
9034 (UINT_FAST8_TYPE) Likewise.
9035 (INT_FAST16_TYPE) Check CHAR_FAST16.
9036 (UINT_FAST16_TYPE) Likewise.
9038 2017-09-30 Jakub Jelinek <jakub@redhat.com>
9041 * config/i386/i386.md
9042 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
9043 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
9044 *divmodsi4_zext_2): New define_insn_and_split.
9045 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
9046 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
9047 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
9048 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
9049 New define_insn_and_split.
9050 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
9051 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
9052 operands[1] having DImode when mode is SImode.
9054 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
9055 always SImode for DIV and MOD in REG_EQUAL notes.
9057 2017-09-29 Yury Gribov <tetra2005@gmail.com>
9060 * match.pd: Fix handling of NaNs in pattern.
9062 2017-09-29 Jeff Law <law@redhat.com>
9064 * sbitmap.c (bitmap_bit_in_range_p): New function.
9065 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
9066 * tree-ssa-dse.c (live_bytes_read): New function.
9067 (dse_classify_store): Ignore reads of dead bytes.
9069 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
9070 typos and whitespace errors.
9071 * config/i386/predicates.md (address_no_seg_operand): Likewise.
9072 * config/s390/s390.c (s390_emit_prologue): Likewise.
9074 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
9077 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
9078 with a symbol for LRA.
9080 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
9082 PR rtl-optimization/82338
9083 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
9085 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
9087 * genmodes.c (calc_wider_mode): Suppress qsort macro.
9088 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
9089 (qsort_chk): Declare.
9090 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
9091 (qsort_chk): New function.
9093 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9095 PR tree-optimization/82337
9096 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
9097 phi definition if the PHI result appears in an abnormal PHI.
9098 (find_basis_for_base_expr): Don't record a basis if the LHS of the
9099 basis appears in an abnormal PHI.
9101 2017-09-29 Richard Biener <rguenther@suse.de>
9103 * graphite-isl-ast-to-gimple.c
9104 (translate_isl_ast_to_gimple::set_codegen_error): New function.
9105 (binary_op_to_tree): Use it.
9106 (get_rename_from_scev): Likewise.
9107 (copy_loop_phi_nodes): Likewise.
9108 (copy_bb_and_scalar_dependences): Likewise.
9109 (translate_pending_phi_nodes): Likewise.
9111 2017-09-29 Jakub Jelinek <jakub@redhat.com>
9114 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
9115 for movabsq $(i32 << shift), r64.
9117 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
9119 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
9120 index when encoding %esp as %rsp to avoid 0x67 prefix.
9122 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
9124 * config/i386/i386.md (*movsf_internal, *movdf_internal):
9125 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
9127 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
9129 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
9130 Extensions with more than 16 double VFP registers.
9131 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
9132 to_clear_mask and all code related to it. Replace the remaining
9133 entry by a sbitmap and adapt code accordingly.
9135 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
9137 * brig-builtins.def: Change pure attributes to const.
9139 2017-09-28 Joseph Myers <joseph@codesourcery.com>
9141 * config.gcc (default_gnu_indirect_function): Default to yes for
9142 sparc*-*-linux* with glibc.
9144 2017-09-28 Joseph Myers <joseph@codesourcery.com>
9146 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
9147 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
9148 when creating .init_array and .fini_array sections with priority
9151 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
9154 * config/aarch64/aarch64.c
9155 (aarch64_builtin_support_vector_misalignment): Always return false
9156 when misalignment is unknown.
9158 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
9160 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
9161 this function to return false if the definition used by the swap
9162 instruction is artificial, or if the memory address from which the
9163 constant value is loaded is not represented by a base address held
9164 in a register or if the base address register is a frame or stack
9165 pointer. Additionally, return false if the base address of the
9166 loaded constant is a SYMBOL_REF but is not considered to be a
9168 (replace_swapped_load_constant): New function.
9169 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
9170 loaded constant vector with a load of a swapped constant vector.
9172 2017-09-27 Carl Love <cel@us.ibm.com>
9174 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
9175 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
9176 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
9179 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
9181 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
9182 first, always call autopref_rank_data otherwise.
9184 2017-09-27 Richard Biener <rguenther@suse.de>
9186 * graphite-scop-detection.c (find_scop_parameters): Move
9187 loop bound handling ...
9188 (gather_bbs::before_dom_children): ... here, avoiding the need
9189 to build scop_info->loop_nest.
9190 (record_loop_in_sese): Remove.
9191 * sese.h (sese_info_t::loop_nest): Remove.
9192 * sese.c (new_sese_info): Do not allocate loop_nest.
9193 (free_sese_info): Do not free loop_nest.
9195 2017-09-27 Jakub Jelinek <jakub@redhat.com>
9198 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
9199 lhs from calls if the lhs has addressable type.
9201 2017-09-27 Richard Biener <rguenther@suse.de>
9203 * graphite.h (scop::max_alias_set): New member.
9204 * graphite-scop-detection.c: Remove references to non-existing
9205 --param in comments.
9206 (build_alias_sets): Record the maximum alias set used for drs.
9207 (build_scops): Support zero as unlimited for
9208 --param graphite-max-arrays-per-scop.
9209 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
9211 (build_poly_sr_1): ... here. Compute alias set based on the
9212 maximum alias set used for drs rather than
9213 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
9215 2017-09-27 Richard Biener <rguenther@suse.de>
9217 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
9218 --param loop-block-tile-size=0 to disable tiling.
9220 2017-09-27 Richard Biener <rguenther@suse.de>
9222 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
9223 (graphite-max-nb-scop-params): Document special value zero.
9224 * domwalk.h (dom_walker::STOP): New symbolical constant.
9225 (dom_walker::dom_walker): Add optional parameter for bb to
9227 (dom_walker::~dom_walker): Declare.
9228 (dom_walker::before_dom_children): Document STOP return value.
9229 (dom_walker::m_user_bb_to_rpo): New member.
9230 (dom_walker::m_bb_to_rpo): Likewise.
9231 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
9232 mapping here if not provided by the user.
9233 (dom_walker::~dom_walker): Free bb to RPO mapping if not
9234 provided by the user.
9235 (dom_walker::STOP): Define.
9236 (dom_walker::walk): Do not compute bb to RPO mapping here.
9237 Support STOP return value from before_dom_children to stop
9239 * graphite-optimize-isl.c (optimize_isl): If the schedule
9240 is the same still generate code if -fgraphite-identity
9241 or -floop-parallelize-all are given.
9242 * graphite-scop-detection.c: Include cfganal.h.
9243 (gather_bbs::gather_bbs): Get and pass through bb to RPO
9245 (gather_bbs::before_dom_children): Return STOP for BBs
9247 (build_scops): Compute bb to RPO mapping and pass it to
9248 the domwalk. Treat --param graphite-max-nb-scop-params=0
9249 as not limiting the number of params.
9250 * graphite.c (graphite_initialize): Remove limit on the
9251 number of basic-blocks in a function.
9252 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
9253 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
9254 default value of 10.
9256 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9258 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
9259 Adjust code to eliminate needing to do the shift right 32-bits
9260 operation after XSCVDPSPN.
9262 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9264 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
9265 ((X / Y) != 0 -> X >= Y): Likewise.
9267 2017-09-26 Carl Love <cel@us.ibm.com>
9269 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
9270 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
9271 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
9272 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
9273 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
9274 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
9275 definitions and overloading.
9276 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
9277 statement for P9V_BUILTIN_XST_LEN_R.
9278 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
9279 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
9280 define_expand and define_insn for the instructions and builtins.
9281 * doc/extend.texi: Update the built-in documentation file for the new
9283 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
9284 define_insn for the instructions
9286 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
9289 * gcc/config/netbsd-protos.h: New file.
9290 * gcc/config/netbsd.c: New file.
9291 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
9292 * gcc/config/t-netbsd: New file.
9293 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
9294 (tmake_file) Add t-netbsd.
9295 (extra_objs) Add netbsd.o.
9297 2017-09-26 Janus Weil <janus@gcc.gnu.org>
9301 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
9303 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9305 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
9306 sign extension from a vector register to a GPR by doing a 32-bit
9307 direct move and then an EXTSW.
9308 (extendsi<mode>2 splitter): Likewise.
9309 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
9310 right or vector extract after doing XSCVDPSPN. Use
9311 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
9313 (movdi_from_sf_zero_ext): Likewise.
9314 (reload_gpr_from_vsxsf): Likewise.
9315 (p8_mfvsrd_4_disf): Delete, no longer used.
9316 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
9317 then needing to move the SFmode to a GPR to use the XSCVDPSP
9318 instruction instead of FRSP and XSCVDPSPN.
9319 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
9320 it is adjacent to the other XSCVSPDP insns.
9321 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
9322 SFmode to be in traditional Altivec registers.
9323 (vsx_xscvdpspn): Eliminate useless alternative constraint.
9324 (vsx_xscvspdpn): Likewise.
9325 (vsx_xscvspdpn_scalar): Likewise.
9327 2017-09-26 Martin Jambor <mjambor@suse.cz>
9329 * tree-sra.c (compare_access_positions): Put integral types first,
9330 stabilize sorting of integral types, remove conditions putting
9331 non-full-precision integers last.
9332 (sort_and_splice_var_accesses): Disable scalarization if a
9333 non-integert would be represented by a non-full-precision integer.
9335 2017-09-26 Joseph Myers <joseph@codesourcery.com>
9337 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
9338 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
9339 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
9340 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
9341 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
9342 conditionals inside the function instead of around it. Call
9343 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
9344 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
9346 2017-09-26 Richard Biener <rguenther@suse.de>
9348 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
9350 (scop_detection::build_scop_breadth): ... this. Removed.
9351 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
9352 (scop_detection::harmful_stmt_in_bb): Likewise.
9353 (scop_detection::graphite_can_represent_stmt): Likewise.
9354 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
9355 (scop_detection::can_represent_loop): Remove recursion, fold in ...
9356 (scop_detection::can_represent_loop_1): ... this. Removed.
9357 (scop_detection::harmful_loop_in_region): Simplify after inlining
9358 the above and remove more quadraticness.
9359 (build_scops): Adjust.
9360 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
9363 2017-09-26 Jakub Jelinek <jakub@redhat.com>
9366 * config/i386/i386.c (ix86_print_operand_address_as): Only test
9367 REGNO (base) == SP_REG if base is a REG.
9370 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
9371 if it is different SSA_NAME.
9372 (optimize_range_tests_cmp_bitwise): New function.
9373 (optimize_range_tests): Call it.
9375 2017-09-26 Richard Biener <rguenther@suse.de>
9377 PR tree-optimization/82321
9378 * graphite.c (canonicalize_loop_closed_ssa): Properly check
9379 for the def being inside the loop.
9381 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9383 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
9385 * config/s390/s390-builtins.def: Fix constraint on op4.
9387 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9389 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
9390 independent expanders.
9391 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
9392 ("vec_ordered", "vec_unordered"): New expanders.
9394 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9396 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
9399 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9401 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
9402 vec_unpacks_lo_v16qi.
9403 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
9405 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9407 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
9408 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
9409 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
9411 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9413 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
9415 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
9417 ("*vec_slb<mode>"): New insn pattern.
9418 ("vec_shr_<mode>"): New expander.
9419 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
9420 and force the shift count operand to V16QImode.
9421 ("vec_srb<mode>"): Set shift count mode to V16QI.
9423 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9425 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
9426 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
9427 ("vec_widen_smult_hi_<mode>"): New expander definitions.
9429 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
9432 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
9434 2017-09-26 Richard Biener <rguenther@suse.de>
9436 PR tree-optimization/82320
9437 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
9440 2017-09-25 Jeff Law <law@redhat.com>
9442 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
9443 prototype for new argument.
9444 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
9445 mostly extracted from rs6000_emit_allocate_stack.
9446 (rs6000_emit_probe_stack_range_stack_clash): New function.
9447 (rs6000_emit_allocate_stack): Call
9448 rs6000_emit_probe_stack_range_stack_clash as needed.
9449 (rs6000_emit_probe_stack_range): Add additional argument
9450 to call to gen_probe_stack_range{si,di}.
9451 (output_probe_stack_range): New.
9452 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
9453 (output_probe_stack_range_stack_clash): New.
9454 (rs6000_emit_prologue): Emit notes into dump file as requested.
9455 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
9456 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
9457 Add additional operand and pass it to output_probe_stack_range.
9459 2017-09-25 Bin Cheng <bin.cheng@arm.com>
9461 PR tree-optimization/82163
9462 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
9463 (checking_verify_loop_closed_ssa): New parameter.
9464 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
9465 (check_loop_closed_ssa_stmt): Delete.
9466 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
9467 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
9468 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
9471 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
9473 * brig-builtins.def: Treat HSAIL barrier builtins as
9474 setjmp/longjump style functions.
9476 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9478 * target.def (constant_alignment): New hook.
9479 * defaults.h (CONSTANT_ALIGNMENT): Delete.
9480 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
9481 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
9482 * doc/tm.texi: Regenerate.
9483 * targhooks.h (default_constant_alignment): Declare.
9484 (constant_alignment_word_strings): Likewise.
9485 * targhooks.c (default_constant_alignment): New function.
9486 (constant_alignment_word_strings): Likewise.
9487 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
9488 instead of CONSTANT_ALIGNMENT.
9489 * varasm.c (align_variable, get_variable_align, build_constant_desc)
9490 (force_const_mem): Likewise.
9491 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
9492 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
9493 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
9494 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9495 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
9497 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
9498 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9499 constant_alignment_word_strings.
9500 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
9501 (CONSTANT_ALIGNMENT): Likewise.
9502 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9503 (arm_constant_alignment): New function.
9504 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
9505 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9506 constant_alignment_word_strings.
9507 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
9508 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9509 constant_alignment_word_strings.
9510 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
9511 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9512 (cris_constant_alignment): New function.
9513 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
9514 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9515 (epiphany_constant_alignment): New function.
9516 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
9517 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9518 constant_alignment_word_strings.
9519 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
9520 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9521 constant_alignment_word_strings.
9522 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
9523 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9524 constant_alignment_word_strings.
9525 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
9526 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
9527 * config/i386/i386.c (ix86_constant_alignment): Make static.
9528 Use the same interface as the target hook.
9529 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9530 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
9531 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9532 constant_alignment_word_strings.
9533 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
9534 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
9535 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9536 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
9537 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9538 constant_alignment_word_strings.
9539 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
9540 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9541 constant_alignment_word_strings.
9542 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
9543 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9544 constant_alignment_word_strings.
9545 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
9546 * config/microblaze/microblaze.c (microblaze_constant_alignment):
9548 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9549 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
9550 * config/mips/mips.c (mips_constant_alignment): New function.
9551 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9552 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
9553 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
9554 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9555 (mmix_constant_alignment): Make static. Use the same interface
9557 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
9558 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9559 constant_alignment_word_strings.
9560 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
9561 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9562 constant_alignment_word_strings.
9563 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
9564 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9565 constant_alignment_word_strings.
9566 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
9567 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9568 (rs6000_constant_alignment): New function.
9569 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
9570 * config/riscv/riscv.c (riscv_constant_alignment): New function.
9571 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9572 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
9573 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9574 (rs6000_constant_alignment): New function.
9575 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
9576 * config/s390/s390.c (s390_constant_alignment): New function.
9577 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9578 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
9579 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9580 constant_alignment_word_strings.
9581 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
9582 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9583 (sparc_constant_alignment): New function.
9584 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
9585 * config/spu/spu.c (spu_constant_alignment): New function.
9586 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9587 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
9588 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9589 constant_alignment_word_strings.
9590 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
9591 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9592 constant_alignment_word_strings.
9593 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
9594 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9595 constant_alignment_word_strings.
9596 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
9597 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9598 (visium_constant_alignment): New function.
9599 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
9600 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9601 (xtensa_constant_alignment): New function.
9602 * system.h (CONSTANT_ALIGNMENT): Poison.
9604 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
9606 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9607 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
9608 (rs6000_builtin_valid_without_lhs): New helper function.
9609 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9610 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
9612 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9614 * target.h (vec_perm_indices): Use unsigned short rather than
9616 (auto_vec_perm_indices): Likewise.
9617 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
9618 Use unsigned int rather than unsigned char.
9619 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
9621 2017-09-25 Richard Biener <rguenther@suse.de>
9623 * cfgloop.h (sort_sibling_loops): Declare.
9624 * cfgloop.c (sort_sibling_loops_cmp): New helper.
9625 (sort_sibling_loops): New function sorting the sibling loop list
9627 * graphite.c (graphite_transform_loops): Sort sibling loops.
9629 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
9631 * target.def (vec_perm_const_ok): Change sel parameter to
9633 * optabs-query.c (can_vec_perm_p): Update accordingly.
9634 * doc/tm.texi: Regenerate.
9635 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
9636 auto_vec_perm_indices and remove separate nelt field.
9637 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
9638 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
9639 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
9640 (aarch64_expand_vec_perm_const): Update accordingly.
9641 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
9642 to vec_perm_indices.
9643 * config/arm/arm.c (expand_vec_perm_d): Change perm to
9644 auto_vec_perm_indices and remove separate nelt field.
9645 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
9646 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
9647 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
9649 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
9650 to vec_perm_indices.
9651 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
9652 sel to vec_perm_indices.
9653 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
9654 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
9655 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
9657 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
9660 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
9663 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
9664 on the FUNCTION_DECL function context if it has a DIE that is a
9667 2017-09-25 Richard Biener <rguenther@suse.de>
9669 PR tree-optimization/82285
9670 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
9673 2017-09-25 Tom de Vries <tom@codesourcery.com>
9677 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
9680 2017-09-25 Richard Biener <rguenther@suse.de>
9682 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
9683 ISL errors other than isl_error_quota happen. Dump if the
9684 schedule is the same.
9685 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
9686 errors instead of aborting inside ISL.
9688 2017-09-25 Iain Sandoe <iain@codesourcery.com>
9691 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
9692 of libgcc_eh for m64.
9693 * config/i386/darwin64.h: Likewise.
9695 2017-09-25 Richard Biener <rguenther@suse.de>
9698 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
9699 attribute for incomplete types nor twice for complete ones.
9701 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
9704 * config/i386/i386.c (ix86_print_operand_address_as): Encode
9705 %esp as %rsp to avoid 0x67 prefix if there is no index or base
9708 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
9711 * config/i386/i386.opt (mprefer-avx256): Use
9712 ix86_target_flags variable.
9713 * config/i386/i386.c (ix86_target_string): Move
9714 -mprefer-avx256 to flag2_opts.
9716 2017-09-22 Jakub Jelinek <jakub@redhat.com>
9719 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
9720 and x != -1 | y != -1 into (x & y) != -1.
9722 2017-09-22 Steve Ellcey <sellcey@cavium.com>
9724 * config.gcc: Add new case statement to set
9725 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
9726 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
9727 s390x-*-linux* case statements. Added aarch64 to the list of
9728 supported architectures.
9730 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9732 PR tree-optimization/82289
9733 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
9734 STMT_VINFO_RELEVANT_P.
9736 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9737 Alan Hayward <alan.hayward@arm.com>
9738 David Sherwood <david.sherwood@arm.com>
9740 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
9741 for VR_RANGE only; don't allow VR_ANTI_RANGE.
9742 (extract_range_from_binary_expr_1): Don't call
9743 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
9745 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9746 Alan Hayward <alan.hayward@arm.com>
9747 David Sherwood <david.sherwood@arm.com>
9749 * target.def (preferred_vector_alignment): New hook.
9750 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
9752 * doc/tm.texi: Regenerate.
9753 * targhooks.h (default_preferred_vector_alignment): Declare.
9754 * targhooks.c (default_preferred_vector_alignment): New function.
9755 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
9757 (DR_TARGET_ALIGNMENT): New macro.
9758 (aligned_access_p): Update commentary.
9759 (vect_known_alignment_in_bytes): New function.
9760 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
9762 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
9763 Calculate the misalignment based on the target alignment rather than
9765 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
9766 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
9767 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
9768 the target alignment, rather than masking the element misalignment
9769 with the number of elements in a vector. Also use the target
9770 alignment when calculating the maximum number of peels.
9771 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
9772 instead of TYPE_ALIGN_UNIT.
9773 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
9774 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
9775 (vect_create_addr_base_for_vector_ref): Update call accordingly.
9776 (vect_create_data_ref_ptr): Likewise.
9777 (vect_setup_realignment): Realign by ANDing with
9778 -DR_TARGET_MISALIGNMENT.
9779 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
9780 the number of peels based on DR_TARGET_ALIGNMENT.
9781 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
9782 with the guaranteed alignment boundary when deciding whether
9784 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
9785 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
9786 (ensure_base_align): Remove stmt_info parameter. Get the
9787 target base alignment from DR_TARGET_ALIGNMENT.
9788 (vectorizable_store): Update call accordingly. Interpret
9789 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
9791 (vectorizable_load): Likewise.
9793 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9794 Alan Hayward <alan.hayward@arm.com>
9795 David Sherwood <david.sherwood@arm.com>
9797 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
9798 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
9799 (vect_enhance_data_refs_alignment): Likewise.
9801 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9803 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9804 error. Only quit immediately if parsing is complete.
9805 (BEGIN): Initialize fatal_err and parse_done.
9806 (begin fpu, end fpu): Check number of arguments.
9807 (begin arch, end arch): Likewise.
9808 (begin cpu, end cpu): Likewise.
9809 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9810 (optalias): Likewise.
9812 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9814 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9815 * config/arm/arm-isa.h: Delete. Move definitions to ...
9816 * arm-cpus.in: ... here. Use new feature and fgroup values.
9817 * config/arm/arm.c (arm_option_override): Use lower case for feature
9819 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9820 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9821 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9823 (print_isa_bits_for): New function.
9824 (gen_isa): New function.
9825 (gen_comm_data): Use print_isa_bits_for.
9826 (define feature): New keyword.
9827 (define fgroup): New keyword.
9828 * config/arm/t-arm (TM_H): Remove.
9829 (GTM_H): Add arm-isa.h.
9830 (arm-isa.h): Add rule to generate file.
9831 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9832 case for feature bit names.
9834 2017-09-22 Richard Biener <rguenther@suse.de>
9836 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
9838 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
9839 print of no dependency loops ...
9840 * graphite.c (graphite_transform_loops): ... here.
9841 (canonicalize_loop_closed_ssa_form): Work from inner to outer
9843 (same_close_phi_node, remove_duplicate_close_phi,
9844 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
9845 (canonicalize_loop_closed_ssa): ... here and simplify.
9846 * graphite-optimize-isl.c: Include tree-vectorizer.h.
9847 (optimize_isl): Use dump_printf_loc to tell when we stopped
9848 optimizing because of an ISL timeout.
9850 2017-09-22 Richard Biener <rguenther@suse.de>
9852 PR tree-optimization/82291
9853 * tree-if-conv.c (predicate_mem_writes): Make sure to
9854 remove writes in blocks predicated with false.
9856 2017-09-22 Richard Biener <rguenther@suse.de>
9858 * sese.c: Include cfganal.h.
9859 (if_region_set_false_region): Remove.
9860 (create_if_region_on_edge): Likewise.
9861 (move_sese_in_condition): Re-implement without destroying
9864 2017-09-22 Richard Biener <rguenther@suse.de>
9866 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
9867 Verify both BBs contain loop PHI nodes before dispatching to
9869 (graphite_regenerate_ast_isl): Do not recompute dominators,
9870 do not verify three times. Restructure for clarity.
9871 * graphite-scop-detection.c (same_close_phi_node,
9872 remove_duplicate_close_phi, make_close_phi_nodes_unique,
9873 defined_in_loop_p, canonicalize_loop_closed_ssa,
9874 canonicalize_loop_closed_ssa_form): Simplify, remove excess
9875 checking and SSA rewrite, move to ...
9876 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
9877 (graphite_initialize): Do not pass in ctx, do not reset the
9878 SCEV cache, compute only dominators.
9879 (graphite_transform_loops): Allocate ISL ctx after
9880 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
9881 Maintain post-dominators only around build_scops.
9882 * sese.c (if_region_set_false_region): Make static. Free
9883 and recompute dominators.
9884 (move_sese_in_condition): Assert we don't get called with
9885 post-dominators computed.
9886 * sese.h (if_region_set_false_region): Remove.
9888 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
9890 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
9891 mode attribute for TARGET_AVX512VL.
9893 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
9895 * config/i386/i386.opt (mprefer-avx256): New option.
9896 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
9898 (ix86_preferred_simd_mode): Return 256-bit AVX modes
9899 for TARGET_PREFER_AVX256.
9900 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
9902 2017-09-21 Jeff Law <law@redhat.com>
9904 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
9905 Fix dump output if the only stack space is for pushed registers.
9907 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9909 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
9912 2017-09-21 Martin Sebor <msebor@redhat.com>
9915 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
9916 code (in C++) or code that triggers warnings.
9918 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
9920 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
9922 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
9924 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
9925 * sched-rgn.c: Ditto.
9926 * sel-sched-ir.c: Ditto.
9928 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
9930 * toplev.h (set_random_seed): Adjust return type.
9931 * toplev.c (init_local_tick): Move eager initialization of random_seed
9932 to get_random_seed. Adjust comment.
9933 (init_random_seed): Inline to get_random_seed, delete.
9934 (get_random_seed): Initialize random_seed lazily.
9935 (set_random_seed): Do not return previous value.
9936 (print_switch_value): Do not call get_random_seed.
9938 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
9940 * cgraph.c (delete_function_version): New, broken out from...
9941 (cgraph_node::delete_function_version): ...here. Rename to
9942 cgraph_node::delete_function_version_by_decl. Update all uses.
9943 (cgraph_node::remove): Call delete_function_version.
9945 2017-09-21 Jakub Jelinek <jakub@redhat.com>
9948 * tree-inline.c (expand_call_inline): Emit clobber stmts for
9949 VAR_DECLs to which addressable non-volatile parameters are mapped
9950 and for id->retvar after the return value assignment. Clear
9951 id->retval and id->retbnd after inlining.
9953 2017-09-21 Richard Biener <rguenther@suse.de>
9955 PR tree-optimization/82276
9956 PR tree-optimization/82244
9957 * tree-vrp.c (build_assert_expr_for): Set
9958 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
9960 (remove_range_assertions): Revert earlier change.
9962 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
9965 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
9967 2017-09-21 Richard Biener <rguenther@suse.de>
9969 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9970 Restore valid IL after code generation errors.
9971 * graphite.c (graphite_transform_loops): Diagnose code
9972 generation issues as MSG_MISSED_OPTIMIZATION and continue
9973 with processing SCOPs.
9975 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9976 Alan Hayward <alan.hayward@arm.com>
9977 David Sherwood <david.sherwood@arm.com>
9979 * calls.c (compute_argument_addresses): Use simplify_gen_binary
9980 rather than choosing between plus_constant and gen_rtx_<CODE>.
9981 * expr.c (emit_push_insn): Likewise.
9982 (expand_expr_real_2): Likewise.
9984 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9985 Alan Hayward <alan.hayward@arm.com>
9986 David Sherwood <david.sherwood@arm.com>
9988 * loop-unroll.c (split_iv): Call copy_rtx on the step.
9990 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9991 Alan Hayward <alan.hayward@arm.com>
9992 David Sherwood <david.sherwood@arm.com>
9994 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
9995 calling tree_to_uhwi.
9997 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9998 Alan Hayward <alan.hayward@arm.com>
9999 David Sherwood <david.sherwood@arm.com>
10001 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
10002 INTEGER_CST rather than a negative test for ADDR_EXPR.
10004 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10005 Alan Hayward <alan.hayward@arm.com>
10006 David Sherwood <david.sherwood@arm.com>
10008 * tree-vrp.c (extract_range_from_binary_expr_1): Check
10009 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
10011 2017-09-21 Richard Biener <rguenther@suse.de>
10013 PR tree-optimization/71351
10014 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
10015 graphite_create_new_loop_guard): Remove, fold remaining parts
10017 (translate_isl_ast_node_for): ... here and simplify.
10019 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10022 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
10023 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
10024 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
10025 alternative always use QI mode, for -Os imov (=R,R) alternative
10026 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
10029 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10030 Jeff Law <law@redhat.com>
10032 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
10033 (allocate_stack_space): New function, partially extracted from
10034 s390_emit_prologue.
10035 (s390_emit_prologue): Track offset to most recent stack probe.
10036 Code to allocate space moved into allocate_stack_space.
10037 Dump actions when no stack is allocated.
10038 (s390_prologue_plus_offset): New function.
10039 (s390_emit_stack_probe): Likewise.
10041 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
10043 * common.opt (Wa, Wl, Wp, g, gz=): Add
10045 (gno-column-info): Remove.
10046 (gcolumn-info): Drop RejectNegative.
10047 (gno-): New prefix.
10048 (gno-record-gcc-switches): Remove.
10049 (grecord-gcc-switches): Drop RejectNegative.
10050 (gno-split-dwarf): Remove.
10051 (gsplit-dwarf): Drop RejectNegative.
10052 (gno-strict-dwarf): Remove.
10053 (gstrict-dwarf): Drop RejectNegative.
10054 * config/darwin.opt (gfull, gused): Add RejectNegative.
10055 * dwarf2out.c (gen_producer_string): Drop
10056 gno-record-gcc-switches handler.
10057 * optc-gen.awk: Add g to prefixes with negative forms.
10058 * opts-common.c (remapping_prefix_p): New.
10059 (find_opt): Check it.
10060 (generate_canonical_option): Test g prefix.
10061 (option_map): Add -gno- mapping.
10062 (add_misspelling_candidates): Check remapping_prefix_p.
10064 2017-09-20 Jeff Law <law@redhat.com>
10066 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
10067 thinko in stack clash protection support.
10069 * explow.c (compute_stack_clash_protection_loop_data): Use
10070 CONST_INT_P instead of explicit test. Verify object is a
10071 CONST_INT_P before looking at INTVAL.
10072 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
10073 instead of explicit test.
10075 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
10078 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
10079 address instead of to r1 and r11.
10081 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
10083 * config.gcc: Support "knm".
10084 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
10085 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
10087 * config/i386/i386.c (m_KNM): Define.
10088 (processor_target_table): Add "knm".
10090 (ix86_option_override_internal): Add "knm".
10091 (ix86_issue_rate): Add PROCESSOR_KNM.
10092 (ix86_adjust_cost): Ditto.
10093 (ia32_multipass_dfa_lookahead): Ditto.
10094 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
10095 (fold_builtin_cpu): Add M_INTEL_KNM.
10096 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
10097 (processor_type): Add PROCESSOR_KNM.
10098 * config/i386/x86-tune.def: Add m_KNM.
10099 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
10101 2017-09-20 Richard Biener <rguenther@suse.de>
10103 PR tree-optimization/80213
10104 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
10105 are allowed in empty BBs as well.
10106 (canonicalize_loop_closed_ssa): Also look for other complex
10108 (scop_detection::get_sese): Include the loop-closed PHI block
10110 (scop_detection::merge_sese): Remove code adding extra blocks.
10111 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
10112 (build_scops): Assert the final returned scop is invalid.
10114 2017-09-20 Richard Biener <rguenther@suse.de>
10116 PR tree-optimization/82264
10117 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
10119 (vn_phi_lookup): Likewise.
10120 (vn_phi_insert): Likewise.
10122 2017-09-20 Jakub Jelinek <jakub@redhat.com>
10124 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
10125 that fits into uhwi or shwi, add DW_AT_const_value regardless
10126 of early_dwarf without going through RTL, using add_AT_unsigned
10129 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
10130 (DEBUG_LTO_ABBREV_SECTION): Likewise.
10131 (DEBUG_LTO_MACINFO_SECTION): Likewise.
10132 (DEBUG_MACRO_SECTION): Likewise.
10133 (DEBUG_LTO_MACRO_SECTION): Likewise.
10134 (DEBUG_STR_DWO_SECTION): Likewise.
10135 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
10136 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
10137 (DEBUG_LTO_DWO_LINE_SECTION): Define.
10138 (DEBUG_LTO_LINE_STR_SECTION): Define.
10139 (init_sections_and_labels): Initialize debug_line_str_section
10140 variable. Initialize debug_loc_section for -gdwarf-5 to
10141 DEBUG_LOCLISTS_SECTION. Formatting fixes.
10143 2017-09-20 Richard Biener <rguenther@suse.de>
10145 * graphite-sese-to-poly.c (extract_affine): Properly handle
10146 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
10148 2017-09-20 Richard Biener <rguenther@suse.de>
10150 PR tree-optimization/81373
10151 * graphite-scop-detection.c (build_cross_bb_scalars_def):
10152 Force SESE live-out defs to be handled even if they are
10155 2017-09-19 Jeff Law <law@redhat.com>
10157 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
10158 nothing for stack adjustments with REG_STACK_CHECK.
10159 * sched-deps.c (parse_add_or_inc): Reject insns with
10160 REG_STACK_CHECK from dependency breaking.
10161 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
10162 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
10163 * reg-notes.def (STACK_CHECK): New note.
10165 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
10166 (ix86_expand_prologue): Dump stack clash info as needed.
10167 Call ix86_adjust_stack_and_probe_stack_clash as needed.
10169 * function.c (dump_stack_clash_frame_info): New function.
10170 * function.h (dump_stack_clash_frame_info): Prototype.
10171 (enum stack_clash_probes): New enum.
10173 * config/alpha/alpha.c (alpha_expand_prologue): Also check
10174 flag_stack_clash_protection.
10175 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
10176 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
10177 (arm_frame_pointer_required): Likewise.
10178 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
10179 (ia64_expand_prologue): Likewise.
10180 * config/mips/mips.c (mips_expand_prologue): Likewise.
10181 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
10182 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
10183 (sparc_flat_expand_prologue): Likewise.
10184 * config/spu/spu.c (spu_expand_prologue): Likewise.
10186 * explow.c: Include "params.h".
10187 (anti_adjust_stack_and_probe_stack_clash): New function.
10188 (get_stack_check_protect): Likewise.
10189 (compute_stack_clash_protection_loop_data): Likewise.
10190 (emit_stack_clash_protection_loop_start): Likewise.
10191 (emit_stack_clash_protection_loop_end): Likewise.
10192 (allocate_dynamic_stack_space): Use get_stack_check_protect.
10193 Use anti_adjust_stack_and_probe_stack_clash.
10194 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
10195 (emit_stack_clash_protection_loop_start): Likewise.
10196 (emit_stack_clash_protection_loop_end): Likewise.
10197 * rtl.h (get_stack_check_protect): Prototype.
10198 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
10199 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
10200 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
10202 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
10204 * doc/tm.texi: Rebuilt.
10205 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
10206 get_stack_check_protect.
10207 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
10208 * config/arm/arm.c (arm_expand_prologue): Likewise.
10209 (arm_frame_pointer_required): Likewise.
10210 * config/i386/i386.c (ix86_expand_prologue): Likewise.
10211 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
10212 * config/mips/mips.c (mips_expand_prologue): Likewise.
10213 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
10214 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
10215 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
10216 (sparc_flat_expand_prologue): Likewise.
10218 * common.opt (-fstack-clash-protection): New option.
10219 * flag-types.h (enum stack_check_type): Note difference between
10220 -fstack-check= and -fstack-clash-protection.
10221 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
10222 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
10223 * toplev.c (process_options): Issue warnings/errors for cases
10224 not handled with -fstack-clash-protection.
10225 * doc/invoke.texi (-fstack-clash-protection): Document new option.
10226 (-fstack-check): Note additional problem with -fstack-check=generic.
10227 Note that -fstack-check is primarily for Ada and refer users
10228 to -fstack-clash-protection for stack-clash-protection.
10229 Document new params for stack clash protection.
10231 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
10233 * config/i386/i386.c (ix86_split_long_move): Do not handle
10234 address used for LEA in a special way.
10236 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
10238 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
10240 2017-09-19 Martin Sebor <msebor@redhat.com>
10243 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
10244 of incompatible types.
10246 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
10248 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10249 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
10250 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10251 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
10253 2017-09-19 Richard Biener <rguenther@suse.de>
10255 PR tree-optimization/82244
10256 * tree-vrp.c (remove_range_assertions): Do not propagate
10257 a constant to abnormals but replace the assert with a copy.
10259 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
10261 PR rtl-optimization/57878
10262 PR rtl-optimization/68988
10263 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
10264 avoidance test involving non_reload_pseudos. Move frequency test
10265 below the general fragmentation avoidance test.
10267 2017-09-19 Richard Biener <rguenther@suse.de>
10269 PR tree-optimization/69728
10270 * graphite-sese-to-poly.c (schedule_error): New global.
10271 (add_loop_schedule): Handle empty domain by failing the
10273 (build_original_schedule): Handle schedule_error.
10275 2017-09-19 Richard Biener <rguenther@suse.de>
10277 * graphite-scop-detection.c (scop_detection::can_represent_loop):
10278 Do not iterate to sibling loops but only to siblings of inner
10281 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
10284 * config/m68k/m68k.md (moveq feeding equality comparison): Check
10285 that the registers are different.
10287 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
10289 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
10290 to processor_model and "amdfam17h" to arch_names_table.
10291 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
10293 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10296 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
10298 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10299 Alan Hayward <alan.hayward@arm.com>
10300 David Sherwood <david.sherwood@arm.com>
10302 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
10304 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
10306 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
10307 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
10308 vect_schedule_slp_instance.
10309 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
10310 Update call to vect_slp_analyze_node_operations. Simplify return
10312 (vect_slp_analyze_bb_1): Update call accordingly.
10313 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
10314 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
10315 (vect_schedule_slp): Update call accordingly.
10317 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10318 Alan Hayward <alan.hayward@arm.com>
10319 David Sherwood <david.sherwood@arm.com>
10321 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
10322 with types that aren't in fact scalar.
10324 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10326 * tree-vect-slp.c (vect_record_max_nunits): New function,
10328 (vect_build_slp_tree_1): ...here.
10329 (vect_build_slp_tree_2): Call it for phis too.
10331 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10333 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
10334 to vect_get_vec_def_for_operand when getting the mask operand.
10336 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10337 Alan Hayward <alan.hayward@arm.com>
10338 David Sherwood <david.sherwood@arm.com>
10340 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
10343 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10344 Alan Hayward <alan.hayward@arm.com>
10345 David Sherwood <david.sherwood@arm.com>
10347 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
10348 calculation for vector booleans.
10350 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10351 Alan Hayward <alan.hayward@arm.com>
10352 David Sherwood <david.sherwood@arm.com>
10354 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
10356 (vect_transform_stmt): ...here.
10357 (vect_analyze_stmt): Use it instead of calling
10358 vectorizable_live_operation directly.
10360 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
10362 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
10363 non-SIMT targets in acc vector loops.
10365 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
10367 * configure.ac: Add arc and check if assembler supports gdwarf2.
10368 * configure: Regenerate.
10370 2017-09-18 Richard Biener <rguenther@suse.de>
10372 PR tree-optimization/82220
10373 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
10374 epilogue niters from the min_profitable_iters compute.
10376 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10379 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
10380 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
10381 (ix86_init_pic_reg): Revert 2017-09-01 changes.
10383 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
10386 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
10387 switching to a new text section.
10389 2017-09-18 Richard Biener <rguenther@suse.de>
10391 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
10393 (build_alias_set): Reject aliases with no access function.
10395 2017-09-18 Richard Biener <rguenther@suse.de>
10397 PR tree-optimization/79622
10398 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
10400 (build_cross_bb_scalars_use): Likewise.
10402 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
10404 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
10406 2017-09-18 Alan Modra <amodra@gmail.com>
10409 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
10410 stack_pointer_rtx for count 0. Update comments. Break up
10411 large rtl expression.
10413 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
10415 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
10416 Increase to 20 bytes.
10417 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
10418 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
10419 or avx version of the stub.
10421 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
10424 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
10425 compute the minimum stack alignment. Also update preferred stack
10426 boundary for leaf functions.
10428 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
10430 PR tree-optimization/82228
10431 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
10434 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10436 * common/config/nds32/nds32-common.c
10437 (nds32_option_optimization_table): Refine formatting.
10438 (nds32_option_optimization_table): Use -fsched-pressure and
10439 -fomit-frame-pointer for specific optimization level.
10441 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10443 * config/nds32/nds32.c: Refine formatting and comments.
10444 * config/nds32/nds32.h: Likewise.
10445 * config/nds32/nds32.md: Likewise.
10446 * config/nds32/nds32-cost.c: Likewise.
10447 * config/nds32/nds32-isr.c: Likewise.
10448 * config/nds32/nds32-md-auxiliary.c: Likewise.
10449 * config/nds32/nds32-multiple.md: Likewise.
10450 * config/nds32/nds32-predicates.c: Likewise.
10452 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
10453 Jakub Jelinek <jakub@redhat.com>
10455 Add support for -std=c++2a.
10456 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
10458 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
10460 2017-09-15 Steve Ellcey <sellcey@cavium.com>
10463 * doc/extend.texi (Common Function Attributes): Add
10464 references to ARM, AArch64, and S/390 specific attributes.
10465 (Function Specific Option Pragmas): Add AArch64 and S/390
10466 to list of back ends that support the target pragma.
10468 2017-09-15 Nathan Sidwell <nathan@acm.org>
10470 * doc/standards.texi: Fix C++17 description. Update URLs for
10473 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
10475 * common.opt (Wcast-align=strict): New warning option.
10476 * doc/invoke.texi: Document -Wcast-align=strict.
10478 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
10480 * cgraph.h (cgraph_thunk_info): Add comments.
10481 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
10482 assert for VIRTUAL_* arguments stricter.
10484 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
10486 PR tree-optimization/71026
10487 * match.pd: Move RDIV patterns from fold-const.c
10488 * fold-const.c (distribute_real_division): Removed.
10489 (fold_binary_loc): Remove calls to distribute_real_divison.
10491 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10493 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
10494 c++1z and gnu++1z as deprecated. Change other references to
10495 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
10496 Change -Wc++1z-compat to -Wc++17-compat.
10497 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
10498 * dwarf2out.c (highest_c_language): Handle C++17.
10499 (gen_compile_unit_die): Likewise.
10501 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10503 PR rtl-optimization/82192
10504 * combine.c (make_extraction): Don't look through non-paradoxical
10505 SUBREGs or TRUNCATE if pos + len is or might be bigger than
10508 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10509 Alan Hayward <alan.hayward@arm.com>
10510 David Sherwood <david.sherwood@arm.com>
10512 * target.def (function_arg_offset): New hook.
10513 * targhooks.h (default_function_arg_offset): Declare.
10514 * targhooks.c (default_function_arg_offset): New function.
10515 * function.c (locate_and_pad_parm): Use
10516 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
10517 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
10518 (TARGET_FUNCTION_ARG_OFFSET): ...this.
10519 * doc/tm.texi: Regenerate.
10520 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
10521 * config/spu/spu.c (spu_function_arg_offset): New function.
10522 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
10523 * system.h (FUNCTION_ARG_OFFSET): Poison.
10525 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10526 Alan Hayard <alan.hayward@arm.com>
10527 David Sherwood <david.sherwood@arm.com>
10529 * target.def (truly_noop_truncation): New hook.
10530 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
10531 than TRULY_NOOP_TRUNCATION.
10532 * hooks.h (hook_bool_uint_uint_true): Declare.
10533 * hooks.c (hook_bool_uint_uint_true): New function.
10534 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
10535 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
10536 * doc/tm.texi: Regenerate.
10537 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
10538 rather than TRULY_NOOP_TRUNCATION in comments.
10539 (simplify_comparison): Likewise.
10540 (record_truncated_value): Likewise.
10541 * expmed.c (extract_bit_field_1): Likewise.
10542 (extract_split_bit_field): Likewise.
10543 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
10544 instead of TRULY_NOOP_TRUNCATION.
10545 * function.c (assign_parm_setup_block): Likewise.
10546 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
10547 * rtlhooks.c: Include target.h.
10548 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
10549 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
10550 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
10551 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
10552 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
10553 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
10554 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
10555 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
10556 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
10557 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
10558 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
10559 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
10560 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
10561 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
10562 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
10563 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
10564 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
10565 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
10566 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
10567 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
10568 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
10569 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
10570 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
10571 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
10572 * config/mips/mips.c (mips_truly_noop_truncation): New function.
10573 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10574 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10575 rather than TRULY_NOOP_TRUNCATION in comments.
10576 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
10577 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
10578 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
10579 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
10580 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
10581 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
10582 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
10583 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
10584 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
10585 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
10586 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
10587 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10588 rather than TRULY_NOOP_TRUNCATION in comments.
10589 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
10590 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
10591 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
10592 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
10593 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
10594 TRULY_NOOP_TRUNCATION condition.
10595 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
10596 (TRULY_NOOP_TRUNCATION): Delete.
10597 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
10598 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
10599 * config/spu/spu.c (spu_truly_noop_truncation): New function.
10600 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10601 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
10602 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
10603 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
10604 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10605 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10606 rather than TRULY_NOOP_TRUNCATION in comments.
10607 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
10608 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
10609 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
10610 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
10611 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
10612 * system.h (TRULY_NOOP_TRUNCATION): Poison.
10614 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
10617 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
10618 (*cmp_ior): Likewise.
10619 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
10620 (*ior_scc_scc_cmp): Likewise.
10621 (*and_scc_scc): Likewise.
10622 (*and_scc_scc_cmp): Likewise.
10624 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10625 Alan Hayard <alan.hayward@arm.com>
10626 David Sherwood <david.sherwood@arm.com>
10628 * target.def (can_change_mode_class): New hook.
10629 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
10630 (hard_regno_nregs): Likewise.
10631 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
10632 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
10633 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
10634 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
10635 (LOAD_EXTEND_OP): Update accordingly.
10636 * doc/tm.texi: Regenerate.
10637 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
10638 CANNOT_CHANGE_MODE_CLASS.
10639 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
10640 (REG_CAN_CHANGE_MODE_P): ...this new macro.
10641 * combine.c (simplify_set): Update accordingly.
10642 * emit-rtl.c (validate_subreg): Likewise.
10643 * recog.c (general_operand): Likewise.
10644 * regcprop.c (mode_change_ok): Likewise.
10645 * reload1.c (choose_reload_regs): Likewise.
10646 (inherit_piecemeal_p): Likewise.
10647 * rtlanal.c (simplify_subreg_regno): Likewise.
10648 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
10649 instead of CANNOT_CHANGE_MODE_CLASS.
10650 (reload_cse_simplify_operands): Likewise.
10651 * reload.c (push_reload): Use targetm.can_change_mode_class
10652 instead of CANNOT_CHANGE_MODE_CLASS.
10653 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
10654 REG_CANNOT_CHANGE_MODE_P.
10655 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10656 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
10657 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10658 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10659 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10660 (arm_can_change_mode_class): New function.
10661 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10662 than CANNOT_CHANGE_MODE_CLASS in comments.
10663 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10664 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
10665 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
10666 (ix86_can_change_mode_class): ...this new function, inverting the
10667 sense of the return value.
10668 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10669 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10670 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10671 (ia64_can_change_mode_class): New function.
10672 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10673 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
10674 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
10675 (m32c_can_change_mode_class): ...this new function, inverting the
10676 sense of the return value.
10677 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10678 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10679 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
10680 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
10681 (mips_can_change_mode_class): ...this new function, inverting the
10682 sense of the return value.
10683 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10684 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10685 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10686 (msp430_can_change_mode_class): New function.
10687 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10688 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
10689 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10690 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10691 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10692 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
10693 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10694 (pa_cannot_change_mode_class): Replace with...
10695 (pa_can_change_mode_class): ...this new function, inverting the
10696 sense of the return value.
10697 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10698 than CANNOT_CHANGE_MODE_CLASS in comments.
10699 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10700 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
10701 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10702 (pdp11_cannot_change_mode_class): Replace with...
10703 (pdp11_can_change_mode_class): ...this new function, inverting the
10704 sense of the return value.
10705 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10706 * config/powerpcspe/powerpcspe-protos.h
10707 (rs6000_cannot_change_mode_class_ptr): Delete.
10708 * config/powerpcspe/powerpcspe.c
10709 (rs6000_cannot_change_mode_class_ptr): Delete.
10710 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10711 (rs6000_option_override_internal): Assign to
10712 targetm.can_change_mode_class instead of
10713 rs6000_cannot_change_mode_class_ptr.
10714 (rs6000_cannot_change_mode_class): Replace with...
10715 (rs6000_can_change_mode_class): ...this new function, inverting the
10716 sense of the return value.
10717 (rs6000_debug_cannot_change_mode_class): Replace with...
10718 (rs6000_debug_can_change_mode_class): ...this new function.
10719 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10720 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
10721 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10722 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10723 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
10725 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
10726 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10727 (rs6000_option_override_internal): Assign to
10728 targetm.can_change_mode_class instead of
10729 rs6000_cannot_change_mode_class_ptr.
10730 (rs6000_cannot_change_mode_class): Replace with...
10731 (rs6000_can_change_mode_class): ...this new function, inverting the
10732 sense of the return value.
10733 (rs6000_debug_cannot_change_mode_class): Replace with...
10734 (rs6000_debug_can_change_mode_class): ...this new function.
10735 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10736 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
10737 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
10738 (s390_can_change_mode_class): ...this new function, inverting the
10739 sense of the return value.
10740 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10741 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10742 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
10743 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10744 (sh_cannot_change_mode_class): Replace with...
10745 (sh_can_change_mode_class): ...this new function, inverting the
10746 sense of the return value.
10747 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10748 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10749 (sparc_can_change_mode_class): New function.
10750 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10751 * config/spu/spu.c (spu_can_change_mode_class): New function.
10752 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10753 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10754 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10755 (visium_can_change_mode_class): New function.
10756 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
10758 2017-09-15 Richard Biener <rguenther@suse.de>
10760 PR tree-optimization/82217
10761 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
10762 but not undefined case.
10764 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10767 * postreload.c (reload_cse_simplify_operands): Skip
10768 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
10770 2017-09-15 Richard Biener <rguenther@suse.de>
10772 PR tree-optimization/68823
10773 * graphite-scop-detection.c (build_alias_set): If we have a
10774 possible dependence check whether we can handle them by just
10775 looking at the DRs DR_ACCESS_FNs.
10776 (build_scops): If build_alias_set fails, fail the SCOP.
10778 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
10780 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
10781 to support float128 built-in functions that require the ISA 3.0
10783 (BU_FLOAT128_3_HW): Likewise.
10784 (SQRTF128): Add support for the IEEE 128-bit square root and fma
10785 built-in functions.
10786 (FMAF128): Likewise.
10788 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
10789 support for built-in functions that need the ISA 3.0 IEEE 128-bit
10790 floating point instructions.
10791 (rs6000_invalid_builtin): Likewise.
10792 (rs6000_builtin_mask_names): Likewise.
10793 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
10794 (RS6000_BTM_FLOAT128_HW): Likewise.
10795 (RS6000_BTM_COMMON): Likewise.
10796 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
10798 * doc/extend.texi (RS/6000 built-in functions): Document the
10799 IEEE 128-bit floating point square root and fused multiply-add
10800 built-in functions.
10802 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
10804 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
10805 reg (r2) isn't in the set of registers defined in the prologue.
10807 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10808 Alan Hayward <alan.hayward@arm.com>
10809 David Sherwood <david.sherwood@arm.com>
10811 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
10812 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
10813 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
10814 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
10815 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
10817 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
10818 max_vectorization_factor.
10819 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
10821 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10822 Alan Hayward <alan.hayward@arm.com>
10823 David Sherwood <david.sherwood@arm.com>
10825 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
10826 (vect_worthwhile_without_simd_p): Declare.
10827 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
10828 (vectorizable_reduction): Use it.
10829 * tree-vect-stmts.c (vectorizable_shift): Likewise.
10830 (vectorizable_operation): Likewise.
10832 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10833 Alan Hayward <alan.hayward@arm.com>
10834 David Sherwood <david.sherwood@arm.com>
10836 * tree-vectorizer.h (vect_get_num_copies): New function.
10837 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
10838 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10839 (vectorizable_induction): Likewise.
10840 (vectorizable_live_operation): Likewise.
10841 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10842 (vectorizable_bswap): Likewise.
10843 (vectorizable_call): Likewise.
10844 (vectorizable_conversion): Likewise.
10845 (vectorizable_assignment): Likewise.
10846 (vectorizable_shift): Likewise.
10847 (vectorizable_operation): Likewise.
10848 (vectorizable_store): Likewise.
10849 (vectorizable_load): Likewise.
10850 (vectorizable_condition): Likewise.
10851 (vectorizable_comparison): Likewise.
10852 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
10854 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10855 Alan Hayward <alan.hayward@arm.com>
10856 David Sherwood <david.sherwood@arm.com>
10858 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
10859 of vect_init_vector.
10861 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10862 Alan Hayward <alan.hayward@arm.com>
10863 David Sherwood <david.sherwood@arm.com>
10865 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
10866 an inline wrapper that provides a location.
10867 (gimple_build_vector): Likewise.
10868 * gimple-fold.c (gimple_build_vector_from_val): New function.
10869 (gimple_build_vector): Likewise.
10870 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
10871 functions to build the initial value. Always return a gimple value.
10872 (get_initial_defs_for_reduction): Likewise. Only compute
10874 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
10875 vect_init_vector on the results from get_initial_def(s)_for_reduction.
10876 (vectorizable_induction): Use gimple_build_vector rather than
10879 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10880 Alan Hayward <alan.hayward@arm.com>
10881 David Sherwood <david.sherwood@arm.com>
10883 * target.h (vec_perm_indices): New typedef.
10884 (auto_vec_perm_indices): Likewise.
10885 * optabs-query.h: Include target.h
10886 (can_vec_perm_p): Take a vec_perm_indices *.
10887 * optabs-query.c (can_vec_perm_p): Likewise.
10888 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
10889 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
10890 * tree-vect-generic.c (lower_vec_perm): Likewise.
10891 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
10892 (vect_grouped_load_supported): Likewise.
10893 (vect_shift_permute_load_chain): Likewise.
10894 (vect_permute_store_chain): Use auto_vec_perm_indices.
10895 (vect_permute_load_chain): Likewise.
10896 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
10897 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
10898 Update uses of can_vec_perm_p.
10899 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
10900 mode with a number of elements. Take a vec_perm_indices *.
10901 (vect_create_epilog_for_reduction): Update accordingly.
10902 Use auto_vec_perm_indices.
10903 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
10904 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
10905 (vect_transform_slp_perm_load): Likewise.
10906 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
10907 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
10908 (vect_gen_perm_mask_checked): Likewise.
10909 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
10910 (vect_gen_perm_mask_checked): Likewise.
10911 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
10912 (vectorizable_store): Likewise.
10913 (vectorizable_load): Likewise.
10914 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
10915 (vectorizable_bswap): Likewise.
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.h (build_vector): Take a vec<tree> instead of a tree *.
10922 * tree.c (build_vector): Likewise.
10923 (build_vector_from_ctor): Update accordingly.
10924 (build_vector_from_val): Likewise.
10925 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
10926 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
10927 * tree-vect-generic.c (add_rshift): Likewise.
10928 (expand_vector_divmod): Likewise.
10929 (optimize_vector_constructor): Likewise.
10930 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
10931 (vect_transform_slp_perm_load): Likewise.
10932 (vect_schedule_slp_instance): Likewise.
10933 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
10934 (vectorizable_call): Likewise.
10935 (vect_gen_perm_mask_any): Likewise. Add elements in order.
10936 * expmed.c (make_tree): Likewise.
10937 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
10938 a vector passed to build_vector.
10939 (fold_convert_const): Likewise.
10940 (exact_inverse): Likewise.
10941 (fold_ternary_loc): Likewise.
10942 (fold_relational_const): Likewise.
10943 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
10944 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
10945 (const_unop): Likewise. Store the reduction accumulator in a
10946 variable rather than an array.
10947 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
10948 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
10949 the new vector, rather than constructing it after the input arrays.
10950 (native_interpret_vector): Use auto_vec<tree> when building
10951 a vector passed to build_vector. Add elements in order.
10952 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
10953 auto_vec<tree> when building a vector passed to build_vector.
10954 (vect_create_epilog_for_reduction): Likewise.
10955 (vectorizable_induction): Likewise.
10956 (get_initial_def_for_reduction): Likewise. Fix indentation of
10958 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
10960 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
10961 passed to build_vector.
10963 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10964 Alan Hayward <alan.hayward@arm.com>
10965 David Sherwood <david.sherwood@arm.com>
10967 * tree-core.h (tree_base::u): Add an "nelts" field.
10968 (tree_vector): Use VECTOR_CST_NELTS as the length.
10969 * tree.c (tree_size): Likewise.
10970 (make_vector): Initialize VECTOR_CST_NELTS.
10971 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
10972 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
10973 TYPE_VECTOR_SUBPARTS.
10974 * expr.c (const_vector_mask_from_tree): Consistently use "units"
10975 as the number of units, setting it from VECTOR_CST_NELTS.
10976 (const_vector_from_tree): Likewise.
10977 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
10978 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
10979 (fold_negate_expr_1): Likewise.
10980 (fold_convert_const): Likewise.
10981 (const_binop): Likewise. Differentiate the number of output and
10983 (const_unop): Likewise.
10984 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
10985 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
10986 in cases that did the opposite.
10988 2017-09-14 Richard Biener <rguenther@suse.de>
10990 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
10993 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
10995 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
10997 2017-09-14 Jakub Jelinek <jakub@redhat.com>
11000 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
11001 if and where to split a bb, except for splitting before debug insn
11002 sequences followed by non-label real insn. Delete debug insns
11003 in between basic blocks.
11005 * combine.c (make_compound_operation_int): Formatting fixes.
11007 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
11008 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11009 * config/netbsd.h (LINK_EH_SPEC): Likewise.
11010 * config/sol2.h (LINK_EH_SPEC): Likewise.
11011 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11012 * config/s390/linux.h (LINK_SPEC): Likewise.
11013 * config/freebsd.h (LINK_EH_SPEC): Likewise.
11014 * config/openbsd.h (LINK_EH_SPEC): Likewise.
11015 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11016 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
11017 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
11018 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11019 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
11020 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
11022 2017-09-13 Jakub Jelinek <jakub@redhat.com>
11024 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
11026 (ENDFILE_LINUX_SPEC): Likewise.
11027 (LINK_EH_SPEC): Likewise.
11028 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
11029 (LINK_OS_LINUX_SPEC32): Likewise.
11030 (LINK_OS_LINUX_SPEC64): Likewise.
11031 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
11032 (LINK_OS_LINUX_SPEC): Likewise.
11034 2017-09-13 Martin Liska <mliska@suse.cz>
11036 PR middle-end/82154
11037 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
11038 CASE_HIGH is NULL_TREE.
11040 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
11041 Alan Hayward <alan.hayward@arm.com>
11042 David Sherwood <david.sherwood@arm.com>
11044 * target.def (secondary_memory_needed): New hook.
11045 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
11046 instead of SECONDARY_MEMORY_NEEDED.
11047 (secondary_memory_needed_mode): Likewise.
11048 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
11049 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
11050 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
11051 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
11052 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
11053 * doc/tm.texi: Regenerate.
11054 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
11055 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
11056 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11057 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
11058 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
11059 * config/i386/i386.c (inline_secondary_memory_needed): Put the
11060 mode argument first and change the reg_class arguments to reg_class_t.
11061 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
11062 Make static. Update the call to inline_secondary_memory_needed.
11063 (ix86_register_move_cost): Update the call to
11064 inline_secondary_memory_needed.
11065 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11066 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
11068 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
11069 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11071 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
11072 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
11073 * config/mips/mips.c (mips_secondary_memory_needed): Make static
11074 and match hook interface. Add comment from mips.h.
11075 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11076 * config/mmix/mmix.md (truncdfsf2): Refer to
11077 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11079 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
11080 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
11081 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11082 (pa_secondary_memory_needed): New function.
11083 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
11084 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
11085 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11086 (pdp11_secondary_memory_needed): Make static and match hook interface.
11087 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
11088 * config/powerpcspe/powerpcspe-protos.h
11089 (rs6000_secondary_memory_needed_ptr): Delete.
11090 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
11092 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11093 (rs6000_option_override_internal): Assign to
11094 targetm.secondary_memory_needed rather than
11095 rs6000_secondary_memory_needed_ptr.
11096 (rs6000_secondary_memory_needed): Match hook interface.
11097 (rs6000_debug_secondary_memory_needed): Likewise.
11098 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
11099 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
11100 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
11101 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11102 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
11103 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
11105 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
11106 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11107 (rs6000_option_override_internal): Assign to
11108 targetm.secondary_memory_needed rather than
11109 rs6000_secondary_memory_needed_ptr.
11110 (rs6000_secondary_memory_needed): Match hook interface.
11111 (rs6000_debug_secondary_memory_needed): Likewise.
11112 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
11113 * config/s390/s390.c (s390_secondary_memory_needed): New function.
11114 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11115 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
11116 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11117 (sparc_secondary_memory_needed): New function.
11118 * lra-constraints.c (check_and_process_move): Refer to
11119 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11121 (curr_insn_transform): Likewise.
11122 (process_alt_operands): Use targetm.secondary_memory_needed
11123 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11124 (check_secondary_memory_needed_p): Likewise.
11125 (choose_split_class): Likewise.
11126 * reload.c: Unconditionally include code that was previously
11127 conditional on SECONDARY_MEMORY_NEEDED.
11128 (push_secondary_reload): Use targetm.secondary_memory_needed
11129 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11130 (push_reload): Likewise.
11131 * reload1.c: Unconditionally include code that was previously
11132 conditional on SECONDARY_MEMORY_NEEDED.
11133 (choose_reload_regs): Use targetm.secondary_memory_needed
11134 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11135 (gen_reload): Likewise.
11136 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
11138 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
11139 Alan Hayward <alan.hayward@arm.com>
11140 David Sherwood <david.sherwood@arm.com>
11142 * target.def (secondary_memory_needed_mode): New hook:
11143 * targhooks.c (default_secondary_memory_needed_mode): Declare.
11144 * targhooks.h (default_secondary_memory_needed_mode): New function.
11145 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
11146 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
11147 * doc/tm.texi: Regenerate.
11148 * lra-constraints.c (check_and_process_move): Use
11149 targetm.secondary_memory_needed_mode instead of
11150 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
11151 (curr_insn_transform): Likewise.
11152 * reload.c (get_secondary_mem): Likewise.
11153 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11154 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
11156 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11157 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11158 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
11159 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11160 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
11162 * config/powerpcspe/powerpcspe-protos.h
11163 (rs6000_secondary_memory_needed_mode): Delete.
11164 * config/powerpcspe/powerpcspe.c
11165 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11166 (rs6000_secondary_memory_needed_mode): Make static.
11167 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11168 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
11170 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
11172 (rs6000_secondary_memory_needed_mode): Make static.
11173 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11174 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
11175 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11176 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11177 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
11179 (sparc_secondary_memory_needed_mode): New function.
11180 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
11182 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
11184 * config/aarch64/constraints.md (Umq): New constraint.
11185 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
11187 (mov<mode>): Update condition.
11189 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11191 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
11192 when bitposition is the same.
11194 2017-09-13 Richard Biener <rguenther@suse.de>
11196 * dwarf2out.c (output_die_symbol): Remove.
11197 (output_die): Do not output a DIEs symbol.
11199 2017-09-13 Richard Biener <rguenther@suse.de>
11201 PR middle-end/82128
11202 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
11203 default-def to avoid breaking iterator update with the weird
11204 interaction with cgraph_update_edges_for_call_stmt_node.
11206 2017-09-13 Richard Biener <rguenther@suse.de>
11208 * tree-cfg.c (verify_gimple_assign_binary): Add verification
11209 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
11210 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
11211 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
11213 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
11215 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
11216 Disable pc relative literal load irrespective of
11217 TARGET_FIX_ERR_A53_84341 for default.
11219 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
11221 * config/sparc/sparc.c (output_return): Output the source location of
11222 the insn in the delay slot, if any.
11223 (output_sibcall): Likewise.
11225 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
11228 * common.opt (-static-pie): New alias.
11229 (shared): Negate static-pie.
11230 (-no-pie): Update help text.
11232 (static-pie): New option.
11233 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
11234 -static-pie support.
11235 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
11236 (LINK_EH_SPEC): Likewise.
11237 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11238 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11239 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11240 * gcc.c (LINK_COMMAND_SPEC): Likewise.
11241 (init_gcc_specs): Likewise.
11242 (init_spec): Likewise.
11243 (display_help): Update help message for -pie.
11244 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
11247 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
11249 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
11250 (movdi_aarch64): Likewise.
11251 (movti_aarch64): Likewise.
11253 2017-09-12 Simon Wright <simon@pushface.org>
11256 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
11257 calculation of the minor version, always output as 0.
11259 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11262 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
11263 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
11264 on it early, rather than manual conversion late. For
11265 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
11266 instead of performing manual conversion.
11268 2017-09-12 Carl Love <cel@us.ibm.com>
11270 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
11271 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
11273 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11274 VMULOSW): Add definitions.
11275 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11276 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11277 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11278 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11279 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11281 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11283 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
11285 (movti_aarch64): Likewise.
11286 (movdf_aarch64): Likewise.
11287 (movtf_aarch64): Likewise.
11288 (load_pairdi): Likewise.
11289 (store_pairdi): Likewise.
11290 (load_pairdf): Likewise.
11291 (store_pairdf): Likewise.
11292 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
11293 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
11294 (ldr_got_small_<mode>): Likewise.
11295 (ldr_got_small_28k_<mode>): Likewise.
11296 (ldr_got_tiny): Likewise.
11297 * config/aarch64/iterators.md (ldst_sz): New.
11298 (ldpstp_sz): Likewise.
11299 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
11301 (thunderx_load): Split load_8 to load_16.
11302 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
11304 (thunderx2t99_storepair_basic): Split store_8 to store_16.
11305 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
11306 (xgene1_store_pair): Split store_8 to store_16.
11307 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
11308 (falkor_st_0_st_sd): Split store_8 to store_16.
11310 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11312 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
11313 and store1/2/3/4 to store_4/8/12/16.
11314 * config/aarch64/aarch64.md: Update for rename.
11315 * config/arm/arm.md: Likewise.: Likewise.
11316 * config/arm/arm.c: Likewise.
11317 * config/arm/thumb1.md: Likewise.
11318 * config/arm/thumb2.md: Likewise.
11319 * config/arm/vfp.md: Likewise.
11320 * config/arm/arm-generic.md: Likewise.
11321 * config/arm/arm1020e.md: Likewise.
11322 * config/arm/arm1026ejs.md: Likewise.
11323 * config/arm/arm1136jfs.md: Likewise.
11324 * config/arm/arm926ejs.md: Likewise.
11325 * config/arm/cortex-a15.md: Likewise.
11326 * config/arm/cortex-a17.md: Likewise.
11327 * config/arm/cortex-a5.md: Likewise.
11328 * config/arm/cortex-a53.md: Likewise.
11329 * config/arm/cortex-a57.md: Likewise.
11330 * config/arm/cortex-a7.md: Likewise.
11331 * config/arm/cortex-a8.md: Likewise.
11332 * config/arm/cortex-a9.md: Likewise.
11333 * config/arm/cortex-m4.md: Likewise.
11334 * config/arm/cortex-m7.md: Likewise.
11335 * config/arm/cortex-r4.md: Likewise.
11336 * config/arm/exynos-m1.md: Likewise.
11337 * config/arm/fa526.md: Likewise.
11338 * config/arm/fa606te.md: Likewise.
11339 * config/arm/fa626te.md: Likewise.
11340 * config/arm/fa726te.md: Likewise.
11341 * config/arm/fmp626.md: Likewise.
11342 * config/arm/iwmmxt.md: Likewise.
11343 * config/arm/ldmstm.md: Likewise.
11344 * config/arm/marvell-pj4.md: Likewise.
11345 * config/arm/xgene1.md: Likewise.
11346 * config/aarch64/thunderx.md: Likewise.
11347 * config/aarch64/thunderx2t99.md: Likewise.
11348 * config/aarch64/falkor.md: Likewise.
11350 2017-09-12 Martin Liska <mliska@suse.cz>
11352 * attribs.c (private_lookup_attribute): New function.
11353 * attribs.h (private_lookup_attribute): Declared here.
11354 (lookup_attribute): Called from this place.
11356 2017-09-12 Richard Biener <rguenther@suse.de>
11358 PR tree-optimization/82157
11359 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
11360 stmts with side-effects.
11362 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11363 Alan Hayward <alan.hayward@arm.com>
11364 David Sherwood <david.sherwood@arm.com>
11366 * target.def (hard_regno_nregs): New hook.
11367 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
11368 * targhooks.h (default_hard_regno_nregs): Declare.
11369 * targhooks.c (default_hard_regno_nregs): New function.
11370 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
11371 (TARGET_HARD_REGNO_NREGS): ...this hook.
11372 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
11373 (CLASS_MAX_NREGS): Likewise.
11374 * doc/tm.texi: Regenerate.
11375 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
11376 instead of HARD_REGNO_NREGS.
11377 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
11378 HARD_REGNO_NREGS in the comment.
11379 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
11380 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
11381 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
11382 Return an unsigned int.
11383 (TARGET_HARD_REGNO_NREGS): Redefine.
11384 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
11385 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
11386 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11387 (arc_hard_regno_nregs): New function.
11388 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
11389 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
11390 (arm_hard_regno_nregs): New function.
11391 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
11392 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
11393 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
11394 (TARGET_HARD_REGNO_NREGS): Redefine.
11395 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
11396 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
11397 (HARD_REGNO_NREGS): Delete.
11398 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
11399 (cr16_hard_regno_nregs): New function.
11400 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
11401 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
11402 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
11403 (cris_hard_regno_nregs): New function.
11404 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
11405 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
11406 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11407 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
11408 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
11409 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
11410 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
11411 (frv_hard_regno_nregs): Make static. Take and return an
11413 (frv_class_max_nregs): Remove outdated copy of documentation.
11414 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
11415 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
11416 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
11417 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
11418 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
11419 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
11420 (TARGET_HARD_REGNO_NREGS): Redefine.
11421 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
11422 (CLASS_MAX_NREGS): Update comment.
11423 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
11424 (ia64_hard_regno_nregs): New function.
11425 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
11426 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
11427 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
11428 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
11429 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
11431 (m32c_hard_regno_nregs): Likewise. Make static.
11432 (TARGET_HARD_REGNO_NREGS): Redefine.
11433 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
11434 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
11435 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
11436 (m68k_hard_regno_nregs): New function.
11437 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
11438 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
11439 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
11440 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
11441 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
11442 Take and return an unsigned int.
11443 (TARGET_HARD_REGNO_NREGS): Redefine.
11444 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
11445 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11446 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
11447 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
11448 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
11449 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
11450 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
11451 (msp430_hard_regno_nregs): Make static. Take and return an
11453 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
11454 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
11455 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
11456 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
11457 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
11458 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
11459 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
11460 (TARGET_HARD_REGNO_NREGS): Redefine.
11461 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
11462 (PA_HARD_REGNO_NREGS): ...this.
11463 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
11464 (PA_HARD_REGNO_NREGS): ...this.
11465 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
11466 (pa_hard_regno_nregs): New function.
11467 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
11468 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
11469 (pdp11_hard_regno_nregs): New function.
11470 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
11471 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
11472 (rs6000_hard_regno_nregs_hook): New function.
11473 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
11474 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
11475 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
11476 Take and return an unsigned int. Move earlier in file.
11477 (TARGET_HARD_REGNO_NREGS): Redefine.
11478 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
11479 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
11480 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
11481 (rl78_hard_regno_nregs): Make static. Take and return an
11483 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
11484 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
11485 (rs6000_hard_regno_nregs_hook): New function.
11486 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
11487 * config/rx/rx.c (rx_hard_regno_nregs): New function.
11488 (TARGET_HARD_REGNO_NREGS): Redefine.
11489 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
11490 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
11491 instead of HARD_REGNO_NREGS.
11492 (s390_hard_regno_nregs): New function.
11493 (s390_hard_regno_mode_ok): Add comment from s390.h.
11494 (TARGET_HARD_REGNO_NREGS): Redefine.
11495 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
11496 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
11497 (sh_hard_regno_nregs): New function.
11498 (sh_pass_in_reg_p): Use it.
11499 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
11500 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11501 (sparc_hard_regno_nregs): New function.
11502 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
11503 * config/spu/spu.c (spu_hard_regno_nregs): New function.
11504 (spu_function_arg_advance): Use it, supplying a valid register number.
11505 (TARGET_HARD_REGNO_NREGS): Redefine.
11506 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
11507 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
11508 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
11509 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
11510 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
11511 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
11512 (CLASS_MAX_NREGS): Remove copy of old documentation.
11513 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
11514 (visium_hard_regno_nregs): New function.
11515 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
11516 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
11517 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
11518 xtensa_hard_regno_nregs): New function.
11519 * system.h (HARD_REGNO_NREGS): Poison.
11521 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11523 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
11524 hard_regno_nregs instead of HARD_REGNO_NREGS.
11525 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
11526 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
11527 (c6x_expand_epilogue): Likewise.
11528 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
11529 (frv_read_iacc_argument): Likewise.
11530 * config/sh/sh.c: Include regs.h.
11531 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
11532 (regs_used): Likewise.
11533 (output_stack_adjust): Likewise.
11534 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
11535 * expmed.c: Include regs.h.
11536 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
11537 * ree.c: Include regs.h.
11538 (combine_reaching_defs): Use hard_regno_nregs instead of
11540 (add_removable_extension): Likewise.
11542 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11544 * regs.h (hard_regno_nregs): Turn into a function.
11545 (end_hard_regno): Update accordingly.
11546 * caller-save.c (setup_save_areas): Likewise.
11547 (save_call_clobbered_regs): Likewise.
11548 (replace_reg_with_saved_mem): Likewise.
11549 (insert_restore): Likewise.
11550 (insert_save): Likewise.
11551 * combine.c (can_change_dest_mode): Likewise.
11552 (move_deaths): Likewise.
11553 (distribute_notes): Likewise.
11554 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
11555 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
11556 (rs6000_split_multireg_move): Likewise.
11557 (rs6000_register_move_cost): Likewise.
11558 (rs6000_memory_move_cost): Likewise.
11559 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
11560 (rs6000_split_multireg_move): Likewise.
11561 (rs6000_register_move_cost): Likewise.
11562 (rs6000_memory_move_cost): Likewise.
11563 * cselib.c (cselib_reset_table): Likewise.
11564 (cselib_lookup_1): Likewise.
11565 * emit-rtl.c (set_mode_and_regno): Likewise.
11566 * function.c (aggregate_value_p): Likewise.
11567 * ira-color.c (setup_profitable_hard_regs): Likewise.
11568 (check_hard_reg_p): Likewise.
11569 (calculate_saved_nregs): Likewise.
11570 (assign_hard_reg): Likewise.
11571 (improve_allocation): Likewise.
11572 (calculate_spill_cost): Likewise.
11573 * ira-emit.c (modify_move_list): Likewise.
11574 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
11575 (ira_hard_reg_in_set_p): Likewise.
11576 * ira.c (setup_reg_mode_hard_regset): Likewise.
11577 (clarify_prohibited_class_mode_regs): Likewise.
11578 (check_allocation): Likewise.
11579 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11580 (lra_setup_reg_renumber): Likewise.
11581 (setup_try_hard_regno_pseudos): Likewise.
11582 (spill_for): Likewise.
11583 (assign_hard_regno): Likewise.
11584 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
11585 * lra-constraints.c (in_class_p): Likewise.
11586 (lra_constraint_offset): Likewise.
11587 (simplify_operand_subreg): Likewise.
11588 (lra_constraints): Likewise.
11589 (split_reg): Likewise.
11590 (split_if_necessary): Likewise.
11591 (invariant_p): Likewise.
11592 (inherit_in_ebb): Likewise.
11593 * lra-lives.c (process_bb_lives): Likewise.
11594 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
11595 (get_hard_regs): Likewise.
11596 (do_remat): Likewise.
11597 * lra-spills.c (assign_spill_hard_regs): Likewise.
11598 * mode-switching.c (create_pre_exit): Likewise.
11599 * postreload.c (reload_combine_recognize_pattern): Likewise.
11600 * recog.c (peep2_find_free_register): Likewise.
11601 * regcprop.c (kill_value_regno): Likewise.
11602 (set_value_regno): Likewise.
11603 (copy_value): Likewise.
11604 (maybe_mode_change): Likewise.
11605 (find_oldest_value_reg): Likewise.
11606 (copyprop_hardreg_forward_1): Likewise.
11607 * regrename.c (check_new_reg_p): Likewise.
11608 (regrename_do_replace): Likewise.
11609 * reload.c (push_reload): Likewise.
11610 (combine_reloads): Likewise.
11611 (find_dummy_reload): Likewise.
11612 (operands_match_p): Likewise.
11613 (find_reloads): Likewise.
11614 (find_equiv_reg): Likewise.
11615 (reload_adjust_reg_for_mode): Likewise.
11616 * reload1.c (count_pseudo): Likewise.
11617 (count_spilled_pseudo): Likewise.
11618 (find_reg): Likewise.
11619 (clear_reload_reg_in_use): Likewise.
11620 (free_for_value_p): Likewise.
11621 (allocate_reload_reg): Likewise.
11622 (choose_reload_regs): Likewise.
11623 (reload_adjust_reg_for_temp): Likewise.
11624 (emit_reload_insns): Likewise.
11625 (delete_output_reload): Likewise.
11626 * rtlanal.c (subreg_get_info): Likewise.
11627 * sched-deps.c (sched_analyze_reg): Likewise.
11628 * sel-sched.c (init_regs_for_mode): Likewise.
11629 (mark_unavailable_hard_regs): Likewise.
11630 (choose_best_reg_1): Likewise.
11631 (verify_target_availability): Likewise.
11632 * valtrack.c (dead_debug_insert_temp): Likewise.
11633 * var-tracking.c (track_loc_p): Likewise.
11634 (emit_note_insn_var_location): Likewise.
11635 * varasm.c (make_decl_rtl): Likewise.
11636 * reginfo.c (choose_hard_reg_mode): Likewise.
11637 (init_reg_modes_target): Refer directly to
11638 this_target_regs->x_hard_regno_nregs.
11640 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11642 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
11643 instead of hard_regno_nregs.
11645 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11647 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
11648 end_hard_regno instead of hard_regno_nregs.
11649 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
11650 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
11651 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
11652 * ira-color.c (improve_allocation): Likewise.
11653 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11654 * lra-lives.c (mark_regno_live): Likewise.
11655 (mark_regno_dead): Likewise.
11656 * lra-remat.c (operand_to_remat): Likewise.
11657 * lra.c (collect_non_operand_hard_regs): Likewise.
11658 * postreload.c (reload_combine_note_store): Likewise.
11659 (move2add_valid_value_p): Likewise.
11660 * reload.c (regno_clobbered_p): Likewise.
11662 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11664 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
11666 * config/v850/v850.c (v850_reorg): Likewise.
11667 * reload.c (refers_to_regno_for_reload_p): Likewise.
11668 (find_equiv_reg): Likewise.
11669 * reload1.c (reload_reg_reaches_end_p): Likewise.
11671 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11673 * caller-save.c (add_used_regs): Use REG_NREGS instead of
11675 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
11676 * config/arm/arm.c (output_move_neon): Likewise.
11677 (arm_attr_length_move_neon): Likewise.
11678 (neon_split_vcombine): Likewise.
11679 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
11680 (c6x_mark_reg_written): Likewise.
11681 (c6x_dwarf_register_span): Likewise.
11682 * config/i386/i386.c (ix86_save_reg): Likewise.
11683 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
11684 (rws_access_reg): Likewise.
11685 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
11686 * mode-switching.c (create_pre_exit): Likewise.
11687 * ree.c (combine_reaching_defs): Likewise.
11688 (add_removable_extension): Likewise.
11689 * regcprop.c (find_oldest_value_reg): Likewise.
11690 (copyprop_hardreg_forward_1): Likewise.
11691 * reload.c (reload_inner_reg_of_subreg): Likewise.
11692 (push_reload): Likewise.
11693 (combine_reloads): Likewise.
11694 (find_dummy_reload): Likewise.
11695 (reload_adjust_reg_for_mode): Likewise.
11696 * reload1.c (find_reload_regs): Likewise.
11697 (forget_old_reloads_1): Likewise.
11698 (reload_reg_free_for_value_p): Likewise.
11699 (reload_adjust_reg_for_temp): Likewise.
11700 (emit_reload_insns): Likewise.
11701 (delete_output_reload): Likewise.
11702 * sel-sched.c (choose_best_reg_1): Likewise.
11703 (choose_best_pseudo_reg): Likewise.
11705 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11706 Alan Hayward <alan.hayward@arm.com>
11707 David Sherwood <david.sherwood@arm.com>
11709 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
11710 * target.def (slow_unaligned_access): New hook.
11711 * targhooks.h (default_slow_unaligned_access): Declare.
11712 * targhooks.c (default_slow_unaligned_access): New function.
11713 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
11714 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
11715 * doc/tm.texi: Regenerate.
11716 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
11717 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
11718 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
11720 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
11721 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
11723 (rs6000_slow_unaligned_access): New function.
11724 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
11725 (expand_block_compare): Likewise.
11726 (expand_strn_compare): Likewise.
11727 (rs6000_rtx_costs): Likewise.
11728 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
11729 (riscv_slow_unaligned_access): Likewise.
11730 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
11731 (riscv_slow_unaligned_access_p): ...this and make static.
11732 (riscv_option_override): Update accordingly.
11733 (riscv_slow_unaligned_access): New function.
11734 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
11735 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
11736 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
11737 (rs6000_slow_unaligned_access): New function.
11738 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
11739 (rs6000_rtx_costs): Likewise.
11740 * config/rs6000/rs6000-string.c (expand_block_compare)
11741 (expand_strn_compare): Use targetm.slow_unaligned_access instead
11742 of SLOW_UNALIGNED_ACCESS.
11743 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
11744 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
11745 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
11746 of SLOW_UNALIGNED_ACCESS.
11747 * expmed.c (simple_mem_bitfield_p): Likewise.
11748 * expr.c (alignment_for_piecewise_move): Likewise.
11749 (emit_group_load_1): Likewise.
11750 (emit_group_store): Likewise.
11751 (copy_blkmode_from_reg): Likewise.
11752 (emit_push_insn): Likewise.
11753 (expand_assignment): Likewise.
11754 (store_field): Likewise.
11755 (expand_expr_real_1): Likewise.
11756 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
11757 * lra-constraints.c (simplify_operand_subreg): Likewise.
11758 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11759 * gimple-ssa-store-merging.c: Likewise in block comment at start
11761 * tree-ssa-strlen.c: Include target.h.
11762 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
11763 of SLOW_UNALIGNED_ACCESS.
11764 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
11766 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11768 PR rtl-optimization/82185
11769 * expmed.c (emit_store_flag_int): Only test tem if it has been
11772 2017-09-12 Richard Biener <rguenther@suse.de>
11774 PR middle-end/82149
11775 * match.pd ((FTYPE) N CMP CST): Fix typo.
11777 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
11779 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
11781 (mips_near_type_p): Add 'short_call' attribute as a synonym
11783 * doc/extend.texi (short_call): Document new function attribute.
11785 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11788 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
11789 assertion check that in the condition.
11790 (get_atomic_generic_size): Likewise. Before testing if parameter
11791 has pointer type, if it has array type, call for C++
11792 default_conversion to perform array-to-pointer conversion.
11794 2017-09-12 Richard Biener <rguenther@suse.de>
11796 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
11797 for operations we cannot scalarize.
11799 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
11801 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
11802 vectors heap vectors. Clean up comments.
11803 Make visited_bbs a reference.
11804 (profitable_jump_thread_path): Make GC
11805 vectors heap vectors. Clean up comments.
11807 (convert_and_register_jump_thread_path): Make GC vectors heap
11809 (check_subpath_and_update_thread_path): Same. Clean up comments.
11810 Make visited_bbs a reference.
11811 (handle_phi): Abstract common code to to
11812 register_jump_thread_path_if_profitable.
11813 Rename VAR_BB to DEF_BB.
11815 Make GC vectors heap vectors.
11816 Make visited_bbs a reference.
11817 (handle_assignment): Same.
11818 (register_jump_thread_path_if_profitable): New.
11819 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
11821 Make GC vectors heap vectors. Clean up comments.
11822 Make visited_bbs a reference.
11823 (find_jump_threads_backwards): Make visited_bbs live in the stack.
11824 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
11827 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
11830 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
11831 words of E_DImode object are reachable by xtensa_uimm8x4 access.
11833 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
11835 Revert r251800 and r251799.
11837 2017-09-11 Martin Jambor <mjambor@suse.cz>
11840 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
11841 arguments in advance.
11842 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
11843 use it to find predecessor edges.
11844 (naive_outof_ssa): Collect vector of predecessors.
11846 2017-09-08 Jason Merrill <jason@redhat.com>
11848 PR c++/70029 - ICE with ref-qualifier and -flto
11849 * langhooks.h (struct lang_hooks_for_types): Add
11850 copy_lang_qualifiers.
11851 * attribs.c (build_type_attribute_qual_variant): Use it.
11852 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
11854 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
11855 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
11857 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
11860 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
11861 (*mulsi3_sp64): New instruction.
11862 (mulsi3): New expander.
11864 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
11866 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
11868 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11870 * sancov.c: Include memmodel.h.
11872 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
11875 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
11878 2017-09-07 Carl Love <cel@us.ibm.com>
11880 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
11881 the sldi instruction.
11883 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
11885 * sancov.c: Include tm_p.h.
11887 2017-09-07 Jakub Jelinek <jakub@redhat.com>
11890 * output.h (switch_to_other_text_partition): New declaration.
11891 * varasm.c (switch_to_other_text_partition): New function.
11892 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
11893 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
11894 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
11895 to the other text partition before emitting LCL label and switch back
11896 after emitting the word after it.
11898 2017-09-07 Richard Biener <rguenther@suse.de>
11900 * passes.def (pass_split_crit_edges): Remove instance before PRE.
11901 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
11902 critical edges here, after loop init.
11903 (pass_data_pre): Remove PROP_no_crit_edges flags.
11904 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
11905 for valueization of call args to avoid leaking VN_TOP.
11906 (visit_use): Assert we do not visit default defs.
11907 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
11908 Use error_mark_node to more easily detect leaking VN_TOP.
11909 All default-defs are varying, not VN_TOP. Mark them visited.
11910 (run_scc_vn): Make code match comment.
11912 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11914 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
11915 OPTION_MASK_FLOAT128_KEYWORD.
11916 (POWERPC_MASKS): Likewise.
11917 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
11918 support for the -mfloat128-type option, and make -mfloat128
11919 default on PowerPC Linux systems. Define or undefine
11920 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
11921 Define __float128 to be __ieee128 if IEEE 128-bit support is
11922 enabled, or undefine it.
11923 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
11924 Delete defining __FLOAT128_TYPE__.
11925 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
11926 -mfloat128-type option and make -mfloat128 default on PowerPC
11928 (TARGET_FLOAT128_TYPE): Likewise.
11929 (-mfloat128-type): Likewise.
11930 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11931 Delete the -mfloat128-type option and make -mfloat128 default on
11932 PowerPC Linux systems. Always use __ieee128 to be the keyword for
11933 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
11934 128-bit floating point is enabled. Change tests from using
11935 -mfloat128-type to -mfloat128.
11936 (rs6000_mangle_type): Use the correct mangling for the __float128
11937 type even if normal long double is restricted to 64-bits.
11938 (floatn_mode): Enable the _Float128 type by default on VSX Linux
11940 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
11941 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
11942 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
11944 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
11945 documentation for -mfloat128.
11947 2017-09-06 Olivier Hainque <hainque@adacore.com>
11949 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
11951 2017-09-06 Wish Wu <wishwu007@gmail.com>
11952 Jakub Jelinek <jakub@redhat.com>
11954 * asan.c (initialize_sanitizer_builtins): Add
11955 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
11956 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
11957 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
11958 BT_FN_VOID_UINT64_PTR variables.
11959 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
11960 (BT_FN_VOID_UINT16_UINT16): Likewise.
11961 (BT_FN_VOID_UINT32_UINT32): Likewise.
11962 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
11963 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
11964 (BT_FN_VOID_UINT64_PTR): Likewise.
11965 * common.opt (flag_sanitize_coverage): New variable.
11966 (fsanitize-coverage=trace-pc): Remove.
11967 (fsanitize-coverage=): Add.
11968 * flag-types.h (enum sanitize_coverage_code): New enum.
11969 * fold-const.c (fold_range_test): Disable non-short-circuit
11970 optimization if flag_sanitize_coverage.
11971 (fold_truth_andor): Likewise.
11972 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
11973 * opts.c (COVERAGE_SANITIZER_OPT): Define.
11974 (coverage_sanitizer_opts): New array.
11975 (get_closest_sanitizer_option): Add OPTS argument, handle also
11976 OPT_fsanitize_coverage_.
11977 (parse_sanitizer_options): Adjusted to also handle
11978 OPT_fsanitize_coverage_.
11979 (common_handle_option): Add OPT_fsanitize_coverage_.
11980 * sancov.c (instrument_comparison, instrument_switch): New function.
11981 (sancov_pass): Add trace-cmp support.
11982 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
11983 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
11984 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
11985 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
11986 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
11987 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
11988 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
11989 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
11990 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
11991 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
11993 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
11995 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11996 error. Only quit immediately if parsing is complete.
11997 (BEGIN): Initialize fatal_err and parse_done.
11998 (begin fpu, end fpu): Check number of arguments.
11999 (begin arch, end arch): Likewise.
12000 (begin cpu, end cpu): Likewise.
12001 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12002 (optalias): Likewise.
12004 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12006 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12007 * config/arm/arm-isa.h: Delete. Move definitions to ...
12008 * arm-cpus.in: ... here. Use new feature and fgroup values.
12009 * config/arm/arm.c (arm_option_override): Use lower case for feature
12011 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12012 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12013 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12015 (print_isa_bits_for): New function.
12016 (gen_isa): New function.
12017 (gen_comm_data): Use print_isa_bits_for.
12018 (define feature): New keyword.
12019 (define fgroup): New keyword.
12020 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
12021 (arm-isa.h): Add rule to generate file.
12022 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12023 case for feature bit names.
12025 2017-09-06 Richard Biener <rguenther@suse.de>
12027 * tree-ssa-pre.c (NECESSARY): Remove.
12028 (create_expression_by_pieces): Do not touch pass-local flags.
12029 (insert_into_preds_of_block): Likewise.
12030 (do_pre_regular_insertion): Likewise.
12031 (eliminate_insert): Likewise.
12032 (eliminate_dom_walker::before_dom_children): Likewise.
12033 (fini_eliminate): Do not look at inserted_exprs.
12034 (mark_operand_necessary): Remove.
12035 (remove_dead_inserted_code): Replace with simple work-list
12036 algorithm based on inserted_exprs and SSA uses.
12037 (pass_pre::execute): Re-order fini_eliminate and
12038 remove_dead_inserted_code.
12040 2017-09-06 Olivier Hainque <hainque@adacore.com>
12042 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
12043 for VxWorks 7. Adjust surrounding comments.
12045 2017-09-06 Richard Biener <rguenther@suse.de>
12047 * gimple-ssa-strength-reduction.c
12048 (find_candidates_dom_walker::before_dom_children): Also allow
12051 2017-09-06 Richard Biener <rguenther@suse.de>
12053 PR tree-optimization/82108
12054 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
12055 for gap in the non-permutation SLP case.
12057 2017-09-06 Martin Jambor <mjambor@suse.cz>
12059 PR tree-optimization/82078
12060 * tree-sra.c (sort_and_splice_var_accesses): Move call to
12061 add_access_to_work_queue...
12062 (build_accesses_from_assign): ...here.
12063 (propagate_all_subaccesses): Make sure racc is the group
12064 representative, if there is one.
12066 2017-09-06 Jakub Jelinek <jakub@redhat.com>
12068 PR middle-end/82095
12069 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
12072 2017-09-06 Richard Biener <rguenther@suse.de>
12074 * gimple-ssa-strength-reduction.c
12075 (find_candidates_dom_walker::before_doom_children): Use a
12076 type and not a mode check.
12078 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
12081 * config/arm/predicates.md (arm_general_adddi_operand): Create new
12083 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
12085 2017-09-05 Jeff Law <law@redhat.com>
12087 PR tree-optimization/64910
12088 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
12089 cases where we have 3 or more operands.
12091 2017-09-05 Jakub Jelinek <jakub@redhat.com>
12093 PR middle-end/81768
12094 * omp-low.c (lower_omp_for): Recompute tree invariant if
12095 gimple_omp_for_initial/final is ADDR_EXPR.
12097 PR middle-end/81768
12098 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
12099 into gimple val before gimplification fo the COND_EXPR.
12101 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
12103 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
12104 REGION_COPY argument.
12105 (thread_through_all_blocks): Remove unused argument to
12106 duplicate_thread_path.
12108 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12109 Alan Hayward <alan.hayward@arm.com>
12110 David Sherwood <david.sherwood@arm.com>
12112 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
12113 Take a scalar_mode rather than a machine_mode.
12114 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12115 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
12116 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12117 (aarch64_gen_adjusted_ldpstp): Likewise.
12118 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
12120 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12121 Alan Hayward <alan.hayward@arm.com>
12122 David Sherwood <david.sherwood@arm.com>
12124 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
12125 Take a scalar_int_mode instead of a machine_mode.
12126 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
12127 (aarch64_output_scalar_simd_mov_immediate): Likewise.
12128 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
12129 (aarch64_simd_attr_length_rglist): Delete.
12130 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
12131 a scalar_int_mode instead of a machine_mode.
12132 (aarch64_add_offset): Likewise.
12133 (aarch64_internal_mov_immediate): Likewise
12134 (aarch64_add_constant_internal): Likewise.
12135 (aarch64_add_constant): Likewise.
12136 (aarch64_movw_imm): Likewise.
12137 (aarch64_rtx_arith_op_extract_p): Likewise.
12138 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
12139 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
12140 Remove assert that the mode isn't a vector.
12141 (aarch64_output_scalar_simd_mov_immediate): Likewise.
12142 (aarch64_expand_mov_immediate): Update calls after above changes.
12143 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
12144 (aarch64_and_bitmask_imm): Check for scalar integer modes.
12145 (aarch64_move_imm): Likewise.
12146 (aarch64_can_const_movi_rtx_p): Likewise.
12147 (aarch64_strip_extend): Likewise.
12148 (aarch64_extr_rtx_p): Likewise.
12149 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
12150 a CONST_INT when the mode parameter is VOIDmode.
12151 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
12153 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12155 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
12156 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12157 (bitwise_type_for_mode): Update accordingly.
12159 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12161 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
12162 * stor-layout.c (mode_for_size_tree): Likewise.
12163 (mode_for_array): Update accordingly.
12164 (layout_decl): Likewise.
12165 (compute_record_mode): Likewise. Only set the mode once.
12167 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12169 * target.def (get_mask_mode): Change return type to opt_mode.
12171 * doc/tm.texi: Regenerate.
12172 * targhooks.h (default_get_mask_mode): Return an opt_mode.
12173 * targhooks.c (default_get_mask_mode): Likewise.
12174 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12175 * optabs-query.c (can_vec_mask_load_store_p): Update use of
12176 targetm.get_mask_mode.
12177 * tree.c (build_truth_vector_type): Likewise.
12179 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12181 * machmode.h (mode_for_vector): Return an opt_mode.
12182 * stor-layout.c (mode_for_vector): Likewise.
12183 (mode_for_int_vector): Update accordingly.
12184 (layout_type): Likewise.
12185 * config/i386/i386.c (emit_memmov): Likewise.
12186 (ix86_expand_set_or_movmem): Likewise.
12187 (ix86_expand_vector_init): Likewise.
12188 (ix86_get_mask_mode): Likewise.
12189 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
12191 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
12192 * expmed.c (extract_bit_field_1): Likewise.
12193 * expr.c (expand_expr_real_2): Likewise.
12194 * optabs-query.c (can_vec_perm_p): Likewise.
12195 (can_vec_mask_load_store_p): Likewise.
12196 * optabs.c (expand_vec_perm): Likewise.
12197 * targhooks.c (default_get_mask_mode): Likewise.
12198 * tree-vect-stmts.c (vectorizable_store): Likewise.
12199 (vectorizable_load): Likewise.
12200 (get_vectype_for_scalar_type_and_size): Likewise.
12202 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12204 * machmode.h (mode_for_int_vector): New function.
12205 * stor-layout.c (mode_for_int_vector): Likewise.
12206 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
12207 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
12208 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12209 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12210 (s390_expand_vcond): Likewise.
12212 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12214 * machmode.h (opt_machine_mode): New type.
12215 (opt_mode<T>): Allow construction from anything that can be
12217 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
12218 (mode_for_size): Return an opt_machine_mode.
12219 * stor-layout.c (mode_for_size): Likewise.
12220 (mode_for_size_tree): Update call accordingly.
12221 (bitwise_mode_for_mode): Likewise.
12222 (make_fract_type): Likewise.
12223 (make_accum_type): Likewise.
12224 * caller-save.c (replace_reg_with_saved_mem): Update call
12226 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12227 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12228 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12229 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12230 * expmed.c (extract_bit_field_1): Likewise.
12231 * reload.c (get_secondary_mem): Likewise.
12232 * varasm.c (assemble_integer): Likewise.
12233 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
12236 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12238 * machmode.h (decimal_float_mode_for_size): New function.
12239 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
12240 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
12241 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
12242 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
12243 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
12244 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
12246 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12248 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
12249 (get_builtin_sync_mode): Likewise.
12250 (expand_ifn_atomic_compare_exchange): Likewise.
12251 (expand_builtin_atomic_clear): Likewise.
12252 (expand_builtin_atomic_test_and_set): Likewise.
12253 (fold_builtin_atomic_always_lock_free): Likewise.
12254 * calls.c (compute_argument_addresses): Likewise.
12255 (emit_library_call_value_1): Likewise.
12256 (store_one_arg): Likewise.
12257 * combine.c (combine_instructions): Likewise.
12258 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
12259 * config/arm/arm.c (arm_function_value): Likewise.
12260 (aapcs_allocate_return_reg): Likewise.
12261 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
12262 * config/i386/i386.c (construct_container): Likewise.
12263 (ix86_gimplify_va_arg): Likewise.
12264 (ix86_expand_sse_cmp): Likewise.
12265 (emit_memmov): Likewise.
12266 (emit_memset): Likewise.
12267 (expand_small_movmem_or_setmem): Likewise.
12268 (ix86_expand_pextr): Likewise.
12269 (ix86_expand_pinsr): Likewise.
12270 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
12271 * config/microblaze/microblaze.c (microblaze_block_move_straight):
12273 * config/mips/mips.c (mips_function_value_1) Likewise.
12274 (mips_block_move_straight): Likewise.
12275 (mips_expand_ins_as_unaligned_store): Likewise.
12276 * config/powerpcspe/powerpcspe.c
12277 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12278 (rs6000_darwin64_record_arg_flush): Likewise.
12279 * config/rs6000/rs6000.c
12280 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12281 (rs6000_darwin64_record_arg_flush): Likewise.
12282 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
12283 (sparc_function_value_1): Likewise.
12284 * config/spu/spu.c (adjust_operand): Likewise.
12285 (spu_emit_branch_or_set): Likewise.
12286 (arith_immediate_p): Likewise.
12287 * emit-rtl.c (gen_lowpart_common): Likewise.
12288 * expr.c (expand_expr_real_1): Likewise.
12289 * function.c (assign_parm_setup_block): Likewise.
12290 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
12291 * reload1.c (alter_reg): Likewise.
12292 * stor-layout.c (mode_for_vector): Likewise.
12293 (layout_type): Likewise.
12295 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12297 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
12298 (spu_convert_move): Likewise.
12299 * lower-subreg.c (resolve_simple_move): Likewise.
12301 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12304 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
12305 define_insn to a define_expand.
12306 (altivec_vsum2sws_direct): New define_insn.
12307 (altivec_vsumsws): Convert from a define_insn to a define_expand.
12309 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12311 * config/arm/arm.c (arm_option_params_internal): Improve setting of
12314 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
12319 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
12320 realign stack if stack alignment needed is less than incoming
12323 2017-09-05 Marek Polacek <polacek@redhat.com>
12326 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
12329 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12331 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
12333 2017-09-05 Richard Biener <rguenther@suse.de>
12335 PR tree-optimization/82084
12336 * fold-const.c (can_native_encode_string_p): Handle wide characters.
12338 2017-09-05 Richard Biener <rguenther@suse.de>
12340 PR tree-optimization/82102
12341 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
12343 2017-09-05 Martin Liska <mliska@suse.cz>
12345 PR tree-optimization/82032
12346 * tree-cfg.c (generate_range_test): New function.
12347 * tree-cfg.h (generate_range_test): Declared here.
12348 * tree-cfgcleanup.c (convert_single_case_switch): New function.
12349 (cleanup_control_expr_graph): Use it.
12350 * tree-switch-conversion.c (try_switch_expansion): Remove
12352 (emit_case_nodes): Use generate_range_test.
12354 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12357 * config/i386/i386.md (*<btsc><mode>_mask): Add
12358 TARGET_USE_BT to insn constraint.
12359 (*btr<mode>_mask): Ditto.
12361 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
12363 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
12364 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
12366 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
12369 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
12370 TARGET_NEON and TARGET_IWMMXT.
12371 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
12372 TARGET_NEON and TARGET_IWMMXT.
12373 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
12375 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12377 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
12378 (ix86_rewrite_tls_address): Ditto.
12379 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
12380 (ix86_rewrite_tls_address_1): Ditto.
12381 (ix86_rewrite_tls_address): Ditto.
12382 * config/i386/predicates.md (tls_address_pattern): New predicate.
12383 * config/i386/i386.md (TLS address splitter): New splitter.
12385 2017-09-04 Richard Biener <rguenther@suse.de>
12387 PR tree-optimization/82084
12388 * fold-const.h (can_native_encode_string_p): Declare.
12389 * fold-const.c (can_native_encode_string_p): Factor out from ...
12390 (native_encode_string): ... here.
12391 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
12392 vectorizing stores from constants we later cannot handle.
12394 2017-09-04 Marek Polacek <polacek@redhat.com>
12397 * doc/invoke.texi: Update -Wtautological-compare documentation.
12399 2017-09-04 Jeff Law <law@redhat.com>
12401 PR tree-optimization/64910
12402 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
12403 swap the first and last operand if the last is a constant.
12405 2017-09-04 Marek Polacek <polacek@redhat.com>
12408 * convert.c (do_narrow): When sanitizing signed integer overflows,
12409 bail out for signed types.
12410 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
12412 2017-09-04 Richard Biener <rguenther@suse.de>
12414 PR tree-optimization/82060
12415 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
12416 Move devirtualization after stmt folding and before EH/AB/noreturn
12417 cleanup to get the stmt refs canonicalized. Use a bool instead
12418 of gimple_modified_p since that doesn't work for NOPs. Schedule
12419 NOPs generated by folding for removal.
12421 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12422 Alan Hayward <alan.hayward@arm.com>
12423 David Sherwood <david.sherwood@arm.com>
12425 * coretypes.h (pad_direction): New enum.
12426 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
12427 (FUNCTION_ARG_PADDING): Likewise.
12428 * target.def (function_arg_padding): New hook.
12429 * targhooks.h (default_function_arg_padding): Declare.
12430 * targhooks.c (default_function_arg_padding): New function.
12431 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
12432 (TARGET_FUNCTION_ARG_PADDING): ...this.
12433 * doc/tm.texi: Regenerate.
12434 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
12435 instead of direction.
12436 (compute_argument_addresses): Likewise.
12437 (load_register_parameters): Likewise.
12438 (emit_library_call_value_1): Likewise.
12439 (store_one_arg): Use targetm.calls.function_arg_padding instead
12440 of FUNCTION_ARG_PADDING.
12441 (must_pass_in_stack_var_size_or_pad): Likewise.
12442 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
12443 (emit_group_store): Likewise.
12444 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
12445 instead of FUNCTION_ARG_PADDING.
12446 (emit_push_insn): Likewise, and propagate enum change throughout
12448 * function.h (direction): Delete.
12449 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
12451 * function.c (assign_parm_find_stack_rtl): Likewise.
12452 (assign_parm_setup_block_p): Likewise.
12453 (assign_parm_setup_block): Likewise.
12454 (gimplify_parameters): Likewise.
12455 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
12456 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
12458 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
12459 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12460 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
12461 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
12462 (aarch64_function_arg_padding): ...this new function.
12463 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
12464 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12465 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
12466 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12467 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
12468 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12469 (arm_pad_arg_upward): Replace with...
12470 (arm_function_arg_padding): ...this new function.
12471 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
12473 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
12474 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
12475 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12476 (ia64_hpux_function_arg_padding): Replace with...
12477 (ia64_function_arg_padding): ...this new function. Use pad_direction
12478 instead of direction. Check for TARGET_HPUX.
12479 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
12480 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12481 (iq2000_function_arg_padding): New function.
12482 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
12483 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
12484 (mips_function_arg_padding): ...this new function.
12485 (mips_pad_reg_upward): Update accordingly.
12486 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12487 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
12488 targetm.calls.function_arg_padding.
12489 (FUNCTION_ARG_PADDING): Delete.
12490 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12491 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
12492 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12493 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
12494 (nios2_block_reg_padding): Return pad_direction instead of direction.
12495 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
12496 instead of direction.
12497 (nios2_function_arg_padding): Likewise. Make static.
12498 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12499 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
12500 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
12501 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
12502 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12503 (pa_function_arg_padding): Make static. Return pad_direction instead
12505 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
12506 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12507 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
12508 instead of direction. Use targetm.calls.function_arg_padding.
12509 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
12510 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
12511 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
12512 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
12513 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
12515 (function_arg_padding): Rename to...
12516 (rs6000_function_arg_padding): ...this. Make static. Return
12517 pad_direction instead of direction.
12518 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
12519 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
12520 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12521 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
12522 instead of direction. Use targetm.calls.function_arg_padding.
12523 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
12524 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
12525 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
12526 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
12527 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12528 (function_arg_padding): Rename to...
12529 (rs6000_function_arg_padding): ...this. Make static. Return
12530 pad_direction instead of direction.
12531 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
12532 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
12533 * config/s390/s390.c (s390_function_arg_padding): New function.
12534 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12535 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
12536 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
12537 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12538 (function_arg_padding): Rename to...
12539 (sparc_function_arg_padding): ...this. Make static. Return
12540 pad_direction instead of direction.
12541 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
12542 * config/spu/spu.c (spu_function_arg_padding): New function.
12543 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12544 * system.h (FUNCTION_ARG_PADDING): Poison.
12546 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12547 Alan Hayward <alan.hayward@arm.com>
12548 David Sherwood <david.sherwood@arm.com>
12550 * target.def (modes_tieable_p): New hook.
12551 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
12552 (TARGET_MODES_TIEABLE_P): ...this.
12553 * doc/tm.texi.in: Regenerate.
12554 * hooks.h (hook_bool_mode_mode_true): Declare.
12555 * hooks.c (hook_bool_mode_mode_true): New function.
12556 * combine.c (subst): Use targetm.modes_tieable_p instead of
12558 * dse.c (find_shift_sequence): Likewise.
12559 * expmed.c (extract_low_bits): Likewise.
12560 * lower-subreg.c: Include target.h.
12561 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
12563 * rtlanal.c (rtx_cost): Likewise.
12564 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
12565 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
12566 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
12567 (TARGET_MODES_TIEABLE_P): Redefine.
12568 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
12569 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
12570 (TARGET_MODES_TIEABLE_P): Redefine.
12571 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
12572 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
12573 (arc_modes_tieable_p): New function.
12574 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
12575 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
12576 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
12577 (arm_modes_tieable_p): Make static.
12578 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
12579 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
12580 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
12581 (TARGET_MODES_TIEABLE_P): Redefine.
12582 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
12583 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
12584 (TARGET_MODES_TIEABLE_P): Redefine.
12585 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
12586 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
12587 (cr16_modes_tieable_p): New function.
12588 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
12589 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
12590 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
12591 (TRULY_NOOP_TRUNCATION): Update comment.
12592 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
12593 (TRULY_NOOP_TRUNCATION): Update comment.
12594 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
12595 (frv_modes_tieable_p): New function.
12596 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
12597 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
12598 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
12599 (TARGET_MODES_TIEABLE_P): Redefine.
12600 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
12601 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
12602 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
12603 (TARGET_MODES_TIEABLE_P): Redefine.
12604 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
12605 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
12606 (ia64_modes_tieable_p): New function.
12607 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
12608 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
12609 (iq2000_modes_tieable_p): New function.
12610 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
12611 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
12612 (lm32_modes_tieable_p): New function.
12613 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
12614 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
12615 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
12616 (TARGET_MODES_TIEABLE_P): Redefine.
12617 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
12618 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
12619 (m32r_modes_tieable_p): New function.
12620 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
12621 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
12622 (m68k_modes_tieable_p): New function.
12623 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
12624 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
12625 (mcore_modes_tieable_p): New function.
12626 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
12627 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
12629 (TARGET_MODES_TIEABLE_P): Redefine.
12630 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
12631 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
12632 * config/mips/mips.c (mips_modes_tieable_p): Make static.
12633 (TARGET_MODES_TIEABLE_P): Redefine.
12634 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
12635 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
12636 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
12637 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
12638 (mn10300_modes_tieable_p): ...this and make static.
12639 (TARGET_MODES_TIEABLE_P): Redefine.
12640 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
12641 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
12642 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
12643 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
12644 (msp430_modes_tieable_p): Make static.
12645 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
12646 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
12647 (TARGET_MODES_TIEABLE_P): Redefine.
12648 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
12649 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
12650 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
12651 (TARGET_MODES_TIEABLE_P): Redefine.
12652 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
12653 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
12654 * config/pa/pa.c (pa_modes_tieable_p): Make static.
12655 (TARGET_MODES_TIEABLE_P): Redefine.
12656 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
12657 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
12658 (pdp11_modes_tieable_p): New function.
12659 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
12660 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
12661 (rs6000_modes_tieable_p): New function.
12662 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12663 * config/powerpcspe/powerpcspe.md: Update comment.
12664 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
12665 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
12666 (TARGET_MODES_TIEABLE_P): Redefine.
12667 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
12668 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
12669 (rl78_modes_tieable_p): New function.
12670 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
12671 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
12672 (rs6000_modes_tieable_p): New function.
12673 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12674 * config/rs6000/rs6000.md: Update comment.
12675 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
12676 * config/rx/rx.c (rx_modes_tieable_p): New function.
12677 (TARGET_MODES_TIEABLE_P): Redefine.
12678 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
12679 * config/s390/s390.c (s390_modes_tieable_p): New function.
12680 (TARGET_MODES_TIEABLE_P): Redefine.
12681 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
12682 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
12683 (sh_modes_tieable_p): New function.
12684 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
12685 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
12686 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
12687 (sparc_modes_tieable_p): Make static.
12688 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
12689 * config/spu/spu.c (spu_modes_tieable_p): New function.
12690 (TARGET_MODES_TIEABLE_P): Redefine.
12691 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
12692 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
12693 (TARGET_MODES_TIEABLE_P): Redefine.
12694 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
12695 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
12696 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
12697 * config/v850/v850.c (v850_modes_tieable_p): New function.
12698 (TARGET_MODES_TIEABLE_P): Redefine.
12699 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
12700 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
12701 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
12702 (visium_modes_tieable_p): New function.
12703 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
12704 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
12705 (xtensa_modes_tieable_p): New function.
12706 * system.h (MODES_TIEABLE_P): Poison.
12708 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12709 Alan Hayward <alan.hayward@arm.com>
12710 David Sherwood <david.sherwood@arm.com>
12712 * target.def (hard_regno_mode_ok): New hook.
12713 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
12714 (TARGET_HARD_REGNO_MODE_OK): ...this.
12715 * doc/tm.texi.in: Regenerate.
12716 * hooks.h (hook_bool_uint_mode_true): Declare.
12717 * hooks.c (hook_bool_uint_mode_true): New function.
12718 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
12719 HARD_REGNO_MODE_OK.
12720 * genpreds.c (write_insn_preds_c): Add an include of target.h.
12721 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
12722 instead of HARD_REGNO_MODE_OK.
12723 * caller-save.c: Include target.h.
12724 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
12725 HARD_REGNO_MODE_OK.
12726 * combine.c (can_combine_p): Likewise.
12727 (combinable_i3pat): Likewise.
12728 (can_change_dest_mode): Likewise.
12729 * expr.c (init_expr_target): Likewise.
12730 (convert_move): Likewise.
12731 (convert_modes): Likewise.
12732 * ira.c (setup_prohibited_class_mode_regs): Likewise.
12733 (setup_prohibited_mode_move_regs): Likewise.
12734 * ira.h (target_ira): Likewise.
12735 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12736 * lra-constraints.c (process_alt_operands): Likewise.
12737 (split_reg): Likewise.
12738 * recog.c (peep2_find_free_register): Likewise.
12739 * ree.c (combine_reaching_defs): Likewise.
12740 * regcprop.c (maybe_mode_change): Likewise.
12741 * reginfo.c (init_reg_sets_1): Likewise.
12742 (choose_hard_reg_mode): Likewise.
12743 (simplifiable_subregs): Likewise.
12744 * regrename.c (check_new_reg_p): Likewise.
12745 * reload.c (find_valid_class): Likewise.
12746 (find_valid_class_1): Likewise.
12747 (reload_inner_reg_of_subreg): Likewise.
12748 (push_reload): Likewise.
12749 (combine_reloads): Likewise.
12750 (find_dummy_reload): Likewise.
12751 (find_reloads): Likewise.
12752 * reload1.c (find_reg): Likewise.
12753 (set_reload_reg): Likewise.
12754 (allocate_reload_reg): Likewise.
12755 (choose_reload_regs): Likewise.
12756 (reload_adjust_reg_for_temp): Likewise.
12757 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
12758 (simplify_subreg_regno): Likewise.
12759 * sel-sched.c (init_regs_for_mode): Likewise.
12760 * varasm.c (make_decl_rtl): Likewise.
12761 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
12762 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
12763 HARD_REGNO_MODE_OK.
12764 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
12765 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
12766 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12767 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
12768 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
12769 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12770 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
12771 (arc_mode_class): Delete.
12772 (HARD_REGNO_MODE_OK): Delete.
12773 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12774 (arc_hard_regno_mode_ok): Rename old array to...
12775 (arc_hard_regno_mode_ok_modes): ...this.
12776 (arc_conditional_register_usage): Update accordingly.
12777 (arc_mode_class): Make static.
12778 (arc_hard_regno_mode_ok): New function.
12779 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
12780 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
12781 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12782 (arm_hard_regno_mode_ok): Make static.
12783 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
12784 HARD_REGNO_MODE_OK.
12785 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
12786 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
12787 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
12789 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12790 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
12791 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
12792 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
12793 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
12794 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12795 * config/bfin/predicates.md (valid_reg_operand): Use
12796 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
12797 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
12798 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
12799 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12800 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
12801 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
12802 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12803 (cr16_hard_regno_mode_ok): Make static and return a bool.
12804 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
12805 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12806 (cris_hard_regno_mode_ok): New function.
12807 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
12808 (epiphany_mode_class): Delete.
12809 (HARD_REGNO_MODE_OK): Delete.
12810 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
12811 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12812 (hard_regno_mode_ok): Rename to...
12813 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
12814 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
12815 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
12816 HARD_REGNO_MODE_OK.
12817 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
12818 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
12819 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12820 (frv_hard_regno_mode_ok): Make static and return a bool.
12821 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
12822 HARD_REGNO_MODE_OK.
12823 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
12824 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
12825 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
12826 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
12828 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12829 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
12830 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
12831 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
12833 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12834 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
12835 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12836 (ia64_hard_regno_mode_ok): New function.
12837 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
12838 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12839 (iq2000_hard_regno_mode_ok): New function.
12840 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
12841 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12842 (lm32_hard_regno_mode_ok): New function.
12843 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
12844 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
12845 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
12846 instead of HARD_REGNO_MODE_OK.
12847 (m32c_hard_regno_ok): Rename to...
12848 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
12849 (m32c_cannot_change_mode_class): Update accordingly.
12850 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12851 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
12852 (m32r_mode_class): Delete.
12853 (HARD_REGNO_MODE_OK): Delete.
12854 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12855 (m32r_hard_regno_mode_ok): Rename to...
12856 (m32r_hard_regno_modes): ...this.
12857 (m32r_mode_class): Make static.
12858 (m32r_hard_regno_mode_ok): New function.
12859 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
12860 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
12861 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12862 (m68k_hard_regno_mode_ok): Make static.
12863 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
12864 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12865 (mcore_hard_regno_mode_ok): New function.
12866 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
12867 (HARD_REGNO_MODE_OK): Delete.
12868 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
12870 (microblaze_hard_regno_mode_ok_p): ...this and make static.
12871 (microblaze_hard_regno_mode_ok): New function.
12872 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12873 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
12874 (mips_hard_regno_mode_ok): Delete.
12875 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
12876 (mips_hard_regno_mode_ok_p): ...this and make static.
12877 (mips_hard_regno_mode_ok_p): Rename to...
12878 (mips_hard_regno_mode_ok_uncached): ...this.
12879 (mips_hard_regno_mode_ok): New function.
12880 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
12881 of HARD_REGNO_MODE_OK.
12882 (mips_option_override): Update after above name changes.
12883 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12884 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
12885 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
12886 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
12887 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
12888 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12889 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
12890 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
12891 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
12892 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12893 (msp430_hard_regno_mode_ok): Make static and return a bool.
12894 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
12895 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
12896 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
12898 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12899 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
12900 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
12901 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
12902 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
12903 (PA_HARD_REGNO_MODE_OK): ...this
12904 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
12905 (PA_HARD_REGNO_MODE_OK): ...this.
12906 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12907 (pa_hard_regno_mode_ok): New function.
12908 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
12909 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12910 (pdp11_hard_regno_mode_ok): New function.
12911 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
12912 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
12914 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
12916 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12917 (rs6000_hard_regno_mode_ok): Rename to...
12918 (rs6000_hard_regno_mode_ok_uncached): ...this.
12919 (rs6000_init_hard_regno_mode_ok): Update accordingly.
12920 (rs6000_hard_regno_mode_ok): New function.
12921 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
12922 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
12923 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
12924 (riscv_hard_regno_mode_ok): ...this and make static.
12925 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12926 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
12927 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
12928 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12929 (rl78_hard_regno_mode_ok): Make static and return bool.
12930 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
12931 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
12933 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
12934 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12935 (rs6000_hard_regno_mode_ok): Rename to...
12936 (rs6000_hard_regno_mode_ok_uncached): ...this.
12937 (rs6000_init_hard_regno_mode_ok): Update accordingly.
12938 (rs6000_hard_regno_mode_ok): New function.
12939 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
12940 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
12941 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12942 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
12943 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
12944 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
12945 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12946 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
12947 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
12948 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12949 (sh_hard_regno_mode_ok): Make static.
12950 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
12951 instead of HARD_REGNO_MODE_OK.
12952 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
12953 (sparc_mode_class): Delete.
12954 (HARD_REGNO_MODE_OK): Delete.
12955 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12956 (hard_regno_mode_classes): Make static.
12957 (sparc_mode_class): Likewise.
12958 (sparc_hard_regno_mode_ok): New function.
12959 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
12960 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
12961 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
12963 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12964 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
12965 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
12966 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
12967 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
12968 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12969 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
12970 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
12971 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12972 (visium_hard_regno_mode_ok): New function.
12973 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
12974 instead of HARD_REGNO_MODE_OK.
12975 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
12976 (HARD_REGNO_MODE_OK): Delete.
12977 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
12978 (xtensa_hard_regno_mode_ok_p): ...this and make static.
12979 (xtensa_option_override): Update accordingly.
12980 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12981 (xtensa_hard_regno_mode_ok): New function.
12982 * system.h (HARD_REGNO_MODE_OK): Poison.
12984 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12985 Alan Hayward <alan.hayward@arm.com>
12986 David Sherwood <david.sherwood@arm.com>
12988 * target.def (hard_regno_call_part_clobbered): New hook.
12989 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
12990 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
12991 * doc/tm.texi: Regenerate.
12992 * hooks.h (hook_bool_uint_mode_false): Declare.
12993 * hooks.c (hook_bool_uint_mode_false): New function.
12994 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12995 * cselib.c (cselib_process_insn): Use
12996 targetm.hard_regno_call_part_clobbered instead of
12997 HARD_REGNO_CALL_PART_CLOBBERED.
12998 * ira-conflicts.c (ira_build_conflicts): Likewise.
12999 * ira-costs.c (ira_tune_allocno_costs): Likewise.
13000 * lra-constraints.c (need_for_call_save_p): Likewise.
13001 * lra-lives.c: Include target.h.
13002 (check_pseudos_live_through_calls): Use
13003 targetm.hard_regno_call_part_clobbered instead of
13004 HARD_REGNO_CALL_PART_CLOBBERED.
13005 * regcprop.c: Include target.h.
13006 (copyprop_hardreg_forward_1): Use
13007 targetm.hard_regno_call_part_clobbered instead of
13008 HARD_REGNO_CALL_PART_CLOBBERED.
13009 * reginfo.c (choose_hard_reg_mode): Likewise.
13010 * regrename.c (check_new_reg_p): Likewise.
13011 * reload.c (find_equiv_reg): Likewise.
13012 * reload1.c (emit_reload_insns): Likewise.
13013 * sched-deps.c (deps_analyze_insn): Likewise.
13014 * sel-sched.c (init_regs_for_mode): Likewise.
13015 (mark_unavailable_hard_regs): Likewise.
13016 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
13017 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13018 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
13020 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13021 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13022 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
13024 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
13026 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13027 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13028 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
13030 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13031 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13032 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
13034 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13035 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
13037 * config/powerpcspe/powerpcspe.c
13038 (rs6000_hard_regno_call_part_clobbered): New function.
13039 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13040 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13041 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
13043 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13044 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13045 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
13047 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13048 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13049 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
13051 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13052 Alan Hayward <alan.hayward@arm.com>
13053 David Sherwood <david.sherwood@arm.com>
13055 * rtl.h (subreg_memory_offset): Declare.
13056 * emit-rtl.c (subreg_memory_offset): New function.
13057 * expmed.c (store_bit_field_1): Use it.
13058 * expr.c (undefined_operand_subword_p): Likewise.
13059 * simplify-rtx.c (simplify_subreg): Likewise.
13061 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
13063 PR rtl-optimization/57448
13066 * optabs.c (expand_atomic_load): Place compiler memory barriers if
13067 using atomic_load pattern.
13068 (expand_atomic_store): Likewise.
13070 2017-09-04 Jakub Jelinek <jakub@redhat.com>
13073 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
13074 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
13075 handling. Use replace_call_with_value with NULL instead of
13076 gsi_replace, unlink_stmt_vdef and release_defs.
13078 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
13081 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
13083 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13086 * rtl.h (emit_library_call_value_1): Declare.
13087 (emit_library_call): Replace declaration with a series of overloads.
13088 Remove the parameter count argument.
13089 (emit_library_call_value): Likewise.
13090 * calls.c (emit_library_call_value_1): Make global. Replace varargs
13091 with an "rtx_mode_t *".
13092 (emit_library_call_value): Delete.
13093 (emit_library_call): Likewise.
13094 * asan.c (asan_emit_stack_protection): Update calls accordingly.
13095 (asan_emit_allocas_unpoison): Likewise.
13096 * builtins.c (expand_builtin_powi): Likewise.
13097 (expand_asan_emit_allocas_unpoison): Likewise.
13098 * cfgexpand.c (expand_main_function): Likewise.
13099 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
13100 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
13101 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
13102 * config/arm/arm.c (arm_trampoline_init): Likewise.
13103 (arm_call_tls_get_addr): Likewise.
13104 (arm_expand_divmod_libfunc): Likewise.
13105 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
13106 (smulsi3_highpart): Likewise.
13107 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
13108 (c6x_expand_compare): Likewise.
13109 (c6x_expand_movmem): Likewise.
13110 * config/frv/frv.c (frv_trampoline_init): Likewise.
13111 * config/i386/i386.c (ix86_trampoline_init): Likewise.
13112 (ix86_expand_divmod_libfunc): Likewise.
13113 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
13114 (ia64_expand_compare): Likewise.
13115 (ia64_profile_hook): Likewise.
13116 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
13117 (nonlocal_goto): Likewise.
13118 (restore_stack_nonlocal): Likewise.
13119 * config/m32r/m32r.c (block_move_call): Likewise.
13120 (m32r_trampoline_init): Likewise.
13121 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
13122 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
13123 (m68k_call_m68k_read_tp): Likewise.
13124 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
13125 (microblaze_expand_divide): Likewise.
13126 * config/mips/mips.h (mips_args): Likewise.
13127 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
13128 (MIPS_ICACHE_SYNC): Likewise.
13129 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
13130 (nios2_trampoline_init): Likewise.
13131 * config/pa/pa.c (hppa_tls_call): Likewise.
13132 (pa_trampoline_init): Likewise.
13133 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
13134 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
13135 (expand_strn_compare): Likewise.
13136 (rs6000_generate_compare): Likewise.
13137 (rs6000_expand_float128_convert): Likewise.
13138 (output_profile_hook): Likewise.
13139 (rs6000_trampoline_init): Likewise.
13140 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
13141 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
13142 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
13143 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
13144 (rs6000_generate_compare): Likewise.
13145 (rs6000_expand_float128_convert): Likewise.
13146 (output_profile_hook): Likewise.
13147 (rs6000_trampoline_init): Likewise.
13148 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
13149 * config/sh/sh.c (sh_trampoline_init): Likewise.
13150 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
13151 (sparc_emit_float_lib_cmp): Likewise.
13152 (sparc32_initialize_trampoline): Likewise.
13153 (sparc64_initialize_trampoline): Likewise.
13154 (sparc_profile_hook): Likewise.
13155 * config/spu/spu.c (ea_load_store): Likewise.
13156 * config/spu/spu.md (floatunssidf2): Likewise.
13157 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
13158 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
13159 * config/visium/visium.c (expand_block_move_4): Likewise.
13160 (expand_block_move_2): Likewise.
13161 (expand_block_move_1): Likewise.
13162 (expand_block_set_4): Likewise.
13163 (expand_block_set_2): Likewise.
13164 (expand_block_set_1): Likewise.
13165 (visium_trampoline_init): Likewise.
13166 (visium_profile_hook): Likewise.
13167 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
13168 (xtensa_setup_frame_addresses): Likewise.
13169 (xtensa_trampoline_init): Likewise.
13170 * except.c (sjlj_emit_function_enter): Likewise.
13171 (sjlj_emit_function_exit): Likewise.
13172 * explow.c (allocate_dynamic_stack_space): Likewise.
13173 (probe_stack_range): Likewise.
13174 * expr.c (convert_mode_scalar): Likewise.
13175 * optabs.c (expand_binop): Likewise.
13176 (expand_twoval_binop_libfunc): Likewise.
13177 (expand_unop): Likewise.
13178 (prepare_cmp_insn): Likewise.
13179 (prepare_float_lib_cmp): Likewise.
13180 (expand_float): Likewise.
13181 (expand_fix): Likewise.
13182 (expand_fixed_convert): Likewise.
13183 (maybe_emit_sync_lock_test_and_set): Likewise.
13184 (expand_atomic_compare_and_swap): Likewise.
13185 (expand_mem_thread_fence): Likewise.
13186 (expand_atomic_fetch_op): Likewise.
13188 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
13190 * doc/generic.texi (OpenACC): Adjust URL.
13191 * doc/invoke.texi (C Dialect Options): Ditto.
13193 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
13195 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
13196 predicate for operand 1. Add (m,<S>) constraint.
13197 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
13198 Prevent memory operand 1 with register operand 2.
13200 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
13202 PR rtl-optimization/82024
13203 * combine.c (try_combine): If the combination result is a PARALLEL,
13204 and we only need to retain the SET in there that would be placed
13205 at I2, check that we can place that at I3 instead, before doing so.
13207 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13210 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
13212 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
13213 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
13216 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
13217 Jeff Law <law@redhat.com>
13219 * varasm.c (bss_initializer_p): Do not put constants into .bss
13220 (categorize_decl_for_section): Handle bss_initializer_p returning
13221 false when DECL_INITIAL is NULL.
13223 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13226 * config/s390/s390.c (s390_can_inline_p): New function.
13228 2017-09-01 Jeff Law <law@redhat.com>
13230 PR tree-optimization/82052
13231 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
13232 Always initialize the returned slot after a hash table miss
13233 when INSERT is true.
13235 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
13237 * config/s390/s390.md (mem_signal_fence): Remove.
13238 * doc/md.texi (mem_signal_fence): Remove.
13239 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
13241 * target-insns.def (mem_signal_fence): Remove.
13243 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13246 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
13249 * asan.c (create_odr_indicator): Strip name encoding from assembler
13250 name before appending it after __odr_asan_.
13252 2017-09-01 Martin Liska <mliska@suse.cz>
13254 PR tree-optimization/82059
13255 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
13256 frequency only when an edge is redirected.
13258 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13260 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
13261 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
13262 (arc_conditional_register_usage): Remove ARC600 lp_count
13264 (arc_file_start): Emit Tag_ARC_CPU_variation.
13265 (arc_can_use_doloop_p): New conditions to use ZOLs.
13266 (hwloop_fail): New function.
13267 (hwloop_optimize): Likewise.
13268 (hwloop_pattern_reg): Likewise.
13269 (arc_doloop_hooks): New struct, to be used with reorg_loops.
13270 (arc_reorg_loops): New function, calls reorg_loops.
13271 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
13272 (arc600_corereg_hazard): Remove ZOL checking, case handled by
13274 (arc_loop_hazard): Remove function, functionality moved into
13276 (arc_hazard): Remove arc_loop_hazard call.
13277 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
13278 into hwloop_optimize.
13279 (arc_label_align): Remove ZOL handling.
13280 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
13281 * config/arc/arc.md (doloop_begin): Remove pattern.
13282 (doloop_begin_i): Likewise.
13283 (doloop_end_i): Likewise.
13284 (doloop_fallback): Likewise.
13285 (doloop_fallback_m): Likewise.
13286 (doloop_end): Reimplement expand.
13287 (arc_lp): New pattern for LP instruction.
13288 (loop_end): New pattern.
13289 (loop_fail): Likewise.
13290 (decrement_and_branch_until_zero): Likewise.
13291 * config/arc/arc.opt (mlpc-width): New option.
13292 * doc/invoke.texi (mlpc-width): Document option.
13294 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13296 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
13297 (arc_ccfsm_advance): Fix checking for delay slots.
13298 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
13300 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13302 * config/arc/arc.md (movqi_insn): Add stores to save constant long
13304 (movhi_insn): Update store instruction constraint which are saving
13305 6-bit short immediates.
13306 (movsi_insn): Consider also short scaled load operations.
13307 (zero_extendhisi2_i): Use Usd constraint instead of T.
13308 (extendhisi2_i): Add q constraint.
13309 (arc_clzsi2): Add type and length attributes.
13310 (arc_ctzsi2): Likewise.
13311 * config/arc/constraints.md (Usc): Update constraint, the
13312 assembler can parse two relocations for a single instruction.
13314 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13316 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
13317 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
13319 2017-08-31 Olivier Hainque <hainque@adacore.com>
13321 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
13322 match as powerpc-wrs-vxworks*.
13324 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
13326 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
13327 register constraint for by-element operand.
13328 (aarch64_mls_elt_merge<mode>): Likewise.
13330 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13332 * config/arc/arc.c (arc_can_follow_jump): Check for short
13335 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13337 * config.gcc: Use g.opt for arc.
13338 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
13339 functionality moved to ...
13340 (legitimate_scaled_address_p): New function, ...here.
13341 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
13342 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
13343 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
13345 (arc_override_options): Handle G option.
13346 (arc_output_pic_addr_const): Correct function definition.
13347 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
13348 (arc_decl_anon_ns_mem_p): Delete.
13349 (arc_in_small_data_p): Overhaul this function to take into
13350 consideration the value given via G option.
13351 (arc_rewrite_small_data_1): Renamed and corrected old
13352 arc_rewrite_small_data function.
13353 (arc_rewrite_small_data): New function.
13354 (small_data_pattern): Don't use pic_offset_table_rtx.
13355 * config/arc/arc.h (CC1_SPEC): Recognize G option.
13356 * config/arc/simdext.md (movmisalignv2hi): Use
13357 prepare_move_operands function.
13359 (movmisalign*): Likewise.
13360 * doc/invoke.texi (ARC options): Document -G option.
13362 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13364 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
13366 * config/arc/arc.c (arc_print_operand): Output scalled address for
13367 sdata whenever is possible.
13368 (arc_in_small_data_p): Allow sdata for 64bit datum when double
13369 load/stores are available.
13370 (compact_sda_memory_operand): Check for the alignment required by
13371 code density instructions.
13372 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
13374 * config/arc/constraints.md (Usd): Update constraint.
13375 (Us0): New constraint.
13376 (Usc): Update constraint.
13378 2017-08-31 Richard Biener <rguenther@suse.de>
13380 PR middle-end/82054
13381 * dwarf2out.c (dwarf2out_early_global_decl): Process each
13382 function only once.
13384 2017-08-31 Tamar Christina <tamar.christina@arm.com>
13386 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
13387 Resize type_signature.
13389 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13390 Alan Hayward <alan.hayward@arm.com>
13391 David Sherwood <david.sherwood@arm.com>
13393 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
13394 subregs whose inner modes can be stored in GPRs.
13395 (aarch64_classify_index): Likewise.
13397 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13398 Alan Hayward <alan.hayward@arm.com>
13399 David Sherwood <david.sherwood@arm.com>
13401 * config/aarch64/iterators.md (V_cmp_result): Rename to...
13402 (V_INT_EQUIV): ...this.
13403 (v_cmp_result): Rename to...
13404 (v_int_equiv): ...this.
13405 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
13406 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
13407 (copysign<mode>3): Likewise.
13408 (aarch64_simd_bsl<mode>_internal): Likewise.
13409 (aarch64_simd_bsl<mode>): Likewise.
13410 (vec_cmp<mode><mode>): Likewise.
13411 (vcond<mode><mode>): Likewise.
13412 (vcond<v_cmp_mixed><mode>): Likewise.
13413 (vcondu<mode><v_cmp_mixed>): Likewise.
13414 (aarch64_cm<optab><mode>): Likewise.
13415 (aarch64_cmtst<mode>): Likewise.
13416 (aarch64_fac<optab><mode>): Likewise.
13417 (vec_perm_const<mode>): Likewise.
13418 (vcond_mask_<mode><v_cmp_result>): Rename to...
13419 (vcond_mask_<mode><v_int_equiv>): ...this.
13420 (vec_cmp<mode><v_cmp_result>): Rename to...
13421 (vec_cmp<mode><v_int_equiv>): ...this.
13423 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13424 Alan Hayward <alan.hayward@arm.com>
13425 David Sherwood <david.sherwood@arm.com>
13427 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
13429 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
13430 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
13431 (UNSPEC_LD4_DREG): New unspecs.
13432 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
13433 (aarch64_ld2<mode>_dreg_be): Replace with...
13434 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
13436 (aarch64_ld3<mode>_dreg_le)
13437 (aarch64_ld3<mode>_dreg_be): Replace with...
13438 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
13440 (aarch64_ld4<mode>_dreg_le)
13441 (aarch64_ld4<mode>_dreg_be): Replace with...
13442 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
13445 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13447 PR tree-optimization/81987
13448 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
13449 insert an initializer in a location not dominated by the stride
13452 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13454 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
13455 on the entire header of the finally block in the fallthru case.
13457 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13459 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
13461 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
13463 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
13464 rs6000_emit_move_from_cr and call renamed function.
13465 (rs6000_emit_prologue): Call renamed functions.
13466 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
13467 movesi_from_cr, remove volatile CRs.
13469 2017-08-30 Jon Beniston <jon@beniston.com>
13470 Richard Biener <rguenther@suse.de>
13472 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
13473 of VECTOR_MODE_P check.
13474 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
13475 element vector types.
13477 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13479 * df.h (df_read_modify_subreg_p): Remove in favor of...
13480 * rtl.h (read_modify_subreg_p): ...this new function. Take a
13481 const_rtx instead of an rtx.
13482 * cprop.c (local_cprop_find_used_regs): Update accordingly.
13483 * df-problems.c (df_word_lr_mark_ref): Likewise.
13484 * ira-lives.c (mark_pseudo_reg_live): Likewise.
13485 (mark_pseudo_reg_dead): Likewise.
13486 (mark_ref_dead): Likewise.
13487 * reginfo.c (init_subregs_of_mode): Likewise.
13488 * sched-deps.c (sched_analyze_1): Likewise.
13489 * df-scan.c (df_def_record_1): Likewise.
13490 (df_uses_record): Likewise.
13491 (df_read_modify_subreg_p): Remove in favor of...
13492 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
13493 const_rtx instead of an rtx.
13495 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13496 Alan Hayward <alan.hayward@arm.com>
13497 David Sherwood <david.sherwood@arm.com>
13499 * rtl.h (partial_subreg_p): New function.
13500 * caller-save.c (save_call_clobbered_regs): Use it.
13501 * calls.c (expand_call): Likewise.
13502 * combine.c (combinable_i3pat): Likewise.
13503 (simplify_set): Likewise.
13504 (make_extraction): Likewise.
13505 (make_compound_operation_int): Likewise.
13506 (gen_lowpart_or_truncate): Likewise.
13507 (force_to_mode): Likewise.
13508 (make_field_assignment): Likewise.
13509 (reg_truncated_to_mode): Likewise.
13510 (record_truncated_value): Likewise.
13511 (move_deaths): Likewise.
13512 * cse.c (record_jump_cond): Likewise.
13513 (cse_insn): Likewise.
13514 * cselib.c (cselib_lookup_1): Likewise.
13515 * expmed.c (extract_bit_field_using_extv): Likewise.
13516 * function.c (assign_parm_setup_reg): Likewise.
13517 * ifcvt.c (noce_convert_multiple_sets): Likewise.
13518 * ira-build.c (create_insn_allocnos): Likewise.
13519 * lra-coalesce.c (merge_pseudos): Likewise.
13520 * lra-constraints.c (match_reload): Likewise.
13521 (simplify_operand_subreg): Likewise.
13522 (curr_insn_transform): Likewise.
13523 * lra-lives.c (process_bb_lives): Likewise.
13524 * lra.c (new_insn_reg): Likewise.
13525 (lra_substitute_pseudo): Likewise.
13526 * regcprop.c (mode_change_ok): Likewise.
13527 (maybe_mode_change): Likewise.
13528 (copyprop_hardreg_forward_1): Likewise.
13529 * reload.c (push_reload): Likewise.
13530 (find_reloads): Likewise.
13531 (find_reloads_subreg_address): Likewise.
13532 * reload1.c (alter_reg): Likewise.
13533 (eliminate_regs_1): Likewise.
13534 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13536 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
13538 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
13539 back to if statements, including unpack.
13541 2017-08-30 Martin Liska <mliska@suse.cz>
13543 PR inline-asm/82001
13544 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
13546 (func_checker::compare_asm_inputs_outputs): ... this function.
13547 (func_checker::compare_gimple_asm): Use the function to compare
13548 also ASM constrains.
13549 * ipa-icf-gimple.h: Rename the function.
13551 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13552 Alan Hayward <alan.hayward@arm.com>
13553 David Sherwood <david.sherwood@arm.com>
13555 * coretypes.h (complex_mode): New type.
13556 * gdbhooks.py (build_pretty_printer): Handle it.
13557 * machmode.h (complex_mode): New class.
13558 (complex_mode::includes_p): New function.
13559 (is_complex_int_mode): Likewise.
13560 (is_complex_float_mode): Likewise.
13561 * genmodes.c (get_mode_class): Handle complex mode classes.
13562 * function.c (expand_function_end): Use is_complex_int_mode.
13564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13565 Alan Hayward <alan.hayward@arm.com>
13566 David Sherwood <david.sherwood@arm.com>
13568 * coretypes.h (scalar_mode_pod): New typedef.
13569 * gdbhooks.py (build_pretty_printer): Handle it.
13570 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
13571 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
13572 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
13573 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
13574 as_a <scalar_mode>.
13576 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13577 Alan Hayward <alan.hayward@arm.com>
13578 David Sherwood <david.sherwood@arm.com>
13580 * machmode.h (mode_for_vector): Take a scalar_mode instead
13582 * stor-layout.c (mode_for_vector): Likewise.
13583 * explow.c (promote_mode): Use as_a <scalar_mode>.
13584 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
13586 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13587 Alan Hayward <alan.hayward@arm.com>
13588 David Sherwood <david.sherwood@arm.com>
13590 * target.def (preferred_simd_mode): Take a scalar_mode
13591 instead of a machine_mode.
13592 * targhooks.h (default_preferred_simd_mode): Likewise.
13593 * targhooks.c (default_preferred_simd_mode): Likewise.
13594 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
13595 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
13596 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
13597 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
13598 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
13599 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
13600 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
13601 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
13603 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
13604 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
13605 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
13606 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
13607 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
13608 * doc/tm.texi: Regenerate.
13609 * optabs-query.c (can_vec_mask_load_store_p): Return false for
13612 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13613 Alan Hayward <alan.hayward@arm.com>
13614 David Sherwood <david.sherwood@arm.com>
13616 * target.def (scalar_mode_supported_p): Take a scalar_mode
13617 instead of a machine_mode.
13618 * targhooks.h (default_scalar_mode_supported_p): Likewise.
13619 * targhooks.c (default_scalar_mode_supported_p): Likewise.
13620 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
13621 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13622 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
13623 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
13624 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
13625 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
13626 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
13627 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
13628 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
13629 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
13630 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
13631 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
13633 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
13634 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
13635 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13636 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13637 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
13639 * doc/tm.texi: Regenerate.
13641 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13642 Alan Hayward <alan.hayward@arm.com>
13643 David Sherwood <david.sherwood@arm.com>
13645 * coretypes.h (opt_scalar_mode): New typedef.
13646 * gdbhooks.py (build_pretty_printers): Handle it.
13647 * machmode.h (mode_iterator::get_2xwider): Add overload for
13649 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
13651 * expr.c (convert_mode_scalar): Likewise.
13652 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13653 * optabs.c (expand_float): Likewise.
13654 (expand_fix): Likewise.
13655 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13658 Alan Hayward <alan.hayward@arm.com>
13659 David Sherwood <david.sherwood@arm.com>
13661 * optabs.c (expand_float): Explicitly check for scalars before
13662 using a branching expansion.
13663 (expand_fix): Likewise.
13665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13666 Alan Hayward <alan.hayward@arm.com>
13667 David Sherwood <david.sherwood@arm.com>
13669 * expr.c (convert_mode): Split scalar handling out into...
13670 (convert_mode_scalar): ...this new function. Treat the modes
13673 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13674 Alan Hayward <alan.hayward@arm.com>
13675 David Sherwood <david.sherwood@arm.com>
13677 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
13679 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
13681 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13682 Alan Hayward <alan.hayward@arm.com>
13683 David Sherwood <david.sherwood@arm.com>
13685 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
13686 rather than a machine_mode.
13687 (fixed_from_string): Likewise.
13688 (fixed_convert): Likewise.
13689 (fixed_convert_from_int): Likewise.
13690 (fixed_convert_from_real): Likewise.
13691 (real_convert_from_fixed): Likewise.
13692 * fixed-value.c (fixed_from_double_int): Likewise.
13693 (fixed_from_string): Likewise.
13694 (fixed_convert): Likewise.
13695 (fixed_convert_from_int): Likewise.
13696 (fixed_convert_from_real): Likewise.
13697 (real_convert_from_fixed): Likewise.
13698 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
13700 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13701 Alan Hayward <alan.hayward@arm.com>
13702 David Sherwood <david.sherwood@arm.com>
13704 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
13705 of separate mode class checks. Do not allow vector modes here.
13706 (immed_wide_int_const): Use as_a <scalar_mode>.
13707 * explow.c (trunc_int_for_mode): Likewise.
13708 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
13709 (wi::shwi): Likewise.
13710 (wi::min_value): Likewise.
13711 (wi::max_value): Likewise.
13712 * dwarf2out.c (loc_descriptor): Likewise.
13713 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
13714 for CONST_WIDE_INT.
13716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13717 Alan Hayward <alan.hayward@arm.com>
13718 David Sherwood <david.sherwood@arm.com>
13720 * tree.h (SCALAR_TYPE_MODE): New macro.
13721 * expr.c (expand_expr_addr_expr_1): Use it.
13722 (expand_expr_real_2): Likewise.
13723 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
13724 (fold_convert_const_fixed_from_int): Likewise.
13725 (fold_convert_const_fixed_from_real): Likewise.
13726 (native_encode_fixed): Likewise
13727 (native_encode_complex): Likewise
13728 (native_encode_vector): Likewise.
13729 (native_interpret_fixed): Likewise.
13730 (native_interpret_real): Likewise.
13731 (native_interpret_complex): Likewise.
13732 (native_interpret_vector): Likewise.
13733 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
13734 (simd_clone_adjust_argument_types): Likewise.
13735 (simd_clone_init_simd_arrays): Likewise.
13736 (simd_clone_adjust): Likewise.
13737 * stor-layout.c (layout_type): Likewise.
13738 * tree.c (build_minus_one_cst): Likewise.
13739 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13740 * tree-inline.c (estimate_move_cost): Likewise.
13741 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
13742 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
13743 (vectorizable_reduction): Likewise.
13744 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
13745 (vect_recog_mixed_size_cond_pattern): Likewise.
13746 (check_bool_pattern): Likewise.
13747 (adjust_bool_pattern): Likewise.
13748 (search_type_for_mask_1): Likewise.
13749 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
13750 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13751 (vectorizable_load): Likewise.
13752 (vectorizable_store): Likewise.
13753 * ubsan.c (ubsan_encode_value): Likewise.
13754 * varasm.c (output_constant): Likewise.
13756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13757 Alan Hayward <alan.hayward@arm.com>
13758 David Sherwood <david.sherwood@arm.com>
13760 * coretypes.h (scalar_mode): New class.
13761 * machmode.h (scalar_mode): Likewise.
13762 (scalar_mode::includes_p): New function.
13763 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
13764 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
13765 * genmodes.c (get_mode_class): Handle remaining scalar modes.
13766 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
13767 * expmed.c (store_bit_field_1): Likewise.
13768 (extract_bit_field_1): Likewise.
13769 * expr.c (write_complex_part): Likewise.
13770 (read_complex_part): Likewise.
13771 (emit_move_complex_push): Likewise.
13772 (expand_expr_real_2): Likewise.
13773 * function.c (assign_parm_setup_reg): Likewise.
13774 (assign_parms_unsplit_complex): Likewise.
13775 * optabs.c (expand_binop): Likewise.
13776 * rtlanal.c (subreg_get_info): Likewise.
13777 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13778 * varasm.c (output_constant_pool_2): Likewise.
13780 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13781 Alan Hayward <alan.hayward@arm.com>
13782 David Sherwood <david.sherwood@arm.com>
13784 * expmed.c (extract_high_half): Use scalar_int_mode and remove
13786 (expmed_mult_highpart_optab): Likewise.
13787 (expmed_mult_highpart): Likewise.
13789 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13790 Alan Hayward <alan.hayward@arm.com>
13791 David Sherwood <david.sherwood@arm.com>
13793 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
13794 instead of a machine_mode.
13795 (builtin_memset_read_str): Likewise.
13796 * builtins.c (c_readstr): Likewise.
13797 (builtin_memcpy_read_str): Likewise.
13798 (builtin_strncpy_read_str): Likewise.
13799 (builtin_memset_read_str): Likewise.
13800 (builtin_memset_gen_str): Likewise.
13801 (expand_builtin_signbit): Use scalar_int_mode for local variables.
13802 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
13803 instead of a machine_mode.
13804 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
13806 (make_extraction): Likewise.
13807 (try_widen_shift_mode): Take and return scalar_int_modes instead
13809 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
13810 a scalar_int_mode instead of a machine_mode.
13811 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
13812 (avr_addr_space_pointer_mode): Likewise.
13813 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
13814 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
13815 (msp430_unwind_word_mode): Likewise.
13816 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
13817 (spu_addr_space_pointer_mode): Likewise.
13818 (spu_addr_space_address_mode): Likewise.
13819 (spu_libgcc_cmp_return_mode): Likewise.
13820 (spu_libgcc_shift_count_mode): Likewise.
13821 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
13822 (rl78_addr_space_pointer_mode): Likewise.
13823 (fl78_unwind_word_mode): Likewise.
13824 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
13826 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
13827 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
13828 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
13829 (mips_valid_pointer_mode): Likewise.
13830 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
13831 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
13832 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
13834 (ft32_addr_space_address_mode): Likewise.
13835 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
13836 scalar_int_mode instead of a machine_mode.
13837 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
13839 (m32c_addr_space_address_mode): Likewise.
13840 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
13841 (rs6000_eh_return_filter_mode): Likewise.
13842 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
13843 (rs6000_eh_return_filter_mode): Likewise.
13844 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
13845 (s390_libgcc_shift_count_mode): Likewise.
13846 (s390_unwind_word_mode): Likewise.
13847 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
13849 * target.def (mode_rep_extended): Likewise.
13850 (valid_pointer_mode): Likewise.
13851 (addr_space.valid_pointer_mode): Likewise.
13852 (eh_return_filter_mode): Return a scalar_int_mode rather than
13854 (libgcc_cmp_return_mode): Likewise.
13855 (libgcc_shift_count_mode): Likewise.
13856 (unwind_word_mode): Likewise.
13857 (addr_space.pointer_mode): Likewise.
13858 (addr_space.address_mode): Likewise.
13859 * doc/tm.texi: Regenerate.
13860 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
13862 (do_jump): Use scalar_int_mode for local variables.
13863 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
13864 rather than a machine_mode.
13865 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
13866 (scompare_loc_descriptor_wide): Likewise.
13867 (scompare_loc_descriptor_narrow): Likewise.
13868 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
13870 * except.c (sjlj_emit_dispatch_table): Likewise.
13871 (expand_builtin_eh_copy_values): Likewise.
13872 * explow.c (convert_memory_address_addr_space_1): Likewise.
13873 Take a scalar_int_mode rather than a machine_mode.
13874 (convert_memory_address_addr_space): Take a scalar_int_mode rather
13875 than a machine_mode.
13876 (memory_address_addr_space): Use scalar_int_mode for local variables.
13877 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
13878 rather than a machine_mode.
13879 * expmed.c (mask_rtx): Likewise.
13880 (init_expmed_one_conv): Likewise.
13881 (expand_mult_highpart_adjust): Likewise.
13882 (extract_high_half): Likewise.
13883 (expmed_mult_highpart_optab): Likewise.
13884 (expmed_mult_highpart): Likewise.
13885 (expand_smod_pow2): Likewise.
13886 (expand_sdiv_pow2): Likewise.
13887 (emit_store_flag_int): Likewise.
13888 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
13890 (extract_low_bits): Likewise.
13891 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
13893 * expr.c (pieces_addr::adjust): Likewise.
13894 (can_store_by_pieces): Likewise.
13895 (store_by_pieces): Likewise.
13896 (clear_by_pieces_1): Likewise.
13897 (expand_expr_addr_expr_1): Likewise.
13898 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
13899 (expand_expr_real_1): Likewise.
13900 (try_casesi): Likewise.
13901 * final.c (shorten_branches): Likewise.
13902 * fold-const.c (fold_convert_const_int_from_fixed): Change the
13903 type of "mode" to machine_mode.
13904 * internal-fn.c (expand_arith_overflow_result_store): Take a
13905 scalar_int_mode rather than a machine_mode.
13906 (expand_mul_overflow): Use scalar_int_mode for local variables.
13907 * loop-doloop.c (doloop_modify): Likewise.
13908 (doloop_optimize): Likewise.
13909 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
13910 than a machine_mode.
13911 (expand_doubleword_shift_condmove): Likewise.
13912 (expand_doubleword_shift): Likewise.
13913 (expand_doubleword_clz): Likewise.
13914 (expand_doubleword_popcount): Likewise.
13915 (expand_doubleword_parity): Likewise.
13916 (expand_absneg_bit): Use scalar_int_mode for local variables.
13917 (prepare_float_lib_cmp): Likewise.
13918 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
13919 rather than a machine_mode.
13920 (convert_memory_address_addr_space): Likewise.
13921 (get_mode_bounds): Likewise.
13922 (get_address_mode): Return a scalar_int_mode rather than a
13924 * rtlanal.c (get_address_mode): Likewise.
13925 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
13926 than a machine_mode.
13927 * targhooks.c (default_mode_rep_extended): Likewise.
13928 (default_valid_pointer_mode): Likewise.
13929 (default_addr_space_valid_pointer_mode): Likewise.
13930 (default_eh_return_filter_mode): Return a scalar_int_mode rather
13931 than a machine_mode.
13932 (default_libgcc_cmp_return_mode): Likewise.
13933 (default_libgcc_shift_count_mode): Likewise.
13934 (default_unwind_word_mode): Likewise.
13935 (default_addr_space_pointer_mode): Likewise.
13936 (default_addr_space_address_mode): Likewise.
13937 * targhooks.h (default_eh_return_filter_mode): Likewise.
13938 (default_libgcc_cmp_return_mode): Likewise.
13939 (default_libgcc_shift_count_mode): Likewise.
13940 (default_unwind_word_mode): Likewise.
13941 (default_addr_space_pointer_mode): Likewise.
13942 (default_addr_space_address_mode): Likewise.
13943 (default_mode_rep_extended): Take a scalar_int_mode rather than
13945 (default_valid_pointer_mode): Likewise.
13946 (default_addr_space_valid_pointer_mode): Likewise.
13947 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
13949 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
13950 rather than a machine_mode.
13951 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
13952 for local variables.
13953 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13954 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
13955 than a machine_mode.
13957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13958 Alan Hayward <alan.hayward@arm.com>
13959 David Sherwood <david.sherwood@arm.com>
13961 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
13962 the mode argument to scalar_int_mode.
13963 (do_jump_by_parts_zero_rtx): Likewise.
13964 (do_jump_by_parts_equality_rtx): Likewise.
13965 (do_jump_by_parts_greater): Take a mode argument.
13966 (do_jump_by_parts_equality): Likewise.
13967 (do_jump_1): Update calls accordingly.
13969 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13970 Alan Hayward <alan.hayward@arm.com>
13971 David Sherwood <david.sherwood@arm.com>
13973 * is-a.h (safe_dyn_cast): New function.
13974 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
13975 (jump_table_for_label): Likewise.
13976 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
13977 instead of an rtx_insn *.
13978 (shorten_branches): Use dyn_cast instead of LABEL_P and
13979 JUMP_TABLE_DATA_P. Use jump_table_for_label and
13980 rtx_jump_table_data::get_data_mode.
13981 (final_scan_insn): Likewise.
13983 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13984 Alan Hayward <alan.hayward@arm.com>
13985 David Sherwood <david.sherwood@arm.com>
13987 * combine.c (try_combine): Use is_a <scalar_int_mode> when
13988 trying to combine a full-register integer set with a subreg
13991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13992 Alan Hayward <alan.hayward@arm.com>
13993 David Sherwood <david.sherwood@arm.com>
13995 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
13996 that is always either address_mode or pointer_mode.
13998 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13999 Alan Hayward <alan.hayward@arm.com>
14000 David Sherwood <david.sherwood@arm.com>
14002 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
14003 when the two are known to be equal.
14005 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14006 Alan Hayward <alan.hayward@arm.com>
14007 David Sherwood <david.sherwood@arm.com>
14009 * simplify-rtx.c (simplify_const_unary_operation): Use
14010 is_a <scalar_int_mode> instead of checking for a nonzero
14011 precision. Forcibly convert op_mode to a scalar_int_mode
14012 in that case. More clearly differentiate the operand and
14013 result modes and use the former when deciding what the value
14014 of a count-bits operation should be. Use is_int_mode instead
14015 of checking for a MODE_INT. Remove redundant check for whether
14016 this mode has a zero precision.
14018 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14019 Alan Hayward <alan.hayward@arm.com>
14020 David Sherwood <david.sherwood@arm.com>
14022 * optabs.c (widen_leading): Change the type of the mode argument
14023 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
14024 (widen_bswap): Likewise.
14025 (expand_parity): Likewise.
14026 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
14027 (expand_ffs): Likewise.
14028 (epand_unop): Check for scalar integer modes before calling the
14031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14032 Alan Hayward <alan.hayward@arm.com>
14033 David Sherwood <david.sherwood@arm.com>
14035 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
14037 (expand_expr_real_1): Update call accordingly.
14039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14040 Alan Hayward <alan.hayward@arm.com>
14041 David Sherwood <david.sherwood@arm.com>
14043 * expmed.c (store_bit_field_using_insv): Add op0_mode and
14044 value_mode arguments. Use scalar_int_mode internally.
14045 (store_bit_field_1): Rename the new integer mode from imode
14046 to op0_mode and use it instead of GET_MODE (op0). Update calls
14047 to store_split_bit_field, store_bit_field_using_insv and
14048 store_fixed_bit_field.
14049 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
14050 Use scalar_int_mode internally. Use a bit count rather than a mode
14051 when calculating the largest bit size for get_best_mode.
14052 Update calls to store_split_bit_field and store_fixed_bit_field_1.
14053 (store_fixed_bit_field_1): Add mode and value_mode arguments.
14054 Remove assertion that OP0 has a scalar integer mode.
14055 (store_split_bit_field): Add op0_mode and value_mode arguments.
14056 Update calls to extract_fixed_bit_field.
14057 (extract_bit_field_using_extv): Add an op0_mode argument.
14058 Use scalar_int_mode internally.
14059 (extract_bit_field_1): Rename the new integer mode from imode to
14060 op0_mode and use it instead of GET_MODE (op0). Update calls to
14061 extract_split_bit_field, extract_bit_field_using_extv and
14062 extract_fixed_bit_field.
14063 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
14064 to extract_split_bit_field and extract_fixed_bit_field_1.
14065 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
14066 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
14067 on the target mode.
14068 (extract_split_bit_field): Add an op0_mode argument. Update call
14069 to extract_fixed_bit_field.
14071 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14072 Alan Hayward <alan.hayward@arm.com>
14073 David Sherwood <david.sherwood@arm.com>
14075 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
14076 * explow.c (hard_function_value): Likewise.
14077 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
14078 convert_to_mode call outside the loop.
14079 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
14080 for the mode iterator. Require the mode specified by max_pieces
14082 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
14084 (copy_blkmode_to_reg): Likewise.
14085 (set_storage_via_setmem): Likewise.
14086 * optabs.c (prepare_cmp_insn): Likewise.
14087 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14088 * stor-layout.c (finish_bitfield_representative): Likewise.
14090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14091 Alan Hayward <alan.hayward@arm.com>
14092 David Sherwood <david.sherwood@arm.com>
14094 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
14095 * expr.c (convert_move): Use them.
14096 (convert_modes): Likewise.
14097 (store_expr_with_bounds): Likewise.
14099 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14100 Alan Hayward <alan.hayward@arm.com>
14101 David Sherwood <david.sherwood@arm.com>
14103 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
14104 parameter for the mode of "x". Remove the "known_x", "known_mode"
14105 and "known_ret" arguments. Change the type of the mode argument
14106 to scalar_int_mode.
14107 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
14108 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
14109 (reg_num_sign_bit_copies_for_combine): Likewise.
14110 * rtlanal.c (nonzero_bits1): Likewise.
14111 (num_sign_bit_copies1): Likewise.
14112 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
14113 (reg_num_sign_bit_copies_general): Likewise.
14114 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
14115 (reg_nonzero_bits_general): Likewise.
14117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14118 Alan Hayward <alan.hayward@arm.com>
14119 David Sherwood <david.sherwood@arm.com>
14121 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
14122 than in subroutines. Return 1 for non-integer modes.
14123 (cached_num_sign_bit_copies): Change the type of the mode parameter
14124 to scalar_int_mode.
14125 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
14126 classes. Handle CONST_INT_P first and then check whether X also
14127 has a scalar integer mode. Check the same thing for inner registers
14128 of a SUBREG and for values that are being extended or truncated.
14130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14131 Alan Hayward <alan.hayward@arm.com>
14132 David Sherwood <david.sherwood@arm.com>
14134 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
14135 in subroutines. Return the mode mask for non-integer modes.
14136 (cached_nonzero_bits): Change the type of the mode parameter
14137 to scalar_int_mode.
14138 (nonzero_bits1): Likewise. Remove early exit for other mode
14139 classes. Handle CONST_INT_P first and then check whether X
14140 also has a scalar integer mode.
14142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14143 Alan Hayward <alan.hayward@arm.com>
14144 David Sherwood <david.sherwood@arm.com>
14146 * expr.c (widest_int_mode_for_size): Make the comment match the code.
14147 Return a scalar_int_mode and assert that the size is greater than
14149 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
14150 (op_by_pieces_d::op_by_pieces_d): Likewise.
14151 (op_by_pieces_d::run): Likewise.
14152 (can_store_by_pieces): Likewise.
14154 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14155 Alan Hayward <alan.hayward@arm.com>
14156 David Sherwood <david.sherwood@arm.com>
14158 * combine.c (extract_left_shift): Add a mode argument and update
14160 (make_compound_operation_int): Change the type of the mode parameter
14161 to scalar_int_mode and update the call to extract_left_shift.
14163 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14164 Alan Hayward <alan.hayward@arm.com>
14165 David Sherwood <david.sherwood@arm.com>
14167 * combine.c (simplify_and_const_int): Change the type of the mode
14168 parameter to scalar_int_mode.
14169 (simplify_and_const_int_1): Likewise. Update recursive call.
14171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14172 Alan Hayward <alan.hayward@arm.com>
14173 David Sherwood <david.sherwood@arm.com>
14175 * combine.c (simplify_compare_const): Check that the mode is a
14176 scalar_int_mode (rather than VOIDmode) before testing its
14178 (simplify_comparison): Move COMPARISON_P handling out of the
14179 loop and restrict the latter part of the loop to scalar_int_modes.
14180 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
14181 and when considering SUBREG_REGs. Use is_int_mode instead of
14182 checking GET_MODE_CLASS against MODE_INT.
14184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14185 Alan Hayward <alan.hayward@arm.com>
14186 David Sherwood <david.sherwood@arm.com>
14188 * combine.c (try_widen_shift_mode): Move check for equal modes to...
14189 (simplify_shift_const_1): ...here. Use scalar_int_mode for
14190 shift_unit_mode and for modes involved in scalar shifts.
14192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14193 Alan Hayward <alan.hayward@arm.com>
14194 David Sherwood <david.sherwood@arm.com>
14196 * combine.c (force_int_to_mode): New function, split out from...
14197 (force_to_mode): ...here. Keep xmode up-to-date and use it
14198 instead of GET_MODE (x).
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 * optabs-query.h (extraction_insn::struct_mode): Change type to
14205 opt_scalar_int_mode and update comment.
14206 (extraction_insn::field_mode): Change type to scalar_int_mode.
14207 (extraction_insn::pos_mode): Likewise.
14208 * combine.c (make_extraction): Update accordingly.
14209 * optabs-query.c (get_traditional_extraction_insn): Likewise.
14210 (get_optab_extraction_insn): Likewise.
14211 * recog.c (simplify_while_replacing): Likewise.
14212 * expmed.c (narrow_bit_field_mem): Change the type of the mode
14213 parameter to opt_scalar_int_mode.
14215 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14216 Alan Hayward <alan.hayward@arm.com>
14217 David Sherwood <david.sherwood@arm.com>
14219 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
14220 to a scalar_int_mode instead of a machine_mode.
14221 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
14222 (get_best_mode): Return a boolean and use a pointer argument to store
14223 the selected mode. Replace the limit mode parameter with a bit limit.
14224 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
14225 for the values returned by bit_field_mode_iterator::next_mode.
14226 (store_bit_field): Update call to get_best_mode.
14227 (store_fixed_bit_field): Likewise.
14228 (extract_fixed_bit_field): Likewise.
14229 * expr.c (optimize_bitfield_assignment_op): Likewise.
14230 * fold-const.c (optimize_bit_field_compare): Likewise.
14231 (fold_truth_andor_1): Likewise.
14232 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
14233 Update for new type of m_mode.
14234 (get_best_mode): As above.
14236 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14237 Alan Hayward <alan.hayward@arm.com>
14238 David Sherwood <david.sherwood@arm.com>
14240 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
14241 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
14242 (store_bit_field): Check is_a <scalar_int_mode> before calling
14243 strict_volatile_bitfield_p.
14244 (extract_bit_field): Likewise.
14246 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14247 Alan Hayward <alan.hayward@arm.com>
14248 David Sherwood <david.sherwood@arm.com>
14250 * target.def (cstore_mode): Return a scalar_int_mode.
14251 * doc/tm.texi: Regenerate.
14252 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
14253 * targhooks.h (default_cstore_mode): Likewise.
14254 * targhooks.c (default_cstore_mode): Likewise, using a forced
14256 * expmed.c (emit_cstore): Expect the target of the cstore to be
14259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14260 Alan Hayward <alan.hayward@arm.com>
14261 David Sherwood <david.sherwood@arm.com>
14263 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
14265 (niter_desc): Likewise mode.
14266 (iv_analyze): Add a mode parameter.
14268 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
14269 and change its type to scalar_int_mode.
14270 * loop-iv.c: Update commentary at head of file.
14271 (iv_constant): Pass the mode paraeter before the rtx it describes
14272 and change its type to scalar_int_mode. Remove VOIDmode handling.
14273 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
14274 (iv_extend): Likewise.
14275 (shorten_into_mode): Likewise.
14276 (iv_add): Use scalar_int_mode.
14277 (iv_mult): Likewise.
14278 (iv_shift): Likewise.
14279 (canonicalize_iv_subregs): Likewise.
14280 (get_biv_step_1): Pass the outer_mode parameter before the rtx
14281 it describes and change its mode to scalar_int_mode. Also change
14282 the type of the returned inner_mode to scalar_int_mode.
14283 (get_biv_step): Likewise, turning outer_mode from a pointer
14284 into a direct parameter. Update call to get_biv_step_1.
14285 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
14286 iv_constant and get_biv_step.
14287 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
14288 and change its type to scalar_int_mode. Don't initialise iv->mode
14289 to VOIDmode and remove later checks for its still being VOIDmode.
14290 Update calls to iv_analyze_op and iv_analyze_expr. Check
14291 is_a <scalar_int_mode> when changing the mode under consideration.
14292 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
14293 Update call to iv_analyze_expr.
14294 (iv_analyze_op): Add a mode parameter. Reject subregs whose
14295 inner register is not also a scalar_int_mode. Update call to
14297 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
14298 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
14299 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
14300 separate mode class checks. Update calls to iv_analyze. Remove
14301 fix-up of VOIDmodes after iv_analyze_biv.
14302 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
14303 don't have a scalar_int_mode. Update call to biv_p.
14305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14306 Alan Hayward <alan.hayward@arm.com>
14307 David Sherwood <david.sherwood@arm.com>
14309 * cfgexpand.c (convert_debug_memory_address): Use
14310 as_a <scalar_int_mode>.
14311 * combine.c (expand_compound_operation): Likewise.
14312 (make_extraction): Likewise.
14313 (change_zero_ext): Likewise.
14314 (simplify_comparison): Likewise.
14315 * cse.c (cse_insn): Likewise.
14316 * dwarf2out.c (minmax_loc_descriptor): Likewise.
14317 (mem_loc_descriptor): Likewise.
14318 (loc_descriptor): Likewise.
14319 * expmed.c (init_expmed_one_mode): Likewise.
14320 (synth_mult): Likewise.
14321 (emit_store_flag_1): Likewise.
14322 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
14323 of a comparison with size.
14324 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
14325 (reduce_to_bit_field_precision): Likewise.
14326 * function.c (expand_function_end): Likewise.
14327 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
14328 * loop-doloop.c (doloop_modify): Likewise.
14329 * optabs.c (expand_binop): Likewise.
14330 (expand_unop): Likewise.
14331 (expand_copysign_absneg): Likewise.
14332 (prepare_cmp_insn): Likewise.
14333 (maybe_legitimize_operand): Likewise.
14334 * recog.c (const_scalar_int_operand): Likewise.
14335 * rtlanal.c (get_address_mode): Likewise.
14336 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14337 (simplify_cond_clz_ctz): Likewise.
14338 * tree-nested.c (get_nl_goto_field): Likewise.
14339 * tree.c (build_vector_type_for_mode): Likewise.
14340 * var-tracking.c (use_narrower_mode): Likewise.
14342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14343 Alan Hayward <alan.hayward@arm.com>
14344 David Sherwood <david.sherwood@arm.com>
14346 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
14347 * builtins.c (expand_builtin_signbit): Use it.
14348 * cfgexpand.c (expand_debug_expr): Likewise.
14349 * dojump.c (do_jump): Likewise.
14350 (do_compare_and_jump): Likewise.
14351 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
14352 * expmed.c (make_tree): Likewise.
14353 * expr.c (expand_expr_real_2): Likewise.
14354 (expand_expr_real_1): Likewise.
14355 (try_casesi): Likewise.
14356 * fold-const-call.c (fold_const_call_ss): Likewise.
14357 * fold-const.c (unextend): Likewise.
14358 (extract_muldiv_1): Likewise.
14359 (fold_single_bit_test): Likewise.
14360 (native_encode_int): Likewise.
14361 (native_encode_string): Likewise.
14362 (native_interpret_int): Likewise.
14363 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
14364 * internal-fn.c (expand_addsub_overflow): Likewise.
14365 (expand_neg_overflow): Likewise.
14366 (expand_mul_overflow): Likewise.
14367 (expand_arith_overflow): Likewise.
14368 * match.pd: Likewise.
14369 * stor-layout.c (layout_type): Likewise.
14370 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14371 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
14372 * tree-ssanames.c (get_range_info): Likewise.
14373 * tree-switch-conversion.c (array_value_type) Likewise.
14374 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
14375 (vect_recog_divmod_pattern): Likewise.
14376 (vect_recog_mixed_size_cond_pattern): Likewise.
14377 * tree-vrp.c (extract_range_basic): Likewise.
14378 (simplify_float_conversion_using_ranges): Likewise.
14379 * tree.c (int_fits_type_p): Likewise.
14380 * ubsan.c (instrument_bool_enum_load): Likewise.
14381 * varasm.c (mergeable_string_section): Likewise.
14382 (narrowing_initializer_constant_valid_p): Likewise.
14383 (output_constant): Likewise.
14385 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14386 Alan Hayward <alan.hayward@arm.com>
14387 David Sherwood <david.sherwood@arm.com>
14389 * machmode.h (NARROWEST_INT_MODE): New macro.
14390 * expr.c (alignment_for_piecewise_move): Use it instead of
14391 GET_CLASS_NARROWEST_MODE (MODE_INT).
14392 (push_block): Likewise.
14393 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
14395 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14398 Alan Hayward <alan.hayward@arm.com>
14399 David Sherwood <david.sherwood@arm.com>
14401 * postreload.c (move2add_valid_value_p): Change the type of the
14402 mode parameter to scalar_int_mode.
14403 (move2add_use_add2_insn): Add a mode parameter and use it instead
14405 (move2add_use_add3_insn): Likewise.
14406 (reload_cse_move2add): Update accordingly.
14408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14409 Alan Hayward <alan.hayward@arm.com>
14410 David Sherwood <david.sherwood@arm.com>
14412 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
14414 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
14415 * optabs.c (expand_unop): Likewise.
14417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14418 Alan Hayward <alan.hayward@arm.com>
14419 David Sherwood <david.sherwood@arm.com>
14421 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
14422 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
14423 (popcount_loc_descriptor): Likewise.
14424 (bswap_loc_descriptor): Likewise.
14425 (rotate_loc_descriptor): Likewise.
14426 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
14427 calling the functions above.
14429 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14430 Alan Hayward <alan.hayward@arm.com>
14431 David Sherwood <david.sherwood@arm.com>
14433 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
14435 (try_combine): Likewise.
14436 (simplify_if_then_else): Likewise.
14437 * cse.c (cse_insn): Likewise.
14438 * dwarf2out.c (mem_loc_descriptor): Likewise.
14439 * emit-rtl.c (gen_lowpart_common): Likewise.
14440 * simplify-rtx.c (simplify_truncation): Likewise.
14441 (simplify_binary_operation_1): Likewise.
14442 (simplify_const_relational_operation): Likewise.
14443 (simplify_ternary_operation): Likewise.
14444 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
14446 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14447 Alan Hayward <alan.hayward@arm.com>
14448 David Sherwood <david.sherwood@arm.com>
14450 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
14451 * reload.c (push_reload): Likewise.
14452 (find_reloads): Likewise.
14454 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14455 Alan Hayward <alan.hayward@arm.com>
14456 David Sherwood <david.sherwood@arm.com>
14458 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
14459 (make_compound_operation_int): Likewise.
14460 (change_zero_ext): Likewise.
14461 * expr.c (convert_move): Likewise.
14462 (convert_modes): Likewise.
14463 * fwprop.c (forward_propagate_subreg): Likewise.
14464 * loop-iv.c (get_biv_step_1): Likewise.
14465 * optabs.c (widen_operand): Likewise.
14466 * postreload.c (move2add_valid_value_p): Likewise.
14467 * recog.c (simplify_while_replacing): Likewise.
14468 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14469 (simplify_binary_operation_1): Likewise. Remove redundant
14470 mode equality check.
14472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14473 Alan Hayward <alan.hayward@arm.com>
14474 David Sherwood <david.sherwood@arm.com>
14476 * combine.c (combine_simplify_rtx): Add checks for
14477 is_a <scalar_int_mode>.
14478 (simplify_if_then_else): Likewise.
14479 (make_field_assignment): Likewise.
14480 (simplify_comparison): Likewise.
14481 * ifcvt.c (noce_try_bitop): Likewise.
14482 * loop-invariant.c (canonicalize_address_mult): Likewise.
14483 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14486 Alan Hayward <alan.hayward@arm.com>
14487 David Sherwood <david.sherwood@arm.com>
14489 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
14490 is_a <scalar_int_mode> instead of != BLKmode.
14492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14493 Alan Hayward <alan.hayward@arm.com>
14494 David Sherwood <david.sherwood@arm.com>
14496 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
14497 instead of != VOIDmode.
14498 * combine.c (if_then_else_cond): Likewise.
14499 (change_zero_ext): Likewise.
14500 * dwarf2out.c (mem_loc_descriptor): Likewise.
14501 (loc_descriptor): Likewise.
14502 * rtlanal.c (canonicalize_condition): Likewise.
14503 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
14505 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14506 Alan Hayward <alan.hayward@arm.com>
14507 David Sherwood <david.sherwood@arm.com>
14509 * simplify-rtx.c (simplify_binary_operation_1): Use
14510 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
14512 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14513 Alan Hayward <alan.hayward@arm.com>
14514 David Sherwood <david.sherwood@arm.com>
14516 * wide-int.h (int_traits<unsigned char>) New class.
14517 (int_traits<unsigned short>) Likewise.
14518 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
14519 Use GET_MODE_UNIT_PRECISION and remove redundant test for
14521 * combine.c (set_nonzero_bits_and_sign_copies): Use
14522 is_a <scalar_int_mode>.
14523 (find_split_point): Likewise.
14524 (combine_simplify_rtx): Likewise.
14525 (simplify_logical): Likewise.
14526 (expand_compound_operation): Likewise.
14527 (expand_field_assignment): Likewise.
14528 (make_compound_operation): Likewise.
14529 (extended_count): Likewise.
14530 (change_zero_ext): Likewise.
14531 (simplify_comparison): Likewise.
14532 * dwarf2out.c (scompare_loc_descriptor): Likewise.
14533 (ucompare_loc_descriptor): Likewise.
14534 (minmax_loc_descriptor): Likewise.
14535 (mem_loc_descriptor): Likewise.
14536 (loc_descriptor): Likewise.
14537 * expmed.c (init_expmed_one_mode): Likewise.
14538 * lra-constraints.c (lra_constraint_offset): Likewise.
14539 * optabs.c (prepare_libcall_arg): Likewise.
14540 * postreload.c (move2add_note_store): Likewise.
14541 * reload.c (operands_match_p): Likewise.
14542 * rtl.h (load_extend_op): Likewise.
14543 * rtlhooks.c (gen_lowpart_general): Likewise.
14544 * simplify-rtx.c (simplify_truncation): Likewise.
14545 (simplify_unary_operation_1): Likewise.
14546 (simplify_binary_operation_1): Likewise.
14547 (simplify_const_binary_operation): Likewise.
14548 (simplify_const_relational_operation): Likewise.
14549 (simplify_subreg): Likewise.
14550 * stor-layout.c (bitwise_mode_for_mode): Likewise.
14551 * var-tracking.c (adjust_mems): Likewise.
14552 (prepare_call_arguments): Likewise.
14554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14555 Alan Hayward <alan.hayward@arm.com>
14556 David Sherwood <david.sherwood@arm.com>
14558 * machmode.h (is_int_mode): New fuction.
14559 * combine.c (find_split_point): Use it.
14560 (combine_simplify_rtx): Likewise.
14561 (simplify_if_then_else): Likewise.
14562 (simplify_set): Likewise.
14563 (simplify_shift_const_1): Likewise.
14564 (simplify_comparison): Likewise.
14565 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
14566 * cse.c (notreg_cost): Likewise.
14567 (cse_insn): Likewise.
14568 * cselib.c (cselib_lookup_1): Likewise.
14569 * dojump.c (do_jump_1): Likewise.
14570 (do_compare_rtx_and_jump): Likewise.
14571 * dse.c (get_call_args): Likewise.
14572 * dwarf2out.c (rtl_for_decl_init): Likewise.
14573 (native_encode_initializer): Likewise.
14574 * expmed.c (emit_store_flag_1): Likewise.
14575 (emit_store_flag): Likewise.
14576 * expr.c (convert_modes): Likewise.
14577 (store_field): Likewise.
14578 (expand_expr_real_1): Likewise.
14579 * fold-const.c (fold_read_from_constant_string): Likewise.
14580 * gimple-ssa-sprintf.c (get_format_string): Likewise.
14581 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
14582 * optabs.c (expand_binop): Likewise.
14583 (expand_unop): Likewise.
14584 (expand_abs_nojump): Likewise.
14585 (expand_one_cmpl_abs_nojump): Likewise.
14586 * simplify-rtx.c (mode_signbit_p): Likewise.
14587 (val_signbit_p): Likewise.
14588 (val_signbit_known_set_p): Likewise.
14589 (val_signbit_known_clear_p): Likewise.
14590 (simplify_relational_operation_1): Likewise.
14591 * tree.c (vector_type_mode): Likewise.
14593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14594 Alan Hayward <alan.hayward@arm.com>
14595 David Sherwood <david.sherwood@arm.com>
14597 * machmode.h (smallest_mode_for_size): Fix formatting.
14598 (smallest_int_mode_for_size): New function.
14599 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
14600 instead of smallest_mode_for_size.
14601 * combine.c (make_extraction): Likewise.
14602 * config/arc/arc.c (arc_expand_movmem): Likewise.
14603 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
14604 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14605 * config/s390/s390.c (s390_expand_insv): Likewise.
14606 * config/sparc/sparc.c (assign_int_registers): Likewise.
14607 * config/spu/spu.c (spu_function_value): Likewise.
14608 (spu_function_arg): Likewise.
14609 * coverage.c (get_gcov_type): Likewise.
14610 (get_gcov_unsigned_t): Likewise.
14611 * dse.c (find_shift_sequence): Likewise.
14612 * expmed.c (store_bit_field_1): Likewise.
14613 * expr.c (convert_move): Likewise.
14614 (store_field): Likewise.
14615 * internal-fn.c (expand_arith_overflow): Likewise.
14616 * optabs-query.c (get_best_extraction_insn): Likewise.
14617 * optabs.c (expand_twoval_binop_libfunc): Likewise.
14618 * stor-layout.c (layout_type): Likewise.
14619 (initialize_sizetypes): Likewise.
14620 * targhooks.c (default_get_mask_mode): Likewise.
14621 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
14623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14624 Alan Hayward <alan.hayward@arm.com>
14625 David Sherwood <david.sherwood@arm.com>
14627 * machmode.h (opt_mode::else_blk): New function.
14628 (int_mode_for_mode): Declare.
14629 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
14630 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
14632 * cfgexpand.c (expand_debug_expr): Likewise.
14633 * combine.c (gen_lowpart_or_truncate): Likewise.
14634 (gen_lowpart_for_combine): Likewise.
14635 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
14636 * config/avr/avr.c (avr_to_int_mode): Likewise.
14637 (avr_out_plus_1): Likewise.
14638 (avr_out_plus): Likewise.
14639 (avr_out_round): Likewise.
14640 * config/i386/i386.c (ix86_split_to_parts): Likewise.
14641 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14642 (s390_expand_vcond): Likewise.
14643 * config/spu/spu.c (spu_split_immediate): Likewise.
14644 (spu_expand_mov): Likewise.
14645 * dse.c (get_stored_val): Likewise.
14646 * expmed.c (store_bit_field_1): Likewise.
14647 (convert_extracted_bit_field): Use int_mode_for_mode instead of
14649 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
14650 (extract_low_bits): Likewise.
14651 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
14652 handling rather than repeating the check.
14653 (emit_group_store): Likewise.
14654 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
14655 * optabs.c (expand_absneg_bit): Likewise.
14656 (expand_copysign_absneg): Likewise.
14657 (expand_copysign_bit): Likewise.
14658 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
14659 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
14660 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
14661 * var-tracking.c (prepare_call_arguments): Likewise.
14662 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
14663 int_mode_for_mode instead of mode_for_size.
14664 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14666 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14667 Alan Hayward <alan.hayward@arm.com>
14668 David Sherwood <david.sherwood@arm.com>
14670 * machmode.h (int_mode_for_size): New function.
14671 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
14672 instead of mode_for_size.
14673 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
14675 * combine.c (expand_field_assignment): Use int_mode_for_size
14676 instead of mode_for_size.
14677 (make_extraction): Likewise.
14678 (simplify_shift_const_1): Likewise.
14679 (simplify_comparison): Likewise.
14680 * dojump.c (do_jump): Likewise.
14681 * dwarf2out.c (mem_loc_descriptor): Likewise.
14682 * emit-rtl.c (init_derived_machine_modes): Likewise.
14683 * expmed.c (flip_storage_order): Likewise.
14684 (convert_extracted_bit_field): Likewise.
14685 * expr.c (copy_blkmode_from_reg): Likewise.
14686 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
14687 * internal-fn.c (expand_mul_overflow): Likewise.
14688 * lower-subreg.c (simple_move): Likewise.
14689 * optabs-libfuncs.c (init_optabs): Likewise.
14690 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14691 * tree.c (vector_type_mode): Likewise.
14692 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
14693 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
14694 * tree-vect-generic.c (expand_vector_parallel): Likewise.
14695 * tree-vect-stmts.c (vectorizable_load): Likewise.
14696 (vectorizable_store): Likewise.
14698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14699 Alan Hayward <alan.hayward@arm.com>
14700 David Sherwood <david.sherwood@arm.com>
14702 * coretypes.h (pod_mode): New type.
14703 (scalar_int_mode_pod): New typedef.
14704 * machmode.h (pod_mode): New class.
14705 (int_n_data_t::m): Change type to scalar_int_mode_pod.
14706 * genmodes.c (emit_mode_int_n): Update accordingly.
14707 * lower-subreg.h (target_lower_subreg): Change type to
14708 scalar_int_mode_pod.
14709 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
14710 scalar_int_mode_pod.
14712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14713 Alan Hayward <alan.hayward@arm.com>
14714 David Sherwood <david.sherwood@arm.com>
14716 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
14717 machine_mode to scalar_int_mode.
14718 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
14719 (rs6000_option_override_internal): Remove cast to int.
14720 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
14721 machine_mode to scalar_int_mode.
14722 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
14723 (rs6000_option_override_internal): Remove cast to int.
14724 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
14725 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
14727 * config/s390/s390.c (s390_expand_builtin): Likewise.
14728 * coretypes.h (scalar_int_mode): New type.
14729 (opt_scalar_int_mode): New typedef.
14730 * machmode.h (scalar_int_mode): New class.
14731 (scalar_int_mode::includes_p): New function.
14732 (byte_mode): Change type to scalar_int_mode.
14733 (word_mode): Likewise.
14734 (ptr_mode): Likewise.
14735 * emit-rtl.c (byte_mode): Likewise.
14736 (word_mode): Likewise.
14737 (ptr_mode): Likewise.
14738 (init_derived_machine_modes): Update accordingly.
14739 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
14740 and MODE_PARTIAL_INT.
14741 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
14742 opt_scalar_int_mode.
14744 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14745 Alan Hayward <alan.hayward@arm.com>
14746 David Sherwood <david.sherwood@arm.com>
14748 * target.def (libgcc_floating_mode_supported_p): Take a
14750 * doc/tm.texi: Regenerate.
14751 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
14753 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
14754 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
14757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14758 Alan Hayward <alan.hayward@arm.com>
14759 David Sherwood <david.sherwood@arm.com>
14761 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
14762 * doc/tm.texi: Regenerate.
14763 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
14764 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
14765 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
14766 * targhooks.h (default_floatn_mode): Likewise.
14767 * targhooks.c (default_floatn_mode): Likewise.
14768 * tree.c (build_common_tree_nodes): Update accordingly.
14770 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14771 Alan Hayward <alan.hayward@arm.com>
14772 David Sherwood <david.sherwood@arm.com>
14774 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
14775 (mode_iterator::iterate_p): Likewise.
14776 (mode_iterator::get_wider): Likewise.
14777 * expr.c (init_expr_target): Use opt_scalar_float_mode.
14779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14780 Alan Hayward <alan.hayward@arm.com>
14781 David Sherwood <david.sherwood@arm.com>
14783 * coretypes.h (opt_scalar_float_mode): New typedef.
14784 * machmode.h (float_mode_for_size): New function.
14785 * emit-rtl.c (double_mode): Delete.
14786 (init_emit_once): Use float_mode_for_size.
14787 * stor-layout.c (layout_type): Likewise.
14788 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
14790 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14791 Alan Hayward <alan.hayward@arm.com>
14792 David Sherwood <david.sherwood@arm.com>
14794 * output.h (assemble_real): Take a scalar_float_mode.
14795 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
14796 * config/arm/arm.md (consttable_4): Likewise.
14797 (consttable_8): Likewise.
14798 (consttable_16): Likewise.
14799 * config/mips/mips.md (consttable_float): Likewise.
14800 * config/s390/s390.c (s390_output_pool_entry): Likewise.
14801 * varasm.c (assemble_real): Take a scalar_float_mode.
14802 (output_constant_pool_2): Update accordingly.
14803 (output_constant): Likewise.
14805 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14806 Alan Hayward <alan.hayward@arm.com>
14807 David Sherwood <david.sherwood@arm.com>
14809 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
14810 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
14811 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
14812 (native_encode_real): Likewise.
14813 (native_interpret_real): Likewise.
14814 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
14815 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14818 Alan Hayward <alan.hayward@arm.com>
14819 David Sherwood <david.sherwood@arm.com>
14821 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
14822 <scalar_float_mode>. Simplify.
14823 (gen_extend_conv_libfunc): Likewise.
14825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14826 Alan Hayward <alan.hayward@arm.com>
14827 David Sherwood <david.sherwood@arm.com>
14829 * coretypes.h (scalar_float_mode): New type.
14830 * machmode.h (mode_traits::from_int): Use machine_mode if
14831 USE_ENUM_MODES is defined.
14832 (is_a): New function.
14834 (dyn_cast): Likewise.
14835 (scalar_float_mode): New class.
14836 (scalar_float_mode::includes_p): New function.
14837 (is_float_mode): Likewise.
14838 * gdbhooks.py (MachineModePrinter): New class.
14839 (build_pretty_printer): Use it for scalar_float_mode.
14840 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
14841 (format_helper::format_helper): Turn into a template.
14842 * genmodes.c (get_mode_class): New function.
14843 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
14844 or machine_mode if none.
14845 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
14846 as_a <scalar_float_mode>.
14847 * dwarf2out.c (mem_loc_descriptor): Likewise.
14848 (insert_float): Likewise.
14849 (add_const_value_attribute): Likewise.
14850 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14851 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
14852 (expand_unop): Update accordingly.
14853 (expand_abs_nojump): Likewise.
14854 (expand_copysign_absneg): Take a scalar_float_mode.
14855 (expand_copysign_bit): Likewise.
14856 (expand_copysign): Update accordingly.
14858 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14859 Alan Hayward <alan.hayward@arm.com>
14860 David Sherwood <david.sherwood@arm.com>
14862 * coretypes.h (opt_mode): New class.
14863 * machmode.h (opt_mode): Likewise.
14864 (opt_mode::else_void): New function.
14865 (opt_mode::require): Likewise.
14866 (opt_mode::exists): Likewise.
14867 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
14868 (GET_MODE_2XWIDER_MODE): Likewise.
14869 (mode_iterator::get_wider): Update accordingly.
14870 (mode_iterator::get_2xwider): Likewise.
14871 (mode_iterator::get_known_wider): Likewise, turning into a template.
14872 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
14873 forcing a wider mode to exist.
14874 * config/cr16/cr16.h (LONG_REG_P): Likewise.
14875 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14876 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
14877 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
14878 * lower-subreg.c (init_lower_subreg): Likewise.
14879 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
14880 on the final iteration.
14881 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
14882 a wider mode exists before asking for a move pattern.
14883 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
14884 forcing a wider mode to exist.
14885 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
14886 returning false if no such mode exists.
14887 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
14888 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
14889 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
14890 Avoid checking for a MODE_INT if we already know the mode is not a
14892 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
14893 forcing a wider mode to exist.
14894 (expmed_mult_highpart_optab): Likewise.
14895 (expmed_mult_highpart): Likewise.
14896 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
14898 * lto-streamer-in.c (lto_input_mode_table): Likewise.
14899 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
14900 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
14901 * internal-fn.c (expand_mul_overflow): Update use of
14902 GET_MODE_2XWIDER_MODE.
14903 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14904 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
14905 GET_MODE_WIDER_MODE.
14906 (convert_plusminus_to_widen): Likewise.
14907 * tree-switch-conversion.c (array_value_type): Likewise.
14908 * var-tracking.c (emit_note_insn_var_location): Likewise.
14909 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14910 Return false inside rather than outside the loop if no wider mode
14912 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
14913 and GET_MODE_2XWIDER_MODE
14914 (can_compare_p): Use else_void.
14915 * gdbhooks.py (OptMachineModePrinter): New class.
14916 (build_pretty_printer): Use it for opt_mode.
14918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14919 Alan Hayward <alan.hayward@arm.com>
14920 David Sherwood <david.sherwood@arm.com>
14922 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
14923 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
14925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14926 Alan Hayward <alan.hayward@arm.com>
14927 David Sherwood <david.sherwood@arm.com>
14929 * machmode.h (mode_traits): New structure.
14930 (get_narrowest_mode): New function.
14931 (mode_iterator::start): Likewise.
14932 (mode_iterator::iterate_p): Likewise.
14933 (mode_iterator::get_wider): Likewise.
14934 (mode_iterator::get_known_wider): Likewise.
14935 (mode_iterator::get_2xwider): Likewise.
14936 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
14937 (FOR_EACH_MODE): Likewise.
14938 (FOR_EACH_MODE_FROM): Likewise.
14939 (FOR_EACH_MODE_UNTIL): Likewise.
14940 (FOR_EACH_WIDER_MODE): Likewise.
14941 (FOR_EACH_2XWIDER_MODE): Likewise.
14942 * builtins.c (expand_builtin_strlen): Use new mode iterators.
14943 * combine.c (simplify_comparison): Likewise
14944 * config/i386/i386.c (type_natural_mode): Likewise.
14945 * cse.c (cse_insn): Likewise.
14946 * dse.c (find_shift_sequence): Likewise.
14947 * emit-rtl.c (init_derived_machine_modes): Likewise.
14948 (init_emit_once): Likewise.
14949 * explow.c (hard_function_value): Likewise.
14950 * expmed.c (extract_fixed_bit_field_1): Likewise.
14951 (extract_bit_field_1): Likewise.
14952 (expand_divmod): Likewise.
14953 (emit_store_flag_1): Likewise.
14954 * expr.c (init_expr_target): Likewise.
14955 (convert_move): Likewise.
14956 (alignment_for_piecewise_move): Likewise.
14957 (widest_int_mode_for_size): Likewise.
14958 (emit_block_move_via_movmem): Likewise.
14959 (copy_blkmode_to_reg): Likewise.
14960 (set_storage_via_setmem): Likewise.
14961 (compress_float_constant): Likewise.
14962 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14963 * optabs-query.c (get_best_extraction_insn): Likewise.
14964 * optabs.c (expand_binop): Likewise.
14965 (expand_twoval_unop): Likewise.
14966 (expand_twoval_binop): Likewise.
14967 (widen_leading): Likewise.
14968 (widen_bswap): Likewise.
14969 (expand_parity): Likewise.
14970 (expand_unop): Likewise.
14971 (prepare_cmp_insn): Likewise.
14972 (prepare_float_lib_cmp): Likewise.
14973 (expand_float): Likewise.
14974 (expand_fix): Likewise.
14975 (expand_sfix_optab): Likewise.
14976 * postreload.c (move2add_use_add2_insn): Likewise.
14977 * reg-stack.c (reg_to_stack): Likewise.
14978 * reginfo.c (choose_hard_reg_mode): Likewise.
14979 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14980 * stor-layout.c (mode_for_size): Likewise.
14981 (smallest_mode_for_size): Likewise.
14982 (mode_for_vector): Likewise.
14983 (finish_bitfield_representative): Likewise.
14984 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
14985 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
14986 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14987 * var-tracking.c (prepare_call_arguments): Likewise.
14989 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14990 Alan Hayward <alan.hayward@arm.com>
14991 David Sherwood <david.sherwood@arm.com>
14993 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
14994 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
14995 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
14996 * machmode.h (mode_size): Move earlier in file.
14997 (mode_precision): Likewise.
14998 (mode_inner): Likewise.
14999 (mode_nunits): Likewise.
15000 (mode_unit_size): Likewise.
15001 (unit_unit_precision): Likewise.
15002 (mode_wider): Likewise.
15003 (mode_2xwider): Likewise.
15004 (machine_mode): New class.
15005 (mode_to_bytes): New function.
15006 (mode_to_bits): Likewise.
15007 (mode_to_precision): Likewise.
15008 (mode_to_inner): Likewise.
15009 (mode_to_unit_size): Likewise.
15010 (mode_to_unit_precision): Likewise.
15011 (mode_to_nunits): Likewise.
15012 (GET_MODE_SIZE): Use mode_to_bytes.
15013 (GET_MODE_BITSIZE): Use mode_to_bits.
15014 (GET_MODE_PRECISION): Use mode_to_precision.
15015 (GET_MODE_INNER): Use mode_to_inner.
15016 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
15017 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
15018 (GET_MODE_NUNITS): Use mode_to_nunits.
15019 * system.h (ALWAYS_INLINE): New macro.
15020 * config/powerpcspe/powerpcspe-c.c
15021 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
15022 int for arg1_mode and arg2_mode.
15024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15025 Alan Hayward <alan.hayward@arm.com>
15026 David Sherwood <david.sherwood@arm.com>
15028 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
15029 Prefix mode names with E_ in case statements.
15030 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15031 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
15032 (aarch64_split_simd_move): Likewise.
15033 (aarch64_gen_storewb_pair): Likewise.
15034 (aarch64_gen_loadwb_pair): Likewise.
15035 (aarch64_gen_store_pair): Likewise.
15036 (aarch64_gen_load_pair): Likewise.
15037 (aarch64_get_condition_code_1): Likewise.
15038 (aarch64_constant_pool_reload_icode): Likewise.
15039 (get_rsqrte_type): Likewise.
15040 (get_rsqrts_type): Likewise.
15041 (get_recpe_type): Likewise.
15042 (get_recps_type): Likewise.
15043 (aarch64_gimplify_va_arg_expr): Likewise.
15044 (aarch64_simd_container_mode): Likewise.
15045 (aarch64_emit_load_exclusive): Likewise.
15046 (aarch64_emit_store_exclusive): Likewise.
15047 (aarch64_expand_compare_and_swap): Likewise.
15048 (aarch64_gen_atomic_cas): Likewise.
15049 (aarch64_emit_bic): Likewise.
15050 (aarch64_emit_atomic_swap): Likewise.
15051 (aarch64_emit_atomic_load_op): Likewise.
15052 (aarch64_evpc_trn): Likewise.
15053 (aarch64_evpc_uzp): Likewise.
15054 (aarch64_evpc_zip): Likewise.
15055 (aarch64_evpc_ext): Likewise.
15056 (aarch64_evpc_rev): Likewise.
15057 (aarch64_evpc_dup): Likewise.
15058 (aarch64_gen_ccmp_first): Likewise.
15059 (aarch64_gen_ccmp_next): Likewise.
15060 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15061 (alpha_emit_xfloating_libcall): Likewise.
15062 (emit_insxl): Likewise.
15063 (alpha_arg_type): Likewise.
15064 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
15065 (arc_preferred_simd_mode): Likewise.
15066 (arc_secondary_reload): Likewise.
15067 (get_arc_condition_code): Likewise.
15068 (arc_print_operand): Likewise.
15069 (arc_legitimate_constant_p): Likewise.
15070 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15071 * config/arc/arc.md (casesi_load): Likewise.
15072 (casesi_compact_jump): Likewise.
15073 * config/arc/predicates.md (proper_comparison_operator): Likewise.
15074 (cc_use_register): Likewise.
15075 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15076 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
15077 (arm_init_iwmmxt_builtins): Likewise.
15078 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
15079 (neon_expand_vector_init): Likewise.
15080 (arm_attr_length_move_neon): Likewise.
15081 (maybe_get_arm_condition_code): Likewise.
15082 (arm_emit_vector_const): Likewise.
15083 (arm_preferred_simd_mode): Likewise.
15084 (arm_output_iwmmxt_tinsr): Likewise.
15085 (thumb1_output_casesi): Likewise.
15086 (thumb2_output_casesi): Likewise.
15087 (arm_emit_load_exclusive): Likewise.
15088 (arm_emit_store_exclusive): Likewise.
15089 (arm_expand_compare_and_swap): Likewise.
15090 (arm_evpc_neon_vuzp): Likewise.
15091 (arm_evpc_neon_vzip): Likewise.
15092 (arm_evpc_neon_vrev): Likewise.
15093 (arm_evpc_neon_vtrn): Likewise.
15094 (arm_evpc_neon_vext): Likewise.
15095 (arm_validize_comparison): Likewise.
15096 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
15097 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
15098 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
15099 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
15100 (c6x_preferred_simd_mode): Likewise.
15101 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
15102 (epiphany_rtx_costs): Likewise.
15103 * config/epiphany/predicates.md (proper_comparison_operator):
15105 * config/frv/frv.c (condexec_memory_operand): Likewise.
15106 (frv_emit_move): Likewise.
15107 (output_move_single): Likewise.
15108 (output_condmove_single): Likewise.
15109 (frv_hard_regno_mode_ok): Likewise.
15110 (frv_matching_accg_mode): Likewise.
15111 * config/h8300/h8300.c (split_adds_subs): Likewise.
15112 (h8300_rtx_costs): Likewise.
15113 (h8300_print_operand): Likewise.
15114 (compute_mov_length): Likewise.
15115 (output_logical_op): Likewise.
15116 (compute_logical_op_length): Likewise.
15117 (compute_logical_op_cc): Likewise.
15118 (h8300_shift_needs_scratch_p): Likewise.
15119 (output_a_shift): Likewise.
15120 (compute_a_shift_length): Likewise.
15121 (compute_a_shift_cc): Likewise.
15122 (expand_a_rotate): Likewise.
15123 (output_a_rotate): Likewise.
15124 * config/i386/i386.c (classify_argument): Likewise.
15125 (function_arg_advance_32): Likewise.
15126 (function_arg_32): Likewise.
15127 (function_arg_64): Likewise.
15128 (function_value_64): Likewise.
15129 (ix86_gimplify_va_arg): Likewise.
15130 (ix86_legitimate_constant_p): Likewise.
15131 (put_condition_code): Likewise.
15132 (split_double_mode): Likewise.
15133 (ix86_avx256_split_vector_move_misalign): Likewise.
15134 (ix86_expand_vector_logical_operator): Likewise.
15135 (ix86_split_idivmod): Likewise.
15136 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
15137 (ix86_build_const_vector): Likewise.
15138 (ix86_build_signbit_mask): Likewise.
15139 (ix86_match_ccmode): Likewise.
15140 (ix86_cc_modes_compatible): Likewise.
15141 (ix86_expand_branch): Likewise.
15142 (ix86_expand_sse_cmp): Likewise.
15143 (ix86_expand_sse_movcc): Likewise.
15144 (ix86_expand_int_sse_cmp): Likewise.
15145 (ix86_expand_vec_perm_vpermi2): Likewise.
15146 (ix86_expand_vec_perm): Likewise.
15147 (ix86_expand_sse_unpack): Likewise.
15148 (ix86_expand_int_addcc): Likewise.
15149 (ix86_split_to_parts): Likewise.
15150 (ix86_vectorize_builtin_gather): Likewise.
15151 (ix86_vectorize_builtin_scatter): Likewise.
15152 (avx_vpermilp_parallel): Likewise.
15153 (inline_memory_move_cost): Likewise.
15154 (ix86_tieable_integer_mode_p): Likewise.
15155 (x86_maybe_negate_const_int): Likewise.
15156 (ix86_expand_vector_init_duplicate): Likewise.
15157 (ix86_expand_vector_init_one_nonzero): Likewise.
15158 (ix86_expand_vector_init_one_var): Likewise.
15159 (ix86_expand_vector_init_concat): Likewise.
15160 (ix86_expand_vector_init_interleave): Likewise.
15161 (ix86_expand_vector_init_general): Likewise.
15162 (ix86_expand_vector_set): Likewise.
15163 (ix86_expand_vector_extract): Likewise.
15164 (emit_reduc_half): Likewise.
15165 (ix86_emit_i387_round): Likewise.
15166 (ix86_mangle_type): Likewise.
15167 (ix86_expand_round_sse4): Likewise.
15168 (expand_vec_perm_blend): Likewise.
15169 (canonicalize_vector_int_perm): Likewise.
15170 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
15171 (expand_vec_perm_1): Likewise.
15172 (expand_vec_perm_interleave3): Likewise.
15173 (expand_vec_perm_even_odd_pack): Likewise.
15174 (expand_vec_perm_even_odd_1): Likewise.
15175 (expand_vec_perm_broadcast_1): Likewise.
15176 (ix86_vectorize_vec_perm_const_ok): Likewise.
15177 (ix86_expand_vecop_qihi): Likewise.
15178 (ix86_expand_mul_widen_hilo): Likewise.
15179 (ix86_expand_sse2_abs): Likewise.
15180 (ix86_expand_pextr): Likewise.
15181 (ix86_expand_pinsr): Likewise.
15182 (ix86_preferred_simd_mode): Likewise.
15183 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
15184 * config/i386/sse.md (*andnot<mode>3): Likewise.
15185 (<mask_codefor><code><mode>3<mask_name>): Likewise.
15186 (*<code><mode>3): Likewise.
15187 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
15188 (ia64_expand_atomic_op): Likewise.
15189 (ia64_arg_type): Likewise.
15190 (ia64_mode_to_int): Likewise.
15191 (ia64_scalar_mode_supported_p): Likewise.
15192 (ia64_vector_mode_supported_p): Likewise.
15193 (expand_vec_perm_broadcast): Likewise.
15194 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
15195 (iq2000_function_arg_advance): Likewise.
15196 (iq2000_function_arg): Likewise.
15197 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
15198 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
15199 (m68k_libcall_value): Likewise.
15200 (m68k_function_value): Likewise.
15201 (sched_attr_op_type): Likewise.
15202 * config/mcore/mcore.c (mcore_output_move): Likewise.
15203 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
15205 (microblaze_function_arg): Likewise.
15206 * config/mips/mips.c (mips16_build_call_stub): Likewise.
15207 (mips_print_operand): Likewise.
15208 (mips_mode_ok_for_mov_fmt_p): Likewise.
15209 (mips_vector_mode_supported_p): Likewise.
15210 (mips_preferred_simd_mode): Likewise.
15211 (mips_expand_vpc_loongson_even_odd): Likewise.
15212 (mips_expand_vec_unpack): Likewise.
15213 (mips_expand_vi_broadcast): Likewise.
15214 (mips_expand_vector_init): Likewise.
15215 (mips_expand_vec_reduc): Likewise.
15216 (mips_expand_msa_cmp): Likewise.
15217 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
15218 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
15219 (cc_flags_for_mode): Likewise.
15220 * config/msp430/msp430.c (msp430_print_operand): Likewise.
15221 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
15222 (nds32_output_casesi_pc_relative): Likewise.
15223 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15224 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
15225 (nvptx_gen_unpack): Likewise.
15226 (nvptx_gen_pack): Likewise.
15227 (nvptx_gen_shuffle): Likewise.
15228 (nvptx_gen_wcast): Likewise.
15229 (nvptx_preferred_simd_mode): Likewise.
15230 * config/pa/pa.c (pa_secondary_reload): Likewise.
15231 * config/pa/predicates.md (base14_operand): Likewise.
15232 * config/powerpcspe/powerpcspe-c.c
15233 (altivec_resolve_overloaded_builtin): Likewise.
15234 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
15236 (rs6000_preferred_simd_mode): Likewise.
15237 (output_vec_const_move): Likewise.
15238 (rs6000_expand_vector_extract): Likewise.
15239 (rs6000_split_vec_extract_var): Likewise.
15240 (reg_offset_addressing_ok_p): Likewise.
15241 (rs6000_legitimate_offset_address_p): Likewise.
15242 (rs6000_legitimize_address): Likewise.
15243 (rs6000_emit_set_const): Likewise.
15244 (rs6000_const_vec): Likewise.
15245 (rs6000_emit_move): Likewise.
15246 (spe_build_register_parallel): Likewise.
15247 (rs6000_darwin64_record_arg_recurse): Likewise.
15248 (swap_selector_for_mode): Likewise.
15249 (spe_init_builtins): Likewise.
15250 (paired_init_builtins): Likewise.
15251 (altivec_init_builtins): Likewise.
15252 (do_load_for_compare): Likewise.
15253 (rs6000_generate_compare): Likewise.
15254 (rs6000_expand_float128_convert): Likewise.
15255 (emit_load_locked): Likewise.
15256 (emit_store_conditional): Likewise.
15257 (rs6000_output_function_epilogue): Likewise.
15258 (rs6000_handle_altivec_attribute): Likewise.
15259 (rs6000_function_value): Likewise.
15260 (emit_fusion_gpr_load): Likewise.
15261 (emit_fusion_p9_load): Likewise.
15262 (emit_fusion_p9_store): Likewise.
15263 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
15264 (fusion_gpr_mem_load): Likewise.
15265 (fusion_addis_mem_combo_load): Likewise.
15266 (fusion_addis_mem_combo_store): Likewise.
15267 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
15268 (fusion_gpr_mem_load): Likewise.
15269 (fusion_addis_mem_combo_load): Likewise.
15270 (fusion_addis_mem_combo_store): Likewise.
15271 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15273 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
15274 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15275 (rs6000_preferred_simd_mode): Likewise.
15276 (output_vec_const_move): Likewise.
15277 (rs6000_expand_vector_extract): Likewise.
15278 (rs6000_split_vec_extract_var): Likewise.
15279 (reg_offset_addressing_ok_p): Likewise.
15280 (rs6000_legitimate_offset_address_p): Likewise.
15281 (rs6000_legitimize_address): Likewise.
15282 (rs6000_emit_set_const): Likewise.
15283 (rs6000_const_vec): Likewise.
15284 (rs6000_emit_move): Likewise.
15285 (rs6000_darwin64_record_arg_recurse): Likewise.
15286 (swap_selector_for_mode): Likewise.
15287 (paired_init_builtins): Likewise.
15288 (altivec_init_builtins): Likewise.
15289 (rs6000_expand_float128_convert): Likewise.
15290 (emit_load_locked): Likewise.
15291 (emit_store_conditional): Likewise.
15292 (rs6000_output_function_epilogue): Likewise.
15293 (rs6000_handle_altivec_attribute): Likewise.
15294 (rs6000_function_value): Likewise.
15295 (emit_fusion_gpr_load): Likewise.
15296 (emit_fusion_p9_load): Likewise.
15297 (emit_fusion_p9_store): Likewise.
15298 * config/rx/rx.c (rx_gen_move_template): Likewise.
15299 (flags_from_mode): Likewise.
15300 * config/s390/predicates.md (s390_alc_comparison): Likewise.
15301 (s390_slb_comparison): Likewise.
15302 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
15303 (s390_vector_mode_supported_p): Likewise.
15304 (s390_cc_modes_compatible): Likewise.
15305 (s390_match_ccmode_set): Likewise.
15306 (s390_canonicalize_comparison): Likewise.
15307 (s390_emit_compare_and_swap): Likewise.
15308 (s390_branch_condition_mask): Likewise.
15309 (s390_rtx_costs): Likewise.
15310 (s390_secondary_reload): Likewise.
15311 (__SECONDARY_RELOAD_CASE): Likewise.
15312 (s390_expand_cs): Likewise.
15313 (s390_preferred_simd_mode): Likewise.
15314 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
15315 * config/sh/sh.c (sh_print_operand): Likewise.
15316 (dump_table): Likewise.
15317 (sh_secondary_reload): Likewise.
15318 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15319 * config/sh/sh.md (casesi_worker_1): Likewise.
15320 (casesi_worker_2): Likewise.
15321 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
15322 (fcc_comparison_operator): Likewise.
15323 * config/sparc/sparc.c (sparc_expand_move): Likewise.
15324 (emit_soft_tfmode_cvt): Likewise.
15325 (sparc_preferred_simd_mode): Likewise.
15326 (output_cbranch): Likewise.
15327 (sparc_print_operand): Likewise.
15328 (sparc_expand_vec_perm_bmask): Likewise.
15329 (vector_init_bshuffle): Likewise.
15330 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15331 (spu_vector_mode_supported_p): Likewise.
15332 (spu_expand_insv): Likewise.
15333 (spu_emit_branch_or_set): Likewise.
15334 (spu_handle_vector_attribute): Likewise.
15335 (spu_builtin_splats): Likewise.
15336 (spu_builtin_extract): Likewise.
15337 (spu_builtin_promote): Likewise.
15338 (spu_expand_sign_extend): Likewise.
15339 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15340 (tilegx_simd_int): Likewise.
15341 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
15342 (tilepro_simd_int): Likewise.
15343 * config/v850/v850.c (const_double_split): Likewise.
15344 (v850_print_operand): Likewise.
15345 (ep_memory_offset): Likewise.
15346 * config/vax/vax.c (vax_rtx_costs): Likewise.
15347 (vax_output_int_move): Likewise.
15348 (vax_output_int_add): Likewise.
15349 (vax_output_int_subtract): Likewise.
15350 * config/visium/predicates.md (visium_branch_operator): Likewise.
15351 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
15352 (visium_print_operand_address): Likewise.
15353 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15354 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
15355 (xtensa_expand_conditional_branch): Likewise.
15356 (xtensa_copy_incoming_a7): Likewise.
15357 (xtensa_output_literal): Likewise.
15358 * dfp.c (decimal_real_maxval): Likewise.
15359 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15361 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15362 Alan Hayward <alan.hayward@arm.com>
15363 David Sherwood <david.sherwood@arm.com>
15365 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
15366 (mode_nunits_inline): Likewise.
15367 (mode_inner_inline): Likewise.
15368 (mode_unit_size_inline): Likewise.
15369 (mode_unit_precision_inline): Likewise.
15370 (emit_insn_modes_h): Likewise. Also emit a #define of the
15372 (emit_mode_wider): Add an E_ prefix to mode names.
15373 (emit_mode_complex): Likewise.
15374 (emit_mode_inner): Likewise.
15375 (emit_mode_adjustments): Likewise.
15376 (emit_mode_int_n): Likewise.
15377 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
15378 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
15379 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
15380 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
15381 (CRC32_BUILTIN, ENTRY): Likewise.
15382 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
15383 (aarch64_pop_regs): Likewise.
15384 (aarch64_process_components): Likewise.
15385 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
15386 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
15387 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
15388 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
15389 * config/arm/arm.c (arm_init_libfuncs): Likewise.
15390 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
15392 * config/i386/i386-builtin.def (pcmpestr): Likewise.
15393 (pcmpistr): Likewise.
15394 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
15395 * config/mmix/mmix.c (mmix_output_condition): Likewise.
15396 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
15398 * config/rl78/rl78.c (mduc_regs): Likewise.
15399 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15400 (htm_expand_builtin): Likewise.
15401 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
15402 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
15403 * config/xtensa/xtensa.c (print_operand): Likewise.
15404 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
15405 (NUM_MODE_VECTOR_INT): Likewise.
15406 * genoutput.c (null_operand): Likewise.
15407 (output_operand_data): Likewise.
15408 * genrecog.c (print_parameter_value): Likewise.
15409 * lra.c (debug_operand_data): Likewise.
15411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15413 * dwarf2out.c (scompare_loc_descriptor_wide)
15414 (scompare_loc_descriptor_narrow): New functions, split out from...
15415 (scompare_loc_descriptor): ...here.
15416 * expmed.c (emit_store_flag_int): New function, split out from...
15417 (emit_store_flag): ...here.
15419 2017-08-30 Richard Biener <rguenther@suse.de>
15421 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
15422 (dwarf2out_early_finish): Move setting of AT_pubnames from
15423 early debug output to early finish.
15425 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15427 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
15428 and -mdata-region to the assembler.
15430 2017-08-30 Richard Biener <rguenther@suse.de>
15432 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
15434 (gen_subprogram_die): Add DW_AT_object_pointer only early.
15435 (dwarf2out_early_global_decl): Only generate a DIE for the
15436 abstract origin if it doesn't already exist or is a declaration DIE.
15437 (resolve_addr): Do not add the linkage name twice when
15438 generating a stub DIE for the DW_TAG_GNU_call_site target.
15440 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15442 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15443 Use machine_mode rather than int for arg1_mode.
15445 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
15448 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
15449 that the second argument of the built-in functions to unpack
15450 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
15451 switch statement instead a lot of if statements.
15452 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
15453 Allow 64-bit values to be in Altivec registers as well as
15454 traditional floating point registers.
15455 (pack<mode>, FMOVE128_VSX iterator): Likewise.
15457 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
15459 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
15460 MAX_REGS_PER_ADDRESS == 1.
15462 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
15464 * config/i386/i386.opt (flag_fentry): Do not init to -1.
15465 * config/i386/i386.c (ix86_option_override_internal): Simplify
15466 setting of opts->x_flag_entry.
15468 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15469 Jakub Jelinek <jakub@redhat.com>
15470 Richard Biener <rguenther@suse.de>
15472 PR tree-optimization/81503
15473 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
15474 folded constant fits in the target type; reorder tests for clarity.
15476 2017-08-29 Martin Liska <mliska@suse.cz>
15478 * passes.def: Include pass_lower_switch.
15479 * stmt.c (dump_case_nodes): Remove and move to
15480 tree-switch-conversion.
15481 (case_values_threshold): Likewise.
15482 (expand_switch_as_decision_tree_p): Likewise.
15483 (emit_case_decision_tree): Likewise.
15484 (expand_case): Likewise.
15485 (balance_case_nodes): Likewise.
15486 (node_has_low_bound): Likewise.
15487 (node_has_high_bound): Likewise.
15488 (node_is_bounded): Likewise.
15489 (emit_case_nodes): Likewise.
15490 (struct simple_case_node): New struct.
15491 (add_case_node): Remove.
15492 (emit_case_dispatch_table): Use vector instead of case_list.
15493 (reset_out_edges_aux): Remove.
15494 (compute_cases_per_edge): Likewise.
15495 (expand_case): Build list of simple_case_node.
15496 (expand_sjlj_dispatch_table): Use it.
15497 * tree-switch-conversion.c (struct case_node): Moved from
15498 stmt.c and adjusted.
15499 (emit_case_nodes): Likewise.
15500 (node_has_low_bound): Likewise.
15501 (node_has_high_bound): Likewise.
15502 (node_is_bounded): Likewise.
15503 (case_values_threshold): Likewise.
15504 (reset_out_edges_aux): Likewise.
15505 (compute_cases_per_edge): Likewise.
15506 (add_case_node): Likewise.
15507 (dump_case_nodes): Likewise.
15508 (balance_case_nodes): Likewise.
15509 (expand_switch_as_decision_tree_p): Likewise.
15510 (emit_jump): Likewise.
15511 (emit_case_decision_tree): Likewise.
15512 (try_switch_expansion): Likewise.
15513 (do_jump_if_equal): Likewise.
15514 (emit_cmp_and_jump_insns): Likewise.
15515 (fix_phi_operands_for_edge): New function.
15516 (record_phi_operand_mapping): Likewise.
15517 (class pass_lower_switch): New pass.
15518 (pass_lower_switch::execute): New function.
15519 (make_pass_lower_switch): Likewise.
15520 (conditional_probability):
15521 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
15522 * tree-pass.h: Add make_pass_lower_switch.
15524 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15527 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
15530 2017-08-29 Richard Biener <rguenther@suse.de>
15532 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
15533 we do not add a DW_AT_inline attribute twice.
15534 (gen_subprogram_die): Remove code setting DW_AT_inline on
15535 DECL_ABSTRACT_P nodes.
15537 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
15539 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
15540 calls to internal functions.
15541 (gimplify_modify_expr): Likewise.
15542 * tree-call-cdce.c (use_internal_fn): Likewise.
15543 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
15544 (convert_to_divmod): Set the nothrow flag.
15545 * tree-if-conv.c (predicate_mem_writes): Likewise.
15546 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15547 (vectorizable_call): Likewise.
15548 (vectorizable_store): Likewise.
15549 (vectorizable_load): Likewise.
15550 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
15551 (vect_recog_mask_conversion_pattern): Likewise.
15553 2017-08-29 Martin Liska <mliska@suse.cz>
15556 * gcc.c (driver_handle_option): Add new argument.
15557 * opts-common.c (handle_option): Pass
15558 target_option_override_hook.
15559 * opts-global.c (lang_handle_option): Add new option.
15560 (set_default_handlers): Add new argument.
15561 (decode_options): Likewise.
15562 * opts.c (target_handle_option): Likewise.
15563 (common_handle_option): Call target_option_override_hook.
15564 * opts.h (struct cl_option_handler_func): Add hook for
15565 target option override.
15566 (struct cl_option_handlers): Likewise.
15567 (set_default_handlers): Add new argument.
15568 (decode_options): Likewise.
15569 (common_handle_option): Likewise.
15570 (target_handle_option): Likewise.
15571 * toplev.c (toplev::main): Pass targetm.target_option.override
15574 2017-08-29 Richard Biener <rguenther@suse.de>
15575 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
15577 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
15578 life to the active subtree.
15580 2017-08-28 Jeff Law <law@redhat.com>
15582 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
15583 derive_equivalences.
15584 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
15585 Code moved into....
15586 (edge_info::derive_equivalences): New private member function
15588 * tree-ssa-dom.c (class edge_info): Changed from a struct
15589 to a class. Add ctor/dtor, methods and data members.
15590 (edge_info::edge_info): Renamed from allocate_edge_info.
15591 Initialize additional members.
15592 (edge_info::~edge_info): New.
15593 (free_dom_edge_info): Delete the edge info.
15594 (record_edge_info): Use new class & associated member functions.
15595 Tighten forms for testing for edge equivalences.
15596 (record_temporary_equivalences): Iterate over the simple
15597 equivalences rather than assuming there's only one per edge.
15598 (cprop_into_successor_phis): Iterate over the simple
15599 equivalences rather than assuming there's only one per edge.
15600 (optimize_stmt): Use operand_equal_p rather than pointer
15601 equality for mini-DSE code.
15603 2017-08-28 Nathan Sidwell <nathan@acm.org>
15605 * gcc.c (execute): Fold SIGPIPE handling into switch
15606 statement. Adjust internal error message.
15608 2017-08-28 Richard Biener <rguenther@suse.de>
15611 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
15612 Do nothing for removed DIEs.
15614 2017-08-28 Richard Biener <rguenther@suse.de>
15616 PR tree-optimization/81977
15617 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
15620 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
15623 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
15624 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
15625 using targetm.gen_mem_thread_fence.
15627 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
15630 * config/i386/i386.md (*<btsc><mode>): Change operand 2
15631 predicate to register_operand. Reorder operands.
15632 (*btr<mode>): Ditto.
15633 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
15634 (*btr<mode>_mask): Ditto.
15636 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
15638 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
15639 * config/rs6000/xmmintrin.h: New file.
15640 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
15642 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15645 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
15646 parameter and_insn and return it.
15647 (recombine_lvx_pattern): Insert a copy to ensure availability of
15648 the base register of the copied masking operation at the point of
15649 the instruction replacement.
15650 (recombine_stvx_pattern): Likewise.
15652 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
15654 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
15655 undocumented switches.
15656 (-mpower9-dform-vector): Likewise.
15657 (-mpower9-dform): Likewise.
15658 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
15659 comments to delete references to -mpower9-dform* switches.
15660 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
15661 Delete reference to -mpower9-dform* switches, test for
15662 -mpower9-vector instead.
15663 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
15664 (OTHER_P9_VECTOR_MASKS): Likewise.
15665 (POWERPC_MASKS): Likewise.
15666 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
15667 tests against -mpower9-dform* to -mpower9-vector. Delete code
15668 that checked for -mpower9-dform* consistancy with other options.
15669 Add test for -mpower9-misc to enable other power9 switches.
15670 (rs6000_init_hard_regno_mode_ok): Likewise.
15671 (rs6000_option_override_internal): Likewise.
15672 (rs6000_emit_prologue): Likewise.
15673 (rs6000_emit_epilogue): Likewise.
15674 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
15675 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
15676 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
15678 (emit_fusion_p9_store): Likewise.
15679 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
15680 resetting these macros if the assembler does not support ISA 3.0
15682 (TARGET_P9_DFORM_VECTOR): Likewise.
15683 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
15684 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
15686 2017-08-25 Alan Modra <amodra@gmail.com>
15688 PR rtl-optimization/81747
15689 * cse.c (cse_extended_basic_block): Don't attempt to record
15690 equivalences for degenerate conditional branches that branch
15691 to their fall-through.
15693 2017-08-24 Martin Sebor <msebor@redhat.com>
15695 PR middle-end/81908
15696 * gimple-fold.c (size_must_be_zero_p): New function.
15697 (gimple_fold_builtin_memory_op): Call it.
15699 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
15701 * config/rs6000/mm_malloc.h: New file.
15703 2017-08-24 Bin Cheng <bin.cheng@arm.com>
15705 PR tree-optimization/81913
15706 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
15707 analysis when either IVs in condition can wrap.
15709 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
15711 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
15712 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
15714 2017-08-24 Richard Biener <rguenther@suse.de>
15717 * targhooks.c (default_target_can_inline_p): Properly
15718 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
15719 is present and always compare.
15720 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
15721 infer -mfpmath=sse from TARGET_SSE_P.
15722 (ix86_can_inline_p): Properly use target_option_default_node when
15723 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
15725 2017-08-24 Richard Biener <rguenther@suse.de>
15728 * dwarf2out.c (output_die): Handle flag_generate_offload like
15730 (output_comp_unit): Likewise.
15731 (gen_array_type_die): Likewise.
15732 (dwarf2out_early_finish): Likewise.
15733 (note_variable_value_in_expr): Likewise.
15734 (dwarf2out_finish): Likewise. Adjust assert.
15735 * cgraphunit.c (symbol_table::compile): Move setting of
15736 flag_generate_offload earlier ...
15737 (symbol_table::finalize_compilation_unit): ... here, before
15738 early debug finalization.
15740 2017-08-24 Richard Biener <rguenther@suse.de>
15742 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
15743 and ipa-fnsummary.h.
15744 (ix86_can_inline_p): When ix86_fpmath flags do not match
15745 check whether the callee uses FP math at all.
15747 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
15749 PR middle-end/81931
15750 * tree-ssanames.c (get_nonzero_bits): Use element_precision
15751 instead of TYPE_PRECISION.
15753 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
15754 Alan Hayward <alan.hayward@arm.com>
15755 David Sherwood <david.sherwood@arm.com>
15757 * combine.c (make_extraction): Use subreg_offset_from_lsb.
15759 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
15761 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
15763 (ix86_frame::stack_realign_allocate): New field.
15764 (struct machine_frame_state): Modify comments.
15765 (machine_frame_state::sp_realigned_fp_end): New field.
15766 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
15767 layout calculation.
15768 (sp_valid_at): Add assertion to assure no attempt to access invalid
15769 offset of a realigned stack.
15770 (fp_valid_at): Likewise.
15771 (choose_baseaddr): Modify comments.
15772 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
15773 ix86_expand_prologue.
15774 (ix86_expand_prologue): Modify stack realignment and allocation.
15775 (ix86_expand_epilogue): Modify comments.
15776 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
15777 avx2_runtime, avx512f, and avx512f_runtime.
15779 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
15781 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
15782 (mstackrealign): Do not init to -1.
15783 * config/i386/i386.c (ix86_option_override_internal):
15784 Check opts_set, not opts when setting default value of
15785 opts->x_ix86_force_align_arg_pointer.
15787 2017-08-23 Richard Biener <rguenther@suse.de>
15789 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
15790 lang_hooks.decl_printable_name.
15791 * print-rtl-function.c (print_rtx_function): Likewise.
15792 * tree-pretty-print.c (dump_function_header): Likewise.
15794 2017-08-23 Richard Biener <rguenther@suse.de>
15797 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
15798 -g0 at compile-time.
15800 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15802 PR middle-end/19706
15803 * doc/sourcebuild.texi (Other hardware attributes):
15806 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15808 PR middle-end/19706
15809 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
15810 Fix single-use check.
15812 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
15814 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
15816 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
15818 * doc/install.texi: Modify to add more details on running selected
15821 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
15823 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
15824 is combined with -mabi=ms.
15825 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
15828 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15830 PR tree-optimization/81488
15831 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
15832 and cached_basis fields.
15833 (MAX_SPREAD): New constant.
15834 (alloc_cand_and_find_basis): Initialize new fields.
15835 (clear_visited): New function.
15836 (create_phi_basis_1): Rename from create_phi_basis, set visited
15837 and cached_basis fields.
15838 (create_phi_basis): New wrapper function.
15839 (phi_add_costs_1): Rename from phi_add_costs, add spread
15840 parameter, set visited field, short-circuit when limits reached.
15841 (phi_add_costs): New wrapper function.
15842 (record_phi_increments_1): Rename from record_phi_increments, set
15844 (record_phi_increments): New wrapper function.
15845 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
15846 (phi_incr_cost): New wrapper function.
15847 (all_phi_incrs_profitable_1): Rename from
15848 all_phi_incrs_profitable, set visited field.
15849 (all_phi_incrs_profitable): New wrapper function.
15851 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
15852 Alan Hayward <alan.hayward@arm.com>
15853 David Sherwood <david.sherwood@arm.com>
15855 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
15856 that takes the outer and inner modes.
15857 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
15858 comparison as the canonical test for a paradoxical subreg.
15859 * combine.c (simplify_set): Use paradoxical_subreg_p.
15860 (make_extraction): Likewise.
15861 (force_to_mode): Likewise.
15862 (rtx_equal_for_field_assignment_p): Likewise.
15863 (gen_lowpart_for_combine): Likewise.
15864 (simplify_comparison): Likewise.
15865 * cse.c (equiv_constant): Likewise.
15866 * expmed.c (store_bit_field_1): Likewise.
15867 * final.c (alter_subreg): Likewise.
15868 * fwprop.c (propagate_rtx): Likewise.
15869 (forward_propagate_subreg): Likewise.
15870 * ira-conflicts.c (ira_build_conflicts): Likewise.
15871 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
15872 * lra-constraints.c (curr_insn_transform): Likewise.
15873 (split_reg): Likewise.
15874 * lra-eliminations.c (move_plus_up): Likewise.
15875 (lra_eliminate_regs_1): Likewise.
15876 * recog.c (general_operand): Likewise.
15877 * ree.c (combine_reaching_defs): Likewise.
15878 * reload.c (push_reload): Likewise.
15879 (find_reloads): Likewise.
15880 * reload1.c (elimination_effects): Likewise.
15881 (compute_reload_subreg_offset): Likewise.
15882 (choose_reload_regs): Likewise.
15883 * rtlanal.c (subreg_lsb_1): Likewise.
15884 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15885 (simplify_subreg): Likewise.
15886 * var-tracking.c (track_loc_p): Likewise.
15887 * emit-rtl.c (byte_lowpart_offset): Likewise.
15888 (paradoxical_subreg_p): Delete out-of-line definition.
15890 2017-08-22 Jeff Law <law@redhat.com>
15892 PR tree-optimization/81741
15893 PR tree-optimization/71947
15894 * tree-ssa-dom.c: Include tree-inline.h.
15895 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
15896 equivalences if one is more expensive to compute than the other.
15897 * tree-ssa-scopedtables.h (class const_or_copies): Make
15898 record_const_or_copy_raw method private.
15899 (class avail_exprs_stack): New method simplify_binary_operation.
15900 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
15901 avail_exprs_stack::simplify_binary_operation as needed.
15902 (avail_exprs_stack::simplify_binary_operation): New function.
15904 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
15906 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
15907 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
15908 (DOT_SYMBOLS): Likewise.
15909 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
15910 (RELOCATABLE_NEEDS_FIXUP): Likewise.
15911 (RS6000_ABI_NAME): Likewise.
15912 (TARGET_CMODEL): Likewise.
15913 (TOC_SECTION_ASM_OP): Likewise.
15914 (SET_CMODEL): New macro.
15915 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
15917 2017-08-22 Richard Biener <rguenther@suse.de>
15919 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
15920 to allow for free-lang-data replacements similar to verify_type_variant.
15922 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
15923 Alan Hayward <alan.hayward@arm.com>
15924 David Sherwood <david.sherwood@arm.com>
15926 * config/aarch64/aarch64.md (casesi): Use DImode rather than
15927 VOIDmode for the LABEL_REF.
15929 2017-08-22 Richard Biener <rguenther@suse.de>
15931 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
15933 2017-08-22 Richard Biener <rguenther@suse.de>
15935 * common.opt (feliminate-dwarf2-dups): Ignore.
15936 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
15937 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
15938 same_die_p_wrap, compute_section_prefix,
15939 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
15940 (comdat_symbol_id, comdat_symbol_number): Likewise.
15941 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
15943 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
15944 (output_die): Mark unreachable path unreachable.
15945 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
15946 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
15947 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
15948 (dwarf2out_early_finish): Likewise.
15950 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
15952 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
15954 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
15957 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
15958 not VAR_P. Filter attribute warnings with OPT_Wattributes.
15959 (avr_attribute_table) <io, io_low, address>: Initialize
15960 .decl_required with true.
15962 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
15964 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
15965 undocumented debugging options.
15966 (-mvsx-scalar-double): Likewise.
15967 (-mallow-df-permute): Likewise.
15968 (-mvectorize-builtins): Likewise.
15969 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15970 (rs6000_builtin_vectorized_function): Likewise.
15971 (rs6000_builtin_md_vectorized_function): Likewise.
15972 (rs6000_opt_vars): Likewise.
15974 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
15977 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
15978 (*btrq_imm): Rename from *btrq.
15979 (*btcq_imm): Rename from *btcq.
15980 (btsc): New code attribute.
15981 (*<btsc><mode>): New insn pattern.
15982 (*btr<mode>): Ditto.
15983 (*<btsc><mode>_mask): New insn_and_split pattern.
15984 (*btr<mode>_mask): Ditto.
15986 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15987 Alan Hayward <alan.hayward@arm.com>
15988 David Sherwood <david.sherwood@arm.com>
15990 * function.c (pad_below): Simplify padding calculation.
15992 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15993 Alan Hayward <alan.hayward@arm.com>
15994 David Sherwood <david.sherwood@arm.com>
15996 * target.def (function_prologue): Remove frame size argument.
15997 (function_epilogue): Likewise.
15998 * doc/tm.texi: Regenerate.
15999 * final.c (final_start_function): Update call to function_prologue.
16000 (final_end_function): Update call to function_epilogue.
16001 (default_function_pro_epilogue): Remove frame size argument.
16002 * output.h (default_function_pro_epilogue): Likewise.
16003 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
16004 (arm_output_function_prologue): Likewise.
16005 * config/frv/frv.c (frv_function_prologue): Likewise.
16006 (frv_function_epilogue): Likewise.
16007 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
16008 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
16009 (ia64_output_function_epilogue): Likewise.
16010 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
16011 (m32r_output_function_epilogue): Likewise.
16012 * config/microblaze/microblaze.c (microblaze_function_prologue)
16013 (microblaze_function_epilogue): Likewise.
16014 * config/mips/mips.c (mips_output_function_prologue): Likewise.
16015 (mips_output_function_epilogue): Likewise.
16016 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
16017 (mmix_target_asm_function_epilogue): Likewise.
16018 * config/msp430/msp430.c (msp430_start_function): Likewise.
16019 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
16020 (nds32_asm_function_epilogue): Likewise.
16021 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
16022 * config/pa/pa.c (pa_output_function_prologue): Likewise.
16023 (pa_output_function_epilogue): Likewise.
16024 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
16025 (rs6000_output_function_epilogue): Likewise.
16026 * config/rl78/rl78.c (rl78_start_function): Likewise.
16027 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
16028 (rs6000_output_function_epilogue): Likewise.
16029 * config/rx/rx.c (rx_output_function_prologue): Likewise.
16030 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
16031 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
16032 (sparc_asm_function_epilogue): Likewise.
16034 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16036 * tree.h (type_has_mode_precision_p): New function.
16037 * convert.c (convert_to_integer_1): Use it.
16038 * expr.c (expand_expr_real_2): Likewise.
16039 (expand_expr_real_1): Likewise.
16040 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
16041 * match.pd: Likewise.
16042 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
16043 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
16044 * tree-tailcall.c (process_assignment): Likewise.
16045 * tree-vect-loop.c (vectorizable_reduction): Likewise.
16046 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
16047 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
16048 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16049 (vectorizable_assignment): Likewise.
16050 (vectorizable_shift): Likewise.
16051 (vectorizable_operation): Likewise.
16052 * tree-vrp.c (register_edge_assert_for_2): Likewise.
16054 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
16056 * match.pd: Add pow (C, x) simplification.
16058 2017-08-21 Richard Biener <rguenther@suse.de>
16060 PR tree-optimization/81900
16061 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
16062 for blocks with abnormal predecessors.
16063 (compute_antic): Do not set visited flag prematurely.
16065 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
16068 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
16070 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16072 * stor-layout.h (vector_type_mode): Move to...
16073 * tree.h (vector_type_mode): ...here.
16074 * stor-layout.c (vector_type_mode): Move to...
16075 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
16077 2017-08-21 Richard Biener <rguenther@suse.de>
16079 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
16080 register_external_die hooks.
16081 (debug_false_tree_charstarstar_uhwistar): Declare.
16082 (debug_nothing_tree_charstar_uhwi): Likewise.
16083 * debug.c (do_nothing_debug_hooks): Adjust.
16084 (debug_false_tree_charstarstar_uhwistar): New do nothing.
16085 (debug_nothing_tree_charstar_uhwi): Likewise.
16086 * dbxout.c (dbx_debug_hooks): Adjust.
16087 (xcoff_debug_hooks): Likewise.
16088 * sdbout.c (sdb_debug_hooks): Likewise.
16089 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16090 * dwarf2out.c (macinfo_label_base): New global.
16091 (dwarf2out_register_external_die): New function for the
16092 register_external_die hook.
16093 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
16094 (dwarf2_debug_hooks): Use them.
16095 (dwarf2_lineno_debug_hooks): Adjust.
16096 (struct die_struct): Add with_offset flag.
16097 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
16098 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
16099 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
16100 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
16101 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
16102 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
16103 defining section names for the early LTO debug variants.
16104 (reset_indirect_string): New helper.
16105 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
16106 (print_dw_val): Add support for offsetted symbol references.
16107 (get_ultimate_context): Split out from is_cxx.
16108 (is_cxx): Use get_ultimate_context.
16109 (is_fortran): Add decl overload.
16110 (compute_comp_unit_symbol): Split out worker from
16111 compute_section_prefix.
16112 (compute_section_prefix): Call compute_comp_unit_symbol and
16113 set comdat_type_p here.
16114 (output_die): Skip DIE symbol output for the LTO added one.
16115 Handle DIE symbol references with offset.
16116 (output_comp_unit): Guard section name mangling properly.
16117 For LTO debug sections emit a symbol at the section beginning
16118 which we use to refer to its DIEs.
16119 (add_abstract_origin_attribute): For DIEs registered via
16120 dwarf2out_register_external_die directly refer to the early
16121 DIE rather than indirectly through the shadow one we created.
16122 Remove obsolete call to dwarf2out_abstract_function for
16123 non-function/block origins.
16124 (gen_array_type_die): When generating early LTO debug do
16125 not emit DW_AT_string_length.
16126 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
16127 late when in LTO. As suggested place a gcc_unreachable for
16128 the DECL_ABSTRACT_P case.
16129 (gen_subprogram_die): Avoid another specification DIE
16130 for early built declarations/definitions for the late LTO case.
16131 (gen_variable_die): Add type references for late duplicated VLA dies
16133 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
16134 we have the abstract instance already.
16135 (process_scope_var): Adjust decl DIE contexts in LTO which
16136 first puts them in limbo.
16137 (gen_decl_die): Do not generate type DIEs late apart from
16138 types for VLAs or for decls we do not yet have a DIE. Do not
16139 call dwarf2out_abstract_function late.
16140 (dwarf2out_early_global_decl): Make sure to create DIEs
16141 for abstract instances of a decl first.
16142 (dwarf2out_late_global_decl): Adjust comment.
16143 (output_macinfo_op): With multiple macro sections use
16144 macinfo_label_base to distinguish labels.
16145 (output_macinfo): Likewise. Update macinfo_label_base.
16146 Pass in the line info label.
16147 (note_variable_value_in_expr): When generating LTO resolve
16148 all variable values here by generating DIEs as needed.
16149 (init_sections_and_labels): Add early LTO debug flag parameter
16150 and generate different sections and names if set. Add generation
16151 counter for the labels so we can have multiple of them.
16152 (reset_dies): Helper to allow DIEs to be output multiple times.
16153 (dwarf2out_finish): When outputting DIEs to the fat part of an
16154 LTO object first reset DIEs.
16155 (dwarf2out_early_finish): Output early DIEs when generating LTO.
16156 (modified_type_die): Check for decl_ultimate_origin being self
16158 (gen_type_die_with_usage): Likewise.
16159 (gen_typedef_die): Allow decl_ultimate_origin being self.
16160 (set_decl_abstract_flags): Remove.
16161 (set_block_abstract_flags): Likewise.
16162 (dwarf2out_abstract_function): Treat the early generated DIEs
16163 as the abstract copy and only add DW_AT_inline and
16164 DW_AT_artificial here and call set_decl_origin_self.
16165 If the DIE has an abstract origin don't do anything.
16166 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
16167 if we have none yet (Go fails to build one, PR78628).
16168 (variably_modified_type_p): Prevent endless recursion for Ada
16169 cyclic pointer types.
16170 * lto-streamer-in.c: Include debug.h.
16171 (dref_queue): New global.
16172 (lto_read_tree_1): Stream in DIE references.
16173 (lto_input_tree): Register DIE references.
16174 (input_function): Stream DECL_DEBUG_ARGS.
16175 * lto-streamer-out.c: Include debug.h.
16176 (lto_write_tree_1): Output DIE references.
16177 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
16178 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
16179 (output_function): Stream DECL_DEBUG_ARGS.
16180 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
16181 Stream DECL_ABSTRACT_ORIGIN.
16182 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
16183 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
16184 DECL_CONTEXT for file-scope decls.
16185 * lto-streamer.h (struct dref_entry): Declare.
16186 (dref_queue): Likewise.
16187 * cfgexpand.c (pass_expand::execute): Do not call the
16188 outlining_inline_function hook here.
16189 * lto-wrapper.c (debug_obj): New global.
16190 (tool_cleanup): Unlink it if required.
16191 (debug_objcopy): New function.
16192 (run_gcc): Handle early debug sections in the IL files by
16193 extracting them to separate files, partially linkin them and
16194 feeding the result back as result to the linker.
16195 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
16196 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
16197 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
16198 sections into a separate segment.
16199 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
16201 (darwin_asm_dwarf_section): Likewise.
16202 (darwin_asm_output_dwarf_offset): Likewise.
16203 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
16205 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16206 Alan Hayward <alan.hayward@arm.com>
16207 David Sherwood <david.sherwood@arm.com>
16209 * read-md.h (md_reader::record_potential_iterator_use): Replace
16210 pointer argument with an rtx and an index.
16211 * read-rtl.c (iterator_group::apply_iterator): Likewise.
16212 (apply_mode_iterator): Likewise.
16213 (apply_code_iterator): Likewise.
16214 (apply_int_iterator): Likewise.
16215 (apply_subst_iterator): Likewise.
16216 (record_iterator_use): Likewise.
16217 (record_attribute_use): Likewise.
16218 (md_reader::record_potential_iterator_use): Likewise. Update calls
16219 to record_iterator_use and apply_iterator.
16220 (iterator_use): Replace ptr with x and index.
16221 (attribute_use): Likewise.
16222 (apply_attribute_uses): Update calls to apply_iterator.
16223 (apply_iterators): Likewise. Update initialization of iterator_use.
16224 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
16225 and record_potential_iterator_use.
16226 (rtx_reader::read_rtx_operand): Likewise.
16228 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16229 Alan Hayward <alan.hayward@arm.com>
16230 David Sherwood <david.sherwood@arm.com>
16232 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
16235 2017-08-21 Richard Biener <rguenther@suse.de>
16237 PR middle-end/81884
16238 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
16239 at struct end conservatively when comparing common bases.
16241 2017-08-21 Richard Biener <rguenther@suse.de>
16243 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
16244 (mem_ref_in_stmt): Remove.
16245 (determine_max_movement): Use ref index to get at the reference.
16246 (invariantness_dom_walker::before_dom_children): Deal with
16247 lim data already initialized.
16248 (gather_mem_refs_stmt): Initialize lim data and record ref index.
16250 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16252 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
16253 (TARGET_ISA_ROUND): Ditto.
16254 (TARGET_ROUND): Ditto.
16255 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
16256 * config/i386/i386.md: Ditto.
16257 * config/i386/sse.md: Ditto.
16258 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
16259 with OPTION_MASK_ISA_SSE4_1.
16261 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16264 * doc/extend.texi (x86 Built-in Functions): Correct the name of
16265 __builtin_ia32_lzcnt_u16.
16267 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
16270 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
16271 (rs6000_set_current_function): Rewrite function to use it.
16273 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
16276 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
16277 and TYPE_WARN_IF_NOT_ALIGN.
16278 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
16279 (handle_warn_if_not_align): New.
16280 (place_union_field): Call handle_warn_if_not_align.
16281 (place_field): Call handle_warn_if_not_align.
16282 Copy TYPE_WARN_IF_NOT_ALIGN.
16283 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
16284 (layout_type): Likewise.
16285 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
16287 (tree_decl_common): Add warn_if_not_align.
16288 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
16289 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
16290 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
16291 (DECL_WARN_IF_NOT_ALIGN): Likewise.
16292 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
16293 * doc/extend.texi: Document warn_if_not_aligned attribute.
16294 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
16296 2017-08-17 Martin Liska <mliska@suse.cz>
16299 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
16300 (get_data_dependence): Use it as pointer type.
16301 (distribute_loop): Likewise.
16303 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16305 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
16306 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
16307 (p8_vmrgow_<mode>_direct): New define_insn.
16308 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
16309 handle endianness for vmrgew and vmrgow permute patterns.
16311 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16313 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
16314 * config/rs6000/rs6000-cpus.def: Remove comment.
16315 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
16316 (POWERPC_MASKS): Likewise.
16317 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
16318 use of TARGET_VSX_TIMODE.
16319 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
16320 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
16321 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
16322 (rs6000_option_override_internal): Remove dead code.
16323 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
16324 (rs6000_legitimize_reload_address): Likewise.
16325 (rs6000_legitimate_address_p): Likewise.
16326 (rs6000_opt_masks): Delete "vsx-timode".
16327 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
16328 from function comment.
16329 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
16330 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
16331 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
16333 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
16334 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
16335 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
16336 (VSX_TI): Likewise.
16338 (define_peephole2): Likewise.
16340 2017-08-17 Martin Sebor <msebor@redhat.com>
16343 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
16344 past the end of an array.
16345 (test_pp_format): Add test cases.
16347 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
16349 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
16350 missing ECF_NOTHROW flags.
16352 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16355 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
16356 operands residing in integer registers.
16357 (*vsx_le_perm_load_<mode>): Likewise.
16358 (*vsx_le_perm_store_<mode>): Likewise.
16359 (define_peephole2): Add peepholes to optimize the above.
16361 2017-08-17 Marek Polacek <polacek@redhat.com>
16363 PR middle-end/81814
16364 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
16365 to mimic what shorten_compare did. Change the return type to bool.
16366 (fold_cond_expr_with_comparison): Update call to
16367 operand_equal_for_comparison_p.
16368 (fold_ternary_loc): Likewise.
16370 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
16372 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
16374 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
16376 2017-08-17 Richard Biener <rguenther@suse.de>
16378 * tree-ssa-structalias.c (solve_graph): When propagating
16379 to successors update the graphs succ edges and avoid duplicate work.
16381 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
16384 * config/i386/i386.c (ix86_option_override_internal): Save target
16385 specific options after ix86_stack_protector_guard_reg was changed.
16387 2017-08-17 Richard Biener <rguenther@suse.de>
16389 PR tree-optimization/81827
16390 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
16391 (new_var_info): Initialize it conservatively.
16392 (get_call_vi): Mark register vars.
16393 (new_scalar_tmp_constraint_exp): Likewise.
16394 (handle_rhs_call): Likewise.
16395 (handle_const_call): Likewise.
16396 (create_function_info_for): Likewise.
16397 (solve_constraints): Sort varinfos to separate register from
16398 non-register vars to pack points-to solution bitmaps during
16401 2017-08-17 Marek Polacek <polacek@redhat.com>
16403 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
16405 2017-08-17 Richard Biener <rguenther@suse.de>
16407 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
16408 to true when overflow is undefined and we saturated the result.
16410 2017-08-17 Alan Modra <amodra@gmail.com>
16413 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
16414 Don't use store multiple if only one reg needs saving.
16415 (interesting_frame_related_regno): New function.
16416 (rs6000_frame_related): Don't emit frame info for regs that
16418 (rs6000_emit_epilogue): Likewise.
16420 2017-08-16 Nathan Sidwell <nathan@acm.org>
16422 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
16423 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
16424 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
16425 * tree.c (free_lang_data_in_type): Use else-if chain. Always
16426 clear TYPE_LANG_1. Remove obsolete member-function stripping.
16427 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
16428 (verify_type): Adjust for TYPE_BINFO move.
16429 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
16430 process TYPE_BINFO directly.
16431 (hash_tree): Likewise.
16432 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16434 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16437 2017-08-16 David Malcolm <dmalcolm@redhat.com>
16439 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
16441 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
16444 * config/i386/i386.md (*anddi_1_btr): Change predicates of
16445 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
16446 Add ix86_binary_operator_ok to insn constraint.
16447 (*iordi_1_bts): Ditto.
16448 (*xordi_1_btc): Ditto.
16449 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
16450 Update corresponding peephole2 pattern.
16454 2017-08-16 Bin Cheng <bin.cheng@arm.com>
16456 PR tree-optimization/81832
16457 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
16458 copy loop header which has IFN_LOOP_DIST_ALIAS call.
16460 2017-08-16 Marek Polacek <polacek@redhat.com>
16463 * fold-const.c (fold_indirect_ref_1): Restore original behavior
16464 regarding size_zero_node.
16466 2017-08-16 Martin Liska <mliska@suse.cz>
16469 * config.gcc: Respect previously set extra_objs in case
16472 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
16474 PR tree-optimization/81835
16475 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
16476 the phi SSA_NAME. Check that the condition in a COND_EXPR does
16477 not depend on the phi.
16479 2017-08-16 Alan Modra <amodra@gmail.com>
16481 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
16484 2017-08-16 Alan Modra <amodra@gmail.com>
16486 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
16487 (save_reg_p): ..into this. Update all callers.
16488 (first_reg_to_save): Simplify.
16490 2017-08-16 Alan Modra <amodra@gmail.com>
16492 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
16495 2017-08-15 Joseph Myers <joseph@codesourcery.com>
16499 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
16500 constant count if that count is less than 32.
16502 2017-08-15 Nathan Sidwell <nathan@acm.org>
16504 * gcc.c (execute): Emit friendlier message if inferior is killed
16505 by an external cause.
16507 2017-08-15 Richard Biener <rguenther@suse.de>
16509 PR tree-optimization/81790
16510 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
16511 CONSTRUCTORs from simplifying and VN.
16513 2017-08-14 Martin Sebor <msebor@redhat.com>
16515 * builtin-attrs.def: Add comments.
16517 2017-08-14 Martin Sebor <msebor@redhat.com>
16520 * doc/extend.texi (attribute nonstring): Document new attribute.
16522 2017-08-14 Martin Sebor <msebor@redhat.com>
16525 * tree-diagnostic.c (default_tree_printer): Handle %G.
16526 * gimple-pretty-print.h (percent_G_format): Declare new function.
16527 * gimple-pretty-print.c (percent_G_format): Define.
16528 * tree-pretty-print.c (percent_K_format): Add argument.
16530 2017-08-14 Martin Sebor <msebor@redhat.com>
16532 PR translation/79998
16533 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
16534 Remove a stray space.
16536 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
16539 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
16540 (*iordi_1_bts): Ditto.
16541 (*xordi_1_btc): Ditto.
16543 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16546 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
16547 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16549 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
16550 quoted strings, and make more translator-friendly.
16551 (darwin_rs6000_override_options): Likewise.
16552 (rs6000_option_override_internal): Likewise.
16553 (rs6000_return_in_memory): Fix overlong line.
16554 (init_cmulative_args): Use quoted strings, and make more
16555 translator-friendly.
16556 (rs6000_pass_by_reference): Fix overlong line.
16557 (def_builtin): Use quoted strings.
16558 (altivec_expand_predicate_builtin): Use quoted strings, and make
16559 more translator-friendly.
16560 (htm_expand_builtin): Use quoted strings.
16561 (cpu_expand_builtin): Use quoted strings, and make more
16562 translator-friendly.
16563 (altivec_expand_builtin): Likewise.
16564 (paired_expand_predicate_builtin): Likewise.
16565 (rs6000_invalid_builtin): Likewise.
16566 (builtin_function_type): Use quoted strings.
16567 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
16568 more translator-friendly.
16569 (rs6000_trampoline_init): Likewise.
16570 (rs6000_handle_altivec_attribute): Likewise.
16571 (rs6000_inner_target_options): Use quoted strings.
16572 (rs6000_disable_incompatible_switches): Likewise.
16573 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
16574 strings, and make more translator-friendly.
16575 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
16577 2017-08-14 Bin Cheng <bin.cheng@arm.com>
16579 PR tree-optimization/81799
16580 * tree-loop-distribution.c (version_loop_by_alias_check): Force
16581 cond_expr to simple gimple operand.
16583 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
16585 PR middle-end/46932
16586 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
16588 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
16592 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
16593 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
16594 TARGET_GASISR_PROLOGUES.
16595 * config/avr/avr.c (avr_option_override): Same.
16596 (avr_pass_pre_proep::execute): Same.
16598 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
16601 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
16602 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
16604 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
16606 * config/i386/i386.md (*load_tp_<mode>): Redefine as
16607 define_insn_and_split. Split to a memory load from 0 in
16608 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
16609 using PTR mode iterator.
16610 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
16611 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
16612 (*add_tp_<mode>): Redefine as define_insn_and_split.
16613 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
16614 address space. Merge with *add_tp_x32 using PTR mode iterator.
16615 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
16616 Split to an add with a memory load from 0 in
16617 DEFAULT_TLS_SEG_REG address space.
16619 2017-08-12 Andrew Pinski <apinski@cavium.com>
16621 * config/aarch64/aarch64-option-extensions.def (rdma):
16622 Fix feature string to what Linux prints out in /proc/cpuinfo.
16624 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
16627 * dwarf2out.c (modified_type_die): For C typedef types that have
16628 an ultimate origin, process the ultimate origin instead of the
16630 (gen_typedef_die): Assert that input DECLs have no ultimate
16632 (gen_type_die_with_usage): For typedef variants that have an
16633 ultimate origin, just call gen_decl_die on the original DECL.
16634 (process_scope_var): Avoid creating DIEs for local typedefs and
16635 concrete static variables.
16637 2017-08-12 Alan Modra <amodra@gmail.com>
16641 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
16642 match gnu-user.h startfile.
16643 (ENDFILE_LINUX_SPEC): Similarly.
16645 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
16648 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
16650 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
16652 2017-08-11 Tamar Christina <tamar.christina@arm.com>
16653 * config/aarch64/aarch64.md (mov<mode>): Change.
16654 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16655 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
16656 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
16658 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
16660 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
16661 for storage order barriers.
16663 2017-08-11 Martin Liska <mliska@suse.cz>
16666 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
16667 variables of void type.
16669 2017-08-11 Martin Liska <mliska@suse.cz>
16671 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
16672 TARGET_SUPPORTS_ALIASES.
16673 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
16674 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
16675 (optimize_weakref): Likewise.
16676 * symtab.c (symtab_node::noninterposable_alias): Likewise.
16677 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
16678 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
16680 2017-08-11 Martin Liska <mliska@suse.cz>
16683 * config/i386/i386.c (make_resolver_func): Do complete
16684 refactoring of the function.
16686 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16689 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
16690 * config/i386/i386.c (ix86_stack_protect_guard): Use
16691 ix86_stack_protect_guard_symbol_str to generate varible declaration.
16692 * doc/invoke.texi (x86 Options): Document
16693 -mstack-protector-guard-symbol= option.
16695 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16697 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
16698 * config/i386/i386.c (ix86_split_stack_guard): New function.
16699 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
16700 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
16701 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
16702 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
16703 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
16704 (split_stack_space_check): Call ix86_split_stack_guard.
16706 2017-08-10 Martin Sebor <msebor@redhat.com>
16708 * print-tree.c (print_node): Print location using the established
16710 Replace spaces with colons.
16711 (debug_raw, debug): Ditto.
16713 2017-08-10 Martin Sebor <msebor@redhat.com>
16716 * pretty-print.c (pp_format): Correct the handling of %s precision.
16718 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
16721 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
16723 (ix86_finalize_stack_frame_flags): This. Also clear
16724 frame_pointer_needed if -fno-omit-frame-pointer is used without
16726 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
16727 with ix86_finalize_stack_frame_flags.
16728 (ix86_expand_epilogue): Likewise.
16729 (ix86_expand_split_stack_prologue): Likewise.
16730 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
16732 2017-08-10 Martin Liska <mliska@suse.cz>
16735 * c-attribs.c (handle_target_attribute):
16736 Report warning for an empty string argument of target attribute.
16738 2017-08-09 Jakub Jelinek <jakub@redhat.com>
16741 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
16743 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
16744 or DECL_NONLOCAL labels.
16745 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
16746 or DECL_NONLOCAL labels here.
16748 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
16750 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
16751 to indicate when early gimple folding has been disabled.
16752 (rs6000_gimple_fold_builtin): Add debug content.
16753 (rs6000_invalid_builtin): Fix whitespace.
16754 (rs6000_expand_builtin): Fix whitespace.
16755 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
16757 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
16760 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
16761 SAVE_MULTIPLE if not all the registers that saves, should be saved.
16763 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
16765 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
16766 (qdf24xx): Likewise.
16767 * config/aarch64/aarch64.md: Include falkor.md.
16768 * config/aarch64/falkor.md: New.
16770 2017-08-09 Marek Polacek <polacek@redhat.com>
16773 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
16774 * diagnostic.c (emit_diagnostic): Add a comment.
16775 (emit_diagnostic_valist): New function.
16777 2017-08-09 Marek Polacek <polacek@redhat.com>
16780 * input.c (make_location): New overload.
16781 * input.h (make_location): Declare.
16783 2017-08-08 Alan Modra <amodra@gmail.com>
16784 H.J. Lu <hongjiu.lu@intel.com>
16787 * gcc.c (NO_PIE_SPEC): Delete.
16788 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
16790 (LINK_PIE_SPEC): ..to here.
16791 (LINK_COMMAND_SPEC): Support -no-pie.
16792 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
16793 chain of crtbegin*.o selection, update for PIE_SPEC changes and
16795 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
16796 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
16797 (ENDFILE_CRTEND_SPEC): Similarly.
16799 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
16802 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
16803 (mstack-protector-guard-offset=): Ditto.
16804 * config/i386/i386.c (ix86_option_override): Handle
16805 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
16807 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
16808 ix86_stack_protect_guard_offset variables.
16809 (TARGET_STACK_PROTECT_GUARD): Always define.
16810 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
16811 and -mstack-protector-guard-offset= options.
16813 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16815 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
16816 boundary case for the last candidate.
16818 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16820 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
16821 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
16823 2017-08-08 Tamar Christina <tamar.christina@arm.com>
16825 PR middle-end/19706
16826 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
16827 * config/aarch64/aarch64-builtins.c
16828 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
16829 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
16830 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
16832 2017-08-08 Tamar Christina <tamar.christina@arm.com>
16833 Andrew Pinski <pinskia@gmail.com>
16835 PR middle-end/19706
16836 * internal-fn.def (XORSIGN): New.
16837 * optabs.def (xorsign_optab): New.
16838 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
16839 (convert_expand_mult_copysign): New.
16840 (pass_optimize_widening_mul::execute): Call
16841 convert_expand_mult_copysign.
16843 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16845 PR tree-optimization/81354
16846 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
16847 Insert on edges rather than explicitly creating landing pads.
16848 (analyze_candidates_and_replace): Commit edge inserts.
16850 2017-08-08 Richard Biener <rguenther@suse.de>
16852 PR middle-end/81719
16853 * tree-ssa-loop-niter.c: Include tree-dfa.h.
16854 (expand_simple_operations): Also look through ADDR_EXPRs with
16855 MEM_REF bases treating them as POINTER_PLUS_EXPR.
16857 2017-08-08 Richard Biener <rguenther@suse.de>
16859 PR tree-optimization/81723
16860 * tree-vect-slp.c (struct bst_traits): New hash traits.
16861 (bst_fail): New global.
16862 (vect_build_slp_tree_2): New worker, split out from ...
16863 (vect_build_slp_tree): ... this now wrapping it with using
16864 bst_fail set to cache SLP tree build fails. Properly handle
16866 (vect_analyze_slp_instance): Allocate and free bst_fail.
16868 2017-08-08 Martin Liska <mliska@suse.cz>
16871 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
16872 LABEL_DECLs that can be from a different function.
16874 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16876 PR tree-optimization/81744
16877 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
16878 loop's number of iterations.
16880 2017-08-08 Martin Liska <mliska@suse.cz>
16882 * asan.c: Include header files.
16883 * attribs.c (build_decl_attribute_variant): New function moved
16885 (build_type_attribute_qual_variant): Likewise.
16886 (cmp_attrib_identifiers): Likewise.
16887 (simple_cst_list_equal): Likewise.
16888 (omp_declare_simd_clauses_equal): Likewise.
16889 (attribute_value_equal): Likewise.
16890 (comp_type_attributes): Likewise.
16891 (build_type_attribute_variant): Likewise.
16892 (lookup_ident_attribute): Likewise.
16893 (remove_attribute): Likewise.
16894 (merge_attributes): Likewise.
16895 (merge_type_attributes): Likewise.
16896 (merge_decl_attributes): Likewise.
16897 (merge_dllimport_decl_attributes): Likewise.
16898 (handle_dll_attribute): Likewise.
16899 (attribute_list_equal): Likewise.
16900 (attribute_list_contained): Likewise.
16901 * attribs.h (lookup_attribute): New function moved from tree.[ch].
16902 (lookup_attribute_by_prefix): Likewise.
16903 * bb-reorder.c: Include header files.
16904 * builtins.c: Likewise.
16905 * calls.c: Likewise.
16906 * cfgexpand.c: Likewise.
16907 * cgraph.c: Likewise.
16908 * cgraphunit.c: Likewise.
16909 * convert.c: Likewise.
16910 * dwarf2out.c: Likewise.
16911 * final.c: Likewise.
16912 * fold-const.c: Likewise.
16913 * function.c: Likewise.
16914 * gimple-expr.c: Likewise.
16915 * gimple-fold.c: Likewise.
16916 * gimple-pretty-print.c: Likewise.
16917 * gimple.c: Likewise.
16918 * gimplify.c: Likewise.
16919 * hsa-common.c: Likewise.
16920 * hsa-gen.c: Likewise.
16921 * internal-fn.c: Likewise.
16922 * ipa-chkp.c: Likewise.
16923 * ipa-cp.c: Likewise.
16924 * ipa-devirt.c: Likewise.
16925 * ipa-fnsummary.c: Likewise.
16926 * ipa-inline.c: Likewise.
16927 * ipa-visibility.c: Likewise.
16929 * lto-cgraph.c: Likewise.
16930 * omp-expand.c: Likewise.
16931 * omp-general.c: Likewise.
16932 * omp-low.c: Likewise.
16933 * omp-offload.c: Likewise.
16934 * omp-simd-clone.c: Likewise.
16935 * opts-global.c: Likewise.
16936 * passes.c: Likewise.
16937 * predict.c: Likewise.
16938 * sancov.c: Likewise.
16939 * sanopt.c: Likewise.
16940 * symtab.c: Likewise.
16941 * toplev.c: Likewise.
16942 * trans-mem.c: Likewise.
16943 * tree-chkp.c: Likewise.
16944 * tree-eh.c: Likewise.
16945 * tree-into-ssa.c: Likewise.
16946 * tree-object-size.c: Likewise.
16947 * tree-parloops.c: Likewise.
16948 * tree-profile.c: Likewise.
16949 * tree-ssa-ccp.c: Likewise.
16950 * tree-ssa-live.c: Likewise.
16951 * tree-ssa-loop.c: Likewise.
16952 * tree-ssa-sccvn.c: Likewise.
16953 * tree-ssa-structalias.c: Likewise.
16954 * tree-ssa.c: Likewise.
16955 * tree-streamer-in.c: Likewise.
16956 * tree-vectorizer.c: Likewise.
16957 * tree-vrp.c: Likewise.
16958 * tsan.c: Likewise.
16959 * ubsan.c: Likewise.
16960 * varasm.c: Likewise.
16961 * varpool.c: Likewise.
16962 * tree.c: Remove functions moved to attribs.[ch].
16963 * tree.h: Likewise.
16964 * config/aarch64/aarch64.c: Add attrs.h header file.
16965 * config/alpha/alpha.c: Likewise.
16966 * config/arc/arc.c: Likewise.
16967 * config/arm/arm.c: Likewise.
16968 * config/avr/avr.c: Likewise.
16969 * config/bfin/bfin.c: Likewise.
16970 * config/c6x/c6x.c: Likewise.
16971 * config/cr16/cr16.c: Likewise.
16972 * config/cris/cris.c: Likewise.
16973 * config/darwin.c: Likewise.
16974 * config/epiphany/epiphany.c: Likewise.
16975 * config/fr30/fr30.c: Likewise.
16976 * config/frv/frv.c: Likewise.
16977 * config/ft32/ft32.c: Likewise.
16978 * config/h8300/h8300.c: Likewise.
16979 * config/i386/winnt.c: Likewise.
16980 * config/ia64/ia64.c: Likewise.
16981 * config/iq2000/iq2000.c: Likewise.
16982 * config/lm32/lm32.c: Likewise.
16983 * config/m32c/m32c.c: Likewise.
16984 * config/m32r/m32r.c: Likewise.
16985 * config/m68k/m68k.c: Likewise.
16986 * config/mcore/mcore.c: Likewise.
16987 * config/microblaze/microblaze.c: Likewise.
16988 * config/mips/mips.c: Likewise.
16989 * config/mmix/mmix.c: Likewise.
16990 * config/mn10300/mn10300.c: Likewise.
16991 * config/moxie/moxie.c: Likewise.
16992 * config/msp430/msp430.c: Likewise.
16993 * config/nds32/nds32-isr.c: Likewise.
16994 * config/nds32/nds32.c: Likewise.
16995 * config/nios2/nios2.c: Likewise.
16996 * config/nvptx/nvptx.c: Likewise.
16997 * config/pa/pa.c: Likewise.
16998 * config/pdp11/pdp11.c: Likewise.
16999 * config/powerpcspe/powerpcspe.c: Likewise.
17000 * config/riscv/riscv.c: Likewise.
17001 * config/rl78/rl78.c: Likewise.
17002 * config/rx/rx.c: Likewise.
17003 * config/s390/s390.c: Likewise.
17004 * config/sh/sh.c: Likewise.
17005 * config/sol2.c: Likewise.
17006 * config/sparc/sparc.c: Likewise.
17007 * config/spu/spu.c: Likewise.
17008 * config/stormy16/stormy16.c: Likewise.
17009 * config/tilegx/tilegx.c: Likewise.
17010 * config/tilepro/tilepro.c: Likewise.
17011 * config/v850/v850.c: Likewise.
17012 * config/vax/vax.c: Likewise.
17013 * config/visium/visium.c: Likewise.
17014 * config/xtensa/xtensa.c: Likewise.
17016 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17019 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
17020 constraints since the -mupper-regs-* switches have been
17022 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
17023 into a vector from a double word element that was extracted from
17024 another vector, and eliminate extra XXPERMDI instructions.
17025 (vsx_concat_<mode>_2): Likewise.
17026 (vsx_concat_<mode>_3): Likewise.
17027 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
17028 concat to allow optimizing inserts from previous extracts.
17030 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
17032 * config/i386/i386.c (ix86_stack_protect_guard): Generate
17033 memory reference to a SSP offset in TLS address space.
17034 (ix86_print_operand) <case '@'>: Remove.
17035 (ix86_print_operand_punct_valid_p): Remove '@' code.
17036 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
17037 UNSPEC_SP_TLS_TEST.
17038 (stack_tls_protect_set_<mode>): Remove.
17039 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
17040 (stack_tls_protect_test_<mode>): Remove.
17041 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
17043 2017-08-07 Olivier Hainque <hainque@adacore.com>
17046 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
17048 2017-08-07 Douglas Rupp <rupp@adacore.com>
17050 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
17051 variable was referenced as multidir in command.
17053 2017-08-07 Jakub Jelinek <jakub@redhat.com>
17056 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
17059 2017-08-07 Martin Liska <mliska@suse.cz>
17061 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
17062 * config/rl78/rl78.c: Add include of attribs.h.
17063 * config/sh/sh.c: Likewise.
17064 * config/v850/v850.c: Likewise.
17066 2017-08-07 Tom de Vries <tom@codesourcery.com>
17068 PR middle-end/78266
17069 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
17071 2017-08-07 Martin Liska <mliska@suse.cz>
17073 * config/mips/mips.c: Include attribs.h.
17075 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
17078 * doc/invoke.texi: Document change in behvaior for -Ofast for
17081 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
17083 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
17085 (aarch64_pop_regs): Likewise.
17086 (aarch64_gen_load_pair): Likewise.
17087 (aarch64_save_callee_saves): Likewise.
17088 (aarch64_restore_callee_saves): Likewise.
17090 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17092 * config/i386/i386.c: Revert the last change.
17094 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17097 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17099 (ix86_finalize_stack_frame_flags): This. Also clear
17100 frame_pointer_needed if -fno-omit-frame-pointer is used without
17102 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17103 with ix86_finalize_stack_frame_flags.
17104 (ix86_expand_epilogue): Likewise.
17105 (ix86_expand_split_stack_prologue): Likewise.
17107 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17110 * config/i386/i386.c (get_builtin_code_for_version): Set priority
17111 to P_AES for Westmere.
17113 2017-08-07 Jonathan Yong <10walls@gmail.com>
17115 * config/i386/mingw.opt (fset-stack-executable): Removed.
17116 * config/i386/cygming.opt (fset-stack-executable): Moved
17118 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
17120 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
17122 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
17124 2017-08-07 Marek Polacek <polacek@redhat.com>
17126 PR middle-end/81737
17127 * fold-const.c (fold_indirect_ref_1): Check type_domain.
17129 2017-08-07 Martin Liska <mliska@suse.cz>
17131 * attribs.h (canonicalize_attr_name): New function.
17132 (cmp_attribs): Move from c-format.c and adjusted.
17133 (is_attribute_p): Moved from tree.h.
17134 * tree-inline.c: Add new includes.
17135 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
17136 (private_is_attribute_p): Remove.
17137 (private_lookup_attribute): Likewise.
17138 (private_lookup_attribute_by_prefix): Simplify.
17139 (remove_attribute): Use is_attribute_p.
17140 * tree.h: Remove removed declarations.
17142 2017-08-07 Jakub Jelinek <jakub@redhat.com>
17144 PR middle-end/81698
17145 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
17146 instead of computing it in the function. Formatting fix.
17147 (expand_case): Don't rely on default_edge being the first edge,
17148 clear it if removing it, pass default_edge to
17149 emit_case_dispatch_table.
17150 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
17153 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
17155 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
17156 insn in the function, emit NOP after the insn.
17158 2017-08-06 Tom de Vries <tom@codesourcery.com>
17160 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
17163 2017-08-06 Tom de Vries <tom@codesourcery.com>
17165 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
17168 2017-08-04 Yury Gribov <tetra2005@gmail.com>
17170 PR tree-optimization/57371
17171 * match.pd: New pattern.
17173 2017-08-04 Marek Polacek <polacek@redhat.com>
17175 PR middle-end/81695
17176 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
17177 perform the computation in offset_int.
17179 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17181 PR tree-optimization/81136
17182 * tree-vectorizer.h: Include tree-hash-traits.h.
17183 (vec_base_alignments): New typedef.
17184 (vec_info): Add a base_alignments field.
17185 (vect_record_base_alignments): Declare.
17186 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
17188 (DR_IS_CONDITIONAL_IN_STMT): New macro.
17189 (create_data_ref): Add an is_conditional_in_stmt argument.
17190 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
17191 the is_conditional_in_stmt field.
17192 (data_ref_loc): Add an is_conditional_in_stmt field.
17193 (get_references_in_stmt): Set the is_conditional_in_stmt field.
17194 (find_data_references_in_stmt): Update call to create_data_ref.
17195 (graphite_find_data_references_in_stmt): Likewise.
17196 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
17197 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
17198 (vect_record_base_alignment): New function.
17199 (vect_record_base_alignments): Likewise.
17200 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
17201 for nested statements even if we fail to compute a misalignment.
17202 Use pooled base alignments for unconditional references.
17203 (vect_find_same_alignment_drs): Compare base addresses instead
17205 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
17206 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
17208 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17210 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
17211 Add an explicit name for the enum. Use auto_vec for slp_instances
17212 and grouped_stores.
17213 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
17215 (_bb_vec_info): Add a constructor and destructor.
17216 (vinfo_for_stmt): Return NULL for uids of -1 as well.
17217 (destroy_loop_vec_info): Delete.
17218 (vect_destroy_datarefs): Likewise.
17219 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
17220 (vec_info::vec_info): New function.
17221 (vec_info::~vec_info): Likewise.
17222 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
17223 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
17224 destroy_loop_vec_info.
17225 * tree-vect-loop.c (new_loop_vec_info): Replace with...
17226 (_loop_vec_info::_loop_vec_info): ...this.
17227 (destroy_loop_vec_info): Replace with...
17228 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
17229 the stmt_vec_infos. Leave handling of vec_info information to its
17230 destructor. Remove explicit vector releases.
17231 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
17232 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
17233 * tree-vect-slp.c (new_bb_vec_info): Replace with...
17234 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
17235 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
17236 (destroy_bb_vec_info): Replace with...
17237 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
17238 information to its destructor.
17239 (vect_slp_analyze_bb_1): Use new and delete instead of
17240 new_bb_vec_info and destroy_bb_vec_info.
17241 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
17244 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17246 * tree-data-ref.h (subscript): Add access_fn field.
17247 (data_dependence_relation): Add could_be_independent_p.
17248 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
17249 (same_access_functions): Move to tree-data-ref.c.
17250 * tree-data-ref.c (ref_contains_union_access_p): New function.
17251 (access_fn_component_p): Likewise.
17252 (access_fn_components_comparable_p): Likewise.
17253 (dr_analyze_indices): Add a reference to access_fn_component_p.
17254 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
17256 (constant_access_functions): Likewise.
17257 (add_other_self_distances): Likewise.
17258 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
17259 (initialize_data_dependence_relation): Use XCNEW and remove
17260 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
17261 of access functions that have the same type. Allow the
17262 subsequence to end with different bases in some circumstances.
17263 Record the chosen access functions in SUB_ACCESS_FN.
17264 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
17265 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
17266 (subscript_dependence_tester_1): Likewise dra and drb.
17267 (build_classic_dist_vector): Update calls accordingly.
17268 (subscript_dependence_tester): Likewise.
17269 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
17270 DDR_COULD_BE_INDEPENDENT_P.
17271 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
17272 comp_alias_ddrs instead of may_alias_ddrs.
17273 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
17275 (vect_analyze_data_ref_dependence): Use it if
17276 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
17277 distance vectors if that fails.
17278 (dependence_distance_ge_vf): New function.
17279 (vect_prune_runtime_alias_test_list): Use it. Don't clear
17280 LOOP_VINFO_MAY_ALIAS_DDRS.
17282 2017-08-04 Richard Biener <rguenther@suse.de>
17284 PR middle-end/81705
17285 * fold-const.c (fold_binary_loc): Properly restrict
17286 minus_var0 && minus_var1 case when associating undefined overflow
17289 2017-08-04 Tom de Vries <tom@codesourcery.com>
17291 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
17293 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17295 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17296 Don't start diagnostic messages with a capital letter.
17297 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17299 (rs6000_invalid_builtin): Likewise.
17300 (rs6000_trampoline_init): Likewise.
17302 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17305 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
17306 after setting changeable df flags.
17308 2017-08-03 Richard Biener <rguenther@suse.de>
17310 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
17311 up if the use is in USE - X.
17313 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
17315 * toplev.c (dumpfile.h): New include.
17316 (internal_error_reentered): New static function. Use it...
17317 (internal_error_function): ...here to handle reentered internal_error.
17319 2017-08-03 Richard Biener <rguenther@suse.de>
17321 PR middle-end/81148
17322 * fold-const.c (split_tree): Add minus_var and minus_con
17323 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
17324 here but always use minus_*.
17325 (associate_trees): Assert we never associate with MINUS_EXPR
17326 and NULL first operand. Do not recurse for PLUS_EXPR operands
17327 when associating as MINUS_EXPR either.
17328 (fold_binary_loc): Track minus_var and minus_con.
17330 2017-08-03 Tom de Vries <tom@codesourcery.com>
17333 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
17334 ACCEL_COMPILER, apply finish_options on
17335 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
17337 2017-08-03 Tom de Vries <tom@codesourcery.com>
17340 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
17341 function_entry_patch_area_size > 0.
17343 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17346 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
17347 instead of 10??LU, perform unit multiplication in wide_int,
17348 don't change alloc_object_size_limit if the limit is larger
17351 PR tree-optimization/81655
17352 PR tree-optimization/81588
17353 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
17354 the case when ranges[i].low and high are 1 for unsigned type with
17357 PR middle-end/81052
17358 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
17359 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
17361 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17363 * tree-vrp.h: Add include guard.
17365 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
17368 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
17369 (ud2): New insn pattern.
17370 * config/i386/i386.c (ix86_expand_epilogue):
17371 For naked functions, generate ud2 instead of trap insn.
17373 2017-08-02 Marek Polacek <polacek@redhat.com>
17376 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
17378 2017-08-02 Tom de Vries <tom@codesourcery.com>
17379 Cesar Philippidis <cesar@codesourcery.com>
17381 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
17382 Add missing edge probabilities.
17384 2017-08-02 Tamar Christina <tamar.christina@arm.com>
17386 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
17387 Correct endianness.
17389 2017-08-02 Jakub Jelinek <jakub@redhat.com>
17391 PR middle-end/79499
17392 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
17393 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
17394 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
17395 prologue_seq sequences - if any.
17397 2017-08-02 Richard Biener <rguenther@suse.de>
17399 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
17400 via vectors if supported, integer extracts via punning if supported
17401 or otherwise vector extracts.
17403 2017-08-02 Richard Biener <rguenther@suse.de>
17405 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
17407 (bitmap_insert_into_set): ... this.
17409 2017-08-02 Richard Biener <rguenther@suse.de>
17411 PR tree-optimization/81633
17413 2015-08-17 Alan Hayward <alan.hayward@arm.com>
17415 PR tree-optimization/71752
17416 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
17418 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
17420 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
17421 (machine_function::call_ms2sysv_pad_out): Remove field.
17422 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
17423 (ix86_compute_frame_layout): Likewise.
17425 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
17428 * config/i386/i386.c (ix86_set_func_type): Disallow naked
17429 attribute with interrupt attribute.
17431 2017-08-01 Andrew Pinski <apinski@cavium.com>
17433 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
17434 BIT_INSERT_EXPR's operand 1
17435 to see if the types precision matches.
17437 2017-08-01 Martin Liska <mliska@suse.cz>
17439 PR middle-end/70140
17440 * builtins.c (expand_builtin_memcpy_args): Remove.
17441 (expand_builtin_memcpy): Call newly added function
17442 expand_builtin_memory_copy_args.
17443 (expand_builtin_memcpy_with_bounds): Likewise.
17444 (expand_builtin_mempcpy): Remove last argument.
17445 (expand_builtin_mempcpy_with_bounds): Likewise.
17446 (expand_builtin_memory_copy_args): New function created from
17447 expand_builtin_mempcpy_args with small modifications.
17448 (expand_builtin_mempcpy_args): Remove.
17449 (expand_builtin_stpcpy): Remove unused argument.
17450 (expand_builtin): Likewise.
17451 (expand_builtin_with_bounds): Likewise.
17453 2017-08-01 Martin Liska <mliska@suse.cz>
17456 Make mempcpy more optimal (PR middle-end/70140).
17458 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17461 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
17462 __builtin_vec_cmpne verify both arguments are compatible vectors
17463 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
17464 verify arg1_type is a pointer or array type. For __builtin_vec_st,
17465 move computation of aligned to after checking the argument types.
17469 * config/rs6000/vsx.md (vextract_fp_from_shorth,
17470 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
17473 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
17474 Doug Rupp <rupp@adacore.com>
17475 Olivier Hainque <hainque@adacore.com>
17477 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
17478 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
17480 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
17481 for TARGET_OS_CPP_BUILTIN.
17482 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
17483 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
17485 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
17486 passing required abi options to the assembler for EABI configurations.
17487 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
17488 of .text.hot and .text.unlikely sections for kernel modules when
17489 using ARM style exceptions.
17490 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
17491 options. Add EXTRA_CC1_SPEC.
17492 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
17494 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
17496 (ARM_TARGET2_DWARF_FORMAT): Define.
17497 * config/arm/t-vxworks: Adjust multilib control to removal of the
17498 Diab command line options.
17500 2017-08-01 Martin Liska <mliska@suse.cz>
17502 PR gcov-profile/81561
17503 * gcov.c (unblock): Make unblocking safe as we need to preserve
17504 index correspondence of blocks and block_lists.
17506 2017-08-01 Richard Biener <rguenther@suse.de>
17508 PR tree-optimization/81181
17509 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
17510 (compute_antic): ... end of iteration here.
17512 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
17514 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
17515 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
17516 (ftree-slp-vectorize): Likewise.
17517 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
17518 can no longer be set independent of flag_tree_loop_vectorize.
17519 * omp-general.c (emp_max_vf): Likewise.
17520 * opts.c (enable_fdo_optimizations): Remove references to
17521 flag_tree_vectorize, these are now implicit.
17522 (common_handle_option): Remove handling for OPT_ftree_vectorize,
17523 and leave it for the options machinery.
17525 2017-08-01 Martin Liska <mliska@suse.cz>
17527 PR middle-end/70140
17528 * builtins.c (expand_builtin_memcpy_args): Remove.
17529 (expand_builtin_memcpy): Call newly added function
17530 expand_builtin_memory_copy_args.
17531 (expand_builtin_memcpy_with_bounds): Likewise.
17532 (expand_builtin_mempcpy): Remove last argument.
17533 (expand_builtin_mempcpy_with_bounds): Likewise.
17534 (expand_builtin_memory_copy_args): New function created from
17535 expand_builtin_mempcpy_args with small modifications.
17536 (expand_builtin_mempcpy_args): Remove.
17537 (expand_builtin_stpcpy): Remove unused argument.
17538 (expand_builtin): Likewise.
17539 (expand_builtin_with_bounds): Likewise.
17541 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
17544 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
17545 print "ds:" only for immediates in generic address space.
17547 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
17550 * config/i386/i386.c (ix86_funciton_naked): New prototype.
17551 (ix86_function_ok_for_sibcall): Return false for naked functions.
17553 2017-08-01 Richard Biener <rguenther@suse.de>
17555 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
17556 (compute_antic): Seed worklist with exit block predecessors.
17557 * cfganal.c (dfs_find_deadend): For a cycle return the source
17558 of the edge closing it.
17560 2017-08-01 Tamar Christina <tamar.christina@arm.com>
17562 * config/aarch64/aarch64.c
17563 (aarch64_can_const_movi_rtx_p): Move 0 check.
17565 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17567 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
17568 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
17570 * match.pd: Ditto in address comparison pattern.
17572 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17574 PR tree-optimization/81627
17575 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
17576 closed ssa form for store-store chain.
17578 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17580 PR tree-optimization/81620
17581 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
17582 for store-store chain.
17584 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17586 PR tree-optimization/81588
17587 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
17588 ranges[i].in_p, invert comparison code ccode. For >/>=,
17589 swap rhs1 and rhs2 and comparison code unconditionally,
17590 for </<= don't do that. Don't swap rhs1/rhs2 again if
17591 ranges[i].in_p, instead invert comparison code ccode if
17592 opcode or oe->rank is BIT_IOR_EXPR.
17595 * optabs.def (vec_extract_optab, vec_init_optab): Change from
17596 a direct optab to conversion optab.
17597 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
17598 with GET_MODE_INNER as last argument instead of optab_handler.
17599 * expmed.c (extract_bit_field_1): Likewise. Use vector from
17600 vector extraction if possible and optab is available.
17601 * expr.c (store_constructor): Use convert_optab_handler instead
17602 of optab_handler. Use vector initialization from smaller
17603 vectors if possible and optab is available.
17604 * tree-vect-stmts.c (vectorizable_load): Likewise.
17605 * doc/md.texi (vec_extract, vec_init): Document that the optabs
17606 now have two modes.
17607 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
17608 of vec_init from half-sized vectors with the same element mode.
17609 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
17610 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
17611 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
17612 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
17613 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
17614 after mode in gen_vec_extract* calls.
17615 (vec_extract<mode>): Renamed to ...
17616 (vec_extract<mode><ssescalarmodelower>): ... this.
17617 (vec_extract<mode><ssehalfvecmodelower>): New expander.
17618 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
17619 element mode after mode in gen_vec_init* calls.
17620 (VEC_INIT_HALF_MODE): New mode iterator.
17621 (vec_init<mode>): Renamed to ...
17622 (vec_init<mode><ssescalarmodelower>): ... this.
17623 (vec_init<mode><ssehalfvecmodelower>): New expander.
17624 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
17625 (vec_extractv2sfsf): ... this.
17626 (vec_initv2sf): Renamed to ...
17627 (vec_initv2sfsf): ... this.
17628 (vec_extractv2si): Renamed to ...
17629 (vec_extractv2sisi): ... this.
17630 (vec_initv2si): Renamed to ...
17631 (vec_initv2sisi): ... this.
17632 (vec_extractv4hi): Renamed to ...
17633 (vec_extractv4hihi): ... this.
17634 (vec_initv4hi): Renamed to ...
17635 (vec_initv4hihi): ... this.
17636 (vec_extractv8qi): Renamed to ...
17637 (vec_extractv8qiqi): ... this.
17638 (vec_initv8qi): Renamed to ...
17639 (vec_initv8qiqi): ... this.
17640 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
17641 (vec_init<mode>): Renamed to ...
17642 (vec_init<mode><VEC_base_l>): ... this.
17643 (vec_extract<mode>): Renamed to ...
17644 (vec_extract<mode><VEC_base_l>): ... this.
17645 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
17646 (vec_initv2sfsf): ... this.
17647 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
17648 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17649 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
17650 element mode after mode in gen_vec_init* calls.
17651 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
17652 (vec_init<mode><Vel>): ... this.
17653 (vec_extract<mode>): Renamed to ...
17654 (vec_extract<mode><Vel>): ... this.
17655 * config/aarch64/iterators.md (Vel): New mode attribute.
17656 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
17657 Add element mode after mode in gen_vec_extract* calls.
17658 * config/s390/vector.md (non_vec_l): New mode attribute.
17659 (vec_extract<mode>): Renamed to ...
17660 (vec_extract<mode><non_vec_l>): ... this.
17661 (vec_init<mode>): Renamed to ...
17662 (vec_init<mode><non_vec_l>): ... this.
17663 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
17664 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
17666 * config/arm/iterators.md (V_elem_l): New mode attribute.
17667 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
17668 (vec_extract<mode><V_elem_l>): ... this.
17669 (vec_extractv2di): Renamed to ...
17670 (vec_extractv2didi): ... this.
17671 (vec_init<mode>): Renamed to ...
17672 (vec_init<mode><V_elem_l>): ... this.
17673 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
17674 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
17675 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
17676 Add element mode after gen_vec_extract* calls.
17677 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
17678 (vec_init<mode><unitmode>): ... this.
17679 (vec_extract<mode>): Renamed to ...
17680 (vec_extract<mode><unitmode>): ... this.
17681 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
17682 (vec_init<mode><unitmode>): ... this.
17683 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
17684 (vec_initv2sfsf): ... this.
17685 (vec_extractv2sf): Renamed to ...
17686 (vec_extractv2sfsf): ... this.
17687 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
17688 Add element mode after gen_vec_extract* calls.
17689 * config/mips/mips.md (unitmode): New mode iterator.
17690 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
17691 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
17692 * config/spu/spu.md (inner_l): New mode attribute.
17693 (vec_init<mode>): Renamed to ...
17694 (vec_init<mode><inner_l>): ... this.
17695 (vec_extract<mode>): Renamed to ...
17696 (vec_extract<mode><inner_l>): ... this.
17697 * config/sparc/sparc.md (veltmode): New mode iterator.
17698 (vec_init<VMALL:mode>): Renamed to ...
17699 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
17700 * config/ia64/vect.md (vec_initv2si): Renamed to ...
17701 (vec_initv2sisi): ... this.
17702 (vec_initv2sf): Renamed to ...
17703 (vec_initv2sfsf): ... this.
17704 (vec_extractv2sf): Renamed to ...
17705 (vec_extractv2sfsf): ... this.
17706 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
17707 (vec_init<mode>): Renamed to ...
17708 (vec_init<mode><VEC_base_l>): ... this.
17709 (vec_extract<mode>): Renamed to ...
17710 (vec_extract<mode><VEC_base_l>): ... this.
17711 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
17712 (vec_initv2sfsf): ... this.
17713 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
17714 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17715 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
17716 gen_vec_init* calls.
17718 2017-08-01 Richard Biener <rguenther@suse.de>
17720 PR tree-optimization/81297
17721 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
17722 TREE_OVERFLOW from INTEGER_CSTs.
17724 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
17726 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
17728 2017-07-31 Carl Love <cel@us.ibm.com>
17730 * config/rs6000/rs6000-c: Add support for built-in functions
17731 vector signed char vec_xl_be (signed long long, signed char *);
17732 vector unsigned char vec_xl_be (signed long long, unsigned char *);
17733 vector signed int vec_xl_be (signed long long, signed int *);
17734 vector unsigned int vec_xl_be (signed long long, unsigned int *);
17735 vector signed long long vec_xl_be (signed long long, signed long long *);
17736 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
17737 vector signed short vec_xl_be (signed long long, signed short *);
17738 vector unsigned short vec_xl_be (signed long long, unsigned short *);
17739 vector double vec_xl_be (signed long long, double *);
17740 vector float vec_xl_be (signed long long, float *);
17741 * config/rs6000/altivec.h (vec_xl_be): Add #define.
17742 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
17743 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
17745 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
17746 (altivec_expand_builtin): Add switch statement to call
17747 altivec_expand_xl_be for each builtin.
17748 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
17749 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
17750 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
17751 __builtin_vsx_le_be_v16qi.
17752 * doc/extend.texi: Update the built-in documentation file for the
17753 new built-in functions.
17755 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17758 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
17760 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17762 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17764 * config.gcc: Add z14.
17765 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
17766 CPU model numbers for z13s and z14.
17767 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
17769 * config/s390/s390-opts.h (enum processor_type): Rename
17770 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
17771 * config/s390/s390.c (processor_table): Add field for CPU name to
17772 be passed to Binutils.
17773 (s390_asm_output_machine_for_arch): Use the new field in
17774 processor_table for Binutils.
17775 (s390_expand_builtin): Replace arch12 with z14.
17776 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
17777 (s390_get_sched_attrmask): Likewise.
17778 (s390_get_unit_mask): Likewise.
17779 * config/s390/s390.opt: Add z14 to processor_type enum.
17781 2017-07-31 Martin Jambor <mjambor@suse.cz>
17784 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
17785 regardless of optimization level.
17787 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
17788 Martin Liska <mliska@suse.cz>
17790 * predict.def: Remove old comment and adjust probability.
17791 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
17792 PREDICT statements.
17794 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17797 * config/i386/i386.c (ix86_function_naked): New function.
17798 (ix86_can_use_return_insn_p): Return false for naked functions.
17799 (ix86_expand_prologue): Skip prologue for naked functions.
17800 (ix86_expand_epilogue): Skip epilogue for naked functions
17801 and emit trap instruction.
17802 (ix86_warn_func_return): New function.
17803 (ix86_attribute_table): Add "naked" attribute specification.
17804 (TARGET_WARN_FUNC_RETURN): Define.
17805 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
17807 2017-07-31 Martin Liska <mliska@suse.cz>
17809 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
17810 (dump_gimple_bb_header): Always dump BB info.
17811 (pp_cfg_jump): Do not append info about BB when dumping a jump.
17813 2017-07-31 Martin Liska <mliska@suse.cz>
17816 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
17817 also with current_function_decl non-null equality.
17819 2017-07-31 Jakub Jelinek <jakub@redhat.com>
17822 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
17823 change type to the element type, instead add eltype variable and
17824 use it where we are interested in the element type.
17826 PR tree-optimization/81603
17827 * ipa-polymorphic-call.c
17828 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
17829 offset arithmetic in offset_int, bail out if the resulting bit offset
17830 doesn't fit into shwi.
17832 2017-07-31 Martin Liska <mliska@suse.cz>
17834 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
17835 (gimplify_save_expr): Fix comment.
17837 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
17840 * config/i386/i386.c (ix86_function_arg): Update arguments for
17842 (ix86_compute_frame_layout): Set the initial stack offset to
17843 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
17844 INCOMING_FRAME_SP_OFFSET.
17845 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
17846 stack before exception handler returns.
17847 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
17848 the 'ERROR_CODE' for exception handler.
17850 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
17852 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
17853 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
17854 (ASM_OUTPUT_REG_POP): Ditto.
17855 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
17856 instead of asm_fprintf to output pure string.
17858 2017-07-29 Jakub Jelinek <jakub@redhat.com>
17860 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
17861 to imported_module_or_decl hook.
17862 (debug_nothing_tree_tree_tree_bool): Remove.
17863 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
17864 * debug.c (do_nothing_debug_hooks): Use
17865 debug_nothing_tree_tree_tree_bool_bool instead of
17866 debug_nothing_tree_tree_tree_bool.
17867 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17868 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
17869 * sdbout.c (sdb_debug_hooks): Likewise.
17870 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
17871 (gen_namespace_die): Add DW_AT_export_symbols attribute if
17873 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
17874 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
17875 attribute, don't add anything.
17877 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17879 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
17880 (fold_build2_stat_loc): Likewise.
17881 (fold_build3_stat_loc): Likewise.
17882 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
17883 (fold_build1_loc): Remove macro.
17884 (fold_build2_loc): Likewise.
17885 (fold_build3_loc): Likewise.
17887 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17889 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
17890 (gimple_build_debug_bind_source_stat): Likewise.
17891 * gimple.h (gimple_build_debug_bind): Remove macro.
17892 (gimple_build_debug_bind_source): Likewise.
17894 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17896 * bitmap.c (bitmap_alloc): Adjust.
17897 (bitmap_gc_alloc): Likewise.
17898 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
17900 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17902 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
17903 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
17904 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
17905 (bitmap_gc_alloc_stat): Likewise.
17906 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
17908 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17910 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
17911 * rtl.h (shallow_copy_rtx): Remove macro.
17913 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17915 * emit-rtl.c (gen_raw_REG): Adjust.
17916 * gengenrtl.c (gendef): Likewise.
17917 * rtl.c (rtx_alloc_stat): Remove _stat from name.
17918 * rtl.h (rtx_alloc): Remove macro.
17920 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17922 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
17923 (build_tree_list_stat): Likewise.
17924 * tree.h (build_tree_list): Remove macro.
17925 (build_tree_list_vec): Likewise.
17927 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17929 * tree.c (make_vector_stat): Remove _stat from name.
17930 (build_vector_stat): Likewise.
17931 * tree.h (make_vector_stat): Remove macro.
17932 (build_vector_stat): Likewise.
17934 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17936 * tree.h (build_var_debug_value): Remove prototype.
17938 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17940 * tree.c (tree_cons_stat): Remove _stat from name.
17941 * tree.h (tree_cons): Remove macro.
17943 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17945 * tree.c (build_vl_exp_stat): Remove _stat from name.
17946 * tree.h (build_vl_exp): Remove macro.
17948 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17950 * tree.c (build_decl_stat): Remove _stat from name.
17951 * tree.h (build_decl): Remove macro.
17953 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17955 * gimple.c (gimple_build_with_ops_stat): Adjust.
17956 (gimple_alloc_stat): Remove _stat from name.
17957 * gimple.h (gimple_alloc): Remove macro.
17959 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17961 * tree.c (make_tree_vec_stat): Remove _stat from name.
17962 (grow_tree_vec_stat): Likewise.
17963 * tree.h (make_tree_vec_stat): Adjust prototype.
17964 (grow_tree_vec_stat): Likewise.
17965 (make_tree_vec): Remove macro.
17966 (grow_tree_vec): Likewise.
17968 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17970 * fold-const.c (fold_build1_stat_loc): Adjust.
17971 (fold_build2_stat_loc): Likewise.
17972 (fold_build3_stat_loc): Likewise.
17973 * tree.c (build0_stat): Remove _stat from name.
17974 (build1_stat): Likewise.
17975 (build2_stat): Likewise.
17976 (build3_stat): Likewise.
17977 (build4_stat): Likewise.
17978 (build5_stat): Likewise.
17979 * tree.h (build1_loc): Remove macro, and rename _stat function
17981 (build2_loc): Likewise.
17982 (build3_loc): Likewise.
17983 (build4_loc): Likewise.
17984 (build5_loc): Likewise.
17986 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17988 * tree.c (make_int_cst_stat): Remove _stat from name.
17989 * tree.h (make_int_cst_stat): Adjust prototype.
17990 (make_int_cst): Remove macro.
17992 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17994 * tree.c (make_tre_binfo_stat): Remove _stat from name.
17995 * tree.h (make_tree_binfo_stat): Adjust prototype.
17996 (make_tree_binfo): Remove.
17998 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18000 * tree.c (copy_node_stat): Rename to copy_node.
18001 (build_distinct_type_copy): Adjust.
18002 * tree.h (copy_node_stat): Adjust prototype.
18003 (copy_node): Remove macro.
18005 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18007 * tree.c (make_node_stat): rename to make_node.
18008 (build_tree_list_stat): Adjust.
18009 (build0_stat): Likewise.
18010 (build2_stat): Likewise.
18011 (build3_stat): Likewise.
18012 (build4_stat): Likewise.
18013 (build5_stat): Likewise.
18014 (build_decl_stat): Likewise.
18015 * tree.h (make_node_stat): Adjust prototype.
18016 (make_node): remove macro.
18018 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
18020 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
18021 (PPC_FEATURE2_SCV): Likewise.
18022 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
18024 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18026 * config/aarch64/aarch64.c
18027 (aarch64_internal_mov_immediate): Add new special pattern.
18028 * config/aarch64/aarch64.md (*movdi_aarch64):
18029 Add reg/32bit const mov case.
18031 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18032 Richard Sandiford <richard.sandiford@linaro.org>
18034 * config/aarch64/aarch64.md (mov<mode>): Generalize.
18035 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18036 Add integer and movi cases.
18037 (movi-split-hf-df-sf split, fp16): New.
18038 (enabled): Added TARGET_FP_F16INST.
18039 * config/aarch64/iterators.md (GPF_HF): New.
18040 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
18042 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18044 * config/aarch64/aarch64.c
18045 (aarch64_simd_container_mode): Add prototype.
18046 (aarch64_expand_mov_immediate): Add HI support.
18047 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
18048 (aarch64_can_const_movi_rtx_p): New.
18049 (aarch64_preferred_reload_class):
18050 Remove restrictions of using FP registers for certain SIMD operations.
18051 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
18052 (aarch64_valid_floating_const): Add integer move validation.
18053 (aarch64_simd_imm_scalar_p): Remove.
18054 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
18055 (aarch64_legitimate_constant_p): Expand list of supported cases.
18056 * config/aarch64/aarch64-protos.h
18057 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
18058 (aarch64_reinterpret_float_as_int): New.
18059 (aarch64_simd_imm_scalar_p): Remove.
18060 * config/aarch64/constraints.md (Uvi): New.
18061 (Dd): Split into Ds and new Dd.
18062 * config/aarch64/aarch64.md (*movsi_aarch64):
18064 (*movdi_aarch64): Add SIMD mov case.
18066 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18068 * tree-predcom.c: (struct chain): Handle store-store chain in which
18069 stores for elimination only store loop invariant values.
18070 (execute_pred_commoning_chain): Ditto.
18071 (prepare_initializers_chain_store_elim): Ditto.
18072 (prepare_finalizers): Ditto.
18073 (is_inv_store_elimination_chain): New function.
18074 (initialize_root_vars_store_elim_1): New function.
18076 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18078 * tree-predcom.c: Revise general description of the pass.
18079 (enum chain_type): New enum type for store elimination.
18080 (struct chain): New field supporting store elimination.
18081 (struct component): Ditto.
18082 (dump_chain): Dump store-stores chain.
18083 (release_chain): Release resources.
18084 (split_data_refs_to_components): Compute and create component
18085 contains only stores for elimination.
18086 (get_chain_last_ref_at): New function.
18087 (make_invariant_chain): Initialization.
18088 (make_rooted_chain): Specify chain type in parameter and record it.
18089 (add_looparound_copies): Skip for store-stores chain.
18090 (determine_roots_comp): Compute type of chain and pass it to
18092 (initialize_root_vars_store_elim_2): New function.
18093 (finalize_eliminated_stores): New function.
18094 (remove_stmt): Handle store for elimination.
18095 (execute_pred_commoning_chain): Execute predictive commoning on
18096 store-store chains.
18097 (determine_unroll_factor): Skip unroll for store-stores chain.
18098 (prepare_initializers_chain_store_elim): New function.
18099 (prepare_initializers_chain): Hanlde store-store chain.
18100 (prepare_finalizers_chain, prepare_finalizers): New function.
18101 (tree_predictive_commoning_loop): Return integer value indicating
18102 if loop is unrolled or lcssa form is corrupted.
18103 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
18105 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18107 * tree-predcom.c (initialize_root): Delete.
18108 (execute_pred_commoning_chain): Initialize root vars and replace
18109 reference of non-combined chain directly, rather than call above
18112 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18114 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
18115 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
18117 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18119 * tree-predcom.c (struct chain): New field init_seq.
18120 (release_chain): Release init_seq.
18121 (prepare_initializers_chain): Record intialization stmts in above
18123 (insert_init_seqs): New function.
18124 (tree_predictive_commoning_loop): Call insert_init_seqs.
18126 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18128 * tree-predcom.c (determine_roots_comp): Skip trivial components.
18130 2017-07-28 Richard Biener <rguenther@suse.de>
18132 * match.pd: Remove superfluous :c.
18133 * genmatch.c (simplify::id): Add member.
18134 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
18136 (current_id): New global.
18137 (dt_node::parent): Move from ...
18138 (dt_operand::parent): ... here. Add for_id member.
18139 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
18140 (decision_tree::find_node): Relax order requirement when
18141 merging DT_TRUE nodes to ones inbetween the current simplify
18142 and the one we try to merge with. Add diagnostic whenever
18143 we need to enforce pattern order by not merging.
18144 (decision_tree::insert): Set current_id.
18145 (decision_tree::print_node): Dump parent node and for_id.
18146 (parser::last_id): Add member.
18147 (parser::push_simplify): Assign unique id.
18148 (parser::parser): Initialize last_id.
18150 2017-07-28 Martin Liska <mliska@suse.cz>
18153 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
18154 gimple_build_debug_bind.
18156 2017-07-28 Richard Biener <rguenther@suse.de>
18158 PR tree-optimization/81502
18159 * match.pd: Add pattern combining BIT_INSERT_EXPR with
18161 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
18163 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
18164 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
18165 for BIT_FIELD_REF args.
18166 * fold-const.c (make_bit_field_ref): Likewise.
18167 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
18169 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18172 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
18173 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
18174 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
18175 Or it into SANITIZER_UNDEFINED.
18176 * ubsan.c: Include gimple-fold.h and varasm.h.
18177 (ubsan_expand_ptr_ifn): New function.
18178 (instrument_pointer_overflow): New function.
18179 (maybe_instrument_pointer_overflow): New function.
18180 (instrument_object_size): Formatting fix.
18181 (pass_ubsan::execute): Call instrument_pointer_overflow
18182 and maybe_instrument_pointer_overflow.
18183 * internal-fn.c (expand_UBSAN_PTR): New function.
18184 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
18185 * sanitizer.def (__ubsan_handle_pointer_overflow,
18186 __ubsan_handle_pointer_overflow_abort): New builtins.
18187 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
18188 * internal-fn.def (UBSAN_PTR): New internal function.
18189 * opts.c (sanitizer_opts): Add pointer-overflow.
18190 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
18191 * fold-const.c (build_range_check): Compute pointer range check in
18192 integral type if pointer arithmetics would be needed. Formatting
18195 2017-07-28 Martin Liska <mliska@suse.cz>
18198 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
18199 parameters that are of a variable-length.
18201 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
18203 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
18205 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
18206 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
18207 (CRT_CALL_STATIC_FUNCTION): Likewise.
18208 (ASM_DEFAULT_SPEC): New define.
18209 (ASM_SPEC32): Likewise.
18210 (ASM_SPEC64): Likewise.
18211 (ASM_SPEC_COMMON): Likewise.
18212 (ASM_SPEC): Likewise.
18213 (INVALID_64BIT): Likewise.
18214 (LINK_OS_DEFAULT_SPEC): Likewise.
18215 (LINK_OS_SPEC32): Likewise.
18216 (LINK_OS_SPEC64): Likewise.
18217 (POWERPC_LINUX): Likewise.
18218 (PTRDIFF_TYPE): Likewise.
18219 (RESTORE_FP_PREFIX): Likewise.
18220 (RESTORE_FP_SUFFIX): Likewise.
18221 (SAVE_FP_PREFIX): Likewise.
18222 (SAVE_FP_SUFFIX): Likewise.
18223 (SIZE_TYPE): Likewise.
18224 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
18225 (TARGET_64BIT): Likewise.
18226 (TARGET_64BIT): Likewise.
18227 (TARGET_AIX): Likewise.
18228 (WCHAR_TYPE_SIZE): Likewise.
18229 (WCHAR_TYPE): Undefine.
18230 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
18231 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
18232 (CPP_OS_RTEMS_SPEC): Delete.
18233 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
18234 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
18236 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
18238 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18240 PR tree-optimization/81578
18241 * tree-parloops.c (build_new_reduction): Bail out if
18242 reduction_code isn't one of the standard OpenMP reductions.
18243 Move the details printing after that decision.
18245 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18247 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
18248 related to reload_in_progress.
18249 (splat_input_operand): Likewise.
18250 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
18252 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
18254 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
18255 (TARGET_INSTANTIATE_DECLS): Likewise.
18256 (legitimate_indexed_address_p): Delete reload_in_progress code.
18257 (rs6000_debug_legitimate_address_p): Likewise.
18258 (rs6000_eliminate_indexed_memrefs): Likewise.
18259 (rs6000_emit_le_vsx_store): Likewise.
18260 (rs6000_emit_move_si_sf_subreg): Likewise.
18261 (rs6000_emit_move): Likewise.
18262 (register_to_reg_type): Likewise.
18263 (rs6000_pre_atomic_barrier): Likewise.
18264 (rs6000_machopic_legitimize_pic_address): Likewise.
18265 (rs6000_allocate_stack_temp): Likewise.
18266 (rs6000_address_for_fpconvert): Likewise.
18267 (rs6000_address_for_altivec): Likewise.
18268 (rs6000_secondary_memory_needed_rtx): Delete function.
18269 (rs6000_check_sdmode): Likewise.
18270 (rs6000_alloc_sdmode_stack_slot): Likewise.
18271 (rs6000_instantiate_decls): Likewise.
18272 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
18273 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
18274 Delete reload_in_progress.
18275 (*vec_reload_and_plus_<mptrsize>): Likewise.
18276 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
18277 (vsx_div_v2di): Likewise.
18278 (vsx_udiv_v2di): Likewise.
18280 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18282 * config/rs6000/rs6000.opt (mlra): Replace with stub.
18283 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
18284 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
18285 (rs6000_debug_reg_global): Delete print of LRA status.
18286 (rs6000_option_override_internal): Delete dead LRA related code.
18287 (rs6000_lra_p): Delete function.
18288 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
18290 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
18292 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
18293 * config/riscv/rtems.h: New file.
18295 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18296 Sudakshina Das <sudi.das@arm.com>
18298 * config/aarch64/aarch64.md
18299 (define_split for and<mode>3nr_compare): Move
18300 non aarch64_logical_operand to a register.
18301 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
18302 register immediate operand to a register.
18303 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
18305 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18307 PR middle-end/81564
18308 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
18310 2017-07-27 Richard Biener <rguenther@suse.de>
18312 PR tree-optimization/81573
18313 PR tree-optimization/81494
18314 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
18315 multi defuse cycle case.
18317 2017-07-27 Richard Biener <rguenther@suse.de>
18319 PR tree-optimization/81571
18320 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
18323 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
18325 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
18326 earlier and only if MASK_FPU is set. Adjust formatting.
18328 2017-07-27 Martin Liska <mliska@suse.cz>
18330 * opt-functions.awk: Add validation of value of Init.
18331 * optc-gen.awk: Pass new argument.
18333 2017-07-27 Martin Liska <mliska@suse.cz>
18335 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
18336 Fix wrong condition.
18338 2017-07-27 Martin Liska <mliska@suse.cz>
18340 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
18341 BBs and edges seen by autoFDO.
18343 2017-07-27 Richard Biener <rguenther@suse.de>
18345 PR tree-optimization/81502
18346 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
18347 with incompatible but same sized type.
18348 (execute_update_addresses_taken): Likewise.
18350 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
18352 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
18353 flag_tree_loop_vectorize rather than flag_tree_vectorize.
18355 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18358 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
18359 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
18360 Change s_operand to memory_operand.
18362 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
18364 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
18365 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
18366 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
18367 Emit instructions rather than returning an expression. Handle TFmode
18368 and KFmode by casting to TImode.
18369 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
18370 (rs6000_emit_le_vsx_store): Likewise.
18371 * config/rs6000/vsx.md (VSX_TI): New iterator.
18372 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
18373 (*vsx_le_undo_permute_<mode>): Likewise.
18374 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
18375 emit the split sequence.
18376 (*vsx_le_perm_store_<mode>): Likewise.
18378 2017-07-27 Jakub Jelinek <jakub@redhat.com>
18380 PR tree-optimization/81555
18381 PR tree-optimization/81556
18382 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
18383 if true, force CHANGED for the recursive invocation.
18384 (reassociate_bb): Remember original length of ops array, pass
18385 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
18387 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
18388 attributes for noipa attribute. For naked attribute use
18389 lookup_attribute first before lookup_attribute_spec.
18390 * final.c (rest_of_handle_final): Disable IPA RA for functions with
18392 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
18393 for functions with noipa attribute.
18394 (cgraph_externally_visible_p): Return true for functions with noipa
18396 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
18397 for functions with noipa attribute.
18398 * doc/extend.texi: Document noipa function attribute.
18399 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
18400 also for functions with noipa attribute.
18401 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
18403 2017-07-26 Andrew Pinski <apinski@cavium.com>
18405 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
18406 vec_unalign_load_cost and vec_unalign_store_cost.
18408 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
18410 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
18411 -mvsx-small-integer option.
18412 (ISA_3_0_MASKS_IEEE): Likewise.
18413 (OTHER_VSX_VECTOR_MASKS): Likewise.
18414 (POWERPC_MASKS): Likewise.
18415 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
18416 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
18417 code, only testing for DImode being allowed in non-VSX floating
18419 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
18420 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
18422 (rs6000_option_override_internal): Delete -mvsx-small-integer.
18423 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
18424 TARGET_P8_VECTOR test.
18425 (rs6000_secondary_reload_simple_move): Likewise.
18426 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
18427 since TARGET_P9_VECTOR was already tested.
18428 (rs6000_opt_masks): Remove -mvsx-small-integer.
18429 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
18430 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18432 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
18433 test for TARGET_VEXTRACTUB was used, and that uses
18435 (p9 extract splitter): Likewise.
18436 (vsx_extract_<mode>_di_p9): Likewise.
18437 (vsx_extract_<mode>_store_p9): Likewise.
18438 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
18439 for TARGET_P9_VECTOR was used. Delete code that is now dead with
18440 the elimination of TARGET_VSX_SMALL_INTEGER.
18441 (vsx_extract_<mode>_p8): Likewise.
18442 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
18443 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
18444 (vsx_set_<mode>_p9): Likewise.
18445 (vsx_set_v4sf_p9): Likewise.
18446 (vsx_set_v4sf_p9_zero): Likewise.
18447 (vsx_insert_extract_v4sf_p9): Likewise.
18448 (vsx_insert_extract_v4sf_p9_2): Likewise.
18449 * config/rs6000/rs6000.md (sign extend splitter): Change
18450 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
18451 (floatsi<mode>2_lfiwax_mem): Likewise.
18452 (floatunssi<mode>2_lfiwzx_mem): Likewise.
18453 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
18454 since a test for TARGET_P9_VECTOR was used.
18455 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18456 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
18457 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18458 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18459 TARGET_P8_VECTOR test.
18460 (fix_trunc<mode>si2_stfiwx): Likewise.
18461 (fix_trunc<mode>si2_internal): Likewise.
18462 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
18463 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18465 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18466 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18467 TARGET_P8_VECTOR test.
18468 (fixuns_trunc<mode>si2_stfiwx): Likewise.
18469 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
18470 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18472 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18473 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
18474 since a test for TARGET_P9_VECTOR was used.
18475 (splitter for loading small constants): Likewise.
18477 2017-07-26 Andrew Pinski <apinski@cavium.com>
18479 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
18482 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
18485 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
18486 (fp_valid_at): Likewise.
18488 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18490 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
18491 (qdf24xx_addrcost_table): Likewise.
18492 (cortexa57_tunings): Update to use generic_branch_cost.
18493 (cortexa72_tunings): Likewise.
18494 (cortexa73_tunings): Likewise.
18495 (qdf24xx_tunings): Likewise.
18497 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18499 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
18500 (thunderx2t99_branch_cost): Likewise.
18501 (cortexa35_tunings): Update to use generic_branch_cost.
18502 (cortexa53_tunings): Likewise.
18503 (cortexa57_tunings): Likewise.
18504 (cortexa72_tunings): Likewise.
18505 (cortexa73_tunings): Likewise.
18506 (thunderx2t99_tunings): Likewise.
18508 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
18510 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
18511 (sparc_option_override): Honour MASK_FSMULD.
18512 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
18513 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
18514 * config/sparc/sparc.opt (mfsmuld): New option.
18515 * doc/invoke.texi (mfsmuld): Document option.
18517 2017-07-26 Marek Polacek <polacek@redhat.com>
18519 PR middle-end/70992
18520 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
18522 2017-07-26 Richard Biener <rguenther@suse.de>
18524 * gimple-match-head.c (do_valueize): Return OP if valueize
18526 (get_def): New helper to get at the def stmt of a SSA name
18527 if valueize allows.
18528 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
18529 do_valueize to get at the def stmt.
18530 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
18532 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
18534 PR middle-end/46932
18535 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
18537 2017-07-26 Martin Liska <mliska@suse.cz>
18540 * function.c (expand_function_start): Make expansion of
18541 nonlocal_goto_save_area after parm_birth_insn.
18543 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
18545 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
18546 from all CPU target flags enable members.
18548 2017-07-26 Richard Biener <rguenther@suse.de>
18550 * genmatch.c (dt_simplify::gen): Make iterator vars const.
18551 (decision_tree::gen): Make 'type' const.
18552 (write_predicate): Likewise.
18554 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
18556 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
18557 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
18558 (rs6000_option_override_internal): Likewise.
18559 (rs6000_expand_vector_set): Likewise.
18560 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
18561 (TARGET_UPPER_REGS_SF): Likewise.
18562 (TARGET_UPPER_REGS_DI): Likewise.
18563 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
18564 (TARGET_DIRECT_MOVE_64BIT): Likewise.
18565 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
18566 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18567 (Splitters for DI constants in Altivec registers): Likewise.
18568 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
18569 (vsx_set_v4sf_p9): Likewise.
18570 (vsx_set_v4sf_p9_zero): Likewise.
18571 (vsx_insert_extract_v4sf_p9): Likewise.
18572 (vsx_insert_extract_v4sf_p9_2): Likewise.
18574 2017-07-25 Carl Love <cel@us.ibm.com>
18576 * doc/extend.texi: Update the built-in documentation file for the
18577 existing built-in functions
18578 vector signed char vec_cnttz (vector signed char);
18579 vector unsigned char vec_cnttz (vector unsigned char);
18580 vector signed short vec_cnttz (vector signed short);
18581 vector unsigned short vec_cnttz (vector unsigned short);
18582 vector signed int vec_cnttz (vector signed int);
18583 vector unsigned int vec_cnttz (vector unsigned int);
18584 vector signed long long vec_cnttz (vector signed long long);
18585 vector unsigned long long vec_cnttz (vector unsigned long long);
18587 2017-07-25 Andrew Pinski <apinski@cavium.com>
18589 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
18590 accesses where the use is for the first operand of a BIT_INSERT.
18592 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18595 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18596 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18598 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18600 * config/i386/gstabs.h: Delete.
18601 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
18603 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
18605 * config/i386/i386.c (ix86_decompose_address): Do not check for
18606 register RTX when looking at index_reg or base_reg.
18607 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
18609 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
18611 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
18612 to update EH info here.
18614 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18616 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
18618 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18620 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
18622 2017-07-25 Torsten Duwe <duwe@suse.de>
18624 * common.opt: Introduce -fpatchable-function-entry
18625 command line option, and its variables function_entry_patch_area_size
18626 and function_entry_patch_area_start.
18627 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
18628 including a two-value parser.
18629 * target.def (print_patchable_function_entry): New target hook.
18630 * targhooks.h (default_print_patchable_function_entry): New function.
18631 * targhooks.c (default_print_patchable_function_entry): Likewise.
18632 * toplev.c (process_options): Switch off IPA-RA if
18633 patchable function entries are being generated.
18634 * varasm.c (assemble_start_function): Look at the
18635 patchable-function-entry command line switch and current
18636 function attributes and maybe generate NOP instructions by
18637 calling the print_patchable_function_entry hook.
18638 * doc/extend.texi: Document patchable_function_entry attribute.
18639 * doc/invoke.texi: Document -fpatchable_function_entry
18640 command line option.
18641 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
18643 * doc/tm.texi: Re-generate.
18645 2017-07-25 Jakub Jelinek <jakub@redhat.com>
18648 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
18649 TARGET_AVX512DQ rather than TARGET_AVX512BW.
18651 2017-07-25 Tamar Christina <tamar.christina@arm.com>
18653 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
18655 2017-07-25 Richard Biener <rguenther@suse.de>
18657 PR tree-optimization/81455
18658 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
18659 not walk in cycles when looking for guards.
18661 2017-07-25 Richard Biener <rguenther@suse.de>
18663 PR tree-optimization/81529
18664 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
18665 when optimizing backedge uses.
18667 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
18669 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
18671 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
18672 to dl_section_ref. On AIX, append an expression to subtract
18673 the size of the section length to dl_section_ref.
18675 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
18677 * configure.ac: If any of the config.* scripts fail, exit 1.
18678 * configure: Regenerate.
18680 2017-07-25 Richard Biener <rguenther@suse.de>
18682 PR middle-end/81546
18683 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
18684 of immediate uses, be more verbose on errors.
18686 2017-07-25 Richard Biener <rguenther@suse.de>
18688 PR tree-optimization/81510
18689 * tree-vect-loop.c (vect_is_simple_reduction): When the
18690 reduction stmt is not inside the loop bail out.
18692 2017-07-25 Richard Biener <rguenther@suse.de>
18694 PR tree-optimization/81303
18695 * tree-vect-loop-manip.c (vect_loop_versioning): Build
18696 profitability check against LOOP_VINFO_NITERSM1.
18698 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18700 * domwalk.c (cmp_bb_postorder): Simplify.
18701 (sort_bbs_postorder): New function. Use it...
18702 (dom_walker::walk): ...here to optimize common cases.
18704 2017-07-25 Martin Liska <mliska@suse.cz>
18707 * ipa-visibility.c (function_and_variable_visibility): Make the
18708 redirection just on target that supports aliasing.
18709 Fix GNU coding style.
18711 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
18714 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
18716 * config/arm/rtems.h: Likewise.
18717 * config/bfin/rtems.h: Likewise.
18718 * config/i386/rtemself.h: Likewise.
18719 * config/lm32/rtems.h: Likewise.
18720 * config/m32c/rtems.h: Likewise.
18721 * config/m68k/rtemself.h: Likewise.
18722 * config/microblaze/rtems.h: Likewise.
18723 * config/mips/rtems.h: Likewise.
18724 * config/moxie/rtems.h: Likewise.
18725 * config/nios2/rtems.h: Likewise.
18726 * config/powerpcspe/rtems.h: Likewise.
18727 * config/rs6000/rtems.h: Likewise.
18728 * config/rtems.h: Likewise.
18729 * config/sh/rtems.h: Likewise.
18730 * config/sh/rtemself.h: Likewise.
18731 * config/sparc/rtemself.h: Likewise.
18733 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
18736 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
18737 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
18738 * tree-ssa-structalias.c (alias_get_name): Same.
18740 2017-07-25 Bin Cheng <bin.cheng@arm.com>
18743 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
18744 instructions if no du chain is found.
18746 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
18748 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
18750 2017-07-25 Richard Biener <rguenther@suse.de>
18752 PR middle-end/81505
18753 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
18756 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
18758 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
18759 upper-regs options.
18760 (ISA_2_7_MASKS_SERVER): Likewise.
18761 (ISA_3_0_MASKS_IEEE): Likewise.
18762 (OTHER_P8_VECTOR_MASKS): Likewise.
18763 (OTHER_VSX_VECTOR_MASKS): Likewise.
18764 (POWERPC_MASKS): Likewise.
18765 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
18766 duplicate list of options.
18767 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
18768 explicit -mupper-regs options.
18769 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
18770 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
18771 alias for -mupper-regs-df.
18772 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18773 (rs6000_init_hard_regno_mode_ok): Likewise.
18774 (rs6000_option_override_internal): Likewise.
18775 (rs6000_opt_masks): Likewise.
18776 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
18777 options in terms of whether -mvsx or -mpower8-vector was used.
18778 (TARGET_UPPER_REGS_DI): Likewise.
18779 (TARGET_UPPER_REGS_SF): Likewise.
18780 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
18781 -mupper-regs-* options.
18783 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
18785 * passes.c (emergency_dump_function): Print some empty lines and a
18786 header before the RTL dump.
18788 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
18790 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
18792 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
18795 * config/aarch64/aarch64.c (aarch64_classify_symbol):
18796 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
18798 2017-07-24 Carl Love <cel@us.ibm.com>
18800 * config/rs6000/rs6000-c.c: Add support for built-in functions
18801 vector float vec_extract_fp32_from_shorth (vector unsigned short);
18802 vector float vec_extract_fp32_from_shortl (vector unsigned short);
18803 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
18804 vec_extract_fp_from_shortl): Add defines for the two builtins.
18805 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
18806 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
18808 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
18809 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
18810 * doc/extend.texi: Update the built-in documentation file for the
18811 new built-in function.
18813 2017-07-24 Jakub Jelinek <jakub@redhat.com>
18816 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
18818 * doc/generic.texi: Likewise.
18819 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18820 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18822 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
18824 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
18825 (aarch64_mls_elt_merge<mode>): Likewise.
18827 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
18829 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18830 having __cxa_atexit.
18832 2017-07-23 Michael Collison <michael.collison@arm.com>
18834 * config/arm/arm.c (arm_option_override): Deprecate
18835 use of -mstructure-size-boundary.
18836 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
18837 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
18839 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18842 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18843 Reduce cost estimate for direct moves.
18845 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
18848 * config/i386/i386.c (ix86_option_override_internal): Disable
18849 BMI, BMI2 and TBM instructions for -m16.
18851 2017-07-21 Carl Love <cel@us.ibm.com>
18853 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18854 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18855 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18856 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18857 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18858 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18859 VMULOSW): New enum "unspec" values.
18860 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18861 altivec_vmulosw): New patterns.
18862 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18863 VMULOSW): Add definitions.
18865 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
18867 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
18868 (qdf24xx): Likewise.
18869 * config/aarch64/aarch64-options-extensions.def (rdma); New.
18870 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
18871 (AARCH64_FL_V8_1): Renumber.
18872 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
18873 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
18874 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
18875 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
18876 rdma to feature modifiers list.
18878 2017-07-21 Yury Gribov <tetra2005@gmail.com>
18880 PR middle-end/56727
18881 * ipa-visibility (function_and_variable_visibility): Convert
18882 recursive PLT call to direct call if appropriate.
18884 2017-07-21 Andrew Pinski <apinski@cavium.com>
18886 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
18887 operand 1 to see if the types precision matches.
18888 * fold-const.c (operand_equal_p): Likewise.
18890 2017-07-21 Richard Biener <rguenther@suse.de>
18892 PR tree-optimization/81303
18893 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
18894 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
18895 (vect_peeling_hash_get_lowest_cost): Adjust.
18896 (vect_enhance_data_refs_alignment): Likewise. Use
18897 vect_get_peeling_costs_all_drs to compute the penalty for no
18898 peeling to match up costs.
18900 2017-07-21 Richard Biener <rguenther@suse.de>
18902 PR tree-optimization/81500
18903 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
18904 we didn't identify a reduction path.
18906 2017-07-21 Tom de Vries <tom@codesourcery.com>
18907 Cesar Philippidis <cesar@codesourcery.com>
18909 PR gcov-profile/81442
18910 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
18913 2017-07-21 Tom de Vries <tom@codesourcery.com>
18916 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
18918 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
18919 nvptx_override_options_after_change.
18921 2017-07-21 Ulrich Drepper <drepper@redhat.com>
18923 * dwarf2out.c (output_file_names): Avoid double testing for
18924 dwarf_version >= 5.
18926 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
18928 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
18930 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
18932 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
18934 (try_crossjump_to_edge): Do not punt on partitioned functions.
18936 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
18938 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
18939 Put all BBs reachable only via paths crossing cold region to cold
18941 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
18943 2016-07-21 Richard Biener <rguenther@suse.de>
18945 PR tree-optimization/81303
18946 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
18947 into account prologue and epilogue iterations when raising
18948 min_profitable_iters to sth at least covering one vector iteration.
18950 2017-07-21 Tamar Christina <tamar.christina@arm.com>
18952 * config/arm/arm.c (arm_test_cpu_arch_dat):
18955 2017-07-20 Nathan Sidwell <nathan@acm.org>
18957 Remove TYPE_METHODS.
18958 * tree.h (TYPE_METHODS): Delete.
18959 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
18960 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
18961 (dbxout_type_methods): Scan TYPE_FIELDS.
18962 (dbxout_type): Don't check TYPE_METHODS here.
18963 * function.c (use_register_for_decl): Always ignore register for
18964 class types when not optimizing.
18965 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
18966 * tree.c (free_lang_data_in_type): Stitch out member functions and
18967 templates from TYPE_FIELDS.
18968 (build_distinct_type_copy, verify_type_variant,
18969 verify_type): Member fns are on TYPE_FIELDS.
18970 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
18971 * tree-pretty-print.c (dump_generic_node): Likewise.
18973 2017-07-20 Jakub Jelinek <jakub@redhat.com>
18976 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
18977 V2TImode and V4TImode.
18978 (ix86_expand_vector_extract): Likewise.
18979 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
18980 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
18981 (ssescalarmode): Handle V4TImode and V2TImode.
18982 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
18983 (*vec_extractv2ti, *vec_extractv4ti): New insns.
18984 (VEXTRACTI128_MODE): New mode iterator.
18985 (splitter for *vec_extractv?ti first element): New.
18986 (VEC_INIT_MODE): New mode iterator.
18987 (vec_init<mode>): Consolidate 3 expanders into one using
18988 VEC_INIT_MODE mode iterator.
18990 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
18992 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
18993 non_spilled_static_chain_regno_p.
18995 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
18997 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
18999 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
19001 * bb-reorder.c (connect_traces): Allow copying of blocks within
19004 2017-07-20 Richard Biener <rguenther@suse.de>
19006 * gimple.h (gimple_phi_result): Add gphi * overload.
19007 (gimple_phi_result_ptr): Likewise.
19008 (gimple_phi_arg): Likewise. Adjust index assert to only
19009 allow actual argument accesses rather than all slots available
19011 (gimple_phi_arg_def): Add gphi * overload.
19012 * tree-phinodes.c (make_phi_node): Initialize only actual
19014 (resize_phi_node): Clear memory not covered by old node,
19015 do not initialize excess argument slots.
19016 (reserve_phi_args_for_new_edge): Initialize new argument slot
19019 2017-07-20 Bin Cheng <bin.cheng@arm.com>
19021 PR tree-optimization/81388
19023 2016-07-21 Bin Cheng <bin.cheng@arm.com>
19025 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
19026 by removing computation of may_be_zero.
19028 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19029 Tom de Vries <tom@codesourcery.com>
19031 PR middle-end/81030
19032 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
19033 when gimple level profile disagrees with what RTL expander did.
19035 2017-07-20 Richard Biener <rguenther@suse.de>
19037 PR tree-optimization/61171
19038 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
19039 (vect_analyze_stmt): Add slp instance parameter.
19040 (vectorizable_reduction): Likewise.
19041 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
19042 (vect_is_simple_reduction): Deal with chains not detected
19043 as SLP reduction chain, specifically not properly associated
19044 chains containing a mix of plus/minus.
19045 (get_reduction_op): Remove.
19046 (get_initial_defs_for_reduction): Simplify, pass in whether
19047 this is a reduction chain, pass in the SLP node for the PHIs.
19048 (vect_create_epilog_for_reduction): Get the SLP instance as
19050 (vectorizable_reduction): Get the SLP instance as arg.
19051 During analysis remember the SLP node with the PHIs in the
19052 instance. Simplify getting at the vectorized reduction PHIs.
19053 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
19054 through SLP instance.
19055 (vect_slp_analyze_operations): Likewise.
19056 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
19057 (vect_transform_stmt): Likewise.
19059 2017-07-20 Tom de Vries <tom@codesourcery.com>
19061 PR tree-optimization/81489
19062 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
19063 read of phi arg location to before loop that modifies phi.
19065 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19067 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
19070 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19072 PR middle-end/81331
19073 * except.c (execute): Fix ordering issue.
19075 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
19077 PR rtl-optimization/81423
19078 * combine.c (make_compound_operation_int): Don't try to optimize
19079 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
19081 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
19083 PR rtl-optimization/81423
19084 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
19085 with a constant that is -1 in the truncated to mode.
19087 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19089 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
19090 (determine_unlikely_bbs): ... here.
19091 * predict.h (propagate_unlikely_bbs_forward): Declare.
19092 * cfgexpand.c (pass_expand::execute): Use it.
19093 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
19095 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
19096 propagate_unlikely_bbs_forward.
19098 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19100 PR middle-end/81331
19101 * except.c (maybe_add_nop_after_section_switch): New function.
19104 2017-07-19 Tom de Vries <tom@codesourcery.com>
19106 * gimple.h (gimple_phi_set_arg): Make assert more strict.
19108 2017-07-19 Tom de Vries <tom@codesourcery.com>
19110 * gimple.h (gimple_phi_arg): Make assert more strict.
19112 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
19114 * config.gcc (powerpc*-*-*): Add mmintrin.h.
19115 * config/rs6000/mmintrin.h: New file.
19116 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
19118 2017-07-19 Jakub Jelinek <jakub@redhat.com>
19120 PR tree-optimization/81346
19121 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
19123 2017-07-19 Tom de Vries <tom@codesourcery.com>
19125 * config/nvptx/nvptx.md (VECIM): Add V2DI.
19127 2017-07-19 Tom de Vries <tom@codesourcery.com>
19129 * config/nvptx/nvptx-modes.def: Add V2DImode.
19130 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
19131 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
19132 (nvptx_output_mov_insn): Handle lack of mov.b128.
19133 (nvptx_print_operand): Handle 'H' and 'L' codes.
19134 (nvptx_vector_mode_supported): Allow V2DImode.
19135 (nvptx_preferred_simd_mode): New function.
19136 (nvptx_data_alignment): New function.
19137 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
19138 nvptx_preferred_simd_mode.
19139 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
19141 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
19143 2017-07-19 Tom de Vries <tom@codesourcery.com>
19145 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
19146 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
19147 (nvptx_vector_mode_supported): New function. Allow V2SImode.
19148 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
19149 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
19150 (mov<VECIM>_insn): New define_insn.
19151 (define_expand "mov<VECIM>): New define_expand.
19153 2017-07-19 Tom de Vries <tom@codesourcery.com>
19155 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
19157 2017-07-19 Jakub Jelinek <jakub@redhat.com>
19159 PR tree-optimization/81346
19160 * fold-const.h (fold_div_compare, range_check_type): Declare.
19161 * fold-const.c (range_check_type): New function.
19162 (build_range_check): Use range_check_type.
19163 (fold_div_compare): No longer static, rewritten into
19164 a match.pd helper function.
19165 (fold_comparison): Don't call fold_div_compare here.
19166 * match.pd (X / C1 op C2): New optimization using fold_div_compare
19167 as helper function.
19169 2017-07-19 Nathan Sidwell <nathan@acm.org>
19171 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
19172 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
19173 * tree.c (find_decls_types_r, verify_type): Use
19174 TYPE_{MIN,MAX}_VALUE_RAW.
19175 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
19176 (hash_tree): Likewise.
19177 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19179 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19182 2017-07-18 Tom de Vries <tom@codesourcery.com>
19184 PR middle-end/81464
19185 * omp-expand.c (expand_omp_for_static_chunk): Handle
19186 equal-argument loop exit phi.
19188 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
19191 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
19192 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
19193 operand 2 predicate.
19194 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
19195 operand 2 predicate.
19196 (ror,rol -> rorx splitters): Use const_int_operand as
19197 operand 2 predicate.
19199 2017-06-18 Richard Biener <rguenther@suse.de>
19201 PR tree-optimization/81410
19202 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
19203 the gap in the ! slp_perm SLP case after each group.
19205 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19207 PR middle-end/81463
19208 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
19211 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19213 PR middle-end/81462
19214 * predict.c (set_even_probabilities): Cleanup; do not affect
19215 probabilities that are already known.
19216 (combine_predictions_for_bb): Call even when count is set.
19218 2017-07-18 Nathan Sidwell <nathan@acm.org>
19220 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
19223 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19226 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
19227 optimization for loop niter analysis.
19229 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
19232 * config/avr/avr.c (avr_optimize_casesi): Don't use
19233 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
19235 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
19237 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
19238 body_cost_vec from _vect_peel_extended_info.
19239 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
19240 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
19243 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19245 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
19247 2017-07-18 Richard Biener <rguenther@suse.de>
19249 PR tree-optimization/80620
19250 PR tree-optimization/81403
19251 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
19252 info when re-using a VN table entry.
19254 2017-07-18 Richard Biener <rguenther@suse.de>
19256 PR tree-optimization/81418
19257 * tree-vect-loop.c (vectorizable_reduction): Properly compute
19258 vectype_in. Verify that with lane-reducing reduction operations
19259 we have a single def-use cycle.
19261 2017-07-17 Carl Love <cel@us.ibm.com>
19263 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
19265 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19266 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19267 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19268 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19269 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19270 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19271 VMULOSW): New enum "unspec" values.
19272 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19273 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19274 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19275 altivec_vmulosw): New patterns.
19276 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19277 VMULOSW): Add definitions.
19278 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
19280 * config/alpha/alpha.c: Include predict.h.
19282 2017-07-17 Yury Gribov <tetra2005@gmail.com>
19284 * tree-vrp.c (compare_assert_loc): Fix comparison function
19285 to return predictable results.
19287 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19289 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
19291 (subdi3): Likewise.
19292 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
19293 * doc/invoke.texi (mexpand-adddi): Update text.
19295 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19297 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
19298 that also clobbers the CC register. The old expand code is moved
19300 (*arc_clzsi2): ... here.
19301 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
19302 the CC register. The old expand code is moved to ...
19303 (arc_ctzsi2): ... here.
19305 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19307 * config/arc/arc.opt (mindexed-loads): Use initial value
19308 TARGET_INDEXED_LOADS_DEFAULT.
19309 (mauto-modify-reg): Use initial value
19310 TARGET_AUTO_MODIFY_REG_DEFAULT.
19311 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19312 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19313 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19314 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19316 2017-07-17 Martin Liska <mliska@suse.cz>
19319 * opts.c (finish_options): Do not allow -fgnu-tm
19320 w/ -fsanitize={kernel-,}address. Say sorry.
19322 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19325 * tree-loop-distribution.c (classify_partition): Only assert on
19326 numer of iterations.
19327 (merge_dep_scc_partitions): Delete prameter. Update function call.
19328 (distribute_loop): Remove code handling loop with unknown niters.
19329 (pass_loop_distribution::execute): Skip loop with unknown niters.
19331 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19334 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
19335 function sort_partitions_by_post_order.
19337 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19339 PR tree-optimization/81374
19340 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
19341 the max index of basic blocks, rather than number of basic blocks.
19343 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19345 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
19347 (arc_legitimate_pic_operand_p): Likewise.
19348 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
19350 (arc_needs_pcl_p): Likewise.
19351 (arc_legitimate_pc_offset_p): Likewise.
19352 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
19353 function is also used in constrains.md.
19354 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
19355 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
19356 PLUS. Only return true/false in known cases, otherwise assert.
19357 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
19358 is already called in arc_legitimate_constant_p.
19359 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
19361 (LEGITIMATE_PIC_OPERAND_P): Use
19362 arc_raw_symbolic_reference_mentioned_p function.
19363 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
19368 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19369 Andrew Burgess <andrew.burgess@embecosm.com>
19371 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
19372 (arc_return_address_register): New function.
19373 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
19374 (arc_handle_fndecl_attribute): Add naked attribute.
19375 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19376 (TARGET_WARN_FUNC_RETURN): Likewise.
19377 (arc_allocate_stack_slots_for_args): New function.
19378 (arc_warn_func_return): Likewise.
19379 (machine_function): Change type fn_type.
19380 (arc_compute_function_type): Consider new naked function type,
19381 change function return type.
19382 (arc_must_save_register): Adapt to handle new
19383 arc_compute_function_type's return type.
19384 (arc_expand_prologue): Likewise.
19385 (arc_expand_epilogue): Likewise.
19386 (arc_return_address_regs): Delete.
19387 (arc_return_address_register): New function.
19388 (arc_epilogue_uses): Use above function.
19389 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
19390 (arc_function_type): Change encoding, add naked type.
19391 (ARC_INTERRUPT_P): Change to handle the new encoding.
19392 (ARC_FAST_INTERRUPT_P): Likewise.
19393 (ARC_NORMAL_P): Define.
19394 (ARC_NAKED_P): Likewise.
19395 (arc_compute_function_type): Delete prototype.
19396 * config/arc/arc.md (in_ret_delay_slot): Use
19397 arc_return_address_register function.
19398 (simple_return): Likewise.
19399 (p_return_i): Likewise.
19401 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19403 PR tree-optimization/81428
19404 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
19405 can't be built for those types.
19407 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19409 Remove stuff dead since r239246.
19411 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
19412 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
19413 (avr_inform_devices): Remove dead stuff.
19415 2017-07-17 Tamar Christina <tamar.christina@arm.com>
19417 * config/arm/arm_neon.h: Fix softp typo.
19419 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19421 PR tree-optimization/81365
19422 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
19423 aggregate moves onto bb predecessor edges, make sure there are no
19424 loads that could alias the lhs in between the start of bb and the
19427 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19430 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
19431 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
19432 [LSHIFTRT, outer_code = TRUNCATE]: Same.
19434 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19436 PR tree-optimization/81396
19437 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
19438 (init_symbolic_number): Initialize it to 1.
19439 (perform_symbolic_merge): Add n_ops from both operands into the new
19441 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
19442 without base_addr as useless if they need more than one operation.
19443 (bswap_replace): Handle !bswap case for NULL base_addr.
19445 2017-07-17 Tom de Vries <tom@codesourcery.com>
19448 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
19451 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
19453 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
19454 conditional builtin define __FIX_LEON3FT_B2BST.
19456 2017-07-17 Daniel Cederman <cederman@gaisler.com>
19458 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
19459 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
19462 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
19464 PR rtl-optimization/81424
19465 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
19466 to remove potential trapping from operands if -fnon-call-exceptions.
19468 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19470 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
19471 profile_proability for scalling.
19472 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
19474 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19476 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
19478 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19480 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
19481 fixpoint arithmetics.
19483 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19485 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
19486 fixpoint arithmetics.
19488 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19490 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
19491 fixpoint arithmetics.
19493 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19495 * profile-count.h (profile_probability::from_reg_br_prob_note,
19496 profile_probability::to_reg_br_prob_note): New functions.
19497 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
19498 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
19499 * predict.c (probability_reliable_p): Update.
19500 (edge_probability_reliable_p): Update.
19501 (br_prob_note_reliable_p): Update.
19502 (invert_br_probabilities): Update.
19503 (add_reg_br_prob_note): New function.
19504 (combine_predictions_for_insn): Update.
19505 * asan.c (asan_clear_shadow): Update.
19506 * cfgbuild.c (compute_outgoing_frequencies): Update.
19507 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
19508 (update_br_prob_note): Update.
19509 (rtl_verify_edges): Update.
19510 (purge_dead_edges): Update.
19511 (fixup_reorder_chain): Update.
19512 * emit-rtl.c (try_split): Update.
19513 * ifcvt.c (cond_exec_process_insns): Update.
19514 (cond_exec_process_if_block): Update.
19515 (dead_or_predicable): Update.
19516 * internal-fn.c (expand_addsub_overflow): Update.
19517 (expand_neg_overflow): Update.
19518 (expand_mul_overflow): Update.
19519 * loop-doloop.c (doloop_modify): Update.
19520 * loop-unroll.c (compare_and_jump_seq): Update.
19521 * optabs.c (emit_cmp_and_jump_insn_1): Update.
19522 * predict.h: Update.
19523 * reorg.c (mostly_true_jump): Update.
19525 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
19526 * config/alpha/alpha.c (emit_unlikely_jump): Update.
19527 * config/arc/arc.c: (emit_unlikely_jump): Update.
19528 * config/arm/arm.c: (emit_unlikely_jump): Update.
19529 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
19530 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
19531 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
19532 (ix86_print_operand): Update.
19533 (ix86_split_fp_branch): Update.
19534 (predict_jump): Update.
19535 * config/ia64/ia64.c (ia64_print_operand): Update.
19536 * config/mmix/mmix.c (mmix_print_operand): Update.
19537 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
19538 (rs6000_expand_split_stack_prologue): Update.
19539 * config/rs6000/rs6000.c: Update.
19540 * config/s390/s390.c (s390_expand_vec_strlen): Update.
19541 (s390_expand_vec_movstr): Update.
19542 (s390_expand_cs_tdsi): Update.
19543 (s390_expand_split_stack_prologue): Update.
19544 * config/sh/sh.c (sh_print_operand): Update.
19545 (expand_cbranchsi4): Update.
19546 (expand_cbranchdi4): Update.
19547 * config/sparc/sparc.c (output_v9branch): Update.
19548 * config/spu/spu.c (get_branch_target): Update.
19549 (ea_load_store_inline): Update.
19550 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
19551 * config/tilepro/tilepro.c: Update.
19553 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
19555 * gimplify.c (mostly_copy_tree_r): Revert latest change.
19556 (gimplify_save_expr): Likewise.
19558 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19560 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
19562 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19564 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
19566 * timevar.def (TV_IPA_FNSUMMARY): Define.
19568 2017-07-16 Daniel Cederman <cederman@gaisler.com>
19570 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
19571 to back store errata sensitive sequence from being generated.
19572 (sqrtdf2_fix): Likewise.
19574 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19576 * tree-ssa-threadupdate.c (compute_path_counts,
19577 update_joiner_offpath_counts): Use profile_probability.
19579 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
19582 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19584 * config/arm/arm-c.c (arm_cpu_builtins): Define
19585 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19587 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
19589 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19590 array entries to represent __ieee128 versions of the
19591 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
19592 scalar_extract_sig, and scalar_insert_exp built-in functions.
19593 (altivec_resolve_overloaded_builtin): Add special case handling
19594 for the __builtin_scalar_insert_exp function, as represented by
19595 the P9V_BUILTIN_VEC_VSIEDP constant.
19596 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
19597 exponent support for __ieee128 argument.
19598 (VSESQP): Add scalar extract signature support for __ieee128
19600 (VSTDCNQP): Add scalar test negative support for __ieee128
19602 (VSIEQP): Add scalar insert exponent support for __int128 argument
19603 with __ieee128 result.
19604 (VSIEQPF): Add scalar insert exponent support for __ieee128
19605 argument with __ieee128 result.
19606 (VSTDCQP): Add scalar test data class support for __ieee128
19608 (VSTDCNQP): Add overload support for scalar test negative with
19609 __ieee128 argument.
19610 (VSTDCQP): Add overload support for scalar test data class
19611 __ieee128 argument.
19612 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
19613 UNSPEC_VSX_SXSIGDP.
19614 (UNSPEC_VSX_SIEXPQP): New constant.
19615 (xsxexpqp): New insn for VSX scalar extract exponent quad
19617 (xsxsigqp): New insn for VSX scalar extract significand quad
19619 (xsiexpqpf): New insn for VSX scalar insert exponent quad
19620 precision with floating point argument.
19621 (xststdcqp): New expand for VSX scalar test data class quad
19623 (xststdcnegqp): New expand for VSX scalar test negative quad
19625 (xststdcqp): New insn to match expansions for VSX scalar test data
19626 class quad precision and VSX scalar test negative quad precision.
19627 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
19628 special case operand checking to enforce that second operand of
19629 VSX scalar test data class with quad precision argument is a 7-bit
19631 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19632 prototypes and descriptions of __ieee128 versions of
19633 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
19634 scalar_test_data_class, and scalar_test_neg built-in functions.
19636 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19638 PR tree-optimization/81162
19639 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
19640 replace a negate with an add.
19642 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
19644 * doc/invoke.texi (arm/-mcpu): Document +crypto.
19646 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19648 * config/arm/arm-c.c (arm_cpu_builtins): Define
19649 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19651 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19653 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
19654 (armv8-r): Set ARM Cortex-R52 as default CPU.
19655 * config/arm/arm-tables.opt: Regenerate.
19656 * config/arm/arm-tune.md: Regenerate.
19657 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19659 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
19660 extension for -mcpu=cortex-r52.
19662 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19664 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
19665 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
19666 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
19667 (fp-armv8): Define it as FP_ARMv8 only.
19668 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
19669 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
19671 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
19672 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
19673 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
19674 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
19675 than TARGET_FPU_ARMV8.
19676 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
19677 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
19678 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
19680 * config/arm/neon.md (neon_vrint): Likewise.
19681 (neon_vcvt): Likewise.
19682 (neon_<fmaxmin_op><mode>): Likewise.
19683 (<fmaxmin><mode>3): Likewise.
19684 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
19685 * config/arm/predicates.md (arm_cond_move_operator): Check against
19686 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
19688 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
19690 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
19691 to top of function.
19693 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19695 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
19696 loop in comment with memset.
19698 2017-07-14 Martin Liska <mliska@suse.cz>
19700 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
19701 * dwarf2out.c (is_java): Remove the function.
19702 (output_pubname): Remove usage of the function.
19703 (lower_bound_default): Remove usage of DW_LANG_Java.
19704 (gen_compile_unit_die): Likewise.
19705 * gcc.c: Remove compiler defaults for .java and .zip files.
19706 * gimple-expr.c (remove_suffix): Change as there's no longer
19707 extension than 4-letter one.
19708 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
19709 (gimplify_save_expr): Likewise.
19710 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
19711 as it's possible even for other languages than Java.
19712 * langhooks.h (struct lang_hooks): Remove Java from a comment.
19713 * lto-opts.c (lto_write_options): Remove reference to Java.
19714 * opts.c (strip_off_ending): Update file extension handling.
19715 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
19716 * tree-eh.c (lower_resx): Likewise.
19717 * tree.c (free_lang_data_in_type): Remove dead code.
19718 (find_decls_types_r): Likewise.
19719 (build_common_builtin_nodes): Remove Java from a comment.
19720 (verify_type): Remove dead code.
19721 * varasm.c (assemble_external): Remove Java from a comment.
19723 2017-07-14 Martin Liska <mliska@suse.cz>
19725 * opts.c (finish_options): Add quotes.
19726 (common_handle_option): Likewise.
19728 2017-07-14 Martin Liska <mliska@suse.cz>
19730 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
19731 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
19732 Remove N_SO_PASCAL.
19733 * dwarf2out.c (lower_bound_default): Do not handle
19735 (gen_compile_unit_die): Likewise.
19736 * gcc.c: Remove default extension binding for GNU Pascal.
19737 * stmt.c: Remove Pascal language from a comment.
19738 * xcoffout.c: Likewise.
19740 2017-07-13 David Malcolm <dmalcolm@redhat.com>
19743 * diagnostic-show-locus.c (fixit_cmp): New function.
19744 (layout::layout): Sort m_fixit_hints.
19745 (column_range::column_range): Assert that the values are valid.
19746 (struct char_span): New struct.
19747 (correction::overwrite): New method.
19748 (struct source_line): New struct.
19749 (line_corrections::add_hint): Add assertions. Reimplement memcpy
19750 calls in terms of classes source_line and char_span, and
19751 correction::overwrite.
19752 (selftest::test_overlapped_fixit_printing_2): New function.
19753 (selftest::diagnostic_show_locus_c_tests): Call it.
19755 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
19757 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
19758 early if there is no lhs.
19760 2017-07-13 Martin Liska <mliska@suse.cz>
19762 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
19763 (gen_reference_type_die): Likewise.
19764 * stor-layout.c: Remove Pascal-related comment.
19766 2017-07-13 Martin Liska <mliska@suse.cz>
19768 * opts.c (finish_options): Add quotes to error messages.
19769 (parse_sanitizer_options): Likewise.
19771 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
19773 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
19775 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
19777 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
19779 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
19781 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
19783 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
19785 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
19788 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
19789 provides the hardware capability bits, define the macro
19790 __BUILTIN_CPU_SUPPORTS__.
19791 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
19792 if GLIBC does not provide the hardware capability bits. Add a
19793 gcc_unreachable call if the built-in cpu function is neither
19794 __builtin_cpu_is nor __builtin_cpu_supports.
19795 (rs6000_get_function_versions_dispatcher): Change the warning
19796 that an old GLIBC is used which does not export the capability
19797 bits to be an error.
19798 * doc/extend.texi (target_clones attribute): Document the
19799 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
19800 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
19801 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
19802 the macros defined by GCC if the newer GLIBC is available.
19804 2017-07-12 Jeff Law <law@redhat.com>
19806 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
19807 remaining includes slightly.
19808 * config/riscv/riscv-builtins.c: Include profile-count.h.
19810 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
19813 * config/avr/avr.c (avr_set_current_function): In diagnostic
19814 messages: Quote keywords and (parts of) identifiers.
19815 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
19818 2017-07-12 Carl Love <cel@us.ibm.com>
19820 * config/rs6000/rs6000-c.c: Add support for built-in functions
19821 vector bool char vec_revb (vector bool char);
19822 vector bool short vec_revb (vector short char);
19823 vector bool int vec_revb (vector bool int);
19824 vector bool long long vec_revb (vector bool long long);
19825 * doc/extend.texi: Update the built-in documentation file for the
19826 new built-in functions.
19828 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19830 * config/s390/s390.md: Remove movcc splitter.
19832 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19834 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
19835 load/store on condition.
19837 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
19840 * config/avr/avr.c (avr_encode_section_info)
19841 [progmem && !TREE_READONLY]: Error if progmem object needs
19844 2017-07-11 Michael Collison <michael.collison@arm.com>
19846 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
19849 2017-07-11 Carl Love <cel@us.ibm.com>
19851 * config/rs6000/rs6000-c.c: Add support for builtins
19852 vector unsigned int vec_parity_lsbb (vector signed int);
19853 vector unsigned int vec_parity_lsbb (vector unsigned int);
19854 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
19855 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
19856 vector unsigned long long vec_parity_lsbb (vector signed long long);
19857 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
19858 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
19859 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
19860 * doc/extend.texi: Update the built-in documentation file for the
19861 new built-in functions.
19863 2017-07-11 David Malcolm <dmalcolm@redhat.com>
19865 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
19866 (layout::m_primary_loc): New field.
19867 (layout::layout): Initialize new field. Move location filtering
19868 logic from here to...
19869 (layout::maybe_add_location_range): ...this new method. Add
19870 support for filtering to just the lines already specified by other
19872 (layout::will_show_line_p): New method.
19873 (gcc_rich_location::add_location_if_nearby): New method.
19874 (selftest::test_add_location_if_nearby): New test function.
19875 (selftest::diagnostic_show_locus_c_tests): Call it.
19876 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
19879 2017-07-11 Tom de Vries <tom@codesourcery.com>
19881 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
19882 (bb_first_real_insn): New function.
19883 (nvptx_single): Add extra initialization of broadcasted condition
19886 2017-07-11 Nathan Sidwell <nathan@acm.org>
19888 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
19890 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
19892 * doc/extend.texi (AVR Function Attributes): Remove weblink to
19893 Binutils doc as TEXI will mess them up.
19894 * doc/invoke.texi (AVR Options): Same here.
19896 2017-07-11 Daniel Cederman <cederman@gaisler.com>
19898 * config/sparc/sparc.opt (mfix-ut700): New option.
19899 (mfix-gr712rc): Likewise.
19900 (sparc_fix_b2bst): New variable.
19901 * doc/invoke.texi (SPARC options): Document them.
19902 (ARM options): Fix warnings.
19903 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
19904 instructions to prevent sequences that can trigger the store-store
19905 errata for certain LEON3FT processors.
19906 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
19907 (sparc_option_override): Set sparc_fix_b2bst appropriately.
19908 * config/sparc/sparc.md (fix_b2bst): New attribute.
19909 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
19911 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
19914 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
19916 (*rsqrtsf2_sse): Ditto.
19918 (div<mode>3): Macroize insn from divdf3 and divsf3
19919 using MODEF mode iterator.
19921 2017-07-10 Martin Sebor <msebor@redhat.com>
19923 PR tree-optimization/80397
19924 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
19925 instead of testing for equality to INTEGER_TYPE.
19927 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
19929 * config.gcc: Remove uclibc from arc target spec.
19931 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
19933 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
19935 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19938 * lto-wrapper.c (remove_option): New function.
19939 (merge_and_complain): Merge PIC/PIE options more realistically.
19941 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
19943 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
19947 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
19948 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
19949 * config.in: Regenerate.
19950 * configure: Regenerate.
19951 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
19952 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
19953 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
19954 (TARGET_GASISR_PROLOGUES): ...target mask.
19955 * common/config/avr/avr-common.c
19956 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
19957 Set -mgas-isr-prologues.
19958 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
19959 INSERT_PASS_BEFORE for it.
19960 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
19961 * config/avr/avr.c (avr_option_override)
19962 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
19963 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
19964 (avr_attribute_table) <no_gccisr>: Add new function attribute.
19965 (avr_set_current_function) <is_no_gccisr>: Init machine field.
19966 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
19968 (make_avr_pass_pre_proep): New function.
19969 (emit_push_sfr) <treg>: Add argument to function and use it
19970 instead of TMP_REG.
19971 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
19972 and set machine->gasisr.yes.
19973 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
19974 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
19975 __gcc_isr.n_pushed to .L__stack_usage.
19976 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
19977 (avr_asm_final_postscan_insn): ...this new static function.
19978 * config/avr/avr.h (machine_function)
19979 <is_no_gccisr, use_L__stack_usage>: New fields.
19980 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
19981 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
19982 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
19983 (gasisr, *gasisr): New expander and insn.
19984 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
19985 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
19986 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
19988 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
19990 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
19993 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
19995 Move jump-tables out of .text again.
19998 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
19999 (ASM_OUTPUT_ADDR_VEC): New function.
20000 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
20001 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
20002 INSN_ADDRESSes as asm comment.
20003 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
20004 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
20005 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
20006 * config/avr/avr.md (*tablejump): Adjust comment.
20007 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
20008 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
20010 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
20011 (avr_output_addr_vec): New proto.
20012 (avr_log_t) <insn_addresses>: New field.
20014 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
20017 * config/i386/i386.c (ix86_function_arg_advance): Set
20018 outgoing_args_on_stack to true if there are outgoing arguments
20020 (ix86_function_arg): Likewise.
20021 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
20022 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
20023 * config/i386/i386.h (machine_function): Add
20024 outgoing_args_on_stack.
20026 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
20028 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20029 supporting pthreds.
20030 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
20032 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
20034 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
20035 (REAL_H): Remove $(MACHMODE_H).
20036 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
20038 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
20039 $(MACHMODE_H) and double-int.h.
20040 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
20042 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
20045 2017-07-07 Andrew Pinski <apinski@cavium.com>
20047 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
20048 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
20050 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20052 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
20053 Add warning if GCC was not configured to link against a GLIBC that
20054 exports the hardware capability bits.
20055 (make_resolver_func): Make resolver function private and not a
20056 COMDAT function. Create the name with clone_function_name instead
20057 of make_unique_name.
20060 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
20061 correct operand in doing the split.
20063 2017-07-07 Carl Love <cel@us.ibm.com>
20065 * config/rs6000/rs6000-c: Add support for built-in function
20066 vector unsigned short vec_pack_to_short_fp32 (vector float,
20068 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
20069 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
20070 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
20071 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
20072 (convert_4f32_8i16): Add define_expand.
20073 * doc/extend.texi: Update the built-in documentation file for the
20074 new built-in function.
20076 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20078 * config/sparc/m8.md: New file.
20079 * config/sparc/sparc.md: Include m8.md.
20081 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20083 * config/sparc/sparc.opt: New option -mvis4b.
20084 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
20085 (sparc_option_override): Handle VIS4B.
20086 (enum sparc_builtins): Define
20087 SPARC_BUILTIN_DICTUNPACK{8,16,32},
20088 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
20089 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
20090 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
20091 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
20092 (check_constant_argument): New function.
20093 (sparc_vis_init_builtins): Define builtins
20094 __builtin_vis_dictunpack{8,16,32},
20095 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
20096 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
20097 __builtin_vis_fpcmpde{8,16,32}shl and
20098 __builtin_vis_fpcmpur{8,16,32}shl.
20099 (sparc_expand_builtin): Check that the constant operands to
20100 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
20101 constant and in range.
20102 * config/sparc/sparc-c.c (sparc_target_macros): Handle
20104 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
20105 (SPARC_IMM5_P): Likewise.
20106 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
20107 (enabled): Handle vis4b.
20108 (UNSPEC_DICTUNPACK): New unspec.
20109 (UNSPEC_FPCMPSHL): Likewise.
20110 (UNSPEC_FPUCMPSHL): Likewise.
20111 (UNSPEC_FPCMPDESHL): Likewise.
20112 (UNSPEC_FPCMPURSHL): Likewise.
20113 (cpu_feature): New CPU feature `vis4b'.
20114 (dictunpack{8,16,32}): New insns.
20115 (FPCSMODE): New mode iterator.
20116 (fpcscond): New code iterator.
20117 (fpcsucond): Likewise.
20118 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
20119 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
20120 (fpcmpde{8,16,32}{si,di}shl): Likewise.
20121 (fpcmpur{8,16,32}{si,di}shl): Likewise.
20122 * config/sparc/constraints.md: Define constraints `q' for unsigned
20123 2-bit integer constants and `t' for unsigned 5-bit integer
20125 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
20127 (imm5_operand_dictunpack16): Likewise.
20128 (imm5_operand_dictunpack32): Likewise.
20129 (imm2_operand): Likewise.
20130 * doc/invoke.texi (SPARC Options): Document -mvis4b.
20131 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
20132 ditunpack* and fpcmp*shl builtins.
20134 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20136 * config.gcc: Handle m8 in --with-{cpu,tune} options.
20137 * config.in: Add HAVE_AS_SPARC6 define.
20138 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
20140 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
20142 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
20143 (CPP_CPU_SPEC): Handle m8.
20144 (ASM_CPU_SPEC): Likewise.
20145 * config/sparc/sparc-opts.h (enum processor_type): Add
20147 * config/sparc/sparc.c (m8_costs): New struct.
20148 (sparc_option_override): Handle TARGET_CPU_m8.
20149 (sparc32_initialize_trampoline): Likewise.
20150 (sparc64_initialize_trampoline): Likewise.
20151 (sparc_issue_rate): Likewise.
20152 (sparc_register_move_cost): Likewise.
20153 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
20154 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
20155 (ASM_CPU64_DEFAULT_SPEC): Likewise.
20156 (CPP_CPU_SPEC): Handle M8.
20157 (ASM_CPU_SPEC): Likewise.
20158 (AS_M8_FLAG): Define.
20159 * config/sparc/sparc.md: Add m8 to the cpu attribute.
20160 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
20161 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
20163 * configure: Regenerate.
20164 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
20167 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20169 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
20171 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
20172 ("*movdi_insn_sp32"): Do not set v3pipe.
20173 ("*movsi_insn"): Likewise.
20174 ("*movdi_insn_sp64"): Likewise.
20175 ("*movsf_insn"): Likewise.
20176 ("*movdf_insn_sp32"): Likewise.
20177 ("*movdf_insn_sp64"): Likewise.
20178 ("*zero_extendsidi2_insn_sp64"): Likewise.
20179 ("*sign_extendsidi2_insn"): Likewise.
20180 ("*mov<VM32:mode>_insn"): Likewise.
20181 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20182 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20183 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
20184 ("<vlop:code><VL:mode>3"): Likewise.
20185 ("*not_<vlop:code><VL:mode>3"): Likewise.
20186 ("*nand<VL:mode>_vis"): Likewise.
20187 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
20188 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
20189 ("one_cmpl<VL:mode>2"): Likewise.
20190 ("faligndata<VM64:mode>_vis"): Likewise.
20191 ("alignaddrsi_vis"): Likewise.
20192 ("alignaddrdi_vis"): Likweise.
20193 ("alignaddrlsi_vis"): Likewise.
20194 ("alignaddrldi_vis"): Likewise.
20195 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20196 ("bmaskdi_vis"): Likewise.
20197 ("bmasksi_vis"): Likewise.
20198 ("bshuffle<VM64:mode>_vis"): Likewise.
20199 ("cmask8<P:mode>_vis"): Likewise.
20200 ("cmask16<P:mode>_vis"): Likewise.
20201 ("cmask32<P:mode>_vis"): Likewise.
20202 ("pdistn<P:mode>_vis"): Likewise.
20203 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20205 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20207 * config/sparc/sparc.md ("subtype"): New insn attribute.
20208 ("*wrgsr_sp64"): Set insn subtype.
20209 ("*rdgsr_sp64"): Likewise.
20210 ("alignaddrsi_vis"): Likewise.
20211 ("alignaddrdi_vis"): Likewise.
20212 ("alignaddrlsi_vis"): Likewise.
20213 ("alignaddrldi_vis"): Likewise.
20214 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
20215 ("fexpand_vis"): Likewise.
20216 ("fpmerge_vis"): Likewise.
20217 ("faligndata<VM64:mode>_vis"): Likewise.
20218 ("bshuffle<VM64:mode>_vis"): Likewise.
20219 ("cmask8<P:mode>_vis"): Likewise.
20220 ("cmask16<P:mode>_vis"): Likewise.
20221 ("cmask32<P:mode>_vis"): Likewise.
20222 ("fchksm16_vis"): Likewise.
20223 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
20224 ("fmean16_vis"): Likewise.
20225 ("fp<plusminus_insn>64_vis"): Likewise.
20226 ("<plusminus_insn>v8qi3"): Likewise.
20227 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20228 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
20229 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
20230 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
20231 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
20232 ("*movqi_insn"): Likewise.
20233 ("*movhi_insn"): Likewise.
20234 ("*movsi_insn"): Likewise.
20235 ("movsi_pic_gotdata_op"): Likewise.
20236 ("*movdi_insn_sp32"): Likewise.
20237 ("*movdi_insn_sp64"): Likewise.
20238 ("movdi_pic_gotdata_op"): Likewise.
20239 ("*movsf_insn"): Likewise.
20240 ("*movdf_insn_sp32"): Likewise.
20241 ("*movdf_insn_sp64"): Likewise.
20242 ("*zero_extendhisi2_insn"): Likewise.
20243 ("*zero_extendqihi2_insn"): Likewise.
20244 ("*zero_extendqisi2_insn"): Likewise.
20245 ("*zero_extendqidi2_insn"): Likewise.
20246 ("*zero_extendhidi2_insn"): Likewise.
20247 ("*zero_extendsidi2_insn_sp64"): Likewise.
20248 ("ldfsr"): Likewise.
20249 ("prefetch_64"): Likewise.
20250 ("prefetch_32"): Likewise.
20251 ("tie_ld32"): Likewise.
20252 ("tie_ld64"): Likewise.
20253 ("*tldo_ldub_sp32"): Likewise.
20254 ("*tldo_ldub1_sp32"): Likewise.
20255 ("*tldo_ldub2_sp32"): Likewise.
20256 ("*tldo_ldub_sp64"): Likewise.
20257 ("*tldo_ldub1_sp64"): Likewise.
20258 ("*tldo_ldub2_sp64"): Likewise.
20259 ("*tldo_ldub3_sp64"): Likewise.
20260 ("*tldo_lduh_sp32"): Likewise.
20261 ("*tldo_lduh1_sp32"): Likewise.
20262 ("*tldo_lduh_sp64"): Likewise.
20263 ("*tldo_lduh1_sp64"): Likewise.
20264 ("*tldo_lduh2_sp64"): Likewise.
20265 ("*tldo_lduw_sp32"): Likewise.
20266 ("*tldo_lduw_sp64"): Likewise.
20267 ("*tldo_lduw1_sp64"): Likewise.
20268 ("*tldo_ldx_sp64"): Likewise.
20269 ("*mov<VM32:mode>_insn"): Likewise.
20270 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20271 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20273 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20275 * config/sparc/sparc.md ("type"): New insn type viscmp.
20276 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
20278 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
20279 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
20280 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20281 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
20283 ("n7_vis_logical_11cycle"): Likewise.
20284 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
20285 * config/sparc/niagara2.md ("niag3_vis": Likewise.
20286 * config/sparc/niagara.md ("niag_vis"): Likewise.
20287 * config/sparc/ultra3.md ("us3_fga"): Likewise.
20288 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
20290 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20292 * config/sparc/sparc.md: New instruction type `bmask'.
20293 (bmaskdi_vis): Use the `bmask' type.
20294 (bmasksi_vis): Likewise.
20295 * config/sparc/ultra3.md (us3_array): Likewise.
20296 * config/sparc/niagara7.md (n7_array): Likewise.
20297 * config/sparc/niagara4.md (n4_array): Likewise.
20298 * config/sparc/niagara2.md (niag2_vis): Likewise.
20299 (niag3_vis): Likewise.
20300 * config/sparc/niagara.md (niag_vis): Likewise.
20302 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20304 * ipa-comdats.c: Remove optimize check from gate.
20305 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
20306 for functions not optimized.
20307 (ipa_fn_summary_read): Skip optimize check.
20308 (ipa_fn_summary_write): Likewise.
20309 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
20311 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
20313 (can_inline_edge_p): Check flag_pcc_struct_return for match.
20314 (check_callers): Give up on caller which is not optimized.
20315 (inline_small_functions): Likewise.
20316 (ipa_inline): Do not give up when not optimizing.
20317 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
20318 away unoptimizes cdtors.
20319 (whole_program_function_and_variable_visibility): Do
20320 ipa_discover_readonly_nonaddressable_vars in LTO mode.
20321 * ipa.c (process_references): Do not check optimize.
20322 (symbol_table::remove_unreachable_nodes): Update optimize check.
20323 (set_writeonly_bit): Update optimize check.
20324 (pass_ipa_cdtor_merge::gate): Do not check optimize.
20325 (pass_ipa_single_use::gate): Remove.
20327 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20329 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
20330 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
20331 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
20332 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
20333 permute_load, permute_store, adjust_extract, adjust_splat,
20334 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
20335 replace_swap_with_copy, dump_swap_insn_table,
20336 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
20337 recombine_lvx_pattern, recombine_stvx_pattern,
20338 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
20339 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
20340 to file rs6000-p8swap.c.
20341 * config/rs6000/rs6000-p8swap.c: New file.
20342 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
20343 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
20344 and rs6000*-*-* targets.
20346 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20348 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
20350 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20352 * lto-wrapper.c (merge_and_complain): Do not merge
20353 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
20354 fsigned_zeros, ftrapping_math, fwrapv.
20355 (append_compiler_options): Do not track these options.
20356 (append_linker_options): Likewie
20358 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20360 * cgraphunit.c (cgraph_node::finalize_function): When
20361 !flag_toplevel_reorde set no_reorder flag.
20362 (varpool_node::finalize_decl): Likewise.
20363 (symbol_table::compile): Drop no toplevel reorder path.
20365 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20367 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
20368 edges; zero probability is not better than uninitialized.
20370 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20372 * asan.h (asan_sanitize_allocas_p): Declare.
20373 * asan.c (asan_sanitize_allocas_p): New function.
20374 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
20375 (handle_builtin_alloca): Likewise.
20376 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
20377 if !asan_sanitize_allocas_p.
20378 * params.def (asan-instrument-allocas): Add new option.
20379 * params.h (ASAN_PROTECT_ALLOCAS): Define.
20380 * opts.c (common_handle_option): Disable allocas sanitization for
20383 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20385 * asan.c: Include gimple-fold.h.
20386 (get_last_alloca_addr): New function.
20387 (handle_builtin_stackrestore): Likewise.
20388 (handle_builtin_alloca): Likewise.
20389 (asan_emit_allocas_unpoison): Likewise.
20390 (get_mem_refs_of_builtin_call): Add new parameter, remove const
20391 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
20392 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
20393 (instrument_builtin_call): Pass gimple iterator to
20394 get_mem_refs_of_builtin_call.
20395 (last_alloca_addr): New global.
20396 * asan.h (asan_emit_allocas_unpoison): Declare.
20397 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
20398 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
20399 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
20400 if function calls alloca.
20401 * gimple-fold.c (replace_call_with_value): Remove static keyword.
20402 * gimple-fold.h (replace_call_with_value): Declare.
20403 * internal-fn.c: Include asan.h.
20404 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
20405 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
20407 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20409 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
20410 (C_SELFTEST_FLAGS): New.
20411 (CPP_SELFTEST_FLAGS): New.
20412 (SELFTEST_DEPS): New, from deps of s-selftest.
20413 (C_SELFTEST_DEPS): New, from deps of s-selftest.
20414 (CPP_SELFTEST_DEPS): New.
20415 (selftest): Add dependency on s-selftest-c++.
20416 (s-selftest): Rename to...
20417 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
20418 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
20419 than SELFTEST_FLAGS.
20420 (selftest-gdb): Rename to...
20421 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
20423 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
20424 (selftest-valgrind): Rename to...
20425 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
20427 (selftest-valgrind): Reintroduce as an alias for
20428 selftest-c-valgrind.
20429 (s-selftest-c++): New.
20430 (selftest-c++-gdb): New.
20431 (selftest-c++-valgrind): New.
20433 2017-07-06 Olivier Hainque <hainque@adacore.com>
20435 * gcc.c (process_command): When deciding if undefined variables
20436 should be ignored when processing specs, accept "gcc -v" as well.
20438 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20440 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
20441 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
20443 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20445 * config/arm/arm-cpus.in (armv8-r): Add new entry.
20446 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
20447 * config/arm/arm-tables.opt: Regenerate.
20448 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
20450 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
20452 2017-07-06 Carl Love <cel@us.ibm.com>
20454 * ChangeLog: Clean up from mid air collision
20456 2017-07-06 Carl Love <cel@us.ibm.com>
20458 * config/rs6000/rs6000-c.c: Add support for built-in functions
20459 vector signed int vec_subc (vector signed int, vector signed int);
20460 vector signed __int128 vec_subc (vector signed __int128,
20461 vector signed __int128);
20462 vector unsigned __int128 vec_subc (vector unsigned __int128,
20463 vector unsigned __int128);
20464 vector signed int vec_sube (vector signed int, vector signed int,
20465 vector signed int);
20466 vector unsigned int vec_sube (vector unsigned int,
20467 vector unsigned int,
20468 vector unsigned int);
20469 vector signed __int128 vec_sube (vector signed __int128,
20470 vector signed __int128,
20471 vector signed__int128);
20472 vector unsigned __int128 vec_sube (vector unsigned __int128,
20473 vector unsigned __int128,
20474 vector unsigned __int128);
20475 vector signed int vec_subec (vector signed int, vector signed int,
20476 vector signed int);
20477 vector unsigned int vec_subec (vector unsigned int,
20478 vector unsigned int,
20479 vector unsigned int);
20480 vector signed __int128 vec_subec (vector signed __int128,
20481 vector signed __int128,
20482 vector signed__int128);
20483 vector unsigned __int128 vec_subec (vector unsigned __int128,
20484 vector unsigned __int128,
20485 vector unsigned __int128);
20486 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
20487 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
20488 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
20489 BU_ALTIVEC_OVERLOAD_X definitions.
20490 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
20491 * doc/extend.texi: Update the built-in documentation file for the new
20492 built-in functions.
20494 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20497 * diagnostic-show-locus.c (layout::layout): Use start and finish
20498 spelling location for the start and finish of each range.
20499 * genmatch.c (linemap_client_expand_location_to_spelling_point):
20500 Add unused aspect param.
20501 * input.c (expand_location_1): Add "aspect" param, and use it
20502 to access the correct part of the location.
20503 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
20505 (expand_location_to_spelling_point): Likewise.
20506 (linemap_client_expand_location_to_spelling_point): Add "aspect"
20507 param, and pass it to expand_location_1.
20509 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
20511 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
20512 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
20513 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
20514 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
20515 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
20516 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
20517 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
20518 _mm_maskz_getmant_ss): New intrinsics.
20519 (__builtin_ia32_getexpss128_mask): Changed to ...
20520 __builtin_ia32_getexpss128_round ... this.
20521 (__builtin_ia32_getexpsd128_mask): Changed to ...
20522 __builtin_ia32_getexpsd128_round ... this.
20523 * config/i386/i386-builtin-types.def
20524 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
20525 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
20526 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
20527 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
20528 __builtin_ia32_getmantss_mask_round): New builtins.
20529 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
20530 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
20531 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
20532 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
20533 * config/i386/sse.md
20534 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
20535 avx512f_sgetexp<mode><mask_scalar_name>
20536 <round_saeonly_scalar_name> ... this.
20537 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
20538 %0, %1, %2<round_saeonly_op3>}): Changed to ...
20539 vgetexp<ssescalarmodesuffix>
20540 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20541 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
20542 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
20543 avx512f_vgetmant<mode><mask_scalar_name>
20544 <round_saeonly_scalar_name> ... this.
20545 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
20546 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
20547 vgetmant<ssescalarmodesuffix>
20548 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
20549 %0<mask_scalar_operand4>, %1, %2
20550 <round_saeonly_scalar_mask_op4>, %3} ... this.
20551 * config/i386/subst.md (mask_scalar_operand4,
20552 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
20553 round_saeonly_scalar_nimm_predicate): New subst attributes.
20555 2017-07-06 Julia Koval <julia.koval@intel.com>
20557 * config/i386/i386.c (ix86_erase_embedded_rounding):
20558 Remove code for old rounding pattern.
20560 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
20562 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
20564 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
20566 * doc/sourcebuild.texi (Test Directives, Variants of
20567 dg-require-support): Add documentation for dg-require-stack-check.
20569 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
20571 * config/i386/subst.md (mask_scalar, round_scalar,
20572 round_saeonly_scalar): New meta-templates.
20573 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
20574 round_scalar_mask_operand3, round_scalar_mask_op3,
20575 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
20576 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
20577 round_saeonly_scalar_constraint,
20578 round_saeonly_scalar_prefix): New subst attribute.
20579 * config/i386/sse.md
20580 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
20581 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
20582 <round_scalar_name> ... this.
20583 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
20584 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
20585 <round_scalar_name> ... this.
20586 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
20587 <sse>_vm<code><mode>3<mask_scalar_name>
20588 <round_saeonly_scalar_name> ... this.
20589 (v<plusminus_mnemonic><ssescalarmodesuffix>
20590 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20591 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20592 v<plusminus_mnemonic><ssescalarmodesuffix>
20593 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20594 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20595 (v<multdiv_mnemonic><ssescalarmodesuffix>
20596 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20597 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20598 v<multdiv_mnemonic><ssescalarmodesuffix>
20599 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20600 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20601 (v<maxmin_float><ssescalarmodesuffix>
20602 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
20603 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
20604 v<maxmin_float><ssescalarmodesuffix>
20605 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20606 %0<mask_scalar_operand3>, %1, %<iptr>2
20607 <round_saeonly_scalar_mask_op3>} ... this.
20609 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
20611 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
20612 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
20614 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
20615 Alan Hayward <alan.hayward@arm.com>
20616 David Sherwood <david.sherwood@arm.com>
20618 * combine.c (simplify_if_then_else): Remove "enum" before
20620 * compare-elim.c (can_eliminate_compare): Likewise.
20621 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
20623 (aarch64_lookup_simd_builtin_type): Likewise.
20624 (aarch64_simd_builtin_type): Likewise.
20625 (aarch64_init_simd_builtin_types): Likewise.
20626 (aarch64_simd_expand_args): Likewise.
20627 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
20629 (aarch64_reverse_mask): Likewise.
20630 (aarch64_simd_emit_reg_reg_move): Likewise.
20631 (aarch64_gen_adjusted_ldpstp): Likewise.
20632 (aarch64_ccmp_mode_to_code): Likewise.
20633 (aarch64_operands_ok_for_ldpstp): Likewise.
20634 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20635 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
20637 (aarch64_min_divisions_for_recip_mul): Likewise.
20638 (aarch64_reassociation_width): Likewise.
20639 (aarch64_get_condition_code_1): Likewise.
20640 (aarch64_simd_emit_reg_reg_move): Likewise.
20641 (aarch64_simd_attr_length_rglist): Likewise.
20642 (aarch64_reverse_mask): Likewise.
20643 (aarch64_operands_ok_for_ldpstp): Likewise.
20644 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20645 (aarch64_gen_adjusted_ldpstp): Likewise.
20646 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
20648 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
20649 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
20650 (arm_lookup_simd_builtin_type): Likewise.
20651 (arm_simd_builtin_type): Likewise.
20652 (arm_init_simd_builtin_types): Likewise.
20653 (arm_expand_builtin_args): Likewise.
20654 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
20655 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
20656 (ft32_setup_incoming_varargs): Likewise.
20657 (ft32_function_arg): Likewise.
20658 (ft32_function_arg_advance): Likewise.
20659 (ft32_pass_by_reference): Likewise.
20660 (ft32_arg_partial_bytes): Likewise.
20661 (ft32_valid_pointer_mode): Likewise.
20662 (ft32_addr_space_pointer_mode): Likewise.
20663 (ft32_addr_space_legitimate_address_p): Likewise.
20664 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
20666 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
20667 (ix86_emit_outlined_ms2sysv_restore): Likewise.
20668 (iamcu_alignment): Likewise.
20669 (canonicalize_vector_int_perm): Likewise.
20670 (ix86_noce_conversion_profitable_p): Likewise.
20671 (ix86_mpx_bound_mode): Likewise.
20672 (ix86_operands_ok_for_move_multiple): Likewise.
20673 * config/microblaze/microblaze-protos.h
20674 (microblaze_expand_conditional_branch_reg): Likewise.
20675 * config/microblaze/microblaze.c
20676 (microblaze_expand_conditional_branch_reg): Likewise.
20677 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20679 (rs6000_reassociation_width): Likewise.
20680 (rs6000_invalid_binary_op): Likewise.
20681 (fusion_p9_p): Likewise.
20682 (emit_fusion_p9_load): Likewise.
20683 (emit_fusion_p9_store): Likewise.
20684 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
20686 (riscv_hard_regno_mode_ok_p): Likewise.
20687 (riscv_address_insns): Likewise.
20688 (riscv_split_symbol): Likewise.
20689 (riscv_legitimize_move): Likewise.
20690 (riscv_function_value): Likewise.
20691 (riscv_hard_regno_nregs): Likewise.
20692 (riscv_expand_builtin): Likewise.
20693 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
20694 (riscv_build_integer): Likewise.
20695 (riscv_split_integer): Likewise.
20696 (riscv_legitimate_constant_p): Likewise.
20697 (riscv_cannot_force_const_mem): Likewise.
20698 (riscv_regno_mode_ok_for_base_p): Likewise.
20699 (riscv_valid_base_register_p): Likewise.
20700 (riscv_valid_offset_p): Likewise.
20701 (riscv_valid_lo_sum_p): Likewise.
20702 (riscv_classify_address): Likewise.
20703 (riscv_legitimate_address_p): Likewise.
20704 (riscv_address_insns): Likewise.
20705 (riscv_load_store_insns): Likewise.
20706 (riscv_force_binary): Likewise.
20707 (riscv_split_symbol): Likewise.
20708 (riscv_force_address): Likewise.
20709 (riscv_legitimize_address): Likewise.
20710 (riscv_move_integer): Likewise.
20711 (riscv_legitimize_const_move): Likewise.
20712 (riscv_legitimize_move): Likewise.
20713 (riscv_address_cost): Likewise.
20714 (riscv_subword): Likewise.
20715 (riscv_output_move): Likewise.
20716 (riscv_canonicalize_int_order_test): Likewise.
20717 (riscv_emit_int_order_test): Likewise.
20718 (riscv_function_arg_boundary): Likewise.
20719 (riscv_pass_mode_in_fpr_p): Likewise.
20720 (riscv_pass_fpr_single): Likewise.
20721 (riscv_pass_fpr_pair): Likewise.
20722 (riscv_get_arg_info): Likewise.
20723 (riscv_function_arg): Likewise.
20724 (riscv_function_arg_advance): Likewise.
20725 (riscv_arg_partial_bytes): Likewise.
20726 (riscv_function_value): Likewise.
20727 (riscv_pass_by_reference): Likewise.
20728 (riscv_setup_incoming_varargs): Likewise.
20729 (riscv_print_operand): Likewise.
20730 (riscv_elf_select_rtx_section): Likewise.
20731 (riscv_save_restore_reg): Likewise.
20732 (riscv_for_each_saved_reg): Likewise.
20733 (riscv_register_move_cost): Likewise.
20734 (riscv_hard_regno_mode_ok_p): Likewise.
20735 (riscv_hard_regno_nregs): Likewise.
20736 (riscv_class_max_nregs): Likewise.
20737 (riscv_memory_move_cost): Likewise.
20738 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
20739 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
20740 (rl78_addr_space_address_mode): Likewise.
20741 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20743 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20744 (rs6000_reassociation_width): Likewise.
20745 (rs6000_invalid_binary_op): Likewise.
20746 (fusion_p9_p): Likewise.
20747 (emit_fusion_p9_load): Likewise.
20748 (emit_fusion_p9_store): Likewise.
20749 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
20750 (ok_for_simple_move_operands): Likewise.
20751 (ok_for_simple_move_strict_operands): Likewise.
20752 (ok_for_simple_arith_logic_operands): Likewise.
20753 (visium_legitimize_reload_address): Likewise.
20754 (visium_select_cc_mode): Likewise.
20755 (output_cbranch): Likewise.
20756 (visium_split_double_move): Likewise.
20757 (visium_expand_copysign): Likewise.
20758 (visium_expand_int_cstore): Likewise.
20759 (visium_expand_fp_cstore): Likewise.
20760 * config/visium/visium.c (visium_pass_by_reference): Likewise.
20761 (visium_function_arg): Likewise.
20762 (visium_function_arg_advance): Likewise.
20763 (visium_libcall_value): Likewise.
20764 (visium_setup_incoming_varargs): Likewise.
20765 (visium_legitimate_constant_p): Likewise.
20766 (visium_legitimate_address_p): Likewise.
20767 (visium_legitimize_address): Likewise.
20768 (visium_secondary_reload): Likewise.
20769 (visium_register_move_cost): Likewise.
20770 (visium_memory_move_cost): Likewise.
20771 (prepare_move_operands): Likewise.
20772 (ok_for_simple_move_operands): Likewise.
20773 (ok_for_simple_move_strict_operands): Likewise.
20774 (ok_for_simple_arith_logic_operands): Likewise.
20775 (visium_function_value_1): Likewise.
20776 (rtx_ok_for_offset_p): Likewise.
20777 (visium_legitimize_reload_address): Likewise.
20778 (visium_split_double_move): Likewise.
20779 (visium_expand_copysign): Likewise.
20780 (visium_expand_int_cstore): Likewise.
20781 (visium_expand_fp_cstore): Likewise.
20782 (visium_split_cstore): Likewise.
20783 (visium_select_cc_mode): Likewise.
20784 (visium_split_cbranch): Likewise.
20785 (output_cbranch): Likewise.
20786 (visium_print_operand_address): Likewise.
20787 * expmed.c (flip_storage_order): Likewise.
20788 * expmed.h (emit_cstore): Likewise.
20789 (flip_storage_order): Likewise.
20790 * genrecog.c (validate_pattern): Likewise.
20791 * hsa-gen.c (gen_hsa_addr): Likewise.
20792 * internal-fn.c (expand_arith_overflow): Likewise.
20793 * ira-color.c (allocno_copy_cost_saving): Likewise.
20794 * lra-assigns.c (find_hard_regno_for_1): Likewise.
20795 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
20796 (process_invariant_for_inheritance): Likewise.
20797 * lra-eliminations.c (move_plus_up): Likewise.
20798 * omp-low.c (lower_oacc_reductions): Likewise.
20799 * simplify-rtx.c (simplify_subreg): Likewise.
20800 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
20801 (TARGET_CHKP_BOUND_MODE): Likewise..
20802 * targhooks.c (default_chkp_bound_mode): Likewise.
20803 (default_setup_incoming_vararg_bounds): Likewise.
20804 * targhooks.h (default_chkp_bound_mode): Likewise.
20805 (default_setup_incoming_vararg_bounds): Likewise.
20806 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
20807 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
20808 (have_whole_vector_shift): Likewise.
20809 * tree-vect-stmts.c (vectorizable_load): Likewise.
20810 * doc/tm.texi: Regenerate.
20812 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
20814 Graceful degrade if Binutils PR21472 is not available.
20817 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
20818 .rodata in flash test fails.
20819 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
20820 * confgure: Regenerate.
20821 * config.in: Regenerate.
20822 * config/avr/avr.c (avr_asm_named_section)
20823 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
20824 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
20825 (avr_asm_init_sections): Same.
20827 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20829 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
20830 (fma<VH:mode>4_intrinsic): Likewise.
20831 (*fmsub<VCVTF:mode>4): Likewise.
20832 (*fmsub<VH:mode>4_intrinsic): Likewise.
20834 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
20837 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
20838 Don't depend on "optimize > 0".
20839 (out_movhi_r_mr, out_movqi_mr_r): Same.
20840 (out_movhi_mr_r, out_movqi_r_mr): Same.
20841 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
20842 io_address_operand on "optimize > 0".
20844 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20846 * tree-loop-distribution.c: Add general explanantion on the pass.
20847 (generate_loops_for_partition): Mark distributed loop.
20848 (pg_add_dependence_edges): New parameter. Handle alias data
20849 dependence specially and record it in the parameter if asked.
20850 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
20851 (init_partition_graph_vertices, add_partition_graph_edge): New.
20852 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
20853 (free_partition_graph_vdata, build_partition_graph): New.
20854 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
20855 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
20856 (data_ref_segment_size, latch_dominated_by_data_ref): New.
20857 (compute_alias_check_pairs, version_loop_by_alias_check): New.
20858 (version_for_distribution_p, finalize_partitions): New.
20859 (distribute_loop): Handle alias data dependence specially. Factor
20860 out loop fusion code as functions and call these functions.
20862 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20864 * tree-loop-distribution.c (classify_partition): New parameter and
20865 better handle reduction statement.
20866 (rdg_build_partitions): Revise comment.
20867 (distribute_loop): Compute statements in all partitions and pass it
20868 to classify_partition.
20870 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20872 * tree-loop-distribution.c (enum partition_type): New.
20873 (struct partition): New field type.
20874 (partition_merge_into): Add parameter. Update partition type.
20875 (data_dep_in_cycle_p, update_type_for_merge): New functions.
20876 (build_rdg_partition_for_vertex): Compute partition type.
20877 (rdg_build_partitions): Dump partition type.
20878 (distribute_loop): Update calls to partition_merge_into.
20880 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20882 * tree-loop-distribution.c (struct ddr_hasher): New.
20883 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
20885 (classify_partition): Call get_data_dependence.
20886 (pg_add_dependence_edges): Ditto.
20887 (distribute_loop): Release data dependence hash table.
20889 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20891 * tree-loop-distribution.c (ref_base_address): Delete.
20892 (similar_memory_accesses): Rename ...
20893 (share_memory_accesses): ... to this. Check if partitions access
20894 the same memory reference.
20895 (distribute_loop): Call share_memory_accesses.
20897 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20899 * tree-loop-distribution.c (struct partition): New field recording
20900 its data reference.
20901 (partition_alloc, partition_free): Init and release data refs.
20902 (partition_merge_into): Merge data refs.
20903 (build_rdg_partition_for_vertex): Collect data refs for partition.
20904 (pg_add_dependence_edges): Change parameters from vector to bitmap.
20906 (distribute_loop): Remve data refs from vertice data of partition
20909 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20911 * tree-loop-distribution.c (params.h): Include header file.
20912 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
20913 (datarefs_vec): New global var.
20914 (create_rdg_vertices): Use datarefs_vec directly.
20915 (free_rdg): Don't free data references.
20916 (build_rdg): Update use. Don't free data references.
20917 (distribute_loop): Compute global variable for data references.
20918 Bail out if there are too many data references.
20920 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20922 * tree-loop-distribution.c (loop_nest): New global var.
20923 (build_rdg): Use loop directly, rather than loop nest.
20924 (pg_add_dependence_edges): Remove loop nest parameter. Use global
20926 (distribute_loop): Compute global variable loop nest. Update use.
20928 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20930 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
20931 (partition_merge_into): New parameter. Dump reason for fusion.
20932 (distribute_loop): Update use of partition_merge_into.
20934 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20936 * tree-loop-distribution.c (bb_top_order_index): New.
20937 (bb_top_order_index_size, bb_top_order_cmp): New.
20938 (stmts_from_loop): Use topological order.
20939 (pass_loop_distribution::execute): Compute and release topological
20940 order for basic blocks.
20942 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20944 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
20947 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20949 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
20950 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
20951 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
20952 * internal-fn.def (LOOP_DIST_ALIAS): New.
20953 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
20954 (fold_loop_internal_call): ... this.
20955 (vect_loop_dist_alias_call): New function.
20956 (set_uid_loop_bbs): Call fold_loop_internal_call.
20957 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
20960 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
20963 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
20964 Require dead FLAGS_REG at the beginning of a peephole.
20966 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
20969 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
20970 arguments in the call to __builtin_ia32_sbb_u32.
20971 (_subborrow_u64): Swap _X and _Y arguments in the call to
20972 __builtin_ia32_sbb_u64.
20974 2017-07-04 Jakub Jelinek <jakub@redhat.com>
20977 * tree-vrp.c (compare_assert_loc): Turn into a function template
20978 with stable template parameter. Only test if a->e is NULL,
20979 !a->e == !b->e has been verified already. Use e == NULL or
20980 e != NULL instead of e or ! e tests. If stable is true, don't use
20981 iterative_hash_expr, on the other side allow a or b or both NULL
20982 and sort the NULLs last.
20983 (process_assert_insertions): Sort using compare_assert_loc<false>
20984 instead of compare_assert_loc, later sort using
20985 compare_assert_loc<true> before calling process_assert_insertions_for
20986 in a loop. Use break instead of continue once seen NULL pointer.
20988 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20990 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20991 Cortex-R7 and Cortex-R8 processors.
20993 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20995 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
20996 uninitialized while src is not.
20998 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
21000 * common/config/arm/arm-common.c: Adjust include path for
21002 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
21003 (arm-cpu.h): Create in build directory. Adjust dependency rules.
21004 (arm-cpu-data.h): Likewise.
21005 (arm-cpu-cdata.h): Likewise.
21006 * config/arm/arm-cpu.h: Delete.
21007 * config/arm/arm-cpu-cdata.h: Delete.
21008 * config/arm/arm-cpu-data.h: Delete.
21010 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
21012 * config/arm/arm-cpus.in (cortex-a55): New.
21013 (cortex-a75): Likewise.
21014 (cortex-a75.cortex-a55): Likewise.
21015 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
21017 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
21018 * config/arm/arm-cpu-cdata.h: Regenerate.
21019 * config/arm/arm-cpu-data.h: Regenerate.
21020 * config/arm/arm-cpu.h: Regenerate.
21021 * config/arm/arm-tables.opt: Regenerate.
21022 * config/arm/arm-tune.md: Regenerate.
21024 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21026 * haifa-sched.c (sched_create_recovery_edges): Update profile.
21028 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21030 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
21033 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
21035 PR tree-optimization/81292
21036 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
21037 full_string_p, also call adjust_related_strinfos if the adjustment
21038 is simple, otherwise invalidate related strinfos.
21040 2017-07-04 Martin Liska <mliska@suse.cz>
21043 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
21044 newly created variable as DECL_IGNORED_P.
21046 2017-07-04 Martin Liska <mliska@suse.cz>
21049 * ipa-inline.c (inline_small_functions):
21050 Use xstrdup_for_dump.
21052 2017-07-04 Tom de Vries <tom@codesourcery.com>
21054 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
21056 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
21059 * config/darwin.c (darwin_function_switched_text_sections):
21062 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
21064 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
21066 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
21068 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
21070 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21072 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
21073 min_profitable_iters, and th as inclusive lower bounds.
21074 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
21075 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
21076 for min_profitable_iters and min_profitable_estimate.
21077 (vect_transform_loop): Treat th as an inclusive lower bound.
21078 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
21080 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
21083 * config/darwin.c (darwin_function_switched_text_sections):
21084 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
21085 in two pieces, and suppress the use of buf.
21087 2017-07-03 Nathan Sidwell <nathan@acm.org>
21089 * hash-table.h (hash_table_mod1): Fix indentation.
21091 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21093 PR middle-end/81290
21094 * predict.c (force_edge_cold): Be more careful about propagation
21096 * profile-count.h (profile_probability::guessed,
21097 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
21099 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
21101 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
21103 * doc/invoke.texi (rcpc architecture extension): Document it.
21105 2017-07-03 Richard Biener <rguenther@suse.de>
21107 PR tree-optimization/60510
21108 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
21109 the scalar reduction PHI and use it.
21110 (vectorizable_reduction): Properly guard the single_defuse_cycle
21111 path for non-SLP reduction chains where we cannot use it.
21112 Rework reduc_def/index and vector type deduction. Rework
21113 vector operand gathering during reduction op code-gen.
21114 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
21115 chains dissolve the chain and leave it to non-SLP reduction
21118 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21120 * tree-data-ref.h (dr_alignment): Declare.
21121 * tree-data-ref.c (dr_alignment): New function.
21122 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
21123 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
21125 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
21127 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21129 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
21130 and base_misalignment fields.
21131 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
21132 * tree-data-ref.c: Include builtins.h.
21133 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
21134 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
21135 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
21136 * tree-vect-data-refs.c: Include tree-cfg.h.
21137 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
21138 fields instead of calculating an alignment here.
21139 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
21140 innermost_loop_behavior fields.
21142 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21144 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
21146 (DR_STEP_ALIGNMENT): New macro.
21147 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
21148 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
21149 (create_data_ref): Print it.
21150 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
21151 to tell whether the step preserves vector (mis)alignment.
21152 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
21153 Move the check for an integer step and generalise to all INTEGER_CST.
21154 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
21155 Print the outer step alignment.
21157 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21159 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
21160 with offset_alignment.
21161 (DR_ALIGNED_TO): Delete.
21162 (DR_OFFSET_ALIGNMENT): New macro.
21163 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
21164 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
21165 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
21166 (create_data_ref): Likewise.
21167 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
21168 (vect_analyze_data_refs): Likewise.
21169 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
21170 creating dummy innermost behavior.
21172 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21174 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
21175 with a "innermost_loop_behavior *" and refeence tree.
21176 * tree-data-ref.c (dr_analyze_innermost): Likewise.
21177 (create_data_ref): Update call accordingly.
21178 * tree-predcom.c (find_looparound_phi): Likewise.
21180 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21182 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
21183 fields with dr_wrt_vec_loop.
21184 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
21185 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
21186 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
21187 (vect_dr_behavior): New function.
21188 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
21189 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
21190 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
21191 track whether the step preserves the misalignment.
21192 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
21193 Use vect_dr_behavior.
21194 (vect_setup_realignment): Update call accordingly.
21195 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
21196 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
21197 call to vect_create_addr_base_for_vector_ref.
21198 (vect_create_cond_for_align_checks): Likewise.
21199 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
21200 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
21201 (vect_recog_mask_conversion_pattern): Likewise.
21202 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
21203 (new_stmt_vec_info): Remove redundant zeroing.
21205 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
21207 * common/config/arm/arm-common.c (arm_be8_option): New function.
21208 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
21209 (ISA_ARMv6): Add isa_bit_be8.
21210 * config/arm/arm.h (arm_be8_option): Add prototype.
21211 (BE8_SPEC_FUNCTION): New define.
21212 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
21213 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
21214 (mlittle-endian): Similarly.
21215 (mbe8, mbe32): New options.
21216 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
21217 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
21219 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21221 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
21223 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21225 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
21226 (cleanup_tree_cfg_bb): Use it.
21227 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
21229 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
21231 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21234 * loop-doloop.c (add_test): Update profile.
21236 2017-07-03 Martin Liska <mliska@suse.cz>
21239 * sanopt.c (rewrite_usage_of_param): New function.
21240 (sanitize_rewrite_addressable_params): Likewise.
21241 (pass_sanopt::execute): Call rewrite_usage_of_param.
21243 2017-07-03 Richard Biener <rguenther@suse.de>
21245 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
21246 back to using VIEW_CONVERT_EXPR.
21248 2017-07-03 Martin Liska <mliska@suse.cz>
21251 * doc/extend.texi: Document when a resolver function is
21252 generated for target_clones.
21254 2017-07-03 Martin Liska <mliska@suse.cz>
21256 * asan.c (asan_emit_stack_protection): Unpoison just red zones
21257 and shadow memory of auto variables which are subject of
21258 use-after-scope sanitization.
21259 (asan_expand_mark_ifn): Add do set only when is_poison.
21261 2016-07-03 Richard Biener <rguenther@suse.de>
21263 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
21265 (vect_force_simple_reduction): Record reduction def -> phi mapping.
21266 (vectorizable_reduction): Perform reduction PHI creation when
21267 visiting a reduction PHI and adjust and simplify code generation
21268 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
21269 (vect_transform_loop): Visit reduction PHIs.
21270 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
21271 defs into the SLP tree.
21272 (vect_build_slp_tree): Reduction defs terminate the recursion.
21273 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
21275 (vect_get_vec_defs_for_stmt_copy): Export.
21276 (vect_get_vec_defs): Likewise.
21277 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
21279 (vect_get_vec_defs_for_stmt_copy): Declare.
21280 (vect_get_vec_defs): Likewise.
21282 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21284 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
21285 parameter with a "loop" parameter and use it instead of the
21286 loop containing DR_STMT. Don't check simple_iv when doing
21287 BB analysis. Describe the two analysis modes in the comment.
21289 2017-07-03 Tom de Vries <tom@codesourcery.com>
21291 PR tree-optimization/69468
21292 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
21293 (find_same_succ_bb): Handle ignore_edge_flags.
21295 2017-07-03 Tom de Vries <tom@codesourcery.com>
21297 PR tree-optimization/81192
21298 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
21300 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
21302 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
21304 2017-07-03 Tom de Vries <tom@codesourcery.com>
21306 PR tree-optimization/81192
21307 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
21308 BB_SAME_SUCC (bb) == NULL.
21310 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21312 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
21315 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21317 * dumpfile.c: Include profile-count.h
21318 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
21320 (insert_cond_bb): Update profile.
21321 * tree-cfg.h (insert_cond_bb): Update prototype.
21322 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
21323 * tree-dump.c: Do not include tree-cfg.
21325 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21327 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
21329 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21331 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
21334 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21336 * tree-complex.c (expand_complex_div_wide): update profile.
21338 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21339 Alan Hayward <alan.hayward@arm.com>
21340 David Sherwood <david.sherwood@arm.com>
21342 * Makefile.in (MACHMODE_H): Remove insn-modes.h
21343 (CORETYPES_H): New define.
21344 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
21345 (insn-modes-inline.h, s-modes-inline-h): New rules.
21346 (generated_files): Add insn-modes-inline.h.
21347 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
21348 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
21349 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
21350 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
21351 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
21352 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
21353 (build/gencodes.o, build/genconditions.o): Likewise.
21354 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
21355 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
21356 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
21357 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
21358 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
21359 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
21360 * coretypes.h: Include everything up to real.h for generators.
21361 Include insn-modes.h first. Include wide-int-print.h after
21362 wide-int.h. Include insn-modes-inline.h and then machmode.h.
21363 * machmode.h: Don't include insn-modes.h here.
21364 * function-tests.c: Remove includes of signop.h, machmode.h,
21365 double-int.h and wide-int.h.
21367 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
21369 * optc-save-gen.awk: Likewise.
21370 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
21371 * godump.c: Remove include of wide-int-print.h.
21372 * pretty-print.h: Likewise.
21373 * wide-int-print.cc: Likewise.
21374 * wide-int.cc: Likewise.
21375 * hash-map-tests.c: Remove include of signop.h.
21376 * hash-set-tests.c: Likewise.
21377 * rtl-tests.c: Likewise.
21378 * mkconfig.sh: Remove include of machmode.h.
21379 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
21381 (emit_insn_modes_inline_h): ...this new function. Emit the code
21382 into an insn-modes-inline.h header file, adding appropriate
21383 include guards and end comments.
21384 (emit_insn_modes_c_header): Remove include of machmode.h.
21385 (emit_min_insn_modes_c_header): Include coretypes.h rather than
21387 (main): Handle -i flag and call emit_insn_modes_inline_h when
21390 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21392 * tree-ssa-strlen.c (strinfo): Rename the length field to
21393 nonzero_chars. Add a full_string_p field.
21394 (compare_nonzero_chars, zero_length_string_p): New functions.
21395 (get_addr_stridx): Add an offset_out parameter.
21396 Use compare_nonzero_chars.
21397 (get_stridx): Update accordingly. Use compare_nonzero_chars.
21398 (new_strinfo): Update after above changes to strinfo.
21399 (set_endptr_and_length): Set full_string_p.
21400 (get_string_length): Update after above changes to strinfo.
21401 (unshare_strinfo): Update call to new_strinfo.
21402 (maybe_invalidate): Likewise.
21403 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
21404 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
21405 as a uhwi instead of an shwi. Update after above changes to
21406 strinfo and new_strinfo.
21407 (zero_length_string): Assert that chainsi contains full strings.
21408 Use zero_length_string_p. Update call to new_strinfo.
21409 (adjust_related_strinfos): Update after above changes to strinfo.
21410 Copy full_string_p from origsi.
21411 (adjust_last_stmt): Use zero_length_string_p.
21412 (handle_builtin_strlen): Update after above changes to strinfo and
21413 new_strinfo. Install the lhs as the string length if the previous
21414 entry didn't describe a full string.
21415 (handle_builtin_strchr): Update after above changes to strinfo
21417 (handle_builtin_strcpy): Likewise.
21418 (handle_builtin_strcat): Likewise.
21419 (handle_builtin_malloc): Likewise.
21420 (handle_pointer_plus): Likewise.
21421 (handle_builtin_memcpy): Likewise. Track nonzero characters
21422 that aren't necessarily followed by a nul terminator.
21423 (handle_char_store): Likewise.
21425 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21427 PR tree-optimization/80769
21428 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
21429 for malloc and calloc. Document the new invariant that all related
21430 strinfos have delayed lengths or none do.
21431 (verify_related_strinfos): Move earlier in file.
21432 (set_endptr_and_length): New function, split out from...
21433 (get_string_length): ...here. Also set the lengths of related
21435 (zero_length_string): Assert that chainsi has known (rather than
21437 (adjust_related_strinfos): Likewise.
21439 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21441 PR tree-optimization/81136
21442 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
21443 assert that two references with the same misalignment have the same
21444 compile-time misalignment if those compile-time misalignments
21447 2017-07-01 Andi Kleen <ak@linux.intel.com>
21449 * print-tree.c (print_node): Print all attributes.
21451 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21453 * cfg.c (scale_bbs_frequencies): New function.
21454 * cfg.h (scale_bbs_frequencies): Declare it.
21455 * cfgloopanal.c (single_likely_exit): Cleanup.
21456 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
21458 (scale_loop_profile): Likewise.
21459 (loop_version): Likewise.
21460 (create_empty_loop_on_edge): Update.
21461 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
21462 scale_loop_frequencies, scale_loop_profile, loopify,
21463 loop_version): Update prototypes.
21464 * modulo-sched.c (sms_schedule): Update.
21465 * predict.c (unlikely_executed_edge_p): Also check probability.
21466 (probably_never_executed_edge_p): Fix typo.
21467 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21468 * tree-parloops.c (gen_parallel_loop): Update.
21469 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
21470 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21471 * tree-ssa-loop-split.c (split_loop): Update.
21472 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21473 * tree-vect-loop-manip.c (vect_do_peeling): Update.
21474 (vect_loop_versioning): Update.
21475 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21477 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21479 * trans-mem.c (split_bb_make_tm_edge): Update profile.
21481 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21483 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
21484 to keep profile consistent.
21486 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21488 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
21489 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
21490 * profile-count.h (max_safe_multiplier): Make unsigned.
21491 (profile_count::guessed_zero): New.
21493 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21495 * bb-reorder.c (fix_up_crossing_landing_pad,
21496 fix_crossing_conditional_branches): Use make_single_succ_edge
21497 to keep profile consistent.
21499 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21501 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
21504 2017-07-01 Jakub Jelinek <jakub@redhat.com>
21507 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
21508 the right scopes, make sure cond_jump isn't preserved between multiple
21509 iterations. Search for fallthru edge whenever there are 3+ edges and
21510 use find_fallthru_edge for it.
21512 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21514 Patch by Alexander Monakov <amonakov@ispras.ru>
21515 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
21516 probabilities consistently.
21518 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21520 * pa.c (pa_expand_compare_and_swap_loop): Update call of
21521 emit_cmp_and_jump_insns.
21523 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21526 * tree-inline.c (expand_call_inline): Combine profile statuses.
21528 2017-06-30 Andrew Pinski <apinski@cavium.com>
21530 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
21531 fold_stmt returned true.
21533 2017-06-30 Nathan Sidwell <nathan@acm.org>
21535 * ggc.h (empty_string): Delete.
21536 * cfgexpand.c (expand_asm_stmt): Use plain "".
21537 * optabs.c (expand_asm_memory_barrier): Likewise.
21538 * stringpool.c (empty_string): Delete.
21539 (digit_vector, digit_string): Delete.
21540 (ggc_alloc_string): Use plain "", don't optimize single digit
21541 strings. Use ggc_alloc_atomic.
21543 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
21545 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
21546 comparison set and one other set, use the cost of the non-comparison
21549 2017-06-30 Nathan Sidwell <nathan@acm.org>
21551 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
21554 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
21556 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
21557 loops. Remove now unneeded calls to gimple_switch_set_label() that
21558 just set removed labels to NULL_TREE.
21560 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
21562 * tree-ssanames.c (set_range_info_raw): Abstract from ...
21563 (set_range_info): ...here. Only call set_range_info_raw if domain
21565 (set_nonzero_bits): Call set_range_info_raw.
21566 * tree-ssanames.h (set_range_info_raw): New.
21568 2017-06-30 Jakub Jelinek <jakub@redhat.com>
21571 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
21572 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
21573 of nonimmediate_operand and <store_mask_constraint> instead of m
21574 for the input operand. For V8FI iterator, always split if input
21575 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
21576 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
21577 <store_mask_predicate> instead of register_operand and
21578 <store_mask_constraint> instead of v for the input operand. Make
21579 sure both operands aren't MEMs for if not <mask_applied>.
21581 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
21583 * lto-wrapper.c (copy_file) Close both file descriptors before
21586 2017-06-30 Martin Liska <mliska@suse.cz>
21589 * multiple_target.c (create_dispatcher_calls): Make ifunc
21590 also for function that don't have calls or are not referenced.
21592 2017-06-30 Richard Biener <rguenther@suse.de>
21594 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
21595 analyze the first scalar stmt. Move vector type computation
21596 for the BB case here from ...
21597 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
21598 live operation processing in the SLP case properly.
21600 2017-06-30 Richard Biener <rguenther@suse.de>
21602 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
21604 2017-06-30 Martin Liska <mliska@suse.cz>
21607 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
21608 before BUILT_IN_UNWIND_RESUME when ASAN is used.
21610 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
21612 * doc/invoke.texi (AArch64): Add missing options and remove redundant
21615 2017-06-30 Richard Biener <rguenther@suse.de>
21617 PR tree-optimization/81249
21618 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
21619 condition reduction result to original scalar type.
21621 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21623 * profile-count.h (enum profile_quality): Fix typos and whitespace
21626 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21628 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
21629 type for branch probabilities.
21631 2017-06-29 Julian Brown <julian@codesourcery.com>
21632 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21634 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
21635 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
21636 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
21637 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
21639 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21641 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
21642 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
21643 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
21644 CC usage from generic code to here.
21645 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
21646 CC usage into the target macros.
21648 2017-06-29 Maya Rashish <coypu@sdf.org>
21650 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
21653 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21655 * arm/arm-builtins.c: Include profile-count.h
21656 * except.c (sjlj_emit_function_enter): Use
21657 profile_probability::unlikely.
21659 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21661 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
21662 and tocrel_offset be pointer args rather than implicitly using
21664 (legitimate_constant_pool_address_p, rs6000_emit_move,
21665 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
21666 tocrel_offset and use in toc_relative_expr_p call.
21667 (print_operand, print_operand_address): Use static tocrel_base_oac
21668 and tocrel_offset_oac.
21669 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
21672 2017-06-29 Maya Rashish <coypu@sdf.org>
21674 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
21676 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
21678 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
21679 objects, take into account only the alignment of 'op0' and 'mode1' if
21682 2017-06-29 Steve Ellcey <sellcey@cavium.com>
21684 * ccmp.c (ccmp_tree_comparison_p): New function.
21685 (ccmp_candidate_p): Update to use above function.
21686 (get_compare_parts): New function.
21687 (expand_ccmp_next): Update to use new functions.
21688 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
21690 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
21691 take mode as argument.
21692 * ccmp.h (expand_ccmp_expr): Add mode as argument.
21693 * expr.c (expand_expr_real_1): Pass mode as argument.
21695 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
21697 * combine.c (combine_instructions): Print insns to dump_file, together
21700 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21702 * asan.c (asan_emit_stack_protection): Update.
21703 (create_cond_insert_point): Update.
21704 * auto-profile.c (afdo_propagate_circuit): Update.
21705 * basic-block.h (struct edge_def): Turn probability to
21706 profile_probability.
21707 (EDGE_FREQUENCY): Update.
21708 * bb-reorder.c (find_traces_1_round): Update.
21709 (better_edge_p): Update.
21710 (sanitize_hot_paths): Update.
21711 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
21712 (make_single_succ_edge): Update.
21713 (check_bb_profile): Update.
21714 (dump_edge_info): Update.
21715 (update_bb_profile_for_threading): Update.
21716 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
21717 probabilitycount to 0.
21718 * cfgbuild.c (compute_outgoing_frequencies): Update.
21719 * cfgcleanup.c (try_forward_edges): Update.
21720 (outgoing_edges_match): Update.
21721 (try_crossjump_to_edge): Update.
21722 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
21723 (expand_gimple_tailcall): Update.
21724 (construct_init_block): Use make_single_succ_edge.
21725 (construct_exit_block): Use make_single_succ_edge.
21726 * cfghooks.c (verify_flow_info): Update.
21727 (redirect_edge_succ_nodup): Update.
21728 (split_edge): Update.
21729 (account_profile_record): Update.
21730 * cfgloopanal.c (single_likely_exit): Update.
21731 * cfgloopmanip.c (scale_loop_profile): Update.
21732 (set_zero_probability): Remove.
21733 (duplicate_loop_to_header_edge): Update.
21734 * cfgloopmanip.h (loop_version): Update prototype.
21735 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
21736 (force_nonfallthru_and_redirect): Update.
21737 (update_br_prob_note): Update.
21738 (rtl_verify_edges): Update.
21739 (purge_dead_edges): Update.
21740 (rtl_lv_add_condition_to_bb): Update.
21741 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
21742 * cgraphunit.c (init_lowered_empty_function): Update.
21743 (cgraph_node::expand_thunk): Update.
21744 * cilk-common.c: Include profile-count.h
21745 * dojump.c (inv): Remove.
21746 (jumpifnot): Update.
21747 (jumpifnot_1): Update.
21748 (do_jump_1): Update.
21750 (do_jump_by_parts_greater_rtx): Update.
21751 (do_compare_rtx_and_jump): Update.
21752 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
21753 do_jump_1. do_compare_rtx_and_jump): Update prototype.
21754 * dwarf2cfi.c: Include profile-count.h
21755 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
21756 (sjlj_emit_dispatch_table): Likewise.
21757 * explow.c: Include profile-count.h
21758 * expmed.c (emit_store_flag_force): Update.
21759 (do_cmp_and_jump): Update.
21760 * expr.c (compare_by_pieces_d::generate): Update.
21761 (compare_by_pieces_d::finish_mode): Update.
21762 (emit_block_move_via_loop): Update.
21763 (store_expr_with_bounds): Update.
21764 (store_constructor): Update.
21765 (expand_expr_real_2): Update.
21766 (expand_expr_real_1): Update.
21767 * expr.h (try_casesi, try_tablejump): Update prototypes.
21768 * gimple-pretty-print.c (dump_probability): Update.
21769 (dump_profile): New.
21770 (dump_gimple_label): Update.
21771 (dump_gimple_bb_header): Update.
21772 * graph.c (draw_cfg_node_succ_edges): Update.
21773 * hsa-gen.c (convert_switch_statements): Update.
21774 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
21775 (find_if_case_1): Update.
21776 (find_if_case_2): Update.
21777 * internal-fn.c (expand_arith_overflow_result_store): Update.
21778 (expand_addsub_overflow): Update.
21779 (expand_neg_overflow): Update.
21780 (expand_mul_overflow): Update.
21781 (expand_vector_ubsan_overflow): Update.
21782 * ipa-cp.c (good_cloning_opportunity_p): Update.
21783 * ipa-split.c (split_function): Use make_single_succ_edge.
21784 * ipa-utils.c (ipa_merge_profiles): Update.
21785 * loop-doloop.c (add_test): Update.
21786 (doloop_modify): Update.
21787 * loop-unroll.c (compare_and_jump_seq): Update.
21788 (unroll_loop_runtime_iterations): Update.
21789 * lra-constraints.c (lra_inheritance): Update.
21790 * lto-streamer-in.c (input_cfg): Update.
21791 * lto-streamer-out.c (output_cfg): Update.
21792 * mcf.c (adjust_cfg_counts): Update.
21793 * modulo-sched.c (sms_schedule): Update.
21794 * omp-expand.c (expand_omp_for_init_counts): Update.
21795 (extract_omp_for_update_vars): Update.
21796 (expand_omp_ordered_sink): Update.
21797 (expand_omp_for_ordered_loops): Update.
21798 (expand_omp_for_generic): Update.
21799 (expand_omp_for_static_nochunk): Update.
21800 (expand_omp_for_static_chunk): Update.
21801 (expand_cilk_for): Update.
21802 (expand_omp_simd): Update.
21803 (expand_omp_taskloop_for_outer): Update.
21804 (expand_omp_taskloop_for_inner): Update.
21805 * omp-simd-clone.c (simd_clone_adjust): Update.
21806 * optabs.c (expand_doubleword_shift): Update.
21807 (expand_abs): Update.
21808 (emit_cmp_and_jump_insn_1): Update.
21809 (expand_compare_and_swap_loop): Update.
21810 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
21811 * predict.c (predictable_edge_p): Update.
21812 (edge_probability_reliable_p): Update.
21813 (set_even_probabilities): Update.
21814 (combine_predictions_for_insn): Update.
21815 (combine_predictions_for_bb): Update.
21816 (propagate_freq): Update.
21817 (estimate_bb_frequencies): Update.
21818 (force_edge_cold): Update.
21819 * profile-count.c (profile_count::dump): Add missing space into dump.
21820 (profile_count::debug): Add newline.
21821 (profile_count::differs_from_p): Explicitly convert to unsigned.
21822 (profile_count::stream_in): Update.
21823 (profile_probability::dump): New member function.
21824 (profile_probability::debug): New member function.
21825 (profile_probability::differs_from_p): New member function.
21826 (profile_probability::differs_lot_from_p): New member function.
21827 (profile_probability::stream_in): New member function.
21828 (profile_probability::stream_out): New member function.
21829 * profile-count.h (profile_count_quality): Rename to ...
21830 (profile_quality): ... this one.
21831 (profile_probability): New.
21832 (profile_count): Update.
21833 * profile.c (compute_branch_probabilities): Update.
21834 * recog.c (peep2_attempt): Update.
21835 * sched-ebb.c (schedule_ebbs): Update.
21836 * sched-rgn.c (find_single_block_region): Update.
21837 (compute_dom_prob_ps): Update.
21838 (schedule_region): Update.
21839 * sel-sched-ir.c (compute_succs_info): Update.
21840 * stmt.c (struct case_node): Update.
21841 (do_jump_if_equal): Update.
21842 (get_outgoing_edge_probs): Update.
21843 (conditional_probability): Update.
21844 (emit_case_dispatch_table): Update.
21845 (expand_case): Update.
21846 (expand_sjlj_dispatch_table): Update.
21847 (emit_case_nodes): Update.
21848 * targhooks.c: Update.
21849 * tracer.c (better_p): Update.
21850 (find_best_successor): Update.
21851 * trans-mem.c (expand_transaction): Update.
21852 * tree-call-cdce.c: Update.
21853 * tree-cfg.c (gimple_split_edge): Upate.
21854 (move_sese_region_to_fn): Upate.
21855 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
21856 * tree-eh.c (lower_resx): Upate.
21857 (cleanup_empty_eh_move_lp): Upate.
21858 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21859 * tree-inline.c (copy_edges_for_bb): Update.
21860 (copy_cfg_body): Update.
21861 * tree-parloops.c (gen_parallel_loop): Update.
21862 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
21863 (gimple_gen_time_profiler): Update.
21864 * tree-ssa-dce.c (remove_dead_stmt): Update.
21865 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21866 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
21867 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
21868 (unloop_loops): Update.
21869 (try_peel_loop): Update.
21870 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21871 * tree-ssa-loop-split.c (connect_loops): Update.
21872 (split_loop): Update.
21873 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21874 (hoist_guard): Update.
21875 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
21876 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
21877 (value_replacement): Update.
21878 * tree-ssa-reassoc.c (branch_fixup): Update.
21879 * tree-ssa-tail-merge.c (replace_block_by): Update.
21880 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
21881 (create_edge_and_update_destination_phis): Update.
21882 (compute_path_counts): Update.
21883 (recompute_probabilities): Update.
21884 (update_joiner_offpath_counts): Update.
21885 (freqs_to_counts_path): Update.
21886 (duplicate_thread_path): Update.
21887 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
21888 (struct switch_conv_info): Update.
21889 (gen_inbound_check): Update.
21890 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21891 (vect_do_peeling): Update.
21892 (vect_loop_versioning): Update.
21893 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21894 (optimize_mask_stores): Update.
21895 * ubsan.c (ubsan_expand_null_ifn): Update.
21896 * value-prof.c (gimple_divmod_fixed_value): Update.
21897 (gimple_divmod_fixed_value_transform): Update.
21898 (gimple_mod_pow2): Update.
21899 (gimple_mod_pow2_value_transform): Update.
21900 (gimple_mod_subtract): Update.
21901 (gimple_mod_subtract_transform): Update.
21902 (gimple_ic): Update.
21903 (gimple_stringop_fixed_value): Update.
21904 (gimple_stringops_transform): Update.
21905 * value-prof.h: Update.
21907 2017-06-29 Carl Love <cel@us.ibm.com>
21909 * config/rs6000/rs6000-c.c: Add support for built-in functions
21910 vector signed int vec_signed (vector float);
21911 vector signed long long vec_signed (vector double);
21912 vector signed int vec_signed2 (vector double, vector double);
21913 vector signed int vec_signede (vector double);
21914 vector signed int vec_signedo (vector double);
21915 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
21916 instruction generator.
21917 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
21918 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
21919 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
21921 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
21922 vunsignede_v2df): Add define_expands.
21923 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
21924 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
21925 VEC_UNSIGNEDO): Add definitions.
21926 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
21927 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
21928 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
21929 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
21930 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
21931 * config/rs6000/altivec.h (vec_signed, vec_signed2,
21932 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
21933 vec_unsignede, vec_unsignedo): Add builtin defines.
21934 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
21936 * doc/extend.texi: Update the built-in documentation file for the
21937 new built-in functions.
21939 2017-06-29 Richard Biener <rguenther@suse.de>
21941 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
21942 reduction chains to LOOP_VINFO_REDUCTIONS.
21943 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
21944 SLP reductions after processing reduction chains.
21946 2017-06-29 Nathan Sidwell <nathan@acm.org>
21948 * builtins.c (fold_builtin_FUNCTION): Use
21949 lang_hooks.decl_printable_name.
21951 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
21953 PR middle-end/81194
21954 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
21955 with only one label.
21956 * stmt.c (expand_case): Assert NCASES is greater than one.
21958 2017-06-29 Richard Biener <rguenther@suse.de>
21960 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
21962 (group_case_labels): Likewise.
21963 (find_taken_edge): Push sanity checking on val to workers...
21964 (find_taken_edge_cond_expr): ... here
21965 (find_taken_edge_switch_expr): ... and here, handle cases
21966 with just a default label.
21967 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
21968 (group_case_labels): Likewise.
21969 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
21970 group_case_labels does anything cleanup the CFG again.
21972 2017-06-29 Bin Cheng <bin.cheng@arm.com>
21974 PR tree-optimization/81196
21975 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
21976 exit condition comparing two IVs.
21978 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
21980 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
21981 profile to the dummy entry at the end of the list of architectures.
21982 * config/arm/arm-cpu-cdata.h: Regenerated.
21984 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21985 Michael Collison <michael.collison@arm.com>
21988 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
21990 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
21991 (*aarch64_reg_<mode>3_minus_mask): New pattern.
21992 (*aarch64_<optab>_reg_di3_mask2): New pattern.
21993 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
21994 of shift when the shift amount is masked with constant equal to
21995 the size of the mode.
21996 * config/aarch64/predicates.md (subreg_lowpart_operator): New
21999 2017-06-29 Martin Liska <mliska@suse.cz>
22001 * config/i386/i386.opt: Change range from [1,5] to [0,5].
22003 2017-06-29 Yury Gribov <tetra2005@gmail.com>
22006 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
22009 (edge_growth_cache_entry::edge_growth_cache_entry): New
22011 (reset_edge_growth_cache): Update to use constructor.
22013 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22015 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
22016 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22017 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22019 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
22021 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
22022 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
22024 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
22026 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
22027 (*-linux-uclibc*): Add t-uclibc tmake_file.
22028 * config/t-musl: New.
22029 * config/t-uclibc: New.
22031 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
22033 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
22035 (gen_comm_data): Emit architectural setting of arch_prof.
22036 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
22038 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
22039 (armv8-m.base, armv8-m.main): Likewise.
22040 * arm-protos.h (arm_build_target): Add profile field.
22041 (arch_option): Likewise.
22042 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
22044 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
22045 arm_active_target.profile.
22047 2017-06-28 Richard Biener <rguenther@suse.de>
22049 PR middle-end/81227
22050 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
22051 TYPE_OVERFLOW_WRAPS.
22052 * match.pd (negate_expr_p): Likewise.
22053 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
22054 fold_build2, not fold_binary.
22056 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
22058 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
22059 Convert memory address to Pmode.
22060 (aarch64_print_operand): Assert MEM operands are always Pmode.
22062 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
22065 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
22066 Remove redundant if.
22067 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
22068 * config/arm/aarch-common-protos.h
22069 (aarch_forward_to_shift_is_not_shifted_re): Remove.
22070 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
22072 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
22075 * multiple_target.c (create_dispatcher_calls): Set the default
22076 clone to be static, not public.
22078 2017-06-28 Richard Biener <rguenther@suse.de>
22080 * tree-vect-loop.c (vectorizable_reduction): Move special
22081 cond reduction IV var creation ...
22082 (vect_create_epilog_for_reduction): ... here. Remove induction_index
22083 parameter. Use STMT_VINFO_VECTYPE.
22084 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
22087 2017-06-28 Martin Liska <mliska@suse.cz>
22090 * ipa-visibility.c (non_local_p): Handle visibility.
22092 2017-06-28 Martin Liska <mliska@suse.cz>
22095 * common.opt: Add IntegerRange to various options.
22096 * opt-functions.awk (integer_range_info): New function.
22097 * optc-gen.awk: Add integer_range_info to cl_options struct.
22098 * opts-common.c (decode_cmdline_option): Handle
22099 CL_ERR_INT_RANGE_ARG.
22100 (cmdline_handle_error): Likewise.
22101 * opts.c (print_filtered_help): Show valid interval in
22102 when --help is provided.
22103 * opts.h (struct cl_option): Add range_min and range_max fields.
22104 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
22106 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
22108 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
22109 (x * C EQ/NE y * C): New transformation.
22111 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
22113 * genmultilib (combination_space): Accept '+' in option names.
22115 2017-06-28 Martin Liska <mliska@suse.cz>
22118 * asan.c (instrument_derefs): Bail out inner references
22119 that are hard register variables.
22121 2017-06-28 Jakub Jelinek <jakub@redhat.com>
22124 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
22125 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
22127 2017-06-28 Richard Biener <rguenther@suse.de>
22129 * tree-vectorizer.h (vect_get_vec_defs): Remove.
22130 (vect_get_slp_defs): Adjust.
22131 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
22133 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
22135 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
22136 get_initial_defs_for_reduction instead of vect_get_vec_defs.
22137 (vectorizable_reduction): Adjust.
22138 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
22140 (vect_get_slp_defs): Likewise.
22141 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
22142 (vectorizable_bswap): Adjust.
22143 (vectorizable_call): Likewise.
22144 (vectorizable_conversion): Likewise.
22145 (vectorizable_assignment): Likewise.
22146 (vectorizable_shift): Likewise.
22147 (vectorizable_operation): Likewise.
22148 (vectorizable_store): Likewise.
22149 (vectorizable_condition): Likewise.
22150 (vectorizable_comparison): Likewise.
22152 2017-06-28 Michael Collison <michael.collison@arm.com>
22155 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
22157 (arm_gen_movmemqi): Removed unused variable 'i'.
22158 Convert 'for' loop into 'while' loop.
22159 (arm_expand_prologue): Remove last unnecessary set of insn.
22160 (thumb_pop): Remove unused variable 'pushed_words'.
22161 (thumb_exit): Remove last unnecessary set of regs_to_pop.
22163 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22165 * config/s390/predicates.md: Use s390_rel_address_ok_p.
22166 * config/s390/s390-protos.h: Add prototype of
22167 s390_rel_address_ok_p.
22168 * config/s390/s390.c (s390_got_symbol): New function.
22169 (s390_rel_address_ok_p): New function.
22170 (legitimize_pic_address): Use s390_rel_address_ok_p.
22171 (s390_load_got): Use s390_got_symbol.
22172 (s390_option_override): Issue error if
22173 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
22174 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
22176 * config/s390/s390.opt: New option mpic-data-is-text-relative.
22178 2017-06-27 Andrew Pinski <apinski@cavium.com>
22180 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
22181 (X * copysign (1.0, X)): New pattern.
22182 (X * copysign (1.0, -X)): New pattern.
22183 (copysign (-1.0, CST)): New pattern.
22185 2017-06-27 Joseph Myers <joseph@codesourcery.com>
22187 * genmultilib (combination_space): Remove variable.
22188 Validate reuse rules against regular expression for any sequence
22189 of multilib options in any order.
22191 2017-06-27 Michael Collison <michael.collison@arm.com>
22193 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
22194 call aarch64_split_simd_combine.
22195 * (aarch64_combine_internal<mode>): Delete pattern.
22196 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
22197 Allow register and subreg operands.
22199 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22201 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
22202 specific need, just fallback on defaults.
22203 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
22205 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22206 Olivier Hainque <hainque@adacore.com>
22208 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
22210 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
22211 targets. Pick a default if no particular attempt applied.
22212 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
22215 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22217 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
22218 (x86_64-wrs-vxworks7): Likewise.
22220 2017-06-27 Marek Polacek <polacek@redhat.com>
22223 * ubsan.c (instrument_null): Check get_base_address's result for null.
22225 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22227 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
22229 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22231 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
22232 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
22233 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
22234 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
22235 New function types.
22236 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
22237 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
22238 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
22239 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
22240 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
22241 BUILT_IN_FEUPDATEENV): New builtins.
22242 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
22243 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
22244 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
22245 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
22247 (builtin_structptr_types): Adjust size.
22248 * tree.c (builtin_structptr_types): Add four entries.
22250 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22251 Olivier Hainque <hainque@adacore.com>
22253 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
22254 (TLS_SYM): New local macro, forcing reference to __tls__ on
22255 link command lines for VxWorks 7 RTPs, triggering initialization
22257 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
22258 OS features TLS support, true for RTPs on VxWorks 7.
22259 * config/vxworks.c (vxworks_override_options): Setup emutls
22262 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22264 * predict.c (test_prediction_value_range): Use -1U instead of -1
22265 to avoid narrowing conversion warning.
22266 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
22267 to avoid narrowing conversion warning.
22268 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
22270 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
22272 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22274 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
22275 64bit configurations.
22276 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
22277 (SIZE_TYPE): Likewise.
22278 * config/vxworks.c (vxworks_emutls_var_fields): Use
22279 long_unsigned_type_node instead of unsigned_type_node as the offset
22280 field type, which is "pointer" mode in emutls.c.
22282 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22285 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
22287 PR middle-end/81207
22288 * gimple-fold.c (replace_call_with_call_and_fold): Handle
22289 gimple_vuse copying separately from gimple_vdef copying.
22291 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22293 * value-prof.c (free_hist): Remove call to memset and the enclosing if
22296 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22297 Olivier Hainque <hainque@adacore.com>
22299 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
22300 for all vxworks7 targets.
22301 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
22302 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
22303 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
22304 variations for VX6/VX7 and 32/64bits later on in ...
22305 (VXWORKS_LIB_SPEC): Leverage new macros.
22306 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
22307 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
22309 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22311 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
22312 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
22314 2017-06-26 Carl Love <cel@us.ibm.com>
22316 * config/rs6000/rs6000-c.c: Add support for built-in functions
22317 vector bool char vec_reve (vector bool char);
22318 vector signed char vec_reve (vector signed char);
22319 vector unsigned char vec_reve (vector unsigned char);
22320 vector bool int vec_reve (vector bool int);
22321 vector signed int vec_reve (vector signed int);
22322 vector unsigned int vec_reve (vector unsigned int);
22323 vector bool long long vec_reve (vector bool long long);
22324 vector signed long long vec_reve (vector signed long long);
22325 vector unsigned long long vec_reve (vector unsigned long long);
22326 vector bool short vec_reve (vector bool short);
22327 vector signed short vec_reve (vector signed short);
22328 vector double vec_reve (vector double);
22329 vector float vec_reve (vector float);
22330 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
22331 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
22332 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
22333 (altivec_vreve): New pattern.
22334 * config/rs6000/altivec.h (vec_reve): New define.
22335 * doc/extend.texi (vec_rev): Update the built-in documentation file
22336 for the new built-in functions.
22338 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22340 PR tree-optimization/71815
22341 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
22343 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
22345 (slsr_process_phi): Likewise.
22346 (replace_uncond_cands_and_profitable_phis): Don't replace a
22347 multiply candidate with a stride of 1 (copy or cast).
22348 (phi_incr_cost): Call uses_consumed_by_stmt rather than
22350 (lowest_cost_path): Likewise.
22351 (total_savings): Likewise.
22353 2017-06-26 Richard Biener <rguenther@suse.de>
22356 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
22357 Use def_builtin_pure for all gather builtins.
22359 2017-06-26 Richard Biener <rguenther@suse.de>
22361 PR tree-optimization/81203
22362 * tree-tailcall.c (find_tail_calls): Do not move stmts into
22363 non-dominating BBs.
22365 2017-06-26 Marek Polacek <polacek@redhat.com>
22368 * doc/invoke.texi: Document -Wmultistatement-macros.
22370 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
22372 * doc/sourcebuild.texi (ARM-specific attributes): Document new
22373 arm_neon_ok_no_float_abi effective target.
22375 2017-06-26 Richard Biener <rguenther@suse.de>
22377 PR tree-optimization/80928
22378 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
22379 (copy_bbs): Set BB_DUPLICATED flag early.
22380 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
22382 (execute_on_shrinking_pred): Likewise.
22383 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
22384 BB_DUPLICATED blocks.
22385 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
22386 iterate over all PHIs considering removal of *gsi.
22388 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
22390 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
22393 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22395 * config/rs6000/rs6000-string.c: (expand_block_clear,
22396 do_load_for_compare, select_block_compare_mode,
22397 compute_current_alignment, expand_block_compare,
22398 expand_strncmp_align_check, expand_strn_compare,
22399 expand_block_move, rs6000_output_load_multiple)
22400 Move functions related to string/block move/compare
22401 to a separate file.
22402 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
22403 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
22404 for this function which is now used in two files.
22405 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
22406 * config.gcc: Add rs6000-string.o to extra_objs for
22407 targets powerpc*-*-* and rs6000*-*-*.
22409 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
22412 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
22413 32-bit, since indexed is not valid for DImode.
22414 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
22415 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
22416 (define_peephole2 for Altivec d-form load): Add 32-bit support.
22417 (define_peephole2 for Altivec d-form store): Likewise.
22420 * multiple_target.c (create_dispatcher_calls): Only create the
22421 dispatcher call if the function is the default clone of a
22422 versioned function.
22424 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
22426 PR middle-end/80902
22427 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
22428 a call, force the call to not be a tail call.
22430 2017-06-23 Jeff Law <law@redhat.com>
22432 * doc/contrib.texi: Add entry for Steven Pemberton's work on
22435 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
22437 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
22438 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
22439 handling for early expansion of vector shifts (sl,sr,sra,rl).
22440 (builtin_function_type): Add vector shift right instructions
22441 to the unsigned argument list.
22443 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
22445 rtl-optimizatoin/79286
22446 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
22447 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
22448 trap. PIC register plus a const unspec without offset can never trap.
22450 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
22452 * tree.h (builtin_structptr_type): New type.
22453 (builtin_structptr_types): Declare new array.
22454 * tree.c (builtin_structptr_types): New array.
22455 (free_lang_data, build_common_tree_nodes): Use it.
22457 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
22460 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
22463 2017-06-22 Matt Turner <mattst88@gmail.com>
22465 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
22466 Lake models to skylake case. Assume skylake for unknown
22467 models with clflushopt.
22469 2017-06-22 Jeff Law <law@redhat.com>
22471 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
22472 frame sizes that do not satisfy aarch64_uimm12_shift.
22474 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
22476 * profile-count.h (apply_probability,
22477 apply_scale, probability_in): Fix checks for zero.
22479 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22481 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
22482 * doc/cppdiropts.texi (-I @var{dir}): Document it.
22484 2016-06-22 Richard Biener <rguenther@suse.de>
22486 * tree-vect-loop.c (vect_model_reduction_cost): Handle
22487 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
22488 REDUC_MAX_EXPR support.
22489 (vectorizable_reduction): Likewise.
22490 (vect_create_epilog_for_reduction): Likewise.
22492 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
22494 * match.pd (A / (1 << B) -> A >> B): New.
22495 * generic-match-head.c: Include optabs-tree.h.
22496 * gimple-match-head.c: Likewise.
22497 * optabs-tree.h (target_supports_op_p): New.
22498 * optabs-tree.c (target_supports_op_p): New.
22500 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22502 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
22503 $gcc_cv_ld --help output.
22504 (gcc_cv_ld_demangle): Likewise.
22505 (gcc_cv_ld_eh_frame_hdr): Likewise.
22506 (gcc_cv_ld_pie): Likewise.
22507 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
22508 (gcc_cv_ld_buildid): Likewise.
22509 (gcc_cv_ld_sysroot): Likewise.
22510 (ld_bndplt_support): Likewise.
22511 (ld_pushpopstate_support): Likewise.
22512 * configure: Regenerate.
22513 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
22515 2017-06-21 Jakub Jelinek <jakub@redhat.com>
22518 * config/i386/sse.md (round<mode>2): Renumber match_dup and
22519 operands indexes to avoid gap between operands and match_dups.
22521 2017-06-21 Andrew Pinski <apinski@cavium.com>
22523 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
22524 Increment Arith_shift and Arith_shift_reg by 1.
22525 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
22527 * config/aarch64/aarch64.c (thunderx_tunings): Enable
22528 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
22529 (aarch64_strip_extend): Add new argument and test for it.
22530 (aarch64_cheap_mult_shift_p): New function.
22531 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
22532 add a cost if it is true.
22533 Update calls to aarch64_strip_extend.
22534 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
22536 2017-06-21 Andrew Pinski <apinski@cavium.com>
22538 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
22540 (thunderxt88): Likewise.
22541 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
22542 (thunderx_prefetch_tune): New variable.
22543 (thunderx2t99_prefetch_tune): Update for the correct values.
22544 (thunderxt88_tunings): New variable.
22545 (thunderx_tunings): Use thunderx_prefetch_tune instead of
22546 generic_prefetch_tune.
22547 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
22549 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22551 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
22552 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
22553 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
22554 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
22555 (aarch64_atomic_cas<mode>, GPI): Likewise.
22557 2017-06-21 Martin Liska <mliska@suse.cz>
22559 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
22560 statements on cold and hot labels.
22561 * predict.c (tree_estimate_probability_bb): Remove the
22562 prediction from this place.
22564 2017-06-21 Martin Liska <mliska@suse.cz>
22566 PR tree-optimization/79489
22567 * gimplify.c (maybe_add_early_return_predict_stmt): New
22569 (gimplify_return_expr): Call the function.
22570 * predict.c (tree_estimate_probability_bb): Remove handling
22572 * predict.def: Update comment about early return predictor.
22573 * gimple-predict.h (is_gimple_predict): New function.
22574 * predict.def: Change default value of early return to 66.
22575 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
22577 * passes.def: Put pass_strip_predict_hints to the beginning of
22580 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
22582 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
22583 FUNCTION_DECL declarations.
22584 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
22586 (dwaf2out_decl): Likewise.
22587 * godump.c (go_early_global_decl): Skip call to the real debug hook
22588 for FUNCTION_DECL declarations.
22589 * passes.c (rest_of_decl_compilation): Skip call to the
22590 early_global_decl debug hook for FUNCTION_DECL declarations, unless
22591 -fdump-go-spec is passed.
22593 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22595 * config/i386/i386.c (struct builtin_isa): New field pure_p.
22596 Reorder for compactness.
22597 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
22598 (def_builtin_pure, def_builtin_pure2): New functions.
22599 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
22601 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22603 * match.pd (nop_convert): New predicate.
22604 ((A +- CST1) +- CST2): Allow some NOP conversions.
22606 2017-06-21 Jakub Jelinek <jakub@redhat.com>
22609 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
22610 with ctors/dtors if GOVD_SHARED is set.
22612 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22614 * config/aarch64/aarch64.md (movti_aarch64):
22615 Emit mov rather than orr.
22616 (movtf_aarch64): Likewise.
22617 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
22618 Emit mov rather than orr.
22620 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22622 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
22623 Swap alternatives, make integer dup more expensive.
22625 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22627 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
22628 Return true for non-tls symbols.
22630 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
22632 * config/aarch64/aarch64-cores.def (cortex-a55): New.
22633 (cortex-a75): Likewise.
22634 (cortex-a75.cortex-a55): Likewise.
22635 * config/aarch64/aarch64-tune.md: Regenerate.
22636 * doc/invoke.texi (-mtune): Document new values for -mtune.
22638 2017-06-21 Tom de Vries <tom@codesourcery.com>
22640 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
22641 stack_size feature.
22642 (Effective-Target Keywords, Other attributes): Suggest using
22643 dg-add-options stack_size feature to get stack limit in stack_size
22644 effective target documentation.
22646 2017-06-21 Julian Brown <julian@codesourcery.com>
22647 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22649 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
22650 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
22651 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
22653 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
22654 attribute type list for neon_multiply.
22655 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
22656 attribute type list for neon_multiply.
22657 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
22658 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
22659 attribute type list for neon_multiply.
22660 * config/arm/types.md (crypto_pmull): Add.
22661 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
22662 attribute type list.
22664 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
22666 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
22669 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22671 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
22672 to make sure not to dereference a NULL cost_classes_ptr pointer.
22674 2017-06-20 Carl Love <cel@us.ibm.com>
22676 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22677 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22678 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22679 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22680 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22681 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22682 VMULOSW): New enum "unspec" values.
22683 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22684 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22685 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22686 altivec_vmulosw): New patterns.
22687 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22688 VMULOSW): Add definitions.
22690 2017-06-20 Julia Koval <julia.koval@intel.com>
22692 * config/i386/i386.c: Fix rounding expand for new pattern.
22693 * config/i386/subst.md: Fix pattern (parallel -> unspec).
22695 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22697 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
22698 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
22700 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22702 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
22705 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22707 * config/aarch64/aarch64-cores.def: Rearrange to sort by
22708 architecture, then by implementer ID.
22709 * config/aarch64/aarch64-tune.md: Regenerate.
22711 2017-06-20 Richard Biener <rguenther@suse.de>
22713 PR middle-end/81097
22714 * fold-const.c (split_tree): Fold to type before negating.
22716 2017-06-20 David Malcolm <dmalcolm@redhat.com>
22718 * diagnostic-show-locus.c
22719 (selftest::test_fixit_deletion_affecting_newline): New function.
22720 (selftest::diagnostic_show_locus_c_tests): Call it.
22722 2017-06-20 Andreas Schwab <schwab@suse.de>
22725 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
22728 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22730 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
22731 __ARM_FEATURE_COPROC according to support.
22733 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22735 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22736 Rewritten to avoid overflow for > 32-bit pointers.
22739 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
22740 by removing enum keyword.
22741 (ubsan_type_descriptor): Likewise. Formatting fix.
22744 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
22745 splitter): Require TARGET_SSE2 in the condition.
22747 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
22750 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
22751 for doing vector set of SFmode on ISA 3.0.
22752 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
22753 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
22755 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
22756 SFmode value into a V4SF variable that was extracted from another
22757 V4SF variable without converting the element to double precision
22758 and back to single precision vector format.
22759 (vsx_insert_extract_v4sf_p9_2): Likewise.
22761 2017-06-19 Jakub Jelinek <jakub@redhat.com>
22763 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
22764 in UWHI to avoid undefined overflow.
22767 * ubsan.h (enum ubsan_encode_value_phase): New.
22768 (ubsan_encode_value): Change second argument to
22769 enum ubsan_encode_value_phase with default value of
22770 UBSAN_ENCODE_VALUE_GENERIC.
22771 * ubsan.c (ubsan_encode_value): Change second argument to
22772 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
22773 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
22774 create_tmp_var_raw instead of create_tmp_var and use a
22776 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
22777 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
22778 ubsan_encode_value callers.
22781 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
22782 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
22783 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
22785 2017-06-19 Richard Biener <rguenther@suse.de>
22787 PR middle-end/81118
22788 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
22789 estimates if we changed anything.
22791 2017-06-19 Richard Biener <rguenther@suse.de>
22793 PR tree-optimization/80887
22794 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
22795 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
22796 simplified lookups, then reset mprts_hook.
22797 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
22799 (try_to_simplify): Likewise.
22801 2017-06-19 Martin Liska <mliska@suse.cz>
22804 * gimplify.c (gimplify_switch_expr):
22805 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
22807 2017-06-19 Martin Liska <mliska@suse.cz>
22809 * doc/install.texi: Document that PGO runs in 4 stages.
22811 2017-06-19 Martin Liska <mliska@suse.cz>
22814 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
22815 to dispatcher function name.
22816 * multiple_target.c (replace_function_decl): New function.
22817 (create_dispatcher_calls): Redirect both edges and references.
22819 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
22821 * profile-count.c (profile_count::dump): Dump quality.
22822 (profile_count::differs_from_p): Update for unsigned val.
22823 * profile-count.h (profile_count_quality): New enum.
22824 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
22826 2017-06-19 Richard Biener <rguenther@suse.de>
22828 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
22829 struct function as arg.
22830 (estimate_numbers_of_iterations): Export overload with loop arg.
22831 (free_numbers_of_iterations_estimates_loop): Use an overload of
22832 free_numbers_of_iterations_estimates instead.
22833 * tree-cfg.c (remove_bb): Adjust.
22834 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
22835 * tree-parloops.c (gen_parallel_loop): Likewise.
22836 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
22838 (tree_unroll_loops_completely): Likewise.
22839 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
22840 Use an overload instead and export.
22841 (estimated_loop_iterations): Adjust.
22842 (max_loop_iterations): Likewise.
22843 (likely_max_loop_iterations): Likewise.
22844 (estimate_numbers_of_iterations): Take struct function as arg
22846 (loop_exits_before_overflow): Adjust.
22847 (free_numbers_of_iterations_estimates_loop): Use an overload.
22848 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
22849 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
22851 2017-06-19 Richard Biener <rguenther@suse.de>
22854 * ipa-prop.c (find_constructor_constant_at_offset): Handle
22855 RANGE_EXPR conservatively.
22857 2017-06-16 Carl Love <cel@us.ibm.com>
22859 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22860 definitions for vec_float, vec_float2, vec_floato,
22861 vec_floate built-ins.
22862 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
22863 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
22865 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
22866 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
22867 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
22868 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
22869 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
22870 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
22871 vec_floato): Add builtin defines.
22872 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
22873 Update the built-in documentation file for the new built-in
22876 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22878 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
22879 (mthumb): Mark as the negative of -marm.
22881 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22883 * doc/invoke.texi (ARM Options, -mcpu): Document supported
22885 (ARM Options, -mtune): Document that this accepts the same
22886 extension options as -mcpu.
22887 (ARM Options, -mfpu): Document addition of -mfpu=auto.
22889 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22891 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
22892 permitted extensions.
22894 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22896 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
22897 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
22898 (armv8-m.main): Add option +nodsp.
22899 * config/arm/arm-cpu-cdata.h: Regenerated.
22901 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22903 * config/arm/t-fuchsia: New file.
22904 * config.gcc (arm*-*-fuchsia*): Use it.
22906 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22908 * config/arm/t-symbian: Rewrite for new option infrastructure.
22910 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22912 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
22913 (MULTILIB_REQUIRED): Likewise.
22915 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22917 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
22918 (MULTILIB_RESUE): Likewise.
22919 (MULTILIB_MATCHES): Likewise.
22920 (MULTLIB_REQUIRED): Likewise.
22922 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22924 * config/arm/t-rtems: Rewrite for new option framework.
22926 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22928 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
22929 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
22930 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
22931 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
22932 * config/arm/t-multilib: ... here.
22933 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
22934 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
22935 armv7-a and armv8*-a when A-profile libraries have not been built.
22936 * config/arm/t-rmprofile: Rewrite.
22938 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22940 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
22941 with a backslash. Remove the backslash after substituting unescaped
22943 * doc/fragments.texi (MULTILIB_REUSE): Document it.
22945 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22947 * config.gcc: (arm*-*-*): When building a-profile libraries, force
22948 the driver to pass through the default setting of -mfloat-abi.
22949 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
22951 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
22952 (all_feat_combs): New rule.
22953 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
22955 * config/arm/t-aprofile: Rewrite.
22957 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22959 * config/arm/arm.h (FPUTYPE_AUTO): Define.
22960 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
22961 fpu is not specified by the user/command-line.
22962 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
22963 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
22964 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
22965 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
22966 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
22967 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
22969 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22971 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
22972 * config/arm/t-arm-elf: Rewritten.
22974 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22976 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
22977 have some floating-point instructions.
22978 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
22979 (TARGET_MAYBE_HARD_FLOAT): New macro.
22980 * config/arm/arm-builtins.c (arm_init_builtins): Use
22981 TARGET_MAYBE_HARD_FLOAT.
22982 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
22984 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22986 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
22987 (configargs.h): Include it.
22988 (arm_print_hint_for_fpu_option): New function.
22989 (arm_parse_fpu_option): New function.
22990 (candidate_extension): New class.
22991 (arm_canon_for_multilib): New function.
22992 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
22993 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
22994 (ARCH_CANONICAL_SPECS): New macro.
22995 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
22997 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22999 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
23000 are set after handling multilib fragments. Set target_cpu_default2
23003 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23005 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
23007 (arm*-*-*): Set target_cpu_default2 to a quoted string.
23008 * config/arm/parsecpu.awk (check_cpu): Validate any extension
23010 (check_arch): Likewise.
23011 * config/arm/arm.c (arm_configure_build_target): Handle
23012 TARGET_CPU_DEFAULT being a string constant. Scan any feature
23013 options in the default.
23015 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23017 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
23018 when an option is an alias of another.
23019 * config/arm/parsecpu.awk (optalias): New parser token.
23020 (gen_comm_data): Mark non-alias options as such. Emit entries
23021 for extension aliases.
23022 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
23023 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
23024 (armv6kz, armv6zk, armv6t2): Likewise.
23025 (armv7): Make vfpv3-d16 an alias.
23026 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
23028 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
23029 Sort in canonical order.
23030 (armv8-a): Sort in canonical order.
23031 (armv8.1-a, armv8.2-a): Likewise.
23032 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
23034 (cortex-a9): Sort in canonical order.
23035 * config/arm/arm.c (selftests.h): Include it.
23036 (arm_test_cpu_arch_data): New function.
23037 (arm_run_self_tests): New function.
23038 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
23039 (targetm): Move declaration to the end of the file.
23040 * arm-cpu-cdata.h: Regenerated.
23042 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23044 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
23045 call to target_mode_check describing the type of option passed.
23046 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
23047 (arm_target_thumb_only): Use arm_parse_arch_option_name or
23048 arm_parse_cpu_option_name to match parameters against list of
23050 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
23051 arm_arch_core_flags data structure.
23052 * config/arm/arm-cpu_cdata.h: Regenerated.
23054 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23056 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
23058 (arm_print_hint_for_cpu_option): Likewise.
23059 (arm_print_hint_for_arch_option): Likewise.
23060 (arm_parse_cpu_option_name): Likewise.
23061 (arm_parse_arch_option_name): Likewise.
23062 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
23063 of entries in the all_fpus list.
23064 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
23065 (arm_parse_cpu_option_name): Declare.
23066 (arm_parse_arch_option_name): Declare.
23067 (arm_parse_option_features): Declare.
23068 (arm_intialize_isa): Declare.
23069 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
23071 (gen_comm_data): ... here. Make definitions non-static.
23072 * config/arm/arm-cpu-data.h: Regenerated.
23073 * config/arm/arm-cpu-cdata.h: Regenerated.
23075 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23077 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
23078 (cpu_arch_extension): New structure.
23079 (cpu_arch_option, arch_option, cpu_option): New structures.
23080 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
23081 architecture types.
23082 (gen_data): Generate new format data tables.
23083 * config/arm/arm.c (cpu_tune): New structure.
23084 (cpu_option, processors): Delete.
23085 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
23086 (arm_print_hint_for_cpu_option): ... this and ...
23087 (arm_print_hint_for_arch_option): ... this.
23088 (arm_parse_arch_cpu_name): Delete. Replace with ...
23089 (arm_parse_cpu_option_name): ... this and ...
23090 (arm_parse_arch_option_name): ... this.
23091 (arm_unrecognized_feature): Change type of target parameter to
23093 (arm_parse_arch_cpu_features): Delete. Replace with ...
23094 (arm_parse_option_features): ... this.
23095 (arm_configure_build_target): Rework to use new configuration data
23097 (arm_print_tune_info): Rework for new configuration data tables.
23098 * config/arm/arm-cpu-data.h: Regenerated.
23099 * config/arm/arm-cpu.h: Regenerated.
23101 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23103 * Makefile.in (OBJS): Move sbitmap.o from here ...
23104 (OBJS-libcommon): ... to here.
23106 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23108 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
23109 (ISA_ALL_CRYPTO): New macro.
23110 (ISA_ALL_SIMD): New macro
23111 (ISA_ALL_FP): New macro.
23112 * config/arm/arm.c (fpu_bitlist): Update initializer.
23113 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
23115 (arm9e): Add fpu. Add option for nofp
23116 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
23117 (arm926ej-s, arm1026ej-s): Likewise.
23118 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
23119 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
23120 neon-fp16, neon-vfpv4, nofp and nosimd.
23121 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
23122 (cortex-a8): Add fpu. Add option for nofp.
23123 (cortex-a9): Add fpu. Add options for nosimd and nofp.
23124 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
23125 (cortex-r4f): Add fpu.
23126 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
23127 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
23129 (cortex-r8): Likewise.
23130 (cortex-m4): Add fpu. Add option for nofp.
23131 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
23132 (cortex-a17.cortex-a7): Likewise.
23133 (cortex-a32): Add fpu. Add options for crypto and nofp.
23134 (cortex-a35, cortex-a53): Likewise.
23135 (cortex-a57): Add fpu. Add option for crypto.
23136 (cortex-a72, cortex-a73): Likewise.
23137 (exynos-m1): Likewise.
23138 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
23139 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
23140 (cortex-m33): Add fpu. Add option for nofp.
23141 * config/arm/arm-cpu-cdata.h: Regenerated
23142 * config/arm/arm-cpu-data.h: Regenerated.
23144 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23146 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
23147 (armv5te, armv5tej): Likewise.
23148 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
23149 (armv7): Add options fp and vfpv3-d16.
23150 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
23151 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
23153 (armv7ve): Likewise.
23154 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
23155 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
23156 (armv8-a): Add nocrypto option.
23157 (armv8.1-a, armv8.2-a): Likewise.
23158 (armv8-m.main): add options fp, fp.dp and nofp.
23160 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23162 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
23164 (armv8-a+crc): Delete.
23165 (armv8.1-a): Add options simd, crypto and nofp.
23166 (armv8.2-a): Add options fp16, simd, crypto and nofp.
23167 (armv8.2-a+fp16): Delete.
23168 (armv8-m.main): Add option dsp.
23169 (armv8-m.main+dsp): Delete.
23170 (cortex-a8): Add fpu. Add nofp option.
23171 (cortex-a9): Add fpu. Add nofp and nosimd options.
23172 * config/arm/parsecpu.awk (gen_data): Generate option tables and
23173 link to main cpu and architecture data structures.
23174 (gen_comm_data): Only put isa attributes from the main architecture
23176 (option): New statement for architecture and CPU entries.
23177 * arm.c (struct cpu_option): New structure.
23178 (struct processors): Add entry for options.
23179 (arm_unrecognized_feature): New function.
23180 (arm_parse_arch_cpu_name): Ignore any characters after the first
23182 (arm_parse_arch_cpu_feature): New function.
23183 (arm_configure_build_target): Separate out any CPU and architecture
23184 features and parse separately. Don't error out if -mfpu=auto is
23185 used with only an architecture string.
23186 (arm_print_asm_arch_directives): New function.
23187 (arm_file_start): Call it.
23188 * config/arm/arm-cpu-cdata.h: Regenerated.
23189 * config/arm/arm-cpu-data.h: Likewise.
23190 * config/arm/arm-tables.opt: Likewise.
23192 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23194 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
23195 assembler when it is not -mfpu=auto.
23197 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23199 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
23200 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
23201 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
23202 (ASM_CPU_SPEC): Rewrite.
23203 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
23204 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
23205 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
23206 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
23207 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
23208 copied string is NUL-terminated. Also strip any characters prefixed
23210 (arm_rewrite_selected_arch): New function.
23211 (arm_rewrite_march): New function.
23213 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23215 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
23216 (x_arm_cpu_string, x_arm_tune_string): Likewise.
23217 (march, mcpu, mtune): Convert to string-based options.
23218 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
23219 (arm_parse_arch_cpu_name): New function.
23220 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
23221 identify selected architecture or CPU.
23222 (arm_option_save): New function.
23223 (TARGET_OPTION_SAVE): Redefine.
23224 (arm_option_restore): Restore string options.
23225 (arm_option_print): Print string options.
23227 2017-06-16 Martin Sebor <msebor@redhat.com>
23229 PR tree-optimization/80933
23230 PR tree-optimization/80934
23231 * builtins.c (fold_builtin_3): Do not handle bcmp here.
23232 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
23233 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
23234 (gimple_fold_builtin): Call them.
23236 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23238 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
23239 as unlikely; update profile.
23241 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23243 * predict.c (force_edge_cold): Handle declaring edges impossible
23246 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23248 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
23250 (try_unroll_loop_completely): Fix reporting.
23252 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23254 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
23256 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
23259 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
23260 if given a non-constant argument for an intrinsic which requires a
23263 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23265 * profile.c (compare_freqs): New function.
23266 (branch_prob): Sort edge list.
23267 (find_spanning_tree): Assume that the list is priority sorted.
23269 2017-06-16 Richard Biener <rguenther@suse.de>
23271 PR tree-optimization/81090
23272 * passes.def (pass_record_bounds): Remove.
23273 * tree-pass.h (make_pass_record_bounds): Likewise.
23274 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
23275 make_pass_record_bounds): Likewise.
23276 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
23277 not free niter estimates at the beginning but at the end.
23278 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
23280 2017-06-16 Richard Biener <rguenther@suse.de>
23282 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
23283 initializer to workaround ICE in host GCC 4.8.
23285 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23287 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
23289 (clone_inlined_nodes): Update.
23291 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23293 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
23294 prefetch settings, and enable prefetching by default at -O3.
23296 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23298 * config/aarch64/aarch64.c (aarch64_override_options_internal):
23299 Set flag_prefetch_loop_arrays according to tuning data.
23301 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23303 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
23304 New tune structure.
23305 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
23306 [Unrelated to main purpose of the patch] Place the pointer field last
23307 to enable type checking errors when tune structure are wrongly merged.
23308 * config/aarch64/aarch64.c (generic_prefetch_tune,)
23309 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
23310 (thunderx2t99_prefetch_tune): New tune constants.
23311 (tune_params *_tunings): Update all tunings (no functional change).
23312 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
23313 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
23314 from tunings structures.
23316 2017-06-16 Jakub Jelinek <jakub@redhat.com>
23319 * ubsan.c (instrument_null): Add T argument, use it instead
23320 of computing it based on IS_LHS.
23321 (instrument_object_size): Likewise.
23322 (pass_ubsan::execute): Adjust instrument_null and
23323 instrument_object_size callers to pass gimple_get_lhs or
23324 gimple_assign_rhs1 result to it. Use instrument_null instead of
23325 calling get_base_address and instrument_mem_ref. Handle
23326 aggregate call arguments for object-size sanitization.
23328 2017-06-16 Yury Gribov <tetra2005@gmail.com>
23330 PR tree-optimization/81089
23331 * tree-vrp.c (is_masked_range_test): Validate operands of
23334 2017-06-15 Martin Sebor <msebor@redhat.com>
23337 * dumpfile.c (dump_register): Avoid calling memset to initialize
23338 a class with a default ctor.
23339 * gcc.c (struct compiler): Remove const qualification.
23340 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
23341 * hash-table.h: Ditto.
23342 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
23344 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
23345 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
23347 * params.h (struct param_info): Make struct members non-const.
23348 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
23349 with default initialization.
23350 * vec.h (vec_copy_construct, vec_default_construct): New helper
23352 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
23353 with vec_copy_construct.
23354 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
23355 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
23356 * doc/invoke.texi (-Wclass-memaccess): Document.
23358 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23360 * emit-rtl.h (is_leaf): Update comment about local
23361 register allocator.
23363 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23366 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
23367 for a variable to have a section before checking if the section has a
23369 Set section to.persistent if persistent attribute is set.
23370 Warn if .persistent attribute is used on an automatic variable.
23372 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
23374 PR rtl-optimization/80474
23375 * reorg.c (update_block): Do not ignore instructions in a delay slot.
23377 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
23379 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
23382 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
23384 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
23385 (casesi): Emit bounds checking as RTL.
23386 (casesi_internal_mips16_<mode>): Remove bounds checking.
23388 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
23390 * config/xtensa/xtensa.c (xtensa_option_override): Append
23391 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
23392 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
23393 xtensa_doloop_hooks): Define unconditionally.
23394 (xtensa_reorg_loops): Only call reorg_loops in the presence of
23396 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
23397 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
23398 for it in xtensa_option_override.
23399 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
23400 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
23402 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
23404 * doc/cppopts.texi: Document '-' special value to -MF.
23406 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
23408 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
23409 (cortex_a53_fconst): Likewise.
23410 (cortex_a53_fpmul): Likewise.
23411 (cortex_a53_f_load_64): Likewise.
23412 (cortex_a53_f_load_many): Likewise.
23413 (cortex_a53_advsimd_alu): Likewise.
23414 (cortex_a53_advsimd_alu_q): Likewise.
23415 (cortex_a53_advsimd_mul): Likewise.
23416 (cortex_a53_advsimd_mul_q): Likewise.
23417 (fpmac bypass): Add new bypass for fpmac-fpmac case.
23418 Add missing fmul, r2f_cvt and fconst cases.
23420 2017-06-14 Richard Biener <rguenther@suse.de>
23422 PR middle-end/81088
23423 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
23425 (fold_binary_loc): When associating do not treat pre-existing
23426 TREE_OVERFLOW on literal constants as a reason to allow
23427 TREE_OVERFLOW on associated literal constants.
23429 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23431 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
23432 (MASK_FEATURES): New macro.
23433 * config/sparc/sparc.c (sparc_option_override): Remove the special
23434 handling of -mfpu and generalize it to all MASK_FEATURES switches.
23436 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23438 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
23439 a division of 0 if non-call exceptions are enabled.
23441 2017-06-14 Andrew Pinski <apinski@cavium.com>
23442 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23445 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
23446 Improve vector initialization code gen for only variable case.
23448 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23450 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
23452 2017-06-14 Richard Biener <rguenther@suse.de>
23454 PR tree-optimization/81083
23455 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
23458 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23460 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
23461 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
23462 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
23463 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
23464 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
23465 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
23467 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23469 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
23470 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
23472 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23474 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
23476 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23478 * config/rs6000/t-rtems: Don't handle SPE.
23480 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23482 * config/rs6000/t-linux: Don't handle SPE.
23484 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23486 * config/rs6000/eabispe.h: Delete file.
23488 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23490 * config/rs6000/t-spe: Delete file.
23492 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23494 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
23495 (rs6000_legitimate_offset_address_p): Return false for anything in
23496 V2SImode or V2SFmode.
23498 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23500 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
23501 except V2SF and V2SI. Rearrange the vector modes, and add comments.
23502 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
23504 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
23505 (rs6000_legitimate_offset_address_p): Ditto.
23506 (rs6000_emit_move): Ditto.
23507 (rs6000_init_builtins): Remove V4HI_type_node.
23509 2017-06-13 Martin Liska <mliska@suse.cz>
23512 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
23513 (gate_asan): Likewise.
23514 * asan.h (asan_no_sanitize_address_p): Remove the function.
23515 (sanitize_flags_p): New function.
23516 * builtins.def: Fix coding style.
23517 * common.opt: Use renamed enum value.
23518 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
23519 * doc/extend.texi: Document no_sanitize attribute.
23520 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
23521 to SANITIZE_UNDEFINED_NONDEFAULT.
23522 * gcc.c (sanitize_spec_function): Use the renamed enum value.
23523 * gimple-fold.c (optimize_atomic_compare_exchange_p):
23524 Use sanitize_flags_p.
23525 * gimplify.c (gimplify_function_tree): Likewise.
23526 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
23527 * opts.c (parse_no_sanitize_attribute): New function.
23528 (common_handle_option): Use renamed enum value.
23529 * opts.h (parse_no_sanitize_attribute): Declare.
23530 * tree.c (sanitize_flags_p): New function.
23531 * tree.h: Declared here.
23532 * tsan.c: Use sanitize_flags_p.
23533 * ubsan.c (ubsan_expand_null_ifn): Likewise.
23534 (instrument_mem_ref): Likewise.
23535 (instrument_bool_enum_load): Likewise.
23536 (do_ubsan_in_current_function): Remove the function.
23537 (pass_ubsan::execute): Use sanitize_flags_p.
23538 * ubsan.h: Remove do_ubsan_in_current_function
23539 * tree-cfg.c (print_no_sanitize_attr_value): New function.
23540 (dump_function_to_file): Use it here.
23542 2017-06-13 Martin Jambor <mjambor@suse.cz>
23544 PR tree-optimization/80803
23545 PR tree-optimization/81063
23546 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
23547 (propagate_subaccesses_across_link): Enqueue subtree whenever
23548 necessary instead of relying on the caller.
23550 2017-06-13 Martin Jambor <mjambor@suse.cz>
23552 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
23553 that have a first_link.
23554 (sort_and_splice_var_accesses): Do not check first_link before
23556 (subtree_mark_written_and_enqueue): Likewise.
23557 (propagate_all_subaccesses): Likewise and do not stop at first
23558 parent with a first_link.
23560 2017-06-13 Martin Jambor <mjambor@suse.cz>
23562 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
23565 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23567 * match.pd: New pattern.
23569 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23571 * tree-vrp.c (is_masked_range_test): New function.
23572 (register_edge_assert_for): Determine ranges for
23575 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23577 PR tree-optimization/67328
23578 * fold-const.c (maskable_range_p): New function.
23579 (build_range_check): Generate bittests if possible.
23581 2017-06-13 Martin Liska <mliska@suse.cz>
23583 * gimple-pretty-print.c (dump_probability): Add new argument.
23584 (dump_edge_probability): Dump both probability and count.
23585 (dump_gimple_label): Likewise.
23586 (dump_gimple_bb_header): Likewise.
23588 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
23591 * config/avr/avr-devices.c: Fix indentation.
23592 * config/avr/gen-avr-mmcu-specs.c: Dito.
23594 2017-06-13 Richard Biener <rguenther@suse.de>
23596 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
23597 instead get vector type from stmt_info.
23598 (vectorizable_reduction): Adjust. Remove dead code.
23600 2017-06-13 Richard Biener <rguenther@suse.de>
23602 PR middle-end/81065
23603 * fold-const.c (extract_muldiv_1): Remove bogus distribution
23604 case of C * (x * C2 + C3).
23605 (fold_addr_of_array_ref_difference): Properly fold index difference.
23607 2017-06-12 David S. Miller <davem@davemloft.net>
23610 * config/sparc/sparc.md (return expander): Emit frame blockage if
23611 function uses alloca.
23613 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
23615 * combine.c (make_field_assignment): Check len rather than the mode
23616 precision when calling force_to_mode.
23618 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
23620 Support multilibs and devices that see flash in RAM address range.
23623 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
23624 (avr_mcu_t) <flash_pm_offset>: New field.
23625 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
23626 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
23627 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
23628 (AVR_TINY_PM_OFFSET): Remove macro.
23629 * config/avr/avr.opt (-mshort-calls): New option.
23630 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23631 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
23632 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
23633 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
23634 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
23635 instead of avr_arch->have_jmp_call.
23636 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
23637 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
23638 avr_arch->flash_pm_offset to define.
23639 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
23640 new field flash_pm_offset. Add entry for avrxmega3.
23641 (avr_texinfo): Add entry for avrxmega3.
23642 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
23643 attiny212, attiny214,
23644 attiny412, attiny414, attiny416, attiny417,
23645 attiny814, attiny816, attiny817,
23646 attiny1614, attiny1616, attiny1617,
23647 attiny3214, attiny3216, attiny3217.
23648 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
23649 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
23650 (avr_print_operand_address) [AVR_TINY]: Same.
23651 (avr_asm_init_sections) <readonly_data_section>: Only patch
23652 callback if avr_arch->flash_pm_offset = 0.
23653 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
23654 for rodata if avr_arch->flash_pm_offset != 0.
23655 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
23656 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
23657 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
23658 (m_options): Append opt_rcall.
23659 (m_dirnames): Append dir_rcall.
23660 * config/avr/t-multilib: Regenerate.
23662 * configure.ac [target=avr]: Check whether avrxmega3 default
23663 linker description file works as needed.
23664 * configure: Regenerate.
23665 * doc/avr-mmcu.texi: Regenerate.
23666 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
23667 <__AVR_ARCH__>: Document avrxmega3 and 103.
23668 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
23669 <__AVR_SHORT_CALLS__>: Document it.
23670 <__AVR_PM_BASE_ADDRESS__>: Document it.
23671 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
23672 (AVR Variable Attributes) <progmem>: Document this is
23673 not needed for avrxmega3.
23674 (AVR Named Address Spaces) <__flash>: Dito.
23676 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
23678 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
23680 2017-06-12 Doug Rupp <rupp@adacore.com>
23682 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
23683 Append vxworks-stdint.h to the tm_file list.
23684 * config/vxworks-stdint.h: New file.
23686 2017-06-12 Martin Liska <mliska@suse.cz>
23688 PR tree-optimization/81041
23689 * tree-profile.c (gimple_gen_ic_func_profiler):
23690 Create an extra BB in profile-generate
23691 (gimple_gen_time_profiler): Likewise.
23693 2017-06-12 Jakub Jelinek <jakub@redhat.com>
23695 PR tree-optimization/81003
23696 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
23697 (update_range_test): Use it instead of force_gimple_operand_gsi.
23699 2017-06-12 Richard Biener <rguenther@suse.de>
23701 PR tree-optimization/81053
23702 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
23703 with backedge value not defined in loop. Simplify def stmt
23706 2017-06-11 Tom de Vries <tom@codesourcery.com>
23709 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
23711 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
23712 nvptx_cannot_force_const_mem.
23714 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23716 * opts.c (finish_options): Move test for flag_split_stack after
23717 it has been initialized.
23719 2017-06-11 Jason Merrill <jason@redhat.com>
23721 * tree.h (id_equal): New.
23722 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
23723 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
23724 instead of strcmp of IDENTIFIER_POINTER.
23726 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23728 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
23729 (mark_all_inlined_calls_cdtor): Fix formating.
23730 (inline_transform): Rescale profile before inlining.
23732 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23734 * cgraph.h (cgraph_edge::clone): Update prototype.
23735 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
23736 (cgraph_node::create_clone): Update.
23737 (cgraph_node::create_version_clone): Update.
23738 * tree-inline.c (copy_bb): Update.
23739 (expand_call_inline): Update.
23741 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
23743 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
23744 factored out from ...
23745 (rs6000_emit_prologue): ... here.
23747 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
23749 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
23750 factored out from ...
23751 (rs6000_emit_prologue): ... here.
23753 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23755 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
23757 (handle_missing_profiles): Fix computation of tp_first_run.
23758 (counts_to_freqs): Do not touch freqs when count is 0.
23760 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23762 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
23765 2017-06-10 Tom de Vries <tom@codesourcery.com>
23767 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23768 attributes): Document signal effective target.
23770 2017-06-10 Tom de Vries <tom@codesourcery.com>
23772 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23773 Document effective target stack_size.
23775 2017-06-09 David Malcolm <dmalcolm@redhat.com>
23777 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
23778 to the edit_context if they can be auto-applied.
23780 2017-06-9 Ian Lance Taylor <iant@golang.org>
23782 * opts.c (finish_options): If -fsplit-stack, disable implicit
23783 -forder-blocks-and-partition.
23784 * doc/invoke.texi (Optimize Options): Document that when using
23785 -fsplit-stack -forder-blocks-and-partition is not implicitly
23788 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23790 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
23791 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
23792 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
23793 * builtins.def (abort, trap, unreachable): Declare cold.
23794 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
23795 * tree-core.h (ECF_COLD): New.
23796 * tree.c (set_call_expr_flags): Handle ECF_COLD.
23797 (build_common_builtin_nodes): Mark unreachable and abort as cold.
23799 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23801 * predict.c (unlikely_executed_stmt_p): Cleanup.
23803 2017-06-09 Richard Biener <rguenther@suse.de>
23805 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
23806 model if the ref is always written to.
23808 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23810 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
23812 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23814 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
23817 2017-06-09 Tom de Vries <tom@codesourcery.com>
23820 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
23821 "target cannot support label values" when encountering LABEL_REF.
23823 2017-06-09 Martin Liska <mliska@suse.cz>
23825 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
23826 (gimple_gen_ic_func_profiler): Emit direct comparison
23827 of __gcov_indirect_call_callee with NULL.
23828 (gimple_gen_time_profiler): Change probability from
23829 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
23831 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23833 * profile.c (edge_gcov_counts): Turn to pointer.
23834 (compute_branch_probabilities, compute_branch_probabilities): Update.
23835 (branch_prob): Do not clear edge_gcov_count.
23836 * profile.h (edge_gcov_counts): Turn to pointer.
23837 (edge_gcov_count): Update.
23839 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23841 * gimple.h (gimple_check_failed): Mark cold.
23843 2017-06-09 Richard Biener <rguenther@suse.de>
23845 PR tree-optimization/66623
23846 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
23847 refactor check_reduction into two parts, properly computing
23848 whether we have to check reduction validity for outer loop
23851 2017-06-09 Richard Biener <rguenther@suse.de>
23853 PR tree-optimization/79483
23854 * graphite-scop-detection.c (order): New global.
23855 (get_order): Compute bb to order mapping that satisfies code
23856 generation constraints.
23857 (cmp_pbbs): New helper.
23858 (build_scops): Start domwalk at entry block, sort generated
23861 2017-06-09 Richard Biener <rguenther@suse.de>
23863 PR middle-end/81007
23864 * ipa-polymorphic-call.c
23865 (ipa_polymorphic_call_context::restrict_to_inner_class):
23866 Skip FIELD_DECLs with error_mark_node type.
23867 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
23870 2017-06-09 Martin Liska <mliska@suse.cz>
23872 * predict.c (struct branch_predictor): New struct.
23873 (test_prediction_value_range): New test.
23874 (predict_c_tests): New function.
23875 * selftest-run-tests.c (selftest::run_tests): Run the function.
23876 * selftest.h: Declare new tests.
23878 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
23881 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
23882 gen_add3_insn did not fail.
23883 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
23884 r0, construct that number in a temporary reg and add that reg to r0.
23885 If asked to put the result in r0 as well, fail.
23887 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
23889 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23890 for early expansion of vec_eqv.
23892 2017-06-08 Jakub Jelinek <jakub@redhat.com>
23894 PR middle-end/81005
23895 * ubsan.c (instrument_null): Avoid pointless code temporary.
23896 (pass_ubsan::execute): Instrument aggregate arguments of calls.
23898 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
23902 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
23905 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
23908 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23910 * predict.c (unlikely_executed_edge_p): Move ahead.
23911 (probably_never_executed_edge_p): Use it.
23913 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
23915 PR middle-end/79988
23916 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
23917 gimple_call_builtin_p call.
23919 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23921 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
23922 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
23923 rtl_check_failed_type2, rtl_check_failed_code1,
23924 rtl_check_failed_code2, rtl_check_failed_code_mode,
23925 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
23926 rtvec_check_failed_bounds, rtl_check_failed_flag,
23927 _fatal_insn_not_found, _fatal_insn): Likewise.
23928 * tree.h (tree_contains_struct_check_failed,
23929 tree_check_failed, tree_not_check_failed,
23930 tree_class_check_failed, tree_range_check_failed,
23931 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
23932 tree_vec_elt_check_failed, phi_node_elt_check_failed,
23933 tree_operand_check_failed, omp_clause_check_failed,
23934 omp_clause_operand_check_failed, omp_clause_range_check_failed):
23937 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23939 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
23940 flag_branch_probabilities.
23941 * ipa-inline.c (edge_badness): Likewise.
23942 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
23943 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
23944 * predict.c (maybe_hot_frequency_p): Likewise.
23945 (probably_never_executed): Likewise.
23946 * sched-ebb.c (schedule_ebbs): Likewise.
23947 * sched-rgn.c (find_single_block_region): Likewise.
23948 * tracer.c (tail_duplicate): Likewise.
23950 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23952 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
23953 longer requires x_flag_profile_use.
23955 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23957 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
23958 instead of flag_reorder_blocks_and_partition.
23959 * dbxout.c (dbxout_function_end): Likewise.
23960 * dwarf2out.c (gen_subprogram_die): Likewise.
23961 * haifa-sched.c (sched_create_recovery_edges): Likewise.
23962 * hw-doloop.c (reorg_loops): Likewise.
23963 * varasm.c (assemble_start_function,
23964 assemble_end_function): Likewise.
23965 (decide_function_section): Do not check for
23966 flag_reorder_blocks_and_partition.
23968 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
23970 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23972 (chkp_get_hard_register_fake_addr_expr): Ditto.
23973 (chkp_build_addr_expr): Add check for hard reg case.
23974 (chkp_parse_array_and_component_ref): Ditto.
23975 (chkp_find_bounds_1): Ditto.
23976 (chkp_process_stmt): Don't generate bounds store for
23979 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23981 * predict.c (maybe_hot_bb_p): Do not check profile status.
23982 (maybe_hot_edge_p): Likewise.
23983 (probably_never_executed): Check for zero counts even if profile
23985 (unlikely_executed_edge_p): New function.
23986 (unlikely_executed_stmt_p): New function.
23987 (unlikely_executed_bb_p): New function.
23988 (set_even_probabilities): Use unlikely predicates.
23989 (combine_predictions_for_bb): Likewise.
23990 (predict_paths_for_bb): Likewise.
23991 (predict_paths_leading_to_edge): Likewise.
23992 (determine_unlikely_bbs): New function.
23993 (estimate_bb_frequencies): Use it.
23994 (compute_function_frequency): Use zero counts even if profile is
23996 * profile-count.h: Fix typo.
23998 2017-08-08 Julia Koval <julia.koval@intel.com>
24000 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
24001 _mm512_mask_cvtsepi16_storeu_epi8,
24002 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
24003 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
24004 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
24005 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
24006 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
24007 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
24008 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
24009 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
24010 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
24011 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
24012 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
24013 __builtin_ia32_pmovuswb256mem_mask,
24014 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
24015 __builtin_ia32_pmovwb512mem_mask): New builtins.
24017 2017-08-08 Julia Koval <julia.koval@intel.com>
24019 PR target/73350,80862
24020 * config/i386/subst.md (round): Fix round pattern.
24021 * config/i386/i386.c (ix86_erase_embedded_rounding):
24022 Fix erasing rounding for the fixed pattern.
24024 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24026 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
24028 2017-06-08 Martin Liska <mliska@suse.cz>
24030 PR gcov-profile/80911
24031 * gcov.c (block_info::block_info): New constructor.
24033 2017-06-07 Carl Love <cel@us.ibm.com>
24035 * config/rs6000/rs6000-c: The return type of the following
24036 built-in functions was implemented as int not long long. Fix sign
24037 of return value for the unsigned version of vec_mulo and vec_mule.
24038 vector unsigned long long vec_bperm (vector unsigned long long,
24039 vector unsigned char)
24040 vector signed long long vec_mule (vector signed int,
24042 vector unsigned long long vec_mule (vector unsigned int,
24043 vector unsigned int)
24044 vector signed long long vec_mulo (vector signed int,
24046 vector unsigned long long vec_mulo (vector unsigned int,
24047 vector unsigned int)
24048 * doc/extend.texi: Fix the documentation for the built-in
24051 2017-06-07 Carl Love <cel@us.ibm.com>
24054 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
24057 2017-06-07 Carl Love <cel@us.ibm.com>
24059 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
24060 support, Generate doublehv for signed int/float for BE case only.
24062 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
24064 * doc/invoke.texi (mcx16): Rewrite.
24066 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24068 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
24069 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
24070 *mov<mode>_softfloat, and an anonymous splitter): Use
24071 nonimmediate_operand instead of rs6000_nonimmediate_operand.
24073 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24075 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
24077 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
24078 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
24079 (rs6000_debug_reg_global): Adjust.
24080 (rs6000_init_hard_regno_mode_ok): Adjust.
24081 (rs6000_dbx_register_number): Adjust.
24082 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
24083 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
24084 Remove SPE_ACC and SPEFSCR.
24085 (REG_ALLOC_ORDER): Ditto.
24086 (FRAME_POINTER_REGNUM): Change to 111.
24087 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
24088 (REG_CLASS_NAMES): Ditto.
24089 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
24090 (REGISTER_NAMES): Ditto.
24091 (ADDITIONAL_REG_NAMES): Ditto.
24092 (rs6000_reg_names): Ditto.
24093 * config/rs6000/rs6000.md: Renumber some register number
24096 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24098 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
24100 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
24101 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
24103 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
24104 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
24105 Delete the SPE high registers.
24106 (REG_ALLOC_ORDER): Ditto.
24107 (enum reg_class): Remove SPE_HIGH_REGS.
24108 (REG_CLASS_NAMES): Ditto.
24109 (REG_CLASS_CONTENTS): Delete the SPE high registers.
24110 (REGISTER_NAMES): Ditto.
24111 (rs6000_reg_names): Ditto.
24112 * doc/tm.texi.in: Remove SPE as example.
24113 * doc/tm.texi: Regenerate.
24115 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24117 * config/rs6000/8540.md (ppc8540_brinc): Delete.
24118 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
24119 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
24120 * config/rs6000/rs6000.md (type): Remove "brinc".
24122 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24124 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
24125 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
24126 * config/rs6000/linuxspe.h: Delete file.
24127 * config/rs6000/rs6000.md: Don't include spe.md.
24128 * config/rs6000/spe.h: Delete file.
24129 * config/rs6000/spe.md: Delete file.
24130 * config/rs6000/t-rs6000: Remove spe.md.
24132 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24134 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
24135 (reg_or_none500mem_operand): Delete.
24136 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
24137 instead of reg_or_none500mem_operand.
24139 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24141 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
24142 handling of SPE flags.
24143 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
24145 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24147 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
24149 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
24150 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
24151 paired_divv2sf3): Similar.
24152 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
24153 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
24154 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
24156 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
24157 Rename the paired_* instruction patterns.
24158 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
24160 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
24161 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
24162 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
24163 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
24164 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
24165 PAIRED_VECTOR_MODE.
24166 (struct machine_function): Delete field spe_insn_chain_scanned_p.
24167 (spe_func_has_64bit_regs_p): Delete.
24168 (spe_expand_predicate_builtin): Delete.
24169 (spe_expand_evsel_builtin): Delete.
24170 (TARGET_DWARF_REGISTER_SPAN): Do not define.
24171 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
24172 (invalid_e500_subreg): Delete.
24173 (rs6000_legitimize_address): Always force_reg op2 as well, for
24174 paired single memory accesses.
24175 (rs6000_member_type_forces_blk): Delete.
24176 (rs6000_spe_function_arg): Delete.
24177 (rs6000_expand_unop_builtin): Delete SPE handling.
24178 (rs6000_expand_binop_builtin): Ditto.
24179 (spe_expand_stv_builtin): Delete.
24180 (bdesc_2arg_spe): Delete.
24181 (spe_expand_builtin): Delete.
24182 (spe_expand_predicate_builtin): Delete.
24183 (spe_expand_evsel_builtin): Delete.
24184 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
24185 (spe_init_builtins): Delete.
24186 (spe_func_has_64bit_regs_p): Delete.
24187 (savres_routine_name): Delete "info" parameter. Adjust callers.
24188 (rs6000_emit_stack_reset): Ditto.
24189 (rs6000_dwarf_register_span): Delete.
24190 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
24191 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
24192 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
24194 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
24196 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
24197 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
24198 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
24199 mulv2sf3, divv2sf3): Delete expanders.
24201 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24203 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
24205 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24207 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
24208 * config/rs6000/rs6000.c: Ditto.
24210 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24212 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
24213 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
24214 comparison_operator.
24216 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24218 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
24219 * config/rs6000/rs6000.opt: Ditto.
24220 * config/rs6000/t-rtems: Ditto.
24222 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24224 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
24225 TARGET_E500_SINGLE by 0, simplify.
24226 * config/rs6000/rs6000.c: Ditto.
24227 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
24228 (spe_build_register_parallel): Delete.
24229 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
24230 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
24231 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
24232 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
24233 (E500_CONVERT): Delete.
24234 * config/rs6000/spe.md: Remove many patterns and all define_constants.
24236 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24238 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
24239 * config/rs6000/dfp.md: Ditto.
24240 (negdd2, *negdd2_fpr): Merge.
24241 (absdd2, *absdd2_fpr): Merge.
24242 (negtd2, *negtd2_fpr): Merge.
24243 (abstd2, *abstd2_fpr): Merge.
24244 * config/rs6000/e500.h: Delete file.
24245 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
24246 TARGET_FPRS by 1 and simplify.
24247 * config/rs6000/rs6000-c.c: Ditto.
24248 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
24249 TARGET_DF_SPE by 0.
24250 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
24252 * config/rs6000/rs6000.md: Ditto.
24253 (floatdidf2, *floatdidf2_fpr): Merge.
24254 (move_from_CR_gt_bit): Delete.
24255 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
24256 (E500_CR_IOR_COMPARE): Delete.
24257 (All patterns that require !TARGET_FPRS): Delete.
24258 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
24260 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24262 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
24264 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24266 * graphds.c (add_edge): Intitialize edge's attached data.
24267 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
24268 pointer parameter. Call pointed function on each edge during
24269 graph traversing. Skip traversing the edge when the function
24271 (graphds_dfs, graphds_scc): Ditto.
24272 (for_each_edge): New parameter. Pass the new parameter to callback
24274 * graphds.h (skip_edge_callback): New function pointer type.
24275 (graphds_dfs, graphds_scc): New function pointer parameter.
24276 (graphds_edge_callback, for_each_edge): New parameter.
24278 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24280 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
24281 out code checking if runtime alias check is possible to below ...
24282 Call the new function.
24283 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
24284 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
24286 2017-06-07 Marek Polacek <polacek@redhat.com>
24289 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
24290 TYPE_OVERFLOW_WRAPS check.
24292 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24294 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
24295 if versioning is required.
24296 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
24297 peeling with the check for versioning.
24299 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24301 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
24302 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
24303 Set true to new parameter if new ssa variable is defined.
24304 (vect_gen_vector_loop_niters): Refactor. Set range information
24305 for the new vector loop bound variable.
24306 (vect_do_peeling): Ditto.
24308 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24310 * tree-affine.c (ssa.h): Include header file.
24311 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
24312 has wrapping overflow behavior.
24314 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24316 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
24318 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24320 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
24321 (tree_to_aff_combination): ... here.
24323 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24325 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
24326 reg_pressure model function.
24327 (ivopts_global_cost_for_size): Delete.
24328 (determine_set_costs, iv_ca_recount_cost): Call new model function
24329 ivopts_estimate_reg_pressure.
24331 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24333 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
24334 expensive than udiv. Remove floating point cases from mod.
24336 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24338 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
24339 Increase idiv cost.
24341 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24343 * config/aarch64/aarch64.md
24344 (copysignsf3): Fix mask generation.
24346 2017-06-07 Jakub Jelinek <jakub@redhat.com>
24348 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
24350 (class dump_manager): Add register_dumps method.
24351 * dumpfile.c: Include langhooks.h.
24352 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
24353 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
24354 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
24355 (dump_manager::dump_register): Start with 512 entries instead of 32.
24356 (dump_manager::register_dumps): New method.
24357 * toplev.c (general_init): Instead of invoking register_dumps
24358 langhook, invoke register_dumps method on the dump manager.
24359 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
24362 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
24364 * doc/md.texi: Clarify the restrictions on a define_insn condition.
24365 Say that # requires an associated define_split to exist, and that
24366 the define_split must be suitable for use after register allocation.
24368 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24370 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
24371 (compute_outgoing_frequencies): Also initialize zero counts.
24372 (find_many_sub_basic_blocks): Do not produce uninitialized profile
24373 around loops; preserve more of profile when nothing changes.
24375 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
24377 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
24379 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
24380 * config/arm/arm-cpu-cdata.h: Regenerate.
24381 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
24382 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
24383 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
24384 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
24385 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
24387 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
24389 * config/arm/t-rmprofile: Likewise.
24390 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
24392 2017-06-06 David S. Miller <davem@davemloft.net>
24395 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
24396 blockage if function uses alloca.
24398 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24400 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
24401 New "uid" fields to hold pretty-print IDs of group and ref.
24402 Memory references are now identified as <group_id>:<ref_id>
24403 instead of using [random] addresses.
24404 (dump_mem_details): Simplify, no functional change.
24405 (dump_mem_ref): Simplify and make output more concise.
24406 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
24407 (find_or_create_group): Initialize group uid.
24408 (record_ref): Initialize ref uid. Improve debug output.
24409 (prune_group_by_reuse, should_issue_prefetch_p,)
24410 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
24411 (mark_nontemporal_store, determine_loop_nest_reuse):
24412 Improve debug output.
24414 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24416 * dbgcnt.def (prefetch): New debug counter.
24417 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
24418 (schedule_prefetches): Stop issueing prefetches if debug counter
24421 2017-06-06 Tom de Vries <tom@codesourcery.com>
24423 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
24424 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
24426 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24428 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
24429 Use aarch64_reg_or_zero predicate for operand 4.
24430 (aarch64_compare_and_swap<mode> define_insn_and_split):
24431 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
24432 (aarch64_store_exclusive<mode>): Likewise for operand 2.
24434 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24436 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
24437 (arm_compute_save_core_reg_mask): This.
24438 (thumb1_compute_save_reg_mask): Rename into ...
24439 (thumb1_compute_save_core_reg_mask): This.
24440 (arm_compute_save_reg0_reg12_mask): Adapt comment.
24441 (arm_compute_frame_layout): Likewise.
24443 2017-06-06 Richard Biener <rguenther@suse.de>
24445 PR tree-optimization/80974
24446 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
24447 keep or clear leaders SSA info.
24449 2017-06-06 Tom de Vries <tom@codesourcery.com>
24451 * config/nvptx/nvptx.c (split_mode_p): New function.
24452 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
24454 2017-06-06 Tom de Vries <tom@codesourcery.com>
24456 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
24458 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24461 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
24464 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24466 * shrink-wrap.c (handle_simple_exit): Update profile.
24467 (try_shrink_wrapping): Upate profile.
24469 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24471 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
24472 (tree_guess_outgoing_edge_probabilities): New.
24473 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
24474 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
24476 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24478 * ipa-split.c (split_function): Initialize return bb profile.
24480 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24482 * profile.c (compute_branch_probabilities): Also initialize
24483 EXIT_BLOCK profile.
24485 2017-06-06 Richard Biener <rguenther@suse.de>
24487 PR tree-optimization/80928
24488 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
24489 (vect_analyze_loop_operations): Properly guard analysis for
24491 (vect_transform_loop): Likewise.
24492 (vect_analyze_loop_2): Also reset SLP type on PHIs.
24493 (vect_model_induction_cost): Do not cost for pure SLP.
24494 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
24495 of induction in inner loop vectorization.
24496 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
24497 (vect_get_and_check_slp_defs): Handle vect_induction_def.
24498 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
24500 (vect_analyze_slp_cost_1): Cost induction.
24501 (vect_detect_hybrid_slp_stmts): Handle PHIs.
24502 (vect_get_slp_vect_defs): Likewise.
24503 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
24504 (vect_transform_stmt): Handle SLP reductions.
24505 * tree-vectorizer.h (vectorizable_induction): Adjust.
24507 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24509 * config/rs6000/rs6000.c (make_resolver_func): Update
24510 init_lowered_empty_function call.
24512 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
24514 * doc/invoke.texi: Document the -fprofile-abs-path option.
24515 * common.opt (fprofile-abs-path): New option.
24516 * gcov-io.h (gcov_write_filename): Declare.
24517 * gcov-io.c (gcov_write_filename): New function.
24518 * coverage.c (coverage_begin_function): Use gcov_write_filename.
24519 * profile.c (output_location): Likewise.
24521 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24523 * shring-wrap.c: Revert accidental commit.
24525 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
24527 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
24529 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24531 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
24533 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
24534 profile in callgraph edge.
24535 * profile-count.h (apply_probability): If THIS is 0, then result is 0
24536 (apply_scale): Likewise.
24537 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
24538 Also scale profile when inlining function with zero profile.
24539 (initialize_cfun): Update exit block profile even when it is zero.
24540 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
24541 when profile is read.
24543 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24545 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
24546 (CLONE_*): New constants to define the processors we can generate
24547 code for with the target_clone attribute.
24548 (rs6000_clone_map): New array to identify which clone processors
24549 the current program is running on.
24550 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
24551 target_clone attribute.
24552 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
24553 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
24554 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
24555 (cpu_expand_builtin): Add support for target_clone attribute.
24556 (rs6000_valid_attribute_p): Allow "default" attribute.
24557 (get_decl_name): New debug function to simplify printing the
24558 current function name in debugging statements.
24559 (rs6000_clone_priority): New functions to support the target_clone
24560 attribute, and be able to generate code to switch between ISA 2.05
24561 through ISA 3.0 (power6 through power9).
24562 (rs6000_compare_version_priority): Likewise.
24563 (rs6000_get_function_versions_dispatcher): Likewise.
24564 (make_resolver_func): Likewise.
24565 (add_condition_to_bb): Likewise.
24566 (dispatch_function_versions): Likewise.
24567 (rs6000_generate_version_dispatcher_body): Likewise.
24568 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
24569 (fusion_gpr_load_p): Fix a spacing issue.
24570 * doc/extend.texi (Common Function Attributes): Document that the
24571 PowerPC supports the target_clone attribute.
24573 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24575 * config/arm/arm.h: explain F symbol found in description of ARM
24576 register allocation in its legend.
24578 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24580 * config/mips/frame-header-opt.c: Include profile-count.h.
24581 * config/riscv/riscv.c: Include profile-count.h
24583 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24585 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
24587 (sm_set_flag_if_changed): Add bbs field.
24588 (execute_sm_if_changed_flag_set): Pass BBS.
24589 (execute_sm): Update.
24591 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24593 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
24596 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24598 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
24599 (peephole2): New peephole2 to emit the above.
24600 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
24602 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24604 * config/aarch64/aarch64.c (define_peephole2 above
24605 *sub_<shift>_<mode>): New peephole.
24607 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24609 * config/i386/i386.c (make_resolver_func): Update.
24610 * Makefile.in: Add profile-count.h and profile-count.o
24611 * auto-profile.c (afdo_indirect_call): Update to new API.
24612 (afdo_set_bb_count): Update.
24613 (afdo_propagate_edge): Update.
24614 (afdo_propagate_circuit): Update.
24615 (afdo_calculate_branch_prob): Update.
24616 (afdo_annotate_cfg): Update.
24617 * basic-block.h: Include profile-count.h
24618 (struct edge_def): Turn count to profile_count.
24619 (struct basic_block_def): Likewie.
24620 (REG_BR_PROB_BASE): Move to profile-count.h
24621 (RDIV): Move to profile-count.h
24622 * bb-reorder.c (max_entry_count): Turn to profile_count.
24623 (find_traces): Update.
24624 (rotate_loop):Update.
24625 (connect_traces):Update.
24626 (sanitize_hot_paths):Update.
24627 * bt-load.c (migrate_btr_defs): Update.
24628 * cfg.c (RDIV): Remove.
24629 (init_flow): Use alloc_block.
24630 (alloc_block): Uninitialize count.
24631 (unchecked_make_edge): Uninitialize count.
24632 (check_bb_profile): Update.
24633 (dump_edge_info): Update.
24634 (dump_bb_info): Update.
24635 (update_bb_profile_for_threading): Update.
24636 (scale_bbs_frequencies_int): Update.
24637 (scale_bbs_frequencies_gcov_type): Update.
24638 (scale_bbs_frequencies_profile_count): New.
24639 * cfg.h (update_bb_profile_for_threading): Update.
24640 (scale_bbs_frequencies_profile_count): Declare.
24641 * cfgbuild.c (compute_outgoing_frequencies): Update.
24642 (find_many_sub_basic_blocks): Update.
24643 * cfgcleanup.c (try_forward_edges): Update.
24644 (try_crossjump_to_edge): Update.
24645 * cfgexpand.c (expand_gimple_tailcall): Update.
24646 (construct_exit_block): Update.
24647 * cfghooks.c (verify_flow_info): Update.
24648 (dump_bb_for_graph): Update.
24649 (split_edge): Update.
24650 (make_forwarder_block): Update.
24651 (duplicate_block): Update.
24652 (account_profile_record): Update.
24653 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
24654 (get_estimated_loop_iterations): Update.
24655 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
24656 (single_likely_exit): Update.
24657 * cfgloopmanip.c (scale_loop_profile): Update.
24659 (set_zero_probability): Update.
24660 (lv_adjust_loop_entry_edge): Update.
24661 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24662 (purge_dead_edges): Update.
24663 (rtl_account_profile_record): Update.
24664 * cgraph.c (cgraph_node::create): Uninitialize count.
24665 (symbol_table::create_edge): Uninitialize count.
24666 (cgraph_update_edges_for_call_stmt_node): Update.
24667 (cgraph_edge::dump_edge_flags): Update.
24668 (cgraph_node::dump): Update.
24669 (cgraph_edge::maybe_hot_p): Update.
24670 * cgraph.h: Include profile-count.h
24671 (create_clone), create_edge, create_indirect_edge): Update.
24672 (cgraph_node): Turn count to profile_count.
24673 (cgraph_edge0: Likewise.
24674 (make_speculative, clone): Update.
24675 (create_edge): Update.
24676 (init_lowered_empty_function): Update.
24677 * cgraphclones.c (cgraph_edge::clone): Update.
24678 (duplicate_thunk_for_node): Update.
24679 (cgraph_node::create_clone): Update.
24680 * cgraphunit.c (cgraph_node::analyze): Update.
24681 (cgraph_node::expand_thunk): Update.
24682 * final.c (dump_basic_block_info): Update.
24683 * gimple-streamer-in.c (input_bb): Update.
24684 * gimple-streamer-out.c (output_bb): Update.
24685 * graphite.c (print_global_statistics): Update.
24686 (print_graphite_scop_statistics): Update.
24687 * hsa-brig.c: Include basic-block.h.
24688 * hsa-dump.c: Include basic-block.h.
24689 * hsa-gen.c (T sum_slice): Update.
24690 (convert_switch_statements):Update.
24691 * hsa-regalloc.c: Include basic-block.h.
24692 * ipa-chkp.c (chkp_produce_thunks): Update.
24693 * ipa-cp.c (struct caller_statistics): Update.
24694 (init_caller_stats): Update.
24695 (gather_caller_stats): Update.
24696 (ipcp_cloning_candidate_p): Update.
24697 (good_cloning_opportunity_p): Update.
24698 (get_info_about_necessary_edges): Update.
24699 (dump_profile_updates): Update.
24700 (update_profiling_info): Update.
24701 (update_specialized_profile): Update.
24702 (perhaps_add_new_callers): Update.
24703 (decide_about_value): Update.
24704 (ipa_cp_c_finalize): Update.
24705 * ipa-devirt.c (struct odr_type_warn_count): Update.
24706 (struct decl_warn_count): Update.
24707 (struct final_warning_record): Update.
24708 (possible_polymorphic_call_targets): Update.
24709 (ipa_devirt): Update.
24710 * ipa-fnsummary.c (redirect_to_unreachable): Update.
24711 * ipa-icf.c (sem_function::merge): Update.
24712 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
24713 * ipa-inline.c (compute_uninlined_call_time): Update.
24714 (compute_inlined_call_time): Update.
24715 (want_inline_small_function_p): Update.
24716 (want_inline_self_recursive_call_p): Update.
24717 (edge_badness): Update.
24718 (lookup_recursive_calls): Update.
24719 (recursive_inlining): Update.
24720 (inline_small_functions): Update.
24721 (dump_overall_stats): Update.
24722 (dump_inline_stats): Update.
24723 * ipa-profile.c (ipa_profile_generate_summary): Update.
24724 (ipa_propagate_frequency): Update.
24725 (ipa_profile): Update.
24726 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24727 * ipa-utils.c (ipa_merge_profiles): Update.
24728 * loop-doloop.c (doloop_modify): Update.
24729 * loop-unroll.c (report_unroll): Update.
24730 (unroll_loop_runtime_iterations): Update.
24731 * lto-cgraph.c (lto_output_edge): Update.
24732 (lto_output_node): Update.
24733 (input_node): Update.
24734 (input_edge): Update.
24735 (merge_profile_summaries): Update.
24736 * lto-streamer-in.c (input_cfg): Update.
24737 * lto-streamer-out.c (output_cfg): Update.
24738 * mcf.c (create_fixup_graph): Update.
24739 (adjust_cfg_counts): Update.
24740 (sum_edge_counts): Update.
24741 * modulo-sched.c (sms_schedule): Update.
24742 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
24743 * predict.c (maybe_hot_count_p): Update.
24744 (probably_never_executed): Update.
24745 (dump_prediction): Update.
24746 (combine_predictions_for_bb): Update.
24747 (propagate_freq): Update.
24748 (handle_missing_profiles): Update.
24749 (counts_to_freqs): Update.
24750 (rebuild_frequencies): Update.
24751 (force_edge_cold): Update.
24752 * predict.h: Include profile-count.h
24753 (maybe_hot_count_p, counts_to_freqs): UPdate.
24754 * print-rtl-function.c: Do not include cfg.h
24755 * print-rtl.c: Include basic-block.h
24756 * profile-count.c: New file.
24757 * profile-count.h: New file.
24758 * profile.c (is_edge_inconsistent): Update.
24759 (correct_negative_edge_counts): Update.
24760 (is_inconsistent): Update.
24761 (set_bb_counts): Update.
24762 (read_profile_edge_counts): Update.
24763 (compute_frequency_overlap): Update.
24764 (compute_branch_probabilities): Update; Initialize and deinitialize
24766 (branch_prob): Update.
24767 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
24768 (edge_gcov_count): New.
24769 (bb_gcov_count): New.
24770 * shrink-wrap.c (try_shrink_wrapping): Update.
24771 * tracer.c (better_p): Update.
24772 * trans-mem.c (expand_transaction): Update.
24773 (ipa_tm_insert_irr_call): Update.
24774 (ipa_tm_insert_gettmclone_call): Update.
24775 * tree-call-cdce.c: Update.
24776 * tree-cfg.c (gimple_duplicate_sese_region): Update.
24777 (gimple_duplicate_sese_tail): Update.
24778 (gimple_account_profile_record): Update.
24779 (execute_fixup_cfg): Update.
24780 * tree-inline.c (copy_bb): Update.
24781 (copy_edges_for_bb): Update.
24782 (initialize_cfun): Update.
24783 (freqs_to_counts): Update.
24784 (copy_cfg_body): Update.
24785 (expand_call_inline): Update.
24786 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24787 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
24788 (try_unroll_loop_completely): Update.
24789 (try_peel_loop): Update.
24790 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24791 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
24792 * tree-ssa-loop-split.c (connect_loops): Update.
24793 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
24794 * tree-ssa-reassoc.c (branch_fixup): Update.
24795 * tree-ssa-tail-merge.c (replace_block_by): Update.
24796 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
24797 (compute_path_counts): Update.
24798 (update_profile): Update.
24799 (recompute_probabilities): Update.
24800 (update_joiner_offpath_counts): Update.
24801 (estimated_freqs_path): Update.
24802 (freqs_to_counts_path): Update.
24803 (clear_counts_path): Update.
24804 (ssa_fix_duplicate_block_edges): Update.
24805 (duplicate_thread_path): Update.
24806 * tree-switch-conversion.c (case_bit_test_cmp): Update.
24807 (struct switch_conv_info): Update.
24808 * tree-tailcall.c (decrease_profile): Update.
24809 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24810 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24811 * value-prof.c (check_counter): Update.
24812 (gimple_divmod_fixed_value): Update.
24813 (gimple_mod_pow2): Update.
24814 (gimple_mod_subtract): Update.
24815 (gimple_ic_transform): Update.
24816 (gimple_stringop_fixed_value): Update.
24817 * value-prof.h (gimple_ic): Update.
24819 2017-06-02 Carl Love <cel@us.ibm.com>
24821 * config/rs6000/rs6000-c: Add support for built-in functions
24822 vector double vec_doublee (vector signed int);
24823 vector double vec_doublee (vector unsigned int);
24824 vector double vec_doublee (vector float);
24825 vector double vec_doubleh (vector signed int);
24826 vector double vec_doubleh (vector unsigned int);
24827 vector double vec_doubleh (vector float);
24828 vector double vec_doublel (vector signed int);
24829 vector double vec_doublel (vector unsigned int);
24830 vector double vec_doublel (vector float);
24831 vector double vec_doubleo (vector signed int);
24832 vector double vec_doubleo (vector unsigned int);
24833 vector double vec_doubleo (vector float);.
24834 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
24835 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
24837 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
24838 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
24839 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
24841 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
24842 vec_doublel, vec_doubleh.
24843 * doc/extend.texi: Update the built-in documentation file for the
24844 new built-in functions.
24846 2017-06-02 David Malcolm <dmalcolm@redhat.com>
24849 * ipa-inline-analysis.c (free_growth_caches): Set
24850 edge_removal_hook_holder to NULL after removing it.
24852 2017-06-02 Sudakshina Das <sudi.das@arm.com>
24854 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
24855 comparision with zero.
24857 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
24858 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24859 for early expansion of vec_min and vec_max builtins.
24860 (builtin_function_type): Add min/max unsigned variants to those
24861 identified as having unsigned arguments.
24863 2017-06-02 Olivier Hainque <hainque@adacore.com>
24865 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
24867 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24869 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
24870 Use VALL_F16 iterator rather than VALL.
24872 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24874 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
24875 Emit CBNZ inside loop when doing a strong exchange and comparing
24876 against zero. Generate the CC flags after the loop.
24878 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
24880 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
24881 (dl_section_ref): New.
24882 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
24883 On AIX, append an expression to subtract the size of the
24884 section length to dl_section_ref.
24886 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
24888 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24889 for early expansion of vector absolute builtins.
24891 2017-06-02 Richard Biener <rguenther@suse.de>
24893 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
24894 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
24896 2017-06-02 Richard Biener <rguenther@suse.de>
24898 PR tree-optimization/80948
24899 * tree-tailcall.c (find_tail_calls): Track stmts to move in
24900 stmt order as well.
24902 2017-06-02 Richard Biener <rguenther@suse.de>
24904 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
24906 * tree-vect-stmts.c (process_use): Do not mark backedge defs
24907 for inductions as relevant.
24909 2017-06-02 Richard Biener <rguenther@suse.de>
24911 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
24912 (vectorizable_induction): ... this. Remove dead code.
24914 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
24916 * builtins. (expand_builtin_alloca): Remove second parameter and
24917 infer its value from the first parameter instead.
24918 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
24920 2017-06-02 Jakub Jelinek <jakub@redhat.com>
24922 PR rtl-optimization/80903
24923 * loop-doloop.c (add_test): Unshare sequence.
24925 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
24927 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
24929 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
24931 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
24933 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
24934 xlogue_layout::get_instance, logue_layout::xlogue_layout,
24935 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
24936 (xlogue_layout::get_stub_rtx): Make static.
24937 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
24938 (xlogue_layout::compute_stub_managed_regs): Rename to...
24939 (xlogue_layout::count_stub_managed_regs): ...this.
24940 (xlogue_layout::is_stub_managed_reg): New function.
24941 (xlogue_layout::m_stub_names): Rename to...
24942 (xlogue_layout::s_stub_names): ...this, make static.
24943 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
24944 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
24945 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
24946 xlogue_layout::s_stub_names): Instantiate statics.
24947 (stub_managed_regs): Remove.
24948 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
24949 (disable_call_ms2sysv_xlogues): Rename to...
24950 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
24951 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
24953 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
24954 change after reload_completed.
24955 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
24957 (ix86_expand_prologue): Likewise.
24958 (ix86_expand_epilogue): Likewise.
24959 (ix86_expand_split_stack_prologue): Likewise.
24960 (ix86_compute_frame_layout): Remove frame parameter ...
24961 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
24962 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
24964 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
24965 (ix86_frame): Move from here ...
24966 * config/i386/i386.h (ix86_frame): ... to here.
24967 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
24968 complete ix86_frame data structure instead. Remove some_ld_name.
24970 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
24972 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
24973 symbols that hold a DECL_VALUE_EXPR.
24975 2017-06-01 Martin Jambor <mjambor@suse.cz>
24977 PR tree-optimization/80898
24978 * tree-sra.c (process_subtree_disqualification): Removed.
24979 (disqualify_candidate): Do not acll
24980 process_subtree_disqualification.
24981 (subtree_mark_written_and_enqueue): New function.
24982 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
24983 RHS has been disqualified and re-queue LHS if necessary. Apart
24984 from that, ignore disqualified RHS.
24986 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24988 * config/s390/s390.c (s390_emit_epilogue): Disable early return
24989 address fetch for z10 or later.
24991 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24993 * config/arc/arc.md (tst_movb): Add guard when splitting.
24995 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24997 * config/arc/arc.c (arc_can_eliminate): Test against
24998 arc_frame_pointer_needed.
25000 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25002 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
25003 to prevent store reordering.
25004 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
25005 (type): Add block type.
25006 (stack_tie): Define special instruction to be used in
25009 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25011 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
25012 constraint. It is not valid for the pattern.
25013 (noncommutative_binary_comparison): Likewise.
25015 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25017 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
25020 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25022 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
25023 be used by the reg-alloc.
25025 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25027 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
25028 reg-alloc when having mul64 or mul32x16 instructions.
25029 (mulsidi3): Likewise.
25030 (umulsidi3): Likewise.
25031 (mulsi32x16): New pattern.
25032 (mulsi64): Likewise.
25033 (mulsidi64): Likewise.
25034 (umulsidi64): Likewise.
25035 (MUL32x16_REG): Define.
25036 (mul64_600): Use MUL32x16_REG.
25037 (mac64_600): Likewise.
25038 (umul64_600): Likewise.
25039 (umac64_600): Likewise.
25041 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25043 * config/arc/arc.md (mulsi3_700): Make it commutative.
25045 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
25047 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
25049 (*sign_extendsidi2_insn): Likewise for movstosw.
25051 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
25053 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
25054 the type of the input discriminant value. Convert the
25055 discriminant value of signedness vary.
25057 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
25059 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
25060 Add to -Wall section.
25062 2017-06-01 Richard Biener <rguenther@suse.de>
25064 PR middle-end/66313
25065 * fold-const.c (fold_plusminus_mult_expr): If the factored
25066 factor may be zero use a wrapping type for the inner operation.
25067 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
25068 and handle moved defs.
25069 (process_assignment): Properly guard the unary op case. Return a
25070 tri-state indicating that moving the stmt before the call may allow
25071 to continue. Pass through to_move.
25072 (find_tail_calls): Handle moving unrelated defs before
25075 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
25078 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
25079 splitter result in the canonical way.
25081 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
25083 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
25084 also for 32bit target. Update insn attributes.
25085 (zero-extendsidi2 splitter): Allow all registers for operand 1.
25087 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
25089 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
25090 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
25091 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
25092 (_mm_maskz_min_ss): New intrinsics.
25094 2017-05-31 Martin Liska <mliska@suse.cz>
25096 * tree-vect-loop.c (vect_create_epilog_for_reduction):
25097 Change comment style to one we normally use.
25098 (vectorizable_reduction): Likewise.
25099 (vectorizable_induction): Likewise.
25100 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
25101 (vectorizable_call): Likewise.
25102 (vectorizable_simd_clone_call): Likewise.
25103 (vectorizable_conversion): Likewise.
25104 (vectorizable_assignment): Likewise.
25105 (vectorizable_shift): Likewise.
25106 (vectorizable_operation): Likewise.
25107 (vectorizable_store): Likewise.
25108 (vectorizable_load): Likewise.
25109 * tree-vectorizer.h: Likewise.
25111 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
25113 * passes.c (emergency_dump_function): New.
25114 * tree-pass.h (emergency_dump_function): Declare.
25115 * plugin.c (plugins_internal_error_function): Remove.
25116 * plugin.h (plugins_internal_error_function): Remove declaration.
25117 * toplev.c (internal_error_function): New static function. Use it...
25118 (general_init): ...here.
25120 2017-05-31 Graham Markall <graham.markall@embecosm.com>
25122 * config/arc/arc.c (arc_print_operand): Handle constant operands.
25123 (arc_rtx_costs): Add costs for new patterns.
25124 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
25125 * config/arc/predicates.md: Add _1_2_3_operand predicate.
25127 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
25129 * tree-ssa-strlen.c (get_next_strinfo): New function.
25130 (get_stridx_plus_constant): Use it.
25131 (zero_length_string): Likewise.
25132 (adjust_related_strinfos): Likewise.
25133 (adjust_last_stmt): Likewise.
25135 2017-05-31 Richard Biener <rguenther@suse.de>
25138 * config/i386/i386.c (ix86_expand_builtin): Remove assert
25139 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
25141 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
25143 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
25144 loop_vinfo argument and use of dependence distance vectors.
25145 Check instead whether the two references differ only in their
25146 initial value and assume that they have the same alignment if the
25147 difference is a multiple of the vector alignment.
25148 (vect_analyze_data_refs_alignment): Update call accordingly.
25150 2017-05-31 Martin Liska <mliska@suse.cz>
25153 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
25155 2017-05-31 Bin Cheng <bin.cheng@arm.com>
25157 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
25158 (create_intersect_range_checks): Move from ...
25159 * tree-data-ref.c (create_intersect_range_checks_index)
25160 (create_intersect_range_checks): ... to here.
25161 (create_runtime_alias_checks): New function factored from ...
25162 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
25163 here. Call above function.
25164 * tree-data-ref.h (create_runtime_alias_checks): New function.
25166 2017-05-31 Bin Cheng <bin.cheng@arm.com>
25168 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
25169 segment length for dr_b and compute it in wide_int.
25171 2017-05-31 Richard Biener <rguenther@suse.de>
25173 PR tree-optimization/80906
25174 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
25175 and pass through iv_map.
25176 (copy_bb_and_scalar_dependences): Adjust.
25177 (translate_pending_phi_nodes): Likewise.
25178 (copy_loop_close_phi_args): Handle code-generating IVs instead
25181 2017-05-30 David Malcolm <dmalcolm@redhat.com>
25183 * diagnostic-color.c (color_dict): Add "type-diff".
25184 (parse_gcc_colors): Update comment.
25185 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
25186 -fdiagnostics-show-template-tree and -fno-elide-type.
25187 (GCC_COLORS): Add type-diff to example.
25189 (-fdiagnostics-show-template-tree): New.
25190 (-fno-elide-type): New.
25191 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
25192 the pp_format_decoder callback. Call any m_format_postprocessor's
25194 (pretty_printer::pretty_printer): Initialize
25195 m_format_postprocessor.
25196 (pretty_printer::~pretty_printer): Delete any
25197 m_format_postprocessor.
25198 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
25199 (class format_postprocessor): New class.
25200 (struct pretty_printer::format_decoder): Document the new parameters.
25201 (struct pretty_printer::m_format_postprocessor): New field.
25202 * tree-diagnostic.c (default_tree_printer): Update for new
25203 bool and const char ** params.
25204 * tree-diagnostic.h (default_tree_printer): Likewise.
25206 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
25208 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
25209 (lwa_operand): Delete rs6000_gen_cell_microcode test.
25210 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25211 rs6000_gen_cell_microcode code.
25212 (rs6000_final_prescan_insn): Delete.
25213 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
25214 "warn-cell-microcode" entries.
25215 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
25216 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
25217 throughout. Change cc_reg_not_micro_cr0_operand to
25218 cc_reg_not_cr0_operand throughout.
25219 (*extendhi<mode>2_noload): Delete.
25220 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
25221 (mwarn-cell-microcode): Delete.
25222 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
25223 -mgen-cell-microcode and -mwarn-cell-microcode.
25225 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
25228 * config/i386/constraints.md (Yd): New constraint.
25230 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
25231 and (?Yd, r) alternatives. Update insn attributes.
25232 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
25233 and (?*Yd, r) alternatives. Update insn attributes.
25234 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
25236 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
25238 * gimplify.c (gimplify_modify_expr): Don't create a
25239 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
25242 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
25244 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
25246 2017-05-30 Richard Biener <rguenther@suse.de>
25248 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
25249 and reduc_def fields.
25250 (STMT_VINFO_REDUC_TYPE): New define.
25251 (STMT_VINFO_REDUC_DEF): Likewise.
25252 (vect_force_simple_reduction): Adjust prototype.
25253 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
25254 (vect_is_simple_reduction): Remove check_reduction argument.
25255 (vect_force_simple_reduction): Adjust and set
25256 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25257 (vectorizable_reduction): Do not re-do reduction analysis
25258 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25259 * tree-parloops.c (gather_scalar_reductions): Adjust.
25261 2017-05-30 Richard Biener <rguenther@suse.de>
25263 PR middle-end/80901
25264 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
25267 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
25269 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
25270 Introduce unknown_misalignment parameter and remove vf.
25271 (vect_peeling_hash_get_lowest_cost):
25272 Pass unknown_misalignment parameter.
25273 (vect_enhance_data_refs_alignment):
25274 Fix unsupportable data ref treatment.
25276 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25278 * tree-vect-data-refs.c (vect_get_data_access_cost):
25279 Workaround for SLP handling.
25280 (vect_enhance_data_refs_alignment):
25281 Compute costs for doing no peeling at all, compare to the best
25282 peeling costs so far and avoid peeling if cheaper.
25284 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25286 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
25287 Return peeling info and set costs to zero for unlimited cost
25289 (vect_enhance_data_refs_alignment): Also inspect all datarefs
25290 with unknown misalignment. Compute and costs for unknown
25291 misalignment, compare them to the costs for known misalignment
25292 and choose the cheapest for peeling.
25294 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25296 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
25297 (vect_get_peeling_costs_all_drs): Create function.
25298 (vect_peeling_hash_get_lowest_cost):
25299 Use vect_get_peeling_costs_all_drs.
25300 (vect_peeling_supportable): Create function.
25301 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
25303 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25305 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
25306 DR_HAS_NEGATIVE_STEP.
25307 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
25308 (vect_enhance_data_refs_alignment): Use.
25309 (vect_duplicate_ssa_name_ptr_info): Use.
25310 * tree-vectorizer.h (dr_misalignment): Use.
25311 (known_alignment_for_access_p): Use.
25313 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25316 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
25318 (has_section_name): New function.
25320 2017-05-30 Martin Liska <mliska@suse.cz>
25323 * auto-profile.c (get_function_decl_from_block): Fix
25326 2017-05-30 Richard Biener <rguenther@suse.de>
25328 PR middle-end/80876
25329 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
25331 2017-05-30 Martin Liska <mliska@suse.cz>
25333 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
25334 * dumpfile.h (struct dump_file_info): Remove ctors.
25336 2017-05-30 Martin Liska <mliska@suse.cz>
25338 * predict.def: Fix GNU coding style.
25340 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25342 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25343 Mark 'to' argument with ATTRIBUTE_UNUSED.
25345 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25347 * config/xtensa/xtensa.c (xtensa_emit_call): Use
25348 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
25349 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
25352 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
25354 * doc/install.texi (Options specification): Restore entry of
25355 --enable-sjlj-exceptions.
25357 2017-05-27 Michael Eager <eager@eagercon.com>
25360 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
25362 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
25364 * config/microblaze/microblaze.h
25365 (FIXED_REGISTERS): Update in macro.
25366 (CALL_USED_REGISTERS): Update in macro.
25368 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
25370 * doc/install.texi: Add links to macOS binary distributions.
25372 2017-05-27 Jakub Jelinek <jakub@redhat.com>
25376 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25378 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25380 2017-05-26 Martin Liska <mliska@suse.cz>
25382 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
25384 2017-05-26 Martin Liska <mliska@suse.cz>
25386 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
25387 always leading ';; '.
25388 (dump_bb_info): Likewise.
25389 (brief_dump_cfg): Likewise.
25390 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
25391 * dumpfile.c: Remove usage of TDF_VERBOSE.
25392 * dumpfile.h (enum dump_kind): Likewise.
25393 (dump_gimple_bb_header): Do not use TDF_COMMENT.
25394 * print-tree.c (debug_verbose): Remove.
25395 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
25396 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
25397 * tree-diagnostic.c (default_tree_printer): Replace
25398 TDF_DIAGNOSTIC with TDF_SLIM.
25400 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25402 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
25403 in parameter loop, rather than loop_vinfo.
25404 (create_intersect_range_checks): Ditto.
25405 (vect_create_cond_for_alias_checks): Update call to above functions.
25407 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25409 PR tree-optimization/80815
25410 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
25411 for merging runtime alias checks. Handle negative DR_STEPs.
25413 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25415 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
25417 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
25418 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
25419 out code pruning runtime alias checks.
25420 * tree-data-ref.c (prune_runtime_alias_test_list): New function
25421 factored out from above.
25422 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25424 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25426 (prune_runtime_alias_test_list): New decalaration.
25428 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25430 * tree-vect-data-refs.c (compare_tree): Rename and move ...
25431 * tree-data-ref.c (data_ref_compare_tree): ... to here.
25432 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
25433 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
25434 (operator==, comp_dr_with_seg_len_pair): Ditto.
25435 (vect_prune_runtime_alias_test_list): Ditto.
25437 2017-05-26 Martin Liska <mliska@suse.cz>
25440 * params.def: Bound partial-inlining-entry-probability param.
25442 2017-05-26 Marek Polacek <polacek@redhat.com>
25445 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
25448 2017-05-26 Richard Biener <rguenther@suse.de>
25450 PR tree-optimization/80842
25451 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
25454 2017-05-26 Richard Biener <rguenther@suse.de>
25456 PR tree-optimization/80844
25457 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
25459 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
25461 * doc/md.texi (Machine Constraints): Update x86 family
25462 machine constraints section to match 'config/i386/constraints.md'.
25464 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
25466 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
25468 2017-05-25 Nathan Sidwell <nathan@acm.org>
25470 * doc/invoke.texi (--enable-languages): Update documentation.
25472 2017-05-25 Martin Liska <mliska@suse.cz>
25474 * dumpfile.c: Add TDF_FOLDING.
25475 * dumpfile.h (enum dump_kind): Likewise.
25476 * genmatch.c (dt_simplify::gen_1): Use it.
25478 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25480 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
25482 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25484 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25485 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
25487 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25489 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
25490 * match.pd (X == C): Rewrite it here.
25491 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
25492 with_certain_nonzero_bits2): New predicates.
25493 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
25495 2017-05-24 Nathan Sidwell <nathan@acm.org>
25497 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
25500 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
25503 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
25505 * config/powerpcspe: New port. Files are copied from the rs6000
25506 port, with "rs6000" in filenames replaced by "powerpcspe".
25508 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
25510 PR rtl-optimization/80754
25511 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
25513 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
25515 * config/sparc/sparc.md (length): Return the correct value for -mflat
25516 sibcalls to match output_sibcall.
25518 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
25522 * config/rs6000/rs6000.c (struct machine_function): Add new field
25524 (rs6000_get_separate_components): Init that field, use it.
25525 (rs6000_components_for_bb): Use the field.
25527 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
25529 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
25531 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
25533 PR middle-end/80823
25534 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
25536 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25539 * config/s390/s390.c (s390_check_qrst_address): Check incoming
25540 address against address_operand predicate.
25541 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
25543 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
25545 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
25546 parameters passed indirectly.
25548 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
25550 * config/i386/i386.md (*movdi_internal): Remove SSE4
25551 alternative 18 (?r, *v). Update insn attributes.
25552 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
25553 Update insn attributes.
25554 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
25555 Update insn attributes.
25556 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
25557 alternative 1 (r, v). Remove isa attribute.
25558 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
25559 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
25560 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
25562 2017-05-23 Tom de Vries <tom@codesourcery.com>
25564 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
25567 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25569 * cgraphunit.c (symbol_table::process_new_functions): Update.
25570 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
25571 (inline_generate_summary): Rename to ...
25572 (ipa_fn_summary_generate): ... this one.
25573 (inline_read_summary): Rename to ...
25574 (ipa_fn_summary_read): ... this one.
25575 (inline_write_summary): Rename to ...
25576 (ipa_fn_summary_write): ... this one.
25577 (inline_free_summary): Rename to ...
25578 (ipa_free_fn_summary): ... this one.
25579 (pass_data_local_fn_summary, pass_local_fn_summary,
25580 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
25581 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
25582 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
25583 make_pass_ipa_fn_summary): New.
25584 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
25585 inline_write_summary, inline_free_summary): Remove.
25586 (ipa_free_fn_summary) : New.
25587 * ipa-inline.c (ipa_inline): Update.
25588 (pass_ipa_inline): Do not generate summaries.
25589 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
25591 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
25592 and add pass_ipa_fn_summary.
25593 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
25595 (make_pass_inline_parameters): Remove.
25597 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
25599 * omp-low.c (struct omp_context): Remove "default_kind" member.
25602 * omp-offload.c (execute_oacc_device_lower): Remove the
25603 parallelism dimensions function attributes for unparallelized
25604 OpenACC kernels constructs.
25606 2017-05-23 Martin Liska <mliska@suse.cz>
25608 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
25610 (cgraph_edge::make_speculative): Likewise.
25611 (cgraph_edge::resolve_speculation): Likewise.
25612 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25613 (cgraph_node::dump): Likewise.
25614 * cgraph.h: Likewise.
25615 * cgraphunit.c (analyze_functions): Likewise.
25616 (symbol_table::compile): Likewise.
25617 * ipa-cp.c (print_all_lattices): Likewise.
25618 (determine_versionability): Likewise.
25619 (initialize_node_lattices): Likewise.
25620 (ipcp_verify_propagated_values): Likewise.
25621 (estimate_local_effects): Likewise.
25622 (update_profiling_info): Likewise.
25623 (create_specialized_node): Likewise.
25624 (perhaps_add_new_callers): Likewise.
25625 (decide_about_value): Likewise.
25626 (decide_whether_version_node): Likewise.
25627 (identify_dead_nodes): Likewise.
25628 (ipcp_store_bits_results): Likewise.
25629 * ipa-devirt.c (dump_targets): Likewise.
25630 (ipa_devirt): Likewise.
25631 * ipa-icf.c (sem_item::dump): Likewise.
25632 (sem_function::equals): Likewise.
25633 (sem_variable::equals): Likewise.
25634 (sem_item_optimizer::read_section): Likewise.
25635 (sem_item_optimizer::execute): Likewise.
25636 (congruence_class::dump): Likewise.
25637 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
25638 (dump_inline_summary): Likewise.
25639 (estimate_node_size_and_time): Likewise.
25640 (inline_analyze_function): Likewise.
25641 * ipa-inline-transform.c (inline_call): Likewise.
25642 * ipa-inline.c (report_inline_failed_reason): Likewise.
25643 (want_early_inline_function_p): Likewise.
25644 (edge_badness): Likewise.
25645 (update_edge_key): Likewise.
25646 (inline_small_functions): Likewise.
25647 * ipa-profile.c (ipa_profile): Likewise.
25648 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
25649 (ipa_make_edge_direct_to_target): Likewise.
25650 (remove_described_reference): Likewise.
25651 (ipa_impossible_devirt_target): Likewise.
25652 (propagate_controlled_uses): Likewise.
25653 (ipa_print_node_params): Likewise.
25654 (ipcp_transform_function): Likewise.
25655 * ipa-pure-const.c (pure_const_read_summary): Likewise.
25656 (propagate_pure_const): Likewise.
25657 * ipa-reference.c (generate_summary): Likewise.
25658 (read_write_all_from_decl): Likewise.
25659 (propagate): Likewise.
25660 (ipa_reference_read_optimization_summary): Likewise.
25661 * ipa-utils.c (ipa_merge_profiles): Likewise.
25662 * ipa.c (walk_polymorphic_call_targets): Likewise.
25663 (symbol_table::remove_unreachable_nodes): Likewise.
25664 (ipa_single_use): Likewise.
25665 * passes.c (execute_todo): Likewise.
25666 * predict.c (drop_profile): Likewise.
25667 * symtab.c (symtab_node::get_dump_name): New function.
25668 (symtab_node::dump_name): Likewise.
25669 (symtab_node::dump_asm_name): Likewise.
25670 (symtab_node::dump_references): Likewise.
25671 (symtab_node::dump_referring): Likewise.
25672 (symtab_node::dump_base): Likewise.
25673 (symtab_node::debug_symtab): Likewise.
25674 * tree-sra.c (convert_callers_for_node): Likewise.
25675 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
25676 * value-prof.c (init_node_map): Likewise.
25678 2017-05-23 Martin Liska <mliska@suse.cz>
25680 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
25681 and symtab_node::debug_symtab to symbol_table::debug.
25682 * cgraphunit.c (analyze_functions): Use the renamed function.
25683 (symbol_table::compile): Likewise.
25684 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
25685 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
25686 * passes.c (execute_todo): Likewise.
25687 * symtab.c (symbol_table::dump): New function.
25688 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
25690 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25692 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
25693 that nonconst implies exec.
25695 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25697 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
25698 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
25699 (inline_edge_summary_vec): Turn into ...
25700 (ipa_call_summaries): ... this one.
25701 (redirect_to_unreachable, edge_set_predicate,
25702 evaluate_properties_for_edge, inline_summary_alloc,
25703 reset_ipa_call_summary, reset_inline_summary,
25704 inline_summary_t::duplicate): Update.
25705 (inline_edge_duplication_hook): Turn to ...
25706 (ipa_call_summary_t::duplicate): ... this one.
25707 (inline_edge_removal_hook): Turn to ...
25708 (ipa_call_summary_t::remove): ... this one.
25709 (dump_inline_edge_summary): Turn to ...
25710 (dump_ipa_call_summary): ... this one.
25711 (estimate_function_body_sizes): Update.
25712 (inline_update_callee_summaries): Update.
25713 (remap_edge_change_prob): Update.
25714 (remap_edge_summaries): Update.
25715 (inline_merge_summary): Update.
25716 (do_estimate_edge_time): Update.
25717 (inline_generate_summary): Update.
25718 (inline_read_section): Update.
25719 (inline_read_summary): Update.
25720 (inline_free_summary): Update.
25721 * ipa-inline.c (can_inline_edge_p): Update.
25722 (compute_inlined_call_time): Update.
25723 (want_inline_small_function_p): Update.
25724 (edge_badness): Update.
25725 (early_inliner): Update.
25726 * ipa-inline.h (inline_edge_summary): Turn to ...
25727 (ipa_call_summary): ... this one.
25728 (ipa_call_summary_t): New class.
25729 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
25730 (ipa_call_summaries): New.
25731 (inline_edge_summary): Remove.
25732 (estimate_edge_growth): Update.
25733 * ipa-profile.c (ipa_propagate_frequency_1): Update.
25734 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25735 * ipa-split.c (execute_split_functions): Update.
25736 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
25738 2017-05-23 Tom de Vries <tom@codesourcery.com>
25740 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
25741 attributes): Document rdrand effective target.
25743 2017-05-23 Tom de Vries <tom@codesourcery.com>
25745 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
25746 attributes): Sort alphabetically.
25748 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
25750 * config/avr/genmultilib.awk: Use gsub instead of gensub.
25752 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
25755 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
25756 V2DF/V2DI splat into two separate patterns, one that handles
25757 registers, and the other that only handles memory. Drop support
25758 for splatting from a GPR on ISA 2.07 and then splitting the
25759 splat into direct move and splat.
25760 (vsx_splat_<mode>_reg): Likewise.
25761 (vsx_splat_<mode>_mem): Likewise.
25763 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
25765 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
25767 2017-05-22 Jakub Jelinek <jakub@redhat.com>
25769 PR middle-end/80809
25770 * omp-low.c (finish_taskreg_remap): New function.
25771 (finish_taskreg_scan): If unit size of ctx->record_type
25772 is non-constant, unshare the size expression and replace
25773 decls in it with possible outer var refs.
25775 PR middle-end/80809
25776 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
25777 GOVD_SHARED rather than GOVD_PRIVATE with it.
25778 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
25779 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
25781 PR middle-end/80853
25782 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
25783 as last argument to build_outer_var_ref for pointer bases of array
25784 section reductions.
25786 2017-05-19 Martin Sebor <msebor@redhat.com>
25788 * print-tree.c (print_node): Print DECL_READ_P flag.
25790 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25792 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
25793 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
25794 * cgraph.c: Likewise.
25795 * cgraphunit.c: Likewise.
25796 * gengtype.c: Likewise.
25797 * ipa-cp.c: Likewise.
25798 * ipa-devirt.c: Likewise.
25799 * ipa-icf.c: Likewise.
25800 * ipa-predicate.c: Likewise.
25801 * ipa-profile.c: Likewise.
25802 * ipa-prop.c: Likewise.
25803 * ipa-split.c: Likewise.
25805 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
25806 edge_predicate_pool, dump_inline_hints,
25807 inline_summary::account_size_time, redirect_to_unreachable,
25808 edge_set_predicate, set_hint_predicate,
25809 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
25810 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
25811 inline_summary_t::remove, remap_hint_predicate_after_duplication,
25812 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
25813 ipa_call_summary_t::remove, initialize_growth_caches,
25814 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
25815 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
25816 mark_modified, unmodified_parm_1, unmodified_parm,
25817 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
25818 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
25819 compute_bb_predicates, will_be_nonconstant_expr_predicate,
25820 will_be_nonconstant_predicate, record_modified_bb_info,
25821 get_minimal_bb, record_modified, param_change_prob,
25822 phi_result_unknown_predicate, predicate_for_phi_result,
25823 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
25824 estimate_function_body_sizes, compute_inline_parameters,
25825 compute_inline_parameters_for_curren, pass_data_inline_parameters,
25826 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
25827 inline_update_callee_summaries, remap_edge_change_prob,
25828 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
25829 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
25830 inline_analyze_function, inline_summary_t::insert,
25831 inline_generate_summary, read_ipa_call_summary, inline_read_section,
25832 inline_read_summary, write_ipa_call_summary, inline_write_summary,
25833 inline_free_summary): Move to ipa-fnsummary.h
25834 (predicate_t): Remove.
25835 * ipa-fnsummary.c: New file.
25836 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
25837 (enum inline_hints_vals, inline_hints, agg_position_info,
25838 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
25839 inline_summaries, ipa_call_summary, ipa_call_summary_t,
25840 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
25841 dump_inline_summary, dump_inline_hints, inline_generate_summary,
25842 inline_read_summary, inline_write_summary, inline_free_summary,
25843 inline_analyze_function, initialize_inline_failed,
25844 inline_merge_summary, inline_update_overall_summary,
25845 compute_inline_parameters): Move to ipa-fnsummary.h
25846 * ipa-fnsummary.h: New file.
25847 * ipa-inline-transform.h: Include ipa-inline.h.
25848 * ipa-inline.c: LIkewise.
25850 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25852 * ipa-inline.c (edge_badness): Use inlined_time instead of
25853 inline_summaries->get.
25855 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25857 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
25859 2017-05-22 Nathan Sidwell <nathan@acm.org>
25861 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
25862 (fdump-lang): Document 'raw' option.
25863 * dumpfile.h (TDI_tu): Delete.
25864 * dumpfile.c (dump_files): Remove translation-unit.
25865 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
25867 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
25869 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
25870 command option from $(AWK) call.
25871 * config/avr/genmultilib.awk: Simplify and rewrite so that it
25872 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
25873 [FORMAT]: Remove handling of variable.
25874 * config/avr/t-multilib: Regenerate.
25876 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25878 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
25880 (dump_inline_summary): Do not print self_time.
25881 (estimate_function_body_sizes): Do not set self_time.
25882 (compute_inline_parameters): Likewise.
25883 (inline_read_section, inline_write_summary): Do not stream self_time.
25884 * ipa-inline.h (inline_summary): Drop self_time.
25886 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25888 * ipa-inline-analysis.c (account_size_time): Rename to ...
25889 (inline_summary::account_size_time): ... this one.
25890 (reset_ipa_call_summary): Turn to ...
25891 (ipa_call_summary::reset): ... this one.
25892 (reset_inline_summary): Turn to ...
25893 (inline_summary::reset): ... this one.
25894 (inline_summary_t::remove): Update.
25895 (inline_summary_t::duplicate): Update.
25896 (ipa_call_summary_t::remove): Update.
25897 (dump_inline_summary): Update.
25898 (estimate_function_body_sizes): Update.
25899 (compute_inline_parameters): Update.
25900 (estimate_node_size_and_time): Update.
25901 (inline_merge_summary): Update.
25902 (inline_update_overall_summary): Update.
25903 (inline_read_section): Update.
25904 (inline_write_summary): Update.
25905 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
25906 add account_size_time and reset member functions.
25907 (ipa_call_summary): Add reset function.
25908 * ipa-predicate.h (predicate::operator &): Constify.
25910 2017-05-22 Richard Biener <rguenther@suse.de>
25912 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
25914 2017-05-19 Jason Merrill <jason@redhat.com>
25916 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
25918 2017-05-19 Marek Polacek <polacek@redhat.com>
25921 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
25922 TYPE_OVERFLOW_WRAPS checks.
25924 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
25926 * tree-core.h (enum omp_clause_default_kind): Add
25927 "OMP_CLAUSE_DEFAULT_PRESENT".
25928 * tree-pretty-print.c (dump_omp_clause): Handle it.
25929 * gimplify.c (enum gimplify_omp_var_data): Add
25930 "GOVD_MAP_FORCE_PRESENT".
25931 (gimplify_adjust_omp_clauses_1): Map it to
25932 "GOMP_MAP_FORCE_PRESENT".
25933 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
25935 * gimplify.c (oacc_default_clause): Clarify.
25937 2017-05-19 Nathan Sidwell <nathan@acm.org>
25939 LANG_HOOK_REGISTER_DUMPS
25940 * toplev.c (general_init): Call register dump lang hook.
25941 * doc/invoke.texi: Document -fdump-lang option family.
25942 * dumpfile.c (dump_files): Remove class dump here.
25943 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
25944 * dumpfile.h (tree_dump_index): Remove TDI_class.
25945 * langhooks-def.h (lhd_register_dumps): Declare.
25946 (LANG_HOOKS_REGISTER_DUMPS): Define.
25947 (LANG_HOOKS_INITIALIZER): Add it.
25948 * langhooks.c (lhd_register_dumps): Define.
25949 * langhooks.h (struct lang_hooks): Add register_dumps.
25951 2017-05-19 Nathan Sidwell <nathan@acm.org>
25953 * context.h (context::set_passes): New.
25954 * context.c (context::context): Do not create pass manager.
25955 * toplev.c (general_init): Create pass manager here.
25957 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
25959 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
25960 use this splitter if two add or or instructions would also work for
25961 the constant we want to generate.
25963 2017-05-19 Richard Biener <rguenther@suse.de>
25966 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
25967 predicate evaluation.
25969 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
25971 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
25973 * ipa-inline.c (want_inline_small_function_p): Do not cast to
25976 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
25978 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
25979 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
25980 (inline_edge_summary_vec): Turn into ...
25981 (ipa_call_summaries): ... this one.
25982 (redirect_to_unreachable, edge_set_predicate,
25983 evaluate_properties_for_edge, inline_summary_alloc,
25984 reset_ipa_call_summary, reset_inline_summary,
25985 inline_summary_t::duplicate): Update.
25986 (inline_edge_duplication_hook): Turn to ...
25987 (ipa_call_summary_t::duplicate): ... this one.
25988 (inline_edge_removal_hook): Turn to ...
25989 (ipa_call_summary_t::remove): ... this one.
25990 (dump_inline_edge_summary): Turn to ...
25991 (dump_ipa_call_summary): ... this one.
25992 (estimate_function_body_sizes): Update.
25993 (inline_update_callee_summaries): Update.
25994 (remap_edge_change_prob): Update.
25995 (remap_edge_summaries): Update.
25996 (inline_merge_summary): Update.
25997 (do_estimate_edge_time): Update.
25998 (inline_generate_summary): Update.
25999 (inline_read_section): Update.
26000 (inline_read_summary): Update.
26001 (inline_free_summary): Update.
26002 * ipa-inline.c (can_inline_edge_p): Update.
26003 (compute_inlined_call_time): Update.
26004 (want_inline_small_function_p): Update.
26005 (edge_badness): Update.
26006 (early_inliner): Update.
26007 * ipa-inline.h (inline_edge_summary): Turn to ...
26008 (ipa_call_summary): ... this one.
26009 (ipa_call_summary_t): New class.
26010 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26011 (ipa_call_summaries): New.
26012 (inline_edge_summary): Remove.
26013 (estimate_edge_growth): Update.
26014 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26015 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26016 * ipa-split.c (execute_split_functions): Update.
26017 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26019 2017-05-19 Richard Biener <rguenther@suse.de>
26021 PR middle-end/80764
26022 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
26024 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
26026 * config/rs6000/rs6000.c (struct machine_function): Add field
26027 fpr_is_wrapped_separately.
26028 (rs6000_get_separate_components): Use 64 components. Handle the
26029 new FPR components.
26030 (rs6000_components_for_bb): Handle the FPR components.
26031 (rs6000_emit_prologue_components): Handle the FPR components.
26032 (rs6000_emit_epilogue_components): Handle the FPR components.
26033 (rs6000_set_handled_components): Handle the FPR components.
26034 (rs6000_emit_prologue): Don't output prologue code for those FPRs
26035 that are already separately shrink-wrapped.
26036 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
26037 that are already separately shrink-wrapped.
26039 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26042 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
26045 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
26046 (define_peephole2 for Altivec d-form load): Add peepholes to catch
26047 cases where the register allocator uses a move and an offsettable
26048 memory operation to/from a FPR register on ISA 2.06/2.07.
26049 (define_peephole2 for Altivec d-form store): Likewise.
26051 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
26054 * config/i386/mmx.md (*mov<mode>_internal): Enable
26055 alternatives 11, 12, 13 and 14 also for 32bit targets.
26056 Remove alternatives 15, 16, 17 and 18.
26057 * config/i386/sse.md (vec_concatv2di): Change
26058 alternative (!x, *y) to (x, ?!*Yn).
26060 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
26062 * dumpfile.h (enum dump_kind): Remove stray comma.
26064 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
26066 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
26067 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
26068 predicate::num_conditions
26069 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
26070 (CHANGED): turn into predicate::changed.
26071 (agg_position_info): Move to ipa-predicate.h
26072 (add_condition, predicate::add_clause, predicate::operator &=,
26073 predicate::or_with, predicate::evaluate, predicate::probability,
26074 dump_condition, dump_clause, predicate::dump,
26075 predicate::remap_after_duplication, predicate::remap_after_inlining,
26076 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
26077 (evaluate_conditions_for_known_args): Update.
26078 (set_cond_stmt_execution_predicate): Update.
26079 * ipa-inline.h: Include ipa-predicate.h
26080 (condition, inline_param_summary, conditions, agg_position_info,
26081 predicate): Move to ipa-predicate.h
26082 * ipa-predicate.c: New file.
26083 * ipa-predicate.h: New file.
26085 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
26087 * final.c (leaf_function_p): Check we are not in a sequence.
26089 2017-05-18 Martin Liska <mliska@suse.cz>
26091 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
26092 * dumpfile.c (dump_register): Use new enum dump_kind.
26093 (get_dump_file_name): Likewise.
26094 (dump_enable_all): Likewise.
26095 (dump_switch_p_1): Likewise.
26096 (enable_rtl_dump_file): Remove usage of TDF_RTL.
26097 * dumpfile.h (enum dump_kind): New enum type.
26098 (struct dump_file_info): Create constructor and
26099 format fields and comments.
26100 * passes.c (pass_manager::register_one_dump_file):
26102 * statistics.c (statistics_early_init): Likewise.
26103 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
26104 TDF_TREE with TDF_SLIM.
26105 (gather_memory_references_ref): Likewise.
26107 2017-05-18 Martin Liska <mliska@suse.cz>
26109 * vec.h (struct vnull): Use it.
26111 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
26113 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
26114 (true_predicate, false_predicate, true_predicate_p,
26115 false_predicate_p): Remove.
26116 (single_cond_predicate, not_inlined_predicate): Turn to member function
26118 (add_condition): Update.
26119 (add_clause): Turn to...
26120 (predicate::add_clause): ... this one; update; allow passing NULL
26122 (and_predicates): Turn to ...
26123 (predicate::operator &=): ... this one.
26124 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
26125 (or_predicates): Turn to ...
26126 (predicate::or_with): ... this one.
26127 (evaluate_predicate): Turn to ...
26128 (predicate::evaluate): ... this one.
26129 (predicate_probability): Turn to ...
26130 (predicate::probability): ... this one.
26131 (dump_condition): Update.
26132 (dump_predicate): Turn to ...
26133 (predicate::dump): ... this one.
26134 (account_size_time): Update.
26135 (edge_set_predicate): Update.
26136 (set_hint_predicate): UPdate.
26137 (evaluate_conditions_for_known_args): Update.
26138 (evaluate_properties_for_edge): Update.
26139 (remap_predicate_after_duplication): Turn to...
26140 (predicate::remap_after_duplication): ... this one.
26141 (remap_hint_predicate_after_duplication): Update.
26142 (inline_summary_t::duplicate): UPdate.
26143 (dump_inline_edge_summary): Update.
26144 (dump_inline_summary): Update.
26145 (set_cond_stmt_execution_predicate): Update.
26146 (set_switch_stmt_execution_predicate): Update.
26147 (compute_bb_predicates): Update.
26148 (will_be_nonconstant_expr_predicate): Update.
26149 (will_be_nonconstant_predicate): Update.
26150 (phi_result_unknown_predicate): Update.
26151 (predicate_for_phi_result): Update.
26152 (array_index_predicate): Update.
26153 (estimate_function_body_sizes): Update.
26154 (estimate_node_size_and_time): Update.
26155 (estimate_ipcp_clone_size_and_time): Update.
26156 (remap_predicate): Rename to ...
26157 (predicate::remap_after_inlining): ... this one.
26158 (remap_hint_predicate): Update.
26159 (inline_merge_summary): Update.
26160 (inline_update_overall_summary): Update.
26161 (estimate_size_after_inlining): Update.
26162 (read_predicate): Rename to ...
26163 (predicate::stream_in): ... this one.
26164 (read_inline_edge_summary): Update.
26165 (write_predicate): Rename to ...
26166 (predicate::stream_out): ... this one.
26167 (write_inline_edge_summary): Update.
26168 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
26169 (clause_t): Turn to uint32_t
26170 (predicate): Turn to class; implement constructor and operators
26172 (size_time_entry): Update.
26173 (inline_summary): Update.
26174 (inline_edge_summary): Update.
26176 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
26178 * fold-const.c (fold_binary_loc): Move transformation...
26179 * match.pd (C - X CMP X): ... here.
26181 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
26183 * config/sparc/sparc.c (sparc_option_override): Set function
26184 alignment for -mcpu=niagara7 to 64 to match the I$ line.
26185 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
26187 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
26189 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
26191 2017-05-18 Marek Polacek <polacek@redhat.com>
26194 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
26195 (pass_ubsan::execute): Call gimple_assign_single_p instead of
26196 gimple_assign_load_p.
26198 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
26200 PR middle-end/80692
26201 * real.c (do_compare): Give decimal_do_compare preference over
26202 comparing just the signs.
26204 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
26206 * doc/md.texi (Canonicalization of Instructions): Describe the
26207 canonical form of instructions that inherently set a condition
26210 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
26212 PR middle-end/80775
26213 * tree-cfg.c: Move deletion of unreachable case statements to after
26214 the merging of consecutive case labels.
26216 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
26218 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
26219 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
26220 restoring of callee-saved registers.
26222 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
26224 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
26225 * config/visium/visium.c (single_set_and_flags): Likewise.
26226 * config/visium/visium.md (Substitutions): Likewise.
26228 2017-05-17 Martin Liska <mliska@suse.cz>
26230 * cfg.c: Introduce dump_flags_t type and
26231 use it instead of int type.
26233 * cfghooks.c: Likewise.
26234 * cfghooks.h (struct cfg_hooks): Likewise.
26235 * cfgrtl.c: Likewise.
26236 * cfgrtl.h: Likewise.
26237 * cgraph.c (cgraph_node::get_body): Likewise.
26238 * coretypes.h: Likewise.
26239 * domwalk.c: Likewise.
26240 * domwalk.h: Likewise.
26241 * dumpfile.c (struct dump_option_value_info): Likewise.
26242 (dump_enable_all): Likewise.
26243 (dump_switch_p_1): Likewise.
26244 (opt_info_switch_p): Likewise.
26245 * dumpfile.h (enum tree_dump_index): Likewise.
26246 (struct dump_file_info): Likewise.
26247 * genemit.c: Likewise.
26248 * generic-match-head.c: Likewise.
26249 * gengtype.c (open_base_files): Likewise.
26250 * gimple-pretty-print.c: Likewise.
26251 * gimple-pretty-print.h: Likewise.
26252 * graph.c (print_graph_cfg): Likewise.
26253 * graphite-scop-detection.c (dot_all_sese): Likewise.
26254 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
26255 * loop-unroll.c (report_unroll): Likewise.
26256 * passes.c (pass_manager::register_one_dump_file): Likewise.
26257 * print-tree.c: Likewise.
26258 * statistics.c: Likewise.
26259 * tree-cfg.c: Likewise.
26260 * tree-cfg.h: Likewise.
26261 * tree-dfa.c: Likewise.
26262 * tree-dfa.h: Likewise.
26263 * tree-dump.c (dump_function): Likewise.
26264 * tree-dump.h (struct dump_info): Likewise.
26265 * tree-pretty-print.c: Likewise.
26266 * tree-pretty-print.h: Likewise.
26267 * tree-ssa-live.c: Likewise.
26268 * tree-ssa-live.h: Likewise.
26269 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
26270 * tree-vect-loop.c: Likewise.
26271 * tree-vect-slp.c: Likewise.
26273 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
26274 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26276 PR tree-optimization/80457
26277 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
26278 of all arguments to a statement as scalar_to_vec operations.
26279 (vectorizable_call): Adjust call to vect_model_simple_cost for
26281 (vectorizable_conversion): Likewise.
26282 (vectorizable_assignment): Likewise.
26283 (vectorizable_shift): Likewise.
26284 (vectorizable_operation): Likewise.
26285 (vectorizable_comparison): Likewise.
26286 (vect_is_simple_cond): Record the def types for operands.
26287 (vectorizable_condition): Likewise, call vect_model_simple_cost.
26288 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
26289 for statement argument count.
26291 2017-05-16 Carl Love <cel@us.ibm.com>
26293 * config/rs6000/rs6000-c: Add support for built-in functions
26294 vector unsigned long long vec_bperm (vector unsigned long long,
26295 vector unsigned char)
26296 vector signed long long vec_mule (vector signed int,
26298 vector unsigned long long vec_mule (vector unsigned int,
26299 vector unsigned int)
26300 vector signed long long vec_mulo (vector signed int,
26302 vector unsigned long long vec_mulo (vector unsigned int,
26303 vector unsigned int)
26304 vector signed char vec_sldw (vector signed char,
26305 vector signed char,
26307 vector unsigned char vec_sldw (vector unsigned char,
26308 vector unsigned char,
26310 vector signed short vec_sldw (vector signed short,
26311 vector signed short,
26313 vector unsigned short vec_sldw (vector unsigned short,
26314 vector unsigned short,
26316 vector signed int vec_sldw (vector signed int,
26319 vector unsigned int vec_sldw (vector unsigned int,
26320 vector unsigned int,
26322 vector signed long long vec_sldw (vector signed long long,
26323 vector signed long long,
26325 vector unsigned long long vec_sldw (vector unsigned long long,
26326 vector unsigned long long,
26328 * config/rs6000/rs6000-c: Add support for built-in functions
26329 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
26330 * config/rs6000/altivec.h: Add defintion for vec_sldw.
26331 * doc/extend.texi: Update the built-in documentation for the
26332 new built-in functions.
26334 2017-05-16 Marek Polacek <polacek@redhat.com>
26338 * tree.c (save_expr): Don't fold the expression.
26340 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
26342 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
26343 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
26344 and (?*y,m). Update insn attributes.
26346 2017-05-16 Martin Liska <mliska@suse.cz>
26348 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
26349 flags argument of print_gimple_stmt, print_gimple_expr,
26350 print_generic_stmt and print_generic_expr.
26351 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
26352 * coretypes.h: Likewise.
26353 * except.c (dump_eh_tree): Likewise.
26354 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
26355 * gimple-pretty-print.h: Likewise.
26356 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
26357 (backprop::push_to_worklist): Likewise.
26358 (backprop::pop_from_worklist): Likewise.
26359 (backprop::process_use): Likewise.
26360 (backprop::intersect_uses): Likewise.
26361 (note_replacement): Likewise.
26362 * gimple-ssa-store-merging.c
26363 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
26364 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
26365 (pass_store_merging::execute): Likewise.
26366 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
26367 (ssa_base_cand_dump_callback): Likewise.
26368 (dump_incr_vec): Likewise.
26369 (replace_refs): Likewise.
26370 (replace_mult_candidate): Likewise.
26371 (create_add_on_incoming_edge): Likewise.
26372 (create_phi_basis): Likewise.
26373 (insert_initializers): Likewise.
26374 (all_phi_incrs_profitable): Likewise.
26375 (introduce_cast_before_cand): Likewise.
26376 (replace_one_candidate): Likewise.
26377 * gimplify.c (gimplify_expr): Likewise.
26378 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
26379 (set_rename): Likewise.
26380 (rename_uses): Likewise.
26381 (copy_loop_phi_nodes): Likewise.
26382 (add_close_phis_to_merge_points): Likewise.
26383 (copy_loop_close_phi_args): Likewise.
26384 (copy_cond_phi_args): Likewise.
26385 (graphite_copy_stmts_from_block): Likewise.
26386 (translate_pending_phi_nodes): Likewise.
26387 * graphite-poly.c (print_pdr): Likewise.
26388 (dump_gbb_cases): Likewise.
26389 (dump_gbb_conditions): Likewise.
26390 (print_scop_params): Likewise.
26391 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
26392 (build_cross_bb_scalars_use): Likewise.
26393 (gather_bbs::before_dom_children): Likewise.
26394 * hsa-dump.c (dump_hsa_immed): Likewise.
26395 * ipa-cp.c (print_ipcp_constant_value): Likewise.
26396 (get_replacement_map): Likewise.
26397 * ipa-inline-analysis.c (dump_condition): Likewise.
26398 (estimate_function_body_sizes): Likewise.
26399 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
26400 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
26401 * ipa-prop.c (ipa_dump_param): Likewise.
26402 (ipa_print_node_jump_functions_for_edge): Likewise.
26403 (ipa_modify_call_arguments): Likewise.
26404 (ipa_modify_expr): Likewise.
26405 (ipa_dump_param_adjustments): Likewise.
26406 (ipa_dump_agg_replacement_values): Likewise.
26407 (ipcp_modif_dom_walker::before_dom_children): Likewise.
26408 * ipa-pure-const.c (check_stmt): Likewise.
26409 (pass_nothrow::execute): Likewise.
26410 * ipa-split.c (execute_split_functions): Likewise.
26411 * omp-offload.c (dump_oacc_loop_part): Likewise.
26412 (dump_oacc_loop): Likewise.
26413 * trans-mem.c (tm_log_emit): Likewise.
26414 (tm_memopt_accumulate_memops): Likewise.
26415 (dump_tm_memopt_set): Likewise.
26416 (dump_tm_memopt_transform): Likewise.
26417 * tree-cfg.c (gimple_verify_flow_info): Likewise.
26418 (print_loop): Likewise.
26419 * tree-chkp-opt.c (chkp_print_addr): Likewise.
26420 (chkp_gather_checks_info): Likewise.
26421 (chkp_get_check_result): Likewise.
26422 (chkp_remove_check_if_pass): Likewise.
26423 (chkp_use_outer_bounds_if_possible): Likewise.
26424 (chkp_reduce_bounds_lifetime): Likewise.
26425 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
26426 (chkp_mark_completed_bounds): Likewise.
26427 (chkp_register_incomplete_bounds): Likewise.
26428 (chkp_mark_invalid_bounds): Likewise.
26429 (chkp_maybe_copy_and_register_bounds): Likewise.
26430 (chkp_build_returned_bound): Likewise.
26431 (chkp_get_bound_for_parm): Likewise.
26432 (chkp_build_bndldx): Likewise.
26433 (chkp_get_bounds_by_definition): Likewise.
26434 (chkp_generate_extern_var_bounds): Likewise.
26435 (chkp_get_bounds_for_decl_addr): Likewise.
26436 * tree-chrec.c (chrec_apply): Likewise.
26437 * tree-data-ref.c (dump_data_reference): Likewise.
26438 (dump_subscript): Likewise.
26439 (dump_data_dependence_relation): Likewise.
26440 (analyze_overlapping_iterations): Likewise.
26441 * tree-inline.c (expand_call_inline): Likewise.
26442 (tree_function_versioning): Likewise.
26443 * tree-into-ssa.c (dump_defs_stack): Likewise.
26444 (dump_currdefs): Likewise.
26445 (dump_names_replaced_by): Likewise.
26446 (dump_update_ssa): Likewise.
26447 (update_ssa): Likewise.
26448 * tree-object-size.c (pass_object_sizes::execute): Likewise.
26449 * tree-parloops.c (build_new_reduction): Likewise.
26450 (try_create_reduction_list): Likewise.
26451 (ref_conflicts_with_region): Likewise.
26452 (oacc_entry_exit_ok_1): Likewise.
26453 (oacc_entry_exit_single_gang): Likewise.
26454 * tree-pretty-print.h: Likewise.
26455 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
26456 (get_scalar_evolution): Likewise.
26457 (add_to_evolution): Likewise.
26458 (get_loop_exit_condition): Likewise.
26459 (analyze_evolution_in_loop): Likewise.
26460 (analyze_initial_condition): Likewise.
26461 (analyze_scalar_evolution): Likewise.
26462 (instantiate_scev): Likewise.
26463 (number_of_latch_executions): Likewise.
26464 (gather_chrec_stats): Likewise.
26465 (final_value_replacement_loop): Likewise.
26466 (scev_const_prop): Likewise.
26467 * tree-sra.c (dump_access): Likewise.
26468 (disqualify_candidate): Likewise.
26469 (create_access): Likewise.
26470 (reject): Likewise.
26471 (maybe_add_sra_candidate): Likewise.
26472 (create_access_replacement): Likewise.
26473 (analyze_access_subtree): Likewise.
26474 (analyze_all_variable_accesses): Likewise.
26475 (sra_modify_assign): Likewise.
26476 (initialize_constant_pool_replacements): Likewise.
26477 (find_param_candidates): Likewise.
26478 (decide_one_param_reduction): Likewise.
26479 (replace_removed_params_ssa_names): Likewise.
26480 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
26481 * tree-ssa-copy.c (dump_copy_of): Likewise.
26482 (copy_prop_visit_cond_stmt): Likewise.
26483 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
26484 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
26485 (record_equivalences_from_stmt): Likewise.
26486 * tree-ssa-dse.c (compute_trims): Likewise.
26487 (delete_dead_call): Likewise.
26488 (delete_dead_assignment): Likewise.
26489 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
26490 (forward_propagate_into_cond): Likewise.
26491 (pass_forwprop::execute): Likewise.
26492 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
26493 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
26495 (move_computations_worker): Likewise.
26496 (execute_sm): Likewise.
26497 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
26498 (remove_exits_and_undefined_stmts): Likewise.
26499 (remove_redundant_iv_tests): Likewise.
26500 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
26501 (adjust_iv_update_pos): Likewise.
26502 * tree-ssa-math-opts.c (bswap_replace): Likewise.
26503 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
26504 (value_replacement): Likewise.
26505 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
26506 * tree-ssa-pre.c (print_pre_expr): Likewise.
26507 (get_representative_for): Likewise.
26508 (create_expression_by_pieces): Likewise.
26509 (insert_into_preds_of_block): Likewise.
26510 (eliminate_insert): Likewise.
26511 (eliminate_dom_walker::before_dom_children): Likewise.
26512 (eliminate): Likewise.
26513 (remove_dead_inserted_code): Likewise.
26514 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
26515 * tree-ssa-reassoc.c (get_rank): Likewise.
26516 (eliminate_duplicate_pair): Likewise.
26517 (eliminate_plus_minus_pair): Likewise.
26518 (eliminate_not_pairs): Likewise.
26519 (undistribute_ops_list): Likewise.
26520 (eliminate_redundant_comparison): Likewise.
26521 (update_range_test): Likewise.
26522 (optimize_range_tests_var_bound): Likewise.
26523 (optimize_vec_cond_expr): Likewise.
26524 (rewrite_expr_tree): Likewise.
26525 (rewrite_expr_tree_parallel): Likewise.
26526 (linearize_expr): Likewise.
26527 (break_up_subtract): Likewise.
26528 (linearize_expr_tree): Likewise.
26529 (attempt_builtin_powi): Likewise.
26530 (attempt_builtin_copysign): Likewise.
26531 (transform_stmt_to_copy): Likewise.
26532 (transform_stmt_to_multiply): Likewise.
26533 (dump_ops_vector): Likewise.
26534 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
26535 (print_scc): Likewise.
26536 (set_ssa_val_to): Likewise.
26537 (visit_reference_op_store): Likewise.
26538 (visit_use): Likewise.
26539 (sccvn_dom_walker::before_dom_children): Likewise.
26540 (run_scc_vn): Likewise.
26541 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
26543 (expr_hash_elt::print): Likewise.
26544 (const_and_copies::pop_to_marker): Likewise.
26545 (const_and_copies::record_const_or_copy_raw): Likewise.
26546 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
26547 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
26548 (dump_predicates): Likewise.
26549 (find_uninit_use): Likewise.
26550 (warn_uninitialized_phi): Likewise.
26551 (pass_late_warn_uninitialized::execute): Likewise.
26552 * tree-ssa.c (verify_vssa): Likewise.
26553 (verify_ssa): Likewise.
26554 (maybe_optimize_var): Likewise.
26555 * tree-vrp.c (dump_value_range): Likewise.
26556 (dump_all_value_ranges): Likewise.
26557 (dump_asserts_for): Likewise.
26558 (register_edge_assert_for_2): Likewise.
26559 (vrp_visit_cond_stmt): Likewise.
26560 (vrp_visit_switch_stmt): Likewise.
26561 (vrp_visit_stmt): Likewise.
26562 (vrp_visit_phi_node): Likewise.
26563 (simplify_cond_using_ranges_1): Likewise.
26564 (fold_predicate_in): Likewise.
26565 (evrp_dom_walker::before_dom_children): Likewise.
26566 (evrp_dom_walker::push_value_range): Likewise.
26567 (evrp_dom_walker::pop_value_range): Likewise.
26568 (execute_early_vrp): Likewise.
26570 2017-05-16 Richard Biener <rguenther@suse.de>
26572 * dwarf2out.c (loc_list_from_tree_1): Do not create
26573 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
26575 2017-05-16 Richard Biener <rguenther@suse.de>
26577 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
26579 (note_variable_value_in_expr): If we resolved the decl ref
26580 do not push to the stack.
26582 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
26584 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
26585 operations in fast-math mode.
26586 (vaddq_f16): Likewise.
26587 (vmul_f16): Likewise.
26588 (vmulq_f16): Likewise.
26589 (vsub_f16): Likewise.
26590 (vsubq_f16): Likewise.
26591 * config/arm/neon.md (add<mode>3): New.
26593 (fma:<VH:mode>3): New. Also remove outdated comment.
26596 2017-05-16 Martin Liska <mliska@suse.cz>
26600 * ipa-devirt.c (warn_types_mismatch): Fix typo.
26601 (odr_types_equivalent_p): Likewise.
26603 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
26605 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
26607 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
26610 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
26611 non-interunit SSE move alternatives with '?'.
26612 (zero-extendsidi peephole2): New peephole to skip intermediate
26613 general register in SSE zero-extend sequence.
26615 2017-05-15 Jeff Law <law@redhat.com>
26617 * reorg.c (relax_delay_slots): Create a new variable to hold
26618 the temporary target rather than clobbering TARGET_LABEL.
26620 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
26621 missing argument to extract_bit_field call.
26622 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
26624 2017-05-15 Martin Liska <mliska@suse.cz>
26627 * gcc.c (process_command): Do not allow empty argument of -o option.
26629 2017-05-15 Renlin Li <renlin.li@arm.com>
26631 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
26632 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
26633 * config/aarch64/constraints.md (Usf): Add long call check.
26634 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
26635 (call_value): Likewise.
26636 (sibcall): Likewise.
26637 (sibcall_value): Likewise.
26639 (call_value_insn): New.
26640 (sibcall_insn): Update rtx pattern.
26641 (sibcall_value_insn): Likewise.
26642 (call_internal): Remove.
26643 (call_value_internal): Likewise.
26644 (sibcall_internal): Likewise.
26645 (sibcall_value_internal): Likewise.
26646 (call_reg): Likewise.
26647 (call_symbol): Likewise.
26648 (call_value_reg): Likewise.
26649 (call_value_symbol): Likewise.
26651 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
26654 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
26656 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
26658 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
26659 compatible with CCGOCmode and with CCZmode.
26661 2017-05-14 Martin Sebor <msebor@redhat.com>
26663 PR middle-end/77671
26664 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
26665 (gimple_fold_builtin_snprintf): Same.
26666 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
26667 (gimple_fold_builtin_snprintf): Same.
26668 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
26669 of character types.
26670 (is_call_safe): New function.
26671 (try_substitute_return_value): Call it.
26672 (try_simplify_call): New function.
26673 (pass_sprintf_length::handle_gimple_call): Call it.
26675 2017-05-14 Martin Sebor <msebor@redhat.com>
26677 PR middle-end/80669
26678 * builtins.c (expand_builtin_stpncpy): Simplify.
26680 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
26682 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
26683 * config/i386/i386.h
26684 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
26685 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
26686 (struct machine_function): Add new members call_ms2sysv,
26687 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
26688 (struct machine_frame_state): New fields sp_realigned and
26689 sp_realigned_offset.
26690 * config/i386/i386.c
26691 (enum xlogue_stub): New enum.
26692 (enum xlogue_stub_sets): New enum.
26693 (class xlogue_layout): New class.
26694 (struct ix86_frame): New fields stack_realign_allocate_offset,
26695 stack_realign_offset and outlined_save_offset. Modify comments to
26696 detail stack layout when using out-of-line stubs.
26697 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
26698 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
26699 -mcall-ms2sysv-xlogues.
26700 (stub_managed_regs): New static variable.
26701 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
26702 registers managed by out-of-line stub.
26703 (disable_call_ms2sysv_xlogues): New function.
26704 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
26705 m->call_ms2sysv when appropriate and compute frame layout for
26707 (sp_valid_at, fp_valid_at): New inline functions.
26708 (choose_basereg): New function.
26709 (choose_baseaddr): Add align parameter, use choose_basereg and modify
26711 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
26712 Use align parameter of choose_baseaddr to generated aligned SSE movs
26714 (pro_epilogue_adjust_stack): Modify to track
26715 machine_frame_state::sp_realigned.
26716 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
26717 (ix86_nsaved_sseregs): Likewise.
26718 (ix86_emit_save_regs): Likewise.
26719 (ix86_emit_save_regs_using_mov): Likewise.
26720 (ix86_emit_save_sse_regs_using_mov): Likewise.
26721 (get_scratch_register_on_entry): Likewise.
26722 (gen_frame_set): New function.
26723 (gen_frame_load): Likewise.
26724 (gen_frame_store): Likewise.
26725 (emit_outlined_ms2sysv_save): Likewise.
26726 (emit_outlined_ms2sysv_restore): Likewise.
26727 (ix86_expand_prologue): Modify stack re-alignment code and call
26728 emit_outlined_ms2sysv_save when appropriate.
26729 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
26730 parameter rtx_insn *insn, which allows the function to be used to only
26731 generate the notes.
26732 (ix86_expand_epilogue): Modify validity checks of frame and stack
26733 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
26734 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
26735 * config/i386/predicates.md
26736 (save_multiple): New predicate.
26737 (restore_multiple): Likewise.
26738 * config/i386/sse.md
26739 (save_multiple<mode>): New pattern.
26740 (save_multiple_realign<mode>): Likewise.
26741 (restore_multiple<mode>): Likewise.
26742 (restore_multiple_and_return<mode>): Likewise.
26743 (restore_multiple_leave_return<mode>): Likewise.
26744 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
26746 2017-05-14 Julia Koval <julia.koval@intel.com>
26748 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
26749 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
26750 (__builtin_ia32_xsetbv): New builtins.
26751 * config/i386/i386.c (ix86_expand_special_args_builtin):
26753 (ix86_expand_builtin): Special expand for new intrinsics.
26754 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
26755 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
26756 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
26758 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26760 * cfganal.c (inverted_post_order_compute): Change argument type
26762 * cfganal.h (inverted_post_order_compute): Adjust prototype.
26763 * df-core.c (rest_of_handle_df_initialize): Adjust.
26764 (rest_of_handle_df_finish): Likewise.
26765 (df_analyze_1): Likewise.
26766 (df_analyze): Likewise.
26767 (loop_inverted_post_order_compute): Change argument to be a vec *.
26768 (df_analyze_loop): Adjust.
26769 (df_get_n_blocks): Likewise.
26770 (df_get_postorder): Likewise.
26771 * df.h (struct df_d): Change field to be a vec.
26772 * lcm.c (compute_laterin): Adjust.
26773 (compute_available): Likewise.
26774 * lra-lives.c (lra_create_live_ranges_1): Likewise.
26775 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
26776 * tree-ssa-pre.c (compute_antic): Likewise.
26778 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26780 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
26781 (depth_first_search::depth_first_search): Change structure init
26782 function to this constructor.
26783 (depth_first_search::add_bb): Rename function to this member.
26784 (depth_first_search::execute): Likewise.
26785 (flow_dfs_compute_reverse_finish): Adjust.
26787 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26789 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
26790 (longest_simple_path): Likewise.
26791 * shrink-wrap.c (spread_components): Likewise.
26792 (disqualify_problematic_components): Likewise.
26793 (emit_common_heads_for_components): Likewise.
26794 (emit_common_tails_for_components): Likewise.
26795 (insert_prologue_epilogue_for_components): Likewise.
26797 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26799 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
26802 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26804 * df-core.c (df_set_blocks): Start using auto_bitmap.
26805 (df_compact_blocks): Likewise.
26806 * df-problems.c (df_rd_confluence_n): Likewise.
26807 * df-scan.c (df_insn_rescan_all): Likewise.
26808 (df_process_deferred_rescans): Likewise.
26809 (df_update_entry_block_defs): Likewise.
26810 (df_update_exit_block_uses): Likewise.
26811 (df_entry_block_bitmap_verify): Likewise.
26812 (df_exit_block_bitmap_verify): Likewise.
26813 (df_scan_verify): Likewise.
26814 * lra-constraints.c (lra_constraints): Likewise.
26815 (undo_optional_reloads): Likewise.
26816 (lra_undo_inheritance): Likewise.
26817 * lra-remat.c (calculate_gen_cands): Likewise.
26818 (do_remat): Likewise.
26819 * lra-spills.c (assign_spill_hard_regs): Likewise.
26820 (spill_pseudos): Likewise.
26821 * tree-ssa-pre.c (bitmap_set_and): Likewise.
26822 (bitmap_set_subtract_values): Likewise.
26824 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26826 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
26827 management with auto_bitmap.
26828 (fix_inter_tick): Likewise.
26829 (fix_recovery_deps): Likewise.
26830 * ira.c (add_store_equivs): Likewise.
26831 (find_moveable_pseudos): Likewise.
26832 (split_live_ranges_for_shrink_wrap): Likewise.
26833 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
26834 (rtx_reuse_manager::seen_def_p): Likewise.
26835 (rtx_reuse_manager::set_seen_def): Likewise.
26836 * print-rtl.h (class rtx_reuse_manager): Likewise.
26838 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26840 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
26842 (migrate_btr_def): Likewise.
26843 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
26844 * df-core.c (loop_post_order_compute): Likewise.
26845 (loop_inverted_post_order_compute): Likewise.
26846 * hsa-common.h: Likewise.
26847 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
26848 * init-regs.c (initialize_uninitialized_regs): Likewise.
26849 * ipa-inline.c (resolve_noninline_speculation): Likewise.
26850 (inline_small_functions): Likewise.
26851 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
26852 * ira.c (combine_and_move_insns): Likewise.
26853 (build_insn_chain): Likewise.
26854 * loop-invariant.c (find_invariants): Likewise.
26855 * lower-subreg.c (propagate_pseudo_copies): Likewise.
26856 * predict.c (tree_predict_by_opcode): Likewise.
26857 (predict_paths_leading_to): Likewise.
26858 (predict_paths_leading_to_edge): Likewise.
26859 (estimate_loops_at_level): Likewise.
26860 (estimate_loops): Likewise.
26861 * shrink-wrap.c (try_shrink_wrapping): Likewise.
26862 (spread_components): Likewise.
26863 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
26864 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
26865 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
26866 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
26867 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
26868 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
26869 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
26870 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
26871 (mark_threaded_blocks): Likewise.
26872 (thread_through_all_blocks): Likewise.
26873 * tree-ssa.c (verify_ssa): Likewise.
26874 (execute_update_addresses_taken): Likewise.
26875 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
26877 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26879 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
26881 (post_order_compute): Likewise.
26882 (inverted_post_order_compute): Likewise.
26883 (pre_and_rev_post_order_compute_fn): Likewise.
26885 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26887 * genrecog.c (int_set::int_set): Explicitly construct our
26888 auto_vec base class.
26889 * vec.h (auto_vec::auto_vec): New constructor.
26891 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26893 * bitmap.h (class auto_bitmap): New constructor taking
26894 bitmap_obstack * argument.
26896 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26898 * bitmap.h (class auto_bitmap): Change type of m_bits to
26899 bitmap_head, and adjust ctor / dtor and member operators.
26901 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
26903 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
26904 when returned register mode doesn't match original mode.
26906 2017-05-12 Jeff Law <law@redhat.com>
26907 Jakub Jelinek <jakub@redhat.com>
26909 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
26910 we look for cc setter after the compare-elim changes.
26911 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
26912 within the vector to match what compare-elim now expects.
26913 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
26914 (xorsi3_flags, one_cmplsi2_flags): Likewise.
26916 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
26917 after the compare-elim changes.
26918 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
26919 the vector to match what compare-elim now expects.
26920 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
26921 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
26922 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
26923 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
26924 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
26926 * config/visium/visium.c (single_set_and_flags): Fix where
26927 we look for cc setter after the compare-elim changes.
26928 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
26929 with the vector to match what compare-elim now expects.
26930 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
26931 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
26932 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
26933 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
26934 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
26935 (neg<mode>2_insn_set_overflow): Likewise.
26937 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
26939 PR middle-end/79794
26940 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
26941 maybe_expand_insn call, set ops[0].target. If still set after call,
26942 set alt_rtl. Add extra arg to recursive calls.
26943 (extract_bit_field): Add alt_rtl argument. Pass to
26945 * expmed.h (extract_bit_field): Fix prototype.
26946 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
26947 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
26948 to extract_bit_field_calls.
26949 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
26950 Pass alt_rtl to extract_bit_field calls.
26951 * calls.c (store_unaligned_arguments_into_psuedos)
26952 load_register_parameters): Pass extra NULL to extract_bit_field calls.
26953 * optabs.c (maybe_legitimize_operand): Clear op->target when call
26955 * optabs.h (struct expand_operand): Add target bitfield.
26957 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
26959 * compare-elim.c (try_eliminate_compare): Canonicalize
26960 operation with embedded compare to
26961 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
26962 (set (reg) (operation)].
26964 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
26966 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
26969 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
26970 cost of adding a carry flag for ADC instruction.
26971 [case MINUS]: Ignore the cost of subtracting a carry flag
26972 for SBB instruction.
26974 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
26976 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
26978 * config/rs6000/bmiintrin.h: New file.
26979 * config/rs6000/bmi2intrin.h: New file.
26980 * config/rs6000/x86intrin.h: New file.
26982 2017-05-12 Jeff Law <law@redhat.com>
26984 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
26987 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
26989 PR middle-end/80707
26990 * tree-cfg.c: Remove cfg edges of unreachable case statements.
26992 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26994 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
26995 early expansion of vector divide builtins.
26996 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
26997 builtins identified as having unsigned arguments.
26999 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27001 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
27002 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
27003 expansion of vector logical operations (and, andc, or, xor,
27006 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27008 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
27009 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
27011 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27013 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27014 early GIMPLE expansion of vector multiplies.
27016 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27018 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
27019 TARGET_HAVE_MOVT conditional.
27020 (movt splitter): Likewise.
27022 2017-05-12 Richard Biener <rguenther@suse.de>
27024 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
27025 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27026 Fold all stmts not inplace.
27028 2017-05-12 Richard Biener <rguenther@suse.de>
27030 PR tree-optimization/80713
27031 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
27032 inserted_exprs bit for not removed stmts.
27034 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
27036 PR middle-end/69921
27037 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
27038 parallelized" attribute for parallelized OpenACC kernels.
27039 * omp-offload.c (execute_oacc_device_lower): Use it.
27041 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
27042 Set "oacc kernels" attribute.
27043 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
27044 parameter. Adjust all users.
27045 (oacc_fn_attrib_kernels_p): Remove function.
27046 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
27047 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
27048 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
27049 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
27050 assert "oacc kernels" attribute is set.
27052 2017-05-11 Carl Love <cel@us.ibm.com>
27054 * config/rs6000/rs6000-c: Add support for built-in functions
27055 vector unsigned char vec_popcnt (vector signed char)
27056 vector unsigned char vec_popcnt (vector unsigned char)
27057 vector unsigned short vec_popcnt (vector signed short)
27058 vector unsigned short vec_popcnt (vector unsigned short)
27059 vector unsigned int vec_popcnt (vector signed int)
27060 vector unsigned int vec_popcnt (vector unsigned int)
27061 vector unsigned long long vec_popcnt (vector signed long long)
27062 vector unsigned long long vec_popcnt (vector unsigned long long)
27063 vector signed long long vec_slo (vector signed long long,
27064 vector signed char)
27065 vector signed long long vec_slo (vector signed long long,
27066 vector unsigned char)
27067 vector unsigned long long vec_slo (vector unsigned long long,
27068 vector signed char)
27069 vector unsigned long long vec_slo (vector unsigned long long,
27070 vector unsigned char)
27071 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
27072 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
27073 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
27074 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
27075 * doc/extend.texi: Update the built-in documentation file for the
27076 new built-in functions.
27078 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
27080 * attribs.h (sorted_attr_string): Move machine independent
27081 functions for target clone support from the i386 port to common
27082 code. Rename ix86_function_versions to common_function_versions.
27083 Rename make_name to make_unique_name.
27084 (common_function_versions): Likewise.
27085 (make_unique_name): Likewise.
27086 (make_dispatcher_decl): Likewise.
27087 (is_function_default_version): Likewise.
27088 * attribs.c (attr_strcmp): Likewise.
27089 (sorted_attr_string): Likewise.
27090 (common_function_versions): Likewise.
27091 (make_unique_name): Likewise.
27092 (make_dispatcher_decl): Likewise.
27093 (is_function_default_version): Likewise.
27094 * config/i386/i386.c (attr_strcmp): Likewise.
27095 (sorted_attr_string): Likewise.
27096 (ix86_function_versions): Likewise.
27097 (make_name): Likewise.
27098 (make_dispatcher_decl): Likewise.
27099 (is_function_default_version): Likewise.
27100 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
27102 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27105 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
27106 Account for direct move costs for vec_construct of integer
27109 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
27112 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
27113 (UNSPEC_STX_ATOMIC): Ditto.
27114 (loaddi_via_sse): New insn.
27115 (storedi_via_sse): Ditto.
27116 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
27117 Update corresponding peephole2 patterns.
27118 (atomic_storedi_fpu): Ditto.
27120 2017-05-11 Julia Koval <julia.koval@intel.com>
27122 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
27123 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
27125 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
27126 (__builtin_ia32_rsqrt14ss_mask): New builtins.
27127 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
27129 2017-05-11 Nathan Sidwell <nathan@acm.org>
27131 * graphite-poly.c: Include dumpfile.h.
27133 * dumpfle.h (dump_function): Declare here ...
27134 * tree-dump.h (dump_function): ... not here.
27135 * dumpfile.c: #include tree-cfg.h.
27136 (dump_function): Move here from ...
27137 * tree-dump.c (dump_function): ... here.
27138 * gimplify.c: #include splay-tree.h, not tree-dump.h.
27139 * graphite-poly.c: Don't include tree-dump.h.
27140 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
27141 * print-tree.c: Likewise.
27142 * stor-layout.c: Likewise.
27143 * tree-nested.c: Likewise.
27145 * dumpfile.c (dump_start): Use TDF_FLAGS.
27146 (dump_enable_all): Fix TDF_KIND check thinko.
27148 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
27150 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27151 array entries to represent two legal parameterizations of the
27152 overloaded __builtin_cmpb function, as represented by the
27153 P6_OV_BUILTIN_CMPB constant.
27154 (altivec_resolve_overloaded_builtin): Add special case handling
27155 for the __builtin_cmpb function, as represented by the
27156 P6_OV_BUILTIN_CMPB constant.
27157 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
27158 (BU_P6_64BIT_2): New macro.
27159 (BU_P6_OVERLOAD_2): New macro
27160 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
27161 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
27162 (CMPB): Add overload support to represent both 32-bit and 64-bit
27163 compare-bytes function.
27164 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
27165 support for TARGET_CMPB.
27166 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
27167 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27168 documentation of the __builtin_cmpb overloaded built-in function.
27170 2017-05-11 Richard Biener <rguenther@suse.de>
27172 PR tree-optimization/80705
27173 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
27174 bases are not vectorizable.
27176 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27178 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
27179 when counting register pressure.
27181 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27183 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
27184 (iv_ca_more_deps): Renamed to ...
27185 (iv_ca_compare_deps): ... this.
27186 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
27188 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27190 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
27192 (determine_group_iv_costs): ... here.
27193 (find_inv_vars_cb): Record inv var if it's not recorded before.
27195 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27197 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
27198 (get_shiftadd_cost): Ditto.
27200 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27202 * tree-ssa-address.c: Include header file.
27203 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
27205 (add_to_parts): Refactor.
27206 (addr_to_parts): New parameter. Update use of move_hint_to_base.
27207 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
27210 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27212 PR tree-optimization/53090
27213 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
27215 (extract_cond_operands): Detect condition with IV on both sides
27216 and return COMP_IV_EXPR_2.
27217 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
27218 (rewrite_use_compare): Simplify by removing call to function
27219 extract_cond_operands.
27221 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27223 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
27224 (extract_cond_operands): Detect condition comparing against non-
27225 invariant bound and return appropriate enum value.
27226 (find_interesting_uses_cond): Update use of extract_cond_operands.
27227 Handle its return value accordingly.
27228 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
27230 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27232 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
27233 nonlinear iv_use computation in loop invariant sensitive way.
27235 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27237 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
27238 (find_iv_candidates): Call relate_compare_use_with_all_cands.
27240 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27242 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
27243 (dump_cand): Support iv_cand.inv_exprs.
27244 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
27246 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
27249 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27251 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
27253 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
27254 as local function. Include necessary header files.
27255 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
27257 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27259 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
27261 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27263 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
27264 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
27265 RSHIFT_EXPR and BIT_NOT_EXPR.
27267 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27269 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
27270 (adjust_setup_cost): New parameter supporting round up adjustment.
27271 (struct address_cost_data): Delete.
27272 (force_expr_to_var_cost): Don't bound cost with spill_cost.
27273 (split_address_cost, ptr_difference_cost): Delete.
27274 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
27275 (struct ainc_cost_data): New struct.
27276 (get_address_cost_ainc): New function.
27277 (get_address_cost, get_computation_cost): Reimplement.
27278 (determine_group_iv_cost_address): Record inv_expr for all uses of
27280 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
27281 (iv_ca_has_deps): Reimplemented to ...
27282 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
27284 (iv_ca_extend): Call iv_ca_more_deps.
27286 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27288 * tree-ssa-address.c (struct mem_address): Move to header file.
27289 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
27290 * tree-ssa-address.h (struct mem_address): Move from C file.
27291 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
27293 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27295 * tree-affine.h (aff_combination_type): New interface.
27296 (aff_combination_zero_p): Remove static.
27297 (aff_combination_const_p): New interface.
27298 (aff_combination_singleton_var_p): New interfaces.
27300 2017-05-11 Richard Biener <rguenther@suse.de>
27302 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27303 Skip unreachable blocks and destinations.
27304 (eliminate): Move stmt removal and fixup ...
27305 (fini_eliminate): ... here. Skip inserted exprs.
27306 (pass_pre::execute): Move fini_pre after fini_eliminate.
27307 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
27308 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
27309 PRE to get rid of dead code that has invalid SSA form and
27310 split critical edges again.
27312 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27314 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27316 2017-05-11 Richard Biener <rguenther@suse.de>
27318 * passes.c (execute_function_todo): Verify loops if they are
27319 said to be up-to-date.
27320 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
27321 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
27323 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
27326 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
27327 handle calling assemble_external ourself.
27330 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
27331 modes with zero size. Enhance comment.
27333 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27335 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
27336 built-ins for vec_xl and vec_xst with short and char pointer
27339 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27341 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
27342 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
27343 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
27344 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
27345 (_mm_maskz_min_round_ss): New intrinsics.
27346 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
27347 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
27348 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
27349 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
27350 (__builtin_ia32_minss_mask_round): New builtins.
27351 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27352 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27353 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
27355 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
27356 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
27358 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
27361 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27363 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
27364 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
27365 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
27366 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
27367 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
27368 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
27369 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
27370 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27371 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27372 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
27373 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
27374 (__builtin_ia32_mulss_mask_round): New builtins.
27375 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27376 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27377 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
27379 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
27380 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27382 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27385 2017-05-10 Julia Koval <julia.koval@intel.com>
27387 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
27388 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
27389 (_mm256_setr_m128i): New intrinsics.
27391 2017-05-10 Julia Koval <julia.koval@intel.com>
27393 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
27394 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
27395 (_mm_maskz_rcp14_ss): New intrinsics.
27396 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
27397 (__builtin_ia32_rcp14ss_mask): New builtins.
27398 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
27400 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
27402 PR tree-optimization/51513
27403 * tree-cfg.c (gimple_seq_unreachable_p): New function.
27404 (assert_unreachable_fallthru_edge_p): Use it.
27405 (group_case_labels_stmt): Likewise.
27406 * tree-cfg.h: Prototype it.
27407 * stmt.c: Include cfghooks.h and tree-cfg.h.
27408 (emit_case_dispatch_table) <gap_label>: New local variable.
27409 Use it to fill dispatch table gaps.
27410 Test for default_label before updating probabilities.
27411 (expand_case) <default_label>: Remove unneeded initialization.
27412 Test for unreachable default case statement and remove its edge.
27413 Set default_label accordingly.
27414 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
27416 2017-05-10 Carl Love <cel@us.ibm.com>
27418 * config/rs6000/rs6000-c: Add support for built-in functions
27419 vector signed char vec_neg (vector signed char)
27420 vector signed short int vec_neg (vector short int)
27421 vector signed int vec_neg (vector signed int)
27422 vector signed long long vec_neg (vector signed long long)
27423 vector float vec_neg (vector float)
27424 vector double vec_neg (vector double)
27425 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
27427 * config/rs6000/altivec.h: Add define for vec_neg
27428 * doc/extend.texi: Update the built-in documentation for the
27429 new built-in functions.
27431 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27433 PR tree-optimization/77644
27434 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
27436 2017-05-10 Nathan Sidwell <nathan@acm.org>
27438 * dumpfile.h (TDI_lang_all): New.
27439 (TDF_KIND): New. Renumber others
27440 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
27442 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
27444 (get_dump_file_name): Adjust suffix generation.
27445 (dump_enable_all): Use TDF_KIND.
27446 * doc/invoke.texi (-fdump-lang-all): Document.
27448 * dumpfile.h: Tabify.
27450 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
27453 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
27454 Move member access before delete.
27456 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
27458 * tree-inline.c (expand_call_inline): Split block at stmt
27461 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
27464 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
27465 are now unused after splitting mov{sf,sd}_hardfloat.
27466 (f32_lr2): Likewise.
27467 (f32_lm): Likewise.
27468 (f32_lm2): Likewise.
27469 (f32_li): Likewise.
27470 (f32_li2): Likewise.
27471 (f32_lv): Likewise.
27472 (f32_sr): Likewise.
27473 (f32_sr2): Likewise.
27474 (f32_sm): Likewise.
27475 (f32_sm2): Likewise.
27476 (f32_si): Likewise.
27477 (f32_si2): Likewise.
27478 (f32_sv): Likewise.
27479 (f32_dm): Likewise.
27480 (f32_vsx): Likewise.
27481 (f32_av): Likewise.
27482 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
27483 For movsf, order stores so the VSX stores occur before the GPR
27484 store which encourages the register allocator to use a traditional
27485 FPR instead of a GPR. For movsd, order the stores so that the GPR
27486 store comes before the VSX stores to allow the power6 to work.
27487 This is due to the power6 not having a 32-bit integer store
27488 instruction from a FPR.
27489 (movsf_hardfloat): Likewise.
27490 (movsd_hardfloat): Likewise.
27492 2017-05-09 Martin Sebor <msebor@redhat.com>
27494 PR translation/80280
27495 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
27498 PR translation/80280
27499 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
27500 data member added in r247778.
27501 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
27503 2017-05-09 Nathan Sidwell <nathan@acm.org>
27505 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
27507 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
27510 2017-05-09 Marek Polacek <polacek@redhat.com>
27512 * doc/invoke.texi: Fix typo.
27514 2017-05-09 Richard Biener <rguenther@suse.de>
27516 * tree-vrp.c (vrp_val_is_max): Adjust comment.
27517 (vrp_val_is_min): Likewise.
27518 (set_value_range_to_value): Likewise.
27519 (set_value_range_to_nonnegative): Likewise.
27520 (gimple_assign_nonzero_p): Likewise.
27521 (gimple_stmt_nonzero_p): Likewise.
27522 (vrp_int_const_binop): Likewise. Remove unreachable case.
27523 (adjust_range_with_scev): Adjust comments.
27524 (compare_range_with_value): Likewise.
27525 (extract_range_from_phi_node): Likewise.
27526 (test_for_singularity): Likewise.
27528 2017-05-09 Richard Biener <rguenther@suse.de>
27530 * tree-vrp.c (get_single_symbol): Add assert that we don't
27531 get overflowed constants as invariant part.
27532 (compare_values_warnv): Add comment before the TREE_NO_WARNING
27533 checks. Use wi::cmp instead of recursing for integer constants.
27534 (compare_values): Just ignore whether we assumed undefined
27535 overflow instead of failing the compare.
27536 (extract_range_for_var_from_comparison_expr): Add comment before the
27537 TREE_NO_WARNING sets.
27538 (test_for_singularity): Likewise.
27539 (extract_range_from_comparison): Do not disable optimization
27540 when we assumed undefined overflow.
27541 (extract_range_basic): Remove init of unused var.
27543 2017-05-09 Richard Biener <rguenther@suse.de>
27545 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
27546 (extract_range_from_multiplicative_op_1): Adjust.
27547 (extract_range_from_binary_expr_1): Use int_const_binop.
27549 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
27552 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
27553 rs6000_store_data_bypass_p in seven define_bypass directives and
27554 in several comments.
27555 * config/rs6000/rs6000-protos.h: Add prototype for
27556 rs6000_store_data_bypass_p function.
27557 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
27558 function implements slightly different (rs6000-specific) semantics
27559 than store_data_bypass_p, returning false rather than aborting
27560 with assertion error when arguments do not satisfy the
27561 requirements of store data bypass.
27562 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
27563 rs6000_store_data_bypass_p.
27565 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
27567 * config/xtensa/xtensa-protos.h
27568 (xtensa_initial_elimination_offset): New declaration.
27569 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
27570 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
27571 macro definition, add case for FRAME_POINTER_REGNUM when
27572 FRAME_GROWS_DOWNWARD.
27573 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
27574 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
27575 xtensa_initial_elimination_offset.
27577 2017-05-08 Nathan Sidwell <nathan@acm.org>
27579 * doc/invoke.texi: Alphabetize -fdump options.
27581 2017-05-08 Martin Sebor <msebor@redhat.com>
27583 PR translation/80280
27584 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
27586 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
27588 * target.def (compute_frame_layout): New optional target hook.
27589 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
27590 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
27591 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
27593 * reload1.c (verify_initial_elim_offsets): Likewise.
27594 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
27595 (use_simple_return_p): Call arm_compute_frame_layout if needed.
27596 (arm_get_frame_offsets): Split up into this ...
27597 (arm_compute_frame_layout): ... and this function.
27599 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
27601 * config/aarch64/constraints.md (Usa): New constraint.
27602 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
27604 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
27606 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
27607 with_multilib_list after it has been checked.
27609 2017-05-08 Richard Biener <rguenther@suse.de>
27611 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
27612 (bitmap_set_subtract_values): Likewise.
27614 2017-05-08 Richard Biener <rguenther@suse.de>
27616 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
27617 (gimple_assign_nonzero): ... this and remove strict_overflow_p
27619 (gimple_stmt_nonzero_warnv_p): Rename to ...
27620 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
27622 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
27623 (extract_range_basic): Adjust, do not disable propagation on
27624 strict overflow sensitive simplification.
27625 (vrp_visit_cond_stmt): Likewise.
27627 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
27629 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
27630 body size unconditionally.
27632 2017-05-07 Jeff Law <law@redhat.com>
27635 2017-05-06 Jeff Law <law@redhat.com>
27636 PR tree-optimization/78496
27637 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27640 PR tree-optimization/78496
27641 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27642 (simplify_stmt_using_ranges): Call it.
27643 (vrp_dom_walker::before_dom_children): Extract equivalences
27644 from an ASSERT_EXPR with an equality comparison against a
27647 2017-05-06 Jeff Law <law@redhat.com>
27649 PR tree-optimization/78496
27650 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27653 PR tree-optimization/78496
27654 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27655 (simplify_stmt_using_ranges): Call it.
27656 (vrp_dom_walker::before_dom_children): Extract equivalences
27657 from an ASSERT_EXPR with an equality comparison against a
27660 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27662 * lra-constraints.c (lra_copy_reg_equiv): New function.
27663 (split_reg): Use it to copy equivalence information from the
27664 original register to the spill register.
27666 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27668 PR rtl-optimization/75964
27669 * simplify-rtx.c (simplify_const_relational_operation): Remove
27670 invalid handling of comparisons of integer ABS.
27672 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
27674 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
27675 initialize to zero.
27676 (init_regs): Remove declaration.
27677 (function_arg_advance_32): Initialize error_p as boolean variable.
27679 2017-05-05 Nathan Sidwell <nathan@acm.org>
27681 * store-motion.c (remove_reachable_equiv_notes): Reformat long
27682 lines. Use for (;;).
27684 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27686 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
27687 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
27688 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
27689 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
27690 VF=2 that require versioning.
27692 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27694 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
27697 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27699 * diagnostic.h (diagnostic_override_option_index): Convert from
27700 macro to inline function.
27702 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27704 * diagnostic.c (last_module_changed_p): New function.
27705 (set_last_module): New function.
27706 (diagnostic_report_current_module): Convert macro usage to
27707 the above functions.
27708 * diagnostic.h (diagnostic_context::last_module): Strengthen
27709 from const line_map * to const line_map_ordinary *.
27710 (diagnostic_last_module_changed): Delete macro.
27711 (diagnostic_set_last_module): Delete macro.
27713 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27715 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
27716 with diagnostic_report_diagnostic.
27717 (diagnostic_n_impl_richloc): Likewise.
27718 * diagnostic.h (report_diagnostic): Delete macro.
27719 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
27720 with diagnostic_report_diagnostic.
27721 * substring-locations.c (format_warning_va): Likewise.
27723 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27725 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
27726 save/restor of format_spec. Move option-printing code to...
27727 (print_option_information): ...this new function, and
27728 reimplement by simply printing to the pretty_printer,
27729 rather than appending to the format string.
27731 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27733 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
27734 handling logic into...
27735 (update_effective_level_from_pragmas): ...this new function.
27737 2017-05-04 Andrew Waterman <andrew@sifive.com>
27739 * config/riscv/riscv.opt (mstrict-align): New option.
27740 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
27741 (SLOW_UNALIGNED_ACCESS): Define.
27742 (riscv_slow_unaligned_access): Declare.
27743 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
27745 (riscv_slow_unaligned_access): New variable.
27746 (rocket_tune_info): Set slow_unaligned_access to true.
27747 (optimize_size_tune_info): Set slow_unaligned_access to false.
27748 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
27749 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
27750 (riscv_option_override): Set riscv_slow_unaligned_access.
27751 * doc/invoke.texi: Add -mstrict-align to RISC-V.
27753 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
27755 * config/riscv/riscv.md: Unify indentation.
27757 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27762 * config/rs6000/rs6000.md (u code attribute): Add FIX and
27764 (extendsi<mode>2): Add support for doing sign extension via
27765 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
27766 don't have ISA 3.0 instructions.
27767 (extendsi<mode>2 splitter): Likewise.
27768 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
27769 generate the normal insns since SImode can now go in vector
27770 registers. Disallow the special UNSPECs needed for previous
27771 machines to hide SImode being used. Add new insns
27772 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
27773 (fix_trunc<mode>si2_stfiwx): Likewise.
27774 (fix_trunc<mode>si2_internal): Likewise.
27775 (fixuns_trunc<mode>si2): Likewise.
27776 (fixuns_trunc<mode>si2_stfiwx): Likewise.
27777 (fctiw<u>z_<mode>_smallint): Likewise.
27778 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
27779 of floating point to 32-bit integer from doing a direct move to
27780 the GPR registers to do a store.
27781 (fctiwz_<mode>): Break long line.
27783 2017-05-05 Bin Cheng <bin.cheng@arm.com>
27785 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
27786 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
27787 (addr_list, addr_offset_valid_p): New.
27788 (split_address_groups): Check offset validity with above function.
27789 (gt-tree-ssa-loop-ivopts.h): Include header file.
27791 2017-05-05 Nathan Sidwell <nathan@acm.org>
27793 * config.gcc (arm*-*-*): Add missing 'fi'.
27795 2017-05-05 Steve Ellcey <sellcey@cavium.com>
27797 * doc/invoke.texi (-fopt-info): Explicitly say order of options
27798 included in -fopt-info does not matter.
27799 * doc/optinfo.texi (-fopt-info): Fix description of default
27800 behavour. Explicitly say order of options included in -fopt-info
27803 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27805 * config.gcc: Allow combinations of aprofile and rmprofile values for
27806 --with-multilib-list.
27807 * config/arm/t-multilib: New file.
27808 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
27809 variables. Remove setting of ISA and floating-point ABI in
27810 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
27811 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
27812 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
27813 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
27815 * config/arm/t-rmprofile: Likewise except for the matches changes.
27816 * doc/install.texi (--with-multilib-list): Document the combination of
27817 aprofile and rmprofile values and warn about pitfalls in doing that.
27819 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
27821 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
27822 (movdi_aarch64): Likewise.
27824 2017-05-05 Jakub Jelinek <jakub@redhat.com>
27826 PR tree-optimization/80632
27827 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
27829 (build_arrays): Initialize it for virtual phis.
27830 (fix_phi_nodes): Use it for virtual phis.
27832 PR tree-optimization/80558
27833 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
27834 [x, y] op z into [x op, y op z] for op & or | if conditions
27837 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
27838 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27841 * config/arm/arm.md (use_literal_pool): Remove.
27842 (64-bit immediate split): No longer takes cost into consideration
27843 if arm_disable_literal_pool is enabled.
27844 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
27845 used when arm_disable_literal_pool is enabled.
27846 (arm_max_const_double_inline_cost): Remove use of
27847 arm_disable_literal_pool.
27848 (push_minipool_fix): Add assert.
27849 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
27850 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
27851 (no_literal_pool_sf_immediate): New.
27853 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27855 PR tree-optimization/80613
27856 * tree-ssa-dce.c (propagate_necessity): Remove cases for
27857 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
27859 2017-05-05 Richard Biener <rguenther@suse.de>
27861 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
27863 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
27865 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
27866 of this flag from insn conditions due to removal from r247495.
27868 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
27870 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
27872 (arm_early_store_addr_dep_ptr): Likewise.
27873 * config/arm/aarch-common-protos.h
27874 (arm_early_load_addr_dep_ptr): Add prototype.
27875 (arm_early_store_addr_dep_ptr): Likewise.
27876 * config/arm/cortex-a53.md: Add new bypasses.
27878 2017-05-05 Jakub Jelinek <jakub@redhat.com>
27880 * tree.c (next_type_uid): Change type to unsigned.
27881 (type_hash_canon): Decrement back next_type_uid if
27882 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
27883 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
27886 2017-05-04 Martin Sebor <msebor@redhat.com>
27888 * builtins.c: Fix a trivial typo in a comment.
27890 PR middle-end/79234
27891 * builtins.c (check_sizes): Adjust to handle reading past the end.
27892 Avoid printing excessive upper bound of ranges. Use %E to print
27893 tree nodes instead of converting them to %wu.
27894 (expand_builtin_memchr): New function.
27895 (compute_dest_size): Rename...
27896 (compute_objsize): ...to this.
27897 (expand_builtin_memcpy): Adjust.
27898 (expand_builtin_mempcpy): Adjust.
27899 (expand_builtin_strcat): Adjust.
27900 (expand_builtin_strcpy): Adjust.
27901 (check_strncat_sizes): Adjust.
27902 (expand_builtin_strncat): Adjust.
27903 (expand_builtin_strncpy): Adjust and simplify.
27904 (expand_builtin_memset): Adjust.
27905 (expand_builtin_bzero): Adjust.
27906 (expand_builtin_memcmp): Adjust.
27907 (expand_builtin): Handle memcmp.
27908 (maybe_emit_chk_warning): Check strncat just once.
27910 2017-05-04 Martin Sebor <msebor@redhat.com>
27912 PR preprocessor/79214
27913 PR middle-end/79222
27914 PR middle-end/79223
27915 * builtins.c (check_sizes): Add inlining context and issue
27916 warnings even when -Wno-system-headers is set.
27917 (check_strncat_sizes): Same.
27918 (expand_builtin_strncat): Same.
27919 (expand_builtin_memmove): New function.
27920 (expand_builtin_stpncpy): Same.
27921 (expand_builtin): Handle memmove and stpncpy.
27923 2017-05-04 Bin Cheng <bin.cheng@arm.com>
27925 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
27926 which is not used any more.
27928 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27930 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
27932 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27934 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
27935 (cortexa53_tunings): Likewise.
27936 (cortexa57_tunings): Likewise.
27937 (cortexa72_tunings): Likewise.
27938 (cortexa73_tunings): Likewise.
27940 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27942 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
27943 Set loop alignment to 8.
27945 2017-05-04 Martin Sebor <msebor@redhat.com>
27947 PR translation/80280
27948 * builtins.c (expand_builtin_object_size): Add missing quoting to
27949 %D and like directives.
27950 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
27951 (hsa_type_for_tree_type): Same.
27952 (verify_function_arguments): Same.
27953 * symtab.c (symbol_table::change_decl_assembler_name): Same.
27954 * varasm.c (get_section): Same.
27957 2017-05-04 Martin Sebor <msebor@redhat.com>
27959 PR translation/80280
27960 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
27962 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27964 * config/aarch64/aarch64.c (generic_addrcost_table):
27965 Change HI/TI mode setting.
27967 2017-05-04 Martin Jambor <mjambor@suse.cz>
27969 PR tree-optimization/80622
27970 * tree-sra.c (comes_initialized_p): New function.
27971 (build_accesses_from_assign): Only set write lazily when
27972 comes_initialized_p is false.
27973 (analyze_access_subtree): Use comes_initialized_p.
27974 (propagate_subaccesses_across_link): Assert !comes_initialized_p
27975 instead of testing for PARM_DECL.
27977 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27979 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
27980 constraint on operand 0 to allow more general addressing modes.
27981 Adjust output template.
27982 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
27984 * config/aarch64/aarch64-protos.h
27985 (aarch64_address_valid_for_prefetch_p): Declare prototype.
27986 * config/aarch64/constraints.md (Dp): New address constraint.
27987 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
27990 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
27992 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
27993 update use of estimate_ipcp_clone_size_and_time.
27994 (estimate_local_effects): Update use of
27995 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
27996 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
27997 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
27998 Return nonspecialized time.
28000 2017-05-04 Richard Biener <rguenther@suse.de>
28002 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
28003 for the last VUSE which def dominates the PHI. Directly call
28005 (get_continuation_for_phi_1): Remove.
28007 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
28009 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
28010 to explain the use of truncating division. Cap the number of
28011 iterations to the maximum given by nb_iterations_upper_bound,
28014 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28016 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
28017 * configure: Regenerate.
28018 * config.in: Regenerate.
28019 * config/i386/driver-mingw32.c: new file.
28020 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
28021 * config.host: Link driver-mingw32.o on MinGW host.
28022 * doc/install.texi: Document new --enable-mingw-wildcard configure
28025 2017-05-04 Marek Polacek <polacek@redhat.com>
28027 PR tree-optimization/80612
28028 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
28030 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28031 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
28033 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
28034 (movt splitter): Likewise.
28035 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
28036 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
28037 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
28038 block for Thumb-1 with MOVT.
28039 (thumb2_legitimate_address_p): Move code block ...
28040 (can_avoid_literal_pool_for_label_p): ... into this new function.
28041 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
28043 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
28044 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
28045 "M-profile targets with the MOVT instruction".
28047 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28049 * config/arm/arm-builtins.c (arm_init_builtins): Rename
28050 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
28051 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
28053 2017-05-04 Martin Liska <mliska@suse.cz>
28055 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
28056 variable cond_code.
28058 2017-05-04 Richard Biener <rguenther@suse.de>
28060 * tree.c (array_at_struct_end_p): Handle arrays at struct
28061 end with flexarrays more conservatively. Refactor and treat
28062 arrays of arrays or aggregates more strict. Fix
28063 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
28064 * tree.c (array_at_struct_end_p): Adjust prototype.
28065 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
28066 * gimple-fold.c (get_range_strlen): Likewise.
28067 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
28069 2017-05-04 Richard Biener <rguenther@suse.de>
28071 PR tree-optimization/31130
28072 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
28074 (supports_overflow_infinity): Likewise.
28075 (is_negative_overflow_infinity): Likewise.
28076 (is_positive_overflow_infinity): Likewise.
28077 (is_overflow_infinity): Likewise.
28078 (stmt_overflow_infinity): Likewise.
28079 (overflow_infinity_range_p): Likewise.
28080 (usable_range_p): Remove as always returning true.
28081 (make_overflow_infinity): Remove.
28082 (negative_overflow_infinity): Likewise.
28083 (positive_overflow_infinity): Likewise.
28084 (avoid_overflow_infinity): Likewise.
28085 (set_value_range): Adjust accordingly.
28086 (set_value_range_to_nonnegative): Likewise, remove now unused
28087 overflow_infinity arg.
28088 (vrp_operand_equal_p): Adjust.
28089 (update_value_range): Likewise.
28090 (range_int_cst_singleton_p): Likewise.
28091 (operand_less_p): Likewise.
28092 (compare_values_warnv): Likewise.
28093 (extract_range_for_var_from_comparison_expr): Likewise.
28094 (vrp_int_const_binop): Likewise.
28095 (zero_nonzero_bits_from_vr): Likewise.
28096 (extract_range_from_multiplicative_op_1): Likewise.
28097 (extract_range_from_binary_expr_1): Likewise.
28098 (extract_range_from_unary_expr): Likewise.
28099 (extract_range_from_comparison): Likewise.
28100 (extract_range_basic): Likewise.
28101 (adjust_range_with_scev): Likewise.
28102 (compare_ranges): Likewise.
28103 (compare_range_with_value): Likewise.
28104 (dump_value_range): Likewise.
28105 (test_for_singularity): Likewise, remove strict_overflow_p parameter
28107 (simplify_cond_using_ranges): Adjust.
28109 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28111 * brig-builtins.def: Added a builtin for class_f64.
28112 * builtin-types.def: Added a builtin type needed by class_f64.
28114 2017-05-03 Jason Merrill <jason@redhat.com>
28116 * timevar.def: Add TV_CONSTEXPR.
28118 2017-05-03 David Malcolm <dmalcolm@redhat.com>
28120 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
28122 2017-05-03 Martin Jambor <mjambor@suse.cz>
28124 * ipa-prop.c (ipa_update_after_lto_read): Removed.
28125 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
28126 * ipa-cp.c (ipcp_propagate_stage): Do not call
28127 ipa_update_after_lto_read.
28128 * ipa-inline.c (ipa_inline): Likewise.
28130 2017-05-03 Martin Jambor <mjambor@suse.cz>
28132 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
28133 tag. Added a default constructor and a destructor.
28134 (ipa_edge_args_sum_t): New class;
28135 (ipa_edge_args_sum): Declare.
28136 (ipa_edge_args_vector): Remove declaration.
28137 (IPA_EDGE_REF): Use ipa_edge_args_sum.
28138 (ipa_free_edge_args_substructures): Remove declaration.
28139 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
28140 (ipa_edge_args_info_available_for_edge_p): Likewise.
28141 * ipa-prop.c (ipa_edge_args_vector): Removed.
28142 (edge_removal_hook_holder): Likewise.
28143 (edge_duplication_hook_holder): Likewise.
28144 (ipa_edge_args_sum): New variable.
28145 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
28146 ipa_edge_args_vector.
28147 (ipa_free_edge_args_substructures): Likewise.
28148 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
28149 ipa_edge_args_vector.
28150 (ipa_edge_removal_hook): Turned into method
28151 ipa_edge_args_sum_t::remove.
28152 (ipa_edge_duplication_hook): Turned into method
28153 ipa_edge_args_sum_t::duplicate.
28154 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
28155 registering edge hooks.
28156 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
28157 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
28158 ipa_edge_args_sum instead of ipa_edge_args_vector.
28159 * ipa-profile.c (ipa_profile): Likewise.
28161 2017-05-03 Martin Jambor <mjambor@suse.cz>
28163 * symbol-summary.h (function_summary): New method exists.
28164 (function_summary::symtab_removal): Deallocate through release.
28165 (call_summary): New class.
28166 (gt_ggc_mx): New overload.
28167 (gt_pch_nx): Likewise.
28168 (gt_pch_nx): Likewise.
28170 2017-05-03 Jeff Law <law@redhat.com>
28172 PR tree-optimization/78496
28173 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
28174 from simplify_cond_using_ranges. Split off code to walk
28175 backwards through casts into ...
28176 (simplify_cond_using_ranges_2): New function.
28177 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
28178 (execute_vrp): After identifying jump threads, call
28179 simplify_cond_using_ranges_2.
28181 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
28184 * ipa-inline.h (inline_summary): Add ctor.
28185 (create_ggc): Do not use ggc_cleared_alloc.
28187 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
28188 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28190 * gcc.c (handle_braces): Support escaping in switch matching
28192 * doc/invoke.texi (Spec Files): Document it.
28193 Remove superfluous @code markup in items.
28195 2017-05-03 David Malcolm <dmalcolm@redhat.com>
28197 * diagnostic-show-locus.c (struct column_range): New struct.
28198 (get_affected_columns): New function.
28199 (get_printed_columns): New function.
28200 (struct correction): New struct.
28201 (correction::ensure_capacity): New function.
28202 (correction::ensure_terminated): New function.
28203 (struct line_corrections): New struct.
28204 (line_corrections::~line_corrections): New dtor.
28205 (line_corrections::add_hint): New function.
28206 (layout::print_trailing_fixits): Reimplement in terms of the new
28208 (selftest::test_overlapped_fixit_printing): New function.
28209 (selftest::diagnostic_show_locus_c_tests): Call it.
28211 2017-05-03 Nathan Sidwell <nathan@acm.org>
28213 Canonicalize canonical type hashing
28214 * tree.h (type_hash_canon_hash): Declare.
28215 * tree.c (type_hash_list, attribute_hash_list): Move into
28216 type_hash_canon_hash.
28217 (build_type_attribute_qual_variant): Break out hash code calc into
28218 type_hash_canon_hash.
28219 (type_hash_canon_hash): New. Generic type hash computation.
28220 (build_range_type_1, build_array_type_1, build_function_type,
28221 build_method_type_directly, build_offset_type, build_complex_type,
28222 make_vector_type): Call it.
28224 2017-05-03 Richard Biener <rguenther@suse.de>
28226 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
28227 When all DRs have unknown misaligned do not always peel
28228 when there is a store but apply the same costing model as if
28229 there were only loads.
28231 2017-05-03 Richard Biener <rguenther@suse.de>
28234 PR tree-optimization/80492
28235 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
28236 compare_base_decls returning dont-know properly.
28238 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
28240 * config/arm/iterators.md (CCSI): New mode iterator.
28241 (arch): New mode attribute.
28242 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
28243 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
28244 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
28245 code iterator for success result mode.
28246 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
28247 the corresponding new insn generators.
28249 2017-05-03 Bin Cheng <bin.cheng@arm.com>
28252 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28253 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28255 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
28257 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
28258 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
28259 (DDR_A): Wrap DDR argument in brackets.
28260 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
28261 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
28262 (DDR_REVERSED_P): Likewise.
28264 2017-05-03 Jakub Jelinek <jakub@redhat.com>
28266 PR tree-optimization/79472
28267 * tree-switch-conversion.c (struct switch_conv_info): Add
28268 contiguous_range and default_case_nonstandard fields.
28269 (collect_switch_conv_info): Compute contiguous_range and
28270 default_case_nonstandard fields, don't clear final_bb if
28271 contiguous_range and only the default case doesn't have the required
28273 (check_all_empty_except_final): Set default_case_nonstandard instead
28274 of failing if contiguous_range and the default case doesn't have empty
28276 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
28277 and only the default case doesn't have the required constants. Skip
28279 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
28280 if default_case_nonstandard.
28281 (build_constructors): Build constant 1 just once. Assert that default
28282 values aren't inserted in between cases if contiguous_range. Skip
28284 (build_arrays): Skip virtual phis.
28285 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
28286 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
28287 Handle virtual phis.
28288 (gen_inbound_check): Handle default_case_nonstandard case.
28289 (process_switch): Adjust check_final_bb caller. Call
28290 gather_default_values with the first non-default case instead of
28291 default case if default_case_nonstandard.
28293 2017-05-02 Nathan Sidwell <nathan@acm.org>
28295 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
28296 check. Fix formatting.
28298 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
28300 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
28301 errors when comparing specialized and unspecialized times.
28303 2017-05-02 David Malcolm <dmalcolm@redhat.com>
28305 * diagnostic-show-locus.c
28306 (layout::should_print_annotation_line_p): Make private.
28307 (layout::print_annotation_line): Make private.
28308 (layout::annotation_line_showed_range_p): Make private.
28309 (layout::show_ruler): Make private.
28310 (layout::print_source_line): Make private. Pass in line and
28311 line_width, rather than calling location_get_source_line. Drop
28313 (layout::print_leading_fixits): New method.
28314 (layout::print_any_fixits): Rename to...
28315 (layout::print_trailing_fixits): ...this, and make private.
28316 Don't print newline fixits.
28317 (diagnostic_show_locus): Move logic for printing one row into...
28318 (layout::print_line): ...this new function. Move the
28319 location_get_source_line call and error-handling from
28320 print_source_line to here. Call print_leading_fixits, and rename
28321 print_any_fixits to print_trailing_fixits.
28322 (selftest::test_fixit_insert_containing_newline): Update now that
28323 newlines are partially supported.
28324 (selftest::test_fixit_insert_containing_newline_2): New test.
28325 (selftest::test_fixit_replace_containing_newline): Update comments.
28326 (selftest::diagnostic_show_locus_c_tests): Call the new test.
28327 * edit-context.c (class added_line): New class.
28328 (class edited_line): Describe newline handling in comment.
28329 (edited_line::actually_edited_p): New method.
28330 (edited_line::print_content): Delete redundant decl.
28331 (edited_line::m_predecessors): New field.
28332 (edited_file::print_content): Call edited_line::print_content.
28333 (edited_file::print_diff): Update to support newlines.
28334 (edited_file::print_diff_hunk): Likewise.
28335 (edited_file::print_run_of_changed_lines): New function.
28336 (edited_file::print_diff_line): Convert to...
28337 (print_diff_line): ...this.
28338 (edited_file::get_effective_line_count): New function.
28339 (edited_line::edited_line): Initialize new field m_predecessors.
28340 (edited_line::~edited_line): Clean up m_predecessors.
28341 (edited_line::apply_fixit): Handle newlines.
28342 (edited_line::get_effective_line_count): New function.
28343 (edited_line::print_content): New function.
28344 (edited_line::print_diff_lines): New function.
28345 (selftest::test_applying_fixits_insert_containing_newline): New
28347 (selftest::test_applying_fixits_replace_containing_newline): New
28349 (selftest::insert_line): New function.
28350 (selftest::test_applying_fixits_multiple_lines): Add example of
28352 (selftest::edit_context_c_tests): Call the new tests.
28354 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28356 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
28357 parameter cand. Update dump information.
28358 (get_computation_cost): Update uses.
28360 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28362 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
28363 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
28364 (get_computation_at, rewrite_use_address): Update use of
28365 get_computation_aff.
28367 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28369 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
28370 (get_computation): Delete.
28371 (get_computation_cost): Implement like get_computation_cost_at.
28372 Use get_computation_at.
28373 (get_computation_cost_at): Delete.
28374 (rewrite_use_nonlinear_expr): Use get_computation_at.
28375 (rewrite_use_compare, remove_unused_ivs): Ditto.
28377 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28379 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
28381 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28383 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
28384 (ivopts_global_cost_for_size): Rename parameter and update uses.
28385 (iv_ca_recount_cost): Update uses.
28386 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
28387 candidates seperately in n_invs and n_cands.
28388 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
28390 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28392 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
28393 (find_inv_vars_cb): New.
28394 (find_depends): Renamed to ...
28395 (find_inv_vars): ... this.
28396 (add_candidate_1, force_var_cost): Call find_inv_vars.
28397 (split_address_cost, determine_group_iv_cost_cond): Ditto.
28399 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28401 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
28402 inv_vars. Add inv_exprs.
28403 (struct iv_cand): Rename depends_on to inv_vars.
28404 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
28405 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
28406 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
28407 (dump_cand): Dump inv_vars.
28408 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
28409 (record_invariant, find_depends, add_candidate_1): Ditto.
28410 (set_group_iv_cost, force_var_cost): Ditto.
28411 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
28412 (get_computation_cost_at, get_computation_cost): Ditto.
28413 (determine_group_iv_cost_generic): Ditto.
28414 (determine_group_iv_cost_address): Ditto.
28415 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
28416 (determine_group_iv_costs): Ditto.
28417 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
28418 (iv_ca_set_remove_invariants): Renamed to ...
28419 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
28420 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
28421 (iv_ca_set_add_invariants): Renamed to ...
28422 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
28423 (iv_ca_set_cp): Use iv_ca_set_add_invs.
28424 (iv_ca_has_deps): Support inv_vars and inv_exprs.
28425 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
28426 (create_new_ivs): Remove useless dump.
28428 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28430 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
28432 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
28433 (iv_ca_set_no_cp, create_new_iv): Ditto.
28435 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28437 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28439 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28441 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
28442 function tree_check2.
28444 2017-05-02 Martin Liska <mliska@suse.cz>
28446 * doc/gcov.texi: Add missing preposition.
28447 * gcov.c (function_info::function_info): Properly fill up
28448 all member variables.
28450 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28452 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
28454 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28456 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
28458 2017-05-02 Martin Liska <mliska@suse.cz>
28461 * lto-streamer-in.c (lto_read_tree_1): Remove
28462 LTO_STREAMER_DEBUG.
28463 * lto-streamer.c (struct tree_hash_entry): Likewise.
28464 (struct tree_entry_hasher): Likewise.
28465 (tree_entry_hasher::hash): Likewise.
28466 (tree_entry_hasher::equal): Likewise.
28467 (lto_streamer_init): Likewise.
28468 (lto_orig_address_map): Likewise.
28469 (lto_orig_address_get): Likewise.
28470 (lto_orig_address_remove): Likewise.
28471 * lto-streamer.h: Likewise.
28472 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
28473 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
28475 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
28477 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
28478 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
28479 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
28480 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
28481 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
28482 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
28483 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
28484 (mm_maskz_sub_ss): New intrinsics.
28485 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28486 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28487 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
28488 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
28489 (__builtin_ia32_subss_mask_round): New builtins.
28490 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28491 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28492 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
28494 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
28495 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28497 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28500 2017-05-02 Martin Jambor <mjambor@suse.cz>
28502 PR tree-optimization/78687
28503 * tree-sra.c (access): New field parent.
28504 (process_subtree_disqualification): New function.
28505 (disqualify_candidate): Call it.
28506 (build_accesses_from_assign): Reset write flag if creating an
28508 (build_access_subtree): Fill in parent field and also prpagate
28509 down grp_write flag.
28510 (create_artificial_child_access): New parameter set_grp_write, set
28511 grp_write to its value.
28512 (propagate_subaccesses_across_link): Also propagate grp_write flag
28514 (propagate_all_subaccesses): Push the closest parent back to work
28515 queue if add_access_to_work_queue returned true.
28517 2017-05-02 Richard Biener <rguenther@suse.de>
28519 * common.opt (fstrict-overflow): Alias negative to fwrapv.
28520 * doc/invoke.texi (fstrict-overflow): Remove all traces of
28521 -fstrict-overflow documentation.
28522 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
28523 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
28524 flag_strict_overflow.
28525 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
28526 * lto-opts.c (lto_write_options): Do not stream it.
28527 * lto-wrapper.c (merge_and_complain): Do not handle it.
28528 * opts.c (default_options_table): Do not set -fstrict-overflow.
28529 (finish_options): Likewise do not clear it when sanitizing.
28530 * simplify-rtx.c (simplify_const_relational_operation): Do not
28531 test flag_strict_overflow.
28533 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
28535 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
28536 using enabled attribute.
28537 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
28538 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
28539 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
28540 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
28541 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
28542 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
28543 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
28544 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
28545 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
28546 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
28548 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
28550 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
28552 2017-05-02 Richard Biener <rguenther@suse.de>
28554 PR tree-optimization/80591
28556 2017-04-10 Richard Biener <rguenther@suse.de>
28558 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28561 2017-05-02 Richard Biener <rguenther@suse.de>
28563 PR tree-optimization/80549
28564 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
28565 (cleanup_tree_cfg_noloop): Create forwarders to known loop
28566 headers if they do not have a preheader.
28568 2017-05-02 Martin Liska <mliska@suse.cz>
28571 * common.opt: Fix typo.
28572 * doc/invoke.texi: Likewise.
28574 2017-05-01 Jan Beulich <jbeulich@suse.com>
28576 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
28577 swapping, add (x,x,m,x,n) alternative.
28579 2017-05-01 Nathan Sidwell <nathan@acm.org>
28581 * calls.c (combine_pending_stack_adjustment_and_call): Remove
28582 unnecessary unadjusted_alignment check.
28584 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28587 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
28589 * gimplify.c (gimplify_cilk_detach): New function.
28590 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
28591 * tree-core.h: Document EXPR_CILK_SPAWN.
28592 * tree.h (EXPR_CILK_SPAWN): Define.
28594 2017-05-01 David Malcolm <dmalcolm@redhat.com>
28596 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
28597 to use new fixit_hint representation, using the "replace" logic.
28598 (get_line_span_for_fixit_hint): Likewise.
28599 (layout::print_any_fixits): Likewise.
28600 (selftest::test_one_liner_many_fixits): Rename to...
28601 (selftest::test_one_liner_many_fixits_1): ...this, and update
28602 comment and expected output to reflect that the multiple fix-it
28603 hints are now consolidated into one insertion.
28604 (selftest::test_one_liner_many_fixits_2): New test.
28605 (selftest::test_diagnostic_show_locus_one_liner): Update for
28607 (selftest::test_fixit_consolidation): Update for fix-it API
28609 * diagnostic.c (print_parseable_fixits): Likewise.
28610 * edit-context.c (edited_line::m_line_events): Convert from
28611 auto_vec <line_event *> to auto_vec <line_event>.
28612 (class line_event): Convert from abstract base class to a concrete
28613 class, taking over the role of replace_event.
28614 (class insert_event): Delete.
28615 (class replace_event): Rename to class line_event. Convert to
28617 (edit_context::add_fixits): Reimplement.
28618 (edit_context::apply_insert): Delete.
28619 (edit_context::apply_replace): Rename to...
28620 (edit_context::apply_fixit): ...this. Convert to half-open range.
28621 (edited_file::apply_insert): Delete.
28622 (edited_file::apply_replace): Rename to...
28623 (edited_file::apply_fixit): ...this.
28624 (edited_line::~edited_line): Drop deletion of events.
28625 (edited_line::apply_insert): Delete.
28626 (edited_line::apply_replace): Rename to...
28627 (edited_line::apply_fixit): ...this. Convert to half-open range.
28628 Update for change to type of m_line_events.
28629 * edit-context.h (edit_context::apply_insert): Delete.
28630 (edit_context::apply_replace): Rename to...
28631 (edit_context::apply_fixit): ...this.
28633 2017-05-01 Martin Sebor <msebor@redhat.com>
28635 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
28638 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
28641 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
28642 __get_cpuid_max returns 0.
28643 (__get_cpuid_count): Ditto.
28645 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
28647 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
28648 replacement expression is another instance of one of its arguments.
28650 2017-05-01 Jakub Jelinek <jakub@redhat.com>
28653 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
28654 check for stack push/pop autoinc.
28655 * config/i386/i386.c (ix86_agi_dependent): Return false
28656 if the only reason why modified_in_p returned true is that
28657 addr is SP based and set_insn is a push or pop.
28659 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28661 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
28664 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28667 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
28668 (account_size_time): Use two predicates - exec_pred and
28670 (evaluate_conditions_for_known_args): Compute both clause and
28672 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
28673 (inline_summary_t::duplicate): Update.
28674 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
28676 (compute_inline_parameters): Likewise.
28677 (estimate_edge_size_and_time): Update caluclation of time.
28678 (estimate_node_size_and_time): Compute both time and nonspecialized
28680 (estimate_ipcp_clone_size_and_time): Update.
28681 (inline_merge_summary): Update.
28682 (do_estimate_edge_time): Update.
28683 (do_estimate_edge_size): Update.
28684 (do_estimate_edge_hints): Update.
28685 (inline_read_section, inline_write_summary): Stream both new predicates.
28686 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
28688 (compute_inlined_call_time): Cleanup.
28689 (big_speedup_p): Update.
28690 (edge_badness): Update.
28691 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
28692 (size_time_entry): Replace predicate by exec_predicate and
28693 nonconst_predicate.
28694 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
28695 (estimate_edge_time): Return also nonspec_time.
28696 (reset_edge_growth_cache): Update.
28698 2017-04-29 Jakub Jelinek <jakub@redhat.com>
28700 PR rtl-optimization/80491
28701 * ifcvt.c (noce_process_if_block): When looking for x setter
28702 with missing else_bb, don't check only the insn right before
28703 cond_earliest, but look for the last insn that x is modified in
28704 within the same bb.
28706 PR rtl-optimization/80491
28707 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
28709 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
28711 PR tree-optimization/80487
28712 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
28714 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28716 PR tree-optimization/79697
28717 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
28718 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
28719 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
28721 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
28722 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
28724 2017-04-28 Martin Sebor <msebor@redhat.com>
28726 PR tree-optimization/80523
28727 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
28728 (init_target_to_host_charmap, target_to_host, target_strtol10): New
28730 (maybe_warn, format_directive, parse_directive): Use new functions.
28731 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
28733 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
28735 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
28737 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
28739 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
28740 target_header_dir): Set correctly.
28741 * configure: Regenerated.
28742 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28743 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28744 instead of SYSTEM_HEADER_DIR.
28746 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
28748 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
28749 (estimate_local_effects): Likewise.
28750 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
28751 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
28752 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
28753 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
28754 do_estimate_edge_time, estimate_edge_time): Likewise.
28755 * ipa-inline-analysis.c (estimate_node_size_and_time,
28756 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
28757 (estimate_time_after_inlining): Remove.
28759 2017-04-28 Martin Liska <mliska@suse.cz>
28761 * doc/gcov.texi: Enhance documentation of gcov.
28763 2017-04-28 Martin Liska <mliska@suse.cz>
28765 * doc/gcov.texi: Sort options in alphabetic order.
28766 * doc/gcov-dump.texi: Likewise.
28767 * doc/gcov-tool.texi: Likewise.
28768 * gcov.c (print_usage): Likewise.
28769 * gcov-dump.c (print_usage): Likewise.
28770 * gcov-tool.c (print_merge_usage_message): Likewise.
28771 (print_rewrite_usage_message): Likewise.
28772 (print_overlap_usage_message): Likewise.
28774 2017-04-28 Martin Liska <mliska@suse.cz>
28776 PR gcov-profile/53915
28777 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
28779 2017-04-28 Martin Liska <mliska@suse.cz>
28781 PR gcov-profile/79891
28782 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
28783 is marked by compiler as living on a line.
28784 (get_cycles_count): Remove usage of the union.
28785 (output_intermediate_file): Likewise.
28786 (find_source): Fix GNU coding style.
28787 (accumulate_line_counts): Remove old non-all block mode.
28788 (output_lines): Remove usage of the union.
28789 * profile.c (output_location): Include all BBs, even if
28790 belonging to a same line (and file) as a previous BB.
28792 2017-04-28 Martin Liska <mliska@suse.cz>
28794 * gcov.c (process_args): Handle new argument 'w'.
28795 (read_graph_file): Assign ID to BBs.
28796 (output_branch_count): Display BB # if verbose flag is set.
28797 (output_lines): Likewise for arcs.
28798 (print_usage): Add '--verbose' option help.
28799 * doc/gcov.texi: Document --verbose (-w) option.
28801 2017-04-28 Martin Liska <mliska@suse.cz>
28803 * gcov.c (struct block_location_info): New struct.
28804 (process_file): Fill up the new structure.
28805 (read_graph_file): Replace usage of encoding by the newly added
28807 (add_line_counts): Likewise.
28808 (accumulate_line_counts): Remove usage of the union.
28809 (function_info::function_info): New function.
28810 (function_info::~function_info): Likewise.
28811 (process_file): Call delete instead of release_function.
28812 (release_function): Release the function.
28813 (release_structures): Call delete instead of release_function.
28814 (solve_flow_graph): Replace usage of num_blocks.
28815 (find_exception_blocks): Likewise.
28816 (output_lines): Fix GNU coding style.
28818 2017-04-28 Martin Liska <mliska@suse.cz>
28821 * coverage.c (coverage_remove_note_file): New function.
28822 * coverage.h: Declare the function.
28823 * toplev.c (finalize): Clean if an error has been seen.
28825 2017-04-28 Martin Liska <mliska@suse.cz>
28827 PR gcov-profile/80031
28828 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
28829 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
28830 * gcov.c (read_graph_file): Read just number of blocks.
28831 * profile.c (branch_prob): Do not stream 0 flags per a basic
28834 2017-04-28 Martin Liska <mliska@suse.cz>
28836 * gcov-dump.c (tag_*): Add new argument to declarations.
28837 (dump_gcov_file): Likewise.
28838 (tag_blocks): Add and use new argument depth.
28839 (tag_arcs): Likewise.
28840 (tag_lines): Likewise.
28841 (tag_counters): Likewise.
28842 (tag_summary): Likewise.
28843 (dump_working_sets): Use depth to do a proper indentation.
28845 2017-04-28 Jakub Jelinek <jakub@redhat.com>
28848 * cgraph.h (symtab_node::debug_symtab): No longer inline.
28849 * symtab.c (symtab_node::debug_symtab): Move definition here.
28851 2017-04-28 Richard Biener <rguenther@suse.de>
28853 * lto-streamer.h (LTO_major_version): Bump to 7.
28855 2017-04-28 Richard Biener <rguenther@suse.de>
28857 * tree-vrp.c (assert_info): New struct.
28858 (add_assert_info): New helper.
28859 (register_edge_assert_for_2): Refactor to add asserts to a vector
28861 (register_edge_assert_for_1): Likewise.
28862 (register_edge_assert_for): Likewise.
28863 (finish_register_edge_assert_for): New helper actually registering
28864 asserts where live on edge.
28865 (find_conditional_asserts): Adjust.
28866 (find_switch_asserts): Likewise.
28867 (evrp_dom_walker::try_find_new_range): Generalize.
28868 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
28870 2017-04-27 Marek Polacek <polacek@redhat.com>
28873 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
28874 arg10 and arg11 to itype.
28876 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
28878 * doc/extend.texi (Object Size Checking): Improve grammar.
28880 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
28883 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
28884 that the logic for permitting reciprocal estimates matches that
28887 2017-04-27 Jakub Jelinek <jakub@redhat.com>
28890 * tree.c (type_cache_hasher::equal): Only compare
28891 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
28892 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
28893 non-aggregate element types.
28894 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
28895 about the flag on ARRAY_TYPEs in the comment, formatting fix.
28897 2017-04-27 Richard Biener <rguenther@suse.de>
28899 PR middle-end/80533
28900 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
28901 stripping ARRAY_REFs from MEM_EXPR make sure we're not
28902 keeping a reference to a trailing array.
28904 2017-04-27 Richard Biener <rguenther@suse.de>
28906 PR middle-end/80539
28907 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
28908 being in loop-closed SSA form conservatively.
28909 (chrec_fold_multiply_poly_poly): Likewise.
28911 2017-04-27 Tamar Christina <tamar.christina@arm.com>
28913 PR middle-end/79665
28914 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
28915 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
28917 2017-04-27 Jakub Jelinek <jakub@redhat.com>
28920 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
28921 (aarch64_function_arg_alignment): Return unsigned int again, but still
28922 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
28923 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
28924 Don't emit -Wpsabi note.
28925 (aarch64_function_arg_boundary): Likewise.
28926 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
28929 2017-04-26 Nathan Sidwell <nathan@acm.org>
28931 * tree.h (crc32_unsigned_n): Declare.
28932 (crc32_unsigned, crc32_unsigned): Make inline.
28933 * tree.c (crc32_unsigned_bits): Replace with ...
28934 (crc32_unsigned_n): ... this.
28935 (crc32_unsigned, crc32_byte): Remove.
28936 (crc32_string): Remove unnecessary braces.
28938 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
28940 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
28941 * ipa-inline-analysis.c (MAX_TIME): Remove.
28942 (account_size_time): Use sreal for time.
28943 (dump_inline_summary): Update.
28944 (estimate_function_body_sizes): Update.
28945 (estimate_edge_size_and_time): Update.
28946 (estimate_calls_size_and_time): Update.
28947 (estimate_node_size_and_time): Update.
28948 (inline_merge_summary): Update.
28949 (inline_update_overall_summary): Update.
28950 (estimate_time_after_inlining): Update.
28951 (inline_read_section): Update.
28952 (inline_write_summary): Update.
28953 * ipa-inline.c (compute_uninlined_call_time): Update.
28954 (compute_inlined_call_time): Update.
28955 (recursive_inlining): Update.
28956 (inline_small_functions): Update.
28957 (dump_overall_stats): Update.
28958 * ipa-inline.h: Include sreal.h.
28959 (size_time_entry): Turn time to sreal.
28960 (inline_summary): Turn self_time nad time to sreal.
28962 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
28964 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
28966 (sreal::stream_out, sreal::stream_in): New.
28967 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
28969 2017-04-25 Jakub Jelinek <jakub@redhat.com>
28971 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
28974 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
28977 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
28979 (dimode_scalar_chain::compute_convert_gain): Ditto.
28980 (dimode_scalar_chain::make_vector_copies): Ditto.
28981 (dimode_scalar_chain::convert_reg): Ditto.
28982 (dimode_scalar_chain::convert_insn): Ditto.
28983 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
28984 (VI248_AVX512BW_1): New mode iterator.
28985 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
28986 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
28989 2017-04-25 Martin Sebor <msebor@redhat.com>
28991 PR tree-optimization/80497
28992 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
28993 constants are representable in HOST_WIDE_INT.
28994 (parse_directive): Ditto.
28996 2017-04-25 Martin Sebor <msebor@redhat.com>
28999 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
29000 (new_zero_array): Adjust signature.
29001 (dom_info::dom_init): Used unsigned rather that size_t.
29002 (dom_info::dom_info): Same.
29004 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29005 Jakub Jelinek <jakub@redhat.com>
29008 * config/arm/arm.c: Include gimple.h.
29009 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29010 returns negative, increment ncrn only if it returned positive.
29011 (arm_needs_doubleword_align): Return int instead of bool,
29012 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
29013 members, but if there is any such non-FIELD_DECL
29014 > PARM_BOUNDARY aligned decl, return -1 instead of false.
29015 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29016 returns negative, increment nregs only if it returned positive.
29017 (arm_setup_incoming_varargs): Likewise.
29018 (arm_function_arg_boundary): Emit -Wpsabi note if
29019 arm_needs_doubleword_align returns negative, return
29020 DOUBLEWORD_ALIGNMENT only if it returned positive.
29022 2017-04-25 Marek Polacek <polacek@redhat.com>
29025 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29026 first argument to type.
29028 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
29031 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
29032 type checks to test for compatibility instead of equality.
29034 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29035 Jakub Jelinek <jakub@redhat.com>
29038 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
29040 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
29041 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
29042 the alignment computation, but return their maximum in warn_alignment.
29043 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29044 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
29046 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
29047 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29050 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29052 * config/arc/simdext.md (dmpyh): Fix typo.
29054 2017-04-25 Richard Biener <rguenther@suse.de>
29056 PR tree-optimization/80492
29057 * alias.c (compare_base_decls): Handle registers with asm
29058 specification conservatively.
29059 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29060 compare_base_decls returning dont-know properly.
29062 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29064 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
29065 (legitimate_offset_address_p): New function.
29066 (arc_legitimate_address_p): Use above function.
29068 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29070 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
29072 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29074 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
29075 ACCH registers whenever they are available.
29077 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29079 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
29080 double regs fix when not used.
29082 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29084 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
29086 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
29087 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
29089 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29091 * config/arc/arc.c (arc_output_addsi): Check for h-register class
29092 when emitting short ADD instructions.
29094 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29096 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
29098 (cmpsi_cc_c_insn): Likewise.
29099 (cbranchsi4_scratch): Compute proper instruction length using
29100 compact_hreg_operand.
29101 * config/arc/predicates.md (compact_hreg_operand): New predicate.
29103 2017-04-25 Richard Biener <rguenther@suse.de>
29105 PR middle-end/80509
29106 * passes.c (pass_manager::pass_manager): Initialize
29107 m_name_to_pass_map.
29109 2017-04-25 Richard Biener <rguenther@suse.de>
29111 PR tree-optimization/79201
29112 * tree-ssa-sink.c (statement_sink_location): Handle calls.
29114 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29117 * config/s390/vector.md: Split MEM->GPR vector moves for
29118 non-s_operand addresses.
29120 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29123 * config/s390/predicates.md (reload_const_wide_int_operand): New
29125 * config/s390/s390.md ("movti"): Remove d/P alternative.
29126 ("movti_bigconst"): New pattern definition.
29128 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29131 * s390-protos.h (s390_expand_cs_hqi): Removed.
29132 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
29133 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
29134 modes as well as CCZ1mode and CCZmode.
29135 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
29136 signature of s390_emit_compare_and_swap.
29137 (s390_expand_cs_hqi): Likewise, make static.
29138 (s390_expand_cs_tdsi): Generate an explicit compare before trying
29139 compare-and-swap, in some cases.
29140 (s390_expand_cs): Wrapper function.
29141 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
29143 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
29144 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
29145 patterns for small and large integers. Forbid symref memory operands.
29146 Move expander to s390.c. Require cc register.
29147 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
29148 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
29149 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
29150 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
29151 symref memory operands. Remove CC mode and call s390_match_ccmode
29153 ("atomic_exchange<mode>"): Allow and implement all integer modes.
29155 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29157 * config/s390/s390.md (define_peephole2): New peephole to help
29158 combining the load-and-test pattern with volatile memory.
29160 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29162 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
29163 with CCZmode for TARGET_Z196.
29165 2017-04-25 Jakub Jelinek <jakub@redhat.com>
29167 PR rtl-optimization/80501
29168 * combine.c (make_compound_operation_int): Set subreg_code to SET
29169 even for AND with mask of the sign bit of mode.
29171 PR rtl-optimization/80500
29172 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
29173 sum's initial value.
29175 2017-04-25 Julian Brown <julian@codesourcery.com>
29176 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29178 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
29180 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
29182 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
29184 2017-04-25 Julian Brown <julian@codesourcery.com>
29185 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29187 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
29188 (thunderx2t99_sha): New Reservation.
29190 2017-04-25 Julian Brown <julian@codesourcery.com>
29191 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29193 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
29194 type for 1-element load.
29196 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
29198 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
29200 2017-04-24 Martin Jambor <mjambor@suse.cz>
29202 PR tree-optimization/80293
29203 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
29204 char arrays not totally scalarizable if it is false.
29205 (analyze_all_variable_accesses): Pass correct value in the new
29206 parameter. Add a statistics counter.
29208 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
29210 PR middle-end/79931
29211 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
29213 2017-04-24 Richard Biener <rguenther@suse.de>
29215 PR tree-optimization/80494
29216 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
29217 out for complex types.
29219 2017-04-24 Richard Biener <rguenther@suse.de>
29221 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
29222 * tree-ssa-sccvn.c (print_scc): Print SCC size.
29223 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
29224 (DFS): Adjust and never fail.
29225 (sccvn_dom_walker::fail): Remove.
29226 (sccvn_dom_walker::before_dom_children): Adjust.
29227 (run_scc_vn): Likewise and never fail.
29228 * tree-ssa-pre.c (pass_pre::execute): Adjust.
29229 (pass_fre::execute): Likewise.
29231 2017-04-24 Richard Biener <rguenther@suse.de>
29233 PR tree-optimization/79725
29234 * tree-ssa-sink.c (statement_sink_location): Return whether
29235 failure reason was zero uses. Move that check later.
29236 (sink_code_in_bb): Deal with zero uses by removing the stmt
29239 2017-04-24 Richard Biener <rguenther@suse.de>
29242 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
29243 pointer-based references.
29245 2017-04-24 Richard Biener <rguenther@suse.de>
29248 * pass_manager.h (pass_manager::operator new): Remove.
29249 (pass_manager::operator delete): Likewise.
29250 * passes.c (pass_manager::operator new): Remove.
29251 (pass_manager::operator delete): Likewise.
29252 (pass_manager::pass_manager): Zero individual pass members.
29254 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
29257 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
29258 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
29259 Check "XEXP (src, 1)" operand here.
29260 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
29261 Check "XEXP (src, 1)" operand here.
29262 (dimode_scalar_chain::make_vector_copies): Detect count register
29263 of a shift instruction. Zero extend count register from QImode
29264 to DImode to satisfy vector shift pattern count operand predicate.
29265 Substitute vector shift count operand with a DImode copy.
29266 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
29269 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
29271 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
29272 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29273 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
29274 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29275 (UNSPEC_NOREX_MEM): Remove definition.
29277 2017-04-21 Richard Biener <rguenther@suse.de>
29279 PR tree-optimization/79547
29280 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
29281 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
29282 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
29283 without any constraints.
29285 2017-04-21 Richard Biener <rguenther@suse.de>
29287 PR tree-optimization/78847
29288 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
29290 2017-04-21 Richard Biener <rguenther@suse.de>
29292 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
29293 (build_distinct_type_copy): Likewise.
29294 (build_variant_type_copy): Likewise.
29295 * tree.c (build_qualified_type): Pass down mem-stat info.
29296 (build_distinct_type_copy): Likewise.
29297 (build_variant_type_copy): Likewise.
29299 2017-04-21 Richard Biener <rguenther@suse.de>
29301 PR tree-optimization/80237
29302 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
29304 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
29305 for a simplified result.
29307 2016-04-21 Richard Biener <rguenther@suse.de>
29309 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
29310 sth as strict as a simple_iv but a chrec without symbols and an
29311 operand defined in the loop we are peeling (and not some subloop).
29312 (propagate_constants_for_unrolling): Propagate all constants.
29314 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29317 * config/i386/i386.c (print_reg): Remove assert for disalowed
29318 regno values, call output_operand_lossage instead.
29320 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29323 * config/i386/constraints.md (Yc): New register constraint.
29324 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
29325 Use Yc constraint for alternative 2 of operand 0. Remove
29326 preferred_for_speed attribute.
29328 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29330 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
29331 lastprivate clauses in SIMT case.
29333 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
29335 * doc/invoke.texi (-Wextra-semi): Document new warning option.
29337 2017-04-20 Richard Biener <rguenther@suse.de>
29339 PR tree-optimization/57796
29340 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
29341 as N scalar stores.
29342 (vect_model_load_cost): Cost gathers as N scalar loads.
29344 2017-04-20 Richard Biener <rguenther@suse.de>
29346 * ggc-page.c (ggc_allocated_p): Rename to ...
29347 (safe_lookup_page_table_entry): ... this and return the lookup
29349 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
29351 2017-04-20 Richard Biener <rguenther@suse.de>
29353 PR tree-optimization/80453
29354 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
29355 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
29356 from the conditions.
29357 (vn_phi_eq): Pass them down.
29358 (vn_phi_lookup): Record them.
29359 (vn_phi_insert): Likewise.
29361 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
29363 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
29364 uninitialized variable warning to avoid buffer overrun.
29366 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29369 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
29370 is suppressed for '{ 0 }' in C.
29372 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29374 * BASE-VER: Set to 8.0.0.
29376 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
29378 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
29379 priority .init_array and .fini_array section with SECTION_NOTYPE
29382 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29384 PR middle-end/80423
29385 * tree.h (build_array_type): Add typeless_storage default argument.
29386 * tree.c (type_cache_hasher::equal): Also compare
29387 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
29388 (build_array_type): Add typeless_storage argument, set
29389 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
29391 (build_nonshared_array_type): Adjust build_array_type_1 caller.
29392 (build_array_type): Likewise. Add typeless_storage argument.
29394 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29395 Jakub Jelinek <jakub@redhat.com>
29397 PR tree-optimization/80426
29398 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
29399 operation on symbolic operands, also compute the overflow for the
29400 invariant part when the operation degenerates into a negation.
29402 2017-04-19 Jakub Jelinek <jakub@redhat.com>
29405 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
29406 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
29409 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
29411 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
29414 * config/avr/avr.c (tree.h): Include it.
29415 (cgraph.h): Include it.
29416 (avr_encode_section_info): Don't warn for uninitialized progmem
29417 variable if it's just an alias.
29419 2017-04-19 Richard Biener <rguenther@suse.de>
29422 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
29423 when needed by AutoPGO.
29425 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
29428 * doc/lto.texi: Remove an extra 'that'.
29430 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
29432 PR rtl-optimization/80429
29433 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
29434 are only used in debug insns.
29436 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29437 Vladimir Makarov <vmakarov@redhat.com>
29439 * config/sparc/predicates.md (input_operand): Add comment. Return
29440 true for any memory operand when LRA is in progress.
29441 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
29443 2017-04-18 Jeff Law <law@redhat.com>
29446 * mips.md ({return,simple_return}_internal): Do not overwrite
29449 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29451 PR tree-optimization/80443
29452 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
29453 instead of adding 1, subtract -1 and similarly instead of subtracting
29456 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
29458 PR rtl-optimization/80357
29459 * haifa-sched.c (tmp_bitmap): New variable.
29460 (model_recompute): Handle duplicate use records.
29461 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
29462 (free_global_sched_pressure_data): Free it.
29464 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
29467 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29468 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29469 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29470 instead of SYSTEM_HEADER_DIR.
29472 2017-04-18 Jeff Law <law@redhat.com>
29474 PR middle-end/80422
29475 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
29476 predecessors after walking up the insn chain.
29478 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29481 * dwarf2out.c (modified_type_die): Try harder not to emit internal
29482 sizetype type into debug info.
29484 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29487 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
29488 unneeded test for TARGET_UPPER_REGS_SF.
29489 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
29491 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29494 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
29495 instead of gsi_after_labels.
29497 2017-04-18 Jeff Law <law@redhat.com>
29499 * regcprop.c (maybe_mode_change): Avoid creating copies of the
29503 2017-04-13 Jeff Law <law@redhat.com>
29504 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
29505 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
29507 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
29510 * config/avr/avr.c (intl.h): Include it.
29511 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
29513 2017-04-18 Martin Liska <mliska@suse.cz>
29515 PR gcov-profile/78783
29516 * gcov-tool.c (gcov_output_files): Validate that destination
29517 file is either removed by the tool or by a user.
29519 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
29520 Guy Benyei <guybe@mellanox.com>
29522 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
29523 block, and do not negate it, the stored id is already negative.
29525 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
29527 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
29529 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29532 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
29533 masks of options that should be turned off if the VSX vector
29534 options are turned off.
29535 (OTHER_P8_VECTOR_MASKS): Likewise.
29536 (OTHER_VSX_VECTOR_MASKS): Likewise.
29537 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
29538 rs6000_disable_incompatible_switches to validate no type switches
29540 (rs6000_incompatible_switch): New function to disallow turning on
29541 other vector options if -mno-vsx, -mno-power8-vector, or
29542 -mno-power9-vector are specified.
29544 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29546 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
29548 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29550 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
29551 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
29552 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
29553 (ARG_POINTER_CFA_OFFSET): Likewise.
29555 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29557 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
29558 conditions to take advantage of various optimizations.
29560 2017-04-13 Jeff Law <law@redhat.com>
29562 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
29563 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
29564 (zero_extendsidi2_dext): Likewise.
29566 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29569 * fold-const.c (fold_ternary_loc): Revert
29570 use op0 instead of fold_convert_loc (loc, type, arg0) part of
29573 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
29575 PR rtl-optimization/80343
29576 * lra-remat.c (update_scratch_ops): Assign original hard reg to
29577 new scratch pseudo.
29579 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
29582 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
29583 to ubsan_encode_value.
29585 2017-04-13 Jeff Law <law@redhat.com>
29587 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
29588 appearing in DEBUG_INSNs.
29590 2017-04-13 Martin Liska <mliska@suse.cz>
29592 PR gcov-profile/80413
29593 * gcov-io.c (gcov_write_string): Copy to buffer just when
29594 allocated size is greater than zero.
29596 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29599 * dwarf2out.c (decls_for_scope): Ignore declarations of
29600 current_function_decl in BLOCK_NONLOCALIZED_VARS.
29602 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
29605 * ipa-visibility.c (non_local_p): Fix typos.
29606 (localize_node): When localizing symbol in same comdat group,
29607 dissolve the group only when we know external symbols are going
29609 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
29611 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29613 PR tree-optimization/79390
29614 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
29615 order does not result in usable sequence, retry with reversed operand
29621 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
29622 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
29623 op0 instead of fold_convert_loc (loc, type, arg0).
29625 2017-04-12 Jeff Law <law@redhat.com>
29627 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
29628 has a delay slot in the generated code.
29630 * config/cris/cris.md (cris_preferred_reload_class): Return
29631 GENNONACR_REGS rather than GENERAL_REGS.
29633 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29636 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
29637 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
29638 signedness of the result type.
29640 2017-04-12 Richard Biener <rguenther@suse.de>
29641 Jeff Law <law@redhat.com>
29643 PR tree-optimization/80359
29644 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
29645 trim stores to TARGET_MEM_REFs.
29647 2017-04-12 Richard Biener <rguenther@suse.de>
29649 PR tree-optimization/79390
29650 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
29651 threading case even more.
29653 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
29656 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
29657 for quad_address_p for TImode, instead of just not indexed_address.
29659 2017-04-12 Richard Biener <rguenther@suse.de>
29660 Bernd Edlinger <bernd.edlinger@hotmail.de>
29662 PR middle-end/79671
29663 * alias.c (component_uses_parent_alias_set_from): Handle
29664 TYPE_TYPELESS_STORAGE.
29665 (get_alias_set): Likewise.
29666 * tree-core.h (tree_type_common): Add typeless_storage flag.
29667 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
29668 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
29669 for types containing members with TYPE_TYPELESS_STORAGE.
29670 (place_field): Likewise.
29671 (layout_type): Likewise for ARRAY_TYPE.
29672 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
29673 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
29674 TYPE_TYPELESS_STORAGE.
29675 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
29677 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29680 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29681 first argument to type.
29683 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29687 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
29688 CONST0_RTX (mode) rather than const0_rtx where appropriate.
29689 (rs6000_expand_binop_builtin): Likewise.
29690 (rs6000_expand_ternop_builtin): Likewise; also add missing
29691 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
29692 vshasigma built-ins.
29693 * doc/extend.texi: Document that vec_xxpermdi's third argument
29694 must be a constant.
29696 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
29698 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
29699 Use shift_const cost parameter when calculating gain of STV shifts.
29701 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
29703 PR rtl-optimization/70478
29704 * lra-constraints.c (process_alt_operands): Check memory for
29705 disfavoring memory insn operand.
29707 2017-04-11 Jakub Jelinek <jakub@redhat.com>
29709 PR middle-end/80100
29710 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
29711 left shift in unsigned HOST_WIDE_INT type.
29713 PR rtl-optimization/80385
29714 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
29715 (not (neg X)) into (plus X -1) for complex or non-integral modes.
29718 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
29719 if they have any depend clauses.
29721 2017-04-11 Martin Liska <mliska@suse.cz>
29724 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
29725 * ipa-split.c (split_function): Create a local comdat symbol
29726 if caller is in a comdat group.
29728 2017-04-11 Martin Liska <mliska@suse.cz>
29731 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
29734 2017-04-11 Martin Sebor <msebor@redhat.com>
29736 PR middle-end/80364
29737 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
29738 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
29740 (directive::set_width, directive::set_precision, format_character):
29742 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
29745 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
29748 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
29749 conflict, set target->arch_name instead of target->cpu_name.
29751 2017-04-11 Richard Biener <rguenther@suse.de>
29753 PR tree-optimization/80374
29754 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
29755 build_zero_cst, remove fold_convertible_p check again.
29757 2017-04-11 Martin Liska <mliska@suse.cz>
29760 * ubsan.c (instrument_object_size): Do not instrument register
29763 2017-04-11 Jakub Jelinek <jakub@redhat.com>
29766 * config/i386/i386-builtin-types.def
29767 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
29768 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
29769 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
29770 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
29771 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
29772 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
29773 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
29774 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
29775 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
29776 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
29777 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
29778 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
29779 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
29780 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
29781 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
29782 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
29783 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
29784 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
29785 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
29786 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
29787 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
29788 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
29789 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
29790 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
29791 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
29792 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
29793 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
29794 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
29795 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
29796 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
29797 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
29798 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
29799 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
29800 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
29801 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
29802 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
29803 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
29804 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
29805 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
29806 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
29807 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
29808 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
29809 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
29810 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
29811 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
29812 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
29814 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
29815 flag to second_arg_count, handle 4 argument function type _COUNT
29816 aliases, handle second_arg_count on second argument rather than last.
29818 2017-04-10 Jeff Law <law@redhat.com>
29820 PR tree-optimization/80374
29821 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
29822 record anything if we can not convert integer_zero_node to the
29825 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29828 * config/rs6000/rs6000.c (rs6000_option_override_internal):
29829 Enhance special handling given to the TARGET_P9_MINMAX option in
29830 relation to certain other options.
29832 2017-04-10 Bin Cheng <bin.cheng@arm.com>
29834 PR tree-optimization/80153
29835 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
29836 remove POINTER_PLUS_EXPR's base part directly, rather than through
29839 2017-04-10 Richard Biener <rguenther@suse.de>
29840 Bin Cheng <bin.cheng@arm.com>
29842 PR tree-optimization/80153
29843 * tree-affine.c (aff_combination_to_tree): Get base pointer from
29844 the first element of pointer type aff_tree. Build result expr in
29846 (add_elt_to_tree): Convert to type unconditionally. Remove other
29847 fold_convert calls.
29848 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
29849 (rewrite_use_nonlinear_expr): Check invariant using iv information.
29851 2017-04-10 Richard Biener <rguenther@suse.de>
29853 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29856 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
29858 PR rtl-optimization/70478
29859 * lra-constraints.c (curr_small_class_check): New.
29860 (update_and_check_small_class_inputs): New.
29861 (process_alt_operands): Update curr_small_class_check. Disfavor
29862 alternative insn memory operands. Check available regs for small
29865 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
29868 * config/mips/mips.opt (-mvirt): Update description.
29869 * doc/invoke.texi (-mvirt): Likewise.
29871 2017-04-10 Richard Biener <rguenther@suse.de>
29873 PR middle-end/80362
29874 * fold-const.c (fold_binary_loc): Look at unstripped ops when
29875 looking for NEGATE_EXPR in -A / -B to A / B folding.
29877 2017-04-10 Martin Liska <mliska@suse.cz>
29879 PR gcov-profile/80224
29880 * gcov.c (print_usage): Fix usage string.
29881 (get_gcov_intermediate_filename): Remove.
29882 (output_gcov_file): Use both for normal and intermediate format.
29883 (generate_results): Do not initialize special file for
29884 intermediate format.
29886 2017-04-10 Richard Biener <rguenther@suse.de>
29888 PR tree-optimization/80304
29889 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
29892 2017-04-10 Nathan Sidwell <nathan@acm.org>
29895 * config/rs6000/rs6000.c (rs6000_vector_type): New.
29896 (rs6000_init_builtins): Use it.
29898 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29900 * config/arm/arm.md (<mrc>): Add mode to SET source.
29901 (<mrrc>): Likewise.
29903 2017-04-10 Richard Biener <rguenther@suse.de>
29905 PR middle-end/80344
29906 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
29908 2017-04-10 Jakub Jelinek <jakub@redhat.com>
29911 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
29912 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
29913 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
29914 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
29915 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
29916 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
29917 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
29918 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
29919 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
29920 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
29921 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
29922 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
29923 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
29924 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
29925 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
29926 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
29927 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
29928 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
29929 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
29930 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
29931 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
29932 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
29933 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
29935 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
29937 PR rtl-optimization/70478
29938 * lra-constraints.c: Reverse the last patch.
29940 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
29942 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
29943 Add comment for WCHAR_T.
29945 2017-04-08 Martin Liska <mliska@suse.cz>
29948 2017-04-07 Martin Liska <mliska@suse.cz>
29951 * ipa-split.c (split_function): Add function part to a same comdat
29954 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29957 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
29959 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
29961 * rs6000/rs6000.c (vec_load_pendulum): Rename...
29962 (vec_pairing): ...to this.
29963 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
29964 (rs6000_sched_init): Adjust for name change.
29965 (struct rs6000_sched_context): Likewise.
29966 (rs6000_init_sched_context): Likewise.
29967 (rs6000_set_sched_context): Likewise.
29969 2017-04-07 Jakub Jelinek <jakub@redhat.com>
29975 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
29977 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
29978 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
29979 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
29981 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
29983 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
29985 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
29987 PR rtl-optimization/70703
29988 * ira-color.c (update_conflict_hard_regno_costs): Use
29989 int64_t instead of HOST_WIDE_INT.
29991 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
29993 PR rtl-optimization/70478
29994 * lra-constraints.c (process_alt_operands): Disfavor alternative
29995 insn memory operands.
29997 2017-04-07 Jeff Law <law@redhat.com>
29999 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
30000 CALL and NOTE_INSN_CALL_ARG_LOCATION.
30002 2017-04-07 Martin Liska <mliska@suse.cz>
30005 * config/aarch64/aarch64.c (aarch64_process_target_attr):
30006 Show error message instead of an ICE.
30008 2017-04-07 Martin Liska <mliska@suse.cz>
30011 * ipa-split.c (split_function): Add function part to a same comdat
30014 2017-04-07 Richard Biener <rguenther@suse.de>
30016 PR middle-end/80341
30017 * tree.c (get_unwidened): Also handle ! for_type case for
30019 * convert.c (do_narrow): Split out from ...
30020 (convert_to_integer_1): ... here. Do not pass final truncation
30021 type to get_unwidened for TRUNC_DIV_EXPR.
30023 2017-04-07 Richard Biener <rguenther@suse.de>
30025 * tree-affine.c (wide_int_ext_for_comb): Take type rather
30027 (aff_combination_const): Adjust.
30028 (aff_combination_scale): Likewise.
30029 (aff_combination_add_elt): Likewise.
30030 (aff_combination_add_cst): Likewise.
30031 (aff_combination_convert): Likewise.
30032 (add_elt_to_tree): Likewise. Remove unused argument.
30033 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
30035 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
30037 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
30039 * config/arm/arm.c (arm_default_short_enums): Use
30040 ARM_DEFAULT_SHORT_ENUMS.
30041 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
30043 2017-04-06 Jakub Jelinek <jakub@redhat.com>
30046 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
30047 members with redundant out-of-class redeclaration.
30049 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
30052 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
30053 * config/i386/i386.md (*zero_extendsidi2):
30054 Add (?*x,*x) and (?*v,*v) alternatives.
30056 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
30059 * config/i386/i386.c (ix86_expand_builtin)
30060 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
30061 mode from insn data. Convert operands to insn operand mode.
30062 Copy operands that don't satisfy insn predicate to a register.
30064 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
30066 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
30069 2017-04-06 Richard Biener <rguenther@suse.de>
30071 PR tree-optimization/80334
30072 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
30073 preserve alignment of accesses.
30075 2017-04-06 Richard Biener <rguenther@suse.de>
30077 PR tree-optimization/80262
30078 * tree-sra.c (build_ref_for_offset): Preserve address-space
30080 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
30081 Drop useless address-space information on MEM_REF offsets.
30083 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
30085 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
30087 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
30089 PR rtl-optimization/70703
30090 * ira-color.c (update_conflict_hard_regno_costs): Use
30091 HOST_WIDE_INT instead of long.
30093 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
30096 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
30097 not defined for x86_64 target. Add -mmmx target option when __SSE2__
30099 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
30100 for x86_64 target. Handle -m3dnowa option.
30102 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
30104 PR rtl-optimization/70703
30105 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
30106 (update_conflict_hard_regno_costs): Use long instead of unsigned
30107 arithmetic for cost calculation.
30109 2017-04-05 Jakub Jelinek <jakub@redhat.com>
30110 Bernd Edlinger <bernd.edlinger@hotmail.de>
30113 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
30116 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
30119 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
30120 ptr_mode with Pmode throughout.
30121 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
30122 into probe_stack_range and use DImode.
30124 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
30127 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
30128 call_eh_return is true.
30130 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30132 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
30133 Initialize last_match_fntype_index.
30135 2017-04-05 Jakub Jelinek <jakub@redhat.com>
30138 * tree-nvr.c: Include internal-fn.h.
30139 (pass_return_slot::execute): Ignore internal calls without
30142 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30143 Richard Biener <rguenther@suse.de>
30146 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
30147 captures used multiple times, except for the last use.
30148 * generic-match-head.c: Include gimplify.h.
30150 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30152 PR tree-optimization/79390
30153 * target.h (struct noce_if_info): Declare.
30154 * targhooks.h (default_noce_conversion_profitable_p): Declare.
30155 * target.def (noce_conversion_profitable_p): New target hook.
30156 * ifcvt.h (struct noce_if_info): New type, moved from ...
30157 * ifcvt.c (struct noce_if_info): ... here.
30158 (noce_conversion_profitable_p): Renamed to ...
30159 (default_noce_conversion_profitable_p): ... this. No longer
30161 (noce_try_store_flag_constants, noce_try_addcc,
30162 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
30163 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
30164 instead of noce_conversion_profitable_p.
30165 * config/i386/i386.c: Include ifcvt.h.
30166 (ix86_option_override_internal): Don't override
30167 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
30168 (ix86_noce_conversion_profitable_p): New function.
30169 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
30170 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
30171 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
30172 * doc/tm.texi: Regenerated.
30174 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30176 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
30179 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
30182 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
30183 instructions for small multiply cores.
30185 2017-04-04 Jeff Law <law@redhat.com>
30187 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
30189 (mips_expand_vec_perm_const): Initialize elements in orig_perm
30190 that are not set by the loop over the elements.
30192 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30195 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
30196 int mode, convert_modes it to mode as unsigned, otherwise use
30197 lowpart_subreg to mode rather than SImode.
30198 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
30199 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
30200 Use DImode instead of SImode for the shift count operand.
30201 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
30204 2017-04-04 Richard Biener <rguenther@suse.de>
30206 PR middle-end/80281
30207 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
30208 arithmetic done for the negate or the plus. Simplify.
30209 (A - (-B) -> A + B): Likewise.
30210 * fold-const.c (split_tree): Make sure to not negate pointers.
30212 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
30214 PR rtl-optimization/60818
30215 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
30216 a compare of comparisons with the thing compared if this results
30217 in a different machine mode.
30219 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
30221 * alias.c (base_alias_check): Fix typo in comment.
30222 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
30223 * cgraphunit.c (symbol_table::compile): Likewise.
30224 * collect2.c (maybe_run_lto_and_relink): Likewise.
30225 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
30226 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
30227 * config/avr/avr.c (avr_map_op_t): Likewise.
30228 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
30229 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
30230 * config/epiphany/epiphany.md (movcc): Likewise.
30231 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
30232 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
30234 * config/mips/mips.c (mips_save_restore_reg): Likewise.
30235 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
30236 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
30237 * config/sh/sh.c (sh_rtx_costs): Likewise.
30238 * fold-const.c (fold_truth_andor): Likewise.
30239 * genautomata.c (collapse_flag): Likewise.
30240 * gengtype.h (struct type::u::s): Likewise.
30241 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
30242 * input.c (FORMAT_AMOUNT): Likewise.
30243 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
30244 (known_aggs_to_agg_replacement_list): Likewise.
30245 * ipa-inline-analysis.c: Likewise.
30246 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
30247 * ipa-polymorphic-call.c
30248 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
30249 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
30250 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
30252 * modulo-sched.c (apply_reg_moves): Likewise.
30253 * omp-expand.c (build_omp_regions_1): Likewise.
30254 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
30255 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
30256 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
30257 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
30258 * value-prof.c: Likewise.
30259 * var-tracking.c (val_reset): Likewise.
30261 2017-04-03 Richard Biener <rguenther@suse.de>
30263 PR tree-optimization/80275
30264 * fold-const.c (split_address_to_core_and_offset): Handle
30267 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
30269 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
30270 descriptors is at least equal to that of functions.
30272 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30274 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
30276 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30279 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
30280 (mov<IMOD4:mode>): New expander.
30281 (*mov<IMOD4:mode>_internal): New insn and split pattern.
30283 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
30285 PR rtl-optimization/79405
30286 * fwprop.c (propagations_left): New variable.
30287 (forward_propagate_into): Decrement it.
30288 (fwprop_init): Initialize it.
30289 (fw_prop): If the variable has reached zero, stop propagating.
30290 (fwprop_addr): Ditto.
30292 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30295 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
30296 a FUNCTION_DECL, pass it as decl instead of origin to
30299 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
30301 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
30304 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
30307 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
30308 TARGET_VSX_SMALL_INTEGER.
30310 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30312 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30313 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
30315 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30317 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
30318 extraction from odd-numbered MSA register.
30320 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30322 PR middle-end/80173
30323 * expmed.c (store_bit_field_1): Don't attempt to create
30324 a word subreg out of hard registers wider than word if they
30325 have HARD_REGNO_NREGS of 1 for their mode.
30327 PR middle-end/80163
30328 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
30329 conversions to integer types wider than word and pointer.
30332 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
30333 (rtx_equal_for_cselib_p): Pass 0 to it.
30334 * cselib.c (cselib_hasher::equal): Likewise.
30335 (rtx_equal_for_cselib_1): Add depth argument. If depth
30336 is 128, don't look up VALUE locs and punt. Increment
30337 depth in recursive calls when walking VALUE locs.
30339 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
30341 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
30342 (make_gcov_file_name): Use the canonical path name for generating
30344 (read_line): Fix handling of files with ascii null bytes.
30346 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
30348 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
30349 to initialise a vector register instead
30350 of using a const_int.
30352 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30354 PR translation/80189
30355 * gimplify.c (omp_default_clause): Use %qs instead of %s in
30356 diagnostic messages.
30358 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
30361 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
30362 (dfp_diex_<mode>): Update mode of operand 1.
30363 * doc/extend.texi (dxex, dxexq): Document change to return type.
30364 (diex, diexq): Document change to argument type.
30366 2017-03-30 Martin Jambor <mjambor@suse.cz>
30369 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
30370 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
30371 it reflects the signature changes performed at the callee side.
30372 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
30373 to cgraph_build_function_type_skip_args.
30374 (build_function_decl_skip_args): Adjust call to the above function.
30376 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30379 * config/i386/sse.md
30380 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
30381 register as dest whenever it is a MEM not rtx_equal_p to the
30382 corresponding dup operand, and when forcing into reg move the
30383 reg into the memory afterwards.
30384 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
30385 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
30386 for the force_reg mode.
30387 (avx512vl_vextractf128<mode>): Use register as dest either
30388 always when a MEM, or when it is a MEM not rtx_equal_p to the
30389 corresponding dup operand, or even not when it is a CONST_VECTOR
30390 depending on the mode and lo vs. hi.
30391 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
30393 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
30394 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
30395 Likewise. Require that operands[2] is even.
30396 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
30397 Remove extraneous parens. Require that operands[2] is a multiple
30399 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
30400 operands[0] is a MEM if <mask_applied>, the predicates/constraints
30401 disallow memory then.
30403 2017-03-30 Richard Biener <rguenther@suse.de>
30405 PR tree-optimization/77498
30406 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
30407 to non-constants over backedges.
30409 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
30411 PR rtl-optimization/80233
30412 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
30413 as last_combined_insn. Do not test for BARRIER_P separately.
30415 2017-03-29 Andreas Schwab <schwab@suse.de>
30418 * calls.c (prepare_call_address): Convert funexp to Pmode before
30419 copying to temp reg.
30421 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30423 PR tree-optimization/80158
30424 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
30425 Handle possible future case of more than one alternate
30427 (replace_rhs_if_not_dup): Likewise.
30428 (replace_one_candidate): Likewise.
30430 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
30432 PR rtl-optimization/80193
30433 * ira.c (ira): Do not check allocation for LRA.
30435 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
30437 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
30438 (nvptx_output_simt_exit): Declare.
30439 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
30440 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
30441 (init_softstack_frame): Move initialization of crtl->is_leaf to...
30442 (nvptx_declare_function_name): ...here. Emit declaration of local
30443 memory space buffer for omp_simt_enter insn.
30444 (nvptx_output_unisimt_switch): New.
30445 (nvptx_output_softstack_switch): New.
30446 (nvptx_output_simt_enter): New.
30447 (nvptx_output_simt_exit): New.
30448 * config/nvptx/nvptx.h (struct machine_function): New fields
30449 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
30450 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
30451 (UNSPECV_SIMT_EXIT): Ditto.
30452 (omp_simt_enter_insn): New insn.
30453 (omp_simt_enter): New expansion.
30454 (omp_simt_exit): New insn.
30455 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
30457 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
30458 (expand_GOMP_SIMT_ENTER_ALLOC): New.
30459 (expand_GOMP_SIMT_EXIT): New.
30460 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
30461 (GOMP_SIMT_ENTER_ALLOC): Ditto.
30462 (GOMP_SIMT_EXIT): Ditto.
30463 * target-insns.def (omp_simt_enter): New insn.
30464 (omp_simt_exit): Ditto.
30465 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
30467 (lower_rec_simd_input_clauses): Implement SIMT privatization.
30468 (lower_rec_input_clauses): Likewise.
30469 (lower_lastprivate_clauses): Handle SIMT privatization.
30471 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
30472 (ompdevlow_adjust_simt_enter): New.
30473 (find_simtpriv_var_op): New.
30474 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
30475 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
30477 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
30478 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
30479 (copy_decl_for_dup_finish): Ditto.
30481 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
30483 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
30486 * config/i386/i386.c (ix86_option_override_internal): Always
30487 allow -mpreferred-stack-boundary=3 for 64-bit targets.
30489 2017-03-28 Bin Cheng <bin.cheng@arm.com>
30491 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
30493 2017-03-28 Bin Cheng <bin.cheng@arm.com>
30495 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
30496 mark new edge's irreducible flag accordign to it.
30497 (vect_do_peeling): Check loop preheader edge's irreducible flag
30498 and pass it to function slpeel_add_loop_guard.
30500 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
30502 PR tree-optimization/80218
30503 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30504 Update block frequencies and counts.
30506 2017-03-28 Richard Biener <rguenther@suse.de>
30508 PR tree-optimization/78644
30509 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
30510 of a simplification result we may not use it at all.
30512 2017-03-28 Richard Biener <rguenther@suse.de>
30515 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
30516 without arguments, generate default definition of a SSA name.
30518 2017-03-28 Richard Biener <rguenther@suse.de>
30520 PR middle-end/80222
30521 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
30522 TYPE_REF_CAN_ALIAS_ALL references.
30523 * fold-const.c (fold_indirect_ref_1): Likewise.
30525 2017-03-28 Martin Liska <mliska@suse.cz>
30528 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
30529 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
30531 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
30532 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
30534 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
30535 (EXTRA_SPECS): Define.
30536 (SUBTARGET_EXTRA_SPECS): Likewise.
30537 (SUBTARGET_CPP_SPEC): Likewise.
30538 * config/arc/elf.h (EXTRA_SPECS): Renamed to
30539 SUBTARGET_EXTRA_SPECS.
30540 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
30542 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
30544 * config/arc/simdext.md (vst64_insn): Update pattern.
30545 (vld32wh_insn): Likewise.
30546 (vld32wl_insn): Likewise.
30547 (vld64_insn): Likewise.
30548 (vld32_insn): Likewise.
30550 2017-03-28 Marek Polacek <polacek@redhat.com>
30553 * fold-const.c (fold_comparison): Use protected_set_expr_location
30554 instead of SET_EXPR_LOCATION.
30556 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
30558 * tree.c (add_expr): Avoid name lookup warning.
30560 2017-03-27 Jeff Law <law@redhat.com>
30562 PR tree-optimization/80216
30563 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
30564 function name. Limit recursion depth.
30565 (record_temporary_equivalences): Corresponding changes.
30567 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
30569 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
30572 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30575 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
30577 * cfgcleanup.c (reg_note_cfa_p): New array.
30578 (insns_have_identical_cfa_notes): New function.
30579 (old_insns_match_p): Don't cross-jump in between /f
30580 and non-/f instructions. If both i1 and i2 are frame related,
30581 verify all CFA notes, their order and content.
30583 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
30586 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
30587 HImode and SImode with zero extend to DImode to one insn.
30588 (bswap<mode>2_extenddi): Likewise.
30589 (bswapsi2_extenddi): Likewise.
30590 (bswaphi2_extendsi): Likewise.
30591 (bswaphi2): Combine bswap HImode and SImode into one insn.
30592 Separate memory insns from swapping register.
30593 (bswapsi2): Likewise.
30594 (bswap<mode>2): Likewise.
30595 (bswaphi2_internal): Delete, no longer used.
30596 (bswapsi2_internal): Likewise.
30597 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
30598 store, and gpr<-gpr swap insns.
30599 (bswap<mode>2_store): Likewise.
30600 (bswaphi2_reg): Register only splitter, combine with the splitter.
30601 (bswaphi2 splitter): Likewise.
30602 (bswapsi2_reg): Likewise.
30603 (bswapsi2 splitter): Likewise.
30604 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
30605 the insns into load, store, and register/register insns.
30606 (bswapdi2_ldbrx): Likewise.
30607 (bswapdi2_load): Likewise.
30608 (bswapdi2_store): Likewise.
30609 (bswapdi2_reg): Likewise.
30611 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
30613 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
30614 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
30616 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
30619 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
30621 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30622 special handling for target option conflicts between dform
30623 options (-mpower9-dform, -mpower9-dform-vector,
30624 -mpower9-dform-scalar) and -mno-direct-move.
30626 2017-03-27 Richard Biener <rguenther@suse.de>
30628 PR tree-optimization/80181
30629 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
30631 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30633 * config/arc/predicates.md (move_double_src_operand): Replace the
30634 call to move_double_src_operand with a call to address_operand.
30636 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30638 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
30639 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
30640 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
30642 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30644 * config/arc/predicates.md (long_immediate_loadstore_operand):
30645 Consider scaled addresses cases.
30647 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30649 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
30650 restored when in interrupt.
30651 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
30652 doesn't have delay slot.
30654 2017-03-27 Richard Biener <rguenther@suse.de>
30657 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
30658 inlined thunk clones.
30660 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30663 * asan.c (instrument_derefs): Copy over last operand from
30664 original COMPONENT_REF to the new COMPONENT_REF with
30665 DECL_BIT_FIELD_REPRESENTATIVE.
30666 * ubsan.c (instrument_object_size): Likewise.
30668 2017-03-27 Richard Biener <rguenther@suse.de>
30670 PR tree-optimization/80170
30671 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
30672 sure DR/SCEV didnt fold in constants we do not see when looking
30673 at the reference base alignment.
30675 2017-03-27 Richard Biener <rguenther@suse.de>
30677 PR middle-end/80171
30678 * gimple-fold.c (fold_ctor_reference): Properly guard against
30679 NULL return value from canonicalize_constructor_val.
30681 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
30684 * config/i386/i386.c (ix86_expand_builtin)
30685 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
30686 flags reg setting and flags reg using instructions.
30687 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
30688 clobbering instructions to zero extend op2.
30690 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
30692 * doc/install.texi (Configuration) <--with-aix-soname>:
30693 Update link to AIX ld.
30695 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
30697 PR rtl-optimization/80160
30698 PR rtl-optimization/80159
30699 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
30700 reg_alternate_class into account.
30702 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
30705 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
30706 to consider in curr_insn_transform.
30708 2017-03-24 Jakub Jelinek <jakub@redhat.com>
30710 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
30711 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
30712 and emit_mode_inner.
30714 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30716 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
30717 argument to the overloaded builtin variants. Use the new flag to
30718 deprecate certain builtin variants.
30719 * config/s390/s390-builtin-types.def: Add new builtin types.
30720 * config/s390/s390-builtins.h: Support new flags field for
30721 overloaded builtins.
30722 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
30723 (s390_macro_to_expand): Enable vector float data type.
30724 (s390_cpu_cpp_builtins_internal): Indicate support of the new
30725 builtins by incrementing the __VEC__ version number.
30726 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
30728 (s390_resolve_overloaded_builtin): Emit error messages depending
30729 on the builtin flags.
30730 * config/s390/s390.c (s390_expand_builtin): Support additional
30731 flags argument. Change error message to match the messages
30732 emitted in s390-c.c.
30733 * config/s390/s390.md: New UNSPEC_* constants.
30734 (op_type): Add new instruction types.
30735 * config/s390/vecintrin.h: Add new builtins and test data class
30737 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
30738 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
30739 (VEC_INEXACT, VEC_NOINEXACT): New constants.
30740 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
30741 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
30742 ("vec_mergel<mode>"): V_HW -> VEC_HW.
30744 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
30745 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
30746 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
30747 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
30749 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
30750 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
30751 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
30752 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
30754 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
30755 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
30756 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
30757 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
30758 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
30759 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
30760 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
30762 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
30763 ("vec_scatter_element<V_HW_4:mode>_DI")
30764 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
30765 ("vec_fpint<mode>", "vflls")
30766 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
30767 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
30768 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
30769 ("*vec_cmphe<mode>_cc"): ... these.
30771 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
30772 mode constant instead of magic value.
30774 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30776 * config/s390/s390.c (s390_expand_vec_compare): Support other
30777 vector floating point modes than just V2DF.
30778 (s390_expand_vcond): Likewise.
30779 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
30780 (s390_cannot_change_mode_class): Prevent mode changes between TF
30781 and V1TF in vector registers.
30782 * config/s390/s390.md (DF, SF): New mode attributes.
30783 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
30784 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
30785 SFmode support for VRs.
30786 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
30788 (VFT, VF_HW): New mode iterators.
30789 (vw, sdx): New mode attributes.
30790 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
30791 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
30792 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
30793 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
30794 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
30795 also the new vector floating point modes. Renaming to ...
30797 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
30798 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
30799 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
30800 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
30801 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
30802 ("vec_unordered<mode>"): ... these.
30804 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
30805 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
30806 ("*vec_extendv2df"): New insn definitions.
30808 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30810 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
30811 ("mulditi3_2", "*muldi3_sign"): New patterns.
30812 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
30813 rename the pattern definition.
30815 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30817 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
30819 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
30821 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30823 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
30824 instruction if possible.
30825 * config/s390/vector.md (vec_halfnumelts): New mode
30827 ("*vec_vllezlf<mode>"): New pattern.
30829 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30831 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
30832 ("popcountv4si2", "popcountv2di2"): Rename to ...
30833 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
30834 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
30836 ("popcount<mode>2_vxe"): New pattern.
30838 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30840 * common/config/s390/s390-common.c (processor_flags_table): Add
30842 * config.gcc: Add arch12.
30843 * config/s390/driver-native.c (s390_host_detect_local_cpu):
30844 Default to arch12 for unknown CPU model numbers.
30845 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
30846 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
30847 PROCESSOR_max sanity check.
30848 * config/s390/s390-opts.h (enum processor_type): Add
30850 * config/s390/s390.c (processor_table): Add arch12.
30851 (s390_expand_builtin): Add check for B_VXE flag.
30852 (s390_issue_rate): Add PROCESSOR_ARCH12.
30853 (s390_get_sched_attrmask): Likewise.
30854 (s390_get_unit_mask): Likewise.
30855 (s390_sched_score): Enable z13 scheduling for arch12.
30856 (s390_sched_reorder): Likewise.
30857 (s390_sched_variable_issue): Likewise.
30858 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
30860 (s390_tune_attr): Use z13 scheduling also for arch12.
30861 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
30862 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
30863 (TARGET_VXE_P): New macros.
30864 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
30865 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
30866 * config/s390/s390.opt: Add arch12 as processor_type.
30868 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30870 * config/s390/s390.md
30871 ("fixuns_truncdddi2", "fixuns_trunctddi2")
30872 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
30873 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
30875 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
30876 Rename expanders to ...
30878 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
30879 ("fixuns_truncdddi2_emu"): ... these.
30881 ("fixuns_trunc<mode>si2_emu"): New expander.
30883 ("*fixuns_truncdfdi2_z13"): Rename to ...
30884 ("*fixuns_truncdfdi2_vx"): ... this.
30886 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30888 * config/s390/2964.md: Remove the single element vector compare
30889 instructions which are no longer used.
30890 * config/s390/s390.c (s390_select_ccmode): Remove handling of
30892 (s390_canonicalize_comparison): Remove handling of DFmode
30894 (s390_expand_vec_compare_scalar): Remove function.
30895 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
30896 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
30898 ("*cmp<mode>_ccs"): Add wfcdb instruction.
30900 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30902 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
30904 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
30905 will anyway by matched by mov<mode>_64dfp.
30907 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30909 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
30910 vlef/vstef. Add missing operand to vleif.
30912 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30914 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
30915 pair for all vector types with 64 bit elements.
30916 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
30917 * config/s390/vector.md (V_HW_64): ... here.
30918 (V_128_NOSINGLE): New mode iterator.
30919 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
30920 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
30921 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
30922 ("*vec_load_pairv2di"): Change to ...
30923 ("*vec_load_pair<mode>"): ... this one.
30925 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30927 * config/s390/constraints.md: Add comments.
30928 (jKK): Reject element sizes > 8 bytes.
30929 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
30931 * config/s390/s390.md: Add the s_operand checks formerly in
30932 s390_split_ok_p to various splitters where they are still
30934 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
30935 for 128 bit vectors. Plus two splitters.
30937 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30939 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
30942 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30945 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
30946 error if the boundary argument is not constant.
30948 2017-03-24 Jakub Jelinek <jakub@redhat.com>
30950 PR rtl-optimization/80112
30951 * loop-doloop.c (doloop_condition_get): Don't check condition
30952 if cmp isn't SET with IF_THEN_ELSE src.
30954 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30956 PR tree-optimization/80158
30957 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
30958 replacing a candidate statement, also replace it for the
30959 candidate's alternate interpretation.
30960 (replace_rhs_if_not_dup): Likewise.
30961 (replace_one_candidate): Likewise.
30963 2017-03-24 Richard Biener <rguenther@suse.de>
30965 PR tree-optimization/80167
30966 * graphite-isl-ast-to-gimple.c
30967 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
30969 (translate_isl_ast_to_gimple::get_rename): Likewise.
30971 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
30973 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
30974 handling of certain combinations of target options, including the
30975 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
30976 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
30978 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30981 * config/arm/arm.md (*load_multiple): Add reload_completed to
30982 matching condition.
30984 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30985 Richard Biener <rguenth@suse.de>
30987 PR tree-optimization/79908
30988 PR tree-optimization/80136
30989 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
30990 been cast away, gimplify_and_add suffices.
30992 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
30994 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
30996 2017-03-23 Richard Biener <rguenther@suse.de>
30998 PR tree-optimization/80032
30999 * gimplify.c (gimple_push_cleanup): Forced unconditional
31000 cleanups still have to go to the conditional_cleanups
31003 2017-03-22 Jakub Jelinek <jakub@redhat.com>
31005 PR tree-optimization/80072
31006 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
31008 (next_operand_entry_id): Change type to unsigned int.
31009 (sort_by_operand_rank): Make sure to return the right return value
31010 even if unsigned fields are bigger than INT_MAX.
31011 (struct oecount): Change cnt and id type to unsigned int.
31012 (oecount_hasher::equal): Formatting fix.
31013 (oecount_cmp): Make sure to return the right return value
31014 even if unsigned fields are bigger than INT_MAX.
31015 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
31018 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
31019 TREE_READONLY on result if writing it more than once.
31022 * doc/invoke.texi (-fsanitize=thread): Document that with
31023 -fnon-call-exceptions atomics are not able to throw
31027 * tsan.c: Include tree-eh.h.
31028 (instrument_builtin_call): Call maybe_clean_eh_stmt or
31029 maybe_clean_or_replace_eh_stmt where needed.
31030 (instrument_memory_accesses): Add cfg_changed argument.
31031 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
31032 if it returned true.
31033 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
31035 PR rtl-optimization/63191
31036 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
31037 wrapper function, moved the whole old content into ...
31038 (ix86_delegitimize_address_1): ... this. New inline function.
31039 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
31040 true as last argument instead of ix86_delegitimize_address.
31042 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
31044 * config/aarch64/aarch64.c (generic_branch_cost): Copy
31045 cortexa57_branch_cost.
31047 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
31049 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
31051 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31054 * doc/md.texi (Constraints): Document wA constraint.
31055 * config/rs6000/constraints.md (wA): New.
31056 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
31057 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
31058 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
31059 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
31061 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
31064 * gimplify.c (is_oacc_declared): New function.
31065 (oacc_default_clause): Use it to set default flags for acc declared
31066 variables inside parallel regions.
31067 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
31068 declared variables.
31069 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
31070 declare attribute to any decl as necessary.
31072 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
31075 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
31076 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
31077 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
31078 (arm_arch_lpae): This.
31079 * config/arm/arm.c (arm_arch7ve): Rename into ...
31080 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
31081 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
31084 2017-03-22 Martin Liska <mliska@suse.cz>
31087 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
31088 error message instead of an ICE.
31090 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31092 * doc/extend.texi (6.11 Additional Floating Types): Revise.
31094 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
31096 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
31098 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31101 2017-03-21 Martin Sebor <msebor@redhat.com>
31103 * doc/extend.texi: Use "cannot" instead of "can't."
31104 * doc/hostconfig.texi: Same.
31105 * doc/install.texi: Same.
31106 * doc/invoke.texi: Same.
31107 * doc/loop.texi: Same.
31108 * doc/md.texi: Same.
31109 * doc/objc.texi: Same.
31110 * doc/rtl.texi: Same.
31111 * doc/tm.texi: Same.
31112 * doc/tm.texi.in: Same.
31113 * doc/trouble.texi: Same.
31115 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
31118 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
31119 (collect_checksum_attributes): Set it.
31120 (die_checksum_ordered): Use it.
31122 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31124 PR tree-optimization/79908
31125 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
31126 change: For a VA_ARG whose LHS has been cast away, use
31127 force_gimple_operand to construct the side effects.
31129 2017-03-21 David Malcolm <dmalcolm@redhat.com>
31131 PR translation/80001
31132 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
31133 more amenable to translation.
31134 (oacc_loop_auto_partitions): Likewise.
31136 2017-03-21 Marek Polacek <polacek@redhat.com>
31137 Martin Sebor <msebor@redhat.com>
31139 PR tree-optimization/80109
31140 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
31141 on INTEGRAL_TYPE_P.
31143 2017-03-21 Jakub Jelinek <jakub@redhat.com>
31144 Segher Boessenkool <segher@kernel.crashing.org>
31147 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
31148 check reg_used_between_p between insn and one of succ or succ2
31149 depending on if succ is artificial insn not inserted into insn
31152 2017-03-21 Martin Liska <mliska@suse.cz>
31154 PR gcov-profile/80081
31155 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
31156 * doc/gcc.texi: Include gcov-dump stuff.
31157 * doc/gcov-dump.texi: New file.
31159 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
31161 PR rtl-optimization/79150
31162 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
31163 conditional jump, if the jump is the last insn of the loop.
31165 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31166 Richard Biener <rguenth@suse.de>
31168 PR tree-optimization/79908
31169 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
31170 been cast away, use force_gimple_operand to construct the side
31173 2017-03-21 Martin Liska <mliska@suse.cz>
31175 PR libfortran/79956
31176 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
31179 2017-03-21 Brad Spengler <spender@grsecurity.net>
31182 * plugin.c (htab_hash_plugin): New function.
31183 (add_new_plugin): Use it and adjust.
31184 (parse_plugin_arg_opt): Adjust.
31185 (init_one_plugin): Likewise.
31187 2017-03-21 Richard Biener <rguenther@suse.de>
31189 PR tree-optimization/80032
31190 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
31191 if set force the cleanup to happen unconditionally.
31192 (gimplify_target_expr): Push inserted clobbers with force_uncond
31193 to avoid them being removed by control-dependent DCE.
31195 2017-03-21 Richard Biener <rguenther@suse.de>
31197 PR tree-optimization/80122
31198 * tree-inline.c (copy_bb): Do not expans va-arg packs or
31199 va_arg_pack_len when the inlined call stmt requires pack
31201 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
31203 2017-03-21 Jakub Jelinek <jakub@redhat.com>
31206 * tsan.c (instrument_builtin_call): If the memory model argument
31207 is not a constant, assume it is valid.
31210 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
31213 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
31215 PR rtl-optimization/79910
31216 * combine.c (can_combine_p): Do not allow combining an I0 or I1
31217 if its dest is used by an insn before I2 (other than the combined
31218 insns themselves, which are properly handled already).
31220 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
31223 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31225 * combine.c (record_used_regs): New static function.
31226 (try_combine): Handle situations where there is an additional
31227 instruction between I2 and I3 which needs to have a LOG_LINK
31231 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31233 * combine.c (try_combine): Delete redundant i1 test. Call
31234 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31236 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31239 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
31240 alternatives 13/14.
31242 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31244 PR tree-optimization/80054
31245 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
31246 the optimization if a PHI or any of its arguments is not dominated
31247 by the candidate's basis. Use gphi* rather than gimple* as
31249 (replace_profitable_candidates): Clean up a gimple* variable that
31250 should be a gphi* variable.
31252 2017-03-20 Martin Sebor <msebor@redhat.com>
31255 * doc/extend.texi (attribute constructor): Document present limitation.
31257 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
31260 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
31261 __POWER9_VECTOR__ #ifdef control, change template definition to
31262 use Power9-specific built-in function.
31263 (vec_any_eq): Likewise.
31264 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
31265 to control outcomes from this test.
31266 (vector_ae_<mode>p): For VEC_F modes, likewise.
31268 2017-03-20 Ian Lance Taylor <iant@google.com>
31270 * config/i386/i386.c (ix86_function_regparm): Save an extra
31271 register for -fsplit-stack with DECL_STATIC_CHAIN.
31273 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31276 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
31277 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
31279 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31281 * config/riscv/riscv.c (riscv_print_operand): Use "fence
31283 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
31286 2017-03-20 Marek Polacek <polacek@redhat.com>
31289 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
31291 2017-03-20 Richard Biener <rguenther@suse.de>
31293 PR tree-optimization/80113
31294 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
31295 allocate extra SSA name for PHI def.
31296 (add_close_phis_to_outer_loops): Likewise.
31297 (add_close_phis_to_merge_points): Likewise.
31298 (copy_loop_close_phi_args): Likewise.
31299 (copy_cond_phi_nodes): Likewise.
31301 2017-03-20 Martin Liska <mliska@suse.cz>
31303 PR middle-end/79753
31304 * tree-chkp.c (chkp_build_returned_bound): Do not build
31305 returned bounds for a LHS that's not a BOUNDED_P type.
31307 2017-03-20 Martin Liska <mliska@suse.cz>
31311 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
31312 COMPLEX_CST and VECTOR_CST.
31314 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31317 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
31318 target operand. A new splitter adds the clobber statement in case
31319 the target operand is dead anyway.
31321 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
31323 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
31324 to age-old versions of binutils and glibc.
31326 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
31328 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
31330 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31332 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
31334 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31336 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
31337 requirement for binutils 2.13.
31339 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31341 * combine.c (try_combine): Delete redundant i1 test. Call
31342 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31344 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
31346 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
31347 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
31349 <riscv64-*-elf>: Re-arrange section
31350 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
31351 <riscv32-*-linux>: Likewise.
31352 <riscv64-*-elf>: Likewise
31353 <riscv64-*-linux>: Likewise.
31355 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
31358 * aarch64.opt(verbose-cost-dump): Fix typo.
31360 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
31363 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
31364 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
31366 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31368 * reload.c (find_reloads): When reloading a nonoffsettable address,
31369 use RELOAD_OTHER for it and its address reloads.
31371 PR rtl-optimization/79910
31372 * combine.c (record_used_regs): New static function.
31373 (try_combine): Handle situations where there is an additional
31374 instruction between I2 and I3 which needs to have a LOG_LINK
31377 2017-03-17 Jeff Law <law@redhat.com>
31379 PR tree-optimization/71437
31380 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
31381 conditional in the hash table first.
31382 (vrp_dom_walker::before_dom_children): Extract condition from
31383 ASSERT_EXPR. Record condition, its inverion and any implied
31384 conditions as well.
31386 2017-03-17 Marek Polacek <polacek@redhat.com>
31387 Markus Trippelsdorf <markus@trippelsdorf.de>
31389 PR tree-optimization/80079
31390 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
31393 2017-03-17 Richard Biener <rguenther@suse.de>
31395 PR middle-end/80075
31396 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
31397 Properly verify the LHS before the RHS possibly claims to be
31399 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
31402 2017-03-17 Martin Jambor <mjambor@suse.cz>
31404 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
31405 (List of -O2 options): Likewise.
31406 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
31409 2017-03-17 Tom de Vries <tom@codesourcery.com>
31411 * gcov-dump.c (print_usage): Print bug_report_url.
31413 2017-03-17 Richard Biener <rguenther@suse.de>
31415 PR middle-end/80050
31416 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
31417 (parser::peek): Likewise.
31419 2017-03-17 Richard Biener <rguenther@suse.de>
31421 PR tree-optimization/80048
31422 * sese.c (free_sese_info): Properly release rename_map and
31423 copied_bb_map elements.
31425 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
31427 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
31428 Add linked-list forward and backlinks. Insert on
31429 construction, remove on destruction.
31430 (class pass_store_merging): Add m_stores_head field.
31431 (pass_store_merging::terminate_and_process_all_chains):
31432 Iterate over m_stores_head list.
31433 (pass_store_merging::terminate_all_aliasing_chains):
31435 (pass_store_merging::execute): Check for debug stmts first.
31436 Push new chains onto the m_stores_head stack.
31438 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
31441 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
31442 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
31443 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
31445 2017-03-16 Jeff Law <law@redhat.com>
31447 PR tree-optimization/71437
31448 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
31449 member function. Implementation moved into after_dom_children
31450 member function and into the threader's thread_outgoing_edges
31452 (dom_opt_dom_walker::after_dom_children): Simplify by moving
31453 some code into new thread_outgoing_edges.
31454 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
31455 definition. Simplify marker handling (do it here). Assume we always
31456 have the available expression and the const/copies tables.
31457 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
31459 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
31460 * tree-vrp.c (equiv_stack): No longer file scoped.
31461 (vrp_dom_walker): New class.
31462 (vrp_dom_walker::before_dom_children): New member function.
31463 (vrp_dom_walker::after_dom_children): Likewise.
31464 (identify_jump_threads): Setup domwalker. Use it rather than
31465 walking edges in a random order by hand. Simplify setup/finalization.
31466 (finalize_jump_threads): Remove.
31467 (vrp_finalize): Do not call identify_jump_threads here.
31468 (execute_vrp): Do it here instead and call thread_through_all_blocks
31471 PR tree-optimization/71437
31472 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
31474 (simplify_stmt_for_jump_threading): Add basic_block argument. All
31476 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
31477 (dom_opt_dom_walker::thread_across_edge): Remove
31478 handle_dominating_asserts argument. All callers changed.
31479 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
31480 changes. Remove calls to lhs_of_dominating_assert. Other
31481 uses of handle_dominating_asserts turn into unconditional code
31482 (simplify_control_stmt_condition_1): Likewise.
31483 (simplify_control_stmt_condition): Likewise.
31484 (thread_through_normal_block, thread_across_edge): Likewise.
31485 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
31486 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
31487 object if it is not an SSA_NAME.
31488 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
31489 before calling into the VRP specific simplifiers.
31490 (identify_jump_threads): Remove handle_dominating_asserts
31493 2017-03-16 Jakub Jelinek <jakub@redhat.com>
31496 * tree-diagnostic.c (default_tree_printer): No longer static.
31497 * tree-diagnostic.h (default_tree_printer): New prototype.
31499 2017-03-16 Tamar Christina <tamar.christina@arm.com>
31501 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
31502 Change ins into fmov.
31504 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31506 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
31507 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
31508 Use h_con constraint for operand 1.
31509 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
31510 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
31512 2017-03-15 Jeff Law <law@redhat.com>
31514 PR tree-optimization/71437
31515 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
31516 (record_temporary_equivalences): Use it.
31518 PR tree-optimization/71437
31519 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
31520 tree-ssa-scopedtables.
31521 (lookup_avail_expr, build_and_record_new_cond): Likewise.
31522 (record_conditions, record_cond, vuse_eq): Likewise.
31523 (record_edge_info): Adjust to API tweak of record_conditions.
31524 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
31525 (record_temporary_equivalences, optimize_stmt): Likewise.
31526 (eliminate_redundant_computations): Likewise.
31527 (record_equivalences_from_stmt): Likewise.
31528 * tree-ssa-scopedtables.c: Include options.h and params.h.
31529 (vuse_eq): New function, moved from tree-ssa-dom.c
31530 (build_and_record_new_cond): Likewise.
31531 (record_conditions): Likewise. Accept vector of conditions rather
31532 than edge_equivalence structure for first argument.
31533 for the first argument.
31534 (avail_exprs_stack::lookup_avail_expr): New member function, moved
31535 from tree-ssa-dom.c.
31536 (avail_exprs_stack::record_cond): Likewise.
31537 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
31538 from tree-ssa-dom.c.
31539 (avail_exprs_stack): Add new member functions lookup_avail_expr
31541 (record_conditions): Declare.
31543 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
31546 * lra-constraints.c (process_alt_operands): Increase reject for
31547 reloading an input/output operand.
31549 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
31552 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
31553 insns to convert from signed/unsigned char/short to IEEE 128-bit
31555 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
31557 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
31560 * config/i386/i386.c (ix86_vector_duplicate_value): Create
31561 subreg of inner mode for values already in registers.
31563 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
31565 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
31566 iteration reg is used after the loop.
31568 2017-03-14 Martin Sebor <msebor@redhat.com>
31570 PR tree-optimization/79800
31571 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
31572 precision in negative-positive range.
31573 (format_floating): Call non-const overload with adjusted precision.
31575 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
31578 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
31581 2017-03-14 Martin Sebor <msebor@redhat.com>
31583 PR middle-end/80020
31584 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
31585 * builtins.def (aligned_alloc): Use it.
31588 * Makefile.in (GTFILES): Add calls.c.
31589 * calls.c: Include "gt-calls.h".
31591 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
31593 PR rtl-optimization/79728
31594 * regs.h (struct target_regs): New field
31595 x_contains_allocatable_regs_of_mode.
31596 (contains_allocatable_regs_of_mode): New macro.
31597 * reginfo.c (init_reg_sets_1): Initialize it, and change
31598 contains_reg_of_mode so it includes global regs as well.
31599 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
31600 rather than contains_regs_of_mode.
31602 2017-03-14 Martin Liska <mliska@suse.cz>
31604 * doc/invoke.texi: Document options that can't be combined with
31605 -fcheck-pointer-bounds.
31607 2017-03-14 Martin Liska <mliska@suse.cz>
31609 PR middle-end/79831
31610 * doc/invoke.texi (-Wchkp): Document the option.
31612 2017-03-14 Martin Liska <mliska@suse.cz>
31614 * Makefile.in: Install gcov-dump.
31616 2017-03-14 Martin Liska <mliska@suse.cz>
31618 * multiple_target.c (expand_target_clones): Bail out for
31619 an invalid attribute.
31621 2017-03-14 Richard Biener <rguenther@suse.de>
31623 * alias.c (struct alias_set_entry): Pack properly.
31624 * cfgloop.h (struct loop): Likewise.
31625 * cse.c (struct set): Likewise.
31626 * ipa-utils.c (struct searchc_env): Likewise.
31627 * loop-invariant.c (struct invariant): Likewise.
31628 * lra-remat.c (struct cand): Likewise.
31629 * recog.c (struct change_t): Likewise.
31630 * rtl.h (struct address_info): Likewise.
31631 * symbol-summary.h (function_summary): Likewise.
31632 * tree-loop-distribution.c (struct partition): Likewise.
31633 * tree-object-size.c (struct object_size_info): Likewise.
31634 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
31635 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
31636 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
31637 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
31638 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
31639 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
31640 (struct _stmt_vec_info): Likewise.
31642 2017-03-14 Martin Liska <mliska@suse.cz>
31645 * multiple_target.c (create_dispatcher_calls): Check that
31646 a target can create a function dispatcher.
31648 2017-03-14 Martin Liska <mliska@suse.cz>
31651 * multiple_target.c (expand_target_clones): Drop local.local
31652 flag for default implementation.
31654 2017-03-14 Richard Biener <rguenther@suse.de>
31656 PR tree-optimization/80030
31657 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
31659 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
31661 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
31662 gcc_fallthrough() instead of __attribute__((fallthrough));
31664 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31666 * doc/gcc.texi: Remove "up" link to (DIR).
31667 * doc/gccint.texi: Ditto.
31669 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31671 * doc/install.texi (Specific) <avr>: Remove reference to
31674 2017-03-13 Jeff Law <law@redhat.com>
31676 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
31677 attribute rather than comments.
31679 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
31680 match_scratch operand is highest.
31682 2017-03-13 Martin Liska <mliska@suse.cz>
31684 PR middle-end/78339
31685 * ipa-pure-const.c (warn_function_noreturn): If the declarations
31686 is a CHKP clone, use original declaration.
31688 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31690 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
31691 (arc_conditional_register_usage): Use a different allocation order
31692 when optimizing for size.
31693 * common/config/arc/arc-common.c (arc_option_optimization_table):
31694 Section anchors default on when optimizing for size.
31696 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31698 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
31700 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31702 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
31703 * config/arc/arc.md (cpu_facility): Add cd variant.
31704 (*movqi_insn): Add code density variant.
31705 (*movhi_insn): Likewise.
31706 (*movqi_insn): Likewise.
31707 (*addsi3_mixed): Likewise.
31708 (subsi3_insn): Likewise.
31710 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31712 * config/arc/arc.md (movsi_cond_exec): Update constraint.
31714 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31716 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
31717 expressions with MINUS and UNARY ops.
31719 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31722 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
31724 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
31725 between vec_select and vector argument.
31726 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
31727 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
31728 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
31729 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
31730 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
31731 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
31733 2017-03-13 Richard Biener <rguenther@suse.de>
31736 * params.def (vect-max-peeling-for-alignment): Fix typo.
31738 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31740 * doc/install.texi (Specific) <mips-*-*>: Remove description of
31741 issue that only occurred with binutils below 2.18.
31743 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31745 * doc/install.texi (Specific) <cris-axis-elf>: No longer
31746 refer to binutils 2.11/2.12 minimum.
31748 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31750 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
31751 ftp.kernel.org and simplify binutils requirement.
31753 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
31755 * doc/invoke.texi (Warning Options): Fix spelling of link-time
31757 (Optimize Options): Ditto. Also remove redundancy.
31759 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31761 PR translation/79848
31762 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
31764 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
31765 to G_ to avoid double translation.
31767 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31769 PR translation/79923
31770 * auto-profile.c (get_combined_location): Convert leading
31771 character of diagnostics to lower case and remove trailing period.
31772 (read_profile): Likewise for various diagnostics.
31773 * config/arm/arm.c (arm_option_override): Remove trailing period
31774 from various diagnostics.
31775 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
31776 (msp430_expand_delay_cycles): Likewise.
31778 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31781 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
31782 full command-line argument, rather than just "str".
31783 (aarch64_validate_march): Likewise.
31784 (aarch64_validate_mtune): Likewise.
31786 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
31788 PR rtl-optimization/78911
31789 * lra-assigns.c (must_not_spill_p): New function.
31790 (spill_for): Use it.
31792 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31794 PR tree-optimization/79981
31795 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
31796 ATOMIC_COMPARE_EXCHANGE ifn result.
31797 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
31798 IFN_ATOMIC_COMPARE_EXCHANGE.
31800 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31803 * opts.c (parse_sanitizer_options): Add missing question mark to
31804 "did you mean" message.
31806 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31808 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
31810 (VMULEUH_UNS): Likewise.
31811 (VMULOUB_UNS): Likewise.
31812 (VMULOUH_UNS): Likewise.
31813 * config/rs6000/rs6000.c (builtin_function_type): Remove
31814 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
31816 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31819 * read-rtl-function.c (function_reader::read_rtx_operand): Update
31820 x with result of extra_parsing_for_operand_code_0.
31821 (function_reader::extra_parsing_for_operand_code_0): Convert
31822 return type from void to rtx, returning x. When reading
31823 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
31824 larger size containing struct block_symbol.
31826 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
31828 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
31829 -mfloat128-hardware without -m64.
31831 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
31834 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
31835 entries to the case statement that marks unsigned arguments to
31836 overloaded functions.
31838 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31840 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
31841 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
31843 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
31846 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
31847 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
31849 2017-03-10 Martin Liska <mliska@suse.cz>
31853 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
31854 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
31857 2017-03-10 Olivier Hainque <hainque@adacore.com>
31859 * tree-switch-conversion (array_value_type): Start by resetting
31860 candidate type to it's main variant.
31862 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31864 PR rtl-optimization/79909
31865 * combine.c (try_combine): Use simplify_replace_rtx on individual
31866 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
31867 of the whole CALL_INSN_FUNCTION_USAGE.
31869 PR tree-optimization/79972
31870 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
31871 get_range_info on SSA_NAMEs. Formatting fixes.
31873 2017-03-10 Richard Biener <rguenther@suse.de>
31874 Jakub Jelinek <jakub@redhat.com>
31876 PR tree-optimization/77975
31877 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
31878 edge to be constant.
31879 (get_val_for): For constant x return it. Formatting fix.
31880 (loop_niter_by_eval): Avoid pointless looping if the next iteration
31881 would use the same bases as the current one.
31883 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31885 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
31886 instead of vec_select for V1TImode.
31887 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
31889 (VSX_LE_128): Add V1TI to this mode iterator.
31890 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
31891 (*vsx_le_perm_store_<mode>): Likewise.
31892 (pre-reload splitter for VSX stores): Likewise.
31893 (post-reload splitter for VSX stores): Likewise.
31894 (*vsx_xxpermdi2_le_<mode>): Likewise.
31895 (*vsx_lxvd2x2_le_<mode>): Likewise.
31896 (*vsx_stxvd2x2_le_<mode>): Likewise.
31898 2017-03-09 Michael Eager <eager@eagercon.com>
31900 Correct failures with --enable-checking=yes,rtl.
31902 * config/microblaze/microblaze.c (microblaze_expand_shift):
31903 Replace GET_CODE test with CONST_INT_P and INTVAL test with
31904 test for const0_rtx.
31905 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
31906 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
31908 2017-03-09 Richard Biener <rguenther@suse.de>
31910 PR tree-optimization/79977
31911 * graphite-scop-detection.c (scop_detection::merge_sese):
31912 Handle the case of extra exits to blocks dominating the entry.
31914 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
31916 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
31919 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
31921 PR rtl-optimization/79949
31922 * lra-constraints.c (process_alt_operands): Check memory when
31923 trying to predict a cycle. Print about the overall increase.
31925 2017-03-09 Richard Biener <rguenther@suse.de>
31927 PR middle-end/79971
31928 * gimple-expr.c (useless_type_conversion_p): Preserve
31929 TYPE_SATURATING for fixed-point types.
31931 2017-03-09 Richard Biener <rguenther@suse.de>
31934 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
31935 alignment of BLKmode params.
31937 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31940 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
31941 (VALL_NO_V2Q): Likewise.
31943 * config/aarch64/aarch64-simd.md
31944 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
31946 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
31947 VALL_NO_V2Q mode iterator.
31948 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
31950 2017-03-09 Martin Liska <mliska@suse.cz>
31952 PR tree-optimization/79631
31953 * tree-chkp-opt.c (chkp_is_constant_addr): Call
31954 tree_int_cst_sign_bit just for INTEGER constants.
31956 2017-03-09 Martin Liska <mliska@suse.cz>
31960 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
31963 2017-03-09 Marek Polacek <polacek@redhat.com>
31966 * tree.c (inchash::add_expr): Handle TREE_VEC.
31968 2017-03-09 Martin Liska <mliska@suse.cz>
31971 (chkp_narrow_size_and_offset): New function.
31972 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
31973 (void chkp_parse_bit_field_ref): New function.
31974 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
31975 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
31977 2017-03-09 Martin Liska <mliska@suse.cz>
31980 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
31981 (chkp_find_bounds_1): Remove gcc_unreachable.
31983 2017-03-09 Jakub Jelinek <jakub@redhat.com>
31986 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
31987 BUILT_IN_SYNC*, determine the access type from the size suffix and
31988 always build a MEM_REF with that type. Handle forgotten
31989 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
31992 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
31993 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
31994 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
31995 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
31996 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
31997 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
31998 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
31999 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
32000 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
32001 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
32002 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
32003 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
32004 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
32005 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
32006 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
32007 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
32008 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
32009 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
32010 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
32011 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
32012 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
32013 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
32014 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
32015 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
32016 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
32017 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
32018 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
32019 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
32020 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
32021 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
32022 definitions outside of __OPTIMIZE__ guarded section.
32025 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
32026 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
32027 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
32028 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
32031 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32033 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
32034 ("vfenez<mode>"): Add missing constraints.
32036 2017-03-08 Martin Sebor <msebor@redhat.com>
32039 * config/nds32/nds32.c (nds32_option_override):
32040 Fix misspelled diagnostic.
32042 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32045 * gimplify.c (gimplify_omp_for): Replace index var in outer
32046 taskloop statement with an artificial variable and add
32047 OMP_CLAUSE_PRIVATE clause for it.
32049 2017-03-08 Richard Biener <rguenther@suse.de>
32051 PR tree-optimization/79955
32052 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
32053 for accesses that are completely outside of the variable.
32055 2017-03-08 Andrew Haley <aph@redhat.com>
32057 PR tree-optimization/79943
32058 * tree-ssa-loop-split.c (compute_new_first_bound): When
32059 calculating the new upper bound, (END-BEG) should be added, not
32062 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32064 * config/avr/avr.md (setmemhi): Make sure match_dup
32065 operand number comes before match_scratch.
32067 2017-03-08 Richard Biener <rguenther@suse.de>
32069 PR tree-optimization/79920
32070 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
32071 with ncopies == 1 to ...
32072 (vect_transform_slp_perm_load): ... here. Properly compute
32073 all element loads by iterating VF times over the group. Do
32074 not handle ncopies (computed in a broken way) in
32075 vect_create_mask_and_perm.
32077 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32080 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
32081 is a uniform vector, use uniform_vector_p return value instead of
32082 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
32084 2017-03-07 Marek Polacek <polacek@redhat.com>
32086 PR middle-end/79809
32087 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
32088 (alloca_call_type): Likewise.
32090 2017-03-07 Martin Liska <mliska@suse.cz>
32092 * gcov.c (process_args): Put comment to correct location.
32094 2017-03-07 Martin Liska <mliska@suse.cz>
32096 PR middle-end/68270
32097 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
32098 Use array_at_struct_end_p instead of DECL_CHAIN (field).
32099 (chkp_narrow_bounds_for_field): Likewise.
32100 (chkp_parse_array_and_component_ref): Pass one more argument to
32103 2017-03-07 Richard Biener <rguenther@suse.de>
32105 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
32108 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
32110 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
32111 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
32113 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32116 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
32117 to end of description.
32118 (PARAM_MAX_STORES_TO_MERGE): Likewise.
32120 2017-03-07 Jakub Jelinek <jakub@redhat.com>
32122 PR rtl-optimization/79901
32123 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
32125 (*avx512f_<code><mode>3<mask_name>): ... this.
32126 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
32127 iterator instead of VI8_AVX2_AVX512BW.
32129 PR rtl-optimization/79901
32130 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
32131 min/max expander, expand it using expand_vec_cond_expr.
32134 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
32137 2017-03-06 Jakub Jelinek <jakub@redhat.com>
32140 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
32141 to void * for PCH reasons.
32142 * dwarf2out.c (output_loc_operands, output_die): Cast
32143 v.val_vec.array to unsigned char *.
32145 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
32148 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
32151 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
32153 PR rtl-optimization/79571
32154 * lra-constraints.c (process_alt_operands): Calculate static
32155 reject and subtract it from overall when only addresses will be
32158 2017-03-06 Julia Koval <julia.koval@intel.com>
32161 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
32162 incoming stack boundary to 128 for 64-bit targets.
32164 2017-03-06 Richard Biener <rguenther@suse.de>
32166 PR tree-optimization/79894
32167 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
32168 to NULL after folding it.
32170 2017-03-06 Richard Biener <rguenther@suse.de>
32172 PR tree-optimization/79824
32173 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
32174 check disabling peeling for gaps.
32176 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
32178 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
32179 attributes): Document gettimeofday.
32181 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
32183 * config/s390/s390.c (s390_option_override_internal): Set
32184 PARAM_MIN_VECT_LOOP_BOUND
32186 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
32188 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
32189 * config/s390/s390.md: Likewise.
32191 2017-03-06 Jakub Jelinek <jakub@redhat.com>
32194 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
32195 (<avx2_avx512>_perm<mode>): Rename to ...
32196 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
32198 (<avx512>_perm<mode>_mask): Rename to ...
32199 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
32201 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
32202 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
32203 instead of VI8F_256_512.
32204 (avx512f_perm<mode>): New define_expand.
32205 (avx512f_perm<mode>_mask): Likewise.
32206 (avx512f_perm<mode>_1<mask_name>): New define_insn.
32207 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
32209 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32211 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
32212 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
32214 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
32216 2017-03-06 Martin Liska <mliska@suse.cz>
32219 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
32220 when having a SSA NAME w/o VAR_DECL assigned to it.
32222 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32224 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
32225 msa_dpsub_<su>_d): Fix MODE for vec_select.
32227 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32229 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
32231 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
32233 2017-03-06 Richard Biener <rguenther@suse.de>
32235 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
32236 * plugin.c (register_plugin_info): Likewise.
32237 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
32239 2017-03-05 Jakub Jelinek <jakub@redhat.com>
32241 * config/i386/sse.md (sse_storehps, sse_storelps,
32242 avx_<castmode><avxsizesuffix>_<castmode>,
32243 avx512f_<castmode><avxsizesuffix>_<castmode>,
32244 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
32245 in condition that at least one operand is not a MEM.
32247 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32249 PR middle-end/79805
32250 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
32251 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
32253 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
32254 gimple_call_nothrow_p flag based on whether original builtin can throw.
32255 If it can, emit following stmts on the fallthrough edge.
32256 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
32257 don't create new bb if inserting just debug stmts on the edge, try to
32258 insert them on the fallthru bb or just reset debug stmts.
32260 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
32263 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
32264 restore recog_data (including the operand rtxes inside it) around
32265 the call to get_insn_template.
32267 2017-03-03 Martin Sebor <msebor@redhat.com>
32269 PR tree-optimization/79699
32270 * context.c (context::~context): Free MPFR caches to avoid
32271 a memory leak on program exit.
32273 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32275 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
32276 Use wide_int::ulow () instead of .elt (0).
32278 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32280 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
32281 (*pushxf): Limit oF constraint to 32bit targets and add oC
32282 constraint for 64bit targets.
32283 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
32284 (*pushdf): Change rmF constraint to rmC.
32286 2017-03-03 Martin Liska <mliska@suse.cz>
32288 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
32289 Remove unused variable.
32291 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32294 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
32295 is a memory operand, increase num_memory.
32296 (ix86_expand_args_builtin): Likewise.
32298 2017-03-03 Jan Hubicka <jh@suse.cz>
32301 * ipa-devirt.c (maybe_record_node): Properly handle
32302 __cxa_pure_virtual visibility.
32304 2017-03-03 Martin Liska <mliska@suse.cz>
32306 PR tree-optimization/79803
32307 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
32309 (pass_loop_prefetch::execute): Disabled optimization if an
32310 assumption about L1 cache size is not met.
32312 2017-03-03 Martin Liska <mliska@suse.cz>
32314 PR rtl-optimization/79574
32315 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
32316 (hash_scan_set): Likewise.
32317 (dump_hash_table): Likewise.
32318 (hoist_code): Likewise.
32320 2017-03-03 Richard Biener <rguenther@suse.de>
32322 * fixed-value.c (fixed_from_string): Restore use of elt (1)
32323 in place of uhigh ().
32324 (fixed_convert_from_real): Likewise.
32326 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32329 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
32331 2017-03-03 Richard Biener <rguenther@suse.de>
32333 PR middle-end/79818
32334 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
32335 TYPE_OVERFLOW_UNDEFINED check.
32337 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32339 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
32341 (vector_ae_<mode>_p): Likewise.
32342 (vector_nez_<mode>_p): Likewise.
32343 (vector_ne_v2di_p): Likewise.
32344 (vector_ae_v2di_p): Likewise.
32345 (vector_ne_<mode>_p): Likewise.
32346 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
32348 (vsx_tsqrt<mode>2_fe): Likewise.
32350 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
32353 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
32355 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32357 PR rtl-optimization/79780
32358 * cprop.c (one_cprop_pass): When second and further conditional trap
32359 in a single basic block is turned into an unconditional trap, turn it
32360 into a deleted note to avoid RTL verification failures.
32362 2017-03-02 Richard Biener <rguenther@suse.de>
32364 * fold-const.c (const_binop): Use ulow () instead of elt (0).
32366 2017-03-02 Richard Biener <rguenther@suse.de>
32368 PR tree-optimization/79345
32370 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
32371 param and abort the walk, returning -1 if it is hit.
32372 (walk_aliased_vdefs): Take a limit param and pass it on.
32373 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
32374 defaulting to 0 and return a signed int.
32375 * tree-ssa-uninit.c (struct check_defs_data): New struct.
32376 (check_defs): New helper.
32377 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
32378 about uninitialized memory.
32379 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
32380 bogus uninitialized warning.
32381 (fixed_convert_from_real): Likewise.
32383 2017-03-02 Bin Cheng <bin.cheng@arm.com>
32385 PR tree-optimization/66768
32386 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
32387 iv_use if base object can't be determined.
32389 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32391 PR tree-optimization/79345
32392 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
32393 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
32394 (get_pattern_stats): Initialize it.
32395 * genemit.c (gen_expand): Verify match_scratch numbers come after
32396 match_operand/match_dup numbers.
32397 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
32398 match_scratch numbers.
32399 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
32401 * config/s390/s390.md (trunctdsd2): Likewise.
32403 2017-03-02 Richard Biener <rguenther@suse.de>
32405 * wide-int.h (wide_int_storage::operator=): Implement in terms
32408 2017-03-02 Richard Biener <rguenther@suse.de>
32410 PR tree-optimization/79777
32411 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
32412 the to insert expression to sth existing.
32414 2017-03-01 Martin Sebor <msebor@redhat.com>
32416 PR middle-end/79692
32417 * gimple-ssa-sprintf.c
32418 (directive::known_width_and_precision): New function.
32419 (format_integer): Use it.
32420 (get_mpfr_format_length): Consider the full range of precision
32421 when computing %g output with the # flag. Set the likely byte
32422 count to 3 rather than 1 when precision is indeterminate.
32423 (format_floating): Correct the lower bound of precision.
32425 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32427 * doc/invoke.texi: Document default code model for 64-bit Linux.
32429 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32432 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
32433 udiv rather than div since input pattern is unsigned.
32435 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
32437 * config/i386/i386.c (print_reg): Warn for values of
32438 unsupported size in integer register.
32440 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
32443 * config/rs6000/predicates.md (current_file_function_operand): Do
32444 not allow self calls to be local if the function is replaceable.
32446 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
32449 * config/rs6000/altivec.h (vec_ctz and others): Change the
32450 preprocessor macro that controls conditional compilation from
32451 _ARCH_PWR9 to __POWER9_VECTOR__.
32452 (vec_all_ne): Change parameterization of __altivec_scalar_pred
32453 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
32454 control (instead of _ARCH_PWR9 control) so that template
32455 definition uses power9-specific function.
32456 (vec_any_eq): Likewise.
32457 (vec_all_ne): Change macro definition to use a power9-specific
32458 expansion under #ifdef __POWER9_VECTOR__ control (instead of
32459 _ARCH_PWR9 control).
32460 (vec_any_eq) Likewise.
32461 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
32462 expansion for CMPNEF to remove support for xvcmpnesp instruction.
32463 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
32464 support for xvcmpnedp instruction.
32465 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
32466 macro expansion so that Power9 implementation of vec_all_ne does
32467 not use the AltiVec predicate framework.
32468 (VCMPNEH_P): Likewise.
32469 (VCMPNEW_P): Likewise.
32470 (VCMPNED_P): Likewise.
32471 (VCMPNEFP_P): Likewise.
32472 (VCMPNEDP_P): Likewise.
32473 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
32474 implementation of vec_any_eq to not use AltiVec predicate
32476 (VCMPAEH_P): Likewise.
32477 (VCMPAEW_P): Likewise.
32478 (VCMPAED_P): Likewise.
32479 (VCMPAEFP_P): Likewise.
32480 (VCMPAEDP_P): Likewise.
32481 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
32482 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
32483 not use the AltiVec predicate framework.
32484 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
32485 of vec_any_eq to not use AltiVec predicate framework.
32486 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32487 support for predefined __POWER9_VECTOR__ macro to indicate that
32488 Power9 instruction selection is enabled.
32489 (altivec_overloaded_builtins): Remove extraneous
32490 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
32491 function argument types RS6000_BTI_bool_V16QI and
32492 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
32493 entry for overloaded function argument types RS6000_BTI_bool_V4SI
32494 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
32495 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
32496 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
32497 Power9 for implementations of vec_cmpne. Change the signature for
32498 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
32499 (representing vec_all_ne) to remove the previously described first
32500 argument of type RS6000_BTI_INTSI, as this was an artifact of
32501 reliance on the AltiVec predicate framework, which is no longer
32502 used in the implementation of these functions. Add
32503 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
32504 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
32505 since, unlike the AltiVec predicate framework implementation, we
32506 do not share function descriptors between vec_alle and vec_anyeq.
32507 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
32508 set of modes that receive special treatment even when
32509 TARGET_P9_VECTOR is true. The special treatment emits code that
32510 does not depend on Power9 instructions.
32511 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
32512 define_expand to not rely on AltiVec predicate framework.
32513 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
32515 (vector_ne_v2di_p): Change this define_expand to not rely on
32516 AltiVec predicate framework.
32517 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
32519 (vector_ne_<mode>_p): Change this define_expand to not rely on
32520 AltiVec predicate framework.
32521 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
32523 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
32524 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
32525 define_insn pattern.
32526 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
32527 define_insn pattern because the xvcmpne<VSs>. instruction is not
32529 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
32530 instruction is not supported.
32532 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32534 * config/nvptx/nvptx.c: Include intl.h.
32536 2017-03-01 Martin Jambor <mjambor@suse.cz>
32539 * ipa-prop.h (ipa_bits): Removed field known.
32540 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
32541 to pointers. Adjusted their comments to warn about their sharing.
32542 (ipcp_transformation_summary): Change bits to a vector of pointers.
32543 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
32544 (ipa_get_ipa_bits_for_value): Declare.
32545 * tree-vrp.h (value_range): Mark as GTY((for_user)).
32546 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
32547 (ipa_bits_hash_table): Likewise.
32548 (ipa_vr_ggc_hash_traits): Likewise.
32549 (ipa_vr_hash_table): Likewise.
32550 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
32551 being pointers and vr_known being removed.
32552 (ipa_set_jf_unknown): Likewise.
32553 (ipa_get_ipa_bits_for_value): New function.
32554 (ipa_set_jfunc_bits): Likewise.
32555 (ipa_get_value_range): New overloaded functions.
32556 (ipa_set_jfunc_vr): Likewise.
32557 (ipa_compute_jump_functions_for_edge): Use the above functions to
32558 construct bits and vr parts of jump functions.
32559 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
32560 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
32562 (ipcp_grow_transformations_if_necessary): Also allocate
32563 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
32565 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
32566 them. Fix too long lines.
32567 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
32568 vr_known being removed.
32569 (ipa_read_jump_function): Use new setter functions to construct bits
32570 and vr parts of jump functions or set them to NULL.
32571 (write_ipcp_transformation_info): Adjust for bits being pointers.
32572 (read_ipcp_transformation_info): Likewise.
32573 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
32575 Include gt-ipa-prop.h.
32576 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
32578 (ipcp_store_bits_results): Likewise.
32579 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
32580 Do not write to existing jump functions but use a temporary instead.
32582 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32585 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
32586 attempt to use its first operand as BIT_FIELD_REF base.
32588 2017-03-01 Richard Biener <rguenther@suse.de>
32590 PR middle-end/79721
32591 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
32592 interpolating formula in wrapping arithmetic.
32593 (chrec_apply): Convert chrec_evaluate return value to wanted type.
32595 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32597 PR tree-optimization/79734
32598 * tree-vect-generic.c (expand_vector_condition): Optimize
32599 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
32600 Handle VEC_COND_EXPR where comparison has different inner width from
32601 type's inner width.
32603 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
32605 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
32606 markup, and similar issues. Remove @opindex entries for things
32607 that aren't options. Add missing -mmpy-option entries.
32609 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32611 PR tree-optimization/79737
32612 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
32613 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
32614 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
32615 instead of byte_size. Formatting fix.
32616 (shift_bytes_in_array_right): Formatting fix.
32618 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
32621 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
32622 condition on optimize for the leaf function test.
32624 2017-02-28 Martin Liska <mliska@suse.cz>
32627 * read-rtl-function.c (function_reader::handle_unknown_directive):
32628 Bail out when one uses -flto.
32630 2017-02-28 Martin Liska <mliska@suse.cz>
32632 * common.opt: Replace space with tabular for options of <number>
32634 * config/i386/i386.opt: Show <number> value for
32635 -mlarge-data-threshold.
32636 * opts.c (print_filtered_help): Do not display number in hexadecimal
32639 2017-02-28 Martin Liska <mliska@suse.cz>
32641 * common.opt: Fix --help=option -Q for options which are of
32644 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
32646 * config/i386/i386.c (print_reg): Error out for values
32647 of 8-bit size in invalid integer register.
32649 2017-02-28 Martin Sebor <msebor@redhat.com>
32651 PR tree-optimization/79691
32652 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
32654 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32657 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
32658 gcc_unreachable with output_operand_lossage.
32660 2017-02-28 Richard Biener <rguenther@suse.de>
32662 PR tree-optimization/79740
32663 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
32665 (visit_nary_op): Insert the nary into the hashtable if we
32666 pattern-matched sth.
32667 * tree-ssa-pre.c (eliminate_insert): Robustify.
32669 2017-02-28 Richard Biener <rguenther@suse.de>
32671 PR middle-end/79731
32672 * fold-const.c (decode_field_reference): Reject out-of-bound
32675 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32677 * config/i386/i386.c: Include intl.h.
32678 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
32679 instead of just cond ? "..." : "...".
32680 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
32681 * coverage.c (read_counts_file): Likewise.
32682 * omp-offload.c: Include intl.h.
32683 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
32684 of just cond ? "..." : "...".
32685 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
32686 of just cond ? "..." : "...".
32688 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
32691 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
32693 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
32694 'tune for' CPU name.
32695 * config/arm/arm-cpu-data.h: Regenerated.
32697 2017-02-28 Richard Biener <rguenther@suse.de>
32699 PR tree-optimization/79732
32700 * tree-inline.c (expand_call_inline): Do not shadow var.
32702 2017-02-28 Richard Biener <rguenther@suse.de>
32704 PR tree-optimization/79723
32705 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
32706 address-space properly.
32708 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
32710 * doc/optinfo.texi (Optimization groups): Fix option used for
32712 * doc/invoke.texi (-fopt-info): Document "omp".
32713 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
32714 (OPTGROUP_ALL): Add OPTGROUP_OMP.
32715 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
32716 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
32717 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
32719 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
32721 * dumpfile.c (optgroup_options): Instead of "openmp", associate
32722 OPTGROUP_OMP with "omp".
32724 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
32727 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
32728 for arithmetic shift of unsigned V2DI.
32730 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
32732 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
32733 arc/linux.h headers.
32734 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
32735 (LINK_SPEC): Likewise.
32736 (ARC_TLS_EXTRA_START_SPEC): Likewise.
32737 (EXTRA_SPECS): Likewise.
32738 (STARTFILE_SPEC): Likewise.
32739 (ENDFILE_SPEC): Likewise.
32740 (LIB_SPEC): Likewise.
32741 (TARGET_SDATA_DEFAULT): Likewise.
32742 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
32743 (MULTILIB_DEFAULTS): Likewise.
32744 (DWARF2_UNWIND_INFO): Likewise.
32745 * config/arc/big.h: New file.
32746 * config/arc/elf.h: Likewise.
32747 * config/arc/linux.h: Likewise.
32748 * config/arc/t-uClibc: Remove.
32750 2017-02-27 Bin Cheng <bin.cheng@arm.com>
32752 PR tree-optimization/77536
32753 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
32754 (tree_transform_and_unroll_loop): Use above function to compute the
32755 estimated niter of unrolled loop and use it when scaling profile.
32756 Also use count info rather than frequency if it's non-zero.
32757 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
32758 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
32759 (vect_transform_loop): Call above function.
32761 2017-02-27 Richard Biener <rguenther@suse.de>
32763 PR tree-optimization/45397
32764 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
32765 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
32766 (visit_nary_op): Add pattern matching for CSEing sign-changed
32767 or truncated operations with wider ones.
32769 2017-02-27 Richard Biener <rguenther@suse.de>
32771 PR tree-optimization/79690
32772 * tree-vect-stmts.c (vectorizable_store): Use vector type
32773 built from the DR with address-space.
32775 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
32777 * doc/invoke.texi (Optimize Options): Refine the description
32778 of asan-use-after-return.
32780 2017-02-25 Alan Modra <amodra@gmail.com>
32782 PR rtl-optimization/79584
32783 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
32784 base, not ad->base_term, the reg within base. Remove assertion
32785 that ad->base == ad->base_term. Replace gen_int_mode using
32786 bogus mode with const0_rtx.
32788 2017-02-25 Jakub Jelinek <jakub@redhat.com>
32790 PR middle-end/79396
32791 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
32792 FMA_EXPR like tcc_binary or tcc_unary.
32794 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
32797 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
32799 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
32800 (output_loc_operands): Handle DW_OP_call_ref and
32801 DW_OP_GNU_variable_value.
32802 (struct variable_value_struct): New type.
32803 (struct variable_value_hasher): Likewise.
32804 (variable_value_hash): New variable.
32805 (string_types): Remove.
32806 (copy_loc_descr): New function.
32807 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
32808 (prepend_loc_descr_to_each): New function.
32809 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
32810 instead of add_loc_descr_to_each if the first argument is single
32811 location list and the second has multiple.
32812 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
32813 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
32814 when looking for variable value which doesn't have other location info.
32815 (loc_list_from_tree): Formatting fix.
32816 (gen_array_type_die): Simplify DW_AT_string_length handling.
32817 (adjust_string_types): Remove.
32818 (gen_subprogram_die): Don't call adjust_string_types nor test/set
32819 string_types. Call resolve_variable_values.
32820 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
32821 (resolve_addr_in_expr): Likewise. Add A argument.
32822 (copy_deref_exprloc): Remove deref argument. Adjust for the
32823 original expression being DW_OP_GNU_variable_value with optionally
32824 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
32825 optionally after it.
32826 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
32827 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
32828 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
32829 (variable_value_hasher::hash, variable_value_hasher::equal): New
32831 (resolve_variable_value_in_expr, resolve_variable_value,
32832 resolve_variable_values, note_variable_value_in_expr,
32833 note_variable_value): New functions.
32834 (dwarf2out_early_finish): Call note_variable_value on all toplevel
32837 2017-02-24 Jakub Jelinek <jakub@redhat.com>
32840 * opts.h (handle_generated_option): Add GENERATED_P argument.
32841 * opts-common.c (handle_option): Adjust function comment.
32842 (handle_generated_option): Add GENERATED_P argument, pass it to
32844 (control_warning_option): Pass false to handle_generated_option
32846 * opts.c (maybe_default_option): Pass true to handle_generated_option
32848 * optc-gen.awk: Likewise.
32850 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32852 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
32853 a REG, look at the REG it is a SUBREG of.
32854 (splitter for cmpeqsi_t): Ditto.
32856 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32858 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
32859 the special USEs with the pattern of the insn, not the insn itself.
32861 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
32864 * doc/invoke.texi: Document -mload-store-pairs.
32866 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32867 Sandra Loosemore <sandra@codesourcery.com>
32869 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
32870 argument isn't a CONST_INT.
32871 (nios2_alternate_compare_const): Assert op is a CONST_INT.
32872 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
32873 (nios2_validate_compare): Bypass alternate compare logic if *op2
32874 is not a CONST_INT.
32875 (ldstwm_operation_p): Return false if first_base is not a REG or
32876 if first_offset is not a CONST_INT.
32878 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32880 * config/cris/cris.md: Use correct operand in a define_peephole2.
32882 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32884 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
32886 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32888 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
32889 this_insn if it is an INSN or JUMP_INSN.
32890 (force_offsettable): Look at base, not at addr.
32891 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
32892 on things that aren't necessarily CONST_INTs.
32894 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
32896 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
32897 -mfpmath=sse is the default also for x86-32 targets with SSE2
32898 instruction set when @option{-ffast-math} is enabled
32900 2017-02-24 Jeff Law <law@redhat.com>
32902 PR rtl-optimizatoin/79286
32903 * ira.c (update_equiv_regs): Drop may_trap_p exception to
32906 2017-02-24 Richard Biener <rguenther@suse.de>
32908 PR tree-optimization/79389
32909 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
32912 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
32914 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
32915 function comment to reflect reality.
32916 (loop_exits_before_overflow): Fix typo in function description.
32918 2017-02-24 Richard Biener <rguenther@suse.de>
32920 PR tree-optimization/79389
32921 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
32922 properly that a threading opportunity exists. Detect conditional
32923 copy/constant propagation opportunities.
32925 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
32927 * config/visium/visium.md (type): Add trap.
32928 (b): New mode attribute.
32929 (*btst): Rename into...
32930 (*btst<mode>): ...this and adjust.
32931 (*cbranchsi4_btst_insn): Rename into...
32932 (*cbranch<mode>4_btst_insn): ...this and adjust.
32933 (trap): New define_insn.
32935 2017-02-23 Jakub Jelinek <jakub@redhat.com>
32937 PR tree-optimization/79389
32938 * ifcvt.c (struct noce_if_info): Add rev_cond field.
32939 (noce_reversed_cond_code): New function.
32940 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
32941 reversed_comparison_code. Formatting fix.
32942 (noce_try_store_flag): Test rev_cond != NULL in addition to
32943 reversed_comparison_code.
32944 (noce_try_store_flag_constants): Likewise.
32945 (noce_try_store_flag_mask): Likewise.
32946 (noce_try_addcc): Use rev_cond if non-NULL instead of
32947 reversed_comparison_code.
32948 (noce_try_cmove_arith): Likewise. Formatting fixes.
32949 (noce_try_minmax, noce_try_abs): Clear rev_cond.
32950 (noce_find_if_block): Initialize rev_cond.
32951 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
32952 instead of false as last argument never attempt to reverse it
32955 2017-02-23 Bin Cheng <bin.cheng@arm.com>
32957 PR tree-optimization/79663
32958 * tree-predcom.c (combine_chains): Process refs in reverse order
32959 only for ZERO length chains, and add explaining comment.
32961 2017-02-23 Jeff Law <law@redhat.com>
32963 PR tree-optimization/79578
32964 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
32965 in call to operand_equal_p.
32967 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
32970 * config/i386/cpuid.h: Fix another undefined behavior.
32972 2017-02-23 Richard Biener <rguenther@suse.de>
32974 PR tree-optimization/79683
32975 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
32976 vector types for data-refs.
32978 2017-02-23 Martin Liska <mliska@suse.cz>
32980 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
32982 2017-02-23 Jakub Jelinek <jakub@redhat.com>
32984 PR middle-end/79665
32985 * internal-fn.c (get_range_pos_neg): Moved to ...
32986 * tree.c (get_range_pos_neg): ... here. No longer static.
32987 * tree.h (get_range_pos_neg): New prototype.
32988 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
32989 are known to be in between 0 and signed maximum inclusive, try to
32990 expand both unsigned and signed divmod and use the cheaper one from
32993 2017-02-22 Jeff Law <law@redhat.com>
32995 PR tree-optimization/79578
32996 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
32997 to compare base operands.
32999 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
33002 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
33003 gpc_reg_operand instead of fpr_reg_operand.
33005 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
33007 * config/mips/mips.c (mips_return_in_memory): Force FP
33008 vector types to be returned in memory for o32 ABI.
33010 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33012 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
33013 instead of DW_TAG_member for static data member declarations and don't
33014 set no_linkage_name for static inline data members.
33015 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
33018 2017-02-22 Martin Liska <mliska@suse.cz>
33020 * doc/invoke.texi: Replace inequality signs with square brackets
33023 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
33026 * lra-constraints.c (simplify_operand_subreg): Handle
33027 WORD_REGISTER_OPERATIONS targets.
33029 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33032 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
33033 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
33034 elimination by swapping fld*.
33036 2017-02-22 Richard Biener <rguenther@suse.de>
33038 PR tree-optimization/79673
33039 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
33040 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
33041 irrelevant address-space qualifiers and avoiding a
33042 ADDR_SPACE_CONVERT_EXPR from fold_convert.
33044 2017-02-22 Richard Biener <rguenther@suse.de>
33046 PR tree-optimization/79666
33047 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
33048 to not symbolically negate if that may introduce undefined
33051 2017-02-22 Martin Liska <mliska@suse.cz>
33054 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
33055 * data-streamer-out.c (streamer_write_gcov_count_stream):
33057 * value-prof.c (stream_out_histogram_value): Make assert more
33058 precise based on type of counter.
33060 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
33063 * config/i386/i386.md (standard_x87sse_constant_load splitter):
33064 Use nonimmediate_operand instead of memory_operand for operand 1.
33065 (float-extend standard_x87sse_constant_load splitter): Ditto.
33067 2017-02-21 Jeff Law <law@redhat.com>
33069 PR tree-optimization/79621
33070 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
33071 blocks with edges to themselves.
33073 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33076 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
33077 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
33078 Use gimple_call_builtin_p.
33081 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
33082 on temporarily removed DEBUG_INSNs.
33084 PR tree-optimization/79649
33085 * tree-loop-distribution.c (classify_partition): Give up on
33086 non-generic address space loads/stores.
33088 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
33090 * doc/loop.texi (Loop manipulation): Remove nonexistent
33091 tree_ssa_loop_version from the documentation.
33092 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
33094 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33097 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
33098 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
33099 * config/rs6000/rs6000.c: Include except.h.
33100 (rs6000_expand_split_stack_prologue): Call
33101 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
33103 2017-02-21 Martin Jambor <mjambor@suse.cz>
33106 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
33107 have been analyzed.
33109 2017-02-21 Martin Jambor <mjambor@suse.cz>
33111 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
33112 for backward compatibility only.
33113 * doc/invoke.texi (Option Summary): Remove all references to
33114 -fipa-cp-alignment.
33116 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
33120 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33122 * lra-constraints.c (curr_insn_transform): Handle
33123 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
33125 2017-02-21 Martin Liska <mliska@suse.cz>
33127 * config/i386/i386.opt: Replace -masm-dialect with -masm.
33129 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
33131 PR translation/79638
33132 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
33134 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
33137 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
33138 (arm_function_ok_for_sibcall): Return false for an indirect call by
33139 descriptor if all the argument registers are used.
33140 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
33141 alignment of the function.
33143 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33145 PR tree-optimization/61441
33146 * simplify-rtx.c (simplify_const_unary_operation): For
33147 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
33148 the sNaN unmodified.
33150 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33152 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33153 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33154 instead of SYSTEM_HEADER_DIR.
33156 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
33157 Martin Liška <mliska@suse.cz>
33159 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
33160 Fix typos and grammar, use active voice, and clarify.
33162 2017-02-20 Marek Polacek <polacek@redhat.com>
33164 PR middle-end/79537
33165 * gimplify.c (gimplify_expr): Handle unused *&&L;.
33168 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
33170 2017-02-20 Jakub Jelinek <jakub@redhat.com>
33173 * config/i386/i386.c (ix86_expand_builtin): Handle
33174 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
33175 ix86_builtins_isa[fcode].isa as a requirement of those
33176 flags and any other flag in the bitmask.
33177 (ix86_init_mmx_sse_builtins): Use 0 instead of
33178 ~OPTION_MASK_ISA_64BIT as mask.
33179 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
33180 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
33181 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
33182 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
33184 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33187 * lra-constraints.c (split_reg): Check requested split mode
33188 is supported by the register.
33190 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33192 * lra-constraints.c (simplify_operand_subreg): Remove early
33195 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33198 * lra-constraints.c (curr_insn_transform): Tighten condition
33199 for converting SUBREG reloads from OP_OUT to OP_INOUT.
33201 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33204 * lra-constraints.c (curr_insn_transform): Handle
33205 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
33207 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
33210 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
33212 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
33214 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
33217 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
33220 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33222 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
33225 2017-02-18 Jakub Jelinek <jakub@redhat.com>
33227 * final.c (last_columnnum, override_columnnum): New variables.
33228 (final_start_function): Set last_columnnum, pass it to begin_prologue
33229 hook and pass 0 to dwarf2out_begin_prologue.
33230 (final_scan_insn): Update override_columnnum. Pass last_columnnum
33231 to source_line debug hook.
33232 (notice_source_line): Compute last_columnnum and for debug_column_info
33233 return true on column changes.
33234 * debug.h (struct gcc_debug_hooks): Add column argument to
33235 source_line and begin_prologue hooks.
33236 (debug_nothing_int_charstar_int_bool): Remove prototype.
33237 (debug_nothing_int_int_charstar,
33238 debug_nothing_int_int_charstar_int_bool): New prototypes.
33239 (dwarf2out_begin_prologue): Add column argument.
33240 * debug.c (do_nothing_debug_hooks): Adjust source_line and
33241 begin_prologue hooks.
33242 (debug_nothing_int_charstar_int_bool): Remove.
33243 (debug_nothing_int_int_charstar,
33244 debug_nothing_int_int_charstar_int_bool): New functions.
33245 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
33246 through to dwarf2out_source_line.
33247 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
33248 (dwarf2out_source_line): Add column argument, emit it if requested.
33249 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
33251 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33252 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33253 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
33254 through to dwarf2out_begin_prologue.
33255 (vmsdbgout_source_line): Add column argument, pass it through to
33256 dwarf2out_source_line.
33257 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
33258 dbxout_source_line caller.
33259 (dbxout_source_line): Add column argument.
33261 * common.opt (gno-column-info, gcolumn-info): New options.
33262 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
33263 (check_die): Also test for multiple DW_AT_decl_column attributes.
33264 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
33265 DW_AT_decl_column if requested.
33266 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
33268 (gen_variable_die): Likewise.
33269 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
33270 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
33273 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
33274 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
33275 (ix86_handle_option): Handle OPT_m3dnowa.
33276 * doc/invoke.texi (-m3dnowa): Document.
33277 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
33278 -m3dnowa instead of -m3dnow -march=athlon.
33281 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
33282 instead of gcc_assert for K, r and R code checks. Formatting fixes.
33284 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33287 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
33288 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
33289 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
33290 generator for vsx_xxpermdi_<mode>_be.
33291 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
33292 force big-endian semantics.
33293 (vsx_xxpermdi_<mode>_be): New define_expand with same
33294 implementation as previous version of vsx_xxpermdi_<mode>.
33296 2017-02-17 Jakub Jelinek <jakub@redhat.com>
33298 PR tree-optimization/79327
33299 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
33300 variable, its initialization and use.
33302 2017-02-17 Julia Koval <julia.koval@intel.com>
33304 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
33305 (OPTION_MASK_ISA_PKU_UNSET): New.
33306 (ix86_handle_option): Handle -mrdpid.
33307 * config/i386/cpuid.h (bit_RDPID): New.
33308 * config/i386/driver-i386.c (host_detect_local_cpu):
33309 Detect RDPID feature.
33310 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
33311 * config/i386/i386-c.c (ix86_target_macros_internal):
33313 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
33314 (ix86_valid_target_attribute_inner_p): Add "rdpid".
33315 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
33316 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
33317 * config/i386/i386.md (define_insn "rdpid"): New.
33318 * config/i386/i386.opt Add -mrdpid.
33319 * config/i386/immintrin.h (_rdpid_u32): New.
33321 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
33323 PR rtl-optimization/79541
33324 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
33325 instead of transforming it into USE.
33327 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
33329 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
33330 If HONOR_SNANS (SFmode) force the input to a register.
33331 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
33332 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
33333 an frsp or similar insn.
33335 2017-02-17 Martin Liska <mliska@suse.cz>
33337 PR rtl-optimization/79577
33338 * params.def (selsched-max-sched-times): Increase minimum to 1.
33340 2017-02-17 Martin Liska <mliska@suse.cz>
33342 PR rtl-optimization/79574
33343 * gcse.c (want_to_gcse_p): Prevent integer overflow.
33345 2017-02-17 Martin Liska <mliska@suse.cz>
33347 PR tree-optimization/79529
33348 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
33349 ssa_defined_default_def_p to handle cases which are implicitly
33351 * tree-ssa.c (ssa_defined_default_def_p): New function.
33352 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
33353 which are implicitly defined.
33354 * tree-ssa.h (ssa_defined_default_def_p): Declare.
33356 2017-02-17 Richard Biener <rguenther@suse.de>
33358 PR middle-end/79576
33359 * params.def (max-ssa-name-query-depth): Limit to 10.
33361 2017-02-17 Richard Biener <rguenther@suse.de>
33363 PR tree-optimization/79552
33364 * tree-ssa-structalias.c (visit_loadstore): Properly verify
33367 2017-02-17 Richard Biener <rguenther@suse.de>
33370 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
33372 2017-02-17 Marek Polacek <polacek@redhat.com>
33374 PR middle-end/79536
33375 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
33376 (fold_negate_expr): New wrapper.
33378 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
33380 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
33381 Correct terminology and de-emphasize pre-standard behavior.
33383 2017-02-16 Alan Modra <amodra@gmail.com>
33385 PR rtl-optimization/79286
33386 * ira.c (def_dominates_uses): New function.
33387 (update_equiv_regs): Don't create an equivalence for insns that
33388 may trap where the register def does not dominate the use.
33390 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
33392 PR rtl-optimization/78127
33393 * lra.c (lra): Call lra_eliminate before finish the loop after
33396 2017-02-16 Richard Biener <rguenther@suse.de>
33398 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
33400 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
33401 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
33402 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
33403 (isl_val_int_from_wi): New function.
33404 (extract_affine_gmp): Rename to ...
33405 (extract_affine_wi): ... this, take a widest_int.
33406 (extract_affine_int): Just wrap extract_affine_wi.
33407 (add_param_constraints): Use isl_val_int_from_wi.
33408 (add_loop_constraints): Likewise, and extract_affine_wi.
33410 2017-02-15 Jeff Law <law@redhat.com>
33412 PR middle-end/79521
33413 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
33414 ira_init_register_move_cost_if_necessary.
33416 2017-02-15 Martin Sebor <msebor@redhat.com>
33418 PR middle-end/32003
33419 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
33420 removed in a prior commit.
33422 2017-02-15 Bin Cheng <bin.cheng@arm.com>
33424 PR tree-optimization/79347
33425 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
33426 counters during peeling.
33428 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
33430 * Makefile.in (site.exp): Remove "set ISLVER".
33432 2017-02-15 Jakub Jelinek <jakub@redhat.com>
33435 * real.c (real_from_integer): Call real_convert even for decimal.
33437 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
33440 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
33442 2017-02-14 Andrew Pinski <apinski@cavium.com>
33444 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
33445 cores and change the partno/implementer to be correct.
33446 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
33447 the 'B" as the implementer.
33448 * config/aarch64/aarch64-tune.md: Regenerate.
33450 2017-02-14 Carl Love <cel@us.ibm.com>
33452 * config/rs6000/rs6000.c: Add case statement entry to make the
33453 xvcvuxdsp built-in argument unsigned.
33454 * config/rs6000/vsx.md: Fix the source and return operand types so they
33455 match the instruction definitions from the ISA document. Fix typo
33456 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
33459 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
33462 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
33463 member early_clobber_alts.
33464 * lra-lives.c (reg_early_clobber_p): New.
33465 (process_bb_lives): Use it.
33466 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
33467 (debug_operand_data): Initialize early_clobber_alts.
33468 (setup_operand_alternative): Set up early_clobber_alts.
33469 (collect_non_operand_hard_regs): Ditto. Pass early clobber
33470 alternatives to new_insn_reg.
33471 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
33473 (lra_update_insn_regno_info): Pass the new arg.
33475 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33477 PR middle-end/79505
33478 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
33479 (new_oacc_loop_raw): Don't clear already cleared fields.
33482 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
33483 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
33484 _mm512_prefetch_i64gather_ps): New inline functions and macros.
33486 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
33489 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
33491 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
33494 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
33495 the extra instruction to the right place to store 128-bit constant
33498 2017-02-14 Martin Sebor <msebor@redhat.com>
33500 PR middle-end/79448
33501 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
33502 warning for strings of unknown length.
33504 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
33506 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
33508 2017-02-14 Jeff Law <law@redhat.com>
33511 * ira-costs.c (scan_one_insn): Initialize register move costs
33512 for pseudos seen in USE/CLOBBER insns.
33514 PR tree-optimization/79095
33515 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
33516 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
33517 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
33518 if the operands are known to be not equal, then the resulting range
33520 (intersect_ranges): If the new range is ~[0,0] and the old range is
33521 wide, then prefer ~[0,0].
33522 * tree-vrp.c (overflow_comparison_p_1): New function.
33523 (overflow_comparison_p): New function.
33524 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
33525 if NAME is used in an overflow test.
33526 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
33527 overflow check that can be expressed as an equality test, then adjust
33528 ops to be that equality test.
33530 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33532 * config/s390/s390-builtin-types.def: Remove flags argument.
33533 * config/s390/s390.c (s390_init_builtins): Likewise.
33535 2017-02-14 Martin Liska <mliska@suse.cz>
33537 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
33538 vector. Fix trailing white spaces.
33540 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
33542 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
33545 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33547 PR rtl-optimization/68664
33548 * config/arm/arm.c (arm_sched_can_speculate_insn):
33549 New function. Declare prototype.
33550 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
33552 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33554 PR rtl-optimization/68664
33555 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
33557 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
33559 2017-02-14 Amit Pawar <amit.pawar@amd.com>
33561 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
33562 max skip bytes for function, loop and jump.
33564 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33566 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
33567 ABS_EXPR for gimple dump.
33569 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33572 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
33574 PR tree-optimization/79408
33575 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
33576 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
33577 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
33578 also if rhs1 is INTEGER_CST.
33580 2017-02-14 Richard Biener <rguenther@suse.de>
33582 PR middle-end/79432
33583 * tree-into-ssa.c (insert_phi_nodes): When the function can
33584 have abnormal edges rewrite SSA names with broken use-def
33585 dominance out of SSA and register them for PHI insertion.
33587 2017-02-13 Martin Sebor <msebor@redhat.com>
33589 PR middle-end/79496
33590 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
33591 clearing info.nowrite flag when snprintf size argument is a range.
33593 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33595 * cprop.c (cprop_jump): Add missing space in string literal.
33596 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
33597 (get_constraint_for_component_ref): Likewise.
33598 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
33599 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
33600 * lra-constraints.c (process_alt_operands): Likewise.
33601 * ipa-inline.c (inline_small_functions): Likewise.
33602 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
33603 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
33604 * trans-mem.c (diagnose_tm_1_op): Likewise.
33605 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
33606 (grid_parallel_clauses_gridifiable): Likewise.
33608 * config/nvptx/mkoffload.c (process): Add space in between
33611 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
33612 "MOD4_SSE_REGS" and "ALL_REGS".
33614 * spellcheck.c (test_data): Add , in between "foo" and "food".
33616 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33619 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
33620 boundary crossing check and subsequent code generation agree.
33622 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33624 * config/aarch64/aarch64.c (has_memory_op): Delete.
33625 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
33628 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33630 PR rtl-optimization/79388
33631 PR rtl-optimization/79450
33632 * combine.c (distribute_notes): When removing TEM_INSN for which
33633 corresponding dest has last value recorded, invalidate that last
33636 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33638 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
33639 of explicit '@'. Add missing assembly comment marker on branch costs
33642 2017-02-13 Nathan Sidwell <nathan@acm.org>
33644 * gengtype-lex.l (<in_struct>): Add '/'.
33646 2017-02-13 Martin Liska <mliska@suse.cz>
33649 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
33651 2017-02-13 Richard Biener <rguenther@suse.de>
33653 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
33655 * configure: Re-generate.
33656 * config.in: Likewise.
33657 * graphite-dependences.c: Simplify as if
33658 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
33659 * graphite-isl-ast-to-gimple.c: Likewise.
33660 * graphite-optimize-isl.c: Likewise.
33661 * graphite-poly.c: Likewise.
33662 * graphite-sese-to-poly.c: Likewise.
33663 * graphite.h: Likewise.
33664 * toplev.c: Include isl/version.h and use isl_version () for
33665 printing the ISL version.
33666 * doc/install.texi: Update ISL requirement.
33668 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33670 * doc/standards.texi (Standards): Update reference to
33673 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33675 * doc/extend.texi (Named Address Spaces): sourceware.org now
33677 * doc/install.texi (Binaries): Ditto.
33680 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33682 * doc/cpp.texi: Replace "stringify"/"stringification" with C
33683 standard terminology "stringize"/"stringizing" throughout.
33684 * doc/cppinternals.texi: Likewise.
33686 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33688 * doc/extend.texi: Fix some spelling mistakes and typos.
33689 * doc/invoke.texi: Likewise.
33691 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
33694 * params.def (inline-min-speedup) Change from 10 to 8.
33696 2017-02-11 Jakub Jelinek <jakub@redhat.com>
33698 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
33701 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
33704 * ipa-inline-analysis.c (get_minimal_bb): New function.
33705 (record_modified): Use it.
33706 (remap_edge_change_prob): Handle also ancestor functions.
33708 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
33710 * doc/contrib.texi (Contributors): Remove broken link into
33711 the Mauve CVS repository.
33713 2017-02-11 Jakub Jelinek <jakub@redhat.com>
33715 PR middle-end/79454
33716 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
33717 result computation whenever lhs doesn't have vector mode, not
33718 just when it has BLKmode.
33720 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
33722 * doc/makefile.texi (profiledbootstrap): Refer to the
33723 installation instructions only in textual form.
33725 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33728 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
33730 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
33732 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
33733 (Specific): Update mingw-w64 reference.
33735 (Specific): Remove broken link to Renesas RX processor.
33737 2017-02-10 Richard Biener <rguenther@suse.de>
33739 * toplev.c (process_options): Do not mention obsolete graphite
33740 options when printing sorry message about missing graphite support.
33741 Mention -floop-nest-optimize.
33743 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
33745 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
33746 (vtst_p16): Likewise.
33747 (vtstq_p8): Likewise.
33748 (vtstq_p16): Likewise.
33750 (vtstq_p64): Likewise.
33751 * config/arm/arm_neon.h (vgetq_lane_p64): New.
33752 (vset_lane_p64): New.
33753 (vsetq_lane_p64): New.
33755 2017-02-10 Jakub Jelinek <jakub@redhat.com>
33757 PR tree-optimization/79411
33758 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
33759 stmt operands are SSA_NAMEs used in abnormal phis.
33760 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
33763 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
33766 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
33769 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
33771 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
33773 2017-02-09 Jakub Jelinek <jakub@redhat.com>
33775 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
33779 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
33780 not arbitrary TREE_CONSTANT.
33783 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
33784 "omp declare target link" attribute unless is_global_var.
33785 * omp-offload.c (find_link_var_op): Likewise.
33787 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
33788 Chung-Lin Tang <cltang@codesourcery.com>
33790 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
33792 (gimplify_adjust_omp_clauses): Don't delete TILE.
33793 (gimplify_omp_for): Deal with TILE.
33794 * internal-fn.c (expand_GOACC_TILE): New function.
33795 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
33797 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
33798 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
33800 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
33801 avoid DIV for outermost collapse var.
33802 (expand_oacc_for): Insert tile element loop as needed. Adjust.
33803 Remove out of date comments, fix whitespace.
33804 * omp-general.c (omp_extract_for_data): Deal with tiling.
33805 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
33806 adjust OLF_DIM_BASE value.
33807 (struct omp_for_data): Add tiling field.
33808 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
33809 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
33810 for auto loops. Remove default auto determining, moved to
33811 oacc_loop_fixed_partitions.
33812 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
33813 stmts, add e_mask field.
33814 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
33815 (oacc_thread_numbers): Use oacc_dim_call.
33816 (oacc_xform_tile): New.
33817 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
33818 (finish_oacc_loop): Adjust for ifns vector.
33819 (oacc_loop_discover_walk): Append loop abstraction sites to list,
33820 add case for GOACC_TILE fns.
33821 (oacc_loop_xform_loop): Delete.
33822 (oacc_loop_process): Iterate over call list directly, and add
33823 handling for GOACC_TILE fns.
33824 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
33826 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
33827 vector partitioning to outer loops. Assign 2 partitions to loops
33828 when available. Add TILE handling.
33829 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
33830 (execite_oacc_device_lower): Process GOACC_TILE fns,
33831 ignore unknown specs.
33832 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
33833 * tree.c (omp_clause_num_ops): Adjust TILE ops.
33834 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
33836 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
33838 * configure.ac (ACX_BUGURL): Update.
33839 * configure: Regenerate.
33841 2017-02-09 Richard Biener <rguenther@suse.de>
33843 PR tree-optimization/69823
33844 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
33845 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
33847 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
33849 * config/arc/arc-c.def: Add __NPS400__ definition.
33850 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
33851 (TARGET_NPS400): Define.
33853 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
33855 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
33857 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
33858 pointer, arch_info.
33859 (arc_cpu_types): Fill the arch_info field with a pointer into the
33860 arc_arch_types table.
33861 (arc_selected_cpu): Declare.
33862 * config/arc/arc.c (arc_selected_cpu): Make global.
33863 (arc_selected_arch): Delete.
33864 (arc_base_cpu): Delete.
33865 (arc_override_options): Remove references to deleted variables,
33866 update access to arch information.
33867 (ARC_OPT): Update access to arch information.
33868 (ARC_OPTX): Likewise.
33869 * config/arc/arc.h (arc_base_cpu): Remove declaration.
33870 (TARGET_ARC600): Update access to arch information.
33871 (TARGET_ARC601): Likewise.
33872 (TARGET_ARC700): Likewise.
33873 (TARGET_EM): Likewise.
33874 (TARGET_HS): Likewise.
33875 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
33878 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
33881 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
33882 condition/operands for integer GE/LE/GEU/LEU operations.
33884 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
33886 PR translation/79397
33887 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
33890 2017-02-08 Martin Jambor <mjambor@suse.cz>
33893 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
33894 whether allocation happened.
33895 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
33896 nothing was allocated.
33898 2017-02-08 Jakub Jelinek <jakub@redhat.com>
33900 PR tree-optimization/79408
33901 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
33902 constant, but SSA_NAME with a known integer range, use the minimum
33903 of that range instead of op1 to determine if modulo can be replaced
33904 with its first operand.
33906 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33908 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
33910 2017-02-08 Richard Biener <rguenther@suse.de>
33912 PR tree-optimization/71824
33913 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
33914 Check all loops contained in the merged region.
33916 2017-02-07 Andrew Pinski <apinski@cavium.com>
33918 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
33920 2017-02-07 Andrew Pinski <apinski@cavium.com>
33922 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
33923 (thunderxt88): Likewise.
33924 (thunderxt81): Disable LSE and change v8.1 to v8.
33925 (thunderxt83): Likewise.
33927 2017-02-07 Jakub Jelinek <jakub@redhat.com>
33928 Richard Biener <rguenther@suse.de>
33930 PR middle-end/79399
33931 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
33932 type from int to size_t.
33933 * ira-costs.c (struct_costs_size): Change type from int to size_t.
33935 2017-02-07 Jakub Jelinek <jakub@redhat.com>
33937 PR rtl-optimization/79386
33938 * cprop.c (bypass_conditional_jumps): Initialize
33939 bypass_last_basic_block already before splitting bbs after
33940 unconditional traps...
33941 (bypass_conditional_jumps): ... rather than here.
33944 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
33945 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
33946 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
33947 fix -masm=intel patterns.
33949 2017-02-07 Richard Biener <rguenther@suse.de>
33951 PR tree-optimization/79256
33952 PR middle-end/79278
33953 * builtins.c (get_object_alignment_2): Use min_align_of_type
33954 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
33955 and ADJUST_FIELD_ALIGN.
33957 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
33959 * doc/tm.texi: Regenerate.
33960 * stor-layout.c (layout_decl): Adjust.
33961 (update_alignment_for_field): Likewise.
33962 (place_field): Likewise.
33963 (min_align_of_type): Likewise.
33964 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
33965 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
33966 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
33967 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
33968 * config/frv/frv.c (frv_adjust_field_align): Likewise.
33969 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
33970 * config/i386/i386.c (x86_field_alignment): Likewise.
33971 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
33972 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
33973 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
33974 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
33975 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
33976 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
33980 2017-01-30 Richard Biener <rguenther@suse.de>
33982 PR tree-optimization/79256
33983 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
33984 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
33987 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
33989 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
33990 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
33991 builtins to SImode and emit a zero-extend, if necessary.
33993 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33995 * docs/invoke.texi (RISC-V Options): Alphabetize.
33997 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33999 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
34002 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34004 * config/riscv/riscv.c: New file.
34005 * common/config/riscv/riscv-common.c: Likewise.
34006 * config.gcc: Likewise.
34007 * config/riscv/constraints.md: Likewise.
34008 * config/riscv/elf.h: Likewise.
34009 * config/riscv/generic.md: Likewise.
34010 * config/riscv/linux.h: Likewise.
34011 * config/riscv/multilib-generator: Likewise.
34012 * config/riscv/peephole.md: Likewise.
34013 * config/riscv/pic.md: Likewise.
34014 * config/riscv/predicates.md: Likewise.
34015 * config/riscv/riscv-builtins.c: Likewise.
34016 * config/riscv/riscv-c.c: Likewise.
34017 * config/riscv/riscv-ftypes.def: Likewise.
34018 * config/riscv/riscv-modes.def: Likewise.
34019 * config/riscv/riscv-opts.h: Likewise.
34020 * config/riscv/riscv-protos.h: Likewise.
34021 * config/riscv/riscv.h: Likewise.
34022 * config/riscv/riscv.md: Likewise.
34023 * config/riscv/riscv.opt: Likewise.
34024 * config/riscv/sync.md: Likewise.
34025 * config/riscv/t-elf-multilib: Likewise.
34026 * config/riscv/t-linux: Likewise.
34027 * config/riscv/t-linux-multilib: Likewise.
34028 * config/riscv/t-riscv: Likewise.
34029 * configure.ac: Likewise.
34030 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
34031 Waterman as RISC-V maintainers.
34032 * doc/install.texi: Add RISC-V entries.
34033 * doc/invoke.texi: Add RISC-V options section.
34034 * doc/md.texi: Add RISC-V constraints section.
34035 * configure: Regenerated.
34037 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
34040 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
34041 false values to be constant vectors with all 0 or all 1 bits set.
34042 (vcondu<mode><mode>): Likewise.
34043 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
34045 (fpmask_comparison_operator): Update comment.
34046 (vecint_comparison_operator): New predicate.
34047 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
34048 vector conditionals when the true and false values are constant
34049 vectors with all 0 bits or all 1 bits set.
34051 2017-02-06 Martin Sebor <msebor@redhat.com>
34053 PR tree-optimization/79376
34054 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
34056 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
34058 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
34059 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
34060 to simplify split condition.
34062 2017-02-06 Jakub Jelinek <jakub@redhat.com>
34064 * omp-expand.c (oxpand_omp_atomic_fetch_op,
34065 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
34068 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
34070 PR rtl-optimization/68664
34071 * target.def (can_speculate_insn): New hook.
34072 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
34073 * doc/tm.texi: Regenerate.
34074 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
34075 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
34076 (rs6000_sched_can_speculate_insn): New function.
34078 2017-02-06 Jakub Jelinek <jakub@redhat.com>
34080 PR tree-optimization/79284
34081 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
34082 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
34083 vectorizable_mask_load_store, vectorizable_operation,
34084 vect_is_simple_cond, get_same_sized_vectype): Use it instead
34085 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
34086 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
34087 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
34088 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
34089 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
34090 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
34091 is_gimple_assign (stmt). Replace another such test with
34092 is_gimple_assign (stmt).
34094 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
34097 * config/avr/avr.c (rtl-iter.h): Include it.
34098 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
34099 (avr_legitimate_combined_insn): ...and implementation.
34101 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
34103 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
34104 * config/s390/s390.c (s390_const_operand_ok)
34105 (s390_canonicalize_comparison, s390_extract_part)
34106 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
34107 (s390_contiguous_bitmask_p, s390_rtx_costs)
34108 (legitimize_pic_address): Likewise.
34109 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
34110 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
34111 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
34112 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
34113 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
34115 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
34117 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
34118 REGNO($0) == REGNO($1).
34120 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34122 * config/s390/linux.h(SIZE_TYPE): Add comment.
34124 2017-02-06 Julian Brown <julian@codesourcery.com>
34125 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34126 Virendra Pathak <virendra.pathak@broadcom.com>
34128 * config/aarch64/aarch64-cores.def: Change the scheduler
34130 * config/aarch64/aarch64.md: Include thunderx2t99.md.
34131 * config/aarch64/thunderx2t99.md: New file.
34133 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34135 * doc/standards.texi (Go Language): Update link to language
34138 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
34140 * tree-eh.c (lower_resx): Sanitize profile.
34141 (cleanup_empty_eh_move_lp): Likewise.
34143 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
34146 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
34148 * cfgloopmanip.h (loop_version): Update prototype.
34149 * modulo-sched.c (sms_schedule): Update call of loop_version.
34150 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
34151 * tree-parloops.c (gen_parallel_loop): Likewise.
34152 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
34153 * tree-ssa-loop-split.c (split_loop): Likewise.
34154 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
34155 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
34157 2017-02-05 Martin Liska <mliska@suse.cz>
34160 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
34162 (print_operand_address): Initialize a struct to zero.
34164 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34166 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
34167 garbage collector only in textual form.
34169 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34171 * doc/extend.texi (x86 specific memory model extensions for
34172 transactional memory): Simplify a phrase.
34174 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
34177 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
34178 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
34179 (atomic_storedi_1): Likewise.
34181 2017-02-04 Jakub Jelinek <jakub@redhat.com>
34183 PR tree-optimization/79338
34184 * tree-parloops.c (gather_scalar_reductions): Don't call
34185 vect_analyze_loop_form for loop->inner before destroying loop's
34188 2017-02-03 Martin Sebor <msebor@redhat.com>
34190 PR tree-optimization/79327
34191 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
34192 when precision has resulted in leading zeros.
34193 (format_integer): Adjust the likely counter to assume an unknown
34194 argument that may be zero is non-zero.
34196 2017-02-03 Jason Merrill <jason@redhat.com>
34199 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
34200 avoid copying non-taken branch.
34202 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34204 PR tree-optimization/79340
34205 * tree-vect-loop.c (vectorizable_reduction): Release
34206 vec_defs elements after safe_splicing them into other vectors.
34209 PR tree-optimization/79327
34210 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
34211 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
34213 (format_integer): Use wide_int_to_tree instead of build_int_cst
34214 + to_?hwi. If argmin is NULL, just set argmin and argmax to
34215 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
34216 of shortest and longest sequence.
34218 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
34220 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
34221 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
34223 2017-02-03 Walter Lee <walt@tilera.com>
34226 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
34227 after initial stackframe link reg save.
34228 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
34230 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34233 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
34234 wu for stxssp alternative.
34236 2017-02-03 Martin Sebor <msebor@redhat.com>
34238 PR tree-optimization/79352
34239 * gimple-fold.c (get_range_strlen): Add argument.
34240 (get_range_strlen): Change return type to bool.
34241 (get_maxval_strlen): Pass in a dummy argument.
34242 * gimple-fold.h (get_range_strlen): Change return type to bool.
34243 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
34244 * tree.h (array_at_struct_end_p): Add argument.
34245 * tree.c (array_at_struct_end_p): Handle it.
34247 2017-02-03 Martin Liska <mliska@suse.cz>
34250 * multiple_target.c (create_dispatcher_calls): Redirect edge
34251 from a caller of a dispatcher.
34252 (expand_target_clones): Make the clones local.
34253 (ipa_target_clone): Do both target clones and resolvers.
34254 (ipa_dispatcher_calls): Remove the pass.
34255 (pass_dispatcher_calls::gate): Likewise.
34256 (make_pass_dispatcher_calls): Likewise.
34257 * passes.def (pass_target_clone): Put as very first IPA early
34260 2017-02-03 Martin Liska <mliska@suse.cz>
34262 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
34263 in case of a function with ifunc attribute.
34265 2017-02-03 Martin Liska <mliska@suse.cz>
34267 * cgraph.c (cgraph_node::dump): Dump function version info.
34268 * symtab.c (symtab_node::dump_base): Add missing new line.
34270 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34272 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
34273 (ifcombine_ifandif): Use it.
34275 2017-02-03 Martin Liska <mliska@suse.cz>
34277 * doc/invoke.texi: Document default value for
34278 use-after-scope-direct-emission-threshold.
34280 2017-02-03 Martin Liska <mliska@suse.cz>
34282 PR tree-optimization/79339
34283 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
34284 (format_floating): Likewise.
34286 2017-02-03 Martin Liska <mliska@suse.cz>
34289 * ipa-prop.c (ipa_node_params_t::insert): Remove current
34291 (ipa_node_params_t::remove): Likewise.
34292 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
34293 initialization from removed ipa_node_params_t::insert.
34294 (ipa_node_params::~ipa_node_params): Move from removed
34295 ipa_node_params_t::release.
34296 * symbol-summary.h (symbol_summary::m_released): New member.
34297 Do not release a summary twice. Do not allow to call finalizer
34298 for types of a summary that live in GGC memory.
34300 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34302 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
34305 2017-02-02 Martin Sebor <msebor@redhat.com>
34307 PR middle-end/79275
34308 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
34309 (format_string): Tighten up the range of output for non-constant
34310 strings and correct the expected range for wide non-constant strings.
34312 2017-02-02 Martin Sebor <msebor@redhat.com>
34314 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
34316 PR middle-end/32003
34317 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
34319 (-fdump-tree-@var): Add to index and document how to come up
34320 with pass-specific option and dump file names.
34321 (-fdump-passes): Clarify where to look for output.
34323 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34325 PR middle-end/77445
34326 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
34327 statistics of the analyzed path; allow threading for speed when
34328 any of BBs along the path are optimized for speed.
34330 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
34332 PR middle-end/78468
34333 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
34334 settings of the virtual registers.
34337 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
34339 * explow.c (get_dynamic_stack_size): Take known alignment of stack
34340 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
34343 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34345 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
34346 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
34348 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34350 * config/s390/s390.md: Add missing comments with the expanded
34352 * config/s390/vector.md: Likewise.
34353 * config/s390/vx-builtins.md: Likewise.
34355 2017-02-02 Jakub Jelinek <jakub@redhat.com>
34358 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
34359 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
34360 conditions on a single line.
34362 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34364 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
34365 __S390_VX__ to __VX__.
34367 2017-02-01 Andrew Pinski <apinski@cavium.com>
34369 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
34370 stmt_info to record_stmt_cost.
34371 (vect_get_known_peeling_cost): Pass stmt_info if known to
34373 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
34374 cpu_vector_cost field into
34375 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
34376 field into vec_int_stmt_cost and vec_fp_stmt_cost.
34377 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
34378 splitting of scalar_stmt_cost and vec_stmt_cost.
34379 (thunderx_vector_cost): Likewise.
34380 (cortexa57_vector_cost): LIkewise.
34381 (exynosm1_vector_cost): Likewise.
34382 (xgene1_vector_cost): Likewise.
34383 (thunderx2t99_vector_cost): Improve after the splitting of the two
34385 (aarch64_builtin_vectorization_cost): Update for the splitting of
34386 scalar_stmt_cost and vec_stmt_cost.
34388 2017-02-01 Torvald Riegel <triegel@redhat.com>
34389 Richard Henderson <rth@redhat.com>
34391 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
34392 conditional on existance of a fast atomic load.
34393 * optabs-query.c (can_atomic_load_p): New function.
34394 * optabs-query.h (can_atomic_load_p): Declare it.
34395 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
34396 no fast atomic load is available for the particular size of access.
34397 (expand_atomic_compare_and_swap): Likewise.
34398 (expand_atomic_load): Likewise.
34399 (expand_atomic_store): Likewise.
34400 (expand_atomic_fetch_op): Likewise.
34401 * testsuite/lib/target-supports.exp
34402 (check_effective_target_sync_int_128): Remove x86 because it provides
34403 no fast atomic load.
34404 (check_effective_target_sync_int_128_runtime): Likewise.
34406 2017-02-01 Richard Biener <rguenther@suse.de>
34408 * graphite.c: Include tree-vectorizer.h for find_loop_location.
34409 (graphite_transform_loops): Provide opt-info for optimized nests.
34410 * tree-parloop.c (parallelize_loops): Provide opt-info for
34411 parallelized loops.
34413 2017-02-01 Richard Biener <rguenther@suse.de>
34415 PR middle-end/79315
34416 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
34417 was not set before.
34419 2017-02-01 Richard Biener <rguenther@suse.de>
34421 PR tree-optimization/71824
34422 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34423 Verify the loops are valid in the merged SESE region.
34424 (scop_detection::can_represent_loop_1): Check analyzing the
34425 evolution of the number of iterations in the region succeeds.
34427 2017-01-31 Ian Lance Taylor <iant@golang.org>
34429 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
34430 REG_ARGS_SIZE note to 32-bit push insns and call insn.
34432 2017-01-31 David Malcolm <dmalcolm@redhat.com>
34434 PR preprocessor/79210
34435 * input.c (get_substring_ranges_for_loc): Replace line_width
34436 assertion with error-handling.
34438 2017-01-31 Richard Biener <rguenther@suse.de>
34440 PR tree-optimization/77318
34441 * graphite-sese-to-poly.c (extract_affine): Fix assert.
34442 (create_pw_aff_from_tree): Take loop parameter.
34443 (add_condition_to_pbb): Pass loop of the condition to
34444 create_pw_aff_from_tree.
34446 2017-01-31 Jakub Jelinek <jakub@redhat.com>
34448 * config/s390/s390.c (s390_asan_shadow_offset): New function.
34449 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
34451 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
34455 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
34457 (convert_int_to_float128): Likewise.
34458 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
34459 (convert_int_to_float128): Likewise.
34460 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
34461 (UNSPEC_IEEE128_CONVERT): Likewise.
34462 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
34463 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
34464 Use local variables for IBM extended format.
34465 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
34466 (fix_trunc<mode>si2_fprs): Likewise.
34467 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
34468 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
34469 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
34470 to know that we can now have integers of all sizes in vector
34472 (fix<uns>_<mode>di2_hw): Likewise.
34473 (float<uns>_<mode>si2_hw): Likewise.
34474 (fix_<mode>si2_hw): Likewise.
34475 (fixuns_<mode>si2_hw): Likewise.
34476 (float<uns>_<mode>di2_hw): Likewise.
34477 (float_<mode>di2_hw): Likewise.
34478 (float_<mode>si2_hw): Likewise.
34479 (floatuns_<mode>di2_hw): Likewise.
34480 (floatuns_<mode>si2_hw): Likewise.
34481 (xscvqp<su>wz_<mode>): Delete, no longer used.
34482 (xscvqp<su>dz_<mode>): Likewise.
34483 (xscv<su>dqp_<mode>): Likewise.
34484 (ieee128_mfvsrd_64bit): Likewise.
34485 (ieee128_mfvsrd_32bit): Likewise.
34486 (ieee128_mfvsrwz): Likewise.
34487 (ieee128_mtvsrw): Likewise.
34488 (ieee128_mtvsrd_64bit): Likewise.
34489 (ieee128_mtvsrd_32bit): Likewise.
34491 2017-01-31 Martin Liska <mliska@suse.cz>
34494 * ipa-prop.c (ipa_free_all_node_params): Call release method
34495 instead of ~sumbol_summary to not to trigger double times
34498 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
34500 PR tree-optimization/71691
34501 * bitmap.h (class auto_bitmap): New.
34502 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
34503 is_maybe_undefined instead of ssa_undefined_value_p.
34505 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34507 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
34508 __S390_ARCH_LEVEL__ to __ARCH__.
34510 2017-01-31 Jakub Jelinek <jakub@redhat.com>
34512 PR tree-optimization/79267
34513 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
34514 if should_remove_lhs_p is true.
34516 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
34519 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
34520 (add_alignment_attribute): New.
34521 (base_type_die): Add alignment attribute.
34522 (subrange_type_die): Likewise.
34523 (modified_type_die): Likewise.
34524 (gen_array_type_die): Likewise.
34525 (gen_descr_array_type_die: Likewise.
34526 (gen_enumeration_type_die): Likewise.
34527 (gen_subprogram_die): Likewise.
34528 (gen_variable_die): Likewise.
34529 (gen_field_die): Likewise.
34530 (gen_ptr_to_mbr_type_die): Likewise.
34531 (gen_struct_or_union_type_die): Likewise.
34532 (gen_subroutine_type_die): Likewise.
34533 (gen_typedef_die): Likewise.
34534 (base_type_cmp): Compare alignment attribute.
34536 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34539 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
34540 (setb_unsigned) New pattern for setb with CCUNS.
34541 * config/rs6000/rs6000.c (expand_block_compare): Use a different
34542 subfc./subfe sequence to avoid overflow problems. Generate a
34543 shorter sequence with cmpld/setb for power9.
34544 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
34545 for generating subfc. instruction.
34546 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
34547 now uses this instruction.
34549 2017-01-30 Ian Lance Taylor <iant@google.com>
34552 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
34553 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
34555 2017-01-30 Martin Sebor <msebor@redhat.com>
34557 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
34558 Move constant to the right of a relational operator.
34559 (get_mpfr_format_length, format_character, format_string): Ditto.
34560 (should_warn_p, maybe_warn): Same.
34562 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
34564 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
34567 * asan.c (get_translation_unit_decl): Remove function.
34568 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
34570 2017-01-30 Martin Liska <mliska@suse.cz>
34572 PR gcov-profile/79259
34573 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
34574 -fprofile-generate.
34576 2017-01-30 Martin Liska <mliska@suse.cz>
34579 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
34580 Initialize variables with NULL value.
34582 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
34585 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
34587 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
34589 2017-01-30 Richard Biener <rguenther@suse.de>
34591 PR tree-optimization/79276
34592 * tree-vrp.c (process_assert_insertions): Properly adjust common
34593 when removing a duplicate.
34595 2017-01-30 Richard Biener <rguenther@suse.de>
34597 PR tree-optimization/79256
34598 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34599 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34601 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
34603 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
34606 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
34607 ("*r<noxa>sbg_<mode>_sll_bitmask")
34608 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
34609 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
34610 Use contiguous_bitmask_nowrap_operand.
34612 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34615 * config/rs6000/altivec.h (vec_xl): Revise #define.
34616 (vec_xst): Likewise.
34618 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
34620 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
34622 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
34624 PR rtl-optimization/79194
34625 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
34626 traps before call to bypass_conditional_jumps.
34628 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34630 PR tree-optimization/71374
34631 * lra-constraints.c (check_conflict_input_operands): New.
34632 (match_reload): Use it.
34634 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34637 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
34638 account to calculate conflict_set.
34640 2017-01-27 Bin Cheng <bin.cheng@arm.com>
34642 PR rtl-optimization/78559
34643 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
34644 other_insn in combine.
34646 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
34648 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
34649 uint16_type_node for BT_UINT16.
34651 2017-01-27 David Malcolm <dmalcolm@redhat.com>
34653 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
34654 "RTL Tests" to menu.
34655 (GIMPLE Tests): New node.
34656 (RTL Tests): New node.
34658 2017-01-27 Richard Biener <rguenther@suse.de>
34660 PR tree-optimization/79245
34661 * tree-loop-distribution.c (distribute_loop): Apply cost
34662 modeling also to detected patterns.
34664 2017-01-27 Richard Biener <rguenther@suse.de>
34666 PR tree-optimization/71433
34667 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
34668 (compare_assert_loc): New function.
34669 (process_assert_insertions): Sort and optimize assert locations
34670 to remove duplicates and push down identical assertions on
34671 edges to their destination block.
34673 2017-01-27 Richard Biener <rguenther@suse.de>
34675 PR tree-optimization/79244
34676 * tree-vrp.c (remove_range_assertions): Forcefully propagate
34677 out SSA names even if abnormal.
34679 2017-01-27 Jakub Jelinek <jakub@redhat.com>
34681 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
34682 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
34683 instead of MPFR_RNDN.
34685 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
34688 * arm.c (arm_option_override): Don't call build_target_option_node
34689 until after doing all option overrides.
34690 (arm_valid_target_attribute_tree): Likewise.
34692 2017-01-27 Martin Liska <mliska@suse.cz>
34694 * doc/invoke.texi (-fprofile-arcs): Document profiling support
34695 for {cd}tors and C++ {cd}tors.
34697 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
34699 * config/s390/s390.md ("*setmem_long_and")
34700 ("*setmem_long_and_31z"): Use zero_extend instead of and.
34702 2017-01-26 Martin Sebor <msebor@redhat.com>
34704 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
34707 2017-01-26 Martin Sebor <msebor@redhat.com>
34709 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
34710 HAVE_DFmode before using XFmode or DFmode.
34711 (parse_directive): Avoid using the z length modifier to avoid
34712 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
34714 PR middle-end/78703
34715 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
34716 to accept adjustment as an array.
34717 (get_int_range): New function.
34718 (struct directive): Make width and prec arrays.
34719 (directive::set_width, directive::set_precision): Call get_int_range.
34720 (format_integer, format_floating): Handle width and precision ranges.
34721 (format_string, parse_directive): Same.
34723 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34726 * dwarf2out.c (generate_skeleton_bottom_up): For children with
34727 comdat_type_p set, just clone them, but keep the children in the
34731 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
34732 which have direct callers with -fvar-tracking-assignments enabled
34734 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
34735 inside of type units.
34737 2017-01-26 Martin Sebor <msebor@redhat.com>
34739 PR middle-end/78703
34740 * gimple-ssa-sprintf.c (struct result_range): Add likely and
34742 (struct format_result): Replace number_chars, number_chars_min,
34743 and number_chars_max with a single member of struct result_range.
34745 (format_result::operator+=): Adjust.
34746 (struct fmtresult): Remove bounded. Handle likely and unlikely
34748 (fmtresult::adjust_for_width_or_precision): New function.
34749 (fmtresult:type_max_digits): New function.
34750 (bytes_remaining): Handle likely and unlikely counters.
34751 (min_bytes_remaining): Remove.
34752 (format_percent): Simplify.
34753 (format_integer, format_floating): Set likely and unlikely counters.
34754 (get_string_length, format_character, format_string): Same.
34755 (format_plain, should_warn_p): New function.
34756 (maybe_warn): Call should_warn_p. Update diagnostic messages
34757 and handle those for all directives, including plain strings.
34758 (format_directive): Handle likely and unlikely counters.
34759 Remove unnecessary quoting from diagnostics. Add an informational
34761 (add_bytes): Remove.
34762 (pass_sprintf_length::compute_format_length): Simplify.
34763 (try_substitute_return_value): Handle likely and unlikely counters.
34765 2017-01-26 Carl Love <cel@us.ibm.com>
34767 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
34768 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
34770 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
34773 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
34774 endianess for subregs into account.
34775 * lra-constraints.c (lra_constraints): Do risky transformations
34776 always on the first iteration.
34777 * lra-lives.c (check_pseudos_live_through_calls): Add arg
34778 last_call_used_reg_set.
34779 (process_bb_lives): Define and use last_call_used_reg_set.
34780 * lra.c (lra): Always continue after lra_constraints on the first
34783 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
34785 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
34787 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
34789 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34791 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
34792 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
34793 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
34794 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
34795 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
34796 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
34797 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
34798 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
34799 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
34801 2017-01-26 Marek Polacek <polacek@redhat.com>
34804 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
34805 for the third operand.
34807 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34809 PR middle-end/79236
34810 * omp-low.c (struct omp_context): Add simt_stmt field.
34811 (scan_omp_for): Return omp_context *.
34812 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
34813 context to the _simt_ SIMD stmt.
34814 (lower_omp_for): For combined SIMD with sibling _simt_
34815 SIMD, make sure to use the same decls in _looptemp_
34816 clauses as in the sibling.
34818 2017-01-26 David Sherwood <david.sherwood@arm.com>
34820 PR middle-end/79212
34821 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
34824 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34827 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
34828 emit fld b; fld a; if possible.
34830 * brig-builtins.def: Update copyright years.
34831 * config/arm/arm_acle_builtins.def: Update copyright years.
34833 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
34836 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
34837 constraint instead of o for the stxsd instruction.
34839 2017-01-25 Carl Love <cel@us.ibm.com>
34841 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
34842 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
34844 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
34846 * doc/invoke.texi (C++ Dialect Options): Fix typo.
34848 2017-01-25 Richard Biener <rguenther@suse.de>
34850 PR tree-optimization/69264
34851 * target.def (vector_alignment_reachable): Improve documentation.
34852 * doc/tm.texi: Regenerate.
34853 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
34855 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
34856 earlier changes with respect to TYPE_USER_ALIGN.
34857 (vector_alignment_reachable_p): Likewise. Improve dumping.
34859 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34862 * config/arm/arm.md (xordi3): Force constant operand into a register
34865 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34867 * doc/invoke.texi (-fstore-merging): Correct default optimization
34868 levels at which it is enabled.
34869 (-O): Move -fstore-merging from list to...
34872 2017-01-25 Richard Biener <rguenther@suse.de>
34875 * omp-expand.c: Include debug.h.
34876 (expand_omp_taskreg): Make sure to generate early debug before
34877 outlining anything from a function.
34878 (expand_omp_target): Likewise.
34879 (grid_expand_target_grid_body): Likewise.
34881 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
34884 * asan.c (get_translation_unit_decl): New function.
34885 (asan_add_global): Extract modules file name from globals
34886 TRANSLATION_UNIT_DECL name.
34888 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
34891 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
34892 for long calls with APCS frame and VFP.
34894 2017-01-24 David Malcolm <dmalcolm@redhat.com>
34896 * cfg.c (original_copy_tables_initialized_p): New function.
34897 * cfg.h (original_copy_tables_initialized_p): New decl.
34898 * cfgrtl.c (relink_block_chain): Guard the call to
34899 free_original_copy_tables with a call to
34900 original_copy_tables_initialized_p.
34901 * cgraph.h (symtab_node::native_rtl_p): New decl.
34902 * cgraphunit.c (symtab_node::native_rtl_p): New function.
34903 (symtab_node::needed_p): Don't assert for early assembly output
34904 for __RTL functions.
34905 (cgraph_node::finalize_function): Set "force_output" for __RTL
34907 (cgraph_node::analyze): Bail out early for __RTL functions.
34908 (analyze_functions): Update assertion to support __RTL functions.
34909 (cgraph_node::expand): Bail out early for __RTL functions.
34910 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
34912 * function.h (struct function): Update comment for field
34914 * gimple-expr.c: Include "tree-pass.h".
34915 (gimple_has_body_p): Return false for __RTL functions.
34916 * Makefile.in (OBJS): Add run-rtl-passes.o.
34917 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
34919 (gcc::pass_manager::get_clean_slate): New accessor.
34920 * passes.c: Include "insn-addr.h".
34921 (should_skip_pass_p): Add logging. Update logic for running
34922 "expand" to be compatible with both __GIMPLE and __RTL. Guard
34923 property-provider override so it is only done for gimple passes.
34925 (skip_pass): New function.
34926 (execute_one_pass): Call skip_pass when skipping passes.
34927 * read-md.c (md_reader::read_char): Support filtering
34928 the input to a subset of line numbers.
34929 (md_reader::md_reader): Initialize fields
34930 m_first_line and m_last_line.
34931 (md_reader::read_file_fragment): New function.
34932 * read-md.h (md_reader::read_file_fragment): New decl.
34933 (md_reader::m_first_line): New field.
34934 (md_reader::m_last_line): New field.
34935 * read-rtl-function.c (function_reader::create_function): Only
34936 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
34937 curr_properties. Set DECL_INITIAL to a dummy block.
34938 (read_rtl_function_body_from_file_range): New function.
34939 * read-rtl-function.h (read_rtl_function_body_from_file_range):
34941 * run-rtl-passes.c: New file.
34942 * run-rtl-passes.h: New file.
34944 2017-01-24 Jeff Law <law@redhat.com>
34946 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
34949 2017-01-24 Bin Cheng <bin.cheng@arm.com>
34951 PR tree-optimization/79159
34952 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
34953 (record_nonwrapping_iv): Improve boundary using above function if no
34954 value range information.
34956 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
34957 Martin Jambor <mjambor@suse.cz>
34959 * brig-builtins.def: New file.
34960 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
34961 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
34962 (DEF_HSAIL_SAT_BUILTIN): Likewise.
34963 (DEF_HSAIL_INTR_BUILTIN): Likewise.
34964 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
34965 * builtin-types.def (BT_INT8): New.
34966 (BT_INT16): Likewise.
34967 (BT_UINT8): Likewise.
34968 (BT_UINT16): Likewise.
34969 (BT_FN_ULONG): Likewise.
34970 (BT_FN_UINT_INT): Likewise.
34971 (BT_FN_UINT_ULONG): Likewise.
34972 (BT_FN_UINT_LONG): Likewise.
34973 (BT_FN_UINT_PTR): Likewise.
34974 (BT_FN_ULONG_PTR): Likewise.
34975 (BT_FN_INT8_FLOAT): Likewise.
34976 (BT_FN_INT16_FLOAT): Likewise.
34977 (BT_FN_UINT32_FLOAT): Likewise.
34978 (BT_FN_UINT16_FLOAT): Likewise.
34979 (BT_FN_UINT8_FLOAT): Likewise.
34980 (BT_FN_UINT64_FLOAT): Likewise.
34981 (BT_FN_UINT16_UINT32): Likewise.
34982 (BT_FN_UINT32_UINT16): Likewise.
34983 (BT_FN_UINT16_UINT16_UINT16): Likewise.
34984 (BT_FN_INT_PTR_INT): Likewise.
34985 (BT_FN_UINT_PTR_UINT): Likewise.
34986 (BT_FN_LONG_PTR_LONG): Likewise.
34987 (BT_FN_ULONG_PTR_ULONG): Likewise.
34988 (BT_FN_VOID_UINT64_UINT64): Likewise.
34989 (BT_FN_UINT8_UINT8_UINT8): Likewise.
34990 (BT_FN_INT8_INT8_INT8): Likewise.
34991 (BT_FN_INT16_INT16_INT16): Likewise.
34992 (BT_FN_INT_INT_INT): Likewise.
34993 (BT_FN_UINT_FLOAT_UINT): Likewise.
34994 (BT_FN_FLOAT_UINT_UINT): Likewise.
34995 (BT_FN_ULONG_UINT_UINT): Likewise.
34996 (BT_FN_ULONG_UINT_PTR): Likewise.
34997 (BT_FN_ULONG_ULONG_ULONG): Likewise.
34998 (BT_FN_UINT_UINT_UINT): Likewise.
34999 (BT_FN_VOID_UINT_PTR): Likewise.
35000 (BT_FN_UINT_UINT_PTR: Likewise.
35001 (BT_FN_UINT32_UINT64_PTR): Likewise.
35002 (BT_FN_INT_INT_UINT_UINT): Likewise.
35003 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
35004 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
35005 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
35006 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
35007 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
35008 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
35009 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
35010 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
35011 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
35012 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
35013 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
35014 * doc/frontends.texi: List BRIG FE.
35015 * doc/install.texi (Testing): Add BRIG tesring requirements.
35016 * doc/invoke.texi (Overall Options): Mention BRIG.
35017 * doc/standards.texi (Standards): Doucment BRIG HSA version.
35019 2017-01-24 Richard Biener <rguenther@suse.de>
35021 PR translation/79208
35022 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
35024 2017-01-24 Martin Jambor <mjambor@suse.cz>
35027 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
35028 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
35029 and known_contexts.
35031 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
35033 PR middle-end/79123
35034 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
35035 casts from signed to unsigned really don't have a range.
35037 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
35039 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
35040 GMP_RNDx for compatiblity.
35042 2017-01-24 Martin Liska <mliska@suse.cz>
35045 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
35046 that would prevent us to call alloca with -1 as argument.
35048 2017-01-24 Jakub Jelinek <jakub@redhat.com>
35050 * dwarf2out.c (output_compilation_unit_header, output_file_names):
35051 Avoid -Wformat-security warning.
35053 2017-01-23 Andrew Pinski <apinski@cavium.com>
35055 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
35058 2017-01-23 Martin Sebor <msebor@redhat.com>
35060 PR middle-end/78703
35061 * gimple-ssa-sprintf.c (warn_level): New global.
35062 (format_integer): Use it here and throughout the rest of the file.
35063 Use the same switch to compute sign as base.
35064 (maybe_warn): New function.
35065 (format_directive): Factor out warnings into maybe_warn.
35066 Add debugging output. Use warn_level.
35067 (add_bytes): Use warn_level.
35068 (pass_sprintf_length::compute_format_length): Add debugging output.
35069 (try_substitute_return_value): Same.
35070 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
35072 PR middle-end/78703
35073 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
35074 (struct fmtresult, format_integer, format_floating): Adjust.
35075 (fmtresult::fmtresult): Set max correctly in two argument ctor.
35076 (get_string_length, format_string,format_directive): Same.
35077 (pass_sprintf_length::compute_format_length): Same.
35078 (try_substitute_return_value): Simplify slightly.
35080 PR middle-end/78703
35081 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
35082 (fmtresult::operator+=): Outlined.
35083 (struct fmtresult): Add ctors.
35084 (struct conversion_spec): Rename...
35085 (struct directive): ...to this. Add and remove data members.
35086 (directive::set_width, directive::set_precision): New functions.
35087 (format_percent): Use fmtresult ctor.
35088 (get_width_and_precision): Remove.
35089 (format_integer): Make naming changes. Avoid computing width and
35091 (format_floating): Same. Adjust indentation.
35092 (format_character, format_none): New functions.
35093 (format_string): Moved character handling to format_character.
35094 (format_directive): Remove arguments, change return type.
35095 (parse_directive): New function.
35096 (pass_sprintf_length::compute_format_length): Move directive
35097 parsing to parse_directive.
35099 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35101 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
35102 (assign_assembler_name_if_needed): ... this.
35103 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
35104 (assign_assembler_name_if_needed): ... this.
35105 (free_lang_data_in_cgraph): Adjust callers.
35106 * cgraphunit.c (cgraph_node::analyze): Likewise.
35107 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
35110 2017-01-23 Richard Biener <rguenther@suse.de>
35112 PR tree-optimization/79088
35113 PR tree-optimization/79188
35114 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
35115 resetting loop bounds after last path deletion. Reset loop
35116 bounds of the target loop, make code match the comments.
35117 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
35118 Make sure loops need no fixups.
35120 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35122 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
35123 exponent support with double type for first argument.
35124 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
35125 type returned by __builtin_vec_extract_sig,
35126 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
35127 functions from "vector int" to "vector unsigned int" or from
35128 "vector long long int" to "vector unsigned long long int".
35129 Changed type returned by __builtin_vec_extract_exp,
35130 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
35131 functions from "vector int" to "vector unsigned int" or from
35132 "vector long long int" to "vector unsigned long long int".
35133 Changed return type of __builtin_vec_test_data_class,
35134 __builtin_vec_test_data_class_sp, and
35135 __builtin_vec_test_data_class_dp from "vector int" to
35136 "vector bool int" or from "vector long long int" to "vector bool
35137 long long int" and changed second argument type from "unsigned
35138 int" to "int". Added new overloaded function forms "vector float
35139 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
35140 "vector float __builtin_vec_insert_exp_sp (vector float, vector
35141 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
35142 double, vector unsigned long long int)" and "vector double
35143 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
35144 long int)". Changed return type of
35145 __builtin_scalar_test_data_class and
35146 __builtin_scalar_test_data_class_sp and
35147 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
35148 int" and changed second argument from "unsigned int" to "int".
35149 Changed type returned by __builtin_scalar_test_neg,
35150 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
35151 from "int" to "bool int". Added new overloaded function form
35152 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
35153 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
35154 exponent double-precision with floating point first argument.
35155 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
35156 documentation of scalar_test_data_class, scalar_test_neg,
35157 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
35158 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
35159 vec_test_data_class built-in functions to reflect refinements in
35160 their type signatures.
35162 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
35164 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
35166 (aarch64_elf_asm_destructor): Likewise.
35168 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
35170 PR rtl-optimization/78634
35171 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
35172 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
35173 * ifcvt.c (noce_try_cmove): Add missing cost check.
35175 PR rtl-optimization/71724
35176 * combine.c (if_then_else_cond): Look for situations where it is
35177 beneficial to undo the work of one of the recursive calls.
35179 2017-01-23 Bin Cheng <bin.cheng@arm.com>
35181 PR tree-optimization/70754
35182 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
35183 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
35184 combined stmt before it if not NULL.
35185 (combine_chains): Process refs reversely and compute dominance point
35188 2017-01-23 Martin Liska <mliska@suse.cz>
35190 PR tree-optimization/79196
35191 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
35192 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
35194 (strlen_optimize_stmt): Call the renamed function.
35196 2017-01-23 Michael Matz <matz@suse.de>
35198 PR tree-optimization/78384
35199 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
35201 2017-01-23 Richard Biener <rguenther@suse.de>
35203 PR tree-optimization/79186
35204 * tree-vrp.c (register_new_assert_for): Make sure we've seen
35205 both incoming edges before moving an assert.
35207 2017-01-23 Martin Jambor <mjambor@suse.cz>
35209 * ipa-prop.c (load_from_param_1): Removed.
35210 (load_from_unmodified_param): Bits from load_from_param_1 put back
35212 (load_from_param): Removed.
35213 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
35214 with stmt. Reverted back to use of load_from_unmodified_param.
35216 2017-01-23 Martin Jambor <mjambor@suse.cz>
35219 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
35220 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
35221 field a pointer to garbage collected vector, mark lattices and
35222 ipcp_orig_node with GTY((skip)).
35223 (ipa_get_param_count): Adjust to descriptors being a pointer.
35224 (ipa_get_param): Likewise.
35225 (ipa_get_type): Likewise.
35226 (ipa_get_param_move_cost): Likewise.
35227 (ipa_set_param_used): Likewise.
35228 (ipa_get_controlled_uses): Likewise.
35229 (ipa_set_controlled_uses): Likewise.
35230 (ipa_is_param_used): Likewise.
35231 (ipa_node_params_t): Move into garbage collector. New methods insert
35233 (ipa_node_params_sum): Annotate wth GTY(()).
35234 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
35236 (ipa_load_from_parm_agg): Adjust declaration.
35237 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
35238 * ipa-profile.c (ipa_profile): Likewise.
35239 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
35240 (ipa_populate_param_decls): Make descriptors parameter garbage
35242 (ipa_dump_param): Adjust to descriptors being a pointer.
35243 (ipa_alloc_node_params): Likewise.
35244 (ipa_initialize_node_params): Likewise.
35245 (load_from_param_1): Make descriptors parameter garbage collected.
35246 (load_from_unmodified_param): Likewise.
35247 (load_from_param): Likewise.
35248 (ipa_load_from_parm_agg): Likewise.
35249 (ipa_node_params::~ipa_node_params): Removed.
35250 (ipa_free_all_node_params): Remove call to delete operator.
35251 (ipa_node_params_t::insert): New.
35252 (ipa_node_params_t::remove): Likewise.
35253 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
35254 copy known_csts and known_contexts vectors.
35255 (ipa_read_node_info): Adjust to descriptors being a pointer.
35256 (ipcp_modif_dom_walker): Make m_descriptors field garbage
35258 (ipcp_transform_function): Make descriptors variable garbage
35261 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
35263 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
35264 * config/i386/avx512dqintrin.h: Ditto.
35265 * config/i386/avx512fintrin.h: Ditto.
35266 * config/i386/i386.c: Handle new builtins.
35267 * config/i386/i386-builtin.def: Add new builtins.
35268 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
35269 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
35271 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35272 Martin Liska <mliska@suse.cz>
35274 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
35275 * asan.c (asan_expand_poison_ifn): Support stores and use
35276 appropriate ASAN report function.
35277 * internal-fn.c (expand_ASAN_POISON_USE): New function.
35278 * internal-fn.def (ASAN_POISON_USE): Declare.
35279 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
35280 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
35281 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
35282 ASAN_POISON calls w/o LHS.
35283 * tree-ssa.c (execute_update_addresses_taken): Create clobber
35284 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
35285 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
35286 * gimplify.c (asan_poison_variables): Add attribute
35287 use_after_scope_memory to variables that really needs to live
35289 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
35290 having the attribute.
35292 2017-01-23 Martin Liska <mliska@suse.cz>
35294 * asan.c (create_asan_shadow_var): New function.
35295 (asan_expand_poison_ifn): Likewise.
35296 * asan.h (asan_expand_poison_ifn): New declaration.
35297 * internal-fn.c (expand_ASAN_POISON): Likewise.
35298 * internal-fn.def (ASAN_POISON): New builtin.
35299 * sanopt.c (pass_sanopt::execute): Expand
35300 asan_expand_poison_ifn.
35301 * tree-inline.c (copy_decl_for_dup_finish): Make function
35303 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
35304 * tree-ssa.c (is_asan_mark_p): New function.
35305 (execute_update_addresses_taken): Rewrite local variables
35306 (identified just by use-after-scope as addressable) into SSA.
35308 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35310 * doc/install.texi (Specific): opensource.apple.com uses https
35311 now. Remove trailing slash.
35313 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35315 * README.Portability: Remove note on an Irix compatibility issue.
35317 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
35319 * gcov.c (INCLUDE_ALGORITHM): Define.
35320 (INCLUDE_VECTOR): Define.
35321 No longer include <vector> and <algorithm> directly.
35323 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
35325 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
35327 * doc/invoke.texi (Code Gen Options): Ditto.
35329 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
35332 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
35334 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
35336 rtl-optimization/79125
35337 * cprop.c (local_cprop_pass): Handle cases where we make an
35338 unconditional trap.
35340 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
35344 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
35345 read from, for big endian.
35347 2017-01-20 Jiong Wang <jiong.wang@arm.com>
35349 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
35350 register pauth builtins for LP64 only.
35352 2017-01-20 Marek Polacek <polacek@redhat.com>
35355 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
35358 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35360 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
35362 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
35363 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
35364 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
35366 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35369 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
35370 in big-endian mode when they are not a single duplicated value.
35372 2017-01-20 Richard Biener <rguenther@suse.de>
35374 * BASE-VER: Bump to 7.0.1.
35376 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35378 * omp-low.c (omplow_simd_context): New struct. Use it...
35379 (lower_rec_simd_input_clauses): ...here and...
35380 (lower_rec_input_clauses): ...here to hold common data. Adjust all
35381 references to idx, lane, max_vf, is_simt.
35383 2017-01-20 Graham Markall <graham.markall@embecosm.com>
35385 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
35388 2017-01-20 Martin Jambor <mjambor@suse.cz>
35390 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
35391 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
35393 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
35394 (GTFILES): Rename hsa.c to hsa-common.c.
35395 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
35396 * hsa-dump.c: Likewise.
35397 * hsa-gen.c: Likewise.
35398 * hsa-regalloc.c: Likewise.
35399 * ipa-hsa.c: Likewise.
35400 * omp-expand.c: Likewise.
35401 * omp-low.c: Likewise.
35402 * toplev.c: Likewise.
35404 2017-01-20 Marek Polacek <polacek@redhat.com>
35407 * doc/invoke.texi: Document -Wduplicated-branches.
35408 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
35409 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
35410 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
35411 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
35412 return 0 only when not OEP_LEXICOGRAPHIC.
35413 (fold_build_cleanup_point_expr): Use the expression
35414 location when building CLEANUP_POINT_EXPR.
35415 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
35416 * tree.c (add_expr): Handle error_mark_node.
35418 2017-01-20 Martin Liska <mliska@suse.cz>
35421 * tree-profile.c (init_ic_make_global_vars): Do not call
35423 (gimple_init_gcov_profiler): Likewise.
35425 2017-01-20 Martin Liska <mliska@suse.cz>
35428 * cgraph.h (maybe_create_reference): Remove argument and
35430 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
35432 * ipa-cp.c (create_specialized_node): Likewise.
35433 * symtab.c (symtab_node::maybe_create_reference): Handle
35434 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
35436 2017-01-20 Martin Liska <mliska@suse.cz>
35438 * read-rtl-function.c (function_reader::create_function): Use
35439 build_decl instread of build_decl_stat.
35441 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
35443 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
35444 * config/i386/avx512dqintrin.h: Ditto.
35445 * config/i386/avx512fintrin.h: Ditto.
35446 * config/i386/i386-builtin-types.def: Add new types.
35447 * config/i386/i386.c: Handle new types.
35448 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
35449 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
35450 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
35451 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
35452 (__builtin_ia32_kshiftridi): New.
35453 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
35455 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
35459 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
35460 define to rs6000_init_stack_protect_guard.
35461 (rs6000_init_stack_protect_guard): New function.
35463 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
35464 Yunqiang Su <yunqiang.su@imgtec.com>
35466 * config.gcc (supported_defaults): Add madd4.
35467 (with_madd4): Add validation.
35468 (all_defaults): Add madd4.
35469 * config/mips/mips.opt (mmadd4): New option.
35470 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35472 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35474 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
35475 (ISA_HAS_FUSED_MADD4): Likewise.
35476 * doc/invoke.texi (-mmadd4): Document the new option.
35477 * doc/install.texi (--with-madd4): Document the new option.
35479 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35481 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
35482 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
35483 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
35484 (aarch64_init_pauth_hint_builtins): New.
35485 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
35486 (aarch64_expand_builtin): Expand new builtins.
35488 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35490 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
35491 * combine-stack-adj.c (no_unhandled_cfa): Handle
35492 REG_CFA_TOGGLE_RA_MANGLE.
35493 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
35494 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
35495 info for return address signing.
35496 (aarch64_expand_epilogue): Likewise.
35498 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35500 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
35501 * config/aarch64/aarch64-protos.h
35502 (aarch64_return_address_signing_enabled): New declaration.
35503 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
35505 (aarch64_expand_prologue): Sign return address before it's pushed onto
35507 (aarch64_expand_epilogue): Authenticate return address fetched from
35509 (aarch64_override_options): Sanity check for ILP32 and ISA level.
35510 (aarch64_attributes): New function attributes for "sign-return-address".
35511 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
35512 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
35513 ("*do_return"): Generate combined instructions according to key index.
35514 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
35515 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
35517 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
35518 * config/aarch64/aarch64.opt (msign-return-address=): New.
35519 * doc/extend.texi (AArch64 Function Attributes): Documents
35520 "sign-return-address=".
35521 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
35523 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
35525 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
35526 overall option summary.
35528 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35530 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
35531 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
35532 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
35533 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
35535 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
35537 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
35538 -mpower9-minmax by default for -mcpu=power9.
35539 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
35540 128-bit floating point.
35542 2017-01-20 Alan Modra <amodra@gmail.com>
35544 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
35545 optimizing for size.
35547 2017-01-20 Alan Modra <amodra@gmail.com>
35550 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
35551 for strcmp and strncmp from corresponding builtin decl.
35553 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
35555 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
35556 instead of i386/rtems-64.h.
35557 * config/i386/rtems-64.h: Remove.
35559 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
35563 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
35565 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
35567 2017-01-19 Tamar Christina <tamar.christina@arm.com>
35569 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
35570 Change int to HOST_WIDE_INT.
35571 * config/aarch64/aarch64-protos.h
35572 (aarch64_simd_gen_const_vector_dup): Likewise.
35573 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
35575 2017-01-19 David Malcolm <dmalcolm@redhat.com>
35577 * langhooks-def.h (lhd_type_for_size): New decl.
35578 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
35579 * langhooks.c (lhd_type_for_size): New function, taken from
35582 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
35584 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
35585 define_bypass for CR latency.
35586 (power9-cracked-alu): Update bypass latency and remove power9-branch.
35587 (power9-alu2): Add define_bypass for CR latency.
35589 (power9-mul): Update insn latency.
35590 (power9-mul-compare): Update insn latency, bypass latency and remove
35593 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35595 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
35597 * config/aarch64/aarch64.md
35598 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
35599 aarch64_nopcrelative_literal_loads.
35600 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
35602 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
35604 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
35605 TARGET_LOONGSON_3A.
35606 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
35608 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
35611 * config.gcc (supported_defaults): Add lxc1-sxc1.
35612 (with_lxc1_sxc1): Add validation.
35613 (all_defaults): Add lxc1-sxc1.
35614 * config/mips/mips.opt (mlxc1-sxc1): New option.
35615 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35617 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35618 __mips_no_lxc1_sxc1.
35619 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
35620 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
35621 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
35623 2017-01-19 Richard Biener <rguenther@suse.de>
35625 PR tree-optimization/72488
35626 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
35627 sure to restore SSA info.
35628 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
35630 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
35632 PR rtl-optimization/79121
35633 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
35634 of the inner type when shifting an extended value.
35636 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
35639 * symtab.c (symtab_node::equal_address_to): Fix comparing of
35640 interposable aliases.
35642 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
35645 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
35646 Use the evmergelohi instruction.
35647 (mov_si<mode>_e500_subreg4_2_le): Likewise.
35648 (mov_sitf_e500_subreg8_2_be): Likewise.
35649 (mov_sitf_e500_subreg12_2_le): Likewise.
35650 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
35651 (mov_si<mode>_e500_subreg4_2_be): Likewise.
35652 (mov_sitf_e500_subreg8_2_le): Likewise.
35653 (mov_sitf_e500_subreg12_2_be): Likewise.
35655 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35657 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
35658 attribute from vecsimple to vecperm.
35659 (altivec_vbpermq2): Likewise.
35661 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35664 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
35666 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35667 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
35668 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
35669 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
35670 case where N arg is SIZE_MAX.
35671 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
35672 (cmpstrsi): Add pattern.
35674 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
35676 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
35677 __builtin_vec_revb builtins.
35678 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
35679 built-in functions to support generation of the ISA 3.0 XXBR<x>
35680 vector byte reverse instructions.
35681 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
35682 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
35683 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
35684 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
35685 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
35686 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
35687 (P9V_BUILTIN_VEC_REVB): Likewise.
35688 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
35689 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
35690 (p9_xxbrq_v16qi): Likewise.
35691 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
35692 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
35693 (p9_xxbrh_v8hi): Likewise.
35694 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
35695 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
35696 vec_revb built-in functions.
35698 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
35700 PR rtl-optimization/78952
35701 * config/i386/i386.md (any_extract): New code iterator.
35702 (*insvqi_2): Use any_extract for source operand.
35703 (*insvqi_3): Use any_shiftrt for source operand.
35705 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
35707 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
35709 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
35711 2017-01-18 Matthias Klose <doko@ubuntu.com>
35713 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
35715 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35717 * config/rs6000/altivec.h (vec_bperm): Change #define.
35718 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
35719 (altivec_vbpermq2): New define_insn.
35720 (altivec_vbpermd): Likewise.
35721 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
35722 function interface.
35723 (VBPERMD): Likewise.
35724 (VBPERM): New polymorphic function interface.
35725 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
35726 Add entries for P9V_BUILTIN_VEC_VBPERM.
35727 * doc/extend.texi: Add interfaces for vec_bperm.
35729 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35731 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
35732 first letter of error messages.
35733 (s390_resolve_overloaded_builtin): Likewise.
35734 * config/s390/s390.c (s390_expand_builtin): Likewise.
35735 (s390_invalid_arg_for_unprototyped_fn): Likewise.
35736 (s390_valid_target_attribute_inner_p): Likewise.
35737 * config/s390/s390.md ("tabort"): Likewise.
35739 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
35741 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
35742 (ISA_AVOID_DIV_HILO): New macro.
35743 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
35744 (ISA_HAS_DDIV): Likewise.
35746 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35748 * doc/invoke.texi (fabi-version): Correct number of occurrences.
35750 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35752 * doc/invoke.texi (fabi-version): Spelling fix.
35754 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35757 * doc/invoke.texi (fabi-version): Mention mangling fix for
35760 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35763 * doc/invoke.texi (fabi-version): Document discriminator mangling.
35765 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
35768 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
35769 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
35771 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
35773 (stack_protect_test): Ditto.
35774 * config/rs6000/rs6000.opt (mstack-protector-guard=,
35775 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
35777 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
35778 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
35779 -mstack-protector-guard-offset=.
35780 (RS/6000 and PowerPC Options): Ditto.
35782 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
35784 * config/i386/i386.h (MASK_CLASS_P): New define.
35785 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
35786 there are no registers from different register sets also when
35787 mask registers are used. Update function comment.
35788 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
35789 to (*k/*r) and (*k/*km) alternatives.
35791 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
35793 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
35794 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
35795 (EH_RETURN_HANDLER_RTX): New define.
35796 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
35797 Force frame pointer in EH return functions.
35798 (aarch64_expand_epilogue): Add barrier for eh_return.
35799 (aarch64_final_eh_return_addr): Remove.
35800 (aarch64_eh_return_handler_rtx): New function.
35801 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
35803 (aarch64_eh_return_handler_rtx): New prototype.
35805 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35807 * config/rs6000/altivec.h (vec_rlmi): New #define.
35808 (vec_vrlnm): Likewise.
35809 (vec_rlnm): Likewise.
35810 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
35811 (UNSPEC_VRLNM): Likewise.
35812 (VIlong): New mode iterator.
35813 (altivec_vrl<VI_char>mi): New define_insn.
35814 (altivec_vrl<VI_char>nm): Likewise.
35815 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
35817 (VRLDNM): Likewise.
35818 (RLNM): New polymorphic function entry.
35819 (VRLWMI): New monomorphic function entry.
35820 (VRLDMI): Likewise.
35821 (RLMI): New polymorphic function entry.
35822 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
35823 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
35824 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
35827 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35830 * dwarf2out.c (field_byte_offset): Restore the
35831 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
35832 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
35833 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
35836 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
35839 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
35841 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35844 * dwarf2out.c (add_data_member_location_attribute): For constant
35845 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
35846 instead of DW_AT_data_member_location, DW_AT_bit_offset and
35847 DW_AT_byte_size attributes.
35849 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
35851 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
35852 after forcing to constant memory when the code model is medium.
35854 2017-01-17 Julia Koval <julia.koval@intel.com>
35857 * config/i386/avx512fintrin.h
35858 (_mm512_i32gather_ps): Change __addr type to void const*.
35859 (_mm512_mask_i32gather_ps): Ditto.
35860 (_mm512_i32gather_pd): Ditto.
35861 (_mm512_mask_i32gather_pd): Ditto.
35862 (_mm512_i64gather_ps): Ditto.
35863 (_mm512_mask_i64gather_ps): Ditto.
35864 (_mm512_i64gather_pd): Ditto.
35865 (_mm512_mask_i64gather_pd): Ditto.
35866 (_mm512_i32gather_epi32): Ditto.
35867 (_mm512_mask_i32gather_epi32): Ditto.
35868 (_mm512_i32gather_epi64): Ditto.
35869 (_mm512_mask_i32gather_epi64): Ditto.
35870 (_mm512_i64gather_epi32): Ditto.
35871 (_mm512_mask_i64gather_epi32): Ditto.
35872 (_mm512_i64gather_epi64): Ditto.
35873 (_mm512_mask_i64gather_epi64): Ditto.
35874 (_mm512_i32scatter_ps): Change __addr type to void*.
35875 (_mm512_mask_i32scatter_ps): Ditto.
35876 (_mm512_i32scatter_pd): Ditto.
35877 (_mm512_mask_i32scatter_pd): Ditto.
35878 (_mm512_i64scatter_ps): Ditto.
35879 (_mm512_mask_i64scatter_ps): Ditto.
35880 (_mm512_i64scatter_pd): Ditto.
35881 (_mm512_mask_i64scatter_pd): Ditto.
35882 (_mm512_i32scatter_epi32): Ditto.
35883 (_mm512_mask_i32scatter_epi32): Ditto.
35884 (_mm512_i32scatter_epi64): Ditto.
35885 (_mm512_mask_i32scatter_epi64): Ditto.
35886 (_mm512_i64scatter_epi32): Ditto.
35887 (_mm512_mask_i64scatter_epi32): Ditto.
35888 (_mm512_i64scatter_epi64): Ditto.
35889 (_mm512_mask_i64scatter_epi64): Ditto.
35890 * config/i386/avx512pfintrin.h
35891 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
35892 (_mm512_mask_prefetch_i32gather_ps): Ditto.
35893 (_mm512_mask_prefetch_i64gather_pd): Ditto.
35894 (_mm512_mask_prefetch_i64gather_ps): Ditto.
35895 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
35896 (_mm512_prefetch_i32scatter_ps): Ditto.
35897 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
35898 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
35899 (_mm512_prefetch_i64scatter_pd): Ditto.
35900 (_mm512_prefetch_i64scatter_ps): Ditto.
35901 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
35902 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
35903 * config/i386/avx512vlintrin.h
35904 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
35905 (_mm_mmask_i32gather_ps): Ditto.
35906 (_mm256_mmask_i32gather_pd): Ditto.
35907 (_mm_mmask_i32gather_pd): Ditto.
35908 (_mm256_mmask_i64gather_ps): Ditto.
35909 (_mm_mmask_i64gather_ps): Ditto.
35910 (_mm256_mmask_i64gather_pd): Ditto.
35911 (_mm_mmask_i64gather_pd): Ditto.
35912 (_mm256_mmask_i32gather_epi32): Ditto.
35913 (_mm_mmask_i32gather_epi32): Ditto.
35914 (_mm256_mmask_i32gather_epi64): Ditto.
35915 (_mm_mmask_i32gather_epi64): Ditto.
35916 (_mm256_mmask_i64gather_epi32): Ditto.
35917 (_mm_mmask_i64gather_epi32): Ditto.
35918 (_mm256_mmask_i64gather_epi64): Ditto.
35919 (_mm_mmask_i64gather_epi64): Ditto.
35920 (_mm256_i32scatter_ps): Change __addr type to void*.
35921 (_mm256_mask_i32scatter_ps): Ditto.
35922 (_mm_i32scatter_ps): Ditto.
35923 (_mm_mask_i32scatter_ps): Ditto.
35924 (_mm256_i32scatter_pd): Ditto.
35925 (_mm256_mask_i32scatter_pd): Ditto.
35926 (_mm_i32scatter_pd): Ditto.
35927 (_mm_mask_i32scatter_pd): Ditto.
35928 (_mm256_i64scatter_ps): Ditto.
35929 (_mm256_mask_i64scatter_ps): Ditto.
35930 (_mm_i64scatter_ps): Ditto.
35931 (_mm_mask_i64scatter_ps): Ditto.
35932 (_mm256_i64scatter_pd): Ditto.
35933 (_mm256_mask_i64scatter_pd): Ditto.
35934 (_mm_i64scatter_pd): Ditto.
35935 (_mm_mask_i64scatter_pd): Ditto.
35936 (_mm256_i32scatter_epi32): Ditto.
35937 (_mm256_mask_i32scatter_epi32): Ditto.
35938 (_mm_i32scatter_epi32): Ditto.
35939 (_mm_mask_i32scatter_epi32): Ditto.
35940 (_mm256_i32scatter_epi64): Ditto.
35941 (_mm256_mask_i32scatter_epi64): Ditto.
35942 (_mm_i32scatter_epi64): Ditto.
35943 (_mm_mask_i32scatter_epi64): Ditto.
35944 (_mm256_i64scatter_epi32): Ditto.
35945 (_mm256_mask_i64scatter_epi32): Ditto.
35946 (_mm_i64scatter_epi32): Ditto.
35947 (_mm_mask_i64scatter_epi32): Ditto.
35948 (_mm256_i64scatter_epi64): Ditto.
35949 (_mm256_mask_i64scatter_epi64): Ditto.
35950 (_mm_i64scatter_epi64): Ditto.
35951 (_mm_mask_i64scatter_epi64): Ditto.
35952 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
35953 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
35954 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
35955 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
35956 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
35957 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
35958 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
35959 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
35960 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
35961 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
35962 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
35963 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
35964 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
35965 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
35966 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
35967 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
35968 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
35969 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
35970 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
35971 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
35972 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
35973 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
35974 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
35975 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
35976 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
35977 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
35978 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
35979 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
35980 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
35981 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
35982 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
35983 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
35984 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
35985 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
35986 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
35987 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
35988 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
35989 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
35990 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
35991 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
35992 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
35993 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
35994 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
35995 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
35996 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
35997 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
35998 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
35999 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
36000 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
36001 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
36002 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
36003 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
36004 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
36005 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
36006 definitions accordingly.
36008 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
36009 Kuan-Lin Chen <kuanlinchentw@gmail.com>
36012 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
36015 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
36018 * ira-conflicts.c (ira_build_conflicts): Update total conflict
36019 hard regs for inner regno.
36021 2017-01-17 Martin Liska <mliska@suse.cz>
36024 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
36025 assumption and add comment.
36027 2017-01-17 Nathan Sidwell <nathan@acm.org>
36029 * ipa-visibility.c (localize_node): New function, broken out of ...
36030 (function_and_variable_visibility): ... here. Call it.
36032 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36034 PR middle-end/77445
36035 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
36036 correctly set frequency of oudgoing edge.
36037 (duplicate_thread_path): Fix profile updating.
36039 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36042 * configure.ac: Add GCC_BASE_VER.
36043 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
36044 version from BASE-VER file.
36045 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
36046 (gcc.o): Depend on $(BASEVER).
36047 * common.opt (dumpfullversion): New option.
36048 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
36049 * doc/invoke.texi: Document -dumpfullversion.
36050 * doc/install.texi: Document --with-gcc-major-version-only.
36051 * configure: Regenerated.
36053 2017-01-17 Richard Biener <rguenther@suse.de>
36055 PR tree-optimization/71433
36056 * tree-vrp.c (register_new_assert_for): Merge same asserts
36057 on all incoming edges.
36058 (process_assert_insertions_for): Handle insertions at the
36061 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
36063 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
36064 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
36066 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
36069 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
36072 2017-01-17 Alan Modra <amodra@gmail.com>
36075 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
36076 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
36077 symbolic stack limit when pic.
36079 2017-01-16 Martin Sebor <msebor@redhat.com>
36081 PR tree-optimization/78608
36082 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
36084 2017-01-16 Jeff Law <law@redhat.com>
36087 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
36088 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
36089 for several include directories that may be relative to sysroot.
36090 * config/i386/x-mingw32 (gplus_includedir): Define.
36091 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
36092 (native_system_includedir): Likewise.
36093 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
36094 override if TARGET_SYSTEM_ROOT is defined.
36095 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
36097 PR tree-optimization/79090
36098 PR tree-optimization/33562
36099 PR tree-optimization/61912
36100 PR tree-optimization/77485
36101 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
36102 and computed trims into the dump file.
36104 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36106 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
36108 2017-01-16 Jakub Jelinek <jakub@redhat.com>
36111 * gimplify.c (gimplify_init_constructor): If want_value and
36112 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
36116 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
36117 sequence. Formatting fixes.
36118 (doloop_optimize): Formatting fixes.
36121 * gcc.c (debug_level_greater_than_spec_func): New function.
36122 (static_spec_functions): Add debug-level-gt spec function.
36123 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
36125 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
36126 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
36127 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
36128 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
36129 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
36130 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
36132 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
36134 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
36135 QImode fixups to general and mask registers only.
36137 2017-01-16 Carl Love <cel@us.ibm.com>
36139 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
36140 for built-in functions
36141 vector signed char vec_nabs (vector signed char)
36142 vector signed short vec_nabs (vector signed short)
36143 vector signed int vec_nabs (vector signed int)
36144 vector signed long long vec_nabs (vector signed long long)
36145 vector float vec_nabs (vector float)
36146 vector double vec_nabs (vector double)
36147 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
36149 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
36150 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
36151 * doc/extend.texi: Update the documentation file for the new built-in
36154 2017-01-16 Martin Sebor <msebor@redhat.com>
36156 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
36159 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36161 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
36162 UNSPEC_VSX__XXSPLTD to require special splat handling.
36164 2017-01-16 David Malcolm <dmalcolm@redhat.com>
36167 * system.h: Poison strndup.
36169 2017-01-16 Alan Modra <amodra@gmail.com>
36172 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
36175 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
36177 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
36179 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
36181 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
36182 call recog here. Assert that INSN_CODE (insn) is non-negative.
36184 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
36187 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
36189 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
36190 in the currently scheduled RTL fragment.
36192 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
36194 PR rtl-optimization/78751
36195 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
36198 2017-01-14 Jeff Law <law@redhat.com>
36200 PR tree-optimization/79090
36201 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
36202 variable length stores.
36203 (compute_trims): Delete dead assignment to *trim_tail.
36204 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
36207 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
36209 PR rtl-optimization/78626
36210 PR rtl-optimization/78727
36211 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
36212 of a block, and split such blocks after everything else is finished.
36214 2017-01-14 Alan Modra <amodra@gmail.com>
36217 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
36218 target legitimate_combined_insn.
36219 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
36220 (rs6000_legitimate_combined_insn): New function.
36221 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
36223 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
36224 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
36225 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
36227 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
36229 * doc/frontends.texi (G++ and GCC): Remove references to Java.
36231 2017-01-13 Jeff Law <law@redhat.com>
36233 PR tree-optimization/33562
36234 PR tree-optimization/61912
36235 PR tree-optimization/77485
36236 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
36238 (delete_dead_assignment): Likewise.
36239 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
36240 statement to delete_dead_call and delete_dead_assignment.
36242 2017-01-13 David Malcolm <dmalcolm@redhat.com>
36245 * substring-locations.c (format_warning_va): Strengthen case 1 so
36246 that both endpoints of the substring must be within the format
36247 range for just the substring to be printed.
36249 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
36251 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
36252 * config/i386/i386.c (ix86_target_string): Add missing options
36253 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
36254 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
36255 flags_other and ix86_target_other to flags2_other. Display unknown
36257 (ix86_valid_target_attribute_inner_p): Add missing options and
36258 reorder options by implied ISAs, as in ix86_target_string.
36260 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36262 * hash-table.h (hash_table::too_empty_p): New function.
36263 (hash_table::expand): Use it.
36264 (hash_table::traverse): Likewise.
36265 (hash_table::empty_slot): Use sizeof (value_type) instead of
36266 sizeof (PTR) to convert bytes to elements. Shrink the table
36267 if the current size is excessive for the current number of
36270 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36272 * ira-costs.c (record_reg_classes): Break from the inner loop
36273 early once alt_fail is known to be true. Update outer loop
36274 handling accordingly.
36276 2017-01-13 Jeff Law <law@redhat.com>
36278 * tree-ssa-dse.c (decrement_count): New function.
36279 (increment_start_addr, maybe_trim_memstar_call): Likewise.
36280 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
36281 when we know the partially dead statement is a mem* function.
36283 PR tree-optimization/61912
36284 PR tree-optimization/77485
36285 * tree-ssa-dse.c: Include expr.h.
36286 (maybe_trim_constructor_store): New function.
36287 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
36289 PR tree-optimization/33562
36290 PR tree-optimization/61912
36291 PR tree-optimization/77485
36292 * doc/invoke.texi: Document new dse-max-object-size param.
36293 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
36294 * tree-ssa-dse.c: Include params.h.
36295 (dse_store_status): New enum.
36296 (initialize_ao_ref_for_dse): New, partially extracted from
36298 (valid_ao_ref_for_dse, normalize_ref): New.
36299 (setup_live_bytes_from_ref, compute_trims): Likewise.
36300 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
36301 (maybe_trim_partially_dead_store): Likewise.
36302 (maybe_trim_complex_store): Likewise.
36303 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
36304 Track what bytes live from the original store. Return tri-state
36305 for dead, partially dead or live.
36306 (dse_dom_walker): Add constructor, destructor and new private members.
36307 (delete_dead_call, delete_dead_assignment): New extracted from
36309 (dse_optimize_stmt): Make a member of dse_dom_walker.
36310 Use initialize_ao_ref_for_dse.
36312 PR tree-optimization/33562
36313 PR tree-optimization/61912
36314 PR tree-optimization/77485
36315 * sbitmap.h (bitmap_count_bits): Prototype.
36316 (bitmap_clear_range, bitmap_set_range): Likewise.
36317 * sbitmap.c (bitmap_clear_range): New function.
36318 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
36320 2017-01-13 Martin Liska <mliska@suse.cz>
36323 * function.c (set_cfun): Add new argument force.
36324 * function.h (set_cfun): Likewise.
36325 * ipa-inline-transform.c (inline_call): Use the function when
36326 strict alising from is dropped for function we inline to.
36328 2017-01-13 Richard Biener <rguenther@suse.de>
36330 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
36331 for dumping GIMPLE INTEGER_CSTs.
36333 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
36335 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
36336 to 201112L since C++17.
36338 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
36341 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
36342 if -fsanitize=kernel-address is present.
36344 2017-01-13 Richard Biener <rguenther@suse.de>
36346 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
36347 as _Literal ( type ) number in case usual suffixes do not
36348 preserve all information.
36350 2017-01-13 Richard Biener <rguenther@suse.de>
36352 PR tree-optimization/77283
36353 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
36354 and ssa-iterators.h.
36355 (is_feasible_trace): Implement a cost model based on joiner
36358 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
36361 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
36362 char or short to __float128/_Float128 directly.
36364 2017-01-12 Martin Sebor <msebor@redhat.com>
36366 to -Wformat-overflow.
36367 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
36368 (min_bytes_remaining): Same.
36369 (get_string_length): Same.
36370 (format_string): Same.
36371 (format_directive): Same.
36373 (pass_sprintf_length::handle_gimple_call): Same.
36375 2017-01-12 Jakub Jelinek <jakub@redhat.com>
36377 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
36378 info.nowrite calls with no lhs that can't throw. Return bool
36379 whether gsi_remove has been called or not.
36380 (pass_sprintf_length::handle_gimple_call): Return bool whether
36381 try_substitute_return_value called gsi_remove. Formatting fix.
36382 (pass_sprintf_length::execute): Don't use gsi_remove if
36383 handle_gimple_call returned true.
36386 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
36387 be removed due to side-effects, don't remove following barrier nor
36388 turn the successor edge into fallthru edge.
36390 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36393 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
36394 element-reversing loads and stores as not swappable.
36396 2017-01-12 Nathan Sidwell <nathan@acm.org>
36397 Nicolai Stange <nicstange@gmail.com>
36399 * combine.c (try_combine): Don't ignore result of overlap checking
36400 loop. Combine overlap & asm check into single loop.
36402 2017-01-12 Richard Biener <rguenther@suse.de>
36404 * tree-pretty-print.c (dump_generic_node): Provide -gimple
36405 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
36407 2017-01-12 Richard Biener <rguenther@suse.de>
36409 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
36410 and TS_TARGET_OPTION directly derive from TS_BASE.
36411 * tree-core.h (tree_optimization_option): Derive from tree_base.
36412 (tree_target_option): Likewise.
36414 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
36416 * config/i386/i386.c (memory_address_length): Increase len
36417 only when rip_relative_addr_p returns false.
36419 2017-01-11 Julia Koval <julia.koval@intel.com>
36421 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
36422 (OPTION_MASK_ISA_SGX_SET): New.
36423 (ix86_handle_option): Handle OPT_msgx.
36424 * config.gcc: Added sgxintrin.h.
36425 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
36426 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
36427 * config/i386/i386.c (ix86_target_string): Add -msgx.
36429 (ix86_option_override_internal): Handle new options.
36430 (ix86_valid_target_attribute_inner_p): Add sgx.
36431 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
36432 * config/i386/i386.opt: Add msgx.
36433 * config/i386/sgxintrin.h: New file.
36434 * config/i386/x86intrin.h: Add sgxintrin.h.
36436 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36439 * fold-const.c (maybe_nonzero_address): Return 1 for function
36441 (tree_single_nonzero_warnv_p): Don't handle function local objects
36445 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
36448 2017-01-11 David Malcolm <dmalcolm@redhat.com>
36451 * opts.c: Include "spellcheck.h"
36452 (struct string_fragment): New struct.
36453 (struct edit_distance_traits<const string_fragment &>): New
36455 (get_closest_sanitizer_option): New function.
36456 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
36458 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36460 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
36462 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
36463 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
36464 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
36465 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
36466 for initial die_offset if dwarf_split_debug_info.
36467 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
36468 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
36470 (output_skeleton_debug_sections): Formatting fix. Use
36471 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
36472 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
36474 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
36476 * config/arm/cortex-a53.md: Add bypasses for
36477 cortex_a53_r2f_cvt.
36478 (cortex_a53_r2f): Only use for transfers.
36479 (cortex_a53_f2r): Likewise.
36480 (cortex_a53_r2f_cvt): Add reservation for conversions.
36481 (cortex_a53_f2r_cvt): Likewise.
36483 2017-01-11 Tamar Christina <tamar.christina@arm.com>
36485 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
36486 to all inlined functions, change static to extern.
36488 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
36491 * config/arm/arm.c (legitimize_pic_address): Handle reference to
36493 (arm_assemble_integer): Likewise.
36495 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
36497 * config.gcc: Use new awk script to check CPU, FPU and architecture
36498 parameters for --with-... options.
36499 * config/arm/parsecpu.awk: New file
36500 * config/arm/arm-cpus.in: New file.
36501 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
36503 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
36505 * config/arm/t-arm: Update dependency rules.
36506 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
36507 of processing .def files.
36508 * config/arm/genopt.sh: Deleted.
36509 * config/arm/gentune.sh: Deleted.
36510 * config/arm/arm-cores.def: Deleted.
36511 * config/arm/arm-arches.def: Deleted.
36512 * config/arm/arm-fpus.def: Deleted.
36513 * config/arm/arm-tune.md: Regenerated.
36514 * config/arm/arm-tables.opt: Regenerated.
36515 * config/arm/arm-cpu.h: New generated file.
36516 * config/arm/arm-cpu-data.h: New generated file.
36517 * config/arm/arm-cpu-cdata.h: New generated file.
36519 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
36522 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
36524 (input_varpool_node): Unpack dynamically_initialized bit.
36526 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
36528 PR rtl-optimization/79032
36529 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
36530 the alignment of the adjusted memory reference against that of MODE,
36531 instead of the alignment of the original memory reference.
36533 2017-01-11 Martin Jambor <mjambor@suse.cz>
36535 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
36537 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
36538 decorated functions.
36540 2017-01-11 Richard Biener <rguenther@suse.de>
36542 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
36543 set range/nonnull info for PHI results. Do not set it on
36544 stmts marked for removal.
36546 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
36548 * expr.c (store_field): In the bitfield case, fetch the return value
36549 from the registers before applying a single big-endian adjustment.
36550 Always do a final load for a BLKmode value not larger than a word.
36552 2017-01-10 David Malcolm <dmalcolm@redhat.com>
36555 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
36556 that we correctly handle column numbers greater than
36557 LINE_MAP_MAX_COLUMN_NUMBER.
36559 2017-01-10 Martin Sebor <msebor@redhat.com>
36561 PR middle-end/78245
36562 * gimple-ssa-sprintf.c (get_destination_size): Call
36563 {init,fini}object_sizes.
36564 * tree-object-size.c (addr_object_size): Adjust.
36565 (pass_through_call): Adjust.
36566 (pass_object_sizes::execute): Adjust.
36567 * tree-object-size.h (fini_object_sizes): Declare.
36569 2017-01-10 Martin Sebor <msebor@redhat.com>
36571 PR tree-optimization/78775
36572 * builtins.c (get_size_range): Move...
36573 * calls.c: ...to here.
36574 (alloc_max_size): Accept zero argument.
36575 (operand_signed_p): Remove.
36576 (maybe_warn_alloc_args_overflow): Call get_size_range.
36577 * calls.h (get_size_range): Declare.
36579 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
36581 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
36582 from TI's devices.csv file as of September 2016.
36583 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
36585 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
36587 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
36588 * doc/invoke.texi: Likewise.
36589 * doc/md.texi: Likewise.
36590 * doc/objc.texi: Likewise.
36592 2017-01-10 Joshua Conner <joshconner@google.com>
36594 * config/arm/fuchsia-elf.h: New file.
36595 * config/fuchsia.h: New file.
36596 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
36597 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
36599 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
36601 2016-01-10 Richard Biener <rguenther@suse.de>
36603 PR tree-optimization/79034
36604 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
36605 Propagate out degenerate PHIs in the joiner.
36607 2017-01-10 Martin Liska <mliska@suse.cz>
36609 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
36610 (sort_congruence_classes_by_decl_uid): Likewise.
36611 (sort_congruence_class_groups_by_decl_uid): Likewise.
36612 (sem_item_optimizer::merge_classes): Sort class, groups in these
36613 classes and members in the groups by DECL_UID of declarations.
36614 This would make merge operations stable.
36616 2017-01-10 Martin Liska <mliska@suse.cz>
36618 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
36619 usage of m_classes_vec.
36620 (sem_item_optimizer::~sem_item_optimizer): Likewise.
36621 (sem_item_optimizer::get_group_by_hash): Likewise.
36622 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
36623 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
36624 (sem_item_optimizer::verify_classes): Likewise.
36625 (sem_item_optimizer::process_cong_reduction): Likewise.
36626 (sem_item_optimizer::dump_cong_classes): Likewise.
36627 (sem_item_optimizer::merge_classes): Likewise.
36628 * ipa-icf.h (congruence_class_hash): Rename from
36629 congruence_class_group_hash. Remove declaration of m_classes_vec.
36631 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
36633 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
36634 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
36635 * config.gcc: Add avx512vpopcntdqintrin.h.
36636 * config/i386/avx512vpopcntdqintrin.h: New.
36637 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
36638 * config/i386/i386-builtin-types.def: Add new types.
36639 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
36640 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
36641 __builtin_ia32_vpopcountq_v8di_mask): New.
36642 * config/i386/i386-c.c (ix86_target_macros_internal): Define
36643 __AVX512VPOPCNTDQ__.
36644 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
36645 (PTA_AVX512VPOPCNTDQ): Define.
36646 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
36647 TARGET_AVX512VPOPCNTDQ_P): Define.
36648 * config/i386/i386.opt: Add mavx512vpopcntdq.
36649 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
36650 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
36652 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36654 PR middle-end/77484
36655 * predict.def (PRED_CALL): Set to 67.
36657 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
36659 * expr.c (store_field): In the bitfield case, if the value comes from
36660 a function call and is of an aggregate type returned in registers, do
36661 not modify the field mode; extract the value in all cases if the mode
36662 is BLKmode and the size is not larger than a word.
36664 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
36667 * config/i386/cpuid.h: Fix undefined behavior.
36669 2017-01-04 Jeff Law <law@redhat.com>
36671 PR tree-optimization/79007
36672 PR tree-optimization/67955
36673 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
36674 conservative for pt.null when flag_non_call_exceptions is on.
36676 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36678 PR translation/79019
36679 PR translation/79020
36680 * params.def (PARAM_INLINE_MIN_SPEEDUP,
36681 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
36682 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
36684 * config/avr/avr.opt (maccumulate-args): Likewise.
36685 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
36686 * common.opt (freport-bug): Likewise.
36687 * cif-code.def (CIF_FINAL_ERROR): Likewise.
36688 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
36689 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
36690 translatable string.
36691 * config/i386/i386.c (function_value_32): Likewise.
36692 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
36693 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
36695 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
36696 * common/config/msp430/msp430-common.c (msp430_handle_option):
36698 * symtab.c (symtab_node::verify_base): Likewise.
36699 * opts.c (set_debug_level): Likewise.
36700 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
36701 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
36702 missing whitespace to translatable strings.
36703 * config/avr/avr.md (bswapsi2): Fix typo in comment.
36704 * config/sh/superh.h: Likewise.
36705 * config/i386/xopintrin.h: Likewise.
36706 * config/i386/znver1.md: Likewise.
36707 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
36708 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
36709 * double-int.h (struct double_int): Likewise.
36710 * double-int.c (div_and_round_double): Likewise.
36711 * wide-int.cc: Likewise.
36712 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
36713 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
36714 * cfgcleanup.c (crossjumps_occured): Renamed to ...
36715 (crossjumps_occurred): ... this.
36716 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
36719 PR tree-optimization/78899
36720 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
36721 returning bool return struct loop *, NULL for failure and the new
36723 (versionable_outer_loop_p): Don't version outer loop if it has
36724 dont_vectorized bit set.
36725 (tree_if_conversion): When versioning outer loop, ensure
36726 tree_if_conversion is performed also on the inner loop of the
36727 non-vectorizable outer loop copy.
36728 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
36729 LOOP_VECTORIZED in inner loop of the scalar outer loop and
36730 prevent vectorization of it.
36731 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
36732 the outer loop vectorization of the non-scalar version is attempted
36733 before vectorization of the inner loop in scalar version. If
36734 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
36735 vectorization of its inner loop.
36736 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
36737 has 2 inner loops, rename also on edges from bb whose single pred
36738 is outer_loop->header. Fix typo in function comment.
36740 2017-01-09 Martin Sebor <msebor@redhat.com>
36743 * asan.c (asan_emit_stack_protection): Increase local buffer size
36744 to avoid snprintf truncation warning.
36746 2017-01-09 Andrew Pinski <apinski@cavium.com>
36748 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
36749 to reference thunderx2t99 for the tuning structure
36750 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
36752 (thunderx2t99_extra_costs): This.
36753 * config/aarch64/aarch64-tune.md: Regenerate.
36754 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
36755 (vulcan_addrcost_table): This.
36756 (vulcan_regmove_cost): Rename to ...
36757 (thunderx2t99_regmove_cost): This.
36758 (vulcan_vector_cost): Rename to ...
36759 (thunderx2t99_vector_cost): this.
36760 (vulcan_branch_cost): Rename to ...
36761 (thunderx2t99_branch_cost): This.
36762 (vulcan_tunings): Rename to ...
36763 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
36764 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
36766 2017-01-09 Martin Jambor <mjambor@suse.cz>
36770 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
36771 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
36772 (propagate_vr_accross_jump_function): Use the above function for all
36773 value range computations for pass-through jump functions and type
36774 converasion from explicit value range values.
36775 (ipcp_propagate_stage): Do not attempt to deduce types of formal
36776 parameters from TYPE_ARG_TYPES.
36777 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
36778 (ipa_write_node_info): Stream type of the actual argument.
36779 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
36781 2017-01-09 Martin Liska <mliska@suse.cz>
36784 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
36785 (lookup_compiler): Do not show error message with have_E.
36787 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36789 PR tree-optimization/78938
36790 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
36791 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
36792 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
36793 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
36796 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36798 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
36801 2017-01-09 Richard Biener <rguenther@suse.de>
36803 PR tree-optimization/78997
36804 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
36805 name condition properly.
36807 2017-01-09 Richard Biener <rguenther@suse.de>
36810 * dwarf2out.c (is_cxx): New overload with context.
36811 (is_naming_typedef_decl): Use it.
36813 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36815 * invoke.texi (Option Summary): Correct spacing in option lists
36816 and add line breaks to fix over-long lines.
36818 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36820 PR middle-end/17660
36822 * extend.texi (Common Variable Attributes): Add xref to GCC
36823 Internals manual to explain mode attribute keywords.
36825 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36828 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
36829 and Preprocessor Options.
36830 (Options for Linking): Document -pthread here....
36831 (RS/6000 and PowerPC Options): ...not here.
36832 (Solaris 2 Options): ...or here.
36833 * doc/cppopts.texi: Document -pthread.
36835 2017-01-08 Martin Sebor <msebor@redhat.com>
36837 PR middle-end/77708
36838 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
36839 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
36840 New member functions.
36841 (format_directive): Used them.
36843 (pass_sprintf_length::handle_gimple_call): Same.
36844 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
36845 to avoid truncation for any argument.
36846 (extract_affine_mul): Same.
36847 * tree.c (get_file_function_name): Same.
36849 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36851 PR middle-end/77484
36852 * predict.def (PRED_INDIR_CALL): Set to 86.
36854 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36856 PR preprocessor/54124
36857 * doc/cppopts.texi: Reformat -d subtable to list the full name
36858 of the options. Add cross-reference to the docs for the general
36859 compiler -d options.
36860 * doc/invoke.texi (Developer Options): Add cross-reference to the
36861 preprocessor-specific -d option documentation.
36863 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36865 PR preprocessor/13498
36866 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
36867 redudant material, and reflect new command-line options.
36868 (System Headers): Likewise.
36870 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36872 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
36873 -isystem, and -idirafter. Copy-edit.
36874 * doc/cppopts.texi: Copy-edit. Remove contradiction about
36875 default for -ftrack-macro-expansion. Delete obsolete and
36876 badly-formatted implementation details about -fdebug-cpp output.
36877 * doc/cppwarnopts.texi: Copy-edit.
36879 2017-01-07 David Malcolm <dmalcolm@redhat.com>
36882 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
36883 that the transition from a max line width >= 1<<10 to narrower
36884 lines works correctly.
36886 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
36888 * doc/options.texi (PerFunction): New.
36889 * opt-functions.awk (switch_flags): Map both Optimization and
36890 PerFunction to CL_OPTIMIZATION.
36891 * opth-gen.awk: Test for PerFunction flag along with
36893 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
36894 it only when the latter is present. Skip those that don't in
36895 the hash function generator.
36896 * common.opt (fvar-tracking): Mark as PerFunction instead of
36898 (fvar-tracking-assignments): Likewise.
36899 (fvar-tracking-assignments-toggle): Likewise.
36900 (fvar-tracking-uninit): Likewise.
36902 2017-01-07 Jakub Jelinek <jakub@redhat.com>
36904 PR translation/79018
36905 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
36908 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
36911 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
36912 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
36913 TARGET_LONG_JUMP_TABLE_OFFSETS.
36914 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
36915 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
36916 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
36917 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
36918 * config/m68k/m68k.md (tablejump expander): Likewise.
36919 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
36920 TARGET_LONG_JUMP_TABLE_OFFSETS.
36921 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
36922 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
36924 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
36925 David Holsgrove <david.holsgrove@xilinx.com>
36927 * common/config/microblaze/microblaze-common.c
36928 (TARGET_EXCEPT_UNWIND_INFO): Remove.
36929 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
36931 * config/microblaze/microblaze.c (microblaze_must_save_register)
36932 (microblaze_expand_epilogue, microblaze_return_addr): Handle
36934 (microblaze_eh_return): New function.
36935 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
36936 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
36937 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
36938 * config/microblaze/microblaze.md (eh_return): New pattern.
36940 2017-01-06 Jakub Jelinek <jakub@redhat.com>
36942 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
36943 GCC_DIAGNOSTIC_STRINGIFY): Define.
36945 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
36947 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36949 * config/arm/arm.md (<mcrr>): New.
36951 * config/arm/arm.c (arm_arch5te): New.
36952 (arm_option_override): Set arm_arch5te.
36953 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
36955 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
36956 (arm_mcrr_qualifiers): ... this. New.
36957 (MRRC_QUALIFIERS): Define to...
36958 (arm_mrrc_qualifiers): ... this. New.
36959 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
36961 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
36962 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
36963 (MRRCI, mrrc, MRRC): New.
36964 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
36965 VUNSPEC_MRRC2): New.
36967 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36969 * config/arm/arm.md (<mcr>): New.
36971 * config/arm/arm.c (arm_coproc_builtin_available): Add
36972 support for mcr, mrc, mcr2 and mrc2.
36973 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
36974 (arm_mcr_qualifiers): ... this. New.
36975 (MRC_QUALIFIERS): Define to ...
36976 (arm_mrc_qualifiers): ... this. New.
36977 (MCR_QUALIFIERS): Define to ...
36978 (arm_mcr_qualifiers): ... this. New.
36979 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
36981 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
36982 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
36983 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
36984 VUNSPEC_MRC2): New.
36986 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36988 * config/arm/arm.md (*ldc): New.
36992 * config/arm/arm.c (arm_coproc_builtin_available): Add
36993 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
36994 (arm_coproc_ldc_stc_legitimate_address): New.
36995 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
36996 'qualifier_const_pointer'.
36997 (LDC_QUALIFIERS): Define to...
36998 (arm_ldc_qualifiers): ... this. New.
36999 (STC_QUALIFIERS): Define to...
37000 (arm_stc_qualifiers): ... this. New.
37001 * config/arm/arm-protos.h
37002 (arm_coproc_ldc_stc_legitimate_address): New.
37003 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
37004 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
37005 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
37006 stc2, stcl, stc2l): New.
37007 * config/arm/constraints.md (Uz): New.
37008 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
37009 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
37010 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
37011 VUNSPEC_STC2L): New.
37013 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37015 * config/arm/arm.md (<cdp>): New.
37016 * config/arm/arm.c (neon_const_bounds): Rename this ...
37017 (arm_const_bounds): ... this.
37018 (arm_coproc_builtin_available): New.
37019 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
37020 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
37021 (CDP_QUALIFIERS): Define to...
37022 (arm_cdp_qualifiers): ... this. New.
37024 (arm_expand_builtin_args): Add case for 6 arguments.
37025 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
37026 (arm_const_bounds): ... this.
37027 (arm_coproc_builtin_available): New.
37028 * config/arm/arm_acle.h (__arm_cdp): New.
37030 * config/arm/arm_acle_builtins.def (cdp): New.
37032 * config/arm/iterators.md (CDPI,CDP,cdp): New.
37033 * config/arm/neon.md: Rename all 'neon_const_bounds' to
37034 'arm_const_bounds'.
37035 * config/arm/types.md (coproc): New.
37036 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
37037 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
37038 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
37039 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
37041 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37043 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
37044 (UBINOP_QUALIFIERS): New.
37046 (acle_builtin_data): New. Change comment.
37047 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
37048 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
37049 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
37050 arm_acle_builtins.def.
37051 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
37052 (arm_init_acle_builtins): New.
37053 (CRC32_BUILTIN): Remove.
37054 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
37055 crc32cb, crc32ch and crc32cw.
37056 (arm_init_crc32_builtins): Remove.
37057 (arm_init_builtins): Use arm_init_acle_builtins rather
37058 than arm_init_crc32_builtins.
37059 (arm_expand_acle_builtin): New.
37060 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
37061 * config/arm/arm_acle_builtins.def: New.
37063 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37065 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
37066 (arm_builtin_datum): ... this.
37067 (arm_init_neon_builtin): Rename to ...
37068 (arm_init_builtin): ... this. Add a new parameters PREFIX
37069 and USE_SIG_IN_NAME.
37070 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
37071 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
37072 'arm_builtin_datum'.
37073 (arm_init_vfp_builtins): Likewise.
37074 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
37075 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
37076 (arm_expand_neon_args): Rename to ...
37077 (arm_expand_builtin_args): ... this. Rename builtin_arg
37078 enum values and differentiate between ARG_BUILTIN_MEMORY
37079 and ARG_BUILTIN_NEON_MEMORY.
37080 (arm_expand_neon_builtin_1): Rename to ...
37081 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
37082 values, arm_expand_builtin_args and add bool parameter NEON.
37083 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
37084 (arm_expand_vfp_builtin): Likewise.
37085 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
37087 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37089 PR middle-end/77484
37090 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
37091 * predict.c (tree_estimate_probability_bb): Reverse direction of
37092 polymorphic call predictor.
37094 2017-01-06 David Malcolm <dmalcolm@redhat.com>
37096 * passes.c (execute_one_pass): Split out pass-skipping logic into...
37097 (determine_pass_name_match): ...this new function and...
37098 (should_skip_pass_p): ...this new function.
37100 2017-01-06 Nathan Sidwell <nathan@acm.org>
37102 * ipa-visibility.c (function_and_variable_visibility): Reformat
37103 comments and long lines. Remove extrneous if.
37104 * symtab.c (symtab_node::make_decl_local): Fix code format.
37105 (symtab_node::set_section_for_node): Fix comment typo.
37107 2017-01-06 Martin Liska <mliska@suse.cz>
37110 * lra-constraints.c: Rename invariant to lra_invariant.
37111 * predict.c (set_even_probabilities): Initialize e to NULL.
37113 2017-01-05 Martin Sebor <msebor@redhat.com>
37115 PR tree-optimization/78910
37116 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
37117 (format_integer): Correct off-by-one error in the handling
37118 of precision with negative numbers in signed conversions..
37120 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
37122 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
37124 2017-01-05 Jakub Jelinek <jakub@redhat.com>
37126 PR tree-optimization/71016
37127 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
37128 factor_out_conditional_conversion. Formatting fix.
37129 (factor_out_conditional_conversion): Add cond_stmt argument.
37130 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
37131 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
37134 2017-01-05 David Malcolm <dmalcolm@redhat.com>
37136 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
37137 read-rtl-function.o, and selftest-rtl.o.
37138 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
37139 (selftest::aarch64_test_loading_full_dump): New function.
37140 (selftest::aarch64_run_selftests): New function.
37141 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
37142 selftest::aarch64_run_selftests.
37143 * config/i386/i386.c
37144 (selftest::ix86_test_loading_dump_fragment_1): New function.
37145 (selftest::ix86_test_loading_call_insn): New function.
37146 (selftest::ix86_test_loading_full_dump): New function.
37147 (selftest::ix86_test_loading_unspec): New function.
37148 (selftest::ix86_run_selftests): Call the new functions.
37149 * emit-rtl.c (maybe_set_max_label_num): New function.
37150 * emit-rtl.h (maybe_set_max_label_num): New decl.
37151 * function.c (instantiate_decls): Guard call to
37152 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
37153 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
37155 * gensupport.c (gen_reader::gen_reader): Pass "false"
37156 for new "compact" param of rtx_reader.
37157 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
37158 rather than an empty string for NULL strings.
37159 * read-md.c: Potentially include config.h rather than bconfig.h.
37160 Wrap include of errors.h with #ifdef GENERATOR_FILE.
37161 (have_error): New global, copied from errors.c.
37162 (md_reader::read_name): Rename to...
37163 (md_reader::read_name_1): ...this, adding "out_loc" param,
37164 and converting "missing name or number" to returning false, rather
37166 (md_reader::read_name): Reimplement in terms of read_name_1.
37167 (md_reader::read_name_or_nil): New function.
37168 (md_reader::read_string): Handle "(nil)" by returning NULL.
37169 (md_reader::md_reader): Add new param "compact".
37170 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
37171 (md_reader::read_file): New method.
37172 * read-md.h (md_reader::md_reader): Add new param "compact".
37173 (md_reader::read_file): New method.
37174 (md_reader::is_compact): New accessor.
37175 (md_reader::read_name): Convert return type from void to file_location.
37176 (md_reader::read_name_or_nil): New decl.
37177 (md_reader::read_name_1): New decl.
37178 (md_reader::m_compact): New field.
37179 (noop_reader::noop_reader): Pass "false" for new "compact" param
37181 (rtx_reader::rtx_reader): Add new "compact" param.
37182 (rtx_reader::read_rtx_operand): Make virtual and convert return
37183 type from void to rtx.
37184 (rtx_reader::read_until): New decl.
37185 (rtx_reader::handle_any_trailing_information): New virtual function.
37186 (rtx_reader::postprocess): New virtual function.
37187 (rtx_reader::finalize_string): New virtual function.
37188 (rtx_reader::m_in_call_function_usage): New field.
37189 (rtx_reader::m_reuse_rtx_by_id): New field.
37190 * read-rtl-function.c: New file.
37191 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
37192 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
37193 (selftest::verify_three_block_rtl_cfg): New decl.
37194 * read-rtl-function.h: New file.
37195 * read-rtl.c: Potentially include config.h rather than bconfig.h.
37196 For host, include function.h, memmodel.h, and emit-rtl.h.
37197 (one_time_initialization): New function.
37198 (struct compact_insn_name): New struct.
37199 (compact_insn_names): New array.
37200 (find_code): Handle insn codes in compact dumps.
37201 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
37202 (bind_subst_iter_and_attr): Likewise.
37203 (add_condition_to_string): Likewise.
37204 (add_condition_to_rtx): Likewise.
37205 (apply_attribute_uses): Likewise.
37206 (add_current_iterators): Likewise.
37207 (apply_iterators): Likewise.
37208 (initialize_iterators): Guard usage of apply_subst_iterator with
37209 #ifdef GENERATOR_FILE.
37210 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
37211 (md_reader::read_mapping): Likewise.
37212 (add_define_attr_for_define_subst): Likewise.
37213 (add_define_subst_attr): Likewise.
37214 (read_subst_mapping): Likewise.
37215 (check_code_iterator): Likewise.
37216 (rtx_reader::read_rtx): Likewise. Move one-time initialization
37218 (one_time_initialization): New function.
37219 (rtx_reader::read_until): New method.
37220 (read_flags): New function.
37221 (parse_reg_note_name): New function.
37222 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
37223 Handle reuse_rtx ids.
37224 Wrap iterator lookup within #ifdef GENERATOR_FILE.
37225 Add parsing support for RTL dumps, mirroring the special-cases in
37226 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
37227 values, and calling handle_any_trailing_information.
37228 (rtx_reader::read_rtx_operand): Convert return type from void
37229 to rtx, returning return_rtx. Handle case 'e'. Call
37230 finalize_string on XSTR and XTMPL fields.
37231 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
37232 "(nil)" values were omitted. Call the postprocess vfunc on the
37234 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
37235 class ctor. Initialize m_in_call_function_usage. Call
37236 one_time_initialization.
37237 * rtl-tests.c (selftest::test_uncond_jump): Call
37238 set_new_first_and_last_insn.
37239 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
37240 * selftest-rtl.c: New file.
37241 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
37242 (selftest::get_insn_by_uid): New decl.
37243 * selftest-run-tests.c (selftest::run_tests): Call
37244 read_rtl_function_c_tests.
37245 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
37246 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
37249 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
37251 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
37252 operands in a special way. Assert that pos+len <= mode precision.
37254 2017-01-05 Jakub Jelinek <jakub@redhat.com>
37256 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
37257 3 argument Alias with unlimited for the negative form.
37258 (fno-vect-cost-model): Removed.
37260 2017-01-05 Martin Liska <mliska@suse.cz>
37262 * hsa-gen.c (gen_hsa_divmod): New function.
37263 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
37265 2017-01-05 Martin Liska <mliska@suse.cz>
37268 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
37271 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37273 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
37274 small constant length operands.
37276 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37278 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
37279 between loop iterations.
37281 2017-01-05 Martin Liska <mliska@suse.cz>
37284 * gimplify.c (gimplify_decl_expr): Compare to
37285 asan_poisoned_variables instread of checking flags.
37286 (gimplify_target_expr): Likewise.
37287 (gimplify_expr): Likewise.
37288 (gimplify_function_tree): Conditionally initialize
37289 asan_poisoned_variables.
37291 2017-01-04 Jeff Law <law@redhat.com>
37293 PR tree-optimizatin/78812
37294 * rtl.h (contains_mem_rtx_p): Prototype.
37295 * ifcvt.c (containts_mem_rtx_p): Move from here to...
37296 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
37297 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
37298 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
37299 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
37301 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37303 * input.c (assert_char_at_range): Default-initialize actual_range.
37305 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37307 * df-scan.c (df_ref_create_structure): Make regno unsigned,
37308 to match the caller.
37310 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37312 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
37313 insns after final jump in test to emit dummy move.
37315 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37317 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
37318 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
37320 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37322 * multiple_target.c (create_dispatcher_calls): Init e_next.
37323 * tree-ssa-loop-split.c (split_loop): Init border.
37324 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
37327 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
37332 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
37333 (altivec_register_operand): Do not return true if the operand
37334 contains a SUBREG mixing SImode and SFmode.
37335 (vsx_register_operand): Likewise.
37336 (vsx_reg_sfsubreg_ok): New predicate.
37337 (vfloat_operand): Do not return true if the operand contains a
37338 SUBREG mixing SImode and SFmode.
37339 (vint_operand): Likewise.
37340 (vlogical_operand): Likewise.
37341 (gpc_reg_operand): Likewise.
37342 (int_reg_operand): Likewise.
37343 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
37344 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
37345 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
37347 (rs6000_emit_move_si_sf_subreg): New helper function.
37348 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
37349 fixup SUBREGs involving SImode and SFmode.
37350 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
37351 numbers for the new peephole2 optimization.
37352 (peephole2 for SFmode unions): New peephole2 to optimize cases in
37353 the GLIBC math library that do AND/IOR/XOR operations on single
37354 precision floating point.
37355 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
37356 target macros to say whether we need to avoid SUBREGs mixing
37358 (TARGET_ALLOW_SF_SUBREG): Likewise.
37359 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
37360 (UNSPEC_SI_FROM_SF): Likewise.
37361 (iorxor): Change spacing.
37362 (and_ior_xor): New iterator for AND, IOR, and XOR.
37363 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
37364 (movdi_from_sf_zero_ext): Likewise.
37365 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
37366 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
37367 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
37368 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
37369 (fms<mode>4): Likewise.
37370 (fnma<mode>4): Likewise.
37371 (fnms<mode>4): Likewise.
37372 (nfma<mode>4): Likewise.
37373 (nfms<mode>4): Likewise.
37375 2017-01-04 Marek Polacek <polacek@redhat.com>
37378 * doc/invoke.texi: Document -Wpointer-compare.
37380 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37382 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
37385 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
37386 descriptions for -gdwarf-5 and emit them as uleb128 instead of
37389 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
37392 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
37393 documentation of the powerpc_popcntb_ok attribute.
37394 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
37395 code to issue warning messages if a requested CPU configuration is
37396 not supported by the binary (assembler and loader) toolchain.
37397 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
37398 made to define a built-in function that has been disabled.
37399 (paired_init_builtins): Add assertion to prevent ICE if attempt is
37400 made to define a built-in function that has been disabled.
37401 (altivec_init_builtins): Add comment explaining why definition
37402 of the DST built-in functions is not preceded by an assertion
37403 check. Add assertions to prevent ICE if attempts are made to
37404 define an altivec predicate or an abs* built-in function that has
37406 (htm_init_builtins): Add comment explaining why definition of the
37407 htm built-in functions is not preceded by an assertion check.
37409 2017-01-04 Jeff Law <law@redhat.com>
37411 PR tree-optimizatin/67955
37412 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
37413 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
37414 the points-to solution does not include pt_null. Use DECL_PT_UID
37417 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
37419 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
37420 Use gen_int_mode instead of gen_lopwart for const_int operands.
37422 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37424 PR tree-optimization/71563
37425 * match.pd: Simplify X << Y into X if Y is known to be 0 or
37426 out of range value - has low bits known to be zero.
37428 2017-01-04 Alan Modra <amodra@gmail.com>
37430 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
37431 * configure: Regenerate.
37432 * config.in: Regenerate.
37434 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37437 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
37438 a substring of the message, but strcmp with the whole message. Ifdef
37439 ENABLE_NLS, translate the message first using dgettext.
37441 2017-01-03 Jeff Law <law@redhat.com>
37443 PR tree-optimizatin/78856
37444 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
37445 (mark_threaded_blocks): Remove code to truncate thread paths that
37446 cross multiple loop headers. Instead invalidate the cached loop
37447 iteration information and handle case of a thread path walking
37448 into an irreducible region.
37450 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
37453 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
37454 assertions. Add support for doing the signbit if the IEEE 128-bit
37455 floating point value is in a GPR.
37456 * config/rs6000/rs6000.md (Fsignbit): Delete.
37457 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
37458 Update the length attribute if the value is in a GPR.
37459 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
37460 the sign or zero extension instruction, since the value is always 0/1.
37461 (signbit<mode>2_dm2): Delete using <Fsignbit>.
37464 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
37465 extracting SImode to a GPR register so that we can generate a
37466 store, limit the vector to be in a traditional Altivec register
37467 for the vextuwrx instruction.
37469 2017-01-03 Ian Lance Taylor <iant@google.com>
37471 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
37473 2017-01-03 Martin Sebor <msebor@redhat.com>
37475 PR tree-optimization/78696
37476 * gimple-ssa-sprintf.c (format_floating): Correct handling of
37477 precision. Use MPFR for %f for greater fidelity. Correct handling
37479 (pass_sprintf_length::compute_format_length): Set width and precision
37480 specified by asrerisk to void_node for vararg functions.
37481 (try_substitute_return_value): Adjust dump output.
37483 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
37485 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
37487 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
37489 * doc/invoke.texi (SPARC options): Document -mlra as the default.
37490 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
37491 -mlra/-mno-lra was passed to the compiler.
37493 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
37495 PR rtl-optimization/65618
37496 * emit-rtl.c (try_split): Move initialization of "before" and
37497 "after" to just before the call to emit_insn_after_setloc.
37499 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
37501 * doc/md.texi (Standard Names): Remove reference to Java frontend.
37503 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
37505 * dwarf2out.c (gen_enumeration_type_die): When
37506 -gno-strict-dwarf, add a DW_AT_encoding attribute.
37508 2017-01-03 Jakub Jelinek <jakub@redhat.com>
37510 PR tree-optimization/78965
37511 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
37512 Change first argument from const call_info & to call_info &. For %n
37513 set info.nowrite to false.
37515 PR middle-end/78901
37516 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
37517 possibly throwing calls.
37519 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
37520 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
37521 and fns handling, rather than in a separate case SSA_NAME.
37523 2017-01-02 Jeff Law <law@redhat.com>
37525 * config/darwin-driver.c (darwin_driver_init): Const-correctness
37526 fixes for first_period and second_period variables.
37528 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
37531 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
37532 (*insvqi_1): New insn pattern.
37533 (*insvqi_1_mem_rex64): Ditto.
37534 (*insvqi_2): Ditto.
37535 (*insvqi_3): Rename from *insvqi.
37537 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
37539 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
37541 * doc/cfg.texi (Edges): Remove reference to Java.
37542 (Maintaining the CFG): Ditto.
37544 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37546 PR middle-end/77674
37547 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
37548 transparent aliases.
37550 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37552 PR middle-end/77484
37553 * predict.def (PRED_CALL): Update hitrate.
37554 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
37555 * predict.c (tree_estimate_probability_bb): Split CALL predictor
37556 into direct/indirect/polymorphic variants.
37558 2017-01-01 Jakub Jelinek <jakub@redhat.com>
37560 Update copyright years.
37562 * gcc.c (process_command): Update copyright notice dates.
37563 * gcov-dump.c (print_version): Ditto.
37564 * gcov.c (print_version): Ditto.
37565 * gcov-tool.c (print_version): Ditto.
37566 * gengtype.c (create_file): Ditto.
37567 * doc/cpp.texi: Bump @copying's copyright year.
37568 * doc/cppinternals.texi: Ditto.
37569 * doc/gcc.texi: Ditto.
37570 * doc/gccint.texi: Ditto.
37571 * doc/gcov.texi: Ditto.
37572 * doc/install.texi: Ditto.
37573 * doc/invoke.texi: Ditto.
37575 Copyright (C) 2017 Free Software Foundation, Inc.
37577 Copying and distribution of this file, with or without modification,
37578 are permitted in any medium without royalty provided the copyright
37579 notice and this notice are preserved.